3-5 | December | 2024

www.PAafrica.org | #AfCPA2024

PRECISION AGRICULTURE
in ACTION for AFRICA

PROCEEDINGS
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

i
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

PROCEEDINGS OF THE
3rd African Conference on
Precision Agriculture (AfCPA)

www.PAafrica.org | #AfCPA2024

3-5 | December | 2024

PUBLISHED BY:
African Plant Nutrition Institute
Benguérir, Morocco
www.apni.net

CONFERENCE WEBSITE:
www.paafrica.org

This publication contains the full paper submissions from the AfCPA. These papers, along
with all other accepted abstract submissions are available at the conference website.

ii
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

3rd African Conference on

Precision Agriculture (AfCPA)

ORGANIZED BY:

SPONSORED BY:

iii
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

TABLE OF CONTENTS

ADOPTION OF PRECISION AGRICULTURE ............................................................................... 9

A CALL FOR ACTION: IMPLEMENTING THE AFRICAN CERTIFIED AGRONOMY ADVISORY
PROGRAM ........................................................................................................................................................... 10
Canon Engoke Norris Savala, Ivan Adolwa, Esther Mugi, Samuel Njoroge, Kwame Frimpong, James
*

Mutegi, Kaushik Majumdar, Shamie Zingore

HIGH-THROUGHPUT FIELD PHENOTYPING OF ASCOCHYTA BLIGHT DISEASE SEVERITY IN

CHICKPEA USING MULTISPECTRAL IMAGING ..................................................................................... 14
Ibn El Mokhtar Fatima Zahrae, 1*Krimibencheqroun Sanae, 1Dahhane Younes, 1Harkani Assia, 2Houmairi
1,2

Hafsa, 1Idrissi Omar, 1*El-Aissaoui Abdellah

EFFECTS OF ADOPTING FERTILIZER MANAGEMENT PRACTICES ON YIELD IN MAIZE-

BASED SYSTEMS IN EMBU COUNTY, KENYA: AN INSTRUMENTAL VARIABLE REGRESSION
APPROACH ......................................................................................................................................................... 22
1*
Malcolm Gitau, 1Rose Nyikal, 1Evans Chimoita, 2Onesmus Kitonyo

GRAIN YIELD AND NITROGEN UPTAKE OF MAIZE AS AFFECTED BY SOIL MANAGEMENT

PRACTICES AND THEIR INTERACTION ON CAMBISOLS AND CHERNOZEM .............................. 27
1
Ashenafi Nigussie

POLICY IMPLICATIONS OF THE ADOPTION OF PRECISION AGRICULTURE IN NIGERIA ...... 34

1*
Oyedepo, Elizabeth Omolola, 2Oyedepo, John Adebayo

ARTIFICIAL INTELLIGENCE (AI) IN AGRICULTURE ....................................................... 40

AN ENSEMBLE-BASED DEEP LEARNING APPROACH FOR EARLY AND ACCURATE WHEAT
DISEASE DETECTION...................................................................................................................................... 41
1
Tagel Aboneh, 1Prof. Abebe Rorissa, 2Prof. Ramasamy Srinivasagan

MULTIVARIATE REGIONAL DEEP LEARNING PREDICTION OF SOIL PROPERTIES FROM

NEAR-INFRARED, MID-INFRARED AND THEIR COMBINED SPECTRA ........................................... 46
Rumbidzai W. Nyawasha, 3Alexandre M.J.-C. Wadoux, 4,5Pierre Todoroff, 1,6Regis Chikowo, 1,2,5,7Gatien
1,2*

N. Falconnier, 4,5Maeva Lagorsse, 5,8Marc Corbeels, 1,2,5Rémi Cardinael

CLIMATE SMART AGRICULTURE ................................................................................................. 55

PREDICTING THE DISTRIBUTION OF GROUNDNUT PHYTOPATHOGENS UNDER CURRENT
AND FUTURE CLIMATIC SCENARIOS IN ZIMBABWE .......................................................................... 56
Hardlife Chinwa, 1Elizabeth Ngadze, 1Juliet Murimwa, 2,3Eliel Belchrist Sossou, 4Clara Machokoto, 1Regis
1

Chikowo, 1*Casper Nyaradzai Kamutando

DECISION SUPPORT SYSTEMS, SOFTWARE AND MOBILE APPLICATIONS ..... 62

EMPOWERING FARMERS TO SUSTAINABLE AGRICULTURAL PRODUCTIVITY IN WEST

AFRICA: FESERWAM, A DIGITAL ADVISOR FOR FARMERS ............................................................. 63
J. Toviho, I. Brou, K. Kouassi

iv
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

A FRAMEWORK FOR A CROP YIELD PREDICTION MODEL BASED ON DECISION TREE AND
MIN-MAX SCALING ......................................................................................................................................... 67
S.A. Audu, A.A. Rabiu, G.I.O. Aimufua, M. Samaila, A.M. Umar, and M.A. Yau

EDUCATION AND ENGAGEMENT IN PRECISION AGRICULTURE ........................... 72

INTER-COUNTRY COOPERATION CAN TRANSFORM PRECISION AGRICULTURE
EDUCATION AND RESEARCH IN AFRICA ................................................................................................ 73
Kwame Agyei Frimpong

ADVANCING PRECISION AGRICULTURE EDUCATION IN SUB-SAHARAN AFRICA:

EXPLORING FACTORS FOR SUCCESS AND OBSTACLES .................................................................... 78
F.J. Adekoya, B.V. Bamidele, T.E. Adefare

YIELD OF MAIZE (Zea mays L.) VARIETIES AS AFFECTED BY NEEM OIL COATED UREA
APPLICATION IN SOUTHERN BENIN ......................................................................................................... 85
Fassinou Hotegni V. Nicodeme, Houessou C. Gildas, Linkpon F. Orthia, Tchokponhoue A. Dedeou,
*

Houdegbe Carlos, Adje Charlotte, Achigan-Dako Enoch

GEOSPATIAL DATA AND MAPPING .............................................................................................. 92

TOWARD PRECISION AGRICULTURE BY ASSESSING FAO SOIL DATA ACCURACY WITH
LOCALISED SOIL MAPPING IN MID-WESTERN UGANDA ................................................................... 93
1*
Emmanuel Ojambo; 1L. Abasize, 1P. Tamale, 1T. Matila, 1B. Kiconco, 1E. Opolot, 2S. Kusiima, 1P.
Ebanyat, 3J. Mutegi, 3T. Oberthur, 1P. Musinguzi

ASSESSING OF SOIL NUTRIENTS USING LABORATORY AND REMOTE SENSING METHODS

IN NORTHERN GUINEA SAVANNAH........................................................................................................... 99
1
A.R. Garba, 2A.M. Yamusa, 1C.K. Daudu, 2S.L. Yau, 3M.I., Gaya, 2A. Abdulkadir

LAND SUITABILITY PREDICTION FOR MAIZE PRODUCTION IN SOUTHWEST NIGERIA

USING GEOGRAPHICAL INFORMATION SYSTEM AND MOST-LIMITING SOIL NATIVE
FERTILITY FACTORS.................................................................................................................................... 110
1
Taiwo S. Olutoberu, 1Mutiu A. Busari, 1Olusegun Folorunso, 1Muftau Adebayo, 1Sodeeq O. Azeez,
1
Sarafadeen G. Hammed, 1J.A. Oyedepo, 2Oluwafolake Ojo, and 1G.A. Ajiboye

ON-FARM EXPERIMENTATION ..................................................................................................... 117

ON-FARM EXPERIMENTATION PROCESS TRIGGERS FARMERS’ ZEAL TO TEST
TECHNOLOGIES IN MAIZE SYSTEMS OF EMBU COUNTY, KENYA ............................................... 118
1*
Onesmus Kitonyo, Evans Chimoita, Timothy Kamanu, Felister Nzuve, Esther Muindi, Alfred Micheni,
James Muthomi, Vincent Kathumo, Grace Mureithi, George Chemining’wa

THE VALUE AND POTENTIAL OF ON-FARM EXPERIMENTATION TO CATALYZE

AGRICULTURAL TRANSFORMATION ..................................................................................................... 123
1*
Ivan S. Adolwa, 2Eileen Nchanji, 1Joses Muthamia, 2Cosmas Lutomia, 3Simon Cook

PRECISION LIVESTOCK MANAGEMENT ................................................................................ 128

MONITORING GRAZING GOATS' BEHAVIOR USING SENSORS AND SATELLITE REMOTE

SENSING ............................................................................................................................................................ 129
1
Y. Chebli, 1S. El Otmani, 2F. Elame, 3J.F. Cabaraux, 4A. Keli 4, 1M. Chentouf

v
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ADVANCEMENTS IN PRECISION AGRICULTURE: CUTTING-EDGE SOLUTIONS FOR

COCCIDIOSIS DISEASE USING SMART DIAGNOSTIC SYSTEMS ..................................................... 133
1*
F.O. Bamigboye, 2O. Bello, 2O.A. Osaikhuwuomwan, 2P.C. Anakua

PRECISION NUTRIENT MANAGEMENT ................................................................................... 137

ASSESSMENT OF NITROGEN FERTILIZATION IN TUNISIAN WHEAT PRODUCTION USING
PROXIMAL AND REMOTE SENSING......................................................................................................... 138
1*
M. Mechri, O. Alshihabi, H. Angar, I. Nouiri, M.H. Landolsi, M. Söderström, K. Persson, S. Phillips

USE OF “FERTIEDGE” APPLICATION FOR OPTIMIZING WHEAT FERTILIZATION ................ 142

1
I. Sbai, 2M. Hmimou, 1A. Bouslimi, 1M. Hannachi, 2H. Baroudi, 2F. Sehbaoui, 1A. Bousselmi, 1S Arfaoui

COMPARATIVE STUDY ON PRECISION NITROGEN MANAGEMENT FOR WHEAT USING

GREENSEEKER, CHLOROPHYLL METER AND LEAF COLOR CHART BASED ON SPECTRAL
CHARACTERISTICS OF LEAVES ............................................................................................................... 147
1
Abdou A. Soaud, 1A. El-Metwally, 2M. Ali, 1Radwa K. Sayed

OPTIMIZING DURUM WHEAT NITROGEN NUTRITION INDEX (NNI) PREDICTION THROUGH

SENTINEL-2 VEGETATION INDICES INTEGRATIONS ........................................................................ 154
1
N.H Boughattas, 1,2M. Kadri, 2M. Zekri, 3S. Ayadi, 4S. Arraouadi, 5H. Hajlaoui, 3Y. Trifa, 2I. Hbiri

PRECISION PLANTING/HARVESTING ....................................................................................... 160

MODELING OF THE OPTIMAL PARAMETERS, IMPROVEMENT AND PERFORMANCE
EVALUATION OF MARC ENGINE DRIVEN COMMON BEAN (PHASEOLUS Vulgaris L.)
THRESHER IN ETHIOPIA ............................................................................................................................. 161
1*
Biniam Zewdie, 2Adesoji M. Olaniyan, 1Amana Wako, 3Dereje Alemu, 3Tamrat Lema

REMOTE AND PROXIMAL SENSING ........................................................................................... 174

HIGH-RESOLUTION MAPPING OF LONG-TERM SOIL ORGANIC CARBON STOCKS AND

CHANGES IN MOROCCO .............................................................................................................................. 175
Mohamed Bayad, 1Bruno Gerard, 1Abdellah Hamma, 2Malcolm J. Hawkesford, 3Eyal Ben-Dor, 4Henry Wai
1*

Chau

EVALUATING SOIL PARAMETERS USING PROXIMAL SOIL SENSORS (EM38, MSP3) .............. 194
1,2*
H. Hajjaj, 1K. Ibno Namr, 2C. El khalif, 2A. El-Aissaoui

SMALL HOLDERS AND PRECISION AGRICULTURE ....................................................... 200

THE USE OF UNMANNED AERIAL VEHICLE (UAV) REMOTELY SENSED DATA AND
BIOPHYSICAL VARIABLES TO PREDICT MAIZE ABOVE-GROUND BIOMASS (AGB) IN
SMALL-SCALE FARMING SYSTEMS......................................................................................................... 201
1*
Celuxolo M Dlamini, 1John Odindi, 1Onisimo Mutanga, 2Trylee N Matongera

ENHANCING THE ESTIMATION OF EQUIVALENT WATER THICKNESS IN NEGLECTED AND

UNDERUTILIZED TARO CROPS USING UAV ACQUIRED MULTISPECTRAL THERMAL IMAGE
DATA AND INDEX-BASED IMAGE SEGMENTATION ........................................................................... 229
1*
Helen S. Ndlovu, 1John Odindi, 2Mbulisi Sibanda, 1Onisimo Mutanga

vi
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

SUSTAINABILITY OF MAIZE PRODUCTION WITH FARMERS’ PRACTICES AND REDESIGNED

CROP MANAGEMENT PRACTICES IN CRV AND JIMMA OF ETHIOPIA ........................................ 236
1*
Workneh Bekere Kenea, 1Amsalu Nebiyu, 2Tesfaye Balemi

OPTIMIZING MAIZE PRODUCTION THROUGH MINIMUM TILLAGE IN CONSERVATION

AGRICULTURE (CA) SYSTEMS: EVIDENCE FROM MALAWI'S LOWLANDS ............................... 243
1
Donald Nyamayevuab, 2Isaiah Nyagumbob, 1Liang Wei-lia, 1Li Rui-qia

USE OF DIGITAL EXTENSION TOOLS FOR AGRICULTURAL INFORMATION MANAGEMENT

AMONG CASSAVA VALUE CHAIN ACTORS IN IBADAN METROPOLIS, OYO STATE, NIGERIA
.............................................................................................................................................................................. 249
1
Olushola Fadairo, 2Adenike Durojaiye, 3Racheal Adegbayi

EMPOWERING SMALLHOLDER FARMERS: PRECISION AGRICULTURE MODELS IN AFRICA

(A CASE STUDY OF NIGERIA) ..................................................................................................................... 254
Stephen Oluwaseun Popoola

SOCIAL SCIENCE APPLICATIONS IN PRECISION AGRICULTURE ........................ 261

ASSESSMENT OF ACCESS TO AND UTILISATION OF TREADLE AND HIP PUMP
TECHNOLOGY BY FARMERS IN MACHAKOS COUNTY, KENYA .................................................... 262
T. Gitau, G.B. Mose, and C. Masiga

WATER MANAGEMENT FOR PRECISION AGRICULTURE .......................................... 267

REAL-TIME MOISTURE CONTROL IN IRRIGATION SYSTEMS FOR WATER USE EFFICIENCY
AND CLIMATE CHANGE RESILIENCE. A REVIEW .............................................................................. 268
D. Kindikiza

UPLAND RICE YIELD RESPONSE TO SOIL MOISTURE VARIABILITY WITH DEPTH ACROSS
FERRALSOLS AND GLEYSOLS IN WESTERN UGANDA ...................................................................... 272
Matila Tom Micah; 1B. Kiconco; 1L. Abasize; 1P. Tamale; 1E. Opolot; 1P. Ebanyat; 2T. Oberthur, 1P.
1*

Musinguzi

RAINWATER HARVESTING AND NUTRIENT INTENSIFICATION IN MAIZE-LEGUME

FARMING SYSTEMS IN SEMI-ARID ZIMBABWE .................................................................................. 277
*
E. Mutsamba-Magwaza, D. Nyamayevu, I. Nyagumbo, R. Mandumbu, G. Nyamadzawo

FACTORS INFLUENCING FARMERS’ DECISIONS ON TIMING AND APPLICATION RATES OF

IRRIGATION WATER IN MWALA, MACHAKOS COUNTY.................................................................. 282
D.M. Onyancha, C.K.K. Gachene, G. Kironchi

COMPARATIVE ASSESSMENT OF LANDUSE LANDCOVER CHANGES ON WATER QUALITY

OF RIVER KADUNA FROM 2012-2020 AT WUYA, NIGER STATE, NIGERIA ................................... 288
1
S.U. Ibrahim, 2M.O. Dania, 3F.O Yusufu, 4R.J Kolo, 5S.O.E. Sadiku

DESIGN AND DEVELOPMENT OF A LORA COMMUNICATION SYSTEM FOR SCALABLE

SMART IRRIGATION SYSTEMS.................................................................................................................. 295
1*
J. Ikabat, 2P. Hess, 3J. Wanyama, 4E. Bwambale, 5P. Nakawuka, 6I. Kabenge, 7S. Felicioni, 8A. Bühlmann,
9
T. Anken, 10J.H. Muyonga

vii
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

EVALUATING THE IMPACT OF SEASONAL WEATHER VARIABILITY ON SOIL MOISTURE

CONSERVATION UNDER MULCHING SYSTEMS FOR DATE PALM PRODUCTION IN OASES 300
1,3
U. Safi, 2E.J. Mustapha, 1O. Abdallah, 3S. Ahmed

INFLUENCE OF MULCHED-DRIP IRRIGATION SYSTEM ON YIELD AND PHYSIOLOGICAL

ATTRIBUTES OF PEPPER VARIETIES IN SOUTHWEST, NIGERIA .................................................. 305
1
I.A. Kehinde, 1*O.O. Fawibe, 1A.S. Oyelakin

SMARTSPROUT: LEVERAGING IOT FOR SUSTAINABLE AND EFFICIENT SMART

IRRIGATION SYSTEMS IN AFRICA ........................................................................................................... 310
Oyedepo John Adebayo and Oluyege, Damilola Emmanuel

viii
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ADOPTION OF PRECISION AGRICULTURE

9
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

A CALL FOR ACTION: IMPLEMENTING THE AFRICAN CERTIFIED AGRONOMY

ADVISORY PROGRAM
#10939

*Canon Engoke Norris Savala, Ivan Adolwa, Esther Mugi, Samuel Njoroge, Kwame Frimpong,
James Mutegi, Kaushik Majumdar, Shamie Zingore
African Plant Nutrition Institute (APNI), ICIPE Compound, P.O. Box 30772-00100, Nairobi,
Kenya
*
e-mail: [email protected]

ABSTRACT

Agriculture remains the backbone of many African economies, providing livelihood for millions
of people while addressing food security concerns. However, productivity often lags other world
regions due to various challenges, including limited access to modern agronomic practices and
advisory services. In response to the pressing need for enhancing agricultural productivity and
sustainability in African countries, it is imperative to establish a standardized extension training
program at the continental level. The African Certified Agronomy Advisory Program (CAAP)
aims to enhance agricultural productivity, sustainability, and resilience by providing farmers with
access to high quality agronomy advisory services. The CAAP initiative addresses the critical gaps
in access to relevant agronomic knowledge and guidance among smallholder farmers, thereby
promoting sustainable farming practices and improving livelihoods. The CAAP framework
encompasses a multi-faceted approach, including the training and certification of agronomy
professionals, development of strategic partnerships with private, governmental and non-
governmental entities, and the utilization of innovative communication technologies for
information dissemination and sharing. Through country or regional tailored advisory services,
CAAP will provide farmers with practical insights into soil and crop management, pest and disease
control, and climate-resilient farming techniques. Central to the success of CAAP is its emphasis
on local capacity building and community engagement. By empowering agronomic advisors and
farmers with the necessary skills and knowledge, the program seeks to foster a culture of self-
reliance and entrepreneurship within rural communities. Furthermore, monitoring and evaluation
mechanisms will ensure the effectiveness and impact of CAAP interventions, allowing for
continuous refinement and improvement. By collaborative promotion of precision nutrient
management such as 4R practices and empowering smallholder farmers, CAAP has the potential
to enhance food security, mitigate environmental degradation, and contribute to overall economic
development in the region.

Keywords: Adoption, Capacity Development, Climate Change, Communication, Extension,

Partnership, Technology, Training

INTRODUCTION

Agriculture plays a crucial role in the economies of many African nations, serving as a primary
source of livelihood for a significant portion of the population. It contributes substantially to GDP
and employment, particularly in rural regions where subsistence farming predominates.
Agriculture is also vital for export earnings, with commodities such as cocoa, coffee, and tea

10
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

forming the cornerstone of international trade (Jayne et al., 2023). However, despite its
importance, the sector faces a multitude of challenges that limit its full potential. One major
challenge is the limited access to modern agronomic practices and advisory services, which
deprives farmers off essential resources such as knowledge and utilization of high-quality seeds,
fertilizers, and machinery necessary for enhancing productivity and yields (Masangano &
Mthinda, 2017). Furthermore, African agriculture is highly vulnerable to climate change and
variability, with extreme weather events posing significant threats to crop production and livestock
rearing. Strengthened advisory services will improve mitigation technics of smallholder farmers,
who rely heavily on rain-fed agriculture and particularly susceptible to climatic shocks that lead
to food insecurity and economic losses (Alston et al., 2022).

Strengthening of agronomic advisories should accompany financial services to farmers and policy
amendments. Access to financial services and markets is another significant barrier, impeding
farmers' ability to invest in modern inputs and technologies. Policy and regulatory constraints,
bureaucratic inefficiencies, and land tenure issues further hinder agricultural development across
Africa (Bambio et al., 2022). Despite these challenges, there are emerging initiatives aimed at
improving agricultural productivity through sustainable farming practices, enhanced agricultural
extension services, and better access to markets and finance (Camillone et al., 2020). This paper
explores the potential of, and need for the African Certified Agronomy Advisory Program (CAAP)
to address these challenges by building capacity among agronomy professionals, promoting
sustainable agricultural practices, and improving food security and economic resilience in Africa.

Problem Statement and Objectives

Smallholder farmers are the backbone of agricultural production in Africa, yet they often lack
access to the knowledge and resources needed to improve their farming practices. This results in
sub-optimal yields, increased vulnerability to pests and diseases, and heightened susceptibility to
climate variability, all of which negatively impact food security and livelihoods (Alston et al.,
2022). Therefore, CAAP seeks to enhance agricultural productivity, sustainability, and resilience
in Africa by providing farmers with access to high-quality agronomy advisory services. The
program aims to fill critical gaps in agronomic knowledge and guidance, promote the adoption of
sustainable farming practices matching transforming food systems and ultimately improve the
livelihoods of smallholder farmers.

Framework Overview
CAAP is to be built on a comprehensive framework that includes the training and certification of
agronomy professionals, the development of strategic partnerships, the use of innovative
communication technologies, and the provision of tailored advisory services to farmers. A
cornerstone of CAAP is the rigorous training and certification of agronomy professionals. These
individuals will be equipped with the skills and knowledge necessary to provide high-quality
advisory services that help farmers improve their food systems productivity and adopt sustainable
practices (Jayne et al., 2023). CAAP seeks to collaborate with a diverse array of partners, including
private companies, government agencies, research institutions (national and international) and
non-governmental organizations, to leverage resources and expertise. These partnerships are
crucial for knowledge development, policy alignment and effective delivery of agronomy advisory
services and for ensuring the sustainability of CAAP interventions (Masangano & Mthinda, 2017).
The program will utilize innovative communication technologies, such as mobile phones and the

11
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

internet, to train agronomists and disseminate agronomic information to farmers. These

technologies facilitate real-time communication and information sharing, allowing agronomy
advisors and farmers to access timely knowledge, guidance and inturn share feedback for
improvement (Camillone et al., 2020). CAAP will provide tailored advisory services based on the
specific needs and contexts of farmers. These services will include practical insights into soil and
crop management, pest and disease control strategies, and climate-resilient farming techniques
(Bambio et al., 2022).

Case Studies from Africa

In Kenya, extension has worked with smallholder farmers to enhance soil fertility through the
adoption of sustainable soil and crop management practices. By promoting practices such as
conservation agriculture and organic farming, the program helped farmers increase yields while
preserving soil health (Jayne et al., 2023). In Ghana, extension has implemented integrated pest
and disease management strategies to combat common agricultural pests and diseases. Using
biological control methods, crop rotation, and resistant crop varieties, extension reduced reliance
on chemical pesticides while effectively managing pest and disease outbreaks (Masangano &
Mthinda, 2017). In Ethiopia, extension has introduced climate-resilient farming techniques to help
farmers adapt to the challenges posed by climate change. These techniques include the use of
drought-tolerant crop varieties, water harvesting and conservation practices, and agroforestry
systems, enabling farmers to maintain productivity despite changing climatic conditions (Alston
et al., 2022).

Empowerment of Farmers
CAAP emphasizes the empowerment of farmers by providing them with the skills and knowledge
needed to improve their farming practices. By fostering a culture of self-reliance and
entrepreneurship within rural communities, the program aims to enhance agricultural sustainability
and food systems productivity (Bambio et al., 2022). Implementation of the program will also
promote entrepreneurship among farmers by supporting the development of agribusinesses and
value-added enterprises through supply of raw materials. By diversifying income streams and
creating market linkages, CAAP will help farmers improve their economic resilience and
livelihoods (Camillone et al., 2020). CAAP will actively involve local communities in the design
and implementation of its interventions, ensuring that they are culturally appropriate and
contextually relevant. Through activities such as farmer field schools, ‘living labs initiative’ and
participatory research initiatives, CAAP will strengthen social cohesion and ownership of
agricultural development initiatives (Jayne et al., 2023).

Potential Impact of CAAP

Through continuous monitoring and evaluation, CAAP will ensure the effectiveness and impact
of technology and strengthened advisory interventions. These mechanisms allow for the
assessment of progress towards program objectives, identification of challenges and bottlenecks,
and adjustment of strategies as needed (Alston et al., 2022). This strategy will allow CAAP to
refine and improve its approaches over time. With an iterative process approach of learning and
adaptation, CAAP will remain responsive to the evolving needs and priorities of smallholder
farmers in Africa (Masangano & Mthinda, 2017). CAAP will promote sustainable farming
practices and improve agricultural productivity thereby enhancing food security in Africa.
Increased yields and diversified crop production can help ensure a stable and nutritious food supply

12
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

for rural communities (Bambio et al., 2022). CAAP will contribute to environmental conservation
by promoting practices that reduce the use of excessive chemical inputs, conserve soil and water
resources, and mitigate greenhouse gas emissions. The adoption of climate-resilient farming
techniques also helps farmers contribute to climate change mitigation and adaptation efforts
(Camillone et al., 2020). Through the promotion of entrepreneurship and the creation of market
linkages, CAAP will stimulate economic development in rural areas. Also, the program will
promote economic growth and poverty reduction in the region when additional income
opportunities and enhanced agricultural value chains will be created (Jayne et al., 2023).

RECOMMENDATIONS AND CONCLUSION

The African Certified Agronomy Advisory Program (CAAP) represents a comprehensive

approach to enhancing agricultural productivity and sustainability in Africa. CAAP will provide
farmers with access to high-quality agronomy advisory services, promote sustainable farming
practices, and foster community engagement. Therefore, CAAP has the potential to significantly
improve livelihoods and food security across the continent. Further research is needed to assess
the long-term impacts of CAAP interventions and identify areas for improvement. Additionally,
studies on the scalability and replicability of CAAP models across different contexts and regions
would be valuable for informing future program expansion efforts. CAAP offers a promising
solution to the challenges facing smallholder farmers in Africa. By leveraging strategic
partnerships, innovative technologies, and community participation, the program has the potential
to catalyze transformative change in the agricultural food systems sector, leading to more resilient
and sustainable food systems across the continent.

REFERENCES

Alston, J. M., Pardey, P. G., James, J. S., & Andersen, M. A. (2022). Research and Productivity
Growth in Agriculture: A Broader Perspective. American Journal of Agricultural
Economics, 104(3), 763-780.
Bambio, Y., & Agha, H. (2022). Adoption of Sustainable Agricultural Practices in Africa: The
Role of Capacity Building. Journal of Development Studies, 58(6), 925-940.
Camillone, J., de Janvry, A., & Sadoulet, E. (2020). Certification and Agricultural Productivity:
Evidence from a Field Experiment. World Development, 130, 104917.
Jayne, T. S., Chamberlin, J., & Benfica, R. (2023). Agriculture in Africa: Telling Myths from
Facts. Food Policy, 104, 102102.
Masangano, C., & Mthinda, C. (2017). Agricultural Extension in Africa: Experiences from
Malawi. Journal of Agricultural Education and Extension, 23(5), 385-401.

13
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

HIGH-THROUGHPUT FIELD PHENOTYPING OF ASCOCHYTA BLIGHT DISEASE

SEVERITY IN CHICKPEA USING MULTISPECTRAL IMAGING
#11660

1,2Ibn
El Mokhtar Fatima Zahrae, 1*Krimibencheqroun Sanae, 1Dahhane Younes, 1Harkani Assia,
2Houmairi Hafsa, 1Idrissi Omar, 1* El-Aissaoui Abdellah
1 Regional Center of Agricultural Research of Settat, National Institute of Agricultural Research

(INRA), Settat, Morocco

2
Agri-food and Health Laboratory, Faculty of Science and Technology, Hassan First University
of Settat, Settat, Morocco
*e-mail: [email protected]; [email protected]

ABSTRACT

Ascochyta blight (AB) caused by Ascochyta rabiei (Pass.) Labr. is an important and widespread
disease of chickpea (Cicer arietinum L.) worldwide. The disease is particularly severe under cool
and humid weather conditions, leading to crop losses at all stages of chickpea growth.
Screening for resistant cultivars remains the most effective, economic and ecological method
of disease management. However, traditional phenotyping methods that relying on trained experts
are slow, costly, labor-intensive, subjective, often involve destructive sampling. The
development of high-throughput phenotyping methods for Ascochyta blight disease holds promise
for precise and rapid data. In this study, 216 chickpea genotypes were screened in field trials
to investigate the use of digital imaging to implement reliable phenotyping of Ascochyta blight
resistance. An unmanned aerial system equipped with a 5-band multispectral camera was used
to capture imagery of the tested genotype plots. Digital image processing was employed to
extract the NDVI index. Our aim was to explore the correlation between the NDVI index and
visual disease severity ratings for Ascochyta blight. Results revealed a consistent correlation
between the NDVI index extracted from image features and disease severity with R 2 of 0.936.
Genotypes were classified into resistant (R), moderately resistant (MR) susceptible (S) and highly
susceptible categories based on their responses. These differences in genotypes response were
utilized to developpe a predictive model for monitoring Ascochyta blight. Our findings highlight
that rapid and precise image-based, high-throughput phenotyping can effectively differentiate
responses to Ascochyta blight across many chickpea genotypes.

Keywords: Drone imaging, Ascochyta blight, severity, phenotyping, chickpea, NDVI,

INTRODUCTION

Chickpea (Cicer arietinum L.) is a highly valuable crop, providing an important source of protein
and improving soil health through nitrogen fixation. However, its production is severely affected
by various abiotic and biotic stresses, including drought and diseases. Among these, Ascochyta
blight (AB), caused by Ascochyta rabiei (Kovatsch.) Arx, 1962, is a major biotic threat that
significantly limits chickpea yield [1]. AB primarily affects the plant’s foliar parts, causing lesions
and tissue necrosis that reduce seed quality and overall crop productivity. The disease often begins
in small patches within the field but can rapidly spread under favorable conditions of temperature
and rainfall [2-3]. Weather plays a crucial role in AB development, particularly in cooler (15-

14
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

25°C) and humid environments (>70%) during the growing season. Additionally, factors such as
inoculum type, virulence, concentration, and the plant's growth stage and resistance level influence
the severity and spread of the disease [3].

Due to the polycyclic nature of AB, control often requires multiple fungicide applications, which
are costly and pose risks to human health, wildlife, and ecosystems [4, 5]. Additionally, the overuse
of fungicides may lead to contamination and the development of resistant pathogens.
Consequently, the sustainable method for managing Ascochyta blight is based on breeding to
select resistant cultivars. However, the traditional methods of phenotyping disease resistance,
relying on human expertise, are often time and labor consuming, not cost-effective, and sometimes
requires destructive sampling of plants. In this context, the use of high-throughput phenotyping
(HTP) methods for Ascochyta blight is promising for developing precise and rapid disease
assessment digital tool. The HTP based on using digital imaging, such as drones mounted thermal,
multispectral (MSI), or hyperspectral (HIS) cameras, offers a non-invasive and consistent imaging
process to monitor plant stresses and disease severity. Use of drone technologies, capturing high-
resolution spectral data showed great opportunities to detect both biotic and abiotic stresses in
different crops [6,7]. Different Indices like NDVI (Normalized Difference Vegetation Index) and
GNDVI (Green Normalized Difference Vegetation Index), are commonly used to assess disease
severity among different plant stresses due plant hydric state or plant health state.

Several authors [8-10] have been used digital methods for evaluating plant disease severity to
provide greater accuracy, repeatability, and reproducibility compared to traditional techniques.
This digital process involves image acquisition, analysis, processing, and validation through
specialized software [8-10]. In fact, over the past three decades (1990-2020), significant
advancements have been made in using digital tools for evaluating plants diseases severity. In the
1990s, cameras were first used to distinguish between healthy and diseased plants, such as in
studies on Fusarium in corn [11] and maize streak virus (MSV) in resistant corn [12]. The 2000s
saw the development of image analysis software like Assess [13] and ImageJ [14, 15], which
improved the precision of disease quantification. By the 2010s, advanced imaging techniques,
including thermal, hyperspectral (HIS), and multispectral (MSI) imaging, became widely used,
offering early disease detection and more effective management compared to traditional visible
spectrum imaging [16,17]. These imaging technologies, often mounted on drones, detect plant
stress or disease by capturing temperature variations and multispectral data [18-19]. MSI cameras
calculate spectral indices such as NDVI, which have been shown to strongly correlate with disease
severity and plant health [20,21]. For instance, NDVI exhibited a strong negative correlation with
disease severity in pineapple (-0.83 to -0.88) [22], and in chickpea, it correlated with leaf area
index, chlorophyll content, and biomass [23]. Additionally, the correlation between visual disease
ratings and NDVI in chickpea increased from -0.61 to -0.66 after 58 days, with NDVI’s correlation
with yield ranging from 0.76 to 0.92 [24].

According to the short review stated above, an early and accurate disease detection remain
essential for implementing timely management strategies. Furthermore, the digital methods require
more improvements as it is often difficult to discriminate between biotic and abiotic stresses that
may cause similar symptoms, making visual diagnosis challenging [25]. In fact, the use of these
digital indices cannot differentiate between biotic and abiotic stresses without efforts from
agronomic experts of relying on the indices data information to the main occurring stress and

15
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

avoiding spatial and temporal interference between two different stresses that can be potentially
expressed in the same digital data taken from one image process acquisition [26, 27].
This study explores the use of digital imaging for reliable phenotyping of Ascochyta blight
resistance in chickpea. Specifically, our innovative digital method aims for testing the correlation
between disease severity and NDVI and boosting this correlation trough use of different
plants/genotypes as checks for showing a gradient of resistance to AB severity and using it as
reference model to predict the disease severity among a large sample of plants/genotypes that can
be potentially tested for selection with reference to AB severity using NDVI information. This
innovative digital method aims to developing a precise, automated phenotyping process for an
effective disease management.

MATERIAL AND METHODS

Evaluation of AB disease severity using classic method of visual scale

A field trial was conducted in 2020-2021 at the Sidi El Aidi experimental station of the INRA
Settat. A total of 216 chickpea genotypes with varied resistance to AB was tested, using a
randomized complete block design (RCBD) augmented with nine blocks and four checks. Spores
of A. rabiei was inoculated via foliar spraying, during vegetative stage. Disease severity was
evaluated visually using a 0-9 scale [28]. The visual reading of AB disease severity was taken for
comparison with digital method using drone multispectral imaging to promote as rapid evaluation
of crops health by the plant pathologist.

Evaluation of AB disease severity using digital imaging and NDVI

The disease severity was evaluated also digitally using drone multispectral imaging. NDVI values
were computed from multispectral data to assess plant health. The NDVI values were used to find
potential correlation with the visual reading.

Reference model for calibration and prediction of disease severity

To test the response of the 216 chickpea genotypes to AB disease severity, four checks of chickpea
genotypes were used as references to show a gradient of different responses from low to high
resistance to AB disease severity. A tuning curve is generated from a gradient of response (4
checks, 9 levels of disease severity) to serve as a reference model for testing the responses of 216
genotypes. This model is then used to assess the potential for predicting AB disease severity among
the tested genotypes.

RESULTS

Evaluation of AB disease severity using visual assessment and NDVI

Use of decision-making tools is essential for phytopathologists to effectively manage disease
interventions by utilizing digital NDVI information to evaluate disease severity. To develop this
tool, we assessed the correlation between Ascochyta blight (AB) disease severity and NDVI based
on data from 216 plots of tested genotypes and 36 plots of reference genotypes. Where C1
represents the resistant check, C2 represents the susceptible check, and C3 and C4 represent the
moderately resistant checks (Fig. 1)

16
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Since AB is characterized by color changes, resulting from lesions on plant leaves, the use of
digital imaging data has significantly distinguished the infected plants. The NDVI results of four
checks clearly illustrate the varying responses of chickpea genotype plots, showing a gradual
transition in leaf color from vibrant green (resistant check C1) to yellow, orange (moderately
resistant C3 and C4), and ultimately red (susceptible check C2). This decrease in green coloration
correlates with the severity of AB, highlighting the relative foliar changes in the infected plants
(Fig. 2).

Figure 1. NDVI values relatives to 4 checks having different reaction to AB, resistant, moderate
resistant or susceptible (36 plots relative to four checks with nine repetitions).

Figure 2. NDVI image illustrating 216 genotype plots with visual notation of 4 checks resistance
to AB disease (S: susceptible, MR: Moderately Resistant and R: Resistant).

17
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Reference model for calibration and prediction of disease severity

The correlation between NDVI and visual reading using data from the four checks used as
reference genotypes facilitated the creation of an empirical model that can be used to predict AB
disease severity using NDVI as an indicator (Eq. 1). The model equation is mounted as follows:

Predicted Severity = -0.9816 * NDVI + 9.983 (1)

The results showed existence of a strong correlation (R² = 0.98) between the actual visual rating
scores and the predicted scores.

Prediction test of severity using the reference model

The severity of AB disease among the 216 genotype is predicted using the reference model to show
how it is possible to find fitting between the predicted severity and the actual one.

Among the 216 genotypes, the correlation between measured severity and predicted severity
showed a good fit using RMSE, MAE, and RE metrics (RMSE = 0.27, MAE = 0.15, RE = ±0.04),
indicating its potential for practical application in assessing AB disease severity in crop fields
based solely on NDVI data.

DISCUSSION

In this research, drone multispectral imaging was utilized to evaluate the severity of Ascochyta
blight (AB) in 252 chickpea plots, comprising 216 genotypes and four check varieties. The NDVI
(Normalized Difference Vegetation Index) was calculated to correlate with visual disease ratings,
enabling field-scale assessment of AB severity. High-resolution orthophoto images revealed
distinct differences between heavily infested and healthy plots, with consistent discoloration linked
to increased plant mortality.

The study found a strong correlation between NDVI and AB severity, aligning with previous
research indicating NDVI as a robust index for disease quantification. The relationship between
NDVI and visual ratings showed an impressive R² value of 0.98. An empirical model was
developed through linear regression, successfully predicting disease severity, validated by RMSE,
MAE, and RE metrics.

The potential of NDVI as a decision-making tool for disease management was emphasized,
facilitating timely interventions based on environmental conditions. Other studies highlighted the
use of machine learning and decision support systems in disease detection and management,
demonstrating significant accuracy in monitoring various crop diseases.

However, the accuracy of disease monitoring may be influenced by factors such as plant
senescence, canopy density, and environmental conditions. Our study established a significant
correlation between NDVI (Normalized Difference Vegetation Index) and the disease severity of
four control genotypes with known reactions to Ascochyta blight. This finding highlights the
potential of NDVI as a reliable tool for detecting biotic stress in crops. The use of NDVI for
monitoring plant health has been widely reported as an effective indicator for assessing vegetation

18
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

vigor and stress, particularly in response to pathogens [25]. By providing non-destructive, real-
time monitoring capabilities, NDVI can serve as an early-warning system for managing disease
outbreaks and guiding targeted interventions in precision agriculture [25]. Our results further
validate the growing body of research that supports NDVI as a promising tool for assessing biotic
stress in crops, allowing for efficient and sustainable crop management. In addition, the study
underscored the need for calibrated images to increase accurate analysis and suggested further
research to enhance the differentiation of disease symptoms using hyperspectral and multispectral
sensing techniques.

CONCLUSION

This study showed that it is possible to adequately use NDVI derived from multispectral images
and improve its fitting to effectively detect and assess the severity of Ascochyta blight on
chickpeas. A strong correlation between NDVI and disease ratings allowed for the creation of an
accurate predictive model. The prediction is greatly improved as the model calibration is
referenced to a gradient of disease severity using four genotypes as checks to show a gradual
response of disease severity.

The reference curve of the checks responses to AB disease showed that it is possible to implement
robust predictive model for monitoring disease severity. In fact, the referencing of NDVI
information to disease severity of known genotypes improved the model fitting. The digital
monitoring of chickpea green cover can be greatly improved if the fitting of NDVI response to
disease severity is calibrated with reference to use of checks gradient to assess AB disease severity.
This innovative method based on calibration can potentially help the plant pathologists to
overcome the problem of discriminating between NDVI responses to biotic and/or abiotic stresses
by using specifically NDVI information to assess and control AB disease severity. The results
highlight NDVI's potential for field-scale disease monitoring and high-throughput phenotyping,
with future integration of deep learning offering further advancements in disease management.

REFERENCES

1. Pande S, Kishore GK, Upadhyaya HD, Rao JN. Identification of sources of multiple disease
resistance in mini-core collection of chickpea. Plant Dis. 2006;90(9):1214-8. doi:
10.1094/PD-90-1214. PMID: 30781104.
2. Pande S, Sharma M, Gaur PM, et al. Development of screening techniques and identification
of new sources of resistance to Ascochyta blight disease of chickpea. Australas Plant
Pathol. 2011;40(2):149-56. doi: 10.1007/s13313-010-0024-8.
3. Pande S, Siddique K, Kishore G, Bayaa B, Gaur P, Laxmipathi Gowda C, et al. Ascochyta
blight of chickpea (Cicer arietinum L.): A review of biology, pathogenicity, and disease
management. Crop Pasture Sci. 2005;56(4):317-32. doi: 10.1071/AR04143.
4. Chang KF, Ahmed HU, Hwang SF, Gossen BD, Strelkov SE, Blade SF, et al. Sensitivity of
field populations of Ascochyta rabiei to chlorothalonil, mancozeb, and pyraclostrobin
fungicides, and effects of strobilurin fungicides on the progress of
5. Ascochyta blight of chickpea. Can J Plant Sci. 2007;87:937-44.

19
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

6. Gan YT, Siddique KHM, MacLeod WJ, Jayakumar P. Management options for minimizing
the damage by ascochyta blight (Ascochyta rabiei) in chickpea (Cicer arietinum L.). Field
Crops Res. 2006;97:121-34.
7. Zhao Y, Gu Y, Qin F, Li X, Ma Z, Zhao L, et al. Application of near-infrared spectroscopy
to quantitatively determine relative content of Puccinia striiformis f. sp. tritici DNA in
wheat leaves during the incubation period. J Spectrosc. 2017;2017:9740295. doi:
10.1155/2017/9740295.
8. Mahlein AK. Plant disease detection by imaging sensors—parallels and specific demands
for precision agriculture and plant phenotyping. Plant Dis. 2016;100:241-51.
9. Kwack MS, Kim EN, Lee H, Kim JW, Chun SC, Kim KD. Digital image analysis to measure
lesion area of cucumber anthracnose by Colletotrichum orbiculare. J Gen Plant Pathol.
2005;71:418-21.
10. Clément A, Verfaille T, Lormel C, Jaloux B. A new colour vision system to quantify
automatically foliar discoloration caused by insect pests feeding on leaf cells. Biosyst
Eng. 2015;133:128-40.
11. El Jarroudi M, Kouadio AL, Mackels C, Tychon B, Delfosse P, Bock CH. A comparison
between visual estimates and image analysis measurements to determine Septoria leaf
blotch severity in winter wheat. Plant Pathol. 2015;64:355-64.
12. Todd LA, Kommedahl T. Image analysis and visual estimates for evaluating disease
reactions of corn to Fusarium stalk rot. Plant Dis. 1994;78:876-8.
13. Martin DP, Willment JA, Rybicki EP. Evaluation of maize streak virus pathogenicity in
differentially resistant Zea mays genotypes. Phytopathology. 1999;89:695-700.
14. Sun H, Wei J, Zhang J, Yang W. A comparison of disease severity measurements using
image analysis and visual estimates using a category scale for genetic analysis of
resistance to bacterial spot in tomato. Eur J Plant Pathol. 2014;139:125-36.
15. Stewart EL, McDonald BA. Measuring quantitative virulence in the wheat pathogen
Zymoseptoria tritici using high-throughput automated image analysis. Phytopathology.
2014;104:985-92.
16. Laflamme B, Middleton M, Lo T, Desveaux D, Guttman DS. Image-based quantification of
plant immunity and disease. Mol Plant Microbe Interact. 2016;29:919-24.
17. Rumpf T, Mahlein AK, Steiner U, Oerke EC, Dehne HW, Plümer L. Early detection and
classification of plant diseases with support vector machines based on hyperspectral
reflectance. Comput Electron Agric. 2010;74:91-9.
18. Zhang D, Zhou X, Zhang J, Lan Y, Xu C, Liang D. Detection of rice sheath blight using an
unmanned aerial system with high-resolution color and multispectral imaging. PLoS One.
2018;13
19. Gao D, Sun Q, Hu B, Zhang S. A framework for agricultural pest and disease monitoring
based on internet-of-things and unmanned aerial vehicles. Sensors. 2020;20:148. doi:
10.3390/s20051487.
20. Bauriegel E, Herppich WB. Hyperspectral and chlorophyll fluorescence imaging for early
detection of plant diseases, with special reference to Fusarium spec. infections on wheat.
Agriculture. 2014;4:32-57.
21. Guo J, Hu Y, Xiong Z, Yan X, Ren B, Bu R. Spatiotemporal variations of growing-season
NDVI associated with climate change in northeastern China’s permafrost zone. Polish J
Environ Stud. 2017;26:1521-9.

20
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

22. Pan S, Zhao X, Yue Y. Spatiotemporal changes of NDVI and correlation with meteorological
factors in Northern China from 1985–2015. E3S Web Conf. 2019.
23. Wang M, Fu J, Wu Z, Pang Z. Spatiotemporal variation of NDVI in the vegetation growing
season in the source region of the Yellow River, China. ISPRS Int J Geo Inf. 2020;9:282.
doi: 10.3390/ijgi9040282.
24. Ajai MR, Kamat DS, Chaturvedi GS, Singh AK, Sinha SK. Spectral assessment of leaf area
index, chlorophyll content, and biomass of chickpea. Photogramm Eng Remote Sens.
1983;49
25. Zhang C, Chen W, Sankaran S. High-throughput field phenotyping of Ascochyta blight
disease severity in chickpea. Crop Prot. 2019;125:104885.
26. Prabhakar M, Prasad YG, Rao MN. Remote sensing of biotic stress in crop plants and its
applications for pest management. In: Venkateswarlu B, Shanker A, Shanker C,
Maheswari M, editors. Crop Stress and its Management: Perspectives and Strategies.
Dordrecht: Springer; 2012. p. 449-62.
27. Rayhana R, Ma Z, Liu Z, Xiao G, Ruan Y, Sangha JS. A review on plant disease detection
using hyperspectral imaging. IEEE Trans AgriFood Electron. 2023;1(2):108-34.
28. Chowhan P, Chakraborty AP. Remote sensing technology—A new dimension in detection,
quantification, and tracking of abiotic and biotic stresses. In: Roy S, Mathur P,
Chakraborty AP, Saha SP, editors. Plant Stress: Challenges and Management in the New
Decade. Cham: Springer; 2022. p. 549-66.
29. Chen W, Coyne CJ, Peever TL, Muehlbauer FJ. Characterization of chickpea differentials
for pathogenicity assay of ascochyta blight and identification of chickpea accessions
resistant to Didymella rabiei. Plant Pathol. 2004;53:759-69.

21
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

EFFECTS OF ADOPTING FERTILIZER MANAGEMENT PRACTICES ON YIELD IN

MAIZE-BASED SYSTEMS IN EMBU COUNTY, KENYA: AN INSTRUMENTAL
VARIABLE REGRESSION APPROACH
#11303

1*Malcolm Gitau, 1Rose Nyikal, 1Evans Chimoita, 2Onesmus Kitonyo

1Department of Agricultural Economics, University of Nairobi, Kenya
2Department of Plant Science and Crop Protection, University of Nairobi, Kenya
*
e-mail: [email protected]; +254705534199

ABSTRACT

Global food security is increasingly becoming dependent on judicious fertilizer use. However,
inefficient use by farmers has hindered yield potential and caused environmental pollution. The
principles of “4R Nutrient Stewardship” promote best fertilizer management practices for
enhanced economic, social and environmental outcomes. Despite widespread promotion of various
fertilizer management practices under the 4R- framework, the empirical evaluation of their effects
on yield remains limited. This study therefore employed an Instrumental Variable regression
model to evaluate the nuanced effects of fertilizer management practices on yield. Results of the
IV regression revealed that adoption of manure plus inorganic fertilizer, split application, soil
testing and precision fertilization positively influence maize yield. Furthermore, practices such as
intercropping, soil moisture conservation, crop rotation, and agroforestry positively influence
yield, whereas minimum tillage has a negative effect. Efforts should be directed towards
supporting and expediting the adoption of fertilizer management practices under the 4R- Nutrient
Stewardship to increase maize yields among smallholder farmers.

Keywords: Maize, Fertilizer Management Practices, 4Rs, IV regression, Effects

INTRODUCTION

Food insecurity remains a pressing global issue, disproportionately affecting Africa (FAO et al.,
2023). In Kenya, it was projected that 4.4 million individuals are at a risk of facing acute food
insecurity (IPC, 2022). In a world faced by pervasive hunger and malnutrition, global production
of the world`s major grains like maize must double by 2050 in order to feed the growing
population (Tian et al., 2021). Maize (Zea mays L.), is a valuable cereal crops in Sub-Saharan
Africa contributing significantly to dietary needs and supporting millions of smallholder farmers
(FAO, 2021). Despite its indispensable role, maize productivity has not increased in a
proportionate manner and significant gaps in yields are still evident.

Fertilizers play an important role in maize production and have been acknowledged for their
potential in boosting yields by a substantial margin of between 40 to 60 percent (FAO, 2015).
Nevertheless, escalating fertilizer prices have led to suboptimal utilization (Obour et al., 2015).
Excessive application, on the other hand, raises environmental concerns (Sapkota et al., 2014).
Additionally, farmers lack knowledge on fertilizer use and often apply excess or insufficient
fertilizers (Aryal et al., 2018 ; Kishore et al., 2021). Recognizing the above limitations in fertilizer
use, the Four Rights (4Rs) (right rate, right source, right time, right place) of fertilizer application

22
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

were formulated through cross-sector collaborative efforts by the International Fertilizer Industry
Association (IFA) and the International Plant Nutrition Institute (IPNI) as guidelines for optimal
management of fertilizers worldwide (IPNI, 2014). Best fertilizer management practices under this
study including soil testing, split application, combining manure with inorganic fertilizers,
precision fertilization, and concurrent application of fertilizers and seeds during wet planting are
inherently embedded within the 4R Nutrient Stewardship framework. They have been promoted
for adoption in different regions. However, empirical evidence showing effects of adoption on
maize yield among small-scale maize farmers remains sparse. This study therefore sought to assess
effects of fertilizer management practices under the 4Rs on maize yield. Addressing this gap is
important to get detailed insights into the best fertilizer types, applications rates, timings, and
placement methods for maximum maize yield.

METHODOLOGY

Study area, data sources and sampling procedure

This study was conducted in Embu County, in upper Eastern Kenya. Embu is one of the strategic
areas of African Plant Nutrition Institute owing to the strong interaction between water and
nitrogen as key factors influencing yield in dryland maize-based crop systems. A household survey
was conducted, and data was collected using structured questionnaires anchored on an android
aided platform (Survey to go). A multistage stratified sampling technique was employed in
selecting sample sub-locations and households. The study excluded higher wet zones, and dry
lower zones where maize is not majorly grown. The agro-ecological zones were stratified into
three transects (1, 2, and 3) encompassing the sub-locations.

There are about 5200 farmers within these transects (list provided by the Sub- County agriculture
officers). This created a definite sampling frame and therefore the study adopted the Taro Yamane
𝑁
(1967) formula below. n=1+𝑁𝑒 2 where: n = sample size, N = total population, and e = margin of
5200
error. The sample size was computed with a confidence level of 95%. n= 1+5200∗(0.05)2 = 371
maize farmers. Data cleaning was done thereafter to identify and remove duplicates and address
outliers arriving at a final data set of 365 farmers.

23
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

TAII
UH1

UHO

LHO
LH1

UM1
UM2
LEGEND UM3
Study area UM4 LM5
LM3
IL5
UM3/UM4
LM3
LM4
Embu County LM4
LM5

LM5
LM4

Transect 3 #
## # #
# ## #
NYAGARI # ## # # ## # # #
Nyagari sublocat ion

## # # # #
# # ### #
## #
#
Transect 2 # # #
# # #
KARURUMO #
# # #
# #
#
Transect 1 ##
## #
#
#
## #
# #
#
#
#
#
# ## # # #

#
# # #
#
KAWANJARA
# # ##
#
# ## #
#
#
#
#
#

# # # #

Kithimu
#
#
#

#
#
# #
#
# #
#
# #
# #
# #
# #

#
# #

#
KARIRU
#

#
#
#
#
#
# # # #
# # #
#
# #
# # # #
# # # # #
# #
# # # # # # # #
# # # # #
# # # # # #
# # # # # # # # # #
KIAMURINGA
#
# #
# #
#
#
#
#
#
# #
# #
# # #
#
# # #
# #
#
#
#

LEGEND #
# # #
#
#
#
#
# #
#
# # GITIBURI
# # # #
#
# # # #
#
#
#
# # #
# #
# # # # # # #
# # # # # # # #
# # # #
# # #
# Selected households (371) #
#
#
#
#
#
#
#
#
#
# # #

# # # #
Selected Sub-locations (9) #
# # #
#
#
#
#
# #
# # ##
#
#
#
# #
# # # # #
## # # # #
# #
# # # # # # # #

AEZ GACHOKA
# ## #
# #
#
#
# #
#
# # # #
# # # # GIKIIRO
#
#
#
#
# # # # # # # # # # # #
UM3/UM4 #
# #
#
# #
# # #
#
#
# # #

#
LM3 # #
5 0 5 10 Kilometers
LM4

Figure 1. Map showing the study area. Source: Author`s compilation

RESULTS AND DISCUSSION

Effects of adopting fertilizer management practices on maize yield- IV regression

Results of the Instrumental Variable (IV) regression are presented in table 1 revealing the effects
of four fertilizer management practices on maize yield. Adoption of manure plus inorganic
fertilizer was found to increase maize yield by 306 kilograms/acre. This result highlights the
importance of an integrated nutrient management approach, where organic nutrient sources such
as manure are combined with inorganic fertilizers. It was further revealed that, compared to non-
adopters, farmers adopting precision fertilization experienced significant increases in maize yield,
of 294 kilograms per acre. Precision fertilization through side banding application involves placing
fertilizer near the root zones ensuring efficient nutrient utilization. Adoption of this practice
enhances efficient and targeted nutrient application, ensuring site-specific nutrient supply to maize.
The IV regression analysis further revealed that adoption of soil testing had a positive and
significant effect on maize yield, increasing yield by 1037 kilograms/ acre. Farmers leveraging

24
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

soil testing can align their fertilizer applications precisely to meet specific soil deficiencies and
requirements.

Table 1. Results of IV regression on effects of fertilizer management practices on yield

Variable Manure + fertilizer Precision fertilization Soil testing Split application

Coefficient (Std error)

Land tenure individual -6.4029 (63.5392) -43.3200 (60.4266) -45.1778 (57.9350) -42.9425 (60.7676)
Experience -2.418957 (2.9040) -2.9043 (2.8824) -2.4473 (2.7452) -2.9685 (2.9029)
Intercropping 600.2347 (92.3679) *** 612.856 (91.1396*** 532.7773 (105.9075) *** 583.8539 (94.3519) ***
Minimum tillage -222.1926 (82.6809) *** -204.9583 (80.7194) ** -68.8440 (108.8409) -214.660 9 (82.3724) ***
Monoculture 470.1703 (101.7040) *** 502.0999 (97.6594*** 453.0867 (105.8808) *** 461.2402 (105.4477) ***
Moisture conservation -85.85303 (72.4340) -96.97673 (71.4337) 153.1873 (77.2872) ** -100.9195 (71.8379)
Conventional farming -81.17338 (81.0702) -57.7537 (79.0499) 46.8493 (100.5287) -64.7622 (79.8595)
Crop rotation 691.5752 (83.3769) *** 677.5726 (87.9084) *** 641.7493 (102.1253) *** 670.0214 (90.6610) ***
Agroforestry 439.9163 (69.1065) *** 439.9945 (68.3584) *** 417.3008 (67.9312) *** 443.0685 (68.7047) ***
Farm demonstrations Instrument ** Instrument 1** Instrument 1** Instrument 1**
Soil fertility training 17.0645(78.3552) Instrument 2** Instrument 2** Instrument 2**
Farm experimentation -4.4157(56.5874) -8.5776(56.7589) -31.5137 (60.4823) -23.43488(59.6578)
Effect on maize yield 305.564 (168.339) * 294.2713 (177.0331) * 1037.811 (706.9282) ** 295.1308 (179.2012) *

No. of observations 365 365 365 365

Wald χ2 χ2 (18) 161.33 χ2 (18) = 162.67 χ2 (18) = 171.73 χ2 (18) = 171.73
Prob > χ2 0.0000 0.0000 0.0000 0.0000

R-squared 0.2919 0.31 0.36 0.29

The practice of split application demonstrated a positive effect on maize yield, increasing yields
by 295 kilograms per acre. Nutrient distribution at different growth stages maximizes nutrient
availability when plants require them most, leading to vigorous growth and increased yield. This
study further revealed that, practices such as intercropping, soil moisture conservation, crop
rotation, and agroforestry positively influenced yield, whereas minimum tillage had a negative
effect.

CONCLUSION AND RECOMMENDATIONS

This study assessed effects of adoption of fertilizer management practices on maize yield. The
study used the Instrumental Variable regression model to assess effects of adoption. Average
maize yield stood at 1552.05 kg/ha, which is significantly lower than the global average and
potential yield. Adoption of practices like manure plus inorganic fertilizer, split application,
precision fertilization (side banding) and soil testing positively influence maize yield. Other
practices including intercropping, soil moisture conservation, crop rotation, and agroforestry also
positively influence yield, while minimum tillage was associated with reduced yields. The findings
from this research collectively highlight the important role of different fertilizer management
practice under the 4R-Nutrient Stewardship in addressing the existing yield gaps and enhancing
maize productivity in Embu County, addressing food insecurity concerns. Efforts should be
directed towards supporting and expediating the adoption fertilizer management practices to
increase maize yields among smallholder farmers in different regions.

25
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REFERENCES

Aryal, J. P., Rahut, D. B., Maharjan, S., & Erenstein, O. (2018). Factors affecting the adoption of
multiple climate-smart agricultural practices in the Indo-Gangetic Plains of India: Jeetendra
Prakash Aryal, Dil Bahadur Rahut, Sofina Maharjan and Olaf Erenstein / Natural Resources
Forum. Natural Resources Forum, 42(3), 141–158. https://doi.org/10.1111/1477-
8947.12152
FAO. (2015). FAO statistical pocketbook 2015: World food and agriculture. Food and Agriculture
Organization of the United Nations. http://www.fao.org/3/a-i4691e.pdf
FAO. (2021). FAOSTAT. Food and Agriculture Organization of the United Nations, Rome, Italy.
https://www.fao.org/faostat/
FAO, IFAD, UNICEF, WFP, & WHO. (2023). STATE OF FOOD SECURITY AND
NUTRITION IN THE WORLD 2023: Urbanization, agrifood system transformation. FOOD
& AGRICULTURE ORG. https://doi.org/10.4060/cc3017en
IPC. (2022). KENYA : IPC Food Security & Nutrition Snapshot. https://www.ipcinfo.org/ipc-
country-analysis/details-map/en/c/1155689/
IPNI. (2014, November 24). 4R Extension Handbook for Smallholder Farming Systems in Sub-
Saharan Africa. http://ssa.ipni.net/article/AFR-3075
Kishore, A., Alvi, M., & Krupnik, T. J. (2021). Development of balanced nutrient management
innovations in South Asia: Perspectives from Bangladesh, India, Nepal, and Sri Lanka.
Global Food Security, 28, 100464. https://doi.org/10.1016/j.gfs.2020.100464
Obour, P. B., Dadzie, F. A., Kristensen, H. L., Rubæk, G. H., Kjeldsen, C., & Saba, C. K. S. (2015).
Assessment of farmers’ knowledge on fertilizer usage for peri-urban vegetable production
in the Sunyani Municipality, Ghana. Resources, Conservation and Recycling, 103, 77–84.
https://doi.org/10.1016/j.resconrec.2015.07.018
Sapkota, T. B., Majumdar, K., Jat, M. L., Kumar, A., Bishnoi, D. K., McDonald, A. J., &
Pampolino, M. (2014). Precision nutrient management in conservation agriculture based
wheat production of Northwest India: Profitability, nutrient use efficiency and
environmental footprint. Field Crops Research, 155, 233–244.
https://doi.org/10.1016/j.fcr.2013.09.001
Tian, Z., Wang, J., Li, J., & Han, B. (2021). Designing future crops: Challenges and strategies for
sustainable agriculture. The Plant Journal, 105(5), 1165–1178.
https://doi.org/10.1111/tpj.15107

26
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

GRAIN YIELD AND NITROGEN UPTAKE OF MAIZE AS AFFECTED BY SOIL

MANAGEMENT PRACTICES AND THEIR INTERACTION ON CAMBISOLS AND
CHERNOZEM
#11250

1Ashenafi
Nigussie
1Department of Soil and Water Management, Wondo Genet Agriculture Research Center,
Shashemene, Ethiopia.
*
e-mail: [email protected]; +251 09063256

ABSTRACT

Although numerous factors contribute to wide yield gaps, low external inputs, particularly N, and
poor cropping practices such as soil tillage and monocropping are among the major factors
affecting low maize production. In view of this, field experiments were implemented on two sites
with Cambisols and Chernozem soil types in two consecutive years to evaluate the impacts of
different soil management practices on the grain yield and quality, nitrogen uptake and selected
soil properties. A three-factor experiment was arranged as a split-split plot arrangement
randomized complete block design with three replications. The minimum tillage (MT) and
conventional tillage (CT) were used as the main plot, haricot bean-maize rotation, and maize
monocropping as a subplot, and four levels of nitrogen fertilization (Control, 20 t ha-1 compost, 46
kg N ha-1 + 10 t ha-1 compost, and 92 kg N ha-1) as the sub-sub plot. Analysis of variance showed
that soil management practices were significantly affecting grain yield, N-uptake and soil
properties. In sites, the conventional tillage and rotation system increased the grain yield, and N-
uptake in contrast to the minimum and monocropping, respectively. Similarly, nitrogen evidently
affected the grain yield, N- uptake, and selected soil properties. However, tillage methods differed
in their effects on soil chemical properties; soil organic carbon and total nitrogen
concentrations were improved through MT compared to CT. Grain yield was significantly
associated with NDVI, grain N-content and N-uptake. Therefore, a CT plus haricot bean-maize
rotation system with the addition of solely 92 kg N ha-1 and integrated 46 kg N ha-1 + 10 t compost
ha-1 could be recommended for Hawassa Zuria (Cambisols) and Meskan (Chernozem) districts,
respectively. However, to ensure sustainable maize production in the investigated areas, an
integrated N-treatment with MT and a rotation system may be recommended, which could improve
soil properties.

Keywords: maize, tillage, cropping systems, fertilization, grain yield and quality, nitrogen uptake

INTRODUCTION

Maize or corn (Zea mays L.) is one of the world's leading cereals, ranking second in production
after wheat (FAO, 2019). Ethiopia is the seventh maize-producing country in Africa. It is the
second in area coverage next to tef (Eragrostis tef (Zucc.), with total land area of 10,478,217 ha
being under cereals, of which maize covered about 17.68% (2,274,305.93 ha) (CSA, 2019).
Despite the large area under maize production, its current national average yield is about 4.2 t ha -
1 (CSA, 2019), which is far below the world's average yield of 5.8 t ha-1 (FAO, 2019). Although

numerous factors contribute to wide yield gaps, low external inputs, particularly N, poor soil

27
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

fertility, reduced water-holding capacity of the soil, and poor soil infiltration problems are among
the major factors paid for low maize productivity (Chimdi et al., 2012; Mourice et al., 2015;
Teklewold et al., 2013). Moreover, frequent tillage, monocropping, and complete removal of crop
residues are also the governing factors for low productivity (Kassie et al., 2013). However, there
is scarce information about the effects of tillage, cropping systems, nitrogen fertilization and their
interaction on the yield, nitrogen uptake of maize, as well as soil chemical properties. Therefore,
the present study was instigated to evaluate the effects of different soil management practices on
the maize grain yield and quality, nitrogen uptake and selected soil chemical properties in the
central Rift valley of Ethiopia, under two soil types - Cambisols and Chernozem.

MATERIALS AND METHODS

The field experiments were conducted for two consecutive years (2019 and 2020) in Hawassa
Zuria and Meskan districts of the Central rift valley of Ethiopia. The Hawassa Zuria site is
geographically situated at 07° 1' 0.83" N Latitude and 38° 22' 26" E Longitude with an altitude of
1713 m above sea level (asl). The experimental site at Meskan is found at 08° 05' 33" N Latitude
and 38° 26' 75" E Longitude with an altitude of 1841 m asl. The soil types for the field trial were
Cambisols for Hawassa Zuria and Chernozem for Meskan, according to the WRB soil
classification system (IUSS Working Group, 2015).

Two tillage methods (TM) were evaluated: conventional tillage (CT) and minimum tillage (MT).
The two tillage practices were combined with two cropping systems (CS): haricot bean-maize
rotation (RCS) and maize monocropping (MCS). In addition, four levels of nitrogen fertilization
(NF) (0, 20 t compost ha-1, 46 kg N ha-1 + 10 t compost ha-1, and 92 kg N ha-1) were combined
with tillage practices and cropping systems. Treatments were arranged as Split-split plot
arrangement randomized as a RCBD (randomized complete block design), with tillage methods as
the main (whole) plots, cropping systems as sub-plots, and nitrogen fertilization treatments as sub-
sub-plots, with three replicates: making 48 sub-sub-plots for each experimental site.

Yield and yield related data were collected from a net plot area of 4 m2 (1.25 m x 3.2 m) by
rejecting the border rows, from three replications. The harvested grain yield was adjusted to a
12.5% moisture level (Nelson et al., 1985) and it was converted into hectare bases. Twenty grams
of grain samples were taken from each experimental unit. The grains were oven-dried to constant
weight thereafter; and the samples were ground and passed through a 0.5 mm sieve. The nitrogen
content in the grain was analyzed using the Kjeldahl procedure after wet digestion by H2SO4/H2O2
(Nelson and Sommers, 1982).

Before the analysis of variance (ANOVA), the normality of the data was checked using the
Shapiro-Wilk normality test. Despite the two experimental sites were distinctly different in their
soil fertility status, subsequently the statistically analysis was done independently for each
location, using the SAS 9.3 software package (SAS Institute, 2014), considering the experimental
treatment as a fixed factor and replication as a random factor. At a probability level of P≤0.05,
differences between treatments means were separated using the protected Fisher’s least significant
difference (LSD) (Steel and Torrie, 1980). The LSD differences for the main factors and
interaction effects comparisons were calculated using the appropriate standard error terms.

28
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

RESULTS AND DISCUSSION

Effects of tillage methods, Cropping systems and nitrogen fertilization on NDVI, grain
yield, grain N-content and uptake and grain protein content
Tillage had revealed a statistically significant (P < 0.05) effect on maize grain yield At Hawassa
Zuria, but not Meskan site, despite the higher yield which was gained from the CT (3855.5 kg ha-
1) and (7094.9 kg ha-1) for Hawassa Zuria and Meskan, respectively (Table 1). In this study, grain

yield increased by 5.2 and 0.1% in CT over MT at Hawassa Zuria, and Meskan. The positive result
of CT on maize grain yield was possibly due to improved soil physical conditions, root growth,
infiltration of water, nutrient mineralization and suppressing weed growth. Correspondingly,
Simić et al. (2020); Salem et al. (2015); Wang et al. (2015) reported that CT in a short-term study
increased corn grain yield compared to a minimum or zero tillage due to less soil compaction,
which improved soil aeration and organic matter mineralization. In both locations, the N-content
and N-uptake parameters responded positively to CT, possibly due to the stimulation of N-
mineralization from organic matter and thereby improved soil mineral N-availability for crop
uptake. Similarly, Simić et al. (2020) verified the benefit of conventional tillage for better maize
grain yield and enhancement in grain protein content.

At Hawassa Zuria, the cropping system had a significant (P < 0.05) effect on grain yield, grain
nitrogen content, nitrogen uptake, and protein content. However, at Meskan, while grain yield was
affected, the other parameters did not show statistically significant differences (Table 1). The
haricot bean-maize rotation system increased maize grain yield, N-content, N-uptake and protein
content by 1.1, 2.7, 17.8, 21.1 and 17.9% in Hawassa Zuria and 1.3, 0.25, 10, 12.1 and 13.7% in
Meskan, respectively, compared to maize monocropping (Table 1). This was possibly due to the
change in inorganic N-availability in the soil solution caused by previous atmospheric N2 fixation
and legume residue decomposition since legume residues had better quality and a narrow C:N
ratio, which results in rapid release of N from the residues (Adesoji et al., 2015; Lupwayi et al.,
2011; Tolera et al., 2009). Our result is in covenant with Lafond et al. (2006) who stated that
legumes offer a positive contribution to soil TN and thus improved its availability.

Analysis of variance depicted that the grain yield differed significantly (P < 0.001) among N-
treatments in both sites. The highest grain yields of 1180.5 kg ha-1 and 8169.1 kg ha-1 were obtained
from the application of 92 kg N ha-1 and 46 kg N ha-1 + 10 t ha-1 compost at Hawassa Zuria and
Meskan sites, respectively. Similarly, Kaplan et al. (2019) proved that the grain yield increased
with increasing the N level. Like grain yield, N-fertilization had revealed significant effects on
GNC, GNU, and GPC (Table 1). In both locations, the integrated use of inorganic nitrogen and
compost at a rate of 46 kg N ha-1 + 10 t ha-1 remarkably increased GNC, GNU and GPC by 35.1,
61.6 and 35.3% at Hawassa Zuria and 23.2, 68.2 and 21.6% at Meskan, respectively, when
compared to the unfertilized treatment. Our result is in covenant with findings of Dunjana et al.
(2012); Negassa et al. (2005); Rusinamhodzi et al. (2013), who stated that integrated application
of organic and mineral fertilizers at appropriate rates can be an effective approach to improve
maize N uptake.

29
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 1. The effects of tillage methods, cropping systems, and N-fertilization on NDVI, grain
yield, N-content, N-uptake and grain protein content of maize at the two sites

Investigated factors Hawassa Zuria Meskan

GY GNC GNU GPC GY GNC g kg- GNU-1kg GPC %
kg ha-1 g kg-1 kg ha-1 % kg ha-1 1 ha
Tillage methods
MT 3662.6b 9.6 35.6b 6.0 7061.0 12.1 89.0 7.8
CT 3855.5 a
9.9 39.1a 6.2 7091.9 12.5 89.7 7.8
LSD (0.05) 112.6 ns 1.89 ns ns ns ns ns
Cropping systems
RCS 3835.1a 10.6a 10.9a 6.6a 7101.0 13.2a 91.5a 8.3a
MCS 3683.5 b
9.0 b
33.7 b
5.6 b
7083.2 12b 81.3b 7.3b
LSD (0.05) 128.8 0.1 3.8 0.21 ns 0.03 5.6 0.2
Nitrogen fertilization
Control 3273.3d 8.2d 26.8c 5.1d 5987.9d 11.2d 66.9d 6.9d
20 t ha Compost
-1
3628.2 c
9.1 c
31.2 b
5.9 c
6169.2 c
11.8c 75.8c 7.1c
46 kg N ha-1 +10 t ha-1 3951.1b 11.1a 11.1a 6.9a 8169.1a 13.8a 112.5a 8.6a
Compost
92 kg N ha-1 1180.5a 10.5b 11.1a 6.6b 7711.8b 13b 102.1b 8.3b
LSD (0.05) 113.7 0.6 2.62 0.35 310.1 0.01 1.2 0.25
Values of a parameter means followed by the same letter did not differ significantly across the tillage methods,
cropping systems and N-fertilization at P ≤ 0.05 according to LSD test.

Effects of tillage, cropping systems and nitrogen fertilization on soil organic carbon and
total nitrogen
There were no significant changes in organic carbon concentrations across tillage methods and
cropping systems in either location (Table 2). This could be because the samples were gathered
two years after the field trial, which is a short time to oversee the effect of tillage on soil OC. A
similar observation was reported by Geisseler and Horwath (2009). Conversely, organic carbon
content was significantly affected by N fertilization (Table 2). The addition of 20 t ha-1 compost
provided the higher OC at Hawassa Zuria, which was statistically comparable with the integrated
N-treatment. The increase in soil OC after the application of compost is due to the composting
material and the rich microbial community, which contributes to the formation of soil organic
carbon (Deepak et al., 2017; Dhillon et al., 2018; Lorenz and Lal., 2016).

Tillage practices had a significant effect on soil total N in both locations, with minimum tillage
contributing more to total N than the conventional tillage (Table 2). This could be owing to
enhanced N protection inside micro and macro aggregates, resulting in lower N losses due to
leaching and organic matter decomposition (Wyngaard et al., 2012). Likewise, the cropping
systems was significantly affected the soil total N at Meskan, but remarkable variation not
observed at Hawassa Zuria. However, in both sites there was a tendency for better soil total N in
the RCS compared to MCS (Table 2). The findings of this study agree with those of Kirkegaard et
al. (2008) and Lupwayi et al. (2011). Total nitrogen content was significantly affected by nitrogen
fertilization (Table 2). The integrated N treatment had the highest TN (0.26% and 0.39% for
Hawassa Zuria and Meskan, respectively), indicating that more N was released through
mineralization of the compost added to the soil and due to the existence of high levels of respective
total N in the compost. Our findings are in line with those of Ashenafi et al. (2021) and Yan et al.

30
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

(2007), who found that inorganic nitrogen influences most soil biological processes by promoting
microbial carbon use, which is critical for mineralization and nutrient transformation activities.

Table 2. Main effects of tillage, cropping systems and nitrogen fertilization on soil reaction,
organic carbon and total nitrogen contents of the surface layer of soils (0-20 cm)

Treatments Hawassa Zuria (Cambisols) Meskan (Chernozem)

pH OC (%) TN (%) C: N pH OC (%) TN (%) C: N
Tillage methods
MT 6.1b 2.61 0.25a 10.52b 6.8 1.00 0.36a 11.31b
CT 6.2 a
2.59 0.23 b
11.18 a
6.8 3.98 0.33b 12.11a
LSD (0.05) 0.06 ns 0.01 0.33 ns ns 0.02 0.83
Cropping systems
RCS 6.1 2.61 0.25 10.72 6.8 3.99 0.35a 11.18b
MCS 6.2 2.60 0.21 10.97 6.8 3.99 0.31 b
11.97a
LSD (0.05) ns ns ns ns ns ns 0.01 0.37
Nitrogen fertilization
Control 6.0b 2.51c 0.21c 11.81a 6.6d 3.83d 0.32c 11.91b
20 t ha Compost
-1
6.3 a
2.68 a
0.21 b
10.73 b
7.0 a
1.11 b
0.32 c
13.02a
16 kg N ha +10 t ha Compost 6.2
-1 -1 a
2.66 a
0.26 a
10.19 c
6.9 b
1.11 a
0.39 a
10.57d
92 kg N ha 1 6.1b 2.58b 0.25ab 10.63bc 6.7c 3.88c 0.31b 11.37c
LSD (0.05) 0.1 0.01 0.01 0.17 0.08 0.03 0.01 0.53
Within the columns, means followed by the same letters are not significantly different at P < 0.05 according to LSD
test.

CONCLUSION

Soil management practices play a crucial role in influencing grain yield, nitrogen content, nitrogen
uptake, and grain protein content, as well as certain soil chemical properties. At both sites, the
conventional tillage and crop rotation system resulted in higher grain yield, nitrogen content,
uptake, and protein content compared to minimum tillage and monocropping systems.
Additionally, nitrogen fertilization had a significant impact on grain yield, nitrogen content, and
uptake, with the application of 92 kg N ha-1 and 46 kg N ha-1 + 10 t compost ha-1 showing superior
results at the Hawassa Zuria and Meskan sites, respectively. Consequently, a conventional tillage
combined with a haricot bean-maize rotation system, supplemented with either 92 kg N ha-1 alone
or 46 kg N ha-1 + 10 t compost ha-1, is recommended for the Hawassa Zuria (Cambisols) and
Meskan (Chernozem) districts, respectively, to achieve optimal yield and nitrogen uptake.
However, for sustainable maize production in these areas, it is advisable to adopt an integrated
nitrogen treatment along with minimum tillage and legume-based crop rotation to enhance soil
properties, ultimately improving yields and nitrogen uptake.

REFERENCES

Adesoji, A.G., Abubakar, I.U. and Labe, D.A. 2015. Influence of incorporated legumes and
nitrogen fertilization on maize (Zea mays L.) nutrient uptake in a Semi-Arid environment.
Journal of Agriculture and Veterinary Science 8(3): 01-08.
Ashenafi, N., H. Wassie, A. Getachehu, and K. Alemayehu. 2021. Growth, Nitrogen Uptake of
Maize (Zea mays L.) and Soil Chemical Properties, and Responses to Compost and

31
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Nitrogen Rates and Their Mixture on Different Textured Soils: Pot Experiment. Applied
and Environmental Soil Science. Pp. 12 https://doi.org/10.1155/2021/9931763
Benton, J.J. 2003. Agronomic handbook: Management of crops, soils, and their fertility. CRC
Press LLC, 2000 N.W. Corporate Blvd., Boca Raton, Florida.
Chimdi, A., Gebrekidan, H., Kibret, K., Tadesse, A. 2012. Status of selected physio-chemical
properties of soil under different land use system of Western Oromia, Ethiopia. Journal of
Biological and Environmental Sciences 2(3):57-71.
CSA (Central Statistics Agency). 2019. Agricultural Sample Survey 2018/2019. Volume I. Report
on Area and Production of Major Crops (Private Peasant Holdings, Meher Season).
Statistical Bulletin 586. Addis Ababa.
Deepak, K.G., M.B. Roshan and K.S. Bishal. 2017. Effects of Biochar and Farmyard Manure on
Soil Properties and Crop Growth in an Agroforestry System in the Himalaya. Sustainable
Agriculture Research Journal 6: 1.
Dhillon, J., M. R. Del Corso, B. Figueiredo, E. Nambi, and W. Rau. 2018. Soil organic carbon,
total nitrogen, and soil pH in a long-term continuous winter wheat (Triticum Aestivum L.)
experiment. Communications in Soil Science and Plant Analysis. 19:803–13.
Dunjana, N.; Nyamugafata, P.; Shumba, A.; Nyamangara, J.; Zingore, S. 2012. Effects of cattle
manure on selected soil physical properties of smallholder farms on two soils of Murewa,
Zimbabwe. Soil Use Manag. 28: 221–228.
FAO (Food and Agricultural Organization) 2019. FAOSTAT statistical database and data sets of
the Food and Agriculture Organization of the United Nations. Available at:
http://www.faostat.fao.org.
Geisseler, D., W.R. Horwath. 2009. Short-term dynamics of soil carbon, microbial biomass, and
soil enzyme activities as compared to longer-term effects of tillage in irrigated row crops.
Biol. Fertil. Soils 16:65-72.
Hazelton P and Murphy B. 2007. Interpreting soil test results: What do all numbers mean? CSIRO
publishing: Melbourne.
Kaplan, M., Karaman, K., Kardes, Y.M., Kale, H. 2019. Phytic acid content and starch properties
of maize (Zea mays L.): Effects of irrigation process and nitrogen fertilizer. Food Chem. 283:
375–380.
Kassie, B.T., Hengsdijk, H., Rötter, R.P., Kahiluto, H., Asseng, S., and Ittersum, M.K.V. 2013.
Adapting to climate variability and change: Experiences from cereal-based farming in the
Central Rift and Kobo valleys, Ethiopia. Environ. Manag. 52:1115–1131.
Kirkegaard, J.A., O. Christen, J. Krupinsky, D. B. Layzell. 2008. Break crop benefits in temperate
wheat production. Field Crops Research 107, 185–195. doi:10.1016/j.fcr.2008.02.010.
Lafond, G.P., May, W.E., Stevenson, F.C., Derksen, D.A. 2006. Effects of Tillage Systems and
Rotations on Crop Production for a Thin Black Chernozem in the Canadian Prairies. Soil
Tillage Res. 89: 232–215.
Lorenz, K. and R. Lal. 2016. Environmental Impact of Organic Agriculture. Adv. Agron. 139: 99-
152.
Lupwayi, N. Z., A. C. Kennedy, and R. M. Chirwa. 2011. Grain legume impacts on soil biological
processes in sub-Saharan Africa. Afr. J. Plant Sci. 5(1): 1-7.
Mourice, S.K., Tumbo, S.D., Nyambilila, A., Rweyemamu, C.L. 2015. Modeling potential rain-
fed maize productivity and yield gaps in the Wami River subbasin, Tanzania. Acta
Agriculturae Scandinavica, Section B: Soil and Plant Science, 65(2):132-110.

32
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Negassa, W., Gebrekidan, H., Friesen, D.K. 2005. Integrated use of farmyard manure and NP
fertilizers for maize on farmers’ fields. J. Agric. Rural Dev. Trop. Subtrop. 106: 131–111.
Nelson, D.W. and Sommers, L.E. 1982. Total carbon, organic carbon, and organic matter. In: A.
L. Page, R. H. Miller & D. R. Keeney, (eds.). Methods of Soil Analysis, Part 2 Chemical
and microbiological properties. Agronomy monograph 9, Am. Society of Agronomy and
Soil Science Society of America, Madison, Wisconsin, USA, pp. 539-591.
Nelson, L.A., Voss, R.D. and Pesek, J. 1985. Agronomic and statistical evaluation of fertilizer
response. Pp. 89.
Rusinamhodzi, L., Corbeels, M., Zingore, S., Nyamangara, J., Giller, K.E. 2013. Pushing the
envelope? Maize production intensification and the role of cattle manure in the recovery
of degraded soils in smallholder farming areas of Zimbabwe. Field Crop. Res. 117: 10–53.
Salem, H.M., Valero, C., Muñoz, M.A., Rodríguez, M.G. and Silva, L.L. 2015. Short-term effects
of four tillage practices on soil physical properties, soil water potential, and maize yield.
Geoderma 237-238: 60-70.
SAS (Statistical Analysis System) 2001. SAS/STAT Software Syntax, Version 9.0. SAS Institute,
Cary, NC. USA. 751pp.
Simić M, Dragičević V, Mladenović Drinić S, Vukadinović J, Kresović B, Tabaković M, Brankov
M. 2020. The Contribution of Soil Tillage and Nitrogen Rate to the Quality of Maize Grain.
Agronomy. 10(7):976. https://doi.org/10.3390/agronomy10070976.
Steel, R.G.D. and Torrie, J.H. 1980. Principles and Procedures of Statistics: a Biometrical
Approach. 2nd Edition. McGraw-Hill. New York. Pp. 631.
Teklewold, H., Kassie, M., Shiferaw, B. 2013. Adoption of multiple sustainable agricultural
practices in rural Ethiopia. J. Agric. Econ. 61: 597–623.
Tolera, A., Daba, F. and Friesen, D.K. 2009. Effects of Crop rotation and N-P Fertilizer Rate on
Grain Yield and related characteristics of Maize and Soil Fertility at Bako Western Oromia,
Ethiopia. East African Journal of Science 3: 70-79.
Wang, X., Zhou, B., Sun, X., Yue, Y., Ma, W., Zhao, M. 2015. Soil Tillage Management Affects
Maize Grain Yield by Regulating Spatial Distribution Coordination of Roots, Soil Moisture
and Nitrogen Status. Plops ONE 10 e0129231.
Wyngaard, N., H. Echeverrıa, H. R. Sainz Rozas, and G. A. Divito. 2012. Fertilization and tillage
effects on soil properties and maize yield in a Southern Pampas Argiudoll. Soil and Tillage
Research. 119: 22–30.
Yan D, Wang D and Yang L, 2007. Long-term effect of chemical fertilizer, straw, and manure on
labile organic matter fractions in paddy soil. Biol. Fertil. Soils, 11(1):93-101.

33
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

POLICY IMPLICATIONS OF THE ADOPTION OF PRECISION AGRICULTURE IN

NIGERIA
1*Oyedepo, Elizabeth Omolola, 2Oyedepo, John Adebayo
1Augustine University, Ilara-Epe; 2Federal University of Agriculture, Abeokuta, Nigeria
e-mail: [email protected]; +2348038648941

ABSTRACT

This paper examines the policy implications of adopting Precision Agriculture (PA) in Nigeria,
emphasizing infrastructure, capacity-building, data regulation, and financial support. It highlights
the transformative potential of PA and recommends policy frameworks to facilitate its sustainable
integration into the agricultural sector. In Nigeria, where agriculture faces challenges including
outdated practices, inadequate infrastructure, and climate vulnerabilities, PA offers a promising
solution. Southwest States with soil health and irrigation technologies, lead adoption while regions
like Sokoto in the Northwest leverage smart irrigation through donor-supported projects. Despite
these advancements, significant barriers such as limited internet penetration, high costs, and low
digital literacy hinder widespread uptake. Policy reforms are crucial, focusing on infrastructure
development, financial incentives, specialized training, and data governance. Socioeconomic
benefits of PA include improved yields, reduced waste, and environmental conservation, but
equitable access for smallholder farmers is essential to ensure inclusive growth. Recommendations
include public-private partnerships for infrastructure, pilot programs in agricultural belts, and
capacity-building collaborations with international organizations. This will help Nigeria to transit
towards sustainable agriculture.

Keywords: Precision Agriculture, Policy, Nigeria, Sustainable Development, Agricultural

Technology

INTRODUCTION

Precision agriculture (PA) represents a paradigm shift in modern farming, leveraging technology
for improved productivity, efficient resource use, and sustainability [1]. In Nigeria, agricultural
productivity has remained suboptimal due to outdated farming practices, inadequate infrastructure,
and climate vulnerabilities [2]. The adoption of PA could without doubt, revolutionize Nigerian
agriculture, this is because precision agriculture integrates advanced technologies such as GPS,
IoT, and data analytics to enhance farming efficiency and environmental sustainability [3].

However, these technologies are inadequate in Nigeria; and to successfully adopt PA in Nigeria
will therefore require that robust policy frameworks addressing infrastructural deficits, financial
support, and human capital development be put in place [4, 5]. Nigeria, is a country still grappling
with food insecurity, land degradation, and climate change impacts and economic downturn.
Adoption of PA therefore presents an opportunity as well a challenge.

34
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Currently, data on the precise number of farmers using precision agriculture in Nigeria has not
been systematically aggregated across the country. However, there are indications of growing
adoption in specific locations and contexts.

While these examples highlight advancements, barriers such as limited internet penetration, high
technology costs, and low digital literacy remain significant challenges. Some locations might be
seeing progress due to private initiatives in solar-powered irrigation and market access platforms.
This paper is therefore an attempt to do a political economic analysis of the adoption of Precision
Agriculture in Nigeria. There is equally the need to do an x-ray of policy challenges and needs of
transition from traditional agriculture to hi-tech agriculture like PA.

Digital agriculture tools such as AI-powered crop monitoring, mobile apps, and precision farming
methods are gaining traction but remain concentrated in areas with better connectivity and
infrastructure [6]. Statistics on farmers using high-tech agricultural practices in Nigeria vary by
state and depend on factors like technology availability, education, and crop focus. In Northern
Nigeria, High-tech adoption is driven by large-scale farmers’ programs like the Anchor Borrowers.
In South-West Nigeria, there is moderate adoption of precision agriculture, with emphasis on soil
testing kits, mobile apps, and IoT tools for smart irrigation, largely facilitated by research
collaborations and local government interventions. In the South-East farmers are notable for
engaging in digital agriculture practices. The South-South region however, exhibit varied adoption
rates. Niger and Benue States demonstrate efforts in deploying agricultural drones and mobile
platforms for cassava farming, reflecting state-supported projects aimed at boosting crop yields
[7]. Despite these efforts, challenges like inadequate infrastructure, cost barriers, and lack of
technical training hinder broader adoption across most states.

Policy Challenges and Needs

• Infrastructure Development: A major bottleneck is the lack of basic infrastructure such as
reliable power, high-speed internet, and rural connectivity. Policies must prioritize
infrastructure investments that support PA technologies.
• Education and Training: A skilled workforce is critical for PA adoption. Government and
private sector collaboration is needed to develop specialized training programs and integrate
PA concepts into agricultural curricula.
• Financial Support: Many smallholder farmers lack the financial capacity to invest in PA
technologies. Policies should introduce subsidies, grants, and low-interest loans to make
these technologies accessible.
• Regulatory Framework for Data Governance: PA relies heavily on data collection and
analysis. A regulatory framework is needed to address data ownership, privacy, and sharing
protocols, ensuring farmer protection and fostering innovation.

Socioeconomic Impacts
Adopting PA in Nigeria could increase crop yields, reduce waste, and optimize input use,
addressing food insecurity while conserving environmental resources. However, the transition
must consider the digital divide and ensure equitable access for smallholder farmers, who
constitute the bulk of Nigeria’s agricultural workforce.

35
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

MATERIALS AND METHODS

Thirteen states within Nigeria were selected based on the higher level of food crop production
taking place in these regions compared to other areas in Nigeria and due to the relative proximity
of the ADPs (Agricultural Development Programmes) in the states which helped give direct access
to the farmers who were the main target group of the research. Primary data collected with
electronic means (smart phones with installed ODK apps) were mainly on level of adoption of
climate smart agriculture, barriers mitigating the adoption of precision agriculture and level of
internet connectivity in farming areas. Three local government areas (LGAs) with the highest
record of climate-induced stress were purposively selected in each state based on the advice of
state Agricultural Development Programme ADPs). In each selected LGA, four (4) villages were
randomly selected and thirty (30) farming households were selected per village. With twelve
villages per state and thirty farm households per village. A total of three hundred and sixty farm
households were selected per state and this makes up a total sample a sample size of 4,680
households (respondents) interviewed for the study. Descriptive statistics such as frequencies,
means and percentages were used to analyse the data and the results are presented below.

RESULTS AND DISCUSSION

Table 1 shows the distribution of respondents according to their personal characteristics. The
results revealed that majority of the farm household heads were male (67.43 %). This showed that
men dominated farming activities in the study area. This result is also in line with [9] who implied
that men dominated farming activities in Nigeria. In terms of age, majority of the respondents
(about 80%) are young, falling between ages 18 and 59 years. About 21% of the respondents are
within the age group of 60 years and above. This result show that majority of the farmers are able
bodied young people within their economically active years of life. In terms of level of literacy,
17.5% of the respondents revealed that they had no form of formal education. However, 30.97%
of the respondents had primary education, 28.13 % had secondary education while only 17.43%
had tertiary education. This result shows that majority of the respondents were literate with at least
primary education. Literacy will enable farmers to easily adopt new techniques of solving
problems of climate change impact on their farms.

Table 2 show distribution of respondents according to access to telecommunication services and

type of climate change risks experienced in the study area. Results revealed that in terms of access
to telecommunication services in the study area, more than half of the respondents had no regular
access to telecommunication services and 17.78 % of the respondents had smart phones. Majority
of the respondents had no regular access to telecommunication services.

Results on variability to climate change revealed that 64.8 % of the respondents reported they
experienced increased temperature and 78.3% reported they experienced decreased rainfall
duration and intensity. About 34.3% of the respondents reported a disappearance in vegetation
cover due to the climate change experienced. The respondents noted they experienced prolonged
dry spell after the early rains which led to increased temperature and loss of crops.

36
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 1. Distribution of respondents according to socioeconomic characteristics.

Variables Frequency Percentages Mean category

Sex
Male 3,156 67.43
Female 1,524 32.57
Age (years)
18 – 31 500 10.68
32 – 45 1,798 38.41 32 -45
46 – 59 1,339 28.61
60 – 73 864 18.47
74 and above 179 3.82
Level of Education
No formal education 858 18.33
Primary education 1,449 30.96
Secondary education 1,557 33.27
Tertiary education 815 17.43
Household size
0–4 809 17.29
5-9 2,613 55.83 5–9
10 – 14 998 21.32
15 & above 260 5. 56
Years of farming experience
1 – 10 1,021 21.81
11 – 20 1,670 35.69 11 – 20
21 – 30 974 20.83
31 – 40 624 13.34
41 and above 390 8.33
Source: Field Survey

Table 2. Distribution of respondents according to access to telecommunication services and type

of climate change risks experienced in the study area.

Variables Frequency Percentages

Category of mobile phone used
Ordinary phone 3,760 80.34
Smart phone 832 17.78
No phone 88 1.88
Access and availability of telecommunication services
No 3,878 82.86
Yes 802 17.15
Variability in climate change experienced?
Increased temperature 3,033 64.8
Decreased temperature 744 15.9
Increased rainfall duration and intensity 1,048 22.4
Decreased rainfall duration and intensity 3,664 78.3
Increased sunlight intensity 2,3112 49.4
Decreased sunlight intensity 491 10.5
Disappearance of vegetation cover 1,605 34.3
Increased changes in vegetation type 997 21.3
Source: Field Survey

37
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 3 reveals the extent to which climate change risks is experienced in the study area. About
43.64% of the respondents reported they always experience inadequate rainfall while 47.57
reported an occasional dry spell. More than half of the farmers (55.14 %) reported they
occasionally experienced frequent crop failure due to erratic rain distribution of rainfall. About
65% reported an occasional shortage of food for households. Majority of the respondents reported
an always or occasional experience of climate change risks in the study area. Only few of the
respondents indicated they have never experienced any of the climate change risks identified
except in cases of death of livestock and extinction of fishes and aquatic life due to climate change.

Table 3. Extent to which climate change risks is experienced.

Variables Always Occasionally Never

(%) (%) (%)
Inadequate rainfall 43.54 47.57 8.89
Frequent crop failure due to erratic distribution and 37.01 55.14 7.85
dwindling rainfall
Incessant flood 37.57 55.83 6.60
Frequent crop failure due to incessant flood 8.88 59.72 31.39
Shortage of food for households 17.78 65.0 0 17.22
Reduction of forest cover and degradation of natural 22.29 62.6 15.07
resources
Extinction of fishes and aquatic life 11.39 54.79 33.82
Death of livestock 11.04 38.40 50.56
Cattle invasion on farm due to inadequate grasses 31.04 48.82 20.14
Source: Field Survey

Recommendations
• Develop public-private partnerships to fund rural broadband and power projects.
• Establish centers for PA research and demonstration across the country.
• Create incentive schemes for local startups to develop affordable PA solutions tailored to
Nigeria’s agro-ecological zones.
• Implement pilot PA programs in key agricultural belts to assess scalability and economic
impact.
• Encourage collaborations with international organizations for knowledge transfer and
capacity-building initiatives.

CONCLUSION

The adoption of precision agriculture in Nigeria presents a pathway to achieving sustainable

agricultural intensification, food security, and environmental conservation. Policymakers must
create an enabling environment that bridges infrastructure gaps, fosters innovation, and ensures
equitable access to PA technologies. The future of Nigerian agriculture depends on its ability to
harness the transformative potential of precision farming.

38
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REFERENCES

1. Somashekar, K. S., Belagalla, N., Srinatha, T. N., Abhishek, G. J., Kumar, V., & Tiwari, A.
(2024). Revolutionizing Agriculture: Innovative Techniques, Applications, and Future
Prospects in Precision Farming. Journal of Scientific Research and Reports, 30(8), 405-419.
2. Oladoyinbo, O. B. (2023). Comprehensive synthesis and integrative review of agricultural
dynamics in southwest Nigeria: Assessing economic viability, technological advances, and
rural development approaches. Technological Advances, and Rural Development
Approaches (December 4, 2023).
3. Ani, O., Chukwuma, K., Okoro, G., & Oyiga, A. (2019). Applications of Mechatronics and
Autonomous Systems in Smart Agriculture: Potentials and Challenges for Nigeria. no.
November.
4. Abdullahi, H. S., & Sheriff, R. E. (2017). Case study to investigate the adoption of precision
agriculture in Nigeria using simple analysis to determine variability on a maize
plantation. Journal of Agricultural Economics and Rural Development, 3(3), 279-292.
5. John, D., Hussin, N., Shahibi, M. S., Ahmad, M., Hashim, H., & Ametefe, D. S. (2023). A
systematic review on the factors governing precision agriculture adoption among small-scale
farmers. Outlook on Agriculture, 52(4), 469-485.
6. Lawal, S. (2023). Digital agriculture: Transforming the landscape in Nigeria. Available online
at: https://farmingfarmersfarms.com/2023/12/digital-agriculture-transforming-the-
landscape-in-nigeria/
7. Nkwuagba, J. A., & Nkamnebe, A. D. (2024). Predictors of Agricultural Technology Adoption
Among the Cassava Farmers in the North-Central Nigeria. International Journal of Small
Business and Entrepreneurship Research, 12(2), 72-124
8. Nyaga, J.M., Onyango, C.M., Wetterlind, J. et al. Precision agriculture research in sub-Saharan
Africa countries: a systematic map. Precision Agric 22, 1217–1236 (2021).
https://doi.org/10.1007/s11119-020-09780-w
9. Sadiq MS, Yakassai MT, Ahmad MM, Lakpene TY, Abubakar M. Profitability and production
efficiency of small-scale maize production in Niger state, Nigeria. IOSR Journal of
Applied Physics (IOSR-JAP). 2013;3(4):19-23.

39
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ARTIFICIAL INTELLIGENCE (AI) IN AGRICULTURE

40
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

AN ENSEMBLE-BASED DEEP LEARNING APPROACH FOR EARLY AND

ACCURATE WHEAT DISEASE DETECTION
#11635

1Tagel
Aboneh, 1Prof. Abebe Rorissa, 2Prof. Ramasamy Srinivasagan
1Addis Ababa Science and Technology University
2Professor & Director, School of Information Sciences Associate Dean for Faculty Affairs

College of Communication and Information the University of Tennessee, Knoxville

ABSTRACT

Crop diseases are the primarily cause for yield loss and a factor for food security issue around the
globe. Crop diseases caused by pathogens pose a significant threat to global food security, the
challenge become worst particularly in developing countries like Ethiopia. Rapid population
growth and accurate disease identification is crucial for timely intervention and minimizing crop
losses. However, traditional methods often rely on expert analysis, which can be time-consuming
and resource intensive. The state of the art in agriculture employed AI enable crop diseases early
detection technologies to support the agriculture domain area. Currently machine learning based
solutions plays significant role to detect and classify crop diseases as early as possible. In this
study, we proposed ensemble-based deep learning approaches for crop diseases classification
purpose. Ensemble Deep Learning is a cutting-edge technique in machine learning that combines
the strengths of multiple deep learning models to achieve superior performance compared to any
individual model. The proposed model leverages the strengths of pre-trained models such as
ResNet50, EfficientNetB4, DenseNet, ViT-Base and VGG19. To train the base models, more
23,000 crop images are acquired from various sources. The trained models are combined using
ensemble learning method adjustable weighted average techniques to create a robust and
generalizable model.We done model performance assessment based on optimization, scalability,
and mitigation of model drifting issues to enhance the overall generalizability. From experimental
results, the proposed ensemble model demonstrated a promising performance with 99.48% for
training data and 99.23% for validation accuracy respectively. This research signifies a crucial step
towards developing a practical and reliable tool for early crop disease detection in resource-
constrained environments.

INTRODUCTION

The rapid population growth, a constant decline in arable land size per capital, and dynamic
environmental change are the main constraints in the agriculture process. On the other hand,
fighting against plant diseases is crucial activity in the agriculture sector to maintain crop
productivity. In this regards, researchers are attempting new methods and technology to support
the identification of crop diseases(Jasim & Al-Tuwaijari, 2020). The research findings in the
domain area reveal that the application of technology would enhance agriculture production
quality. In this regard, classical farming approaches, resources optimization, dynamic weather
condition, severity of different pathogens are the main cause for y(Habiba & Islam, 2021)ield
variation in the case of Ethiopia. In this study, we proposed ensemble based deep learning approach
to classify crop diseases into the respective categories. Based on different literature review reports,
about 20 to 40\% yield loss because of crop diseases. This is a significant effect which demand

41
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

appropriate action by stakeholder. On the other hand, climate change also a critical factor for crop
production and a cause for food insecurity issue. Thus, the sector demands an AI enabled system
at least to minimize its consequence on food security.

CONTRIBUTION

The proposed ensemble-based deep learning model holds significant potential to empower farmers
and agricultural stakeholders with the ability to rapidly identify and address crop diseases,
ultimately contributing to improved food security and agricultural sustainability. The proposed
model are converted smart phone application to support smallholder farmers. An ensemble-based
deep learning approaches are designed to develop generalizable models to improve the limitations
of crop diseases early detection(Fuentes et al., 2017). The proposed model can be used for others
crop disease detection and reference for wheat disease management work. Similarly, the proposed
model can be used as decision support tool for different stakeholder in the domain area.

MATERIALS AND METHODS

In this research, an empirical research approach has been used to implement an ensemble deep
learning framework to efficiently detect plant diseases(Reddy et al., 2021). In this work our main
focus is to build generalizable(Ferreira et al., 2020) machine learning model for the purpose of
crop diseases detection as early as possible. Figure 4.4 below illustrate the proposed system general
system architecture.

Figure 1. Ensemble based deep learning model data flow

DATASETS

To implement the experiment work, we have collected more than 25 thousand image datasets from
kaggle repository and 1500 wheat image (RGB) dataset from our previous research study entitle
with ’computer vision approach for wheat diseases classification using GPU infrastructure’. The
datasets are structure into 4 major categories namely corn, wheat, potatoes and tomato. These
categories are further classified into 20 classes (Figure 2).

42
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 2.

Data processing for feature extraction (Jasim & Al-Tuwaijari, 2020) and selection are the
important task before build the proposed model. In this regard, we have covered dimensional,
removal of the least relevant features, image normalization, formatting, removal of poor-quality
images, re-scaling or image resizing, and cropping of irrelevant parts of the image. Similarly, re-
scaling pixel intensities values ranging from 0 to 255. Furthermore, we transformed the data by re
scaling and setting the dimensions of the images at 224 × 224andchannel = 3 to standardize the
data set. We have used well annotated crop image data to train our model, and the data sources are
organized into training, testing and validation dataset. Then, we have selected five different deep
learning model as a base learner namely ResNet50, EfficientNetB4, DenseNet, ViT-Base and
VGG19 models (Rasti & Bleakley, 2020) (Figure 3). We have considered the following criteria,
such as the size and quality of the data set, computational resources, disease types and crops, and
accuracy requirements to select the base models. In line with ensemble learning method, the issue
of computation infrastructure is very critical. To solve this challenge, we have used NVIDIA
GeforceRTX3036 GPU facility to handle computational cost. Adjustable weighted average
method has been used to readjust the weight of individual based models based on the validation
loss accuracy.

43
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 3.

In case of ensemble learning method, evaluating pairwise correlation between base models is
very important before proceeding to build an ensemble model. In this regard, we computed the
pairwise correlation between the output probabilities or predicted labels of each model. On
Figure 4, we illustrated the model crop diseases prediction accuracy of 99.48% training and
99.23% validation accuracy respectively. From the experiment results, we can conclude that
the proposed model has learned the underlying patterns in the on the unseen dataset perfectly.

Figure 4. Ensemble model training and validation accuracy

We have used confusion matrix to understand the distribution of predictions across different
classes which helps us to make informed decisions, such as adjusting the threshold for specific
classes or focusing efforts on improving the model’s performance for critical classes. Similarly,
44
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

we have computed model drifting based the validation dataset to assess how the model’s
performance has changed over time. Finally, we have made comparison analysis between the
proposed model and similar studies in the domain area.

Ensemble learning is a powerful technique in machine learning where multiple models are
combined to produce a single superior model.

CONCLUSION

Crop diseases also remain as a major threat to food security. However, rapid disease detection
remains a home job for many developing countries such as Ethiopia. Different evidence reveal
that proportion of yield loss due crop diseases is significant. Currently, machine learning based
solutions plays significant role to detect and classify crop diseases as early as possible. In this
study, we proposed ensemble-based deep learning approaches for crop diseases classification
purpose. The underlying principle is that the collective decision of the committee tends to
exhibit superior overall accuracy compared to any individual member. From the experiment
results, the proposed model classifies the type of crop diseases with optimal accuracy. To
further justify model’s performance, we have used different scalability assessment on the
proposed model.

REFERENCES

Ferreira, J. R., Armando Cardona Cardenas, D., Moreno, R. A., De Fatima De Sa Rebelo, M.,
Krieger, J. E., & Antonio Gutierrez, M. (2020). Multi-View Ensemble Convolutional
Neural Network to Improve Classification of Pneumonia in Low Contrast Chest X-Ray
Images. Proceedings of the Annual International Conference of the IEEE Engineering in
Medicine and Biology Society, EMBS, 2020-July, 1238–1241.
https://doi.org/10.1109/EMBC44109.2020.9176517
Fuentes, A., Yoon, S., Kim, S. C., & Park, D. S. (2017). A robust deep-learning-based detector
for real-time tomato plant diseases and pests recognition. Sensors (Switzerland), 17(9).
https://doi.org/10.3390/s17092022
Habiba, S. U., & Islam, M. K. (2021). Tomato Plant Diseases Classification Using Deep
Learning Based Classifier from Leaves Images. 2021 International Conference on
Information and Communication Technology for Sustainable Development, ICICT4SD
2021 - Proceedings, 82–86. https://doi.org/10.1109/ICICT4SD50815.2021.9396883
Jasim, M. A., & Al-Tuwaijari, J. M. (2020). Plant Leaf Diseases Detection and Classification
Using Image Processing and Deep Learning Techniques. Proceedings of the 2020
International Conference on Computer Science and Software Engineering, CSASE 2020,
259–265. https://doi.org/10.1109/CSASE48920.2020.9142097
Rasti, S., & Bleakley, C. J. (2020). Crop growth stage estimation prior to canopy closure using
deep learning algorithms. 6. https://doi.org/10.1007/s00521-020-05064-6
Reddy, D. B. R. S., Madhavi, D. G. B., Lakshmi, C. H. S., Nagendra, D. K. V., & Sridevi, D.
R. (2021). Detection of Disease in Maize Plant Using Deep Learning. Alinteri Journal of
Agriculture Sciences, 36(2), 82–88. https://doi.org/10.47059/alinteri/v36i2/ajas21118
More references are available in the main content ….

45
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

MULTIVARIATE REGIONAL DEEP LEARNING PREDICTION OF SOIL

PROPERTIES FROM NEAR-INFRARED, MID-INFRARED AND THEIR
COMBINED SPECTRA
#11647

1,2*Rumbidzai W. Nyawasha, 3Alexandre M.J.-C. Wadoux, 4,5Pierre Todoroff, 1,6Regis

Chikowo, 1,2,5,7Gatien N. Falconnier, 4,5Maeva Lagorsse, 5,8Marc Corbeels, 1,2,5Rémi Cardinael
1Plant Production Sciences and Technology, University of Zimbabwe, Harare, Zimbabwe
2CIRAD, UPR AIDA, Harare, Zimbabwe
3LISAH, Université de Montpellier, INRAE, IRD, Institut Agro, Montpellier, France
4CIRAD, UPR AIDA, F-97410 Saint-Pierre, Réunion, France
5AIDA, Univ Montpellier, CIRAD, Montpellier, France
6Plant, Soil and Microbial Sciences Department, Michigan State University, East Lansing, MI

48824, USA
7
International Maize and Wheat Improvement Centre (CIMMYT)-Zimbabwe, Harare,
Zimbabwe
8IITA, International Institute of Tropical Agriculture, PO Box 30772, Nairobi 00100, Kenya
*e-mail: [email protected]; +263 776 163 414

ABSTRACT

Artificial neural network (ANN) models have been successfully used in infrared spectroscopy
research for the prediction of soil properties. They often show better performance than
conventional methods such as partial least squares regression (PLSR). In this paper we develop
and evaluate a multivariate extension of ANN for predicting correlated soil properties: total
carbon (C), total nitrogen (N), clay, silt, and sand contents, using visible near-infrared (vis-
NIR), mid-infrared (MIR) or combined spectra (vis-NIR + MIR). We hypothesize that
accounting for the correlation through joint modelling of soil properties with a single model
can eliminate “pedological chimera”: unrealistic values that may arise when properties are
predicted independently such as when calculating ratio or soil texture values. We tested two
types of ANN models, a univariate (ANN-UV) and a multivariate model (ANN-MV), using a
dataset of 228 soil samples collected from Murehwa district in Zimbabwe at two soil depth
intervals (0 – 20 and 20 - 40 cm). The models were compared with results from a univariate
PLSR (PLSR-UV) model. We found that the multivariate ANN model was better at conserving
the observed correlations between properties and consequently gave realistic soil C:N and
C:Clay ratios, but that there was no improvement in prediction accuracy over using a univariate
model (ANN or PLSR). The use of combined spectra (vis-NIR + MIR) did not make any
significant improvements in prediction accuracy of the multivariate ANN model compared to
using the vis-NIR or MIR only. We conclude that the multivariate ANN model is better suited
for the prediction of multiple correlated soil properties and that it is flexible and can account
for compositional constrains. The multivariate ANN model helps to keep realistic ratio values
– with strong implications for assessment studies that make use of such predicted soil values.

INTRODUCTION

Soils and soil properties vary over space in relation to the parent material, climate, topography,
among others, and change over time in response to natural processes and human activities
(Jenny, 1994; Beillouin et al., 2023). Sampling and monitoring of soils is costly and time
consuming, as it usually requires a large number of measurements and laboratory analyses
(Webster and Lark, 2013). To adequately capture the spatial and temporal variations of soils,
46
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

effective and less costly methods of data collection and analysis have been developed, including
the use of visible and near-infrared (vis-NIR) and mid-infrared (MIR) spectroscopy (Nocita et
al., 2015). Statistical models can then be employed to establish a predictive relationship
between the spectral characteristics and values of soil properties for which corresponding
laboratory measurements are available (Barra et al., 2021). Partial least squares regression
(PLSR) has become the most popular regression model in soil spectroscopy (Viscarra Rossel
and Lark, 2009; Soriano-Disla et al., 2014). It has been shown to perform well in many
situations (Janik et al., 1998; Viscarra Rossel et al., 2006; Cambou et al., 2016; Allo et al., 2020;
Bachion de Santana and Daly, 2022). Usually, each soil property is modelled independently,
ignoring the correlations that exist between properties. In cases where multiple dependent
properties are predicted, this can result in inconsistent predictions and the occurrence of
“pedological chimera” as defined by Lagacherie et al. (2022). As a solution, multivariate
counterparts of PLSR have been developed, the most common being the PLS2 regression
model, a modification of PLSR developed by Wold et al. (1983) and Martens and Naes (1987).
However, in terms of predictive accuracy, PLS2 usually performs worse than a model fitted for
an individual variable. Several studies, (Pedro and Ferreira, 2007; Blanco and Peguero, 2008;
Mishra and Passos, 2022), acknowledged that the univariate model gave higher prediction
accuracy than PLS2.

Recently, data-driven models and algorithmic tools from the field of machine learning have
become popular for predicting soil properties from spectral data (Meza Ramirez et al., 2021).
Commonly used algorithms in soil spectroscopy are support vector machines (Demattê and da
Silva Terra, 2014; Deiss et al., 2020), cubist (Minasny and McBratney, 2008; Clergue et al.,
2023), random forest (Viscarra Rossel and Behrens, 2010; McDowell et al., 2012; Wadoux,
2023), and artificial neural networks (ANNs) (Daniel et al., 2003; Wijewardane et al., 2018).
The use of ANNs has been successful for soil property prediction and showed better
performance than conventional methods such as PLSR in several studies (Daniel et al., 2003;
Viscarra Rossel and Behrens, 2010; Ng et al., 2019; Padarian et al., 2019). The main advantages
of ANNs over conventional regression models are the ability to extract relevant information in
high-dimensional datasets, the modelling of non-linear relationships between spectra and soil
properties, and a flexibility in the definition of the algorithm and objective function (Ludwig et
al., 2019; Margenot et al., 2020). Despite its flexibility, to date very few studies have attempted
to understand whether a multivariate ANN model accounts for the correlations that exist
amongst soil properties, although promising results were found in Mishra and Passos (2022),
Ng et al. (2019), and Ramsundar et al. (2015).

In this paper we develop, further expand, and test the multivariate extension of ANNs for
predicting soil properties from their vis-NIR, MIR and combined spectra (vis-NIR + MIR).
After model development, we investigate the ability of the multivariate model to predict
correlated soil properties, as compared to a model that predicts each property individually. The
methodology is tested for total carbon, total nitrogen, sand, silt, and clay contents in soils from
Murehwa district located in the sub-humid region of Zimbabwe. We hypothesize that combined
modelling of several soil properties can eliminate “pedological chimera” by accounting for the
correlations between the properties. The comparison between observed and predicted soil
properties from a univariate and a multivariate model is made using vis-NIR, MIR or combined
vis-NIR + MIR spectra.

47
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

METHODOLOGY

The study was done in Murehwa district (17°39'S, 31°47'E), a smallholder farming area situated
about 80 km northeast of Harare, the capital city of Zimbabwe. Soil samples were collected in
three villages randomly selected from Ward 28 of the district. 50 % of the households in the
three villages were then randomly selected to give a total of 183 farming households. Soil
samples were collected from all agricultural fields belonging to the selected households.
Samples were also collected from common lands – lands that are available for all villagers and
used for grazing, collecting firewood, litter, and wild fruits. Soil samples were collected
between June and July 2021 at two depths i) 0 – 20 cm ii) 20 – 40 cm. Sampling was carried
out following a zig-zig transect covering each field, with a sub-sample being collected at 10 m
distance using an auger and all the sub-samples were mixed to obtain a composite per field and
depth.

Spectra were acquired at the laboratory of the French Agricultural Research Centre for
International Development (CIRAD) in Saint Denis, La Réunion, on all soil samples ground to
200µm. The MIR spectra were measured using an Agilent 4300 handheld FTIR spectrometer
(Agilent Technologies, Santa Clara, CA) over a wavenumber range between 650 – 4000 cm-1
with a measurement interval of 4 cm-1, vis-NIR spectra were measured using a LabSpec 5000
(Analytical Spectral Devices, Inc. Boulder, CO, USA) with an optical fibre connected to the
internal light (adapted to small sample sizes) over a wavelength of 350 – 2500nm and spectral
resolution of 3 nm at 700 nm and 10 nm at 1400/2100 nm. Spectral pre-processing was done to
ensure the removal of any variations caused by light scattering and to enhance some features
within the spectra (Wadoux et al., 2021). The MIR spectra were trimmed to remove the noise
at the edges leaving the range between 800-4000 cm-1 whereas vis-NIR spectra were trimmed
to 20000 – 4080 cm-1. The MIR and vis-NIR datasets were then combined using spectra
concatenation to create a third dataset (vis-NIR + MIR) ranging between 10000 - 800 cm-1. A
For laboratory analysis, a subset of 230 soil samples, corresponding to 17 % of the total number
of samples, was selected. The selection was based on spectra similarity and the most
representative spectra were chosen using the Kennard Stone algorithm as implemented in the
Unscrambler X 10.5 Software (CAMO Software Inc., Oslo, Norway). Total carbon and total
nitrogen were determined by the Dumas elemental dry combustion method using an Elementar
VarioMax Cube. Soil texture analysis was done using the hydrometer method following Gee
and Bauder (1986).

Two types of ANN models were built, a univariate model which predicts one soil property at a
time, and a multivariate model which predicts more than one property at the same time. The
univariate model was made up of one input layer, three hidden layers and one output layer. The
multivariate model was made up of one input layer, four hidden layers and an output layer
predicting five outputs simultaneously. The models were trained using vis-NIR, MIR and the
combined vis-NIR + MIR data. The two ANN models were compared to a univariate PLSR
model to gauge their performance against a conventional model. The ANN models in this study
were built using the keras package (Allaire and Chollet, 2023) in R with tensorflow as backend
(Allaire and Tang, 2023) and the PLSR was built using the pls package (Liland et al., 2023)
also in R. The measured values of the soil properties from the laboratory analyses were used to
fit the models. The measured values were split into training and validation sets using k-fold
cross-validation to assess prediction accuracy of the model predictions on unseen data.
Validation statistics – i.e. mean error (ME), the root mean square error (RMSE) and the
coefficient of determination R2 - were calculated from the pairwise comparison of measured

48
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

and predicted values obtained from all folds as each represents a specific aspect of prediction
quality.

RESULTS AND DISCUSSION

The best prediction models were obtained using MIR spectra, followed by vis-NIR + MIR
spectra and lastly by vis-NIR spectra. Model predictions based on MIR spectra had consistently
higher R2 values and lower RMSE values, and this difference was significant when compared
to predictions based on vis-NIR spectra (Table 1). This can be attributed to the presence of
fundamental vibrations in the MIR region whereas only overtones and combinations bands are
present in the vis-NIR regions. Other studies report similar results, particularly for soil carbon
predictions where MIR outperforms vis-NIR (Viscarra Rossel et al., 2006; Vohland et al., 2014;
Wijewardane et al., 2018). The use of combined vis-NIR + MIR spectra did not improve the
predictive accuracy of soil properties in this study. There are varying results on this - a study
conducted by Johnson et al. (2019) reported an improved accuracy with combined spectra for
several soil properties whereas others report that because the predictions with MIR spectra
alone are already highly accurate, combining spectra either results in slightly worse results
(Viscarra Rossel et al., 2006; Shao and He, 2011; Ng et al., 2019) or produces results that are
equally comparable to MIR alone (Knox et al., 2015).

Table 1. Comparison of the PLSR-UV, ANN-UV and ANN-MV models for three spectral
datasets, vis-NIR, MIR and combined vis-NIR + MIR using mean error (ME), root mean square
error (RMSE) and coefficient of determination (R2)

We also studied the predictions of two key ratios: (1) soil C:N ratio, which is calculated using
total carbon and total nitrogen values and is a sensitive indicator of soil quality and for assessing
the carbon and nitrogen nutrition balance of soils; and (2) the C:Clay ratio, calculated using soil
carbon and clay content, which has been proposed as an indicator for soil organic carbon status
and soil structure quality (Poeplau and Don, 2023). The range of values for the soil C:N ratio
was all within the range between 10 – 25, comparable to the measured values, whereas the
ANN-UV model gave more unrealistic values including some negative ones. Previous studies
in the study area have shown that soil carbon concentrations in the most fertile soils rarely
exceed 10 g C kg-1 (Masvaya et al., 2010; Zingore et al., 2011). For the C:Clay ratios the
predictions made by the ANN-MV model gave significantly better results (Figure 2). Soil clay
content plays an important role in the stabilization of SOC since clay minerals have a high
specific surface area and carry a charge, enabling them to bind, and thereby chemically
stabilize, organic matter. Clay aggregates also provide micropores for the physical protection
of soil organic carbon (Wattel-Koekkoek et al., 2001). The C:Clay ratios obtained in this study
range between 1:10 – 1:13 and sometimes even lower, which suggests that these soils are

49
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

degraded (Poeplau and Don, 2023). This is accurate as these soils are granitic derived. A low
clay plus silt fraction usually provides little physical protection of organic matter to influence
soil physical properties (Feller and Beare, 1997; Nyamangara et al., 2014). Moreover, clay
content is not an accurate predictor of SOC, particularly in tropical soils with high
concentrations of aluminium and iron oxides (Khomo et al., 2017; Kirsten et al., 2021).

Figure 1. Boxplots of a) soil C:N ratio and b) C:Clay ratio calculated with measured values and
the two ANN models

REFERENCES

Allaire, J.., Chollet, F., 2023. Keras: R interface to ‘Keras’ [WWW Document]. URL
https://github.com/rstudio/keras (accessed 8.23.23).
Allaire, J., Tang, Y., 2023. Tensorflow: R Interface to ‘TensorFlow’ [WWW Document]. R
Packag. version 2.11.0.9000. URL https://github.com/rstudio/tensorflow%0A (accessed
8.23.23).
Allo, M., Todoroff, P., Jameux, M., Stern, M., Paulin, L., Albrecht, A., 2020. Prediction of
tropical volcanic soil organic carbon stocks by visible-near- and mid-infrared
spectroscopy. Catena 189, 104452. https://doi.org/10.1016/j.catena.2020.104452
Bachion de Santana, F., Daly, K., 2022. A comparative study of MIR and NIR spectral models
using ball-milled and sieved soil for the prediction of a range soil physical and chemical
parameters. Spectrochim. Acta - Part A Mol. Biomol. Spectrosc. 279, 121441.
https://doi.org/10.1016/j.saa.2022.121441
50
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Barra, I., Haefele, S.M., Sakrabani, R., Kebede, F., 2021. Soil spectroscopy with the use of
chemometrics, machine learning and pre-processing techniques in soil diagnosis: Recent
advances–A review. TrAC - Trends Anal. Chem. 135, 116166.
https://doi.org/10.1016/j.trac.2020.116166
Beillouin, D., Corbeels, M., Demenois, J., Berre, D., Boyer, A., Fallot, A., Feder, F., Cardinael,
R., 2023. A global meta-analysis of soil organic carbon in the Anthropocene. Nat.
Commun. 14, 1–10. https://doi.org/10.1038/s41467-023-39338-z
Blanco, M., Peguero, A., 2008. An expeditious method for determining particle size distribution
by near infrared spectroscopy: Comparison of PLS2 and ANN models. Talanta 77, 647–
651. https://doi.org/10.1016/j.talanta.2008.07.015
Cambou, A., Cardinael, R., Kouakoua, E., Villeneuve, M., Durand, C., Barthès, B.G., 2016.
Prediction of soil organic carbon stock using visible and near infrared reflectance
spectroscopy (VNIRS) in the field. Geoderma 261, 151–159.
https://doi.org/10.1016/j.geoderma.2015.07.007
Clergue, T.C., Saby, N.P.A., Wadoux, A.M.J.-C., Barthès, B.G., Lacoste, M., 2023. Estimating
soil aggregate stability with infrared spectroscopy and pedotransfer functions. Soil Secur.
11, 100088. https://doi.org/10.1016/j.soisec.2023.100088
Daniel, K.W., Tripathi, N.K., Honda, K., 2003. Artificial neural network analysis of laboratory
and in situ spectra for the estimation of macronutrients in soils of Lop Buri (Thailand).
Aust. J. Soil Res. 41, 47–59. https://doi.org/10.1071/SR02027
Deiss, L., Margenot, A.J., Culman, S.W., Demyan, M.S., 2020. Tuning support vector machines
regression models improves prediction accuracy of soil properties in MIR spectroscopy.
Geoderma 365, 114227. https://doi.org/10.1016/j.geoderma.2020.114227
Demattê, J.A.M., da Silva Terra, F., 2014. Spectral pedology: A new perspective on evaluation
of soils along pedogenetic alterations. Geoderma 217–218, 190–200.
https://doi.org/10.1016/j.geoderma.2013.11.012
Feller, C., Beare, M.H., 1997. Physical control of soil organic matter dynamics in the tropics.
Geoderma 79, 69–116. https://doi.org/10.1016/S0016-7061(97)00039-6
Gee, G.W., Bauder, J.W., 1986. Particle-size Analysis, in: Klute, A. (Ed.), Methods of Soil
Analysis, Part 1. Physical and Mineralogical Methods-Agronomy. Agronomy Society of
America/Soil Science Society of America, Madison, Wisconsin, pp. 384–411.
Janik, L.J., Merry, R.H., Skjemstad, J.O., 1998. Can mid infrared diffuse reflectance analysis
replace soil extractions? Aust. J. Exp. Agric. 38, 681–696.
https://doi.org/10.1071/EA97144
Jenny, H., 1994. Factors of Soil Formation A System of Quantitative Pedology. Courier
Corporation.
Johnson, J.M., Vandamme, E., Senthilkumar, K., Sila, A., Shepherd, K.D., Saito, K., 2019.
Near-infrared, mid-infrared or combined diffuse reflectance spectroscopy for assessing
soil fertility in rice fields in sub-Saharan Africa. Geoderma 354, 113840.
https://doi.org/10.1016/j.geoderma.2019.06.043
Khomo, L., Trumbore, S., Bern, C.R., Chadwick, O.A., 2017. Timescales of carbon turnover
in soils with mixed crystalline mineralogies. Soil 3, 17–30. https://doi.org/10.5194/soil-
3-17-2017
Kirsten, M., Mikutta, R., Vogel, C., Thompson, A., Mueller, C.W., Kimaro, D.N., Bergsma,
H.L.T., Feger, K.H., Kalbitz, K., 2021. Iron oxides and aluminous clays selectively
control soil carbon storage and stability in the humid tropics. Sci. Rep. 11, 1–12.
https://doi.org/10.1038/s41598-021-84777-7
Knox, N.M., Grunwald, S., McDowell, M.L., Bruland, G.L., Myers, D.B., Harris, W.G., 2015.
Modelling soil carbon fractions with visible near-infrared (VNIR) and mid-infrared
(MIR) spectroscopy. Geoderma 239–240, 229–239.
51
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

https://doi.org/10.1016/j.geoderma.2014.10.019
Lagacherie, P., Buis, S., Constantin, J., Dharumarajan, S., Ruiz, L., Sekhar, M., 2022.
Evaluating the impact of using digital soil mapping products as input for spatializing a
crop model: The case of drainage and maize yield simulated by STICS in the Berambadi
catchment (India). Geoderma 406, 115503.
https://doi.org/https://doi.org/10.1016/j.geoderma.2021.115503.
Liland, K.., Mevik, B.-H., Wehrens, R., Hiemstra, P., 2023. pls: Partial Least Squares and
Principal Component Regression [WWW Document]. 2.8-2. URL
https://github.com/khliland/pls (accessed 8.23.23).
Ludwig, B., Murugan, R., Parama, V.R.R., Vohland, M., 2019. Accuracy of Estimating Soil
Properties with Mid-Infrared Spectroscopy: Implications of Different Chemometric
Approaches and Software Packages Related to Calibration Sample Size. Soil Sci. Soc.
Am. J. 83, 1542–1552. https://doi.org/10.2136/sssaj2018.11.0413
Margenot, A., O’ Neill, T., Sommer, R., Akella, V., 2020. Predicting soil permanganate
oxidizable carbon (POXC) by coupling DRIFT spectroscopy and artificial neural
networks (ANN). Comput. Electron. Agric. 168, 105098.
https://doi.org/10.1016/j.compag.2019.105098
Martens, H., Naes, T., 1987. Multivariate calibration by data compression, in: Williams, P..,
Norris, K. (Eds.), Near-Infrared Technology in the Agricultural and Food Industries.
American Association of Cereal Chemists, St. Paul, MN, pp. 57–87.
Masvaya, E.N., Nyamangara, J., Nyawasha, R.W., Zingore, S., Delve, R.J., Giller, K.E., 2010.
Effect of farmer management strategies on spatial variability of soil fertility and crop
nutrient uptake in contrasting agro-ecological zones in Zimbabwe. Nutr. Cycl.
Agroecosystems 88, 111–120. https://doi.org/10.1007/s10705-009-9262-y
McDowell, M.L., Bruland, G.L., Deenik, J.L., Grunwald, S., Knox, N.M., 2012. Soil total
carbon analysis in Hawaiian soils with visible, near-infrared and mid-infrared diffuse
reflectance spectroscopy. Geoderma 189–190, 312–320.
https://doi.org/10.1016/j.geoderma.2012.06.009
Meza Ramirez, C.A., Greenop, M., Ashton, L., Rehman, I. ur, 2021. Applications of machine
learning in spectroscopy. Appl. Spectrosc. Rev. 56, 733–763.
https://doi.org/10.1080/05704928.2020.1859525
Minasny, B., McBratney, A.B., 2008. Regression rules as a tool for predicting soil properties
from infrared reflectance spectroscopy. Chemom. Intell. Lab. Syst. 94, 72–79.
https://doi.org/10.1016/j.chemolab.2008.06.003
Mishra, P., Passos, D., 2022. Multi-output 1-dimensional convolutional neural networks for
simultaneous prediction of different traits of fruit based on near-infrared spectroscopy.
Postharvest Biol. Technol. 183, 111741.
https://doi.org/10.1016/j.postharvbio.2021.111741
Ng, W., Minasny, B., Montazerolghaem, M., Padarian, J., Ferguson, R., Bailey, S., McBratney,
A.B., 2019. Convolutional neural network for simultaneous prediction of several soil
properties using visible/near-infrared, mid-infrared, and their combined spectra.
Geoderma 352, 251–267. https://doi.org/10.1016/j.geoderma.2019.06.016
Nocita, M., Stevens, A., van Wesemael, B., Aitkenhead, M., Bachmann, M., Barthès, B., Dor,
E. Ben, Brown, D.J., Clairotte, M., Csorba, A., Dardenne, P., Demattê, J.A.M., Genot,
V., Guerrero, C., Knadel, M., Montanarella, L., Noon, C., Ramirez-Lopez, L., Robertson,
J., Sakai, H., Soriano-Disla, J.M., Shepherd, K.D., Stenberg, B., Towett, E.K., Vargas,
R., Wetterlind, J., 2015. Soil Spectroscopy: An Alternative to Wet Chemistry for Soil
Monitoring. Adv. Agron. 132, 139–159. https://doi.org/10.1016/bs.agron.2015.02.002
Nyamangara, J., Marondedze, A., Masvaya, E.N., Mawodza, T., Nyawasha, R., Nyengerai, K.,
Tirivavi, R., Nyamugafata, P., Wuta, M., 2014. Influence of basin-based conservation
52
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

agriculture on selected soil quality parameters under smallholder farming in Zimbabwe.

Soil Use Manag. 30, 550–559. https://doi.org/10.1111/sum.12149
Padarian, J., Minasny, B., McBratney, A.B., 2019. Using deep learning to predict soil properties
from regional spectral data. Geoderma Reg. 16, e00198.
https://doi.org/10.1016/j.geodrs.2018.e00198
Pedro, A.M.K., Ferreira, M.M.C., 2007. Simultaneously calibrating solids, sugars and acidity
of tomato products using PLS2 and NIR spectroscopy. Anal. Chim. Acta 595, 221–227.
https://doi.org/10.1016/j.aca.2007.03.036
Poeplau, C., Don, A., 2023. A simple soil organic carbon level metric beyond the organic
carbon-to-clay ratio. Soil Use Manag. 39, 1057–1067. https://doi.org/10.1111/sum.12921
Ramsundar, B., Kearnes, S., Riley, P., Webster, D., Konerding, D., Pande, V., 2015. Massively
Multitask Networks for Drug Discovery, in: International Conference on Machine
Learning. https://doi.org/10.48550/arXiv.1502.02072
Shao, Y., He, Y., 2011. Nitrogen, phosphorus, and potassium prediction in soils, using infrared
spectroscopy. Soil Res. 49, 166–172. https://doi.org/10.1071/SR10098
Soriano-Disla, J.M., Janik, L.J., Viscarra Rossel, R.A., MacDonald, L.M., McLaughlin, M.J.,
2014. The performance of visible, near-, and mid-infrared reflectance spectroscopy for
prediction of soil physical, chemical, and biological properties. Appl. Spectrosc. Rev. 49,
139–186. https://doi.org/10.1080/05704928.2013.811081
Viscarra Rossel, R.A., Behrens, T., 2010. Using data mining to model and interpret soil diffuse
reflectance spectra. Geoderma 158, 46–54.
https://doi.org/10.1016/j.geoderma.2009.12.025
Viscarra Rossel, R.A., Lark, R.M., 2009. Improved analysis and modelling of soil diffuse
reflectance spectra using wavelets. Eur. J. Soil Sci. 60, 453–464.
https://doi.org/10.1111/j.1365-2389.2009.01121.x
Viscarra Rossel, R.A., Walvoort, D.J.J., McBratney, A.B., Janik, L.J., Skjemstad, J.O., 2006.
Visible, near infrared, mid infrared or combined diffuse reflectance spectroscopy for
simultaneous assessment of various soil properties. Geoderma 131, 59–75.
https://doi.org/10.1016/j.geoderma.2005.03.007
Vohland, M., Ludwig, M., Thiele-Bruhn, S., Ludwig, B., 2014. Determination of soil properties
with visible to near- and mid-infrared spectroscopy: Effects of spectral variable selection.
Geoderma 223–225, 88–96. https://doi.org/10.1016/j.geoderma.2014.01.013
Wadoux, A.M.J.-C., 2023. Interpretable spectroscopic modelling of soil with machine learning.
Eur. J. Soil Sci. 74, 1–14. https://doi.org/10.1111/ejss.13370
Wadoux, A.M.J.-C., Malone, B., Minasny, B., Fajardo, M., McBratney, A.B., 2021. Soil
Spectral Inference with R: Analysing Digital Soil Spectra Using the R Programming
Environment, Springer International Publishing AG, Cham.
Wattel-Koekkoek, E.J.W., Van Genuchten, P.P.L., Buurman, P., Van Lagen, B., 2001. Amount
and composition of clay-associated soil organic matter in a range of kaolinitic and
smectitic soils. Geoderma 99, 27–49. https://doi.org/10.1016/S0016-7061(00)00062-8
Webster, R., Lark, R.M., 2013. Field Sampling for Environmental Science and Management.
Routledge, New York. https://doi.org/10.4324/9780203128640
Wijewardane, N.K., Ge, Y., Wills, S., Libohova, Z., 2018. Predicting Physical and Chemical
Properties of US Soils with a Mid‐Infrared Reflectance Spectral Library. Soil Sci. Soc.
Am. J. 82, 722–731. https://doi.org/10.2136/sssaj2017.10.0361
Wold, S., Albano, C., Dunn III, W.J., Esbensen, K., Hellberg, S., Johansson, E., Sjorstrom, S.,
1983. Pattern recognition: Finding and using regularities in multi-variate data, in:
Martens, H., Russwurm, Jr, H. (Eds.), Food Research and Data Analysis. Applied Science
Publishers, London, pp. 147–188.
Zingore, S., Tittonell, P., Corbeels, M., van Wijk, M.T., Giller, K.E., 2011. Managing soil
53
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

fertility diversity to enhance resource use efficiencies in smallholder farming systems: A

case from Murewa District, Zimbabwe. Nutr. Cycl. Agroecosystems 90, 87–103.
https://doi.org/10.1007/s10705-010-9414-0

54
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

CLIMATE SMART AGRICULTURE

55
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

PREDICTING THE DISTRIBUTION OF GROUNDNUT PHYTOPATHOGENS

UNDER CURRENT AND FUTURE CLIMATIC SCENARIOS IN ZIMBABWE
#11736

1Hardlife Chinwa, 1Elizabeth Ngadze, 1Juliet Murimwa, 2,3Eliel Belchrist Sossou, 4Clara
Machokoto, 1Regis Chikowo, 1*Casper Nyaradzai Kamutando
e-mail: [email protected]; 00263713232033

ABSTRACT

Groundnut (Arachis hypogaea L.) is an important oil seed crop with immense nutritional and
economic benefits, but its productivity in sub-Saharan Africa (SSA) is threatened by a plethora
of phytopathogens such as groundnut rosette virus, Alternaria leafspots, early leafspots and
peanut rust. In Zimbabwe, ecological niches and epidemiology of these pathogenic microbial
strains, particularly under the current and predicted climate change scenarios, are still poorly
understood. Yet, this information is vital in crafting effective and sustainable disease
management approaches for production of this important crop. Here, a field survey was
conducted on small-scale farms across a climate gradient during the 2023-24 and 2024-25
summer cropping seasons, to predict the current and future (2050) distribution of groundnut
phytopathogens in Zimbabwe. Three out of the five main agro-ecological regions (AER) of
Zimbabwe (i.e., AERII, AERIII and AERIV) were used as sampling sites. Georeferenced data
was subjected to maximum entropy (MaxEnt) algorithm for species distribution modelling. The
model identified isothermality, precipitation and temperature seasonality as the major
environmental variables governing suitability of groundnut phytopathogens occurrence across
the diverse agro ecologies of Zimbabwe. Distribution of groundnut phytopathogens in
Zimbabwe was predicted to vary with change in climatic conditions, in particular, rainfall and
temperature-based bio-climatic variables. In conclusion, groundnut phytopathogens are widely
distributed in Zimbabwe and climate change under current and future climate scenarios will
influence their distribution posing significant threat to food security.

Keywords: Phytopathogens; Groundnut; Zimbabwe; Climate change; MaxEnt

INTRODUCTION

Foliar diseases are one of the major biotic stresses impacting the productivity of groundnuts, a
potential crop widely grown for food and nutrition security by small scale farmers in sub-
Saharan Africa (SSA) [1]. Renown biotypes of economic importance are (1) groundnut rosette
virus (GRV), the most devastating viral phytopathogen cosmopolitan to SSA, (2)
Nothopassalora arachidicola (late leafspot), (3) Passalora arachidicola (early leafspot), (4)
Puccinia arachidis (peanut rust) and (5) Alternaria species (i.e., A. alternata (Fr.) Keissler., A.
tennuissima, A arachidis and A. longipse (leaf blights)) [2-5]. Groundnut rosette alone can cause
100% yield losses while leafspots can cause >50% yield losses [6, 7]. In Ghana, yield losses
between 78-88% due to GRV on groundnuts was reported by Appiah, Offei (3). In 2003,
Nutsugah, Oti-Boateng (8) reported yield loss of up to 80% on pod yield due to early and late
leafspot in Ghana. Mau and Ndiwa (9) reported yield loss of up to 57% due to the co-infection
of peanut rust and late leafspot in Indonesia. As groundnut is a cash crop and source of nutrition
for thousands of smallholder farmers in Zimbabwe, groundnut phytopathogens through yield
losses can obscure the roadmap towards food and nutrition security patronaged by the
sustainable development goals (SDG2 and SDG3) of the United nations [10-12].

56
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Due to wide range of hosts, exchange of planting materials, spatial production of landraces,
evolution of pathogens as well as the inadvertent sporadic weather patterns, and lack of resistant
varieties, the environmental suitability of groundnut phytopathogens in Zimbabwe is likely to
increase significantly [13, 14]. Exacerbated by climate change, various agro-ecological regions
in Zimbabwe are likely to share similar environmental characteristics favoured by foliage
disease-causing microbes for groundnuts.

Understanding the epidemic and conducive environmental conditions promoting the

pathogenicity of groundnut diseases in Zimbabwe is a first line of defence against future
potential yield losses and inoculum build-up as it helps policy makers in deciding whether to
go for new varieties or improved plant and soil health. State-of-the-art species distribution
models such as correlative (also referred to as bioclimatic), and mechanistic models have been
developed to pacify our comprehension on the spread of a species as a function of
environmental variables through the use of computer algorithms [15]. Predicting the
distribution of species under climate change using mathematical models is useful in: (i)
revealing the biogeographical patterns of a species, (ii) finding suitable areas of re-introduction
of endangered species, (iii) assessing how environmental conditions influence the occurrence
or abundance of species, and (iv) ecological forecasting [13, 14]. In this study, a field survey
was conducted on smallholder groundnut farmers across a climate gradient in Zimbabwe to
predict the potential spatial distribution of groundnut phytopathogens under current and future
climates. We hypothesize that, the distribution of groundnut phytopathogens in Zimbabwe is
influenced by current and future climates.

MATERIALS AND METHODS

Occurrence data
Presence only (PO) data were collected as input data across a climate gradient, encompassing
three out of the five agro-ecological regions of Zimbabwe from a total of 445 occurrence points
during the 2023-24 and 2024-25 summer cropping season.

Disease confirmation under compound microscope

Symptomatic diseased groundnut samples were brought to the laboratory for further
confirmation through the aid of a compound microscope. Fungi specimen was cultured in an
incubator at 25 ℃ using potato-dextrose agar until spores were developed.

Data for environmental variables

Worldclim’s (www.worldclim.org) nineteen (19) bioclimatic variables (BIO1-BIO19) from
1970-2000 were downloaded from database as raster layers. Four socio-economic pathways
(SSPs) namely, SSP5-8.5 (highest scenario), SSP3-7.0 (high scenario), SSP2-4.5 (medium
scenario) and SSP1-2.6 (low scenario) for greenhouse gas (GHG) emissions, released by the
Intergovernmental Panel on Climate Change (IPCC) in 2013, were selected from Australia
ACCESS CM2 of the Coupled Model Intercomparison Project Phase 6 (CMIP6) for the year
2050 (2040-2060).

MaxEnt model simulation

To simulate the model, the maximum entropy (MaxEnt) ver 3.4.3. algorithm was employed
[16]. The software is open source with high accuracy, and it can work with as little as 20
occurrences to predict the distribution of species.

57
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Determination of suitable and unsuitable areas

Output maps from the MaxEnt software in .asc format were uploaded into QGIS together with
the country shapefile downloaded from diva-gis. Uploaded layers under symbology were
changed from single band Gray to single band pseudo-colour in-which turbo colour ramp was
selected to generate colourful continuous maps. To generate binary maps, linear interpolation
was changed to discrete. The raster calculator in QGIS was used to assign the binary values
(classes) of 1 and 0 to the variables.

RESULTS

Model and variable performances

The model had an excellent performance with an accuracy value of 0.965 and a standard
deviation of 0.041. The bioclimatic variables which reduced the model gain the most when
excluded was BIO3 (isothermality, Figure 1).

Figure 1. Model and variable performance based on area under curve (AUC) of the receiver
operating curve (ROC), omission rate and Jackknife tests

Groundnut phytopathogens under current and future climatic scenarios

Under current climatic conditions, the ecological niche model (ENM) revealed that, the
unsuitable area (unsuitable = 0%) was far less than the total suitable area (suitable > 90%). The
analysis based on projecting the final model into the future climate scenarios (2050) revealed
no predicted increase of unsuitable areas for groundnut phytopathogens under low carbon
concentration (Figure 2).

58
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 2. Suitable areas for groundnut phytopathogens in Zimbabwe under current climate and
future scenario (SSP1-2.6)

DISCUSSION

The current and future distribution of groundnut phytopathogens in Zimbabwe could be

attributed to several factors including: (i) host distribution and susceptibility, (ii) pathogen
evolution, (iii) vector distribution and life cycle, (iv) management and farming systems, and (v)
government policy.

Host distribution and susceptibility

The distribution of groundnut phytopathogens under current and future climates is likely to be
influenced by the distribution of the hosts and their susceptibility. The study concurred with the
findings reported in Kenya by Mabele, Were (13) that, legumes including groundnut and non-
legume crops were susceptible to GRV.

Management and farming systems

Other elements which are likely to propel the distribution of groundnut phytopathogens in
Zimbabwe are management and farming systems currently used by farmers. Commonest
farming systems such as crop rotations, intercropping and mulching may be inappropriate in
the future as they can serve as vehicles for spreading disease inoculum [17].

59
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

CONCLUSION

The Maxent algorithm successfully predicted the current and future distribution of groundnut
phytopathogens under the current and future climates in Zimbabwe. Among those
environmental factors, isothermality, precipitation and temperature seasonality emerged as the
top factors associated with groundnut phytopathogens distribution. Therefore, current and
future climatic conditions helped to determine the potential range of the phytopathogens and
hotspot areas where proactive measures may be necessary to prevent severe yield losses

REFERENCES

1. Kumar V, Thirumalaisamy P. Diseases of groundnut. New Delhi: Indian Phytopathological

Society, Today, Tomorrow’s Printers Publishers; 2016. 459-87 p.
2. Prabhu R, Manivannan N, Mothilal A, Ibrahim SM. Artificial screening for foliar fungal
diseases in groundnut (Arachis hypogaea L.). Adv Life Sci. 2016;5(14):5634-8.
3. Appiah AS, Offei SK, Tegg RS, Wilson CR. Varietal response to groundnut rosette disease
and the first report of Groundnut ringspot virus in Ghana. Plant Dis. 2016;100(5):946-52.
4. Sudini H, Upadhyaya HD, Reddy SV, Naga Mangala U, Rathore A, Vijay Krishna Kumar
K. Resistance to late leaf spot and rust diseases in ICRISAT’s mini core collection of
peanut (Arachis hypogaea L.). Australas Plant Pathol. 2015;44:557-66.
5. Kumar N, Ajay BC, Jadon KS, Gangadhara K, Kona P, Rathanakumar AL, et al. Genetic
variability for Alternaria leaf blight resistance in groundnut genotypes. 2020.
6. Amoah RA, Akromah R, Asibuo JY, Oppong A, Nyadanu D, Agyeman A, et al. Genetic
control of resistance to rosette virus disease in groundnut (Arachis hypogaea L.). Journal
of Plant Breeding and Crop Science. 2016;8(6):87-93.
7. Gaikpa DS, Akromah R, Asibuo JY, Appiah-Kubi Z, Nyadanu D. Evaluation of yield and
yield components of groundnut genotypes under Cercospora leaf spots disease pressure.
International Journal of Agronomy and Agricultural Research. 2015;7(3):66-75.
8. Nutsugah SK, Oti-Boateng C, Tsigbey FK, Brandenburg RL. Assessment of yield losses due
to early and late leaf spots of groundnut (Arachis hypogaea L.). Ghana Journal of
Agricultural Science. 2007;40(1):21-7.
9. Mau YS, Ndiwa ASS. Field evaluation of late leaf spot and leaf rust resistance and the
associated yield losses in Indonesian groundnut genotypes. Asian J Plant Sci.
2018;17(3):134-41.
10. Katema T, Hanyani-Mlambo EMBT, Gomera MR, Chamboko T. An Analysis of the
Profitability of Groundnut Production by Small-holder Farmers in Chegutu District,
Zimbabwe. J Econ Sustain Dev. 2017;8:167-75.
11. Allahyari MS, Sadeghzadeh M. Agricultural extension systems toward SDGs 2030: zero
hunger. Zero hunger. 2020:41-52.
12. Mollier L, Seyler F, Chotte J-L, Ringler C. End hunger, achieve food security and
improved nutrition and promote sustainable agriculture: SDG 2. A Guide to SDG
Interactions: From Science to Implementation Paris, France: ICSU; 2017. p. 34-80.
13. Mabele AS, Were HK, Ndong’a MF, Mukoye B. Distribution, molecular detection and
host range of groundnut rosette assistor virus in western Kenya. Journal of Plant Sciences.
2019;7(5):100-5.
14. van Niekerk J, Wynberg R. Traditional seed and exchange systems cement social
relations and provide a safety net: A case study from KwaZulu-Natal, South Africa.
Agroecol Sustain Food Syst. 2017;41(9-10):1099-123.

60
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

15. Kearney MR, Wintle BA, Porter WP. Correlative and mechanistic models of species
distribution provide congruent forecasts under climate change. Conservation letters.
2010;3(3):203-13.
16. Phillips SJ, Dudík M. Modeling of species distributions with Maxent: new extensions and
a comprehensive evaluation. Ecography. 2008;31(2):161-75.
17. Juroszek P, Von Tiedemann A. Potential strategies and future requirements for plant
disease management under a changing climate. Plant Pathol. 2011;60(1):100-12.

61
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

DECISION SUPPORT SYSTEMS, SOFTWARE AND MOBILE APPLICATIONS

62
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

EMPOWERING FARMERS TO SUSTAINABLE AGRICULTURAL

PRODUCTIVITY IN WEST AFRICA: FESERWAM, A DIGITAL ADVISOR FOR
FARMERS
#11359

J. Toviho, I. Brou, K. Kouassi

International Fertilizer Development Center, EnGRAIS project, Ghana
e-mail: [email protected]; (229) 97 27 94 71

ABSTRACT

Smallholder farmers in West Africa frequently face challenges in accessing critical information
about agricultural inputs and practices. To address this issue, EnGRAIS/IFDC and CORAF,
with funding from USAID, developed the Fertilizer and Seed Recommendations Map for West
Africa (FeSeRWAM) mobile application. FeSeRWAM App is a digital agricultural tool for
West Africa, offering tailored Agricultural Input Packages (AIP) with recommendations on
fertilizers, seeds and good agricultural practices based on country specific agroecological zones.
Built on the FeSeRWAM web platform developed by IFDC and CORAF with USAID funding
and technical inputs from National Agricultural Research Structures, the app provides guidance
for over 650 AIP on 21 crops, 578 varieties and 62 fertilizer grades for 15 ECOWAS countries,
as well as Mauritania and Chad. Designed for ease of use, the app allows farmers, extension
agents, and agrodealers to access critical information offline and on the go. The app’s success
is attributed to its regional customization and the collaborative effort of over 350 stakeholders.
Additionally, it aligns with the ECOWAS Regional Agricultural Input Strategy. EnGRAIS has
played a pivotal role in this initiative by training 50 regional trainers and 5,000 national trainers,
reaching a total of 600,000 farmers to promote FeSeRWAM application. Moving forward, the
project plans to further integrate a decision support tool (DST) based on economic returns for
smart AIPs selection, to continue boosting agricultural productivity in the region.

Keywords: FeSeRWAM, Fertilizers, Seeds, Good agricultural practices, Agroecological

zones.

INTRODUCTION

Agriculture remains a cornerstone of the West African economy, providing livelihoods for
millions and contributing significantly to the region's GDP. Many smallholder farmers in West
Africa continue to produce below capacity due to their inability to access the the right
information on appropriate agricultural inputs and practices to unlock existing potential, make
better decisions and get more dividends on their investments. Various efforts have been made
over the years by value chain actors to change the trend, but not many of these efforts have
yielded the expected results. Ideally, extension agents are supposed to provide this information
to farmers. However, there are not enough extension agents, and those available often lack
access to the right information on farm inputs. Meanwhile, the information available in research
centers is not disseminated. In this condition, farmers often get their information from
agrodealers and more educated people in their communities, but this actor is also struggle to
find the correct information.

To address these issues and boost agricultural productivity sustainably, technological

innovations are increasingly recognized as essential. Among these innovations is FeSeRWAM.
FeSeRWAM, an acronym for the Fertilizer and Seed Recommendations Map for West Africa,
63
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

is a comprehensive digital app developed to support farmers in West Africa. This digital tool
aims to empower farmers by providing site-specific recommendations on seeds, fertilizers, and
best agricultural practices, thereby promoting sustainable agricultural productivity in the
region.

MATERIALS AND METHODS

The development of FeSeRWAM involved a collaborative effort between various stakeholders,

including the International Fertilizer Development Center (IFDC) and the West and Central
African Council for Agricultural Research and Development (CORAF). Funded by USAID
through the Feed the Future Enhancing Growth through Regional Agricultural Input Systems
(EnGRAIS) and Partnership for Agricultural Research, Education, and Development
(PAIRED) projects, FeSeRWAM App builds on the FeSeRWAM website's is ton and tons of
agronomic data. It utilizes big data technologies, is mobile, is in the hands of the user, accessible
instantaneously to deliver site-specific recommendations. The application integrates data from
over 350 stakeholders across national and regional organizations. It provides recommendations
for more than 650 agro-input packages (AIPs) on 21 crops, and 578 seed varieties and 62
fertilizer grades for 15 ECOWAS countries as well as Mauritania and Chad. The tool is
designed to be user-friendly, enabling farmers, extension agents, and agro-input dealers to
access critical information easily.

• Technological Infrastructure: It is accessible via mobile applications on Android

(feserwam – Applications Android sur Google Play), and iOS for iPhone and iPad
(FeSeRWAM on the App Store (apple.com) devices. The entire system is hosted on a
cloud platform, allowing for scalable processing power, data storage, and real-time
analytics.
64
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

• Data sources and analysis: It utilizes big data technologies to process vast amounts of
agronomic data to generate accurate and localized recommendations. The information
presented on the FeSeRWAM mobile app is sourced directly from the agricultural
research institute's databases.
• Training Materials and support: Educational resources, including user manuals,
video tutorials, and community workshops, were developed to ensure effective use of
FeSeRWAM by farmers. These materials address both technological and agricultural
literacy. Farmers receive training on using FeSeRWAM through workshops and online
resources. Ongoing support is provided via a dedicated helpdesk and community
forums.
• User Interaction: Farmers interact with FeSeRWAM through a user-friendly interface
on their mobile devices. The app provides personalized advice based on the specific
needs of their farms.
• Cost: The mobile app requires a smartphone and an internet connection to use.
Currently, it is free for users because it was developed as part of a project.

RESULTS AND DISCUSSION

Increased capacity of extension services

The FeSeRWAM app plays a key role in disseminating knowledge and best practices by
extending the reach of agricultural extension services. The mobile application's guidance on
good agricultural practices and nutrient management are particularly valued. FeSeRWAM also
promotes sustainable agricultural practices. The application facilitates better resource
management, reducing the environmental impact of farming activities.

Enhanced Productivity
Since its launch and its implementation, FeSeRWAM has demonstrated significant
improvements in agricultural productivity in West Africa. The app has enabled over 600,000
farmers to access vital information on improved seeds and appropriate fertilizers, leading to
increased crop yields and enhanced food security. By providing specific recommendations,
FeSeRWAM helps farmers make informed decisions, optimize input use, and adopt sustainable
agricultural practices.

The FeSeRWAM mobile application has shown promising potential to enhance agricultural
productivity through improved advice and guidance. While it may be premature to claim
definitive yield increases directly attributable to the app, early indications suggest farmers who
have adopted the recommended practices are experiencing positive outcomes.

For instance, in collaboration with the Institut Togolais de Recherches Agronomiques (ITRA)
and the Institut de Conseil et d'Appui Technique (ICAT), the EnGRAIS project trained many
extension agents to support farmers in using the FeSeRWAM app and adopting the
recommended agricultural best practices. Bakoundi Ayékénam Nadège, a rice and maize farmer
who is part of a 19-member cooperative at Akaglakopé, participated in one of these training
programs covering soil preparation, sowing, fertilizer application, and other good practices. She
reported that, “after adopting these practices, her cooperative's rice yields doubled from one-
and-a-half tons per hectare to three metric tons per hectare.”

Economic Impact
FeSeRWAM has shown positive economic potential for farming communities. There is a direct
cause between using the app and increased incomes has not yet been conclusively demonstrated,
65
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

the improved agricultural practices enabled by FeSeRWAM can reasonably be expected to lead
to enhanced productivity and farm revenues.

DISCUSSION

Empowering Farmers
FeSeRWAM represents a significant advancement in the digitalization of agriculture in West
Africa by empowering farmers with timely and relevant information. By leveraging technology
to provide tailored agricultural advice, the App addresses key challenges faced by farmers, such
as low yields and limited access to quality inputs. The FeSeRWAM is designed to be accessible
to all farmers including women and youth. The FeSeRWAM is especially friendly to women
and youth users. The success of FeSeRWAM underscores the importance of digital tools in
enhancing agricultural productivity and sustainability. However, the FeSeRWAM’s
effectiveness depends on regular updates. Ensuring that farmers are aware of and can access
FeSeRWAM is crucial for its sustained impact. The innovative training programs and
awareness campaigns can help increase the adoption of the tool among farmers and other
stakeholders. Moreover, expanding the app’s capabilities to include more crops and regions can
further enhance its utility. Collaboration with local agricultural organizations and continuous
feedback from users will be essential in refining and improving FeSeRWAM.

Addressing Challenges

While FeSeRWAM has shown considerable benefits, challenges remain. Accessibility to

technology in remote areas, variability in digital literacy, and infrastructure limitations are
notable hurdles. Addressing these challenges requires continued investment in digital
infrastructure, targeted training programs, and support for low-tech solutions.

Prospects
Looking ahead, FeSeRWAM can be enhanced with additional features such as agricultural-
based technologies that improve Nutrient Use Efficiency (NUE) and integration of a decision
support tool (DST) based on economic returns for smart AIPs selection. Expanding its reach
and functionality could further bolster agricultural sustainability and productivity in the region.

REFERENCES

IFDC, 2024. Enhancing Agricultural Productivity in West Africa by Empowering Women

Aggregators. Empowering Women Aggregators through Training - IFDC
IFDC, 2022. IFDC and CORAF Release Updated Online Tool to Promote Use of Innovative
Technologies for West African Farmers to Increase Agricultural Productivity and Profits.
IFDC and CORAF Release Updated Online Tool to Promote Use of Innovative
Technologies for West African Farmers to Increase Agricultural Productivity and Profits
- IFDC
IFDC, 2021. EnGRAIS and PAIRED Validate Strategy and Action Plan for FeSeRWAM and
AIP’s Dissemination. EnGRAIS and PAIRED Validate Strategy and Action Plan for
FeSeRWAM and AIP’s Dissemination - IFDC
IFDC, 2020. FeSeRWAM, a New Digital for Agriculture (D4AG) Tool Launched.
FeSeRWAM, a New Digital for Agriculture (D4AG) Tool Launched - IFDC

66
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

A FRAMEWORK FOR A CROP YIELD PREDICTION MODEL BASED ON

DECISION TREE AND MIN-MAX SCALING

S.A. Audu, A.A. Rabiu, G.I.O. Aimufua, M. Samaila, A.M. Umar, and M.A. Yau
1,2,3,5,6Department
of Computer Science, Nasarawa State University, Keffi, Nigeria
4Department of Computer Science, Federal University Gusau, Zamfara, Nigeria

e-mail: [email protected]; +234 8065670727

ABSTRACT

The prediction of crop yield has become critical for enhancing global food security. This study
presents a framework for agricultural yield prediction, employing a decision tree-based
mathematical model integrated with Min-Max scaling. A dataset comprising 28,241 entries was
collected from the Food and Agriculture Organisation (FAO) and the World Data Bank (WDB).
Variables include nation, crop type, year, rainfall, pesticide usage, and temperature. The model
achieved an accuracy of 97.8%, demonstrating that crop variety significantly influences
agricultural output, while temperature and pesticide usage impact yield more than rainfall. The
findings provide actionable insights into optimizing agricultural productivity by identifying key
influencing factors. This research offers a robust decision-making tool for stakeholders aiming
to enhance food security and agricultural efficiency.

Keywords: Crop yield, decision tree, prediction framework, Min-Max scaling, agricultural
productivity.

INTRODUCTION

The agricultural sector faces numerous challenges, including climate variability, environmental
degradation, labor costs, resource limitations, and conflicts such as herdsmen-farmer clashes
and banditry. Additionally, the need to sustain agricultural practices while preserving the
environment complicates efforts to meet food demands for growing populations. These
challenges underscore the importance of technological innovations for improving crop
productivity without compromising quality (Mittal et al., 2020).

Crop yield forecasting is vital for achieving food security. It supports decision-making for
farmers, industries, and governments by estimating production levels based on various factors,
such as soil properties, fertilizer use, irrigation management, and climatic variables like
temperature and rainfall (Khaki et al., 2020). Optimizing these factors, coupled with effective
policies, can significantly enhance agricultural productivity.

This study focuses on yield management, which integrates all agricultural processes into the
final productivity outcome. By leveraging predictive modeling, stakeholders can improve
agricultural efficiency, meet growing food demands, and inform government decisions
regarding food imports (Juvanna et al., 2021). This research aims to develop a framework for
crop yield prediction, emphasizing the role of climatic and agronomic variables. Beyond a
shadow of a doubt, predicting crop yield is another way of increasing the productivity of
agricultural products to meet the growing demand for food and to advise the government on the
amount of food to be imported based on the estimated yield as outlined in the study by (Juvanna
et al., 2021)

67
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

MATERIALS AND METHODS

The crop yield prediction framework integrates climatic, agronomic, and other variables (Figure
1). Key variables include temperature, precipitation, seed variety, soil water content, and soil
fertility rate.

Figure 1. Crop Yield Prediction Framework.

Climatic Factors
Temperature and precipitation are pivotal to crop yield. Studies reveal that extreme climatic
conditions, such as high temperatures and limited precipitation, adversely affect agricultural
productivity (Beillouin et al., 2020; Guo et al., 2021).

Agronomic Factors
Pesticides play a critical role in mitigating crop losses. Research by Tudi et al. (2021) highlights
that pesticide use prevents substantial losses in global fruit, vegetable, and grain production.
Thus, pesticides significantly contribute to higher agricultural yields by minimizing the impact
of pests and diseases.

Other Variables
Additional factors include the type of crop, the country of cultivation, and historical yield data.
These variables are critical inputs for the proposed framework.

Decision Tree-Based Crop Yield Model Development

The proposed system architecture (Figure 2) outlines the stages from data collection to model
training. Data preprocessing includes normalization using Min-Max scaling to ensure
uniformity. The normalized dataset is split into training and testing subsets, with the decision
tree (DT) regressor trained on the former. The model predicts crop yield using the trained DT
algorithm.

68
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 2. System Architecture for Decision Tree based Crop Prediction.

RESULTS AND DISCUSSION

The model's performance was evaluated using training and validation scores, which
demonstrated its accuracy and generalization capability (Figure 3). Variable importance was
assessed using the regressor feature_importances_ function, revealing that crop type is the most
significant predictor of yield. Temperature and pesticide usage follow closely, with rainfall
contributing less than expected (Figure 4).

Figure 3. The Validation and the Learning Curve

69
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 4. Bar Chart of order of Importance of the Features.

The results underscore the importance of integrating agronomic and climatic factors in
predictive models. The high accuracy (97.8%) indicates the model's potential as a decision-
support tool for policymakers and farmers to enhance agricultural productivity.

CONCLUSION

This study developed a decision tree-based crop yield prediction framework incorporating Min-
Max scaling. The findings highlight the dominant role of crop variety, temperature, and
pesticide usage in determining yield. The framework provides a reliable tool for stakeholders
to optimize agricultural output and address food security challenges. Future research could
explore integrating additional variables, such as soil texture and socio-economic factors, to
further refine predictive accuracy.

REFERENCES

Beillouin, D.; Schauberger, B.; Bastos, A.; Ciais, P.; and Makowski, D. (2020). Impact of
extreme weather conditions on European crop production in 2018: Random forest - Yield
anomalies. Philosophical Transactions of the Royal Society B: Biological Sciences,
375(1810). https://doi.org/10.1098/rstb.2019.0510
Cunha, R. L. D. F.; and Silva, B. (2020). Estimating Crop Yields with Remote Sensing and
Deep Learning. 2020 IEEE Latin American GRSS and ISPRS Remote Sensing
Conference, LAGIRS 2020 - Proceedings, 273–278.
https://doi.org/10.1109/LAGIRS48042.2020.9165608
Guo, Y. ; Xiang, H. ; Li, Z. ; Ma, F. ; and Du, C. (2021). Prediction of rice yield in east China
based on climate and agronomic traits data using artificial neural networks and partial
least squares regression. Agronomy, 11(2). https://doi.org/10.3390/agronomy11020282
Juvanna, I.; Yuvesh, Balaji V. G.; Sri Raam, M. A.; and Karthikeyan, T. (2021). A Novel
Framework Crop Yield Prediction Using Data Mining. Turkish Journal of Computer and
Mathematics Education, 12(10), 350–356.
70
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Kavitha, M.; Anusha, K.; Santhi, M. V. B. T.; Rajesh, N.; and Choragudi, F. (2020). Predicting
Crop Yield using Neural Network with Optimal. 32(2), 3028–3036.
Khaki, S.; Wang, L.; and Archontoulis, S. V. (2020). A CNN-RNN Framework for Crop Yield
Prediction. Frontiers in Plant Science, 10(January), 1–14.
https://doi.org/10.3389/fpls.2019.01750
Martinez-Ferrer, L.; Piles, M.; and Camps-Valls, G. (2020). Crop Yield Estimation and
Interpretability With Gaussian Processes. IEEE Geoscience and Remote Sensing Letters,
XX(1), 1–5. https://doi.org/10.1109/lgrs.2020.3016140
Mittal, D.; Kaur, G.; Singh, P.; Yadav, K.; and Ali, S. A. (2020). Nanoparticle-Based
Sustainable Agriculture and Food Science: Recent Advances and Future Outlook.
Frontiers in Nanotechnology, 2(December). https://doi.org/10.3389/fnano.2020.579954
Nishant P. S.; Venkat P.S,; Avinash B. L.; and Jabber B., ”Crop Yield Prediction based on
Indian Agriculture using Machine Learning,” 2020 International Conference for
Emerging Technology (INCET), 2020, pp. 1-4, DOI:
10.1109/INCET49848.2020.9154036
Nosratabadi, S.; Ardabili, S.; Lakner, Z.; Mako, C.; and Mosavi, A. (2021). Prediction of food
production using machine learning algorithms of multilayer perceptron and anfis.
Agriculture (Switzerland), 11(5) pp. 1-13. https://doi.org/10.3390/agriculture11050408
Tudi, M.; Ruan, H. D.; Wang, L.; Lyu, J.; Sadler, R.; Connell, D.; and Chu, C. (2021).
Agriculture Development, Pesticide Application and Its Impact on the Environment. PP.
1–23.

71
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

EDUCATION AND ENGAGEMENT IN PRECISION AGRICULTURE

72
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

INTER-COUNTRY COOPERATION CAN TRANSFORM PRECISION

AGRICULTURE EDUCATION AND RESEARCH IN AFRICA
#11628

Kwame Agyei Frimpong

African Plant Nutrition Institute (APNI)
e-mail: [email protected]

ABSTRACT

The productivity and profitability of sub-Saharan (SSA) agriculture can be greatly enhanced by
the adoption of precision agriculture technologies and tools. However, until 2020 when the
African Plant Nutrition Institute (APNI) established the African Association for Precision
Agriculture (AAPA), most SSA PA enthusiasts worked in isolation. The AAPA was formed to
innovate Africa’s agricultural industry by connecting PA science to its practice and
disseminating PA tailored to the needs and aspirations of the African people. This paper
highlights the leverage that intra-Africa cooperation such as the AAPA offers to transform PA
education, research, and outreach in Africa. The paper further presents activities and milestones
that the AAPA has chalked in the last two years. So far, the AAPA has provided opportunities
for members to engage in research, extension, education, and training activities to promote the
advancement and dissemination of PA tailored to the needs and aspirations of the African
people and the furtherance of international collaborations. These opportunities include a Global
Challenge Research Fund networking forum, and EU funding to promote leadership skills
development, advocacy, and engagement among five African universities and a European
partner. In conclusion, intra-Africa cooperation through AAPA has catalyzed the enthusiasm
and willingness of universities in African countries to work together to develop and scale
precision agriculture (PA) education, research, and outreach to reduce skill gaps and tackle food
and nutritional security challenges in African countries.

INTRODUCTION

Precision agriculture (PA) has gained considerable recognition as a farm management strategy
to improve resource economy, farm productivity, and profitability to enhance the sustainability
of agricultural production systems (Wang et al., 2023; Ofori and El-Gayar, 2021). However,
the application of PA technologies is yet to be widely adopted in smallholder farms in sub-
Saharan Africa (SA). The World Bank (2021) estimated that agriculture employs 52% of the
SSA adult population and contributes to about 17% of the region’s gross domestic product
(GDP). Despite the critical contribution of the agricultural sector to the economic development
of the continent, the sector is inherently characterized by low productivity and low income for
farmers (Abay et al., 2023; Marinus et al., 2022), and so the agriculture industry is unattractive
for the youth. The burgeoning SSA population growth (World Bank, 2022b), and the impacts
of climate change make it imperative for the adoption of innovative agricultural technologies
to meet food and nutrition demands in the region. Adoption of precision agriculture is a
promising pathway to transform agricultural productivity and profitability in SSA and
incentivize the youth in the region to embrace agricultural production as a profitable livelihood
option. Nyaga et al. (2021) contended that the low adoption rate of PA in SSA could be
attributed to the lack of awareness and information among smallholder farmers and other
agriculture stakeholders in SSA.

73
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Therefore, there is a need to equip farmers, extension agents, and the youth in SSA with the
relevant information and knowledge and provide the necessary tools to drive PA use for
sustainable agriculture in the continent. Studies have been done on the potential for PA to
improve farm productivity and profitability (Wang et al., 2023) adoption decisions (John et al.,
2023; DeLay et al., 2022; Lowenberg-DeBoer and Erickson, 2019), the opportunities and
challenges of precision agriculture for farmers (Khanna et al., 2022; Bosompem, 2021; Nyaga
et al., 2021; Ofori and El-Gayar, 2021) and the role of data management for the deployment of
precision agriculture (Gobezie and Biswas, 2023). However, there is a paucity of information
about how African countries can leverage the power of collaboration and networking to
transform agricultural production in their countries through inter-country collaboration. The
objective of this paper was to highlight the potential of inter-country cooperation to transform
precision agriculture education and research in Africa. Specifically, this paper provides
practical examples of inter-country collaboration in the region, which have amply demonstrated
that these collaborations can foster co-development and co-implementation of PA research,
education, and outreach actions to increase awareness, motivate, and build the capacity of
farmers, extension agents and students in the region.

MATERIALS AND METHODS

The paper presents an overview of three examples of inter-country collaborations – The African
Association for Precision Agriculture (AAPA), the Global Challenge Research Fund (GCRF)
networking forum, and the EU-funded ‘Capacity Building of African Young Scientists in
Precision Agriculture Through Cross-Regional Academic Mobility for Enhanced Climate-
Smart Agri-Food System’ (PATH) project, which have provided opportunities for faculty,
postgraduate students, and staff of some African universities to engage in advocacy and
networking fora to promote the advancement and dissemination of PA tailored to the needs and
aspirations of the African people. The paper highlights some successes that the AAPA has
achieved since its formation in 2020. The overview is supported by information from published
and grey literature.

RESULTS AND DISCUSSION

The inter-country collaborations among African countries within the AAPA is a typical
example of how such cooperation can promote PA education, research and outreach in the
continent. The AAPA was formed by the African Plant Nutrition Institute (APNI) in 2020. The
AAPA boasts of a membership of 628 registered members from 51 countries. The objectives,
activities and board composition of the AAPPA are summarised in Table 1.

The AAPA have organised webinars and published newsletters, as well as supported the
organization of the African Conference for Precision Agriculture in 2020 and 2022. Another
example of inter-country collaboration is the GCRF-funded networking forum that was held
at the University of Cape Coast in Ghana. The forum was a collaborative project between the
Harper Adams University of the United Kingdom, the University of Cape Coast, and the
University for Development Studies, both in Ghana, the University of Abomey Calavi in
Benin, and the Institute of Agricultural Research and Training in Nigeria. The forum attracted
participants from the Ministry of Food and Agriculture, Ghana, Non-governmental
organizations, and agro-input dealers. An important outcome of the networking forum is the
EU-funded PATH. Project, which is a collaborative project among four universities from
West, East, and Southern Africa, a northern African associate partner, and a European
technical partner, respectively, to promote leadership skills development, advocacy,
74
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

postgraduate education, and research. The partnering institutions, objectives, and target crops
of the PATH project are presented in Table 2.

Table 1. Objectives, activities and board membership of the AAPA

Objectives Activities Board membership

The AAPA provides opportunities Publishes a quarterly newsletter; • President (Nigeria)

for members to engage in leads professional community • Past President (Ghana)
research, extension, education, within the International Society • President-elect (Benin):
and training activities to promote of Precision Agriculture (ISPA); • Executive secretary
the advancement and and organizes workshops and (Morocco)
• Regional
dissemination of PA tailored to symposiums at the African
Representatives North
the needs and aspirations of the Conference on Precision Africa (Tunisia), West
African people; the furtherance of Agriculture (AfCPA), and the Africa (Togo), Central
international collaborations; International Conference on Africa (Cameroon),
leadership skills development; Precision Agriculture (ICPA). East Africa (Tanzania),
advocacy; and engagement with AAPA also organizes and hosts and Southern Africa
policymakers toward creating the webinars on various topics in (Zimbabwe):
• Founder: Steve Phillips
enabling environment needed to precision agriculture.
(APNI)
advance the knowledge of the
science and practice of PA in
Africa.

Table 2. Capacity Building of African Young Scientists in Precision Agriculture Through

Cross-Regional Academic Mobility for Enhanced Climate-Smart Agri-Food System (PATH)
Project
Objectives Universities involved Countries Target crops
involved
• Training 32 MSc and 12 PhD • University of • Benin (lead) • soybean, rice, tomato
African scholars in Precision Abomey Calavi • Ghana (Co- and pineapple
Agriculture (PA) to upgrade • University of Cape lead) between Ghana
the continent's capability Coast • Rwanda (Co- and Benin
• Building the capacity of 10 • University of lead) • groundnut between
young African trainees and Rwanda • Eswatini (Co- Benin and Eswatini
10 staff in PA and • University of lead) • rice and potato
entrepreneurship Eswatini • Morocco between Benin and
• Improving PAAC and • University of (Associate Rwanda
ICT4Ag curricula and Mohammed VI partner) • beans between
research at the participating Polytechnic • France Eswatini and
African Higher • Laboratoire Univers (Technical Rwanda
• Education Institutions (HEIs) et Particules de partner) • taro and sweet potato
to address more efficiently Montpellier between Ghana
the current challenges of and Eswatini
agriculture and climate • sorghum and taro
change between Ghana
• Developing a network of and Rwanda
HEIs in Africa involved in
PAAC research and training

75
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Potential benefits from Intra-Africa collaboration

African countries can harness available skills, expertise, and infrastructure in their countries to
drive the PA agenda through knowledge sharing, and a culture of collaborative learning. Inter-
country cooperation can also foster mutual deployment of digital technologies to enhance
agricultural productivity, increase access to markets and finance, and improve food security.
An analysis of the Baobab Insights companies’ 2015 database showed that more than 40
innovative start-ups are established in Africa that provide various precision farming
technologies and services with t 50% of these companies based in Nigeria, Kenya and South
Africa. Universities, and other agricultural organizations in Africa can pool resources,
coordinate their research efforts to avoid duplication of efforts and harmonize common research
actions, and develop comprehensive training programs for farmers, extension workers, and
students. The “Digital Transformation Strategy for Africa (2020-2030) of the African Union
outlines the vision and objectives for Africa to leverage digital technologies for economic and
social development (African Union, 2020). In June 2021, a high-level dialogue among Head of
States of African countries on Feeding Africa pointed out the need for technology-led
agriculture that can increase crop productivity, and improve product quality including post-
harvest losses reduction while mitigating the negative effects of climate changes (drought,
flooding, pest outbreak etc.

Challenges and the way forward

Intra-Africa cooperation in PA education, research and outreach is constrained by the lack of
PA expertise, tools, and facilities in SSA education and research institutions, Further, there are
limited financial commitments and enabling policies towards integrating PA into ongoing
country-specific agricultural development programmes. Other challenges that inter-country
cooperation face include the high costs of mobile technology and internet, which is exacerbated
by unreliable electricity supply, low ICT literacy levels, and lack of financial resources to secure
the use of ICTs.

CONCLUSION

Inter-country cooperation is crucial to support the farmers and other agriculture stakeholders in
SSA. These cooperations can promote co-learning and sharing of scientific, technical, and
market information. Thus, it is imperative to link universities and other agricultural institutions
in Africa to tackle knowledge and skills gaps, while facilitating participatory research, and
enhancing results dissemination. Inter-country cooperation can stimulate investment in digital
infrastructure, technical capacity building, and policy support to unlock the potential of PA to
improve the productivity and profitability of agriculture in Africa.

REFERENCES

1. Abay, K. A., Wossen, T., & Chamberlin, J. (2023). Mismeasurement and efficiency
estimates: Evidence from smallholder survey data in Africa. Journal of Agricultural
Economics, 74(2), 413-434.
2. Bosompem, M. (2021). Potential challenges to precision agriculture technologies
development in Ghana: Scientists’ and cocoa extension agents’ perspectives. Precision
Agriculture, 22(5), 1578-1600
3. DeLay, N. D., Thompson, N. M., & Mintert, J. R. (2022). Precision agriculture technology
adoption and technical efficiency. Journal of Agricultural Economics, 73(1), 195-219.

76
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

4. Khanna, M., Atallah, S. S., Kar, S., Sharma, B., Wu, L., Yu, C., ... & Guan, K. (2022). Digital
transformation for a sustainable agriculture in the United States: Opportunities and
challenges. Agricultural Economics, 53(6), 924-937.
5. Lowenberg-DeBoer, J. (2019). The economics of precision agriculture. In Precision
agriculture for sustainability (pp. 481-502). Burleigh Dodds Science Publishing
6. Marinus, W., Thuijsman, E. S., van Wijk, M. T., Descheemaeker, K., van de Ven, G. W.,
Vanlauwe, B., & Giller, K. E. (2022). What farm size sustains a living? Exploring future
options to attain a living income from smallholder farming in the East African
Highlands. Frontiers in Sustainable Food Systems, 5, 759105.
7. Nyaga, J. M., Onyango, C. M., Wetterlind, J., & Söderström, M. (2021). Precision
agriculture research in sub-Saharan Africa countries: A systematic map. Precision
Agriculture, 22, 1217-1236.
8. Ofori, M., & El-Gayar, O. (2021). Drivers and challenges of precision agriculture: a social
media perspective. Precision Agriculture, 22, 1019-1044.
9. Wang, T., Jin, H., Sieverding, H., Kumar, S., Miao, Y., Rao, X., ... & Cheye, S. (2023).
Understanding farmer views of precision agriculture profitability in the US
Midwest. Ecological Economics, 213, 107950.

77
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ADVANCING PRECISION AGRICULTURE EDUCATION IN SUB-SAHARAN

AFRICA: EXPLORING FACTORS FOR SUCCESS AND OBSTACLES
#12326

F.J. Adekoya, B.V. Bamidele, T.E. Adefare

Integrated Aerial Precision, Lagos, Nigeria
e-mail: [email protected]; 2348136163231

ABSTRACT

Precision Agriculture (PA) is a critical tool for addressing food security challenges, yet its
adoption in Sub-Saharan Africa (SSA) remains low. This study examines the role of education
in advancing PA adoption, focusing on barriers and opportunities for integrating PA into the
agricultural curriculum in SSA, with Nigerian institutions as a case study. A mixed-methods
approach was employed, involving 252 respondents comprising faculty, undergraduate, and
postgraduate students from tertiary institutions. Quantitative surveys assessed awareness,
familiarity, and factors influencing PA adoption, while qualitative interviews explored
curriculum integration and potential solutions.

Findings reveal a significant gap in PA education, with 57.71% of respondents reporting no

exposure to PA-related courses. Familiarity with basic tools such as GPS and drones was
moderate, while advanced technologies like robotics and telemetry were largely unfamiliar.
Key barriers include limited curriculum integration, inadequate digital infrastructure, and
insufficient faculty expertise.

The study underscores the importance of updating agricultural curricula to incorporate PA

concepts, enhancing digital infrastructure for practical learning, and providing faculty
development programs through workshops and certifications. These reforms are essential to
equip future agricultural professionals with the skills needed to adopt and implement PA,
thereby fostering a skilled workforce capable of leveraging advanced technologies for
sustainable agriculture. This research highlights the transformative potential of education in
overcoming barriers to PA adoption, contributing to enhanced agricultural productivity and
food security in SSA.

Keywords: Precision Agriculture, Sub-Saharan Africa, Agricultural Education, Technology

Adoption.

INTRODUCTION

Globally, food production is projected to increase by 50% by 2050 to meet the needs of the
rapidly growing population (FAO, 2020). Achieving this demand is unlikely without advanced
technologies to improve food security, as traditional methods alone may fall short. Precision
Agriculture (PA) has emerged as a vital approach to meet these challenges, focusing on
efficiency, sustainability, and productivity. It relies on advanced technologies, including
Geographic Information Systems (GIS), remote sensing, drone technology, and data analytics,
to improve crop production while minimizing environmental impact (Gebbers & Adamchuk,
2010).

While Precision Agriculture (PA) is gaining momentum in developed countries due to

advancements in technology and infrastructure (Fountas et al., 2020; Roberts et al., 2021), Sub-
78
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Saharan Africa (SSA) has been slower to adopt these innovations due to barriers such as limited
digital infrastructure, high costs, and a lack of technical expertise (Mabaya et al., 2022; Tsan et
al., 2019).

The education sector plays a crucial role in accelerating PA adoption by equipping farmers ,
agronomists, and extension workers with the skills necessary to implement these advanced
technologies. With a growing population and the urgent need for food security, particularly in
Sub-Saharan Africa (SSA), PA presents significant potential for improving agricultural
productivity in the region. However, the successful adoption and implementation of PA largely
depends on the level of education, awareness, and skill proficiency of the agricultural workforce
(Tsan et al., 2019).

Precision agriculture education when incorporated into the Education curricula could increase
the adoption of data-driven technologies and enhance the quality of agricultural education in
Sub-Saharan Africa (SSA). However, for SSA where agriculture is a key livelihood and food
security is a priority, this incorporation would mean that students enter the workforce well-
prepared to contribute to resilient, productive agricultural systems.

This study aims to access the level of knowledge about precision agriculture in the education
system in Sub-Saharan Africa, using university faculty, undergraduate, and postgraduate
students in Nigeria as a case study to examine the familiarity with PA tools and technologies,
and the factors and barriers influencing their willingness to adopt PA or become an expert in
PA tools and the extent of PA integration in the academic curriculum.

MATERIAL AND METHODOLOGY

This study which was conducted in Nigeria adopted a mixed-method approach to retrieving
data from penultimate year students, final-year students, postgraduate students, and lecturers of
the Faculty of Agriculture across tertiary institutions in Nigeria. Quantitative data were
gathered, followed by a qualitative interview. The questionnaire was structured to gather
information regarding their socio-economic characteristics, awareness and familiarity with
Precision agriculture tools, and factors influencing their willingness to adopt PA or become an
expert in PA tools. In addition, the qualitative interview contained questions regarding the
integration of PA into their educational curriculum as well as recommendations to the
government and educational institutions.

RESULT AND DISCUSSION

The discussion shows that most respondents are male (62.55%). This agrees with (Luka et al.,
2023; Omotosho et al., 2020) who conducted a survey and found higher male participation in
agricultural-related careers. A greater proportion of the respondents (62.55%) fall within the
age bracket of 18-30. This is similar to the findings of (Luka et al., 2023) who found the mean
age of agricultural students in Bauchi state to be 27 years indicating that their mean age is
between 18 -30 years.

Based on the proportion of respondents who had been taught or taken any course relating to
Precision Agriculture, most of the respondents (57.71%) had never taught or taken any course
relating to PA. This likely implies that most of the agricultural students in Nigeria have not
been exposed to the concept of precision agriculture via their institutions. This agrees with the
findings of Adepoju et al., 2022 and Nyaga et al., 2021 who found in their studies that a
79
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

significant gap exists in the learning and teaching of precision agriculture in tertiary institutions
in Nigeria as many students have not been exposed to courses relating to PA.

Table 1. Demographic characteristics of respondents.

Variables Categories Frequency Percentage

Sex Female 85 37.44
Male 142 62.56
Age 18-30 142 62.55
31-50 79 34.80
51-65 6 2.6
Min 18
Max 62
Category of the Finalist 62 27.31
Respondents
Lecturer/staff 42 18.50
Penultimate year 28 12.33
Post graduate Student 95 41.85
Taught or taken any No 131 57.71
course in Precision
Agriculture
Yes 96 42.29

Table 2. Level of familiarity with precision agriculture tools.

PA Tool Extremely Moderately Not Familiar

Familiar Familiar At All
Variable Rate Application 9 (3.96) 85 (37.44) 116 (51.10)
GPS/GNSS 9(3.96) 98(43.17) 85(37.44)
Yield and Soil Moisture Sensor 6 (2.64) 78(34.36) 120(52.86)
Drones 6(2.64) 98(43.17) 86(37.89)
GPS Tracker 9 (3.96) 92 (40.53) 80 (35.24)
Digital Fencing 6 (2.64) 81 (35.68) 121 (53.13)
Field Sensor 5 (2.26) 85 (37.44) 113 (49.78)
Data Analytics 7 (3.08) 105 (46.26) 86 (37.88)
Digital Soil maps 6 (2.64) 80 (35.24) 119 (52.42)
Telemetry Systems and Automation Tech 6 (2.64) 48 (21.15) 165 (72.69)
Machine Vision Tech for livestock 4 (1.76) 53 (23.35) 159 (70.04)
Automated Feeding System for Livestock 7 (3.08) 96 (42.29) 89 (39.21)
Robotic Milking Systems 4 (1.76) 7 (33.04) 131 (57.71)
Electronic Identification for Livestock 6 (2.64) 79 (34.80) 12.5 (55.07)

The survey reveals that the respondents are familiar with basic tools such as GPS, Drones, and
Data Analytics in Precision Agriculture and less familiar with advanced tools like Robotic
Milking systems, Telemetry systems, and automation technology

80
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Willingness to adopt precision agriculture tools

Performance expectancy (PE): For every one-unit increase in Performance Expectancy, the
behavioral intention to adopt PA increases by approximately 0.752 units. This strong positive
relationship suggests that respondents believe that adopting PA will significantly enhance their
performance, making this factor crucial in influencing their willingness to engage with PA
technologies. This finding is similar to the result of Eweoya et al., 2021, who conducted a
survey and found that performance expectancy is the most significant factor that influences the
adoption of e-agriculture in Nigeria. In addition, Lee et al. (2023) highlights that performance
expectancy significantly predicts professionals' intention to adopt precision agriculture
technologies. This emphasizes the importance of demonstrating clear and tangible benefits of
such technologies to potential users.

Effort Expectancy (EE): the coefficient of 0.214 indicates that EE positively influences
adoption, though its effect is smaller compared to PE. This suggests that when PA technologies
are perceived as easier to use, the students and lecturers are more likely to consider adoption (p
< .01) or become experts in PA technology. This agrees with Al-zboon et al., 2022, that effort
expectancy is positive and significant to the attitude of science and mathematics teachers
towards integrating ICT in their teaching activities.

Social influence: The result shows that SI, with a coefficient of 0.048, is not statistically
significant (p = .292), implying that social factors—such as recommendations from others—do
not have a strong impact on adoption intentions for Precision agriculture technology. This
agrees with Tey and Brindal (2012), that while social factors like recommendations can play a
role, they are typically weaker predictors of technology adoption in precision agriculture
compared to economic and technical factors, such as perceived financial benefits, ease of use,
and productivity improvements.

Facilitating Conditions: The coefficient of 0.248 for FC is significant, indicating that when
supportive resources or infrastructure are available, the likelihood of adoption or tendency to
become an expert by the respondents, increases (p < .01). This is similar to the findings of
Reichardt and Jurgens, (2009). In their research, Reichardt and Jürgens found that access to
supportive infrastructure, such as technical resources and financial support, significantly
enhances the adoption of precision agriculture technologies. Their findings highlight that when
farmers have the necessary resources and infrastructure, they are more likely to adopt precision
farming practices and improve their expertise over time.

Perceived Challenges: Having a negative coefficient of -0.217, PC significantly reduces the

willingness to adopt or become an expert in PA. This suggests that the more challenges (e.g.,
high costs, technical difficulties) individuals perceive, the less inclined they are to adopt PA (p
< .01). This agrees with Paustian, M., & Theuvsen, L. 2017. In their study, Paustian and
Theuvsen found that perceived challenges (e.g., high initial costs, complexity of use) were
significant barriers to adopting precision agriculture technologies. Their analysis suggests that
as perceived difficulties increase, farmers are less likely to adopt these technologies,
highlighting a negative association between perceived challenges and adoption intentions.

81
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Predictor Variables Coefficient Standard T value P value

Error
Performance Expectancy average 0.752*** 0.048 15.69 0.0000
Effort Expectancy average 0.214*** 0.051 4.18 0.000
Social Influence average 0.048 0.046 1.06 0.292
Facilitating Condition average 0.248*** 0.033 7.61 0.0000
Perceived Challenges average -0.217*** 0.041 -0.526 0.0000
Constant -0.995 0.277 -0.360 0.0000
Mean dep Variable 484.495
R- squared 1.000
F-test 3460540.518

Summary of the Qualitative Interview

Result of the qualitative interview shows the perceived benefits of precision agriculture as it
enhances speed and accuracy in farming, reduces the workload of the farmers and assists in
farm-data collection. The Finalists and Postgraduate students cited drone technology, GPS, and
sensors as the technology they are most familiar with. The respondents gave the high cost, and
gap in knowledge as the major barriers in the adoption of the PA tools. The respondents further
noted that the integration of PA into their educational curriculum was very low. In terms of the
respondent's willingness to adopt or become experts in Precision Agriculture, the respondents
noted that the provision of education and training to the younger generation will attract students
to agriculture. As stated by one of the respondents “I have a passion for training farmers. If I
gain more knowledge, I will be able to train others” shows a willingness to enhance their
knowledge and educate others too on precision agriculture.

CONCLUSION AND RECOMMENDATIONS

This study underscores the critical role of education in advancing Precision Agriculture (PA)
adoption in Sub-Saharan Africa, with Nigeria as a case study. The findings reveal a significant
gap in PA education, as many agricultural students lack exposure to PA concepts and tools due
to insufficient curriculum integration. Despite these challenges, there is growing awareness
among students and faculty about PA’s potential to transform farming practices and enhance
food security. The study highlights performance expectancy, effort expectancy, and facilitating
conditions as key drivers of PA adoption, while perceived challenges deter progress.

To address these gaps, educational institutions in SSA must prioritize updating agricultural
curricula to integrate PA concepts and tools. Providing students with practical training on
technologies such as drones, GPS systems, and data analytics is essential to preparing them for
the evolving demands of modern agriculture. Investment in digital infrastructure is critical to
ensure access to the necessary tools for hands-on learning.

Additionally, supporting faculty development through workshops, seminars, and certifications

will enhance their ability to teach PA effectively and bridge existing knowledge gaps. By
equipping both students and educators with the skills and resources required for PA adoption,
institutions can foster a skilled workforce capable of leveraging advanced technologies for
sustainable agriculture and improved food security.

82
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REFERENCES

Adepoju, O., Esan, O., and Akinyomi, O. (2022). Food Security in Nigeria: Enhancing
Workers’ Productivity in Precision Agriculture. Journal of Digital Food, Energy & Water
Systems. https://doi.org/10.36615/digital_food_energy_water_systems.v3i2.
Ajao, H., Alegbeleye, I., and Westfall-Rudd, D. (2022). Curriculum Design In An Agricultural
Education Program In Nigeria: Towards Advancing Career Readiness. Advancements in
Agricultural Development, 3(2), 17-30. https://doi.org/10.37433/aad.v3i2.159
Chaudhary, S., Patel, J., and Raval, K. (2023). Attitude And Aspiration Of Undergraduate
Agricultural Students. Gujarat Journal of Extension Education.
https://doi.org/10.56572/gjoee.2023.35.2.0007.
Ekea, U. U. (2019). Weed Science Education At The Tertiary Educational Level In Nigeria.
Agriculture, Forestry And Fisheries, 8(1), 27. https://doi.org/10.11648/j.aff.20190801.15
Eweoya, I., Okuboyejo, S., Odetunmibi, O., and Odusote, B. (2021). An empirical investigation
of acceptance, adoption and the use of E-agriculture in Nigeria. Heliyon, 7.
https://doi.org/10.1016/j.heliyon.2021.e07588.
Food and Agriculture Organization (FAO). (2020). The State of Food Security and Nutrition in
the World 2020.
Fountas, S., Mylonas, N., Malounas, I., Gemtos, T. A., & Blackmore, S. (2020). Precision
Agriculture Technology in Fruit Production: Status, Challenges and Future Trends.
Computers and Electronics in Agriculture, 170, 105309
Gebbers, R., & Adamchuk, V. I. (2010). Precision agriculture and food security. Science,
327(5967), 828-831.
Lee, C., Strong, R., Briers, G., Murphrey, T., Rajan, N., and Rampold, S. (2023). A
Correlational Study of Two U.S. State Extension Professionals’ Behavioral Intentions to
Improve Sustainable Food Chains through Precision Farming Practices. Foods, 12.
https://doi.org/10.3390/foods12112208.
Luka, J., Murtala, N., and Sani, M. H. (2023). Socio-Economic Factors Influencing Choice Of
Agribusiness As Career Option After Graduation Among Agricultural Students Of
Tertiary Institutions In Bauchi State, Nigeria. Nigerian Journal of Agriculture and
Agricultural Technology, 3(1), 85-96.
Mabaya, E., & Gahakwa, D. (2022). Precision Agriculture in Africa: Addressing Food Security
and Economic Transformation. African Journal of Science, Technology, Innovation and
Development, 14(1), 50-58.
Nyaga, J., Onyango, C., Wetterlind, J., and Söderström, M. (2021). Precision agriculture
research in sub-Saharan Africa countries: a systematic map. Precision Agriculture, 1-20.
https://doi.org/10.1007/S11119-020-09780-W.
Nguyen, L., Khuu, D., Halibas, A., and Nguyen, T. (2023). Factors That Influence the Intention
of Smallholder Rice Farmers to Adopt Cleaner Production Practices: An Empirical Study
of Precision Agriculture Adoption.. Evaluation review, 193841X231200775.
https://doi.org/10.1177/0193841X231200775.
Obi, B., and Oladokun, B. (2023). Security Measures in Meeting Information Needs and
Seeking Behaviour of Agricultural Science Students in Nigerian Universities. Southern
African Journal of Security. https://doi.org/10.25159/3005-4222/14383.
Omotosho, A., Emmanuel, A., Ayegba, P., and Ayoola, J. (2020). Impact of Agricultural
Education on Students' Career Choice: A Survey. Int. J. Emerg. Technol. Learn., 15, 51-
61. https://doi.org/10.3991/ijet.v15i03.11260.
Paustian, M., & Theuvsen, L. (2017). Adoption of precision agriculture technologies by
German crop farmers. Precision Agriculture, 18(5), 701–716.
https://doi.org/10.1007/s11119-016-9482-5
83
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Reichardt, M., & Jürgens, C. (2009). Adoption and future perspective of precision farming in
Germany: Results of several surveys among different agricultural target groups. Precision
Agriculture, 10(1), 73–94. https://doi.org/10.1007/s11119-008-9101-1
Roberts, M. J., Schimmelpfennig, D., & Livingston, M. (2021). Precision Agriculture Adoption
and Technical Efficiency. American Journal of Agricultural Economics, 103(2), 589–615
Tey, Y. S., and Brindal, M. (2012). Factors influencing the adoption of precision agricultural
technologies: A review for policy implications. Precision Agriculture, 13(6), 713–730.
https://doi.org/10.1007/s11119-012-9273-6
Tsan, M., Totapally, S., Hailu, M., & Addom, B. K. (2019). The Digitalisation of African
Agriculture Report 2018–2019. CTA. Retrieved from
https://cgspace.cgiar.org/handle/10568/101498
Yekinni, O., Ladigbolu, T., Adeniyi, R., and Oluwasusi, S. (2021). Information and
Communication Technologies Usage among Agricultural Students of Tertiary
Institutions in Southwest of Nigeria. The Journal of Agricultural Extension, 24, 1-10.
https://doi.org/10.4314/JAE.V25I1.1.

84
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

YIELD OF MAIZE (Zea mays L.) VARIETIES AS AFFECTED BY NEEM OIL

COATED UREA APPLICATION IN SOUTHERN BENIN
#11624

*Fassinou Hotegni V. Nicodeme, Houessou C. Gildas, Linkpon F. Orthia, Tchokponhoue A.

Dedeou, Houdegbe Carlos, Adje Charlotte, Achigan-Dako Enoch
Genetics, Biotechnology and Seed Science Unit (GBioS), Laboratory of Crop Production,
Physiology and Plant Breeding (PAGEV), Faculty of Agronomic Sciences, University of
Abomey-Calavi, 01 P.O. Box 526, Cotonou, Republic of Benin.
*e-mail: [email protected]

ABSTRACT

Maize is one of the most important cereals grown and consumed in Benin Republic. In maize
cultivation, huge amount of nitrogen is used contributing to soil and environment pollution.
Despite such huge level of applied nitrogen, maize grain yield is still below 2 ton/ha. The aim
of this study was to make available to the maize grain’s farmers, an eco-friendly practice based
on the use of Neem Oil Coated Urea (NOCU) for a high maize grain yield in Benin. An
experiment was carried out from April 2023 to August 2023 using a split plot with four blocks
design. The treatments consisted of the combination of four varieties (V1: DMR ESR-W, V2:
EVDT 97 STR W, V3: 2000 SYN EE-W and V4: ABONTEM) and five levels of fertilization
(T0: the control, T1: Urea only, T2: NOCU at 1%, T3: NOCU at 3% and T4: NOCU at 6%).
Yield data were collected. Results showed that with the application of the NOCU at the rate of
3% (200 Kg/ha of NPK and 75 Kg/ha of urea + 3ml of neem oil), V1 and V4 showed the highest
maize grain yield (3.82 ton/ha) and (4.71 ton/ha) respectively. With the application of the
NOCU rate at 6% (200 Kg/ha of NPK and (75 Kg/ha of urea + 6ml of neem oil), V2 and V3
showed the highest maize grain yield (5 ton/ha) and (4.07 ton/ha) respectively. These results
suggest that the NOCU at 3% (200 Kg/ha of NPK and (75 Kg/ha of urea + 3ml of neem oil)
and 6% (200 Kg/ha of NPK and (75 Kg/ha of urea + 6ml of neem oil) can be used for the maize
production in Benin. Further studies should be conducted to depict the post-harvest behavior of
produced maize from NOCU.

INTRODUCTION

In Benin Republic, maize contributes to 6.54% to the Gross Domestic Product (GDP) and is
listed among the crop being promoted in four Agricultural Development Territorial Agency
(ATDA) (6, 5, 4, 3 and 2) out of the seven ATDAs established by Benin government. Maize
production provides employment for communities in rural area and improves the farmer’s
livelihood (Saïdou et al., 2018). Among the West African countries, maize production volume
is low in countries such as Benin Republic, Côte d’Ivoire and Togo compared to that of Nigeria
and Burkina Faso (FAOSTAT, 2020). At the same time, the yearly per capita consumption of
maize is high in Benin Republic (87 kg), followed by Togo (70 kg) and Ghana (45 kg) (Badu-
Apraku & Fakorede, 2017). The actual maize grain yield is between 0.8 ton and 1.2 ton/ha
which is far below the potential yield of 3 to 4t/ha. According to the DSA (2021) the maize
grain yields in Benin during 2017, 2018, 2019, 2020 production season were respectively 1.29
ton/ha, 1.32 ton/ha, 1.07 ton/ha, 1.27 ton/ha. This reported low yield is caused, by the soil
degradation due to the increasing of the fertilizers amount, by the soil fertility degradation, and
by the climate change effect (low annual rainfall, high temperature) (Tovihoudji et al., 2022).
Considering the importance of maize and huge consumption by the population with the poor
grain yield, it is urgent to develop strategies to increase maize yield.
85
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Attempts have been made by Saïdou et al. (2018) who recommended the use of N-P-K rates at
80 Kg N /ha, 30 Kg P /ha and 25 Kg K /ha and 80 Kg N/ha, 15 Kg P/ha and 40 Kg K/ha (for
Acrisols) and 80.5 Kg N/ha, 22.5 Kg P/ha, 20 Kg K/ha (for Ferric and Plintic Luvisols) in
intensive maize grains production system in the Southern and Center parts of Benin. According
to Tovihoudji et al. (2022), the combination of microdose and drought tolerant varieties
(especially 35 kg/ha of N and 8 kg/ha of P applied to TZE Y Pop STR QPM maize varieties),
was suggested in the current context of declining soil fertility and climatic variability. In the
context of increasing prices of fertilizers (Ali & Azaroual, 2022), of climate change, soil fertility
declines along with consumers tendency to consume safe and quality product, there is an urgent
need to develop eco-friendly practices (agro-ecological practices) respecting consumers
attributes and farmers health.

Many researches have been conducted to develop agro ecological practices in maize production.
Tovihoudji and al. (2019) found that microdose fertilization alone increased maize grain yields
up to 1.145 ton/ha compared to the unfertilized land (1.096 ton/ha) in northern Benin. The same
authors also reported that combining microdose fertilization with farmyard manure increased
yields from 1.834 to 4.475 ton/ha with microdose + farm yard manure compared by 0.420 to
1.687 ton/ha. Akplo et al. (2019) found that the highest growth rate (2.38 cm/day), leaf area
(65.70 cm²), collar diameter (1.39 cm), grain yield (4.148 ton DM/ha), straw yield (5.077 ton
DM/ha) and harvested index (40%) were obtained with the combination of plowing and mulch.
Amogou et al. (2021) showed that the inoculation of maize of seeds with Pseudomonas syringae
+ 50% NPK + urea led to an increase of yield by 30.64 to 32.25%. All these technologies,
although proven by research, are not accessible to farmers. Therefore, there is a need to develop
and rethink other agronomic practices promoting the reduction of urea while increasing maize
grain yield and with high potential to be adopted by maize grains farmers.

In India, Gudge et al. (2019) found that the application of Neem Oil Coated Urea (NOCU)
increased the cob length, the cob width, the numbers of cobs per hectare and 1000 seed weight
in maize plants compared with simple urea application (uncoated urea). Ashraf et al. (2019)
found that the application of NOCU delayed the nitrification up to 30 days and increased the
plant available N pool compared to uncoated urea. The apparent N recovery ranged from 61-
84% between coated urea treatments than ordinary urea. The relative growth rate increased by
11-89% and 30-70% in all natural nitrification inhibitors coated urea. Fagodiya et al. (2019)
found that the NO2-N emission decreased by 16% in NOCU with higher maize grain yield as
compared to the uncoated urea application. The same authors also reported that the greenhouse
gas intensity was reduced by 6% in NOCU. So far, no scientific report has been found dealing
with NOCU in maize production in Benin Republic. Therefore, this study aims at developing
eco-friendly practices based on NOCU application in maize production in Benin. Specific
objectives were to (i) evaluate the effect of the NOCU on the different maize varieties yield and
physiological parameters and (ii) determine the economical rate of NOCU giving a relatively
high grain yield.

MATERIALS AND METHODS

Experimental sites and maize varieties used

The experiment was conducted from April 2023 to August 2023in the southern Benin (Guinean
zone phytogeographical zone) at Zè municipality. The climate is characterized by a tropical
type with a bimodal rainfall with two rainy seasons (March to July and September to October)
and two dry seasons: (August and November to February). Geographically, the site was in
86
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

latitude of 6°36' 36"N and longitude of 2°13' 46"E. The variation of the annual rainfall in
southern Benin is between 1200 mm and 1300 mm (Adigoun et al., 2022). The average annual
temperature is around 25 °C and the maximum is 34.5 °C.

Four maize varieties were used, three local varieties and one hybrid. The local varieties are
those that are well appreciated by the farmers and mostly produced in southern Benin. These
varieties are: DMR ESR-W, EVDT 97 STR W and 2000 SYN EE-W (Adigoun et al., 2022).

Experimental design and treatments

In each experimental site a Split-plot design with two factors was used. The first factor was the
‘’Variety’’ with 4 modalities: V1: DMR ESR-W; V2: EVDT 97 STR W; V3: 2000 SYN EE-
W; V4: ABONTEM, The second factor was the “Rate of NOCU”: For maize cultivation in
the Republic of Benin, the NPK 10-20-20 fertilizer is applied at 200 Kg/ha 21 days after sowing
(DAS) and urea is applied at 100Kg/ha between 30 and 45 DAS depending on the varieties
(Salami et al. 2020). According to Ashraf et al. (2019), 100g of granular urea is coated with
1mL of neem oil. To prepare the NOCU, 75% of the dose of 100 Kg/ha of urea (2.4g per plant)
was mixed with the neem oil at the rate of 1ml/100g of urea. The NOCU was applied at different
doses; 1%, 3% and 6%. The NOCU at 1% was prepared by a mixture of 100g of granular urea
and 1ml of neem oil. The NOCU at 3% was prepared by a mixture of 100g of granular urea and
3ml of neem oil and the NOCU at 6% was prepared by a mixture of 100g of granular urea and
6ml of neem oil. Based on these data, five modalities were identified: (i) The control (T0: 0
Kg/ha of NPK and 0 Kg/ha of urea); (ii) Urea (T1: 200 Kg/ha of NPK and 100 Kg/ha of PU);
(iii) NOCU at 1% (T2: 200 Kg/ha of NPK and NOCU at 1% (75 Kg/ha PU + 1ml of neem oil));
(iv) NOCU at 3% (T3: 200 Kg/ha of NPK and NOCU at 3% (75 Kg/ha PU + 3ml of neem oil)
and (v) NOCU at 6% (T4:200 Kg/ha of NPK and NOCU at 6% (75 Kg/ha PU + 6ml of neem
oil).

The experiment was composed by four replications (4 blocks) with 20 experimental units (5 ×
4 from treatments combination with a total of 80 experimental units per experiment) per block.
The plots size was 3 m × 5.5 m (16.5 m²) and contained 24 maize plants. The block size was
132 m × 4 m (528 m²). The maize seeds were sown at the density of 62,500 plants/hectare (0.80
m × 0.40 m) with 2 seeds per hole.

Collected data and data analysis

Yield data was collected at harvesting. Data were analyzed with R.4.3.1 software (2023) using
ANOVA model. The hierarchization of the means were obtained using “Student-Newman-
Keuls” (SNK), and boxplot function was used to make the boxplot.

RESULTS AND DISCUSSION

Results showed main effects of varieties and Treatments (NOCU) (Table 1). Regarding the
varieties, similar yields were obtained for V1 and V3 and V2 and V4 (Figure 1). All treatments
with NOCU have high yield compared to the farmer’s practice. Although the interaction
between variety and treatment was not significant, results indicated that:
• For the variety V2, the highest maize grain yield (5 ton/ha) was obtained on the plots
which received the treatment T4 and the lowest (4,03 ton/ha) was obtained on the plots
which did not receive any treatment.
• For the variety V3, the highest maize grain yield (4,07 ton/ha) was obtained on the plots
which receive the treatment T4 and the lowest (1,99 ton/ha) on the plot which did not
receive any treatment.
87
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

• For the variety V4, the highest maize grain yield (4,71 ton/ha) was obtained on the plots
which receive the treatment T3 and the lowest (2,76 ton/ha) on the plants which did not
receive any treatment.

Table 1. P-value of the effect of NOCU application on the maize grain yield

Factors Grain yield

Variety (V) 0.000***
Treatment NOCU (T) 0.002**
V*T 0.675
**: Statistically significant at 0.01 > P ≥ 0.001; ***: Statistically significant at P < 0.001

a
a
b b

Figure 1. Effect of NOCU application on maize grain yield per variety

88
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

b a a a
a

Figure 2. Effect of NOCU application on maize grain yield

These previous results can be explained by the fact that the NOCU at 3% and 6% allow the
nitrification inhibition for a long period of time (Ali et al., 2020). This nitrification inhibition
increases the nitrogen content in the soil and in the plant by the decreasing of the NH 4+ losses. The
maize grain yield is increased because the nitrogen availability increases the stem girth which lead
to the better nutrient and water uptake by the plants (Ashraf et al., 2019; Nasar et al., 2021). It is
also due by the increasing of the ear length and diameter which increase the number of grains rows
and the numbers of grains per row (Mu & Chen, 2021). In fact, the maize grain dry weight is
influenced by the main component of the grain which is the starch (Liu et al., 2021).
Physiologically, the maize grain formation is closely related with the starch synthesis in the grain
by the starch enzymes. The starch synthesis enzymes such as ADP-glucose pyrophosphorylase
(AGPase) (C16H25N5O15P2), granule bound starch synthetase (GBSS), soluble starch synthase
(SSS) and starch branching enzyme (SBE) have an crucial role in starch accumulation and grain
weight (Liu et al., 2021). According to the same authors, when the nitrogen supply is high it has a
strong positive effect on this starch enzymes activities which increase the starch synthesis and by
this way the grain weight.

CONCLUSION

Maize grain yield in Benin can be increased by improving the nitrogen use efficiency of the
plants. To allow this it is essential to avoid the ammonium losses in the soil by the coating of the

89
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

nitrogen fertilizer with the natural nitrification inhibitors such as neem oil. The NOCU
application increased the maize leaves chlorophyll content, the maize leaves area index, the
maize plants height and stem girth. The different maize varieties used showed different trends
after the application of the treatment. The aim of this study which is to find an eco-friendly
practice to enhance the maize grain yield in Benin by the reducing of the urea amount is reached.
In this study we used 75% of the recommended dose and obtained good results. We suggest
conducting a second trial to be sure about the results obtained

REFERENCES

Adigoun RFR, Houdegbe AC, Fassinou Hotegni NV, Segnon AC, N'Danikou S, Adjé CAO,
Adadja RPM and Achigan-Dako EG (2022). Enabling effective maize seed system in low-
income countries of West Africa: Insights from Benin. Front. Sustain. Food Syst. 6:1045629.
doi: 10.3389/fsufs.2022.1045629
Akplo, T. M., Kouelo, A. F., Azontonde, H. A., & Houngnandan, P. (2019). Effect of tillage and
mulching on agronomics performances of maize and soil chemical properties in Linsinlin
Watershed of Centre of Benin. African Journal of Agricultural Research, 14(31), 1421‑1431.
https://doi.org/10.5897/AJAR2019.13952
Ali, P. A. A., & Azaroual, F. (2022). Les répercussions économiques de la guerre en Ukraine pour
l’Afrique et le Maroc.
Amogou, O., Noumavo, A. P., Agbodjato, N., & SINA, H. (2021). Rhizobacterial inoculation in
combination with mineral fertilizer improves maize growth and yield in poor ferruginous soil
in central Benin. Journal of Biotechnology, Computational Biology and Bionanotechnology,
102(2), 141‑155. https://doi.org/10.5114/bta.2021.106520
Ashraf, M., Aziz, T., Maqsood, M., Bilal, H., Raza, S., Zia, M., Mustafa, A., Xu, M., Wang, Y.,
Ashraf, M. N., Aziz, T., Maqsood, M., Bilal, H., Raza, S., Zia, M., Mustafa, A., Xu, M., &
Wang, Y. (2019). Evaluating Organic Materials Coating on Urea as Potential Nitrification
Inhibitors for Enhanced Nitrogen Recovery and Growth of Maize (Zea mays L.).
International Journal of Agriculture and Biology, 22(5).
https://doi.org/10.17957/IJAB/15.1175
Badu-Apraku, B., & Fakorede, M. A. B. (2017). Maize in Sub-Saharan Africa : Importance and
Production Constraints. In B. Badu-Apraku & M. A. B. Fakorede, Advances in Genetic
Enhancement of Early and Extra-Early Maize for Sub-Saharan Africa (p. 3‑10). Springer
International Publishing. https://doi.org/10.1007/978-3-319-64852-1_1
DSA. (2021). Direction des Statistiques Agricoles. https://dsa.agriculture.gouv.bj/
Fagodiya, R. K., Pathak, H., Bhatia, A., Jain, N., Gupta, D. K., Kumar, A., Malyan, S. K., Dubey,
R., Radhakrishanan, S., & Tomer, R. (2019). Nitrous oxide emission and mitigation from
maize–wheat rotation in the upper Indo-Gangetic Plains. Carbon Management, 10(5),
489‑499. https://doi.org/10.1080/17583004.2019.1650579
FAOSTAT. (2020). Classement des États d’Afrique par production de maïs.
https://atlasocio.com/classements/economie/agriculture/classement-etats-par-production-
mais-afrique.php
Gudge, A., Rawat, G., JAT, S., & Tiwari, S. (2019). Impact of conservation agriculture and
nitrogen management on growth and productivity of maize (Zea mays L.). Journal of
Pharmacognosy and Phytochemistry, 8(4), 2260‑2264.
Liu, X., Gu, W., Li, C., Li, J., & Wei, S. (2021). Effects of nitrogen fertilizer and chemical
regulation on spring maize lodging characteristics, grain filling and yield formation under
90
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

high planting density in Heilongjiang Province, China. Journal of Integrative Agriculture,

20(2), 511‑526. https://doi.org/10.1016/S2095-3119(20)63403-7
Mu, X., & Chen, Y. (2021). The physiological response of photosynthesis to nitrogen deficiency.
Plant Physiology and Biochemistry, 158, 76‑82.
https://doi.org/10.1016/j.plaphy.2020.11.019
Nasar, J., Khan, W., Khan, M. Z., Gitari, H. I., Gbolayori, J. F., Moussa, A. A., Mandozai, A.,
Rizwan, N., Anwari, G., & Maroof, S. M. (2021). Photosynthetic Activities and
Photosynthetic Nitrogen Use Efficiency of Maize Crop Under Different Planting Patterns
and Nitrogen Fertilization. Journal of Soil Science and Plant Nutrition, 21(3), 2274‑2284.
https://doi.org/10.1007/s42729-021-00520-1
Saïdou, A., Balogoun, I., Ahoton, E. L., Igué, A. M., Youl, S., Ezui, G., & Mando, A. (2018).
Fertilizer Recommendations for Maize Production in the South Sudan and Sudano-Guinean
Zones of Benin. In A. Bationo, D. Ngaradoum, S. Youl, F. Lompo, & J. O. Fening (Éds.),
Improving the Profitability, Sustainability and Efficiency of Nutrients Through Site Specific
Fertilizer Recommendations in West Africa Agro-Ecosystems (p. 215‑234). Springer
International Publishing. https://doi.org/10.1007/978-3-319-58792-9_13
Tovihoudji, P. G., Akponikpè, P. B. I., Agbossou, E. K., & Bielders, C. L. (2019). Variability in
maize yield and profitability following hill-placement of reduced mineral fertilizer and
manure rates under smallholder farm conditions in northern Benin. Field Crops Research,
230, 139‑150. https://doi.org/10.1016/j.fcr.2018.10.018
Tovihoudji, P. G., Bagri, B. M., Batamoussi Hermann, M., Tonnang, Z. E. H., & Akponikpè, P. B.
I. (2022). Interactive Effects of Drought-Tolerant Varieties and Fertilizer Microdosing on
Maize Yield, Nutrients Use Efficiency, and Profitability in the Sub-Humid Region of Benin.
Frontiers in Agronomy, 3. https://www.frontiersin.org/articles/10.3389/fagro.2021.763430

91
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

GEOSPATIAL DATA AND MAPPING

92
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

TOWARD PRECISION AGRICULTURE BY ASSESSING FAO SOIL DATA

ACCURACY WITH LOCALISED SOIL MAPPING IN MID-WESTERN UGANDA
#11669

1*Emmanuel Ojambo; 1L. Abasize, 1P. Tamale, 1T. Matila, 1B. Kiconco, 1E. Opolot, 2S. Kusiima,
1P. Ebanyat, 3J. Mutegi, 3 T. Oberthur, 1P. Musinguzi
1School of Agricultural Sciences, Makerere University, Kampala, Uganda
2Local Government, Kikuube District
3African Plant Nutrition Institute, Benguérir, Morrocco
*e-mail: [email protected]

ABSTRACT

Proper soil mapping remains a major input for efforts that aim at developing context specific soil
and crop management. Soil maps often enable policy makers and other agricultural extension
programs to establish the suitability of land for various production systems. Currently, most soil
maps have been developed basing in the FAO soil databases at a scale of 1:50,000 yet farmers and
other local land users require fine scales to guide decision in the use of fertilizers and selection of
suitable crop enterprises. A study was conducted in midwestern Uganda in Kikuube town council
to establish the accuracy of FAO Soil maps for improving soil information for precision agriculture.
The methodology involved use of ARCGIS 10.8, Google Earth Pro and SAGA GIS creating local
mapping units, which are areas relatively homogeneous in soil composition. The mapping units
were defined using a combination of slope percentage, topographic wetness index, remote sensing
data and land cover. Field morphology surveys were conducted and soil samples collected basing
on WRB guidelines in selected soil profiles along soil mapping units. Soil classification by
harmonisation was conducted. A near-tool analysis in GIS was used to overlay the new soil profile
data on mapped units basing on their proximity to sampled points. Findings from ground truthing
and soil mapping activity indicate that Ferralsols, Plinthisols, and Gleysols cover 44.84%, 37.45%,
and 17.71% of the aredca, respectively. The existing FAO classification identified four soil types:
Dystric Regosols, Gleysols, Acric Ferralsols, and Histosols, covering 80.51%, 12.75%, 6.72%, and
1.78% of the area, respectively. The soil types under FAO-Database were somewhat different from
soil types obtained in the re-mapping activity in Kikuube town council. Overall, local soil mapping
is essential for farmers to accurately identify soil types to aid proper use of fertilizers. The study
highlighted discrepancies between the FAO soil classification and field soil mapping activity,
underscoring the importance of ground-truthing and re-doing soil morphology for accurate soil
maps to guide precision agriculture.

INTRODUCTION

Uganda is in the East region of Africa with a population of about 45.9 million people as per UBOS
2024 census. More than 75% of its population is employed in the agricultural sector mainly
composed of peasant farmers (Okonya, 2014). Uganda was historically known for its fertile soils
in the past decades, but now the fertility is decreasing from low to medium due to land degradation,
soil erosion and population increasing leading to over utilization of land (Apanovich et al., 2018).
Farmers are adopting to the use of fertilizers to boost their yields to cope with soil infertilities as
noted by Rapsomaniki in 2015, though many smallholder farmers are reluctant to purchase

93
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

fertilizer to increase crop yields, despite the evidence that fertilizer use increases yield that in turn
increases income (Larson.2016). Efficient application of fertilizers requires knowing the type and
composition of soils for optimal use of fertilizers (Singh et al., 2015). Proper soil mapping remains
a major input for efforts that aim at developing context specific soil and crop management (Chen
et al., 2022). Soil maps often enable policy makers and other agricultural extension programs to
establish the suitability of land for various production systems (Apokti et al., 2019). Currently,
most soil maps have been developed basing in the FAO soil databases at a scale of 1:50,000 yet
farmers and other local land users require fine scales to guide decision in the use of fertilizers and
selection of suitable crop enterprises (Campbell, 2018). This calls for farmers to map and analysis
the soils before making selecting crops and making fertilizer use decisions.

METHODS AND MATERIALS

Study Area
The study was conducted in Kikuube town council that is in western Uganda in Bunyoro kingdom
in Kikuube district. The district covers an area of about 2,097 Km2, with Lake Albert covering
905.9 Km2 (43.2%) according to Uganda Investment Authority (UIA) in 2021. The district has five
sub-counties namely, Kiziranfumbi, Kabwoya, Buhimba, Bugambe and Kyangwali and two town
councils, Kikuube and Buhimba. The district had a population estimation of 410,000 people in
2019 with 110, 000 refugees (UIA, 2021). The district’s climate varies with altitude, featuring a
bimodal rainfall pattern ranging from 800 mm in the Rift Valley floor to 1,500 mm in the
escarpment. The soils are mainly ferralitic and acidic, with good organic matter in lower slopes
and valleys and varying soil productivity, with some areas having fair to low productivity according
to Kikuube district local government. Agriculture is the main source of livelihood for 90% of
Kikuube's residents, who cultivate key food crops such as maize, rice, cassava, bananas, and beans.
Important cash crops include tobacco, tea, and sugarcane. Livestock farming, including poultry and
piggery, also contributes to the local economy. Fishing on Lake Albert boost the economy of the
locals. Recent developments include oil and gas exploration in the Albertine Rift Valley
(Nuwagaba, 2021), specifically in Kyangwali and Kabwoya Sub-Counties expected to have a major
impact on Uganda's GDP and public revenue (Ddamulira, 2021).

94
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 3. Location of Kikuube district

Data Collection and Analysis

The methodology involved using ARCGIS 10.8, Google Earth Pro, and SAGA GIS to create local
mapping units. Local mapping units were composed of slope percentage derived from digital
elevation model, topographic wetness index derived from digital elevation model in SAGA GIS,
FAO Soil maps, and land use/cover obtained from land use classification of satellite images. The
datasets were then overlaid and soil mapping units created based on homogeneity. Map units were
then used in field morphology soil surveys, and soil samples were collected based on the World
Reference Base guidelines from selected soil profiles (IUSS, 2022). The collected soil samples
were analyzed in the laboratory and results tabulated based on Fao classification (Fao, 2015) visa
vie the sampled mapping units and soil types generated. The laboratory results were then used to
classify the generated mapping units visa vie the Near-tool analysis in GIS used to overlay the new
soil profile data on mapped units, considering their proximity to sampled points.

95
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Ferrallsols Gleysols Plinthisols

Figure 4. Field Soil Type Mapping.

RESULTS AND DISCUSSION

FAO Soil mapping comprised of four soil types namel;-Acric Ferralsols, Dystric Regosols,
Gleysols and Histosols covering 6.72%, 80.51%, 12.75% and 0.02% respectively. Whereas the
Field Soil Mapping comprised of Ferralsols, Gleysols and Plinthisols covering 44.84%, 17.71%
and 37.45%. An overlay of FAO soil map and Field mapping Soil map show FAO Soil Mapping
Classification was generic, and it was noted that each Fao soil type comprised of the three field
mapping soil types except the Histosols that were 100% Classified as Gleysols as shown in Table
1.

Comparison of Soil Classification Area

The FAO maps indicated Dystric Regosols as dominant soil types covering more than three
quarters of the soil types of which 44.47% were Ferralsols, 15.78% were Gleysol and 39.75 were
Plinthisols (Table 2). The Local Field Mapping identified Ferralsols as the most commom soil type
covering 44.84% of the soils in the study area. More than half of the Fao Classified Ferralsols were
on a local precise scale classified as Plinthisol. This was also noted in the Fao Gleysols
classification, that locally on a precise scale comprised of a mixture of three locally mapped soil
types dominated by Ferralsols. The generalization of Fao Soils on a scale of 1:50,000 which on
ground represents several mapping units underscores the importance of ground-truthing and re-
evaluating soil morphology to produce accurate soil maps.

96
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 3. Results from FAO and Field Soil Classification

FAO Soil Map Field Soil Mapping

Soil Type Area (ha) Area (%) Soil Type Area (ha) Area (%)
Acric Ferralsols 695.93 6.72 Ferralsol 4,645.11 44.84
Dystric Regosols 8,341.54 80.51 Gleysol 1,834.86 17.71
Gleysols 1,321.14 12.75 Plinthisol 3,880.42 37.45
Histosols 1.78 0.02 Total 10,360.39 100
Total 10,360.39 100

Table 4. Soils Comparison

Soil Classification Area Area

Fao Dataset Field Mapping (ha) (%)
Acric Ferralsol 296.19 42.56
Ferralsols Gleysol 18.80 2.70
Plinthisol 380.95 54.74
Dystric Ferralsol 3,709.77 44.47
Regosols Gleysol 316.32 15.78
Plinthisol 3,315.45 39.75
Gleysols Ferralsol 639.15 48.38
Gleysol 497.96 37.69
Plinthisol 184.02 13.93
Histosols Gleysol 1.78 100.00

Therefore, accurate soil mapping is essential for farmers to make informed decisions about
fertilizer use and crop selection. The study's findings suggest that localized soil mapping provides
a more accurate representation of soil types, which is crucial for precision agriculture. By
identifying the correct soil types, farmers can optimize their agricultural practices, improve crop
yields, and ensure sustainable land management.

REFERENCES

Akpoti, K., Kabo-bah, A. T., & Zwart, S. J. (2019). Agricultural land suitability analysis: State-of-
the-art and outlooks for integration of climate change analysis. Agricultural systems, 173,
172-208.
Apanovich, N., & Mazur, R. E. (2018). Determinants of seasonal food security among smallholder
farmers in south-central Uganda. Agriculture and Food Security, 7(1).
https://doi.org/10.1186/s40066-018-0237-6
Campbell, G., Corstanje, R., Hannam, J. A., Black, H. I. J., & Lilly, A. (2018). The Application Of
Digital Soil Mapping To Improve The Resolution Of National Soil Properties Across Great
Britain School Of Water, Energy And Environment (Swee).
Chen, S., Arrouays, D., Mulder, V. L., Poggio, L., Minasny, B., Roudier, P., & Walter, C. (2022).
Digital mapping of GlobalSoilMap soil properties at a broad scale: A review. Geoderma,
409, 115567.

97
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Ddamulira, R. (2021). The co-existence of sustainable forest management amidst oil development
analyzing resource policy options for uganda-east africa.
FAO. (2015). World reference base for soil resources 2014: International soil classification system
for naming soils and creating legends for soil maps. Food and Agriculture Organization of
the United Nations.
IUSS Working Group WRB. 2022. World Reference Base for Soil Resources. International soil
classification system for naming soils and creating legends for soil maps. 4th edition.
International Union of Soil Sciences (IUSS), Vienna, Austria.
Kikuube District Local Government. (n.d) Retrieved August 31, 2024, from https://kikuube.go.ug/
Larson, D. F., Savastano, S., Murray, S., & Palacios-López, A. (2016). On the Determinants of
Low Productivity in Maize Farming in Uganda: The Role of Markets, Fertilizer Use and
Gender. In In Pursuit of an African Green Revolution (pp. 165–182). Springer Japan.
https://doi.org/10.1007/978-4-431-55693-0_8
Nuwagaba, I., & Tshombe, L. M. (2021). African Renaissance.
https://journals.co.za/doi/pdf/10.31920/2516-5305/v2021sin1a10
Okonya, J. S., & Kroschel, J. (2014). Gender differences in access and use of selected productive
resources among sweet potato farmers in Uganda. Agriculture & Food Security, 3, 1-10.
Rapsomanikis, G. (2015). Food and Agriculture Organization of the United Nations Rome.
www.fao.org/publications.
Singh, B., & Ryan, J. (2015). Managing fertilizers to enhance soil health. International Fertilizer
Industry Association, Paris, France, 1.
UBOS. (2024). National-Population-and-Housing-Census-2024-Preliminary-Report. Retrieved
August 30, 2024, from https://www.ubos.org/wp-content/uploads/publications/National-
Population-and-Housing-Census-2024-Preliminary-Report.pdf
Uganda Investment Authority. (2021). Kikuube District Investment Profile Brief District Profile.
https://www.ugandainvest.go.ug/wp-content/uploads/2021/08/Kikuube-2021.pdf

98
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ASSESSING OF SOIL NUTRIENTS USING LABORATORY AND REMOTE SENSING

METHODS IN NORTHERN GUINEA SAVANNAH
#11640

1A.R. Garba, 2A.M. Yamusa, 1C.K. Daudu, 2S.L. Yau, 3M.I., Gaya, 2A. Abdulkadir
1. National Agricultural Extension and Research Liaison Services (NAERLS), Ahmadu Bello
University (ABU) Zaria
2. Department of Soil Science, Institute of Agricultural Research (IAR)/ABU, Zaria
3. Samaru College of Agriculture, Division of Agricultural Colleges, Ahmadu Bello University,
Zaria, Nigeria
e-mail: [email protected]; +2348035457216

ABSTRACT

To understand soil properties and how they might be used sustainably, mapping of soil attributes
is a crucial activity. The study was carried out in four local government area of Kaduna State of
Nigeria to map out some soil properties and assess their variability within the area. From the study
area, a total of 16 soil samples (0–20 cm) were collected from different cropping patterns. A
portable global positioning system (GPS) was used to collect the coordinates of each sampling site.
Then, the soil properties, that is, soil organic carbon (SOC), total nitrogen (Total N), soil organic
matter (SOM), and soil available nutrients (P and K) were measured in the laboratory. Correlation
analysis between laboratory and remote sensing data showed positive relationships for carbon
(r=0.23), total nitrogen (r=0.14), and organic matter (r=0.68), but negative correlations for
available phosphorus (r=-0.48) and potassium (r=-0.42). These variable results highlight the greater
reliability of remote sensing for assessing total carbon and organic matter versus limitations in
quantifying phosphorus and potassium availability. Interactive effects of climate variables on soil
nutrients were not directly assessed but remain a critical area for further research.

Keywords: Remote sensing, Laboratory analysis and Soil nutrients

INTRODUCTION

Soil is a complex material that is extremely variable in its physical and chemical composition. The
influence of soil and crop management practices such as fertilization, cropping systems, and land-
use change exert considerable changes to such soil compositions over time. Over the years, routine
analysis of such chemical and physical changes remains the only way to access and maintain the
fertility of soil. The importance of the soil analysis cannot be over-emphasised since low nutrient
values limit plant growth and excessive rainfall may result in loss of nutrients from the soil, causing
soil fertility degradation and water pollution (Chi, et al., 2019). Therefore, soil analysis is the basic
frame for providing the nutrient requirements of every crop.

Comparative assessment of soil nutrients under different cropping systems and changing climate
conditions requires a combination of ground-based soil sensing and laboratory analytical methods
along with remote sensing technologies. Ground-based sensors like portable X-ray fluorescence
(pXRF) analyzers allow rapid in-situ quantification of major and trace nutrients in soils (Towett et.
al. 2015). Laboratory methods using combustion analysis, titrations, and spectroscopic techniques

99
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

offer accurate and precise measurements of total and plant-available nutrient pools (Robertson et.
al. 1999). Satellite and aerial remote sensing provide spatial data on vegetation characteristics and
soil properties related to soil fertility at broader scales (Mulder et al. 2011). Together, these
approaches can provide a comprehensive assessment of soil nutrient dynamics across landscapes.
This study synthesizes research utilizing integrated soil laboratory, ground-based sensing, and
remote sensing methods to evaluate the impacts of climate and agricultural land use on soil
nutrients. The focus is on comparative studies across different cropping systems under current and
projected future climate scenarios, concentrating on research conducted in sub-Saharan Africa.

Monitoring agriculture from remote sensing is a vast subject that has been widely addressed from
multiple viewpoints, sometimes based on specific applications (e.g. precision farming, yield
prediction, irrigation, weed detection), on specific remote sensing platforms (e.g. satellites,
Unmanned Aerial Vehicles - UAV, Unmanned Ground Vehicles - UGV) or sensors (e.g. active or
passive sensing, wavelength domain, spatial sampling) or specific locations and climatic contexts
(e.g. country or continent, wetlands or dry lands).

In recent years, digital soil mapping has been identified as a low-cost and efficient method for
predicting the spatial distribution of soil nutrients. Most digital soil mapping methods are based on
soil-landscape models, which establish mathematical or statistical relationships between soil
properties and related environmental variables (Zhang et. al., 2019) by predicting soil
characteristics and fertility status with the help of remote sensing data. Remote sensing in itself is
the process of detecting and monitoring the physical characteristics of a particular soil by
measuring its reflected and emitted radiation at a distance. The nature and working principle of
remote sensing give it the advantages of being an extensive, non-invasive, timeliness, and flexible
method of soil analysis, and it has the potential to increase the availability of high-resolution remote
sensing data by providing a new opportunity for predicting soil characteristics with acceptable
accuracy.

MATERIALS AND METHODS

Study Area Description

This study was conducted in the northern and southern Guinea savannahs of Kaduna State. The
locations in the northern savannah were Kubau and Makarfi while the southern Guinea savannah
was Kagarko and Lere LGAs.

Nigeria’s climatic zone encompasses the tropical humid forest in the south and the savannah in the
north. Nigeria’s climatic zone encompasses the tropical humid forest in the south and the savannah
in the north. The derived savannah is a transition zone between the rainforest and savannah biomes
caused by forest clearance as stated by Ofomata (1975). The study was carried out in Kaduna state
(Longitude/Latitude 9°26' to 11°13' N and 7°47 to 8°42' E) respectively, which is in the Northwest
of Nigeria (Fig.1). The climate belt of the area is tropical Guinea Savanna, with an annual average
temperature of 25.2℃ and an annual average rainfall of 1,323mm (Akinbode et. al., 2008).

100
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Location of area(s) of interest in Kaduna state (Kubau, Makarfi, Lere, and Kagarko Local
Government Areas) and distribution of samples.

Soil Laboratory Analysis

The chemical properties of the soils were determined at the Soil Science Laboratory, Faculty of
Agriculture, Ahmadu Bello University, Zaria, Nigeria.

The soil samples were determined by using the following methods: The organic carbon was
analyzed by the wet oxidation method of Walkley and Black as modified by (Nelson and Sommers,
1982). Total nitrogen by the micro-kjeldahl distillation procedure according to (Bremmer, 1996),
available phosphorus was determined by the Bray No. 1 acid fluoride method (Nelson and
Sommers, 1982).

Field Sampling and Spatial Analysis

The remote sensing samples were collected in the same 4 LGAs of Kaduna state distributed evenly
between the northern and southern parts of the state; and for each farm, a sample was collected for
each 4 points at 0-20cm depth. This is because most satellite data for soil properties are within the
top-soil range (Hengl et al., 2015). Therefore, restricting the ground-based sampling in this study
to 0-20 cm aligns with the typically sensed depth ranges from satellite platforms. The remote
sensing samples were collected same time during the soil sample collection on the field. A data
streaming pipeline is used to query and download multispectral data from the Sentinel-2 repository
which is then processed using a proprietary algorithm. The result from the satellite image and how
it correlates with those from chemical analysis is the subject and primary objective of this study.

Correlation Analysis
Python programming language version 3.11.4 was used as the correlation analysis tool using
Pearson to compare the laboratory analysis and remote sensing results (Virtanen et al., 2020).
Scatter plots allow visualization of the relationship between two variables, while correlation
analysis provides a quantitative measure of the strength and direction of the relationship (Graham,
2023). Python was selected due to its extensive libraries for statistical analysis and data
visualization along with the flexibility to handle diverse data types from both laboratory and
satellite sources (Qiusheng et al., 2009). Utilizing the Python environment for integrated analysis
of remote sensing imagery and laboratory soil analytics follows established best practices for
digital soil mapping and precision agriculture applications (Padarian et al., 2019).
101
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Python provides a flexible open-source platform for handling diverse datasets and performing
correlation analysis (Hengl et al. 2022). Two datasets were employed, one from remote sensing
and the other from the laboratory, each containing 16 instances of soil chemical properties across
12 columns. These datasets were collected from four distinct communities in Kaduna State:
Gubuchi, Kuli, Krosha, and Kubacha, each located in different Local Government Areas.

RESULTS AND DISCUSSIONS

Correlation Analysis
Correlation between the remote sensing nitrogen and the lab nitrogen result
The result of correlation between total nitrogen of remote sensing data and laboratory analysis is
presented in Figure 1. From the result, there was a weak positive correlation between the
determined parameters, and this indicates an existing relation between nitrogen levels assessed
through the remote sensing and the laboratory analysis. The weak positive correlation (r=0.14)
found between the laboratory and remote sensing soil nitrogen could be associated with the high
mobility and volatilization nature of nitrogen that may encourage leaching, run-off and other
nitrogen losses from the soil, hence very difficult to measure. Towett et al. (2015) found a weak
correlation (r=0.19) between laboratory and portable X-ray Fluorescent sensor nitrogen
measurements in Kenyan soils due to difficulties estimating subsurface nitrogen indirectly from
the spectral response. The low correlation highlights challenges in using remote sensing alone to
accurately predict soil nitrogen across agricultural landscapes. The need for further ground-based
sensing ground-truthing of satellite data to improve nitrogen prediction aligns with Piikki et al.
(2013), who used on-ground sensors to calibrate satellite imagery for soil clay mapping. Vågen and
Winowiecki (2019) also emphasized multi-scale calibration of remote sensing using soil analytical
lab data for accurate digital soil mapping.

The finding that neither remote sensing nor laboratory methods fully capture soil nitrogen
complexity agrees with Hengl et al. (2017), who concluded that integrated approaches are essential
given the intricacies of nitrogen biogeochemistry. The variability between sites also reflects Towett
et al. (2015), who found location-specific differences in remote sensing accuracy for soil nutrients.
Further coordinated research and data integration will help improve soil nitrogen assessment and
enhance remote sensing capabilities for nutrient management.

Correlation between the remote sensing organic matter and the lab organic matter result
The statistical analysis indicates a significant positive correlation between the remote sensing-
derived organic matter data and the laboratory organic matter data. The strong positive correlation
(r=0.68) between remote sensing and laboratory soil organic matter data is consistent with findings
from other studies. Shepherd and Walsh (2002) reported R-values from 0.76 to 0.89 between lab
and field spectroscopy organic matter measurements across diverse African agricultural soils.
Towett et al. (2015) found the highest correlation (r=0.86) between laboratory and portable XRF
sensor organic carbon content compared to other nutrients in Kenyan soils. The reliability of remote
sensing for organic matter mapping aligns with Vågen and Winowiecki (2019), who used MODIS
satellite data to map soil organic carbon across Sub-Saharan Africa with reasonable accuracy
compared to ground-based sensing. The robust relationship between spectral response and organic
matter is attributed to the direct impacts of surface organic content on crop growth patterns
detectable through remote imaging (Hengl et al. 2017). However, some researchers note challenges
102
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

in relating surface organic matter to total profile carbon stocks using remote sensing alone (Piikki
et al. 2013). Integrated approaches incorporating soil sampling, terrain analysis, and digital soil
mapping techniques may further improve organic matter quantification across landscapes (Hengl
et al. 2017). Still, the strong positive correlation demonstrates the potential of remote sensing for
cost-effective wide-area mapping of this important indicator of soil quality and health.

Figure 1. Correlation between the Remote Sensing Nitrogen and the Lab Nitrogen result.

Figure 2. Correlation between the remote sensing organic matter and the lab organic matter result.

Correlation between the remote sensing potassium and the lab potassium result
A negative correlation was observed between the remote sensing-derived potassium data and the
laboratory potassium data, with a correlation value of -0.42 The negative correlation (r=-0.42)
between remote sensing and laboratory soil potassium aligns with other studies showing the
complexity in using spectral data to estimate plant-available potassium. Piikki and Söderström
(2019) found poor correlation (r=0.38) between remote sensing vegetation indices and
exchangeable potassium measured in topsoils across agricultural fields in Sweden. They attributed
this to the dependence of spectral response on multiple soil factors like mineral composition

103
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

influencing potassium availability. Mulder et al. (2011) noted challenges in relating leaf potassium
absorption to total soil potassium pools given intricacies of potassium chemistry and soil
interactions. Vågen and Winowiecki (2019) were unable to map exchangeable potassium at
sufficient accuracy using solely MODIS (moderate resolution imaging spectroradiometer) satellite
data for Sub-Saharan African soils. This could indicate that the remote sensing data might not
accurately capture the true potassium levels in the soil or that there are other factors affecting the
results. The finding highlights the need for integrated approaches combining spectral data with soil
chemistry analysis, geologic surveys, and crop modeling to improve potassium prediction noted by
both Piikki and Söderström (2019) and Vågen and Winowiecki (2019).

While it shows promise for assessing organic matter, it may have limitations in accurately
estimating potassium levels. Understanding these correlations is vital for the appropriate
interpretation of remote sensing data in agricultural and environmental applications. Further
research and validation may be needed to better understand the factors contributing to these
correlations and improve the accuracy of remote sensing techniques for soil property assessments.

Correlation between the remote sensing carbon and the lab carbon data
Based on figure 4 below, it shows that a weak positive correlation of 0.23 was observed between
the remote sensing-derived carbon data and the laboratory carbon data. The weak positive
correlation (r=0.23) between remote sensing and laboratory soil carbon aligns with other studies
showing the limitations of using vegetation indices alone to estimate total soil organic carbon.
Mulder et al. (2011) found poor correlations between satellite data and measured soil carbon, as
remote sensors only detect surface carbon versus total profile stores. Piikki et al. (2013) reported
underestimation of soil carbon by 40-60% using solely remote sensing due to difficulties assessing
subsurface carbon. Hengl et al. (2017) concluded that integrated approaches are needed to improve
carbon mapping, given uncertainties in relating land cover to soil carbon balances and the
importance of environmental covariates like climate, topography and parent material. The potential
reasons for the weak correlation noted here are supported by the literature, including mismatches
between surface and profile carbon and the indirect nature of spectral indicators relying on biomass
proxies (Vågen and Winowiecki 2019). Recommendations for further analysis align with emphasis
on multi-source data integration and digital soil mapping advancements to strengthen carbon
prediction (Towett et al. 2015).

Figure 3. Correlation between the remote sensing organic matter and the lab organic matter result.
104
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Lastly, the carbon correlation analysis reflects consistent findings in the literature on the benefits
and limitations of remote sensing for soil carbon assessment, highlighting the particular importance
of integrating spectral data with soil analytics, terrain attributes, land use data and process-based
models to support carbon monitoring and management.

Figure 4. Correlation between the remote sensing organic matter and the lab organic matter result.

Correlation between the remote sensing phosphorus and the lab phosphorus data
Figure 5 revealed a significant negative correlation of -0.48 between the remote-sensing
phosphorus data and the laboratory phosphorus data. The moderate negative correlation (r=-0.48)
between remote sensing and laboratory soil phosphorus aligns with other studies demonstrating
challenges in using spectral vegetation indices to estimate plant-available phosphorus.

Mulder et. al. (2011) found a poor correlation between remote sensing data and soil test phosphorus
due to difficulties detecting complex soil phosphorus chemistry from leaf reflectance. Piikki and
Söderström (2019) reported an underestimation of Mehlich-3 extractable phosphorus by 80% using
solely remote sensing across agricultural fields in Sweden.

Hengl et. al. (2017) concluded that machine learning approaches combining remote sensing with
soil data, terrain attributes, geology maps, and land use improved the prediction of plant-available
phosphorus compared to spectral data alone. The negative correlation suggests reliance on indirect
plant phosphorus proxies from remote sensing is insufficient to capture dynamics of sorption,
precipitation, and labile phosphorus forms in the soil (Vågen and Winowiecki 2019). Integrating
targeted soil sampling and digital soil mapping techniques could potentially strengthen phosphorus
assessment noted by Towett et al. (2015).

The finding calls for further investigation to ascertain the fundamental reasons for the negative
correlation. It may indicate limitations in the accuracy of remote sensing techniques for assessing
phosphorus levels, or it could be influenced by other factors affecting the data. Understanding and
addressing the reasons for this negative correlation are essential for improving the reliability of
remote sensing-based assessments of phosphorus in soil.

105
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 5. Correlation between the remote sensing organic matter and the lab organic matter result.

Heat-map representation of the correlations among all the variables

A heat map is a powerful visual tool for representing the correlations among variables which is
also known as the “R” value table. The heat map visualization provides a clear overview of the
variable relationships between soil properties measured through remote sensing and laboratory
methods, as noted in other studies. The positive correlations for nitrogen, carbon, and organic
matter reflect the reliability of remote sensing for total concentrations of these parameters found
by Towett et. al. (2015) and Hengl et. al. (2017) in African agricultural soils.

In contrast, the negative correlations for phosphorus and potassium align with the literature on the
challenges of using spectral vegetation indices to estimate plant-available nutrient pools given
complex sorption dynamics (Mulder et al. 2011; Piikki and Söderström 2019).

Vågen and Winowiecki (2019) effectively used similar heat map matrices to represent validation
results between ground-based sensing and laboratory measurement of soil organic carbon and
texture fractions. The visualization format allows clear interpretation of correlations and
discrepancies essential for selecting appropriate remote sensing approaches for different soil
nutrients (Towett et al. 2015).

By summarizing multiple correlation analyses in one figure, the heat map enables the identification
of strengths and limitations across soil parameters to guide integrated data collection and analysis
strategies (Hengl et al. 2017). Conversely, a negative correlation is observed in the Phosphorus (P)
and Potassium (K) data. A negative correlation implies that as one variable increases, the other
tends to decrease. In essence, it means that there is a discrepancy or difference between the
measurements obtained through remote sensing and lab analysis for Phosphorus and Potassium.
This negative correlation could be indicative of some level of inaccuracy in the remote sensing data
for these specific soil properties or perhaps differences in how these properties are measured using
the two methods.

In practical terms, the positive correlations for Nitrogen, Carbon, and Organic Matter suggest that
remote sensing can be a valuable tool for assessing these soil properties, offering a time and cost-

106
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

effective alternative to laboratory analysis. However, for Phosphorus and Potassium, the negative
correlations highlight the need for further investigation into the reasons behind the discrepancies
and whether adjustments are necessary in the remote sensing methodology or calibration.

Figure 6. Correlations among all the variables.

CONCLUSION

This study demonstrated the potential of integrated laboratory and remote sensing techniques for
the comparative assessment of soil nutrients.

1. Laboratory and remote sensing techniques showed varying degrees of correlation and
accuracy for different soil properties. Strong positive correlations were found for carbon
and organic matter having r=0.23 and r=0.68. Weak positive correlation was seen for total
nitrogen having r=0.14. And poor negative correlations existed for phosphorus and
potassium having r=0.48 and r=0.42 respectively.
2. Remote sensing provided useful climate and environmental data to characterize the
cropping systems. But incorporation of additional climate variables could further improve
biophysical crop-soil system characterization.

107
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REFERENCES

A. K. Stum, Random forests applied as a soil spatial predictive model in arid Utah [M.S. thesis],
Utah State University, 2010.
Bielders, C.L., Gérard, B. (2015). Millet response to microdose fertilization in south–western
Niger: Effect of antecedent fertility management and environmental factors. Field Crops
Research, 171, 165-175.
Black C.A. (1965) method of soil analysis Agronomy No 9-part 2 American society Agronomy
Bouyoucos G.J. (1951). Hydrometer method improved for marking particle size analysis of soils.
Brady, N.C. and Well, R.R. (2002). The nature and properties of soil, 13th edition. Prentice Hall,
Brady, N.C., and R.R. Weil. 2000. Elements of the nature and properties of soils. 12th ed. Prentice
RaIl, New Jersey
Bray, R. H., and Kurtz, L.T. (1945). Determination of total organic and available phosphorus in
soils. Soil Science, 59: 39-45.
Bremner, J. M. (1996). Nitrogen total. In: D. L. Sparks et al. (Ed.). Methods of soil analysis, Part
3.Chemical Methods. Soil Science Society of America Book series. 5. Soil Science Society
of America and American Society of Agronomy. Madison, Wisconsin. p. 1085-1121.
Chi, Y.; Zhao, M.; Sun, J.; Xie, Z.; Wang, E. Mapping soil total nitrogen in an estuarine area with
highland scape fragmentation using a multiple-scale approach. Geoderma 2019,339, 70–84.
Graham, A. (2023). Evaluating correlation in Excel, SPSS Statistics and R. Statistics How To.
G. R. Blake, “Bulk density,” in Methods of Soil Analysis Agron, C. A. Black, Ed., Part I, No. 9.,
pp. 374–399, American Society of Agronomy, Madison, Wis, USA, 1965.
Hengl, T. Heuvelink, G.B., Walsh, M.G. (2017). Soil nutrient maps of Sub-Saharan Africa:
assessment of soil nutrient content at 250m spatial resolution using machine learning.
Nutrient Cycle Agroecosystem, 109, 77–102.
Mulder, V.L., de Bruin, S., Schaepman, M.E. and Mayr, T.R. (2011). The use of remote sensing in
soil and terrain mapping—A review. Geoderma, 162(1-2), 1-19.
Nelson, D. W., & Sommers, L. E. (1982). Total carbon, organic carbon, and organic matter. In A.
L. Page (Ed.), Methods of soil analysis. Part 2. Chemical and microbiological properties
(pp. 539-579). American Society of Agronomy, Soil Science Society of America.
Nelson, D.W., and L.E. Sommers. 1996. Total carbon, organic carbon, and organic matter. p. 961-
1010 In J.M. Bartels et al. (ed.) Methods of soil analysis. Part 3 Chemical methods. SSSA
Book Ser No.5. SSSA, Madison, WI. Nigeria. Geoderma, 48: 113-129.
Ofomata, G. E. K. (1987). Soil erosion in Nigeria: The views of a geomorphologist. University of
Nigeria Inaugural Lecture Series.
Padarian, J., Minasny, B. and McBratney, A.B. Using machine learning to predict soil organic
matter and particle size distribution from VIS-NIR spectra. Soil Tillage Res 195, 104400
(2019).
Piikki, K. and Söderström, M. (2019). Digital soil mapping of arable land in Sweden – Validation
of performance at multiple scales. Geoderma, 352, 342-350.
Piikki, K., Söderström, M., and Stenberg, B. (2013). Sensor data fusion for topsoil clay mapping.
Geoderma, 199, 106-116.
Qiusheng, W. (2009). What is Python and Why is it a Good Language Choice for Science and
Scientific Computing.
Towett, E. K., Shepherd, K. D., Sila, A., Aynekulu, E., & Cadisch, G. (2015). Modelling of soil
organic carbon stocks in the northern Rangelands of Kenya using remotely sensed covariates.
Soil Use and Management, 31(1), 81-92.
108
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Towett, E.K., Shepherd, K.D., Sila, A., Aynekulu, E., and Cadisch, G. (2015). Mid-infrared and
total x-ray fluorescence spectroscopy complementarity for assessment of soil properties. Soil
Science Society of America Journal, 79(5), 1375-1385.
Vågen, T. G., & Winowiecki, L. A. (2019). Predicting the spatial distribution and severity of soil
degradation in East Africa. Applied Geography, 94, 130-142.
Vågen, T.G., Winowiecki, L.A., Tondoh, J.E., Desta, L.T. and Gumbricht, T. (2016). Mapping of
soil properties and land degradation risk in Africa using MODIS reflectance. Geoderma, 263,
216-225.
Virtanen, P., Gommers, R., Oliphant, T.E. et al. SciPy 1.0: fundamental algorithms for scientific
computing in Python. Nat Methods 17, 261–272 (2020).
Zhang, M.; Shi, W. Systematic comparison of five machine-learning methods in classification and
interpolation of soil particle size fractions using different transformed data. Hydrol. Earth
Syst. Sci. Discuss.2019,2019,1–39.

109
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

LAND SUITABILITY PREDICTION FOR MAIZE PRODUCTION IN SOUTHWEST

NIGERIA USING GEOGRAPHICAL INFORMATION SYSTEM AND MOST-
LIMITING SOIL NATIVE FERTILITY FACTORS

1Taiwo S. Olutoberu, 1Mutiu A. Busari, 1Olusegun Folorunso, 1Muftau Adebayo, 1Sodeeq O.

Azeez, 1Sarafadeen G. Hammed, 1J.A. Oyedepo, 2Oluwafolake Ojo, and 1G.A. Ajiboye
1Federal University of Agriculture, Abeokuta, Nigeria
2Osun State University, Osogbo, Nigeria
e-mail: [email protected]; +234 8039658671

ABSTRACT

Maize yield per hectare in Southwest Nigeria has been found to be one of the lowest compared to
other regions of the world. Digital land suitability assessment was carried out using indices of
most-limiting soil native fertility and geographical information systems. We explored the
combined use of continuous soil attributes modeling, and ordinal logistic regression through a two-
stage mapping process to accomplish the land suitability assessment for maize production.
Stepwise forward regression analysis of environmental covariates was conducted to reduce the
number of predictors to only those having significant effects at 95% confidence interval. Most area
from northern to southern part of the study area had organic carbon content less than 1%. Larger
portion of the study area northern part had soil native total nitrogen below 0. 1 g/kg. Most portion
of the study area had extractable phosphorus between 23 and 28 mg/kg while the northern part had
extractable potassium between 0.29 and 0.33 g/kg. Only some area around southern part of the
study area had pH less than or equal 5.5 while other parts had values above 5.5. According to FAO
land suitability classification systems, 69.078% of the landmass is moderately suitable, 29.865%
is unsuitable and only 1.056% is suitable for maize production. We recommend that policy should
be enacted and implemented to regulate infrastructural development and protect agricultural land,
non-regenerative agricultural practices should be discouraged, and government and private sectors
should empower the agrarian communities with modern soil management training and subsidized
farm inputs

INTRODUCTION

Youths in Southwest Nigeria are constrained by lack of access to scientific and technological
information that could enhance maize production capacity (Olaniyi and Adewale, 2012). Although
they are said to be the future farmers who are expected to carry on farming as a profession for
sustainable food production in the Country. Fawole (2008) posited that clamor for adoption of
innovation in agricultural development may not be justified without availability and dissemination
of innovative information. Olaniyan also posited that increase in maize production in Nigeria has
been achieved greatly by expansion in area cultivated rather than increase in yield. The author
further stated that the area cultivated and harvested increased from 2.8 million hectares in 1986 to
over 3 million hectares in 2000 and over 6 million in 2010. However, the average yield of maize
in Nigeria is 1.68 tons/hectare while it is average 9.3 t/ha in the United States of America over the
same period (ATA, 2011)

110
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Previous studies on maize production in Southwest Nigeria have done justice to the socio-
economic aspect of maize production but no study has investigated the spatial-temporal land
suitability evaluation of Southwest Nigeria for maize production (otherwise known as digital soil
assessment) to achieve optimum yield and hence maximum profitability and to give credence to
the principle of comparative advantage.

According to Akinbode et al. (2024), the production of systematic digital soil fertility mapping in
Nigeria is of urgent national emergency in this era of digital advancement and precision agriculture
as such has not been previously conducted for the optimum utilization of farm resources towards
improving farm productivity. Until now, the country has maintained its conventional soil maps.
However, digital soil mapping provides in-situ real-time information about the soil in a given
location. Hence, it assists farmers’ decision-making and impacts positively on agricultural
productivity. This study is therefore set up to investigate holistically temporal and spatial land
suitability of Southwest Nigeria for maize production. We explored the combined use of
continuous soil attributes modeling and ordinal logistic regression through a two-stage mapping
process to accomplish the land suitability assessment for maize in the Southwest. This way all
covariate factors necessary for the holistic evaluation such as soil, climate, organism, relief, parent
materials, age, and spatial position have been included as predictors. The results of this study will
have implications for the cultivation of land for maize production in Southwest Nigeria in line with
the sustainable development goals 2 (zero hunger), 11(sustainable cities and communities), and 12
(responsible consumption and production) of the United Nations in Nigeria. This study answered
the following research questions – 1) What is the soil’s native fertility status in Southwest Nigeria
in terms of composite soil nutrients? 2) What is the status of environmental covariates in the study
area? And 3) Which region within Southwest Nigeria is best suited for maize production?

The null hypotheses of this study are:

H10: The native soil nutrients of any two randomly selected geographical locations within
Southwest Nigeria are not significantly different from each other.

H20: Any two randomly selected geographical locations within Southwest Nigeria are not
significantly different from each other in terms of suitability for optimum maize production.

H30: Soil native nutrients have no statistical association with maize production in Southwest
Nigeria.

MATERIALS AND METHODS

The study was carried out in six states of Southwest Nigeria comprising Ekiti, Lagos, Ogun, Osun,
Ondo, and Oyo States between latitude 5ο N and 9 ο N of the Equator and longitudes 2.5 ο and 6 ο
East of the Greenwich Meridian. It is bounded by the Atlantic Ocean in the South, Kwara, and
Kogi states in the North, Anambra state in Eastern Nigeria, and the Republic of Benin in the West
(Fig.1). The study area has a land area of about 114,271 km2 representing about 12 percent of the
country's total land area (Olaniyi and Adewale, 2012). The climate in southwestern Nigeria is
predominantly humid with rainfall from 1500mm to 3000mm per annum. The mean monthly
temperature ranges from 18 οC to 24 οC during the rainy season and 20 οC to 35 οC during the dry
season (Sahib et al,1997).
111
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 1. Maps of the study area. Source: Akinbode et al. (2024).

The 30 m resolution land use and land cover map of the study area from 1st January 2022 to 31st
January 2023 was downloaded from USGS website and was classified by supervised learning using
the maximum likelihood algorithm in Qgis version 3.26.3 (QGIS Development Team, 2021) (Fig.
2). Then stratified sampling method was adopted at 10 km sampling distance within each stratum
like Wang et al (2022). However, some sampling points were shifted to nearby distances from
their georeferenced points when they fell on watersheds, water bodies, built-up areas, or road
networks. At every georeferenced point sampled, the quadrant method of sampling was adopted
by taking 3 composite samples at 0 – 40 cm depths in each quadrant which were then thoroughly
mixed and from which representative sample was taken into a black cellophane sample bag which
was properly labeled with the location’s unique number identifier. Environmental and biophysical
covariates used as predictors include precipitation, annual temperature, elevation, hill shading,
terrain wetness index (TWI), topography positioning index (TPI), altitude above channel network
(AACN), gamma radiometric potassium (radK), normalized difference vegetation index (NDVI),
Mid slope position, slope, light insolation, and terrain ruggedness index. Also, the multiresolution
index for valley bottom flatness (MRVBF) and multiresolution index for terrain top flatness
(MRTTF) were included in the predictor variables because many areas within Southwest Nigeria
are characterized by undulating landscapes. Soil native nutrients such as soil total nitrogen,
phosphorus, potassium, organic matter content, and pH were the target or dependent variables.
Soil total N was determined using the micro Kjeldhal method (Brememer and Mulvancy, 1982).
Available P was analyzed using Bray-1 P extractant and determined colorimetrically by the
molybdenum blue procedure. Exchangeable cations were extracted using 1M Ammonium Acetate
pH 7.0 and the potassium in the extract was determined by atomic absorption spectrophotometer
(AAS). Continuous soil attribute modeling, regressions, and land suitability modeling were carried
out using R statistical software version 4.3 (R Core Team, 2019)

112
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 2. Southwest Nigeria land use and land cover.

RESULTS AND DISCUSSION

The translation of digital soil mapping outputs of soil native nutrients and climatic conditions,
framed by the contextual expert-value system that addresses the optimum conditions for maize
production, revealed that Southwest Nigerian soil fell only within 3 FAO land suitability categories
(FAO, 1976) for maize production. These include Moderately suitable, suitable, and unsuitable
(Fig. 3). The largest portion of Southwest Nigerian soil (69.078%) was within the moderately
suitable category for maize production. This suitability category was found in all cardinal locations
of the study area extending from north to south and east to west (Fig. 3). This agrees with the
findings of Olaniyan who posited that an increase in maize production in Nigeria has been achieved
greatly by expansion in area cultivated rather than increase in yield. The author further reiterated
that Nigeria had to commit over 3 million hectares of land in 2000 which it later increased to 6
million hectares in 2010 to the production of maize as against 2.8 million it committed in 1986 to
have higher quantitative yield. Such a huge land resource could have been allocated to more
profitable crop enterprises if a land suitability assessment like this had been carried out. However,
Olaniyan (2015) advocated that maize production in such areas could improve if fertilizer, land,
and subsidized education could be provided by the government, private investors, and non-
governmental organizations.

Larger percentage (29.865%) of the study area was unsuitable for maize production using the
indices of soil native nutrients and climatic conditions fed into a computer algorithm that was

113
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

guided by agronomist prescriptions. The unsuitable categorized soil was also found localized in
different parts of the study area. (Fig.3)

Figure 3. Land suitability map for maize (Zea mays. Linn) production in Southwest Nigeria using
geographical information system and most-limiting soil nutrients.

The superimposition of land suitability prediction maps (Fig. 3) on initial land use and land cover
maps revealed that the bulk of this unsuitable land fell in built-up areas, flooded vegetation,
and heavily tree canopy-covered areas while some also fell within heavily crop-cultivated regions
(Fig. 2). This revelation that some FAO classified unsuitable areas fell within built-up areas attests
to the fact that the quest for urbanization which takes up space for residential, road networks and
industrial development could render some land unsuitable. In a study conducted to evaluate the
impacts of urbanization in Nigeria Makurdi town, Yusuf et al. (2020) reported that 336 km 2
representing 32% of the total landmass of the study area was taken up by built-up area while 200
km2 representing 19% of the agricultural land was lost to urbanization.

Observation also revealed that only a few regions of the study area (1.056%) were suitable for
maize production using the indices of soil native nutrients and climatic conditions. The suitable
regions appeared as spots scattered at different parts of the study area. Most of the suitable land
was observed as a cluster at the boundary of Osun State with Ekiti State while others were found
as fragments of land in other regions.
114
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

CONCLUSIONS

The findings of this study revealed that the largest part of Southwest Nigeria is moderately suitable
for maize production while the larger part is unsuitable. However, a few regions of the study area
are suitable using the indices of native soil nutrients and climatic conditions. Some of the
unsuitable regions were found in built-up areas suggesting indiscriminate and legitimate activities
of man competing for agricultural land. Government can control this by putting regulations in
place to prevent indiscriminate development and protect Agricultural land. Also, provision of
modern soil management training and supply of farm input like organic fertilizer could encourage
agrarian communities in regenerative agricultural practices that can improve the nutrient status of
moderately suitable lands.

Moreover, digital land suitability assessment like this should be encouraged in other regions of the
Country not only to produce maize but also for other crops to ensure adequate allocations of land
resources for optimum yield and to comply with sustainable development goals of zero hunger,
sustainable cities and communities, responsible consumption and production of United Nations.
This study advocates for more research in digital land evaluation for sustainable agricultural
production while acknowledging that both the resolution of data and methodology of this research
could be improved upon in future studies

REFERENCES

Agricultural Transmission Agenda (ATA) (2011). Maize-Soybean Transmission Action Plan.

Federal Ministry of Agriculture and Rural Development, Abuja, Nigeria.
Akinbode, S.O., Folorunso, O., Olutoberu, T.S., Olowokere, F.A., Adebayo, M., Azeez, S.O.,
Hammed, S.G., and Busari, M.A. (2024). Farmers’ Perception and Practice of Soil Fertility
Management and Conservation in the Era of Digital Soil Information Systems in Southwest
Nigeria. Agriculture. MDPI. 14, 1182. https://doi.org/10.3390/ agriculture14071182
Bremmer, J.M.; Mulvaney, C.S. (1982). Nitrogen-Total. Methods of soil analysis. Part 2. In:
Chemical and Biological Properties; Page, A.L., Miller, R.H., Keeney, D.R., Eds.; American
Society of Agronomy, Soil Science Society of America: Madison, WI, USA, pp. 526–595.
Fawole, O.P. (2008). Pineapple farmers' information sources and usage in Nigeria. Bulgarian
Journal of Agricultural Sciences, 14(4): 381 – 389.
Food and Agriculture Organization (1976). A framework for land evaluation. Soils Bulletin, vol
32. Food and Agriculture Organisation of the United Nations, Rome
Olaniyan, A. B. 2015. Maize: Panacea for hunger in Nigeria. African Journal of Plant Science, 9
(3). Pp 155 -174. Doi: 10.5897//AJPS2014.1203
Olaniyi, O. A. and Adewale, J. G. (2012). Information on maize production among rural youth: A
solution for sustainable food security in Nigeria. Library Philosophy and Practice, ISSN
1522-0222. Doi: http://unllib.unl.edu/LPP/
R Core Team, (2019). R: A Language and Environment for Statistical Computing. URL. R
Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/
QGIS Development Team (2021). QGIS Geographic Information System. Open Source
GeoDPatial Foundation Project. http://qgis.osgeo.org.

115
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Wang W, Duan M, Zhang X, Song X, Liu X, Cui D. (2022). Determining Optimal Sampling
Numbers to Investigate the Soil Organic Matter in a Typical County of the Yellow River
Delta, China. Applied Sciences. 12(12):6062. https://doi.org/10.3390/app12126062
Yusuf Mohammed Bakoji, Elijah Elizabeth, Anita Humshe Philip, Mohammed Salim Isa, Umar
Jauro Abba. (2020). Urbanization and its Impact on Agricultural Lands in Growing Cities in
Developing Countries: A Case Study of Makurdi, Benue State Nigeria. International Journal
of Economy, Energy and Environment. Vol. 5, No. 4, 2020, pp. 41-46. doi:
10.11648/j.ijeee.20200504.11

116
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ON-FARM EXPERIMENTATION

117
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ON-FARM EXPERIMENTATION PROCESS TRIGGERS FARMERS’ ZEAL TO TEST

TECHNOLOGIES IN MAIZE SYSTEMS OF EMBU COUNTY, KENYA
#11665

1*Onesmus Kitonyo, Evans Chimoita, Timothy Kamanu, Felister Nzuve, Esther Muindi, Alfred
Micheni, James Muthomi, Vincent Kathumo, Grace Mureithi, George Chemining’wa
2Department of Plant Science and Crop Protection, University of Nairobi, Nairobi, Kenya;
2Department of Agricultural Economics, University of Nairobi, Nairobi, Kenya; 3Department of

Mathematics, University of Nairobi, Nairobi, Kenya; 4Department of Crop Sciences, Pwani

University, Kilifi, Kenya; 5Kenya Agricultural and Livestock Research Organization, Embu,
Kenya. 6Mount Kenya University, Thika, Kenya; 7BRITS Agribusiness & Farmers Call Centre,
Nairobi, Kenya.
*e-mail: [email protected]; +254721434412

ABSTRACT

Using on-farm experimentation (OFE) approach, this study was carried out to validate a package
of soil moisture and fertilizer nitrogen management practices, and to track farmer adoption of better
agronomic practices in maize systems of Embu County, Kenya. Crop residue mulch in combination
with calcium ammonium nitrate fertilizer, and a soil conditioner (hydrogels) coupled with slow-
release urea were validated against farmer practices. Both management practices increased maize
grain yield compared with farmers’ practice. Stakeholders observed that these practices improved
plant density and vigour, increased grain yield, reduced weed and pest pressure. The OFE process
facilitated quick adoption and testing of technologies by farmers. At the onset of the third
experimentation season, farmers began to experiment on a range of practices, especially mulching
and optimal plant density.

INTRODUCTION

Despite enormous investment in research to improve the productivity of maize systems of Embu
County, farmers hardly adopt high yielding agronomic practices. Low adoption could partly be
attributed to approaches used in the research process. Researchers have traditionally used on-
farm experiments to generate data but without the involvement of the farmer, either at design of
experiments, data collection or interpretation of the results (Kummer et al., 2017). To improve
bridge the gap in knowledge generation and transfer, and promote innovation by both researchers
and farmers, and other stakeholders, it is important to rethink the way experiments are conducted
(Richardson et al., 2021). Besides the large pool of stakeholders in co-creation of knowledge,
OFE creates value proposition that distinguishes it from other participatory approaches in
research. Often, this value arises from farmers being able to access information they can trust
(Lacoste et al., 2022). To accelerate farmer experimentation and innovation, this study co-
designed experiments with farmers and stakeholders to validate water and nitrogen management
practices in maize systems of Embu County, Kenya.

118
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

METHODS

On-farm experiments (OFE) were carried out in two environments in the maize growing region
of Embu County, in eastern Kenya. The OFE sites were in the upper midland (UM) zones UM3
and UM4, and lower midland (LM) zones LM3 and LM4. Prior to the establishment of
experiments, farmers engaged in focus group discussions with researchers to identify relevant
management practices for improved productivity of maize. In a bottom-up consultation process,
farmers prioritized fertilizer and soil moisture management as the most pressing issues.
Subsequently, farmers and researchers co-designed treatment combinations that could address
the identified problems. Due to the large pool of treatments, management packages were
designed in two distinct plots. The highest best management package (BMP1) comprised the use
of soil conditioners (hydrogels) and a slow-release nitrogen (N) source of ‘KynoPlus S®’ while
the next highest management package (BMP2) consisted of the application of 3 t/ha crop residue
as mulch and calcium ammonium nitrate (CAN) as N source. The two researcher-managed plots
were compared with farmers’ business as usual plots. However, it was agreed that practices
would change from season to season depending on experiences gathered. In this case, based on
learning from the first experimental cycle, treatments were amended during the second season
to include a uniform application of 5 t/ha of manure in both BMP1 and BMP2, and the farmer
continued business as usual operations but with integration of knowledge from the OFE process.

While plot sizes varied from farm to farm, BMP1 and BMP2 were each allocated at least 900
m2, a size that is comparable with farmers’ plots. Data were collected in researcher-managed
(BMP1 and BMP 2) and farmer business as usual plots. Prior to harvesting, experiment host
farmers, neighbours and other stakeholders were invited to evaluate the performance of the
experiments. Farmers were asked to select preferred treatment plots based on their own criteria.
The farmers were given three categories of choice per treatment plot, either poor performance,
average performance or best performing treatment. The selection exercise was followed by a
dialogue to document the criteria applied and farmer perception about the demonstrated
management practices.

RESULTS

Yield and farmer selections

Figure 1 presents results from farmer evaluation of experimental plots. Overall, farmers’ plots
were least preferred by the respondents.

These selections were a true reflection of crop yield performance. Generally, plots applied with
hydrogels outperformed those treated with applied with crop residue. However, in UM3/4 sites,
differences in grain yield between hydrogel and mulched plots were small, and sometimes not
significant. However, in the drier LM3/4 sites, hydrogel plots consistently and largely out-
performed mulched plots.

Learning and evolution of farmer practice

Figure 2 shows the evolution of farmer learning from business-as-usual operations to the
implementation of better management practices for improved water and nitrogen management.
Farmers provided diverse feedback on their learning and presented a range of practices they were
willing to test and implement in their plots in the ensuing seasons. At the on-set of the third
119
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

experimentation season (2023 short rains), majority farmers implemented at least one practice
learnt from the engagement with the project. Although farmer perception was not measured,
farmers demonstrated confidence with the experimentation process and trusted the results.
Indeed, more farmers were enthusiastic to either join the project or test technologies in their
farms.

Figure 1. Farmers’ selection of management practices during 2022 short rains (season 1) and 2023
long rains (season 2) in lower midland zones (a) and the upper midland zones (b). ‘Mul’ denotes
mulched plots, ‘hyd’ indicates plots with hydrogels and ‘FP’ is farmer’s practice.

Figure 2. Evolution of farmer management practices as the project entered the third season.

120
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

DISCUSSION

The two researcher-managed plots optimized crop management practices unlike in the farmer
plots where there were delays in weeding, fertilizer application and pest control. However, the
study did not measure significant differences between the two plots. This implies that both
hydrogels and mulch potentially conserved soil moisture at similar efficiency. Similarly, the
application of calcium ammonium nitrate or the slow-release nitrogen fertilizer formulation did
not show differences in maize yield. However, based on the unit price of nitrogen in each
formulation, gross margin analyses (not shown) pointed to significantly higher returns per unit
area with the use of slow-release fertilizer compared with calcium ammonium nitrate.
Nonetheless, either of the fertilizer formulation ought to be applied at an optimal rate, at the right
crop stage and placed near the root zone to maximize plant uptake (Bruulsema, 2021).

Adoption of residue retention among smallholder farmers, and especially those in mixed crop-
livestock systems of Embu is constrained by the competing uses of crop residue (Jaleta et al.,
2012; Baudron et. Al., 2014). In Embu, crop residues are primarily used as animal feed or sold
improve household incomes. Through the OFE project farmers evaluated the benefits of mulch
in improving maize yield, an outcome that fundamentally changed the farmers’ mindset in the
allocation of more crop residue to conserve moisture. Improved moisture conservation would
open a window for better utilization of nutrients and reduce drought stress. In addition, farmers
learnt the importance of better agronomic practices to improve maize yield. Key among the
practices, farmers are willing to experiment are early planting, optimal plant density, early
weeding, optimal fertilization based on the weather outlook, and use of manure.

In this study, the OFE approach accelerated knowledge transfer and practice change. This was
demonstrated in the ability of farmers to take only two seasons of experimentation to start to
adopt and test weather-resilient management practices such as mulching of soils with crop
residue. This was a significant shift from the status quo where farmers remove crop residue for
livestock feed or sale.

REFERENCES

Baudron, F., Jaleta, M., Okitoi, O., & Tegegn, A. (2014). Conservation agriculture in African
mixed crop-livestock systems: expanding the niche. Agriculture, Ecosystems &
Environment, 187, 171-182. https://doi.org/10.1016/j.agee.2013.08.020
Bruulsema, T. (2022). Nutrient stewardship: Taking 4R further. Crops & Soils, 55(1), 34-40.
https://doi.org/10.1002/crso.20165
Jaleta, M., Kassie, M., & Shiferaw, B. (2013). Tradeoffs in crop residue utilization in mixed
crop–livestock systems and implications for conservation agriculture. Agricultural
Systems, 121, 96-105. https://doi.org/10.1016/j.agsy.2013.05.006
Kummer, S., Leitgeb, F., & Vogl, C. R. (2017). Farmers’ own research: Organic farmers’
experiments in Austria and implications for agricultural innovation systems. Sustainable
Agriculture Research, 6(526-2017-2663). https://dx.doi.org/10.5539/sar.v6n1p103
Lacoste, M., Cook, S., McNee, M., Gale, D., Ingram, J., Bellon-Maurel, V., ... & Hall, A. (2022).
On-Farm Experimentation to transform global agriculture. Nature Food, 3(1), 11-18.
https://doi.org/10.1038/s43016-021-00424-4

121
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Laub, M., Corbeels, M., Ndungu, S. M., Mucheru-Muna, M. W., Mugendi, D., Necpalova, M.,
... & Six, J. (2023). Combining manure with mineral N fertilizer maintains maize yields:
Evidence from four long-term experiments in Kenya. Field Crops Research, 291, 108788.
https://doi.org/10.1016/j.fcr.2022.108788
Mwaura, G. G., Kiboi, M. N., Bett, E. K., Mugwe, J. N., Muriuki, A., Nicolay, G., & Ngetich,
F. K. (2021). Adoption intensity of selected organic-based soil fertility management
technologies in the Central Highlands of Kenya. Frontiers in Sustainable Food Systems, 4,
570190. https://doi.org/10.3389/fsufs.2020.570190
Richardson, M., Coe, R., Descheemaeker, K., Haussmann, B., Wellard, K., Moore, M., ... &
Nelson, R. (2022). Farmer research networks in principle and practice. International
Journal of Agricultural Sustainability, 20(3), 247-264.
https://doi.org/10.1080/14735903.2021.1930954
Toffolini, Q. & Jeuffroy, M. (2022). On-farm experimentation practices and associated farmer-
researcher relationships: a systematic literature review. Agronomy for Sustainable
Development, 42 (144), 1-16. https://doi.org/10.1007/s13593-022-00845-w

122
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

THE VALUE AND POTENTIAL OF ON-FARM EXPERIMENTATION TO CATALYZE

AGRICULTURAL TRANSFORMATION
#11253

1*Ivan S. Adolwa, 2Eileen Nchanji, 1Joses Muthamia, 2Cosmas Lutomia, 3Simon Cook
1African Plant Nutrition Institute (APNI), Nairobi
2Bioversity International & CIAT, Nairobi
3Murdoch University, Murdoch, Australia
*e-mail: [email protected]; +254799657652

ABSTRACT

On-farm experimentation (OFE), which inculcates farmers’ agency in knowledge discovery, has
the potential to support and accelerate transformative agronomy at scale. The OFE process within
the Nutrient-Catalyzed Agricultural Transformation in Africa (NUTCAT) project, encompasses
farmer engagements, set-up of simple, easy-to-understand treatment designs, and contextual
analysis of the data to enhance the relevance of the results to farmers. Ultimately, it is envisaged
that this process will unpack the potential of precision nutrient management (PNM) to improve
cereal system production. In this study we focus on the value propositions of OFE for maize-mixed
systems of Kenya. To understand the value of OFE we applied the value proposition canvas (VPC)
tool on three customer segment categories: Farmers, Service providers and Researchers. We also
sought to understand how OFE could be embedded in existing agricultural innovation systems and
contexts in which farmers operate. To this end, a survey was implemented to socioeconomically
characterize Kenyan maize-mixed farming systems. In terms of the value of OFE, we find that
maize farmers are the customer segment that stands to benefit most from the NUTCAT OFE.
Farmers derive functional value in terms of high yields and income obtained, but also personal
value which is realized through learning and internalization of agronomic concepts. It was observed
that farmers value change processes that are holistic with targeted interventions across the value
chain. This implies that for OFE to effectively take root in current innovation systems there is a
need to explore several entry points beyond plant nutrition interventions. Analysis of the survey
data shows that farmers operate under less-than-ideal conditions where fertilizers are costly and
supporting institutional structures tend to be ineffective. For instance, although there have been
efforts to provide subsidized inputs only a small share of farmers is accessing them. Nevertheless,
at least a third of farmers seek and test for relevant solutions to overcome some of the obstacles
they face in their farm enterprises. This goes to show that OFE can play a role in strengthening
farmer agency to lead the innovation process and transform agricultural landscapes.

Keywords: Value proposition, on-farm experimentation, precision nutrient management

INTRODUCTION

The innovation system in Africa is still characterized by top-down or linear approaches that largely
have stifled farmer agency thus contributing to agricultural stagnation. There has been an overt
focus by Agricultural Research for Development (AR4D) organizations to prescribe blanket
nutrient management recommendations to beneficiary farmers (Zingore at al. 2022). A substantial
amount of participatory work has been done with farmers to understand underlying driving factors

123
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

to their decision-making, but this is yet to bear fruits in terms of increasing their innovative
capacity. On-farm experimentation (OFE) is disruptive to this counter-productive process as it
brings agricultural stakeholders together around mutually beneficial experimentation to support
farmers’ own management decisions (Lacoste et al. 2022). Given that OFE is farmer-centric,
farmers are not only passive recipients of technologies but are also experimenters, hence are central
to the innovation processes. The experiments are conducted at scales that include effects of
variations and mimic local conditions as far as possible (Cook et al. 2018). In addition, OFE is
characterized by evidence-driven (standardized data protocols), expert-enabled (added value
through scientific engagement), co-design (of experiments), and scaling by co-learning (sharing of
data, insights, or ideas) principles (Lacoste et al. 2022).

An introspection of the relevant literature ascribes value of agricultural technologies in terms of

productivity and household income. There is evidence that technologies such as fertilizers and
improved seeds have contributed significantly to increased productivity or incomes of smallholder
farmers (Khonje et al. 2015). Nevertheless, the supposed benefits of these technologies have still
not been adequate to support their widespread adoption. An interesting perspective is drawn from
the re-definition of ‘value’ by Christensen et al. (2016). They define the value of a product or
service as the ability to accomplish the intended task of the beneficiary at a specific point in time.
Instead of focusing on the attributes of the beneficiary or ‘customer’, it is more crucial to
concentrate efforts in establishing what the ‘customer’ hopes to accomplish. Hence, borrowing
from this definition we define ‘value’ as the ability of OFE to enable farmers (or other
beneficiaries) to complete ‘jobs to be done’. This paper explores the value propositions of OFE for
maize-mixed systems of Kenya to understand the value of OFE to different customer segments
(farmers, service providers and researchers) and how it could be embedded in existing agricultural
innovation systems.

MATERIALS AND METHODS

The value proposition canvas (VPC) was used to garner data on the value of the OFE process from
the perspective of the farmers, service providers and researchers (Osterwalder et al. 2015). The
value proposition (VP) concept and set of tools is aimed at creating products and services that
customers want. In our case the ‘customers’ were i) farmers who participated in the OFE process
and those that were not a part of the process; ii) service providers, mainly the extension agents,
non-governmental organizations and any other actors involved in sharing information and
knowledge; and iii) researchers drawn from academia and international organizations. The VP is
a useful methodology and approach that furnishes tools to help in creating value for the customers,
helps us to learn what customers want, trains focus on customers rather than on technologies,
products and features, and helps one work with clear processes and tools. Hence, the VPC consists
of the customer profile (‘understanding your customer’) and the value map (‘how to create value
for the customer’). Finally, you try to see where the two parts fit.

A one-day workshop was organized in September 2023 that brought together different ‘customer’
segments to ascertain the value they derived from the OFE process. The participants were taken
through the tool so that could effectively partipate in the exercise. This was followed by interactive
sessions with participants in the late morning and early afternoon sessions where the customer
profile was defined, the value map described and finally the value proposition for OFE determined
and reviewed (Figure 1). The forty-one attendees included 15 farmers with the rest drawn from 9
124
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

organisations including OCP-Africa, APNI, Kenya Agricultural Livestock Research Organization

(KALRO), Tupande by One Acre Fund, Cereal Growers Association (CGA), University of Nairobi,
Pwani University, County Government extension, and the Alliance of Bioversity and CIAT.

RESULTS AND DISCUSSION

The beneficiaries or customer segments of NUTCAT OFE were categorized into four: i. NUTCAT
farmers ii. Non-NUTCAT farmers iii. Service providers iv. Researchers. On application of the VPC
tool, it was clear that farmers were the main beneficiaries, but with service providers and
researchers benefiting as well albeit to a smaller extent.

Figure 5. Interactive session to define customer profiles of service providers (top left), farmers
(top left) and researchers.

In terms of gains, for farmers functional gains were the most important. Farmers learned from the
OFE process on best management practices e.g., crop density, proper use of fertilizer. Farmers
valued the higher income earned because of increased productivity. They were well motivated to
continue their farming enterprise due to anticipated yield increases. For instance, one NUTCAT
farmer reported getting a revenue of US$ 1500 from maize sales. Farmers derived personal value
from NUTCAT OFE given that they were able to learn and internalize new agronomic concepts
ultimately building their human capital. This is important as it improves the management skills of
farmers enabling them to be more adept at innovating. Also, personal value was derived when
there was an improvement in social status. Farmers, for instance, indicated that one gain from the
OFE process was an increase in social standing in the community as they were able to pay their
children’s school fees or go to good hospitals.

125
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

In terms of inputs and treatments, the inputs OFE works with included fertilizers (DAP, CAN,
MoP, Urea, Urea-S, Urea-S Zn), hybrid seeds (Pannar, Duma, DK8033, DK8031), farmyard
manure, herbicides, and pesticides. The treatment OFE works with was precision nutrient
management (PNM), which emphasizes the efficient and appropriate use of fertilizers. It ialso
focuses on site specific nutrient management considering spatial (and temporal) variability. There
is also the aspect of sustainable intensification that includes elements of soil and water conservation
e.g., use of furrows, and integrated soil fertility management (ISFM). The intent of the OFE process
was to increase efficiency of applied inputs through the application of PNM, which in this case is
undergirded by 4R Nutrient Stewardship principles i.e., using the right fertilizer source, planting at
the right time, right rate, and using the right application method. Substitution took place as well
with farmers having stopped using recycled seeds and now using certified, high vigour ones. There
was also the intent to redesign the farming system using soil and water conservation technologies
to help conserve soil moisture. From a science standpoint, the OFE process entailed the collection
of agronomic data (mainly crop cuts). Agronomic principles are adhered to especially where the
input of the scientist was prominent. This would entail use of hybrid seed varieties, recommended
spacing (crop density), gapping, weed control, and pest control. These agronomic practices were
documented for both the scientist-led (OT) and the farmer-led (FP) treatments. All the sampling
points (9-36 per field) for the crop-cuts were georeferenced. This was important for garnering
spectral data that could be used to generate imageries and yield maps and be correlated analytically
with measured yields.

90
80
70
60
50
40
30
20
10
0
Kenya Tanzania Ghana Ivory Coast

Lack of cash for inputs Susceptible to pests/diseases Lack of labor

Low yielding variety Cannot get credit None

Figure 6. Major farm constraints across countries.

Social or qualitative data is collected at engagement events with farmers and other stakeholders
e.g., post-harvest dialogue meetings. Focus group discussion check lists and interviews are applied.
Monitoring is done throughout the season to track farmer learning and uptake of innovations.
Detailed surveys were also conducted to unravel the socio-economic context in more depth. Key
survey results show that farmers operate under less-than-ideal conditions where fertilizers are
costly and supporting institutional structures tend to be ineffective (Figure 2). Despite efforts to
126
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

provide subsidized inputs only a small share of farmers are accessing them. Nevertheless, at least
a third of sampled farmers are experimenting with various options to overcome some of the
obstacles they face in their farm enterprises (Table 1).

Table 5. Level of farmer-led experimentation and knowledge generation in Kenya.

Experimentation
Farmer experimenting /testing solutions (%) 28.10%
Farmers have sought for solutions (%) 33.10%

CONCLUSIONS

The OFE process in principle portends substantive value for farmers. Nevertheless, OFE has not
addressed financial literacy and record keeping needs of farmers. This is important if they are to
derive maximum value from the process as they need to know if their farming enterprise is
profitable. There other issues related to translational uncertainties e.g., access to good, certified
seeds that need to be addressed to further unlock OFE. There is need to strengthen extension given
the high farmer to extension ratio. Gender streamlining, which is important for service providers
and researchers, may have a positive spillover on the OFE process. Government support is crucial
for researchers to derive benefits from an OFE process and to alleviate some of the infrastructural
constraints that generally hinder the change process.

REFERENCES

Christensen, C.M., Hall, T., Dillon, K. and Duncan, D.S., 2016. Know your customers’ jobs to be
done. Harvard business review, 94(9), 54-62.
Cook, S., Lacoste, M., Evans, F., Ridout, M., Gibberd, M., and Oberthür, T. 2018. An On-Farm
Experimental Philosophy for Farmer-centric Digital Innovation. Proceedings of the 14th
International Conference on Precision Agriculture.
Lacoste, M., Cook, S., McNee, M., Gale, D., Ingram, J., Bellon-Maurel, V., MacMillan, T.,
Sylvester-Bradley, R., Kindred, D., Bramley, R., Tremblay, N., Longchamps, L., Thompson,
L., Ruiz, J., García, F. O., Maxwell, B., Griffin, T., Oberthur, T., Huyghe, C., … Hall, A.
2022. On-Farm Experimentation to transform global agriculture. Nature Food, 3(1), 11–18.
https://doi.org/10.1038/s43016-021-00424-4
Khonje, M., Manda, J., Alene, A.D., and Kassie, M., 2015. Analysis of adoption and impacts of
improved maize varieties in eastern Zambia. World Dev. 66, 695–706.
Osterwalder, A., Pigneur, Y., Bernarda, G. and Smith, A., 2015. Value proposition design: How to
create products and services customers want. John Wiley & Sons.
Zingore, S., Adolwa, I.S., Njoroge, S., Johnson, J.-M., Saito, K., Phillips, S., Kihara, J., Mutegi, J.,
Murrell, S., Dutta, S., Chivenge, P., Amouzou, K., Oberthür, T., Chakraborty, S., and Sileshi,
G.W. 2022. Novel insights into factors associated with yield response and nutrient use
efficiency of maize and rice in sub-Saharan Africa. A review. Agronomy for Sustainable
Development, 42, 82.

127
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

PRECISION LIVESTOCK MANAGEMENT

128
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

MONITORING GRAZING GOATS' BEHAVIOR USING SENSORS AND SATELLITE

REMOTE SENSING
#11267
1Y.Chebli, 1S. El Otmani, 2F. Elame, 3J.F. Cabaraux, 4A. Keli 4, 1M. Chentouf
1Regional Center of Agricultural Research of Tangier, INRA, Morocco; 2Regional Center of
Agricultural Research of Agadir, INRA, Morocco; 3Faculty of Veterinary Medicine, University
of Liège, Belgium; 4National School of Agriculture, Meknes, Morocco
e-mail: [email protected]

ABSTRACT

The recent development of new methods for remotely observing animal behavior using electronic
sensors such as global positioning systems (GPS) and three-axis accelerometers to monitor and
record behavior at different spatial and temporal scales presents real opportunities to better
understanding and interpreting behaviors of grazing animal. The goal of this study was to
distinguish different behavioral categories of grazing goats by combining GPS collars,
accelerometers, and satellite remote sensing. It was conducted in the mountainous forest rangeland
of Beni Arouss (Northern Morocco) from an extensive local goat's farm. Eight experimental goats
were fitted with GPS collars and leg sensors to monitor their seasonal grazing activities. A
calibration study and classification tree analysis were used to predict the grazing activities of goats.
According to the results, goats spent most of their time searching for forage in the spring and
autumn. Goats prolonged their resting time in summer (p<0.001) at the expense of grazing time.
The number of steps was numerically similar and significantly higher in both seasons of summer
and autumn (p<0.001). Goats spent 48% of their feeding time grazing (foraging) during the spring
season, in contrast to the summer (27%) and autumn (31%). Analysis of GPS collar data showed a
significant effect of the season on the measured parameters (p<0.001). Monitoring grazing
activities by using GPS collars and sensors provides useful and accurate information, which could
be used to manage grazing strategies and optimize animal performance

INTRODUCTION

In Northern Morocco, forest rangelands ensure abundant and free fodder production for grazing
animals. Livestock in this region is concerned with grazing goats in forest pastures, guaranteeing
free animal feed all year round (Chebli et al., 2023). Grazing in mountainous forest rangelands
generates additional physical activities for the vertical locomotion of goats (Chebli et al., 2022).
These altitudinal motions increase the time and energy required to travel a given distance. This
information is difficult to obtain only through direct observation because observers cannot
accurately measure individual animal behavior, such as movement and activity patterns. Data on
an animal's behavioral activities are critical to understand feeding behavior and interactions with
the environment, and to identify optimal management intervention strategies. The recent
developments in Global Positioning Systems (GPS) and the increasing number of accelerometers
used to monitor and record behavioral activities offer real opportunities to expand databases and
understand animal grazing behavior. Previous studies using sensors and GPS technology to track
animal grazing activities have focused on grazing cattle and sheep (Barbari et al., 2006; González-
Pech et al., 2015; Ungar et al., 2018). The aim of this research is to ensure the sustainability of goat
farming in extensive production systems using GPS collars, sensors, and remote sensing to better
129
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

understand the grazing behaviour of goats to make targeted decisions for management and grazing
strategies.

MATERIALS AND METHODS

This work was carried out in the forest pasture of Beni Arouss in Northern Morocco. Eight
experimental local meat goats of the Beni Arouss breed (local goat, 30 ± 2.6 kg live weight (BW)
and 36 ± 6 months of age) were chosen to conduct this study, during the three-grazing seasons
(spring, summer, and autumn). Goats spend most of their day in the studied forest pasture. At the
end of the grazing day, the animals are confined to a closed and semi-open shed inside the farm. In
winter, access to forest rangelands is very limited and corresponds to the calving period. To ensure
that goats are fed during winter season, herders delimb the branches of evergreen trees in the forest
as fodder and bring them to the goat farm (Chebli et al., 2023). Each experimental goat was fitted
with a GPS collar and an IceTag sensor on the left hind leg for three days during each studied
season. Several days before the actual experimentation, these goats were fitted with GPS collars
and IceTag sensors to accustom them to the devices attached to their bodies. GPS data was used to
estimate location, speed, and horizontal and vertical traveled distances. The data were analyzed by
the GPS3000 Host software. Coordinates were converted from UTM WGS84 to Moroccan
Transverse Mercator using ArcGIS 10.X. Coordinates (x and y) in meters were calculated for each
fixed record using ArcMap. The vertical distance (VD) was derived from the altitude difference
between successive positions 1 (z1) and 2 (z2). IceTag data was analyzed by IceManager software.
The variables provided are the goat is lying (sitting to rest or ruminating), standing (standing
without eating and ruminating), number of steps, and movement index (a proprietary metric of
overall leg activity measured in three dimensions).

Data analyzes were performed using SAS software. The grazing activity data were analyzed
according to the SAS PROCMIXED procedure. Parameters were compared across seasons (i.e.,
spring, summer, and autumn). For all analyzes, the level of significance was declared at p < 0.05.
In case of significance, means were compared using Tukey test.

RESULTS AND DISCUSSION

Figures 1 and 2 represent the seasonal variation of grazing goat activities. Goats spent most of their
time searching for palatable species in the spring and autumn. Goats prolonged their resting time
in summer (p<0.001) to the detriment of resting time. The number of steps was numerically similar
and significantly higher in both seasons of summer and autumn (p<0.001).

Analysis of GPS collar data showed a significant effect of the season on the measured parameters
(p<0.001). During the summer, the forage availability is very limited, which obliges the goats’
herder to move during this season and to settle in another forest pasture in the region (Figure 2).
The horizontal distance traveled by goats was similar and significantly higher in autumn and
summer. A similar trend was observed for the vertical distance. Conversely, goat speed was
significantly higher in spring compared to other seasons (p<0.001). Foraging day length (time spent
grazing) was prolonged (p<0.001) in summer compared to autumn and spring. According to CART
(Classification and Regression Tree) analysis, the time spent grazing (eating) was longer in spring
and similar in summer and autumn (p<0.001). Standing rest was similar between seasons (p =
0.191). Time spent walking without grazing (eating) is classified as fall > summer > spring. The
130
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

findings of this study correlate with seasonal variations in grazing behavior of goats in similar
forest pastures (Chebli et al., 2022). In other regions of Africa, Safari et al. (2011) reported that
goats in the semi-arid zone of Tanzania increased their grazing (eating) time (57–68%) and
decreased their resting time (6.8–1.4%) between rains and late summer, while their time spent
walking was similar (27%). In a similar region of Zimbabwe, goats spent most of their time eating
during the rainy season (52–75%) in contrast to the summer (29–50%) (9. Nyamangara et al.,
1995). Like the current results, goats spent 48% of their feeding time grazing (eating) during the
green season, in contrast to the summer (27%) and autumn (31%) seasons [2]. This result could be
explained by the high abundance of preferred shrubs (Cistus spp. and Lavandula stoechas) and
herbaceous plants during the spring season. In the semi-arid zone of Tanzania, Safari et al. (2011)
reported that goats extended the length of their grazing day in the summer compared to the rainy
season to meet their intake requirements.

Figure 1. Seasonal variation in grazing activities of experimental local goats browsing in Beni
Arouss pasture (Northern Morocco).

131
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

CONCLUSIONS

The combination of GPS collar, accelerometer, and remote sensing to monitor and record the
grazing activities of goats has provided useful data for understanding the grazing behavior of goats
in a complex forest rangeland of Northern Morocco.

Data on individual animal behavior, such as movement and activity patterns, are often important
for their management on pasture. It would be more interesting to extend this type of study to other
livestock systems and other types of animals to develop a guide on the use of forest pastures in
Morocco.

REFERENCES

Chebli Y., Chentouf M., Cabaraux J.F., and El Otmani S. 2023. Floristic Composition, Diversity,
Palatability, and Forage Availability of Forest Rangelands in the Southern Mediterranean
Region of Northern Morocco. Land, 12: 215.
Chebli Y., El Otmani S., Hornick J.L., Bindelle J., Cabaraux J.F., Chentouf M. 2022. Estimation
of grazing activity of dairy goats using accelerometers and global positioning system.
Sensors, 22: 5629.
Barbari M., Conti L., Koostra B.K., Masi G., Guerri F.S., Workman S.R. 2006. The use of global
positioning and geographical information systems in the management of extensive cattle
grazing. Biosyst. Eng. 95: 271–280.
González-Pech P.G., Torres-Acosta J.F.D.J., Sandoval-Castro C.A., Tun-Garrido J., de Jesús
Torres-Acosta J.F., Sando-val-Castro C.A., Tun-Garrido J. 2015. Feeding behavior of sheep
and goats in a deciduous tropical forest during the dry season: The same menu consumed
differently. Small Rumin. Res. 133: 128–134.
Ungar E.D., Nevo Y., Baram H., Arieli A. 2018. Evaluation of the IceTag leg sensor and its
derivative models to predict be-haviour, using beef cattle on rangeland. J. Neurosci. Methods.
15: 127–137.
Safari J., Mushi D.E., Kifaro G.C., Mtenga L.A., Eik L.O. 2011. Seasonal variation in chemical
composition of native forages, grazing behaviour and some blood metabolites of Small East
African goats in a semi-arid area of Tanzania. Anim. Feed Sci. Technol. 164: 62–70.
Nyamangara M.E., Ndlovu L.R. 1995. Feeding behaviour, feed intake, chemical and botanical
composition of the diet of indigenous goats raised on natural vegetation in a semi-arid region
of Zimbabwe. J. Agric. Sci. 124: 455–461.

132
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ADVANCEMENTS IN PRECISION AGRICULTURE: CUTTING-EDGE SOLUTIONS

FOR COCCIDIOSIS DISEASE USING SMART DIAGNOSTIC SYSTEMS

1*F.O.
Bamigboye, 2O. Bello, 2O.A. Osaikhuwuomwan, 2P.C. Anakua
1 Department of Agricultural Sciences, College of Sciences, Afe Babalola University Ado-Ekiti,
Ekiti State; Department of Computer Science, College of Sciences, Afe Babalola University
2

Ado-Ekiti, Ekiti State

e-mail: [email protected]

ABSTRACT

The poultry industry in Nigeria faces critical challenges from diseases like Coccidiosis, which
severely hinder productivity. Traditional farming methods, limited access to veterinary services,
and unreliable diagnostic information exacerbate this issue. To address these challenges, this study
harnesses artificial intelligence (AI) and machine learning (ML) to develop a smart diagnostic
solution for poultry diseases. Using Google Teachable Machine and a dataset of 4,880 faecal
images sourced from an open-access Zenodo repository, this study applied advanced image
augmentation, pre-processing, and segmentation techniques to enhance diagnostic accuracy. The
dataset was split into 85% for training and 15% for testing, achieving an impressive 99% accuracy.
The classification model was integrated into a mobile application built with Flutter and Python,
enabling farmers to easily access diagnostic tools for proactive disease management. This
innovation bridges the gap in veterinary services, providing an efficient and cost-effective solution
for identifying and managing Coccidiosis in poultry. Future work will explore extending the
application to other poultry diseases and validating its impact through real-world field trials. This
study marks a significant advancement towards sustainable poultry farming in Nigeria, fostering
economic growth and improved food security.

Keywords: Machine Learning, Precision Agriculture, Poultry Disease Diagnostics, Mobile

Applications, Image Classification

INTRODUCTION

Nigeria is noted as having the second largest chicken population in Africa, with a standing stock
of about 180 million birds producing more than 14 billion eggs and 454,000 tonnes of meat
annually. The Nigerian poultry industry contributes approximately 25% to agricultural GDP
(Makasi et al., 2020). However, poultry production has not kept pace with the rapid increase in
domestic consumption because it is greatly affected by poultry diseases. Coccidiosis is a poultry
disease ranked as one of the leading causes of death in poultry with Eimeria tenella among the
most pathogenic parasite (Abbas et al., 2019; Williams, 2005). Without timely detection and
intervention, outbreaks of coccidiosis can lead to substantial economic losses in the livestock
sector.

The present and most common diagnostic techniques for coccidiosis rely on clinical indicators,
such as determining if the diarrhoea is bloody or brown, counting the number of oocytes in the
stool, and assessing the intestinal tract to get the lesion score. In addition to taking days to complete,

133
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

all of these procedures are costly and time-consuming since by the time a diagnosis is obtained, the
disease may have progressed, or a high death rate may have occurred.

Recent studies highlight how smart diagnostic systems—powered by sensors, machine learning,
and data analytics—can offer real-time monitoring and early detection of coccidiosis in livestock,
enabling farmers to implement timely interventions and reduce reliance on antibiotics (Ahmad et
al., 2020; Su et al., 2018). These systems utilize data-driven insights to provide continuous health
assessments, helping farmers address disease outbreaks efficiently while supporting sustainable
livestock production practices. Hence, the present study explored the possibilities of cutting-edge
solutions for Coccidiosis disease using Smart Diagnostic Systems in poultry production.

MATERIALS AND METHODS

The study utilized a comprehensive dataset of poultry faecal images, sourced from a Zenodo
repository published by Machuve et al. (2021). The images, collected between February 2020 and
February 2021 from farms in Tanzania, represented two categories: chickens affected by
coccidiosis and healthy chickens. A total of 720 laboratory-labelled images were processed using
Polymerase Chain Reaction (PCR) for disease confirmation, while 4160 farm-labelled images were
annotated by veterinarians and field officers based on visual features like color and shape.

To prepare the data, the images were resized to uniform dimensions (224x224 or 512x512 pixels)
and categorized into training (85.2%), validation (15%), and testing (14.8%) sets. Pre-processing
involved manual labeling and data augmentation techniques, such as image flipping, cropping,
padding, and saturation adjustments, to enhance the training set and prevent overfitting.

Model training was conducted using Google Teachable Machine, employing pre-trained deep
neural networks. Key parameters included a learning rate of 0.001, 50 training epochs, and a batch
size of 16. Performance was evaluated by monitoring training and validation accuracy and loss
across epochs. Once trained, the model was exported as a TensorFlow Lite file for mobile
application deployment.

The final system integrated the AI model with a Flutter-based mobile application, enabling efficient
and portable diagnostics for chicken diseases. This framework provided farmers with an accessible
tool to detect coccidiosis on-site using a lightweight, high-performing classification model.

134
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Automated coccidiosis disease diagnostics research work-flow

RESULTS AND IMPLEMENTATION

The study evaluated the performance of a deep-learning model trained to diagnose coccidiosis in
poultry using a dataset of 4880 labelled faecal images. The dataset included 2476 images of
chickens affected by coccidiosis and 2404 images of healthy chickens. The model, trained using
Google Teachable Machine, demonstrated exceptional accuracy, achieving 99% for the coccidiosis
class and 100% for the healthy class. Performance metrics, including the confusion matrix,
accuracy trends, and loss per epoch over 50 training iterations, confirmed the model’s reliability
and effectiveness in classification tasks.

For practical implementation, the trained model was converted into TensorFlow Lite format to
optimize it for mobile devices. This lightweight format ensured compatibility with devices having
limited computing resources while maintaining the model's high accuracy. The conversion process
included loading the model, preprocessing input images, running predictions, and postprocessing
the output to generate class labels and their probabilities.

The integrated system was deployed as a mobile application developed with Flutter, providing a
user-friendly interface for poultry farmers. The app allows users to capture or upload images of
chicken faeces, which are classified in real-time using the embedded AI model. Results, including
the predicted class (Coccidiosis or Healthy) and prediction accuracy, are displayed alongside
expert-verified treatment recommendations. Treatment options, such as herbal and chemical
remedies, are stored in a centralized database and accessed dynamically within the app.

CONCLUSION AND RECOMMENDATION

The mobile application developed for the detection of coccidiosis in poultry demonstrates the
potential of leveraging artificial intelligence to support early disease diagnosis in agricultural
settings. The integration of a highly accurate deep-learning model (achieving 99% and 100%
accuracy for coccidiosis and healthy classes, respectively) with a user-friendly interface ensures a
practical, scalable tool for poultry farmers. By enabling real-time analysis of faecal images and
135
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

offering reliable treatment recommendations, the system addresses a critical challenge in poultry
farming—minimizing disease-related losses. Its deployment as a lightweight mobile application
ensures accessibility even for small-scale farmers with limited resources.

It is therefore recommended that efforts should focus on promoting the app’s adoption through
training programs and awareness campaigns to educate farmers about its benefits. Regular updates
to the AI model are essential to enhance its robustness, ensuring accuracy across diverse regions
and poultry breeds. Expanding the app’s diagnostic capabilities to include other poultry diseases
would further increase its value as a comprehensive tool for farmers.

To improve accessibility, offline functionality should be developed, to enable farmers in remote

areas with limited internet access to use the app effectively. Additionally, integrating the app with
veterinary services could provide users with expert guidance for advanced diagnosis and treatment.
Data privacy measures must be prioritized to safeguard user information and ensure compliance
with security standards. Finally, conducting economic evaluations to quantify the app’s financial
benefits—such as reduced disease-related losses and improved productivity—will highlight its
value and encourage wider adoption.

REFERENCES

Abbas, R. Z., Colwell, D. D., & Gilleard, J. (2019). Advances in the diagnosis of coccidiosis in
poultry. Veterinary Parasitology, 268, 3-11.
Ahmad, T., Awais, M. M., Akhtar, S., & Saleem, M. (2020). Smart farming: real-time diagnostic
approaches in livestock disease management. Journal of Animal and Plant Sciences, 30(5),
1093-1105.
Machuve, Dina, Nwankwo, Ezinne, Lyimo, Emmanuel, Maguo, Evarist, & Munisi, Charles (2022).
Machine Learning Dataset for Poultry Diseases Diagnostics - PCR annotated (Version 3)
[Data set]. Zenodo. https://doi.org/10.5281/zenodo.5801834
Masaki, M.N., Lee, I., Duns, H., Toromade, F. and Ayo, O. (2020). Poultry Sector Study in Nigeria.
By Netherlands Enterprise Agency (NABC, 2020).
Su, W., Li, Y., & Yang, Y. (2018). Application of big data and artificial intelligence in intelligent
farming of animals. Engineering, 4(5), 560-565.
Williams, R. B. (2005). Intercurrent coccidiosis and necrotic enteritis of chickens: Rational,
integrated disease management by maintenance of gut integrity. Avian Pathology, 34(3),
159-180.

136
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

PRECISION NUTRIENT MANAGEMENT

137
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ASSESSMENT OF NITROGEN FERTILIZATION IN TUNISIAN WHEAT

PRODUCTION USING PROXIMAL AND REMOTE SENSING
#11255

1*M. Mechri, O. Alshihabi, H. Angar, I. Nouiri, M.H. Landolsi, M. Söderström, K. Persson, S.

Phillips
1National Institute of Field Crops (INGC/Tunisia)
*e-mail: [email protected]; 00216 5010712

ABSTRACT

The study aims to develop an approach and a workflow to optimize the in-season nitrogen (N)
application in wheat cultivation in Tunisia, using the remote and proximal sensing techniques. Two
types of trials, N response trials on small plots and large-scale trials, were carried out for two
seasons (2023-2024) in the different climatic zones in the country. Data were collected for several
parameters (e.g. N content, vegetation indices ‘VIs’, yield) and statistically analyzed. The
preliminary results showed wide variations of data around the means and no significant differences,
for the different parameters, between the N rates in the N response trials, even in the zero plot (no
N added). The VIs were quickly saturated at early Growth stage (GS≈31-39), only the first
measurement in February showed slight differences between the measurements of the different N
treatments. The correlation between the N content and the VIs for the first date was weak (r2=0.2).

INTRODUCTION

Tunisia faces water scarcity, and the agricultural sector declines due to climate change, with
projected decreases in rainfall, changes in rainfall pattern and increases in temperature (Ouessar et
al., 2021; Mechri et al., 2023). Tunisian farmers face challenges in effective land management and
optimal N fertilisation due to the crucial role of agriculture in Tunisia's economy and food security
(FAO, 2023). N is one of the key nutrients that limit the crop growth of cereals in many production
systems and is a key factor in achieving the optimum grain yield (Wang et al, 2023). The
importance of N fertilization in increasing wheat production has been well recognized but still
difficult to determine the quantities to apply under water deficit conditions (Kedir, 2020). The
cereal sector in Tunisia covers wide areas in the country from sub-humid to semi-arid zones; where
most of the fields are rainfed (Sadok et al., 2019). The Nutrients availability in the soil are induced
elements in the soil, water shortage is tightly correlated to water management policy in the country
that prioritizes allocating surface water to domestic uses rather than to irrigation. On the other hand,
irrigation using groundwater (e.g. in Kairouan) continues to overuse the water table with an average
drawdown of 5 m year-1. These conditions lead to a low N use efficiency, low national wheat
(Triticum durum L.) average yield – estimated to 1.4 t ha-1, in addition to groundwater pollution
(de Oliveira et al 2020). In Sweden, the zero plot technique to assess the initial soil N supply is
widely used by farmers, often in combination with proximal sensing based digital tools for variable
rate application (VRA) of N (Alshihabi et al., 2020). The risk of nutrients leaching is high during
the after-harvest rainfall, in Sweden catch crops, is a measure commonly used to uptake the residual
nutrients in between growing seasons, which is not applicable in Tunisia due to lack of rainfall
during the summer season. In this study an experimental program was implemented in Tunisia to
assess the N uptake using proximal and remote sensing for better N fertilizing practices.

138
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

MATERIALS AND METHODS

The present project (2023-2024), is a collaboration between the Swedish university of agricultural
sciences (SLU) and two institutions in Tunisia (the National Institute of Field Crops ‘INGC’, and
the National Agronomic Institute of Tunisia ‘INAT’), it aims to develop basic knowledge, methods,
calibration models and workflows for proximal and remote sensing in wheat production in Tunisia,
to be used as the basis for a DSS for optimizing N recommendations to the wheat farmers. The
project collaborates with a wider project on N application management (NUTCAT) covering
several African countries. The joint research program in Tunisia covers all the climatic zones of
cereal cultivation, the rainfed production in the sub-humid (e.g. Beja) and the semi-arid (e.g.
Manouba and Siliana) areas, while irrigation is applied in the arid area in Kairouan. In each climatic
zone, two types of trials were implemented in farmers’ fields: N response trial on small plots
(3×12m), and 2-3 larger trials (1 ha). The N response trials cover 15 different rates of N application
(T1-15), which vary from 0 kg ha-1 for the zero plot in all areas, and for the fertilized plots varies
from 20-80 kg ha-1 in the semi-arid area, 50-190 kg ha-1 in the sub-humid area to 70-280 kg ha-1
in the aridIrrigated area. The fertilization was carried on in 1, 2 or 3 doses (D1 pre-plant, D2 at
growth stage 25 and D3 at GS 30). Although, the amount of added N varies from region to another,
but it follows same strategy:

• _T1, for zero plots (no N added)

• _T2-5, only D1 varies increasingly from T2 to T5
• _T6-8, D1 same amount and D2 varies increasingly from T6 to T8
• _T9-11, D1 same amount, no D2 and D3 varies increasingly from T9 to T11
• _T12-15, D1 same amount, D2 same amount and D3 varies increasingly from T9 to T11

The large-scale trials are split into two parts (1 ha each), one for the optimal practices (OT)
designed by INGC, and the second part is the farmer practices (FP). Data on soil properties, in
season crop status, yield and grain quality, georeferenced vegetation indices (VIs) using simple
radiometer proximal sensors (RapidScan CS-45, Holland Scientific, USA and Green Seeker
handheld, Trimble, USA) and the chlorophyll meter (SPAD-502, Konica Minolta, Japan) were
collected. The satellite images, mainly from Sentinel-2, were downloaded and correlated with the
ground truth measurements and the proximal data in the OT and FP experiments. The approach
and the workflow aim at developing two correlations, one between the ground truth N uptake and
the VIs from the proximal sensors measured in the N response trials, the second is a correlation
between the VIs measured from the proximal sensors in the large trials and those calculated from
the satellite images. The data analysis is still ongoing, in this study only the statistical analysis, the
ANOVA test (at 95% confidence level) results of the yield and the VIs in the different N response
trials for the two seasons 2023-2024 and the correlation between the VIs and the N content for the
year 2023 will be presented.

RESULTS AND DISCUSSION

The results showed a wide range of variations in the yield between treatments in the two years in
all the climatic areas. The year 2024 was worse, in general, in term of yield except for the semi-
arid area. The range of StDev was very high in the semi-arid area when compared to the

139
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

means (70% of the mean in 2024, 90% in 2023), while it was below 50% for the sub-humid and
arid irrigated areas (see Table 1). This can be explained by the severe shortage of water, where the
variations are less when water is more available (the case for sub-humid and arid irrigated areas).
The maximum obtained yield in the sub-humid and semi-arid didn’t exceed 4 t ha-1, while the
minimum values were very low (e.g. 0.14 kg h-1 in the semi-arid in 2023). In the Arid Irrigated
area, the maximum yield reached 6.55 t ha-1, and the minimum obtained yield was 1.26 in 2023,
the maximum yield in the year 2024 was lower than the maximum in 2023 but the lower yield was
noticeably higher (3.1 t ha-1) as shown in table 1. The highness and the stability of the yield in this
area, compared to the sub-humid and semi-arid areas, is attributed to the high N rate application
and the irrigation.

Table 1. Descriptive statistic for the yield in the three regions for the two years.

Region Year Mean StDev Min. Max

Sub-humid 2023 1.99 – 2.88 0.10 – 1.23 0.57 – 2.25 2.25 – 3.96
2024 1.49 –2 .45 0.18 – 1.24 1.03 – 2.05 1.35 – 3.62
Semi-arid 2023 0.27 – 1.63 0.03 – 1.46 0.14 – 0.56 0.44 – 3.32
2024 0.61 – 2.15 0.09 – 1.48 0.20 – 1.03 1.11 – 3.52
Arid 2023 2.75 – 6.21 0.10 – 2.21 1.26 – 4.61 3.81 – 6.55
(Irrigated) 2024 3.64 – 5.24 0.25 – 1.31 3.10 – 4.73 4.10 – 5.66

The zero plots (T1) showed comparable yield values to the other treatments in the two years at
the three zones (this was obvious in the field visits at the different growth stages). Adding the
total amount of the fertilizer at the pre-plant stage (T2-5) was slightly beneficial to the crop in
term of final yield comparing to the other treatments (T6-15), no clear trend was noticed between
the different N rates among T2-5 (Figure 1). Adding N at different amounts (2 or 3 doses) did not
improve the yield in the different climatic zones for the two studying years. The ANOVA test
showed no significance at confidence level 95% between the different treatments, including the
zero plot (T1-15), for the different N rates and application strategy. This likely means that
something other than N supply was more limiting for crop growth and development. In the non-
irrigated trials, it may have been water (small crop N demand). In the irrigated trials, it may
instead be due to a large soil N supply (large soil N supply). This shows that, in a farming
situation, small on-farm trials, like zero N plots, may be useful for the farmer to understand the
balance between soil N supply and crop N demand, and how much is the need for supplemental
N fertilization. If the zero plot cannot be distinguished from the surrounding field, the local soil N
supply is large enough for the current crop and one may consider saving on supplemental N
fertilization (which would mean avoided cost and reduced risk of N leaching to the environment).
If, on the other hand, the zero-plot shows symptoms of N deficiency, one may consider
supplemental N fertilization, considering whether one expects otherwise favorable conditions
such that the crop can use the additional N.

The statistical analysis gave the same result (no significance at 95% confidence level from
ANOVA test between the different treatments T1-15) for the VIs (NDVI, NDRE) measured using
the proximal sensors, the chlorophyll concertation measured using SPAD and the N concentration
from the laboratory analysis. The VIs were saturated (reached maximum values NDVI≈0.93,
NDRE≈0.43) in end of February (GS≈31-39). The crop growth goes very fast in the month of

140
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

February and varies between the climatic zones. The VIs values were lower in March because of
the progress to the heading growth stages (GS≈50-58). The correlation between the N content and
the VIs for the measurements taken in 24th February 2023 was weak (r2=0.2), the correlation was
not studied for the later measurements because of the saturation in VIs’ values.

Figure 1. Yield in t ha-1 (2023,2024), N response trials by climate zone, T1-15 from 0 N kg ha-1
(T1, the one on the left of each group) to 80-280 N kg ha-1 (T15, the one to the right of each group).

The preliminary obtained results in this project showed, for the three climate zones and the two
successive years, there is a need to assess the soil N supply from the previous season as a vital
measure to optimize the N management in term of amount and timing. Simple technics can be
adopted in the farmers’ fields, like the zero plot, to detect the plant-available N storage in the soil,
the technique is easy enough, that the farmer can apply several zero plots in his field to detect the
within field variations for precision agriculture practices.

REFERENCES

Alshihabi, O., Nouiri, I., Mechri, M., Angar, A., Piikk, K., Martinsson, J., Söderström, S., 2020.
CropSAT – opportunities for applications in precision agriculture in Africa. First African
Conference on Precision Agriculture. 1st AfCPA. 8-10 December.
de Oliveira Silva, A.; Ciampitti, I.A.; Slafer, G.A.; Lollato, R.P., 2020. Nitrogen utilization
efficiency in wheat: A global perspective. Eur. J. Agron. 114, 126008. Nitrogen utilization
efficiency in wheat: A global perspective - ScienceDirect
FAO, 2023. Bulletin de la FAO sur l'offre et la demande de céréales. 8 septembre.

141
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

USE OF “FERTIEDGE” APPLICATION FOR OPTIMIZING WHEAT

FERTILIZATION
#11646
1I. Sbai, 2M. Hmimou, 1A. Bouslimi, 1M. Hannachi, 2H. Baroudi, 2F. Sehbaoui, 1A. Bousselmi, 1S
Arfaoui
1National Institute of Field Crops; 2 AgriEdge SA, Benguerir, Morocco

e-mail: [email protected]; 0021628332795

ABSTRACT

Wheat is a crop of global importance, and effective fertilization is crucial to maximize yield and
quality. Traditional methods of fertilization often result in under- or over-application of nutrients,
resulting in environmental problems and suboptimal crop yields. FertiEdge is a digital application
that provides accurate fertilization recommendations based on real-time data, it’s an innovative
tool designed to enhance the efficiency of wheat fertilization. This study evaluates its impact on
optimizing nutrient management, improving crop yield, and promoting sustainable agricultural
practices. The study included a comparative analysis of wheat fields managed with and without the
FertiEdge application. Key metrics measured included soil nutrient levels, fertilizer application
rates, and final yield. Plots using the ‘FertiEdge’ application showed a 22 % and 20.6 % increase
in yield compared to those managed with the conventional method respectively for Site 1 and Site
2 in the North of Tunisia. In addition, the application has allowed to reduce the quantities of
nitrogen fertilizer applied by 56 kg/ha and 113.7 kg/ha compared to those managed with the
conventional method respectively for site 1 and site 2. The use of ‘FertiEdge’ significantly
enhances the precision of wheat fertilization, leading to better resource use efficiency and higher
yields. The application’s ability to integrate diverse data sources and provide real-time
recommendations is a key advantage over traditional methods. Furthermore, the reduction in
excessive fertilizer use contributes to environmental sustainability by minimizing nutrient leaching
and greenhouse gas emissions. FertiEdge is a valuable tool for farmers looking to optimize wheat
fertilization. Its application leads to higher yields, better resource management, and sustainable
farming practices. These results remain preliminary and must be proven by repeating the same
experiment two more agricultural years.

INTRODUCTION

Wheat is one of the most widely cultivated cereal crops globally, serving as a staple food source
for billions of people. The efficient management of nitrogen fertilization is critical to maximizing
wheat yields and ensuring food security. Nitrogen is a vital nutrient for plant growth, influencing
not only the productivity of wheat but also the quality of the grain. However, the application of
nitrogen fertilizers must be carefully managed to avoid environmental degradation, such as
nitrogen leaching, greenhouse gas emissions, and soil acidification, which can result from over-
application or improper timing.

In recent years, precision farming technologies have emerged as transformative tools in agricultural
practices, offering the potential to enhance nitrogen management in wheat cultivation. These
technologies, including satellite imagery, soil sensors, and variable rate application systems, enable
farmers to apply fertilizers more accurately, ensuring that nitrogen is supplied according to the
142
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

specific needs of the crop and the variability of the field. By integrating precision farming
techniques, it is possible to optimize nitrogen use efficiency, reduce input costs, and minimize
environmental impacts, ultimately leading to more sustainable and productive wheat farming
systems.

At a firsthand, this paper reviews the critical role of nitrogen fertilization management in wheat
production and examines the added value of precision farming technologies in enhancing nitrogen
use efficiency. Secondly, this paper includes a comparative analysis of wheat fields managed with
and without the FertiEdge application which is a precision farming technology developed by
AgriEdge to enhance nitrogen fertilization management. The KPIs metrics measured included soil
nutrient levels, fertilizer application rates, and final yield.

Importance of Nitrogen Fertilization Management

Nitrogen fertilization is a critical component in the management of wheat cultivation, directly
influencing crop yield, quality, and environmental impact. The advancement of precision farming
technologies has significantly enhanced the efficiency and effectiveness of nitrogen management
in wheat farming, leading to substantial benefits for both agricultural productivity and
sustainability.

Nitrogen is a vital nutrient for wheat, essential for protein synthesis and overall plant growth.
Proper nitrogen management is crucial in optimizing wheat yield, improving grain quality, and
minimizing environmental impacts such as nitrogen leaching and greenhouse gas emissions.
Recent studies have emphasized the importance of precise nitrogen management strategies. For
instance, long-term studies have shown that increased nitrogen application can significantly
improve wheat yield and soil properties, particularly in systems combining nitrogen fertilization
with organic practices such as straw return (Jaćimović et al., 2023).

Moreover, studies on winter wheat cultivation under waterlogged conditions have demonstrated
that optimizing nitrogen fertilization rates can enhance plant growth and productivity, even in
challenging environmental conditions (He et al., 2024). Similarly, optimal fertilization strategies,
as observed in North China, have shown that tailored nitrogen management can significantly
increase wheat yield and improve environmental sustainability (Jiang et al., 2023).

Added Value of Precision Farming Technologies

Precision farming technologies have revolutionized nitrogen management by allowing site-specific
application, which tailors fertilization to the specific needs of different areas within a field. This
approach not only optimizes nitrogen use efficiency but also reduces the overall amount of nitrogen
required, thereby minimizing environmental impacts.

One of the key advancements in precision farming is the use of Variable Rate Technology (VRT).
This technology incorporates soil property maps and management zones to apply nitrogen more
accurately. Studies have shown that VRT can lead to a 25% reduction in nitrogen fertilizer usage
in wheat cultivation while maintaining or even improving yield and quality characteristics such as
grain protein and gluten content (Denora et al., 2022).

The potential of management zones and geospatial technologies to enhance wheat production by
optimizing site-specific fertilization has been demonstrated in various studies (Haroon et al., 2023).
143
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Remote sensing technologies, such as Sentinel-2 NDVI and hyperspectral imagery, have proven
effective in monitoring crop nitrogen status and adjusting fertilization practices accordingly, further
supporting sustainable intensification in wheat production (Santaga et al., 2021; Song et al., 2007).
Additionally, integrating precision farming technologies with traditional farming practices, such as
utilizing farmyard manure in conjunction with mineral nitrogen rates, has shown significant
improvements in nitrogen use efficiency and crop productivity (Salama et al., 2021).

Summary table of the relevant references.

Study Main Focus Key Findings

Enhanced wheat yield and soil
Long-term straw return and
Jaćimović et al., 2023 properties with combined
nitrogen fertilization
practices.
Improved plant growth and
Nitrogen management under
He et al., 2024 productivity with optimized
waterlogged conditions
nitrogen rates.
Increased yield and sustainability
Optimal fertilization
Jiang et al., 2023 with tailored nitrogen
strategies in North China
management.
25% reduction in nitrogen use
Variable Rate Technology
Denora et al., 2022 while maintaining yield and
(VRT)
quality.
Geospatial technologies and Enhanced wheat production with
Haroon et al., 2023
management zones site-specific fertilization.
Sentinel-2 NDVI in nitrogen Improved nitrogen use efficiency
Santaga et al., 2021
management and sustainable wheat production.
Optimization of fertilization
Hyperspectral imagery in
Song et al., 2007 strategies with precision
nitrogen management
technologies.
Significant improvements in
Integration of precision and
Salama et al., 2021 nitrogen use efficiency and
traditional farming practices
productivity.

The integration of precision farming technologies into nitrogen fertilization management offers
substantial benefits in terms of efficiency, sustainability, and crop productivity. These
technologies enable more precise application of fertilizers, reduce environmental impacts, and
support higher yields and better quality in wheat production. The ongoing advancements in this
field suggest a promising future for sustainable agriculture, where precision management will
play a pivotal role in addressing the challenges of food security and environmental conservation.

MATERIALS AND METHODS

Physical Environment
The experiment was conducted at two experimental stations of the National Institute of Major
Crops (INGC) in Bousalem (Site 1), which has a superior semi-arid climate, and in Béjà (Site 2),
which has a humid climate. Both sites are in the northwest of Tunisia with a rainfall regime.

144
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Plant Material
The study focused on two varieties of durum wheat: the Inrat100 variety for Site 1 and the Maali
variety for Site 2.

Experimental Design
For each site, the trial was conducted on a 1-hectare plot divided into two sections, each
representing a different nitrogen fertilization treatment. Nitrogen was applied as ammonium
nitrate (33.5%) in three fractions (at the 3-5 leaf stage, the tillering stage, and the booting stage)
according to the recommendations of the FertiEdge application (Plot 1). The control (Plot 2)
followed the recommendations of the balance method (Plot 2).

By the end of the season, at each plot for the two treatments, the actual yield was measured using
a combine harvester.

RESULTS AND DISCUSSION

Nitrogen optimization
The results (Table 1) show that the nitrogen fertilization approach recommended by the FertiEdge
application reduced the amounts of applied nitrogen fertilizers by 56 kg/ha and 113.7 kg/ha
compared to those managed with the conventional method, respectively for site 1 and site 2.

Reducing the amount of nitrogen fertilizer helps lower production costs and mitigate the risk of
pollution.

Table 1. Quantity of ammonium nitrate supplied.

Site1 Site 2
Plot 1 Plot 2 Plot 1 Plot 2
Quantity of ammonium nitrate supplied (Kg/ha) 234 290 156.3 270
* For Site 2, the first nitrogen application was missed due to the lack of rain.

Yield Enhancement
The results obtained (Table 2) show that plots using the 'FertiEdge' application experienced a
yield increase of 22% and 20.6% compared to those managed with the conventional method,
respectively for sites 1 and 2. This underscores the importance of precision fertilization, which
ensures that crops receive the right amount of fertilizer at the right time and in the right place.

Table 2. Grain yield.

Site1 Site 2
Plot 1 Plot 2 Plot 1 Plot 2
Grain Yield (qx/ha) 39.34 32.31 45.83 38

145
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REFERENCES

Denora, M., Amato, M., Brunetti, G., De Mastro, F., & Perniola, M. (2022). Geophysical field
zoning for nitrogen fertilization in durum wheat (Triticum durum Desf.). PLOS ONE, 17(4),
e0267219. https://dx.doi.org/10.1371/journal.pone.0267219
Haroon, Z., Cheema, M. J. M., Saleem, S., Amin, M., Anjum, M. N., Tahir, M., Hussain, S., Zahid,
U., & Khan, F. (2023). Potential of precise fertilization through adoption of management
zones strategy to enhance wheat production. Land, 12(3), 540.
https://dx.doi.org/10.3390/land12030540
He, P., Yu, S., Ding, J., Ma, T., Li, J., Dai, Y., Chen, K., Peng, S., Zeng, G., & Guo, S. (2024).
Multi-objective optimization of farmland water level and nitrogen fertilization management
for winter wheat cultivation under waterlogging conditions based on TOPSIS-Entropy.
Agricultural Water Management, 300, 108840.
https://dx.doi.org/10.1016/j.agwat.2024.108840
Jaćimović, G., Aćin, V., Mirosavljević, M., Brbaklić, L., Vujić, S., Dunđerski, D., & Šeremešić, S.
(2023). Effects of Combined Long-Term Straw Return and Nitrogen Fertilization on Wheat
Productivity and Soil Properties in the Wheat-Maize-Soybean Rotation System in the
Pannonian Plain. Agronomy, 13(6), 1529. https://dx.doi.org/10.3390/agronomy13061529
Jiang, X., Li, J., An, Z., Liang, J.-M., Tian, X., Chen, Y., Sun, Y., & Li, Y. (2023). Optimal
Fertilization Strategies for Winter Wheat Based on Yield Increase and Nitrogen Reduction
on the North China Plain. Sustainability, 15(5), 4199. https://dx.doi.org/10.3390/su15054199
Salama, H. S. A., Nawar, A., Khalil, H., & Shaalan, A. (2021). Improvement of Maize Productivity
and N Use Efficiency in a No-Tillage Irrigated Farming System: Effect of Cropping
Sequence and Fertilization Management. Plants, 10(7), 1459.
https://dx.doi.org/10.3390/plants10071459
Santaga, F., Benincasa, P., Toscano, P., Antognelli, S., Ranieri, E., & Vizzari, M. (2021).
Simplified and Advanced Sentinel-2-Based Precision Nitrogen Management of Wheat.
Agronomy, 11(6), 1156. https://dx.doi.org/10.3390/AGRONOMY11061156
Song, X., Yan, G., Wang, J., Liu, L., Xue, X., Li, C., & Huang, W. (2007). Use of airborne
hyperspectral imagery to investigate the influence of soil nitrogen supplies and variable-rate
fertilization to winter wheat growth. Proceedings of SPIE, 6752, 67521Q.
https://dx.doi.org/10.1117/12.736116

146
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

COMPARATIVE STUDY ON PRECISION NITROGEN MANAGEMENT FOR WHEAT

USING GREENSEEKER, CHLOROPHYLL METER AND LEAF COLOR CHART
BASED ON SPECTRAL CHARACTERISTICS OF LEAVES
#11641

1Abdou A. Soaud, 1A. El-Metwally, 2M. Ali, 1Radwa K. Sayed

1Faculty of Agriculture, Cairo University, Giza, Egypt; 2 Desert Research Centre, Cairo, Egypt
e-mail: [email protected]; 00201007507791

ABSTRACT

Collecting results on nitrogen (N) uptake throughout the growing season using tools such as
GreenSeeker optical sensor, chlorophyll meter and Leaf Color Chart (LCC) holds great promise
for optimizing N fertilizer management in cereal crops. To investigate this further, field
experiments were conducted over two consecutive winter seasons (2017/2018 and 2018/2019) on
wheat at the Experimental Farm of the Faculty of Agriculture, Cairo University, located in Giza
Governorate, Egypt. The primary objective of these experiments was to develop effective strategies
for managing N fertilizer in wheat during the growing season using GreenSeeker optical sensor,
chlorophyll meter or LCC. The goal was to enhance N-use efficiency, reduce the N fertilizer input
and achieve optimal results. In the first season, various rates of N fertilizer were applied to create
diversity in the readings obtained from the three tools: GreenSeeker optical sensor, chlorophyll
meter and LCC. Based on the findings from the first season, a strategy was proposed for refining
the application of N fertilizer during the jointing growth stage of wheat. This strategy was guided
by the readings obtained from the three tools and implemented in the second season. For the
GreenSeeker optical sensor, chlorophyll meter and LCC, an initial application of prescriptive N
fertilizer (100 kg N ha-1 in two splits) was recommended, followed by corrective doses based on
the guidance provided by the three tools. The results of this study were remarkable. The N recovery
efficiency that correlated with higher yield values achieved using the GreenSeeker optical sensor,
chlorophyll meter, and LCC was 74.1%, 67.4%, and 55.4%, respectively, compared to only 50.5%
with the general recommendation treatment. Also, the rate of nitrogen application decreased to 160,
180, 190 Kg N ha-1 with the use of GreenSecker, chlorophyll meter and LCC, compared to 240 Kg
N ha-1 with the treatment of general recommendations without affecting grain yield. These findings
clearly demonstrate that the utilization of tools such as the GreenSeeker optical sensor, chlorophyll
meter, and LCC can significantly improve the N-use efficiency and decrease N application rate
without compromising grain yields.

INTRODUCTION

The application of nitrogen (N) fertilizer is widely recognized as a crucial factor in promoting the
growth, yield, and quality of crops. Nitrogen plays a vital role in the formation of various
compounds necessary for plant development, including chlorophyll and enzymes. Nitrogen
fertilizer management in wheat production in Egypt is typically based on a general
recommendation that is applied across large areas. However, to achieve high yields, farmers often
exceed the recommended N application rates. This practice is influenced by the temporal and
spatial variability of N requirements, which leads to either over or under-application of fertilizer,
ultimately reducing its efficiency. It is worth noting that the global nitrogen recovery efficiency in

147
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

cereals is estimated to be around 35% (Omara et al., 2019). This indicates that a significant portion
of the N fertilizer applied is susceptible to losses from the soil-plant system. Consequently,
substantial amounts of N fertilizer are lost from the soil, leading to both environmental degradation
and increased costs. The low recovery efficiency of N fertilizer not only contributes to
environmental concerns but also imposes financial burdens (Bijay and Yadvinder, 2003; Fageria
and Baligar, 2005 and Ali and Habib, 2022). Therefore, it is crucial to address this issue to enhance
both the sustainability and profitability of wheat production in Egypt. Advanced technologies such
as GrS, ChM and LCC have emerged as promising tools for efficient N management (Ali et al.,
2020; Singh et al., 2022 and Ram et al., 2022). These cutting-edge tools offer farmers real-time,
non-destructive measurements of plant health, empowering them to apply N with precision
according to the specific needs of their crops. By utilizing these technologies, farmers gain the
ability to closely monitor the temporal and spatial variations of N levels across their fields. This
invaluable insight allows for the optimization of fertilizer usage, minimizing the risk of nutrient
losses that could harm the environment. In essence, these advanced tools not only provide farmers
with accurate and timely information about their plants' N requirements, but also enable them to
make informed decisions that promote sustainable farming practices. By harnessing the power of
technology, farmers can enhance their productivity while minimizing the environmental impact of
their operations.

The primary objective of this study was to develop effective strategies for managing N fertilizer in
wheat during the growing season using GreenSeeker, chlorophyll meter, or LCC. The goal was to
enhance N-use efficiency, reduce the N fertilizer input and achieve optimal wheat yield.

MATERIALS AND METHODS

The experimental site

In two successive winter seasons (2017/2018 and 2018/2019), field experiments were carried out
on wheat (Triticum aestivum L.) variety Giza 171 at the Experimental Farm of the Faculty of
Agriculture, Cairo University, Giza Governorate, Egypt, latitude 30.0861N, and longitude
31.2122E. Initial soil samples were taken from the experimental site and analyzed using the
procedures outlined by Page et al. (1982) for some physical and chemical properties. Soil texture
was clay loam, pH in soil saturation past was 7.91, EC in soil saturation past was 4.53, organic
matter was 2.3% and available N, P and K were 100.9, 18.5 and 354.0 mg kg-1, respectively.

Experimental design and treatments

The soil has been ploughed and levelled prior to sowing. In both seasons, in mid-November, wheat
(Triticum aestivum L.) of the variety Giza 171 grains was mechanically sown in rows 15 cm apart
and divided into 15 m2 parcels. N fertilizer levels of 0, 40, 80, 120, 160, 200, 240, 280 and 320 kg
N ha-1 were added in three equal split doses in the first season as ammonium sulphate. This range
was used to determine plots with great variability in the wheat uptake and yield of N. The second
season was developed to validate the effectiveness of the GreenSeeker optical sensor, chlorophyll
meter, and LCC for the application of N fine-tuning fertilizer. The treatment consisted of setting
various prescriptive N application scenarios at the early growth stage, followed by a corrective
dose at the joint growth stage, as directed by GreenSeeker optical sensor, chlorophyll meter or
LCC. The experiments were performed with three replications in a randomized complete block
design. Following the general recommendation, phosphorus (as a single superphosphate) was

148
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

applied for sowing. Potassium fertilizer was avoided because enough available K (354 mg kg -1)
were present in the soil.

Plant measurements
The greenness of plant was measured by portable GreenSeeker optical sensor, chlorophyll meter,
and LCC. The GreenSeeker accurately detected spectral reflectance and presented it as NDVI
(Normalized Difference Vegetation Index). The sensor unit was positioned approximately 1 meter
above the plant canopy during the measurement procedure. To ensure precision, the sensor
recorded NDVI measurements at a rate of 10 per second while moving at a slow walking pace. The
chlorophyll meter measured transmittance at wavelengths of 660 and 940 nm. The procedure
entailed inserting the central section of the most fully developed leaf into the aperture of the meter.
From each plot, three plants were randomly selected, and their measurements were collected and
averaged for subsequent analysis. The LCC with six green shades the greenness of plants. The
topmost fully expanded leaf was placed on the LCC and the color of the middle part of the leaf was
matched with greenness of the panels on the LCC.

Plant sampling and analysis

At the joint growth stage, over ground plant samples from an area of 1 m2 were collected from each
plot immediately after obtaining the tools’ readings. The wheat production was manually collected
from a net area of 6 m2 at maturity from the center of each plot. Grain and straw samples were
dried in a hot air oven set at 70º C until they reached a constant weight. Dried samples were digested
in a mixture of H2SO4-H2O2, and total N was determined using the micro-Kjeldahl method (Kalra,
1997).

Calculations and statistical analysis

Using Microsoft excel program (a component in Microsoft Office 2016), regression models were
mounted. Variance analysis (ANOVA) has been used to evaluate the effect of N treatments on the
data collected. As described by Gomez and Gomez (1984), Duncan's multiple range test (DMRT)
at probability value < 0.05 was used to examine the difference between means. As described by
Cassman et al. (1998), the recovery efficiency of N (REN) was computed as:

𝑇𝑜𝑡𝑎𝑙 𝑁 𝑢𝑝𝑡𝑎𝑘𝑒 𝑖𝑛 𝑓𝑒𝑟𝑡𝑖𝑙𝑖𝑧𝑒𝑑 𝑝𝑙𝑜𝑡 − 𝑇𝑜𝑡𝑎𝑙 𝑁 𝑢𝑝𝑡𝑎𝑘𝑒 𝑖𝑛 𝑧𝑒𝑟𝑜 𝑁 𝑝𝑙𝑜𝑡

𝑅𝐸𝑁 (%) =
𝑄𝑢𝑎𝑛𝑡𝑖𝑡𝑦 𝑜𝑓 𝑎𝑝𝑝𝑙𝑖𝑒𝑑 𝑁 𝑓𝑒𝑟𝑡𝑖𝑙𝑖𝑧𝑒𝑟

RESULTS AND DISCUSSION

Effect of N fertilizer application rate on grain yield of wheat

The relationship between increasing N fertilizer rate and grain yields of wheat collected from the
first season exhibited a second-degree response function (y = - 0.1064 x2 + 45.922 x + 3926.2).
Function derivation analysis showed that the highest grain yield of 8881 kg ha -1 was achieved by
applying N fertilizer rate of 215.8 kg N ha-1. Approximately 155 kg N ha-1 was calculated as the
N fertilizer rate required for economic grain yield (8437 kg ha-1, 95% of maximum yield). The
widely adopted general N fertilizer recommendation for wheat in the area is 180-240 kg N ha-1.
These results suggest that there is a need to establish site-specific management strategies in the
season that can adjust the rate of application of N fertilizer according to the actual need for the
crop.

149
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Prediction of N uptake at jointing growth stage

Rapid acquisition of N uptake information where plants can respond to N inputs prior to harvesting
is essential for the development of a successful N fertilizer management plan for precision N
management. Variation in N uptake at the joint growth stage of wheat was created by the increasing
rate of N fertilizer applied in the first season experiment. This variability has been reflected in grain
yield increases. The data derived from the relationship between grain yield and N wheat uptake
were as follow: the estimated maximum uptake was 373 kg N ha-1, the estimated maximum yield
was 7981 kg grain ha-1, the optimum grain yield (95% of the maximum grain yield) was 7582 kg
grain ha-1 and the optimum N uptake = 275 kg N ha-1.

a. Sufficiency index approach for managing N fertilizer using GreenSeeker

By many varietal groups, seasons or regions, leaves greenness may vary. Consequently, one
GreenSeeker fixed threshold value may not work effectively. The strategy to the sufficiency index
(calculated as the ratio of NDVI reading of the evaluated plot and that of a reference N-rich plot)
allows dynamic values instead of a fixed threshold value to be used for precision N management.
According to the variability of soil properties and seasons, this strategy has the potential to be self-
calibrating. The sufficiency index can be calculated as follow:

SI= NDVI of the meausred treatment / NDVI of the reference treatment

The SI was used to calculate the corrective N fertilizer dose in the second season at jointing stage
(Feekes 6) of wheat, as steered by the GreenSeeker algorithm created in first season as follow:

𝟐𝟕𝟓− 𝟐𝟗𝟏.𝟒𝟕× 𝑺𝑰 𝑵𝑫𝑽𝑰𝟏.𝟔𝟖𝟔

N fertilizer dose (kg N ha-1) = 𝟎.𝟔𝟓

b. Sufficiency index approach for managing N fertilizer using chlorophyll meter

To determine the N nutrition index and sufficiency index of the chlorophyll meter, a calculation
was conducted by dividing the N uptake and chlorophyll meter readings in the tested plot by
reference values. These reference values were established using boxplot diagrams, specifically
identifying the upper interquartile values. The resulting reference values for the N nutrition index
and sufficiency index were determined to be 63 and 54.5 kg N ha-1, respectively. Upon analyzing
the relationship between both indexes, a highly significant linear correlation (R2= 0.6) was
obtained. This correlation strongly suggests that the sufficiency index, derived from the chlorophyll
meter, effectively captures variations in the N nutrition index. Based on this finding, a strategy was
developed to optimize the application of N during the jointing stage of wheat in the second season.
This strategy is based on recommended amounts of N fertilizer that should be applied based on the
value of sufficiency index of the chlorophyll meter as follow: if the sufficiency index: >0.95, 0.95
– 0.85, 0.85 – 0.75 and < 0.75, the corrective N dose will be 0, 80, 120, 160, respectively.

c. Establishment of fertilizer N management strategies using LCC

The utilization of different N fertilizer rates has led to significant variations in measurements of
LCC, grain yield, and total nitrogen uptake. A box-and-whisker diagram was constructed to display
the LCC readings at Feekes 6, revealing a threshold score of 4. However, as the leaf greenness
approaches an LCC score below 4, a higher amount of N (120 kg N ha-1) is necessary to achieve
optimal yield levels. It is recommended to apply a moderate dose of N fertilizer (90 kg N ha -1)

150
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

within the range of scores 4-4.5. Additionally, when the leaf greenness corresponds to an LCC
score of 4.5 or more, fertilizer nitrogen application may be omitted (0 kg N ha-1).

Validation of GreenSeeker, chlorophyll meter and Leaf Color Chart in managing N

fertilizer
The experiment performed during the second season has been used to assess the GreenSeeker
sensor, Chlorophyll meter and LCC performance as proposed in this study. Various doses and
timings of N fertilizer were added prior to applying the corrective dose as steered by each one of
the three tools to make growth variance in biomass and N uptake in wheat.

The data presented in table (1) show that the wheat grain yields obtained from Treatment 3 with
the GreenSeeker, chlorophyll meter, and LCC, they were recorded at 7989, 8141, and 7971 kg ha-
1, respectively, with statistically equivalent values. However, there was a significant variance in

the corrective dosage recommendations. While the GreenSeeker proposed a corrective dosage of
60 kg N ha-1, the Chlorophyll meter and LCC advocated for 80 and 90 kg N ha-1, respectively.

The utilization of the Green Seeker sensor, chlorophyll meter, and LCC has yielded N efficiencies
of 74.1%, 67.4%, and 55.4%, respectively, in comparison to the general recommendation's modest
50.5%. These findings underscore the superior performance and reliability of these advanced tools
in optimizing N fertilizer utilization for enhanced crop productivity. This discrepancy indicates the
superiority of the GreenSeeker, as it necessitated the least nitrogen input for the same yield, thereby
positively impacting nitrogen recovery efficiency. This outcome could be attributed to the
GreenSeeker's utilization of red and near-infrared rays to compute the Normalized Difference
Vegetation Index (NDVI), along with its capability to average readings across the entire plot. This
feature renders it a more dependable indicator compared to the Chlorophyll meter and Leaf Color
Chart, contributing to enhanced precision in nitrogen fertilizer management practices.

CONCLUSIONS

The standard recommendation for applying a fixed rate of fertilizer N over large areas is not optimal
for achieving high N use efficiency in wheat grown in diverse soils in Egypt. Field experiments
were conducted to investigate the use of tools such as Green Seeker optical sensor, chlorophyll
meter and LCC for managing N fertilizer in wheat crops. The results showed that the N recovery
efficiency achieved using the tools were higher compared to the general recommendation, with the
GreenSeeker optical sensor achieving the highest efficiency. This study demonstrates that using
these tools can significantly improve N-use efficiency without affecting grain yields.

The Green Seeker sensor, chlorophyll meter, and LCC have been validated as dependable tools for
accurately predicting N uptake in wheat and effectively guiding N fertilizer applications. The
strategies proposed in this study have demonstrated exceptional proficiency in N fertilizer
management, resulting in remarkable yield levels and substantial savings in N fertilizer usage.

Furthermore, the utilization of these advanced tools not only improves N efficiencies but also
contributes to sustainable agriculture practices. By accurately measuring the crop's nitrogen status,
farmers can apply fertilizers more precisely, reducing unnecessary nitrogen application and
minimizing environmental pollution.

151
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 1. Wheat grain yields, total N uptake, and N use efficiencies as influenced by different N fertilizer treatments
as guided by GreenSeeker sensor, chlorophyll meter and LCC.
‘
N fertilizer
Corrective
rate at
Tools’ dose Total amount of Grain Total N
REN***
Treatment 0 30 Sufficiency kg N ha-1 N fertilizer yield uptake
%
index at feekes kg N ha-1 kg ha-1 kg ha-1
DAS DAS
*
6**

N fertilizer treatments as guided by GreenSeeker sensor

T1 (zero-N) 0 0 - 0 0 3118 d 109.6 c -
T2 (general
80 80 0.75 80 (fixed) 240 8023 a 233.4 a 51.5 c
recommendation)
T3 40 60 0.74 60.9 160.9 7989 a 228.9 a 74.1 a
T4 100 0 0.72 77.3 177.3 7373 b 238.7 a 72.8 a
T5 0 100 0.71 85.3 185.3 7742 a 243.2 a 72.1 a
T6 0 0 0.68 109.1 109.1 6114 c 183.5 b 67.7 b
T7 100 100 0.80 10.1 210.1 7871 a 224.6 a 54.7 c
N fertilizer treatments as guided by chlorophyll meter
T1 (zero-N) 0 0 - 0 0 3208 d 96.6 c -
T 2 (general
80 80 - 80 (fixed) 240 8172 a 225.1 a 53.5 c
recommendation)
T3 40 60 0.91 80 180 8141 a 218.6 a 67.4 a
T4 100 0 0.82 120 220 7413 b 230.1 a 60.6 b
T5 0 100 0.87 80 180 7642 b 215.2 a 65.7 a
T6 0 0 0.73 160 160 6215 c 177.3 b 50.4 c
T7 100 100 0.96 0 200 7613 b 199.4 b 51.4 c
N fertilizer treatments as guided by LCC
T1 (zero-N) 0 0 - 0 0 3118 e 105.2 c -
T 2 (general
80 80 4.5 80 (fixed) 240 7942 a 217.1 a 46.6 d
recommendation)
T3 40 60 4.5 90 190 7971 a 210.5 a 55.4 c
T4 100 0 4.5 90 190 7283 c 222.4 a 61.7 b
T5 0 100 4.5 90 190 7452 c 235.1 a 68.4 a
T6 0 0 4 120 120 6015 d 179.3 b 61.8 b
T7 100 100 5 200 200 7671 b 198.5 b 46.7 d
*
DAS = days after sowing. **
Feekes 6 = around 50-60 days after sowing
***
REN = Recovery efficiency of nitrogen fertilizer
Means followed by the same letter within the same column are not significantly different at the 0.05 level of probability
by Duncan’s multiple range test (DMRT)

REFERENCES

Ali, A.M. and Habib, A.A. (2022). Estimation of the economic optimum rates of nitrogen fertilizer
for maize grown in a calcareous soil in combination with organic manure applications.
Communications in Soil Science and Plant Analysis, 53(18):2484-2496.

152
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Ali, A.M.; Ibrahim, S.M. and Bijay, S. (2020). Wheat grain yield and nitrogen uptake prediction
using atLEAF and Green Seeker portable optical sensors at jointing growth stage.
Information Processing in Agriculture, 7(3):375-383.
Bijay, S. and Yadvinder, S. (2003). Environmental implications of nutrient use and crop
management in rice-based ecosystems. In: Rice Science: Innovations and Impact for
Livelihood (Eds. Mew, T.W.; Brar, D.S.; Peng, S.; Dawe, D. and Hardy, B.), IRRI, Los
Banos, Philippines. pp.463-477.
Fageria, N.K. and Baligar, V.C. (2005). Enhancing nitrogen use efficiency in crop plants. Adv.
Agron., 88:97-185.
Gomez, K.A. and Gomez, A.A., 1984. Statistical procedures for agricultural research. John Wiley
& Sons.
Page, A.L. and Page, A.L., 1982. Methods of soil analysis: chemical and microbiological
proerpteis. Amen Society of Agronomy.
Omara, P.; Aula, L.; Oyebiyi, F. and Raun, W.R. (2019). World cereal nitrogen use efficiency
trends: review and current knowledge. Agrosystems, Geosciences & Environment, 2(1):1-8.
Ram, H.; Kaur, H. and Sikka, R. (2022). Need-based nitrogen management of wheat through the
use of green seeker and leaf color chart for enhancing grain yield and quality. J. Plant Nutr.,
45(17):2655-2671
Singh, V.; Singh, M. and Singh, B. (2022). Spectral indices measured with proximal sensing using
canopy reflectance sensor, chlorophyll meter and leaf color chart for in-season grain yield
prediction of basmati rice. Pedosphere, 32(6):812-822.

153
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

OPTIMIZING DURUM WHEAT NITROGEN NUTRITION INDEX (NNI) PREDICTION

THROUGH SENTINEL-2 VEGETATION INDICES INTEGRATIONS
#11335

1N.H Boughattas, 1,2M. Kadri, 2M. Zekri, 3S. Ayadi, 4S. Arraouadi, 5H. Hajlaoui, 3Y. Trifa, 2I.
Hbiri
1 University of Sousse, Higher Institute of agronomy of chott-Meriem (ISA-CM), Sousse,

Tunisia; 2 Robocare, Sfax, Tunisia; 3 National Agronomic Institute of Tunisia (INAT), Tunis,
Tunisia; 4 Regional Center of Agriculture Research of Sidi Bouzid (CRRA), Sidi Bouzid,
Tunisia; 5 Faculty of Sciences and Technology of Sidi Bouzid, Tunisia.
e-mail: [email protected]; 00216 2921361

ABSTRACT

Nitrogen is crucial for durum wheat growth and productivity, but excess or insufficient levels can
harm both the environment and farmers' finances. Remote sensing offers rapid, cost-effective, and
nondestructive ways to assess crop nutrition, with vegetation indices (VIs) indicating plant health.
This study aims to enhance the accuracy of durum wheat nitrogen status prediction by investigating
modified formulations of Nitrogen Nutrition Index (NNI) coupled with various VIs, such as NDVI
Sentinel-2, NDVI by GreenSeeker, GNDVI, NDRE, NRI, RESAVI, REDVI, and MCARI. Two
experimental plots of durum wheat were selected, one in the Medjez El Bab region in the Beja
governorate (Z30) and the other in the Sadaguia region in the Sidi Bouzid governorate (Z60). A
nitrogen dilution curve (Nc) was established for each plot at a specific growth stage to determine
the NNI index. Statistical analysis was performed using RStudio software to obtain a predictive
model for NNI and the VIs extracted by CropCare application established by Robocare. The
performance of this model was evaluated using the coefficient of determination, R 2. The correlation
analysis allowed us to identify a significant correlation between NNI and VIs. The GNDVI index
proved to be the best indicator for estimating NNI (R2=0.972), while the NDVI was excluded
(R2=0.221). In summary, this study underscores the effectiveness of integrating modified NNI
formulations with diverse VIs from remote sensing, offering improved precision in fertilizer
management for precision agriculture.

INTRODUCTION

Nitrogen is a fundamental nutrient for plant growth, playing a pivotal role in photosynthesis,
protein synthesis, and overall crop productivity (Nino et al., 2024). In the case of durum wheat
(Triticum durum), a staple in many agricultural systems, nitrogen management is critical for
achieving optimal yields and grain quality. However, the delicate balance between sufficient and
excessive nitrogen application poses a challenge (P. Chen, 2015) (C. Chen et al., 2023). Over-
application can lead to environmental issues such as nitrate leaching and greenhouse gas emissions,
while under-application can result in reduced yields and economic losses for farmers (Denora et
al., 2023). Traditionally, nitrogen management has relied on soil tests and fixed fertilizer
application rates, which often fail to account for spatial and temporal variability in crop nitrogen
needs (Diacono et al., 2012). This has driven the development of more precise, dynamic
approaches, among which remote sensing has emerged as a powerful tool (Piikki et al., 2022).
Remote sensing technologies offer the ability to monitor crop nutrition over large areas with high

154
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

spatial and temporal resolution (Yu et al., 2023). By analyzing specific spectral bands, VIs can be
derived to assess plant stress levels, and nutrient status (Xue & Su, 2017) (Fabbri et al., 2020). The
Nitrogen Nutrition Index (NNI) is a widely used indicator for assessing the nitrogen status of crops,
providing insights into whether a crop is experiencing nitrogen deficiency or sufficiency (Gée et
al., 2023). However, the accuracy of NNI predictions can vary depending on the methods and
indices used. Recent advancements in remote sensing, particularly with the availability of high-
resolution satellite data like Sentinel-2, have opened new avenues for enhancing NNI prediction
accuracy (Zha et al., 2020) (Gée et al., 2023) (Yu et al., 2023) (Nino et al., 2024). This study
explores the integration of various VIs, including those derived from Sentinel-2, to optimize the
prediction of NNI in durum wheat. By analyzing the performance of different VIs and their
relationship with NNI, this research aims to refine nitrogen management practices, ultimately
contributing to more sustainable and efficient agriculture.

MATERIALS AND METHODS

Site descriptions
During the 2023-2024 durum wheat growing season, this study was conducted at two experimental
sites in Tunisia (Fig.1a), chosen to represent different agro-climatic zones and key phenological
stages Z30 and Z60, per the Zadoks scale (Zadoks et al., 1974). Both sites followed actual field
practices, reflecting the methods and techniques used by farmers in their daily agricultural
activities. The first site, a 47 ha field located in Medjez El Bab, Beja Governorate (Fig.1b), typically
receives 550-600 mm of annual rainfall. However, during the 2023 hydrological year, the region
faced significant challenges due to adverse climatic conditions. From September 2022 to June
2023, the area experienced severe drought, with only 80 mm of rainfall.

(b)

(c)
(a)

Figure 7. (a) Location of study plots in Tunisia. (b) Delineation of the study plot in Medjez el Bab.
(c) Delineation of the study plot in Sadaguia, Sidi Bouzid.

The second site is in the Sadaguia region, Sidi Bouzid Governorate (Fig.3c), with an area of 0.278
ha. This site, cultivating Maali durum wheat, is situated in a semi-arid climate characterized by
lower annual rainfall, averaging between 200 and 300 mm. The site features sandy loam soils,
which pose specific challenges for water retention and nutrient management.
155
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Experimentation
Sample collection points were chosen using the CropCare application, based on NDVI VI maps.
In the field, one-square-meter plots were sampled to determine fresh weight. Chlorophyll content
was measured with a SPAD 502Plus (SPAD), and NDVI values were recorded using a
GreenSeeker (NDVI Green). In the laboratory, samples were dried at 60°C for 48 hours, then
ground and analyzed for total nitrogen (%N measured) using the Kjeldahl method. Critical nitrogen
levels (%Nc) were derived from dry matter values, and the NNI was calculated as the ratio of actual
nitrogen absorption to critical nitrogen absorption.

Data collection and analysis

Sentinel-2 spectral bands were used to derive those traditional VIs commonly used in the literature
(Tab.1) (Hatfield et al., 2019). These indices were computed over plant sample locations and at
two phenological stages Z30 and Z60 using the CropCare application established by Robocare
(Robocare, 2024). Measured variables (SPAD, NDVI Green, %N measured, %Nc, NNI) and VIs
(Table 1)) were analyzed using descriptive statistics, correlation analysis, and multiple linear
regression to determine NNI based on other parameters, all performed with R 4.4.1 statistical
software.

Table 1. Vegetation Index extracted from Sentinel-2 images.

VIs Definition Formula Application References

Normalized
Chlorophyll (Rouse et al.,
NDVI Difference (𝑁𝐼𝑅 − 𝑅𝑒𝑑)⁄(𝑁𝐼𝑅 + 𝑅𝑒𝑑)
content 1974)
Vegetation Index
(Diker &
Nitrogen (𝐺𝑟𝑒𝑒𝑛 − 𝑟𝑒𝑑)⁄(𝐺𝑟𝑒𝑒𝑛 + 𝑅𝑒𝑑)
NRI N content Bausch,
Reflectance Index
2003)
Normalized
(Barnes et al.,
NDRE Difference Red (𝑁𝐼𝑅 − 𝑅𝑒𝑑𝑒𝑑𝑔𝑒 )⁄(𝑁𝐼𝑅 + 𝑅𝑒𝑑𝑒𝑑𝑔𝑒) N content
2000)
Edge
Red Edge Soil
1,5 𝑥((𝑁𝐼𝑅 − 𝑅𝑒𝑑𝑒𝑑𝑔𝑒 )⁄𝑁𝐼𝑅 + 𝑅𝑒𝑑𝑒𝑑𝑔𝑒 − 0.5)) (Cao et al.,
RESAVI Adjusted NNI
2013)
Vegetation Index
Red Edge
(2𝑥𝑁𝑅𝐼 + 1)2 − 8𝑥(𝑁𝑅𝐼 − 𝑅𝑒𝑑𝑒𝑑𝑔𝑒 )
REDVI Difference NNI
Vegetation Index
Green Normalized
(𝑁𝐼𝑅 − 𝐺𝑟𝑒𝑒𝑛)⁄(𝑁𝐼𝑅 + 𝐺𝑟𝑒𝑒𝑛 (Gitelson et
GNDVI Difference NNI
al., 1996)
Vegetation Index
Modified
Chlorophyll (Daughtry et
MCARI (𝑅𝑒𝑑𝑒𝑑𝑔𝑒 − 𝑅𝑒𝑑)⁄(𝑅𝑒𝑑𝑒𝑑𝑔𝑒 + 𝑅𝑒𝑑) NNI
Absorption Ratio al., 2000)
Index

RESULTS AND DISCUSSION

Descriptive Statistical Analysis

The descriptive statistical analysis revealed notable differences between the Z30 and Z60 growth
stages, particularly in key VIs and measured variables. The t-test results indicated significant
variations in NDVI, SAVI, and NDVI Green, with higher values generally observed at the Z60

156
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

stage, reflecting more advanced plant development and increased biomass. Additionally, SPAD
values, which measure chlorophyll content, showed a marked increase at Z60, aligning with the
period of peak nitrogen demand. The %N measured and %Nc also differed significantly between
stages, with higher nitrogen content observed at Z60. The NNI index was significantly higher at
Z60, indicating better nitrogen status. These findings align with recent studies showing that these
indicators increase with plant development and peak nitrogen demand, supporting their use for
optimizing nitrogen management in durum wheat (Yu et al., 2023) (Al-Shammari et al., 2024)
(Nino et al., 2024).

Correlation Analysis
The correlation analysis between various VIs and measured variables at the Z30 and Z60 stages for
the Medjez El Bab and Sidi Bouzid plots revealed several key relationships. At Z30, strong positive
correlations were observed between NDVI and GNDVI, NDVI and NDRE, and NDVI and
RESAVI, indicating a close relationship between these indices. Additionally, NDVI Green showed
strong correlations with SPAD, %N mesuré, and %Nc, while %Nc exhibited negative correlations
with these variables. In contrast, the Z60 stage showed different correlation patterns, with NDRE
strongly correlating with NDVI and RESAVI, while GNDVI and NRI also showed high positive
correlations. Notably, the correlation between NDVI and GNDVI was positive at Z30 but negative
at Z60, indicating a shift in their relationship across stages. Overall, these results highlight the
varying strength and direction of correlations between VIs and N related variables at different
phenological stages, underscoring the complexity of crop-nutrient interactions over time (Nino et
al., 2024) (Yu et al., 2023). The shift in correlation patterns between NDVI and GNDVI across
growth stages reflects similar trends reported in studies of crop development and nutrient uptake
(Zha et al., 2020).

(a) (b)

Figure 8. Correlation Matrix Between Different Variables, (a) of the Medjez El Bab Site (Z30),
(b) of the Sidi Bouzid Site (Z30).

Relationship between NNI and VIs, Multiple Linear Regression Analysis

The multiple linear regression analysis focused on predicting the NNI Index using various VIs
derived from the CropCare application. The model (Eq.1) achieved a high R² value of 0.975,
explaining 97.5% of the variation in NNI. This indicates a strong predictive capability, with most

157
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

VIs contributing significantly to the model. The residuals were randomly distributed around zero,
suggesting that the model accurately captured the relationships between the VIs and NNI.

Eq.1
INN = 220.60 NDVI - 19.10 GNDVI - 160 NDRE + 6.27 NRI - 48.19 RESAVI
+ 0.0376 REDVI - 0.0001456 MCARI - 8.16

CONCLUSION

This study demonstrates the effectiveness of VIs and regression models in optimizing nitrogen
management for durum wheat. Statistical analysis revealed significant differences in VIs between
stages Z30 and Z60, along with strong correlations among the indices. The linear regression model
accounted for 97.5% of the variability in the NNI at stage Z30. Future research should focus on
incorporating additional data sources to improve model accuracy. Additionally, conducting field
validation trials to evaluate practical applicability and developing new VIs could further enhance
nitrogen management strategies.

REFERENCES

Al-Shammari, D., Fuentes, I., Whelan, B. M., Wang, C., Filippi, P., & Bishop, T. F. A. 2024.
Combining Sentinel 1, Sentinel 2 and MODIS data for major winter crop type classification
over the Murray Darling Basin in Australia. Remote Sensing Applications: Society and
Environment, 34, 101200. https://doi.org/10.1016/j.rsase.2024.101200
Barnes, E., Clarke, T. R., Richards, S. E., Colaizzi, P., Haberland, J., Kostrzewski, M., Waller, P.,
Choi, C., Riley, E., & Thompson, T. L. 2000. Coincident detection of crop water stress,
nitrogen status, and canopy density using ground based multispectral data.
Cao, Q., Miao, Y., Wang, H., Huang, S., Cheng, S., Khosla, R., & Jiang, R. 2013. Non-destructive
estimation of rice plant nitrogen status with Crop Circle multispectral active canopy sensor.
Field Crops Research, 154, 133‑144. https://doi.org/10.1016/j.fcr.2013.08.005
Chen, C., Zhou, S., Afshar, R. K., Franck, W., & Zhou, Y. 2023. Durum wheat yield and protein
influenced by nitrogen management and cropping rotation. Journal of Plant Nutrition.
https://www.tandfonline.com/doi/abs/10.1080/01904167.2022.2068432
Chen, P. 2015. A Comparison of Two Approaches for Estimating the Wheat Nitrogen Nutrition
Index Using Remote Sensing. Remote Sensing, 7(4), Art. 4.
https://doi.org/10.3390/rs70404527
Daughtry, C. S. T., Walthall, C. L., Kim, M. S., de Colstoun, E. B., & McMurtrey, J. E. 2000.
Estimating Corn Leaf Chlorophyll Concentration from Leaf and Canopy Reflectance.
Remote Sensing of Environment, 74(2), 229‑239. https://doi.org/10.1016/S0034-
4257(00)00113-9
Denora, M., Candido, V., D’Antonio, P., Perniola, M., & Mehmeti, A. 2023. Precision nitrogen
management in rainfed durum wheat cultivation: Exploring synergies and trade-offs via
energy analysis, life cycle assessment, and monetization. Precision Agriculture, 24(6),
2566‑2591. https://doi.org/10.1007/s11119-023-10053-5
Diacono, M., Rubino, P., & Montemurro, F. 2012. Precision nitrogen management of wheat. A
review. Agronomy for Sustainable Development, 33(1), 219‑241.
https://doi.org/10.1007/s13593-012-0111-z
158
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Diker, K., & Bausch, W. C. 2003. Potential Use of Nitrogen Reflectance Index to estimate Plant
Parameters and Yield of Maize. Biosystems Engineering, 85(4), 437‑447.
https://doi.org/10.1016/S1537-5110(03)00097-7
Fabbri, C., Mancini, M., dalla Marta, A., Orlandini, S., & Napoli, M. 2020. Integrating satellite
data with a Nitrogen Nutrition Curve for precision top-dress fertilization of durum wheat.
European Journal of Agronomy, 120, 126148. https://doi.org/10.1016/j.eja.2020.126148
Gée, C., Denimal, E., de Yparraguirre, M., Dujourdy, L., & Voisin, A.-S. 2023. Assessment of
Nitrogen Nutrition Index of Winter Wheat Canopy from Visible Images for a Dynamic
Monitoring of N Requirements. Remote Sensing, 15(10), Art. 10.
https://doi.org/10.3390/rs15102510
Gitelson, A. A., Kaufman, Y. J., & Merzlyak, M. N. 1996. Use of a green channel in remote sensing
of global vegetation from EOS-MODIS. Remote Sensing of Environment, 58(3), 289‑298.
https://doi.org/10.1016/S0034-4257(96)00072-7
Hatfield, J. L., Prueger, J. H., Sauer, T. J., Dold, C., O’Brien, P., & Wacha, K. 2019. Applications
of Vegetative Indices from Remote Sensing to Agriculture : Past and Future. Inventions, 4(4),
Art. 4. https://doi.org/10.3390/inventions4040071
Nino, P., D’Urso, G., Vanino, S., Di Bene, C., Farina, R., Falanga Bolognesi, S., De Michele, C.,
& Napoli, R. 2024. Nitrogen status of durum wheat derived from Sentinel-2 satellite data in
central Italy. Remote Sensing Applications: Society and Environment, 36, 101323.
https://doi.org/10.1016/j.rsase.2024.101323
Piikki, K., Söderström, M., & Stadig, H. 2022. Remote sensing and on-farm experiments for
determining in-season nitrogen rates in winter wheat – Options for implementation, model
accuracy and remaining challenges. Field Crops Research, 289, 108742.
https://doi.org/10.1016/j.fcr.2022.108742
Robocare. 2024. RoboCare—We Tech-Care Of your Land. RoboCare - We Tech-Care
Of your Land. https://robocare.tn/
Rouse, J. W., Haas, R. H., Schell, J. A., & Deering. 1974. Monitoring Vegetation Systems in the
Great Plains with ERTS. Third ERTS-1 Symposium NASA, 309‑317.
Xue, J., & Su, B. 2017. Significant Remote Sensing Vegetation Indices: A Review of
Developments and Applications. Journal of Sensors, 2017, 1‑17.
https://doi.org/10.1155/2017/1353691
Yu, F., Bai, J., Jin, Z., Zhang, H., Yang, J., & Xu, T. 2023. Estimating the rice nitrogen nutrition
index based on hyperspectral transform technology. Frontiers in Plant Science, 14.
https://doi.org/10.3389/fpls.2023.1118098
Zadoks, J. C., Chang, T. T., & Konzak, C. F. 1974. A decimal code for the growth stages of cereals.
Weed Research, 14(6), 415‑421. https://doi.org/10.1111/j.1365-3180.1974.tb01084.x
Zha, H., Miao, Y., Wang, T., Li, Y., Zhang, J., Sun, W., Feng, Z., & Kusnierek, K. 2020. Improving
Unmanned Aerial Vehicle Remote Sensing-Based Rice Nitrogen Nutrition Index Prediction
with Machine Learning. Remote Sensing, 12(2), Art. 2. https://doi.org/10.3390/rs12020215

159
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

PRECISION PLANTING/HARVESTING

160
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

MODELING OF THE OPTIMAL PARAMETERS, IMPROVEMENT AND

PERFORMANCE EVALUATION OF MARC ENGINE DRIVEN COMMON BEAN
(PHASEOLUS Vulgaris L.) THRESHER IN ETHIOPIA
#11221

1*Biniam Zewdie, 2Adesoji M. Olaniyan, 1Amana Wako, 3Dereje Alemu, 3Tamrat Lema
1*School of Mechanical, Chemical, and Materials Engineering, Departments of Agricultural
Machinery Engineering, Adama Science and Technology University, P.O. Box 1888, Adama,
Ethiopia; 2Department of Agricultural and Bioresources Engineering, Faculty of Engineering,
Federal University Oye-Ekiti, Ikole-Ekiti Campus, Post Code 370001, Ikole-Ekiti, Nigeria;
3Ethiopian Institute of Agricultural Research; Agricultural Engineering Research, Melkassa

Agricultural Research Center, P.O. Box 436, Adama, Ethiopia

*e-mail: [email protected]

ABSTRACT

This executive summary provides an overview of the research study conducted on modeling the
optimal parameters, improvement, and performance evaluation of the engine-driven bean thresher
developed by the Melkassa Agricultural Research Center (MARC) in Ethiopia. The study aimed to
enhance the efficiency and productivity of the bean thresher, contribute to sustainable agricultural
practices, promote technology adoption, and support the economic development of the agricultural
sector. The research study employed a comprehensive approach involving data collection,
modeling, experimentation, and performance evaluation. Key steps undertaken during the study
included: Literature Review: A thorough review of existing literature related to bean threshing,
agricultural machinery, and optimization techniques was conducted to establish a knowledge base
and identify research gaps. Data Collection: Field surveys, interviews, and observations will be
conducted to gather data on the existing bean thresher's performance, challenges faced by farmers,
and specific requirements for improvement. Modeling and Optimization: Mathematical modeling
and simulation techniques will be employed to identify the optimal parameters for the bean
thresher. Parameters such as cylinder speed, concave clearance, and fan speed will be analyzed to
determine their impact on threshing efficiency and grain quality. Experimental Design: Field
experiments will be designed and conducted to evaluate the performance of the optimized
parameters. The experiments involved comparing the modified thresher with the existing version,
measuring key performance indicators, and assessing grain loss, power consumption, and
processing efficiency. Performance Evaluation: The data collected from the experiments will be
analyzed to evaluate the performance of the optimized bean thresher. The evaluation included
metrics such as grain damage, threshing efficiency, fuel consumption, and processing capacity. The
findings of the research study demonstrate the significance and potential impact of the modeled
optimal parameters and improved bean thresher. The key outcomes and recommendations will
include the following: Improved Thresher Performance: The optimized parameters resulted in a
significant improvement in the thresher's performance. Threshing efficiency increased, grain
damage was reduced, and fuel consumption was reduced. These improvements contribute to
enhanced productivity and reduced post-harvest losses. Sustainable Agriculture: The optimized
bean thresher exhibited improved resource efficiency, reducing fuel and energy consumption. This
promotes sustainable agricultural practices and reduces the environmental impact of bean
processing. Economic Benefits: The improved thresher reduces labor requirements, processing

161
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

time, and operational costs. Farmers can achieve cost savings of approximately, leading to
increased profitability and economic benefits for bean producers. Technology Adoption: The
research study provides evidence-based recommendations for the adoption of the improved bean
thresher. Farmers, agricultural extension workers, and policymakers are encouraged to promote the
use of this technology to enhance agricultural productivity and modernize farming practices. Policy
Implications: The study's findings will have policy implications for the agricultural sector.
Policymakers are advised to consider incorporating support mechanisms and incentives for the
adoption of improved agricultural machinery, fostering local manufacturing, and promoting
sustainable agricultural practices.

Keywords: RSM, Thresher, Threshing efficiency, Germination, and Grain Damage

INTRODUCTION

The common bean (Phaseolus vulgaris L.), according to (Joshi et al., 2022), is one of the main
nourishments for both humans and livestock in Africa and the third most consumed legume
worldwide. Its seed is high in carbohydrates and protein, and animal feed originates from the seed
and its pod (Uebersax et al., 2023). With 560,191 hectares of cultivated land and 208,913 tons of
beans produced in 2019, Ethiopia is the world's most significant producer of edible legumes.
Among the most significant legumes are common beans (Bento et al., 2022).

After being harvested by hand, the common bean crop is threshed by a machine. The thresher uses
impact force and pressure to remove grain from pod and stalk (Que et al., 2024). Grains sustain
significant damage from the crop migrating between the thresher unit's stirring components and
from inadequate clearance among static as well as moving portions (Lee et al., 2023). Grain that
has been damaged has the lowest shelf life and is less resilient to pests and diseases (Adewoyin,
2023). Grain grading is the primary factor that determines its marketability; fragmented seeds result
in a lower grain grade (Parker et al., 2022). In addition, damaged grains prevent seeds from
germinating (Chandra et al., 2024).

The most qualitative parameters to determine the efficiency of a thresher operation consist of mixed
chaff with the grain, loss from threshing, and damage to the grain. According to Ghebrekidan et al.
(2024) analysis of a thresher apparatus's design features, the threshing performance was
significantly influenced by the rate at which materials were fed into the device as well as
technological factors including drum speed and concave-to-drum clearance. Additionally, Juraev
et al. (2023) observed that the threshing process was influenced by the crop cultivar, moisture
content, and biometrical indices.

Grain damage, loss from threshing, and mingled the most prestigious are the particles with the
grain metrics to assess a thresher's performance (Strecker et al., 2022). The velocity of material
feeding into the device, along with technological aspects like drum acceleration and convex-to-
drum aperture, had a substantial impact on the shredding performance, as per Ghebrekidan et al.
(2024) analysis of the design elements of a thresher apparatus. The biometrical parameters,
moisture level, and crop genotype were also found to have an impact on the threshing process by
Jan et al. (2021).

162
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Loss of grains, grain impairment, level of separation, as well as size of the pod decrease were all
the parameters that Ejara et al. (2018) differentiated into standard bean threshing quality indices in
an unusual investigation. In the process of threshing common beans, two important factors were
identified to be the aperture along with the wire loop type drum and the convex, as well as drum
peripheral velocity. Ghebrekidan et al. (2024) investigated the parameters of the typical mechanism
used for separating beans. It was shown that the distance between the cylinder and the concaves
and peripheral speed were the main parameters affecting crop quality. The findings of their
experiment using the tangential threshing mechanism indicated that the rate of grain breakage
improved from 3.8 to 6.01% when the cylinder perimeter speed was enhanced from 9.4 to 21.4 ms -
1.

Numerous threshing units were used by Umbataliyev et al. (2023) for common bean seeds. Using
a multitude of sorts of drums, rates, as well as rate of feed, they assessed the thresher's performance
in terms of throughput capacity, threshing effectiveness, damage to the grain, losses of the grain,
differentiation, energy the threshold, as well as specific utilization of energy. They discovered that
the apparent damage to the grain increased along acceleration as well as flow rate. Kidney bean
threshers were examined by Wang and Cichy (2024) using variables such seed moisture level,
clearance rates, and cylinder rpm. The outcomes demonstrated that moisture level, cyl speed, and
convex level all had a major impact on the germination of threshed seeds.

Huertas et al. (2023) found that the feed rate, moisture content, and threshing drum beat all had a
substantial impact on the success rate of threshing, output capacity, and grain damage and losses
of a longitudinal flow barrier used in common beans. The impact of the moisture level, pod size,
with the pace of the drum in a rasp-bean thresher were investigated in relation to the proportion of
damaged grains and threshed pods (Lisciani et al., 2024). The findings showed that the pod size
had the biggest impact on damage intensity, while the drum speed had the least. It was further
suggested that the optimal circumstances for common bean threshing would be a water content
ranging from 12 to 15% and a drum speed of 9.5 ms -1. Although several studies have been
conducted on the threshing of different agricultural crops, none have examined the response surface
methodology's potential for optimizing the threshing of common beans with relation to machine-
crop factors. To enhance technological parameters including cylinder acceleration, convex
clearance levels, rate of feed, and level of moisture, the response surface approach is utilized, the
principal aim of this investigation was to enhance threshing efficiency, minimize grain damage,
and maximize seed germination when threshing common beans.

MATERIALS AND METHODS

Selected improved varieties of common beans from the Oromia regional State in Ethiopia were
provided by the Awash Melkassa Research Center. A digital vernier caliper (TA, M5 0–300 mm,
China) was used to measure the three primary axial dimensions of the beans: With an accuracy of
0.01 mm, the measurements are dimension (L, mm), (W, mm), and (T, mm). The experimental
findings indicated that the average mean values for thickness (4.962 ± 0.50 mm), width (6.316 ±
0.502 mm), and length (9.848 ± 0.802 mm) were, accordingly. After common beans were harvested
by hand, the threshing procedure was carried out using a laboratory wire loop/rasp type drum
thresher. The assembled thresher and a collaborative assessment of it are depicted in Figure 1. With
33 teeth spaced 100 mm apart along each of the device's four axes, the drum measured 730 mm in
length. The concave was made from 720 mm long steel sheets that had been rolled and perforated.
163
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Experimental design
Based on the multifactorial experiment principle with three independent replications, the
experiment utilized a split-split plot design. The main plot was assigned to the two varieties of
crops levels, the sub plot was assigned to the three threshing drum speed levels, and the sub-sub
plot was assigned to the three feeding levels, each with three replications (Table 1). The Response
Surface Method was utilized to maximize the threshing performance, and statistical R-studio
software was utilized to analyze all the data gathered during the laboratory and field performance
evaluations.

Figure 1. MARC-Bean thresher schematic diagram and participatory evaluation assessment.

Response surface method (RSM)

Four independent parameters were considered for optimization: moisture content (5, 10, and 15%
wb), convex aperture (25, 35, 45 mm), speed of cylinder (7.5, 9.17, 10.83 ms-1), and rate of feeding
(550, 650, 750 kgh-1). Germination of seeds percentage, threshing efficiency, and damage to grain
were the three dependent variables in the experimental method of optimization. To fit the
experimental results, a polynomial equation of second order was thus developed using the method
of response surfaces and central composite experiment design.

According to the findings of earlier research and the limitations of the manufactured thresher (Que
et al., 2024) the levels of convex aperture, moisture level, and chamber rate were chosen (Savic et
al. 2019). In the end, 54 experiments were conducted utilizing triplets of implementation for the

164
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

independent variables in a CCD-type experimental design, as shown in Table 1. In a random order,

the trials were carried out. In the latter half of the parameters with encode, three replications were
conducted to determine the relationship model describing the two main parameters' sum of square
errors and lack of fitness (Güvercin and Yıldız, 2018). Design-Expert 12 was used to optimize the
several responses simultaneously.

Table 1. Randomization layout.

R1 R2 R3
S1F1M1 S2F2M2 S3F3M3 S1F1M3 S2F2M1 S3F3M2 S1F1M2 S2F3M3 S1F1M1
S3F2M1 S1F2M2 S2F1M3 S3F2M3 S1F2M1 S2F1M2 S2F2M2 S1F2M3 S2F2M1
S2F3M1 S3F1M2 S1F3M3 S3F1M3 S2F1M1 S2F2M2 S3F3M2 S3F1M3 S3F3M1
S3F3M1 S2F3M2 S1F1M3 S3F3M3 S2F3M1 S1F1M2 S3F1M2 S2F1M3 S3F1M1
S2F1M1 S1F3M2 S2F3M3 S2F1M3 S3F1M1 S2F3M2 S3F2M2 S3F3M3 S2F3M1
S1F2M1 S3F2M2 S3F2M3 S1F2M3 S1F3M1 S3F2M2 S1F2M2 S1F3M3 S1F2M1
S1F3M1 S2F1M2 S3F1M3 S1F3M3 S3F2M1 S3F1M2 S1F3M2 S2F2M3 S1F3M1
S2F2M1 S3F3M2 S1F2M3 S2F3M3 S3F3M1 S1F2M2 S2F1M2 S3F2M3 S3F2M1
S3F1M1 S1F1M2 S2F2M3 S2F2M3 S1F1M1 S1F3M2 S2F3M2 S1F1M3 S2F1M1
S = drum speed, F = feed rate, M =moisture content, & R = replications

Evaluation procedure
The chamber rate, flow rate, Level of moisture, and convex aperture width of the thresher were
evaluated at three different levels on a firm surface after installation and adjustments. With regards
to the trial, the consequence of their separate parameters on sprouting, threshing efficiency and
grain damage was considered. Samples were randomly prepared and put into the thresher once it
was turned on to obtain the thresher performance indices. According to Wang and Cichy (2024),
the effectiveness of threshing (TE), the aptitude for threshing (TC), effective cleaning (CE), and
proportional of losses were determined using the following relationships to assess the threshing
machine's effectiveness.

RESULTS AND DISCUSSION

Threshing Efficiency
The figures 2a–c were prepared using optimal feeding amounts of 672 kg/h, 37.4cm concave
clearance, and 8.25 ms-1 drum speed. Threshing efficiency improved together with concave
geometry clearance and rate of feed, as Figure 2a presented. Threshing efficiency attained a highest
of 98.7% at an average feed rate of 672 kg/h and a convex clearance of 37.4 cm. Figure 2b illustrates
how increasing the rate of feed and speed of the drum led to an enhancement to the effectiveness
of threshing. The most significant threshing efficiency (99.7%) was ascertained with an intake rate
of 672 kg/h and a drum with a speed of 8.25 ms-1. In contrast, the efficiency of threshing climbed
in tandem with the drum speed improved and convex clearance dropped. The drum speed at which
the highest efficiency (99%) was achieved was 8.25ms -1 and a concave clearance of 37.4mm (Fig.
2c). Threshing efficiency improved when the rubbing force between the bean and the canvas
concave increased, corresponding with a decrease in convex clearance between the concave strip
and the concave bar. As perimeter rate climbed, so did momentum and thrust of impact on the
165
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

trembling, which in turn boosted threshing efficiency as drum speed climbed. When it came to
bean threshers, Umbataliyev et al. (2023) discovered similar patterns.

The experimental findings are illustrated in Figures 2-d to -f. throughout the range of input
components examined, the threshing efficiency varied between 95.1 and 99%. At the 1%
confidence level, Table 2 illustrates that threshing efficiency was significantly impacted through
the rate of feed, cylinder speed, level of moisture, and convex clearance. The impact of the chamber
frequency on common bean effectiveness of threshing is illustrated in Figure 2-d. When cylinder
speed was increased from 7.5 to 9.17 ms-1, threshing efficiency climbed from 96.81 to 99.21% with
a moisture level of 11.6%. Furthermore, as anticipated, the highest cylinder speed (10.83 ms -1)
produced the highest threshing efficiency rating (99.69%).

Figure 2. The implications of those parameters on the threshing efficiency: (a) feed rate and concave
clearance; (b) drum speed and feed rate; (c) drum speed and concave clearance; (d) cylinder speed
and moisture content; (e) cylinder speed and concave clearance; and (f) moisture content and feed
rate.

As concave aperture increased, threshing efficiency decreased, as Figure 2-e illustrates. Considering
improvements in convex aperture from 35 to 45 mm, the threshing efficacy reduced from 97.45 to
96.16% at 7.5 ms-1 with the chamber's frequency. Convex space did not significantly influence
performances at speed of drum exceeding 9.17 ms -1. The higher cylinder speed resulted in refined
threshing efficiency because of an increased impact force. The rationale for lowest threshing efficiency
at the highest concave clearance was the insufficient force exerted on the pods, which caused them to
fall out without separating the seeds. At the 1% confidence level, the concave clearance and cylinder
speed influences on the threshing efficiency interacted significantly. There was a negative correlation
between the feed rate and threshing efficiency.

166
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

As the feed consumption rate went up from 550 to 750kgh-1, the average threshing efficiency reduced
from 99.52 to 99.09% in (Figure 2f). Outcomes of the investigation indicated that the detrimental
impact of cylinder speed on crop threshing was mitigated as the rate of feed escalated due to an increase
in the width of the trim slice between the cylinder and concave. For every drum speed level, Huertas
et al. (2023) found that as feed rate climbed the effectiveness of threshing decreased.

The efficiency of threshing dramatically dropped as the input material's level of moisture escalated,
shown in Figure 2-f. There was a correlation between the highest (99.52%) and minimum (98.31%)
effectiveness of threshing and the amounts of water of 5% and 15%, within that sequence. At increasing
levels of water content, there was a greater impact of moisture content on threshing efficiency. Que et
al. (2024) also reported a similar outcome. Pods and seeds are more easily split because there is less
tension holding the pod together and the pods are more brittle due to reduced seed moisture
concentrations. Threshing efficiency dropped because of increased pod cohesion brought on by the
plant materials' increased flexibility at higher moisture contents.

The ANOVA illustrated in Table 2 (p<0.001) implies that the predicted value of F (19.81) is high,
indicating that a model with quadratic parameters could be a good fit for the outcomes of the
experiment. Table 2 illustrates the F-values that demonstrate the significant impact of the rate of feed,
convex clearance, and speed of drum in terms of linear regression on the effectiveness of shredding at
the 1% significance level. In a similar manner at the 5% significance level, the interaction terms
between the drum speed x convex clearances and the drum speed a quadratic term exhibited an
important impact on the threshing effectiveness. The remaining interactive and graphs had no
discernible effect on the threshing effectiveness while not even at the 10% threshold of significance.

Detects the proportion of noise to signal with adequate precision; A value more than four is preferred.
In this case, the ratio changed to 16.577, indicating a strong pulse. Savic and Savic-Gajic (2021) assert
that this framework can be used for maneuvering within the realm of design. This model's predicted
R2 (0.81) and adjusted R2 (0.89) agreed. Using polynomial form fitting, the regression model
illustrating the threshing efficiency change with regard to the independent parameters (𝑓𝑒𝑒𝑑 𝑟𝑎𝑡𝑒, 𝐹𝑟 ),
(𝑑𝑟𝑢𝑚 𝑠𝑝𝑒𝑒𝑑, 𝑣𝑠 ) and (𝑐𝑜𝑛𝑐𝑎𝑣𝑒 𝑐𝑙𝑒𝑎𝑟𝑎𝑛𝑐𝑒, 𝐶𝑐 ) was produced. The simplified polynomial model was
obtained by removing terms from the quadrilateral model that are not significant (Savic et al. 2019)

Grain Damage
The variation among the investigation's outcomes illustrated that the convex aperture, chamber
rate, rate of feeding, and levels of water content are all exhibited a significant impact on the amount
of grain damage (Table 2). The most significant factors were determined to be the cylinder speed,
which was followed by rate of feeding, moisture level, and convex aperture. First-order interactions
were prioritized according to relevance: chamber frequency × level of moisture, feed rate × level
of moisture, chamber frequency × convex aperture, and cylinder speed × cylinder speed. The
implications of convex clearance and speed of cylinder on the percentage of grain damage are
shown in Figure 3a. This figure illustrates how the rotational frequency at which the drum is
threshed enhances the amount of grain impairment.

Damage of grains escalated from 4.98 to 47.97% at the convex of 35 mm when the drum speed
increased from 7.5 to 10.83 ms-1. When cylinder speed was raised from 7.5 to 10.83 ms-1, Grain
breakage escalated from 1.71 to 33.29% at a convex aperture of 35mm. During threshing, the

167
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

common bean was subjected to higher impact levels, which increased damage. However, as
concave clearance improved, grain damage drastically decreased.

Table 2. Response surface quadratic model-based analysis of variance for common bean
threshing.

Source of variation dfa Grain damage Threshing efficiency

Model 54 164.62** 99.73**
Cys 1 1437.46** 930.83**
Fr 1 78.75** 21.01**
Cc 1 70.63** 60.35**
Mc 1 232.06** 144.55**
Cys× Fr 1 15.69** 13.34**
Cc ×Fr 1 0.83ns b
0.082ns
Fr ×Mc 1 1 5.46* 0.92ns
Cc ×Cys 1 1 24.03** 35.36**
Cc ×Mc 1 3.59ns 0.78ns
Cys× Mc 1 94.68** 21.16**
( Mc)2 1 5.38* 2.51ns
(Cys)2 1 93.47** 52.46**
( Fr)2 1 0.46ns 1.60ns
(Cc)2 1 1.45ns 0.090ns
Res. 15
Pe 5
Corr. total 69
*Significant at the 5% level; **highly significant at the 1% level; a Degrees of freedom, b Non-significant,
Fr = Feed rate, Cs = Drum speed, Cc = Convex aperture, Mc = Level of moisture, Res. = Residual, Pe =
pure error, Corr. = Correlation total

Grain damage and rate of feeding interacted inversely with each other across independent variables.
Since the crop was subjected to more intense contact at the lower feed rate, the reduction in grain
damage was approximately 50% (Figure 3b) when the concave clearance of 37.4mm was attained
while upgrading the intake rate from 550 to 750 kg/h. Additionally, according to Ghebrekidan et
al. (2024)), grain damage increased as feed rate diminished.

When the amount of moisture escalated, the proportion of grain damage dropped dramatically, as
shown in Figure 3c. On the other hand, grain loss went from 33.42 to 57.79% when the amount of
moisture decreased at a speed of 10.83 m/s, from 15% to 5%. At lower cylinder speeds, the impact
of moisture content on grain damage was minimal. When moisture content was reduced from 15%
to 5%, grain damage increased from 5.52 to 10.51% at a cylinder speed of 7.5m/s. Grain elastic
behavior increased with increasing moisture content; hence, more energy was needed to crack the
grain. Moisture content has also been identified by several researchers as a significant factor
influencing grain impairment (Huertas et al., 2023; Chandra et al., 2024).

A 672 kg/h rate for feeding, a concave clearance of 35cm, and a drum speed of 8.25m/s were the
optimal parameters for preparing the Fig. 3g–i. As illustrated in Fig. 3g, the greatest damage to the
grain appeared at 35–45 mm convex clearance at rates of feed varied from 650–750 kg/h. There
was no evidence of damage to grains within the 35–38 mm convex clearance range at 650–675

168
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

kg/h amount of intake. The greatest amount of grain impairment has been observed to be 3.5% at
25 cm convex spacing and 750 kg/h amount of intake. Figure 3h showed the proportion of damaged
grains emerged in tandem with raised rate of feed and drum rpm. With an amount of intake 750
kg/h and a drum rate of 10.83 m/s, the ultimate breakdown of grain was achieved, at 3.3%.

Likewise, there was an increase in damage to grains when the drum speed climbed, and the convex
clearance diminished. At a chamber inclination of 10.83 m/s and a convex space of 25 mm, the
highest possible 3.8% loss of grain was seen (Figure 3i). The reduction in convex clearance led to
an increase in the contacting action between the grains and the covering stripe, degrading the grains.
Moreover, it happened because there was more intimate interaction among the beans and the canvas
strip and the segments of the chamber that convex. Significant forces from impacts were detected
when the drum was moving faster. The maximum grain damage was caused by those maximal
impact forces. At lower drum speeds, the maximum grain damage is caused by these maximum
impact forces, and vice versa. Grain damage was found to be decreased at higher feed rates because
maximum feed rates share the power of collision and contacting force produced by drums in
rotation, whereas minimum feed rates handle the greatest the power of collision and contacting
force, which results in highest degree of scratches. Similar findings with respect to the multi-
threshing machine (Huertas et al., 2023; Chandra et al., 2024) published their findings. The greater
feed that was shared by the impact and rubbing power of the revolving drum resulted in less
degradation of grain when the degree of feeding escalated.

Figure 3. The implications on grain damage percentage of (a) cylinder speed and concave
clearance, (b) feed rate and concave clearance, (c) cylinder speed and moisture content, (g) concave
clearance and feed rate, (h) drum speed and feed rate, and (i) drum speed and concave clearance.

The influence of the rate of feeding (kg/h), speed of the drum (m/s), and convex clearance (mm)
on common bean damage was investigated using the implementation of the ANOVA described in

169
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 2. It is evident from emulate F’s considerable value (44.34) (p < 0.001) that an equation with
quadratic might correspond to the empirical information efficiently. The linear parameters of rate
of feeding, drum speed, convex space, interaction coefficient curvature space x speed of drum, and
nonlinear term convex clearance all had a significant impact on grains damage at the 1% level of
significance, based on Table 2's F-values.

At the 5% significance level, the rate feed x drum speed interaction term also significantly
influenced the degree of grain impairment. The damage to the beans was not significantly impacted
by the relationship between the terms flow rate x concave aperture or the quadrilateral in relation
to convex geometry and intake rate variables, irrespective of the significance threshold of 10% (p
< 0.1).

Sufficient accuracy is used to measure the signal to noise proportion. Therefore, the ratio should
be higher than four. In this instance, the ratio changed to 22.74, indicating a strong signal. To
navigate the design space, one can apply this model (Savic et al. 2019; Savic and Savic-Gajic,
2021). This model's predicted R2 (0.89) and adjusted R2 (0.95) agreed. Polynomial form fitting was
used to generate the regression equation that shows a variation of the percentage of grain damage
(GD, %) with respect to the independent parameters (𝑓𝑒𝑒𝑑 𝑟𝑎𝑡𝑒, 𝐹𝑟 ), (𝑑𝑟𝑢𝑚 𝑠𝑝𝑒𝑒𝑑, 𝑣𝑠 ) and
(𝑐𝑜𝑛𝑐𝑎𝑣𝑒 𝑐𝑙𝑒𝑎𝑟𝑎𝑛𝑐𝑒, 𝐶𝑐 ). The exponential model's insignificant terms had been eliminated to
create the simplified multiplication framework (Savic et al. 2019)

Optimization of MARC bean thresher

The graphical optimization and optimal outcomes are shown in Figure 4. The machine's
independent design parameters, which are connected to these outcomes, establish the optimal
ranges of cleaning efficiency, threshing efficiency, and grain damage. The predicted percentages
for cleaning efficiency, grain damage, and threshing efficiency were 85%, 0.086%, and 97.94%,
respectively. By using graphical optimization, the optimal values of several variables were found,
including concave clearances of 25 – 45mm with 87.94% efficiency of threshing, 85% cleaning
effectiveness, and 0.086% fractures.

The marked region of Figure 4a–c displays the collective outcomes of this optimization. The same
values were obtained by the numerical and graphical optimization techniques (Benaseer et al. 2018;
Umbataliyev et al., 2023). These optimal features guided the development of the drum, which was
then finished and its performance assessed to validate the chosen parameters. The findings
indicated that the percentage of cleaning, detrimental to the grain and spinning was 86% compared
to 85%, 99%, and 0.1%, respectively, compared to predictions of 97.94% and 0.086%. As a result,
a cylinder speed of 8.25 ms-1, convex aperture of 37.4mm, rate of feed 672kgh-1, and level of
moisture 11.6% were recommended for threshing common beans.

170
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 4. Graphical optimization of the operating parameters of the threshing drum; (a)
Superimposed contours for threshing efficiency, pearling efficiency, and damage to bean at varying
feed rates and concave clearance; (b) Superimposed contours for threshing efficiency, pearling
efficiency, and speed of the drum at varying feed rates; and (c) Superimposed contours for
threshing efficiency, drum speeds, and concave clearance at varying feed rates.

CONCLUSIONS

The threshing drum of the MARC bean thresher is one of its essential parts and its performance is
depending on its operational parameters. Important variables influencing grain damage, threshing
efficiency and cleaning efficiency in common bean threshed seed quality are rate of feed, moisture
level, convex aperture, and drum speed. The most significant crop and machine measurement was
cylinder speed, which was subsequently the moisture level. The percentage of damaged grain
improved from 45.98 to 47.97% and the overall threshing efficiency elevated from 96.81 to 98.69%
when the speed of drum was varied from 7.5 to 10.83 ms -1. Increased moisture content was
associated with increased grain damage, efficiency of threshing and rates of seed germination. The
proportion of grain impairment, and threshing efficiency were all significantly (P<0.01) impacted
by concave clearance. Within the 550–750 kg h-1 rate of feed range, there was variation in the
average value of damage to grain (16.65–7.67%) and threshing efficiency (96.52-28.09%). As a
result, a cylinder speed of 8.25 ms-1, convex aperture of 37.4mm, rate of feed 672kgh-1, and
moisture level of 11.6% were recommended for threshing common beans.

REFERENCES

Adewoyin, O. B. (2023). Pre-Harvest and Postharvest Factors Affecting Quality and Shelf Life of
Harvested Produce. In New Advances in Postharvest Technology. IntechOpen. 42, 28-420
Benaseer, S., Masilamani, P., Albert, V. A., Govindaraj, M., Selvaraju, P., & Bhaskaran, M. (2018).
Impact of harvesting and threshing methods on seed quality: A review. Agricultural Reviews,
39(3), 183–192.
Bento, J. A. C., Gomes, M. J. C., Bassinello, P. Z., Martino, H. S. D., de Souza Neto, M. A., &
Oomah, B. D. (2022). Benefits of Carioca Beans (Phaseolus vulgaris) Molecular
Mechanisms to Human Health and Nutrition. Molecular Mechanisms of Functional Food,
112-141.
Chandra, R. J., Masilamani, P., Suthakar, B., Rajkumar, P., Sivakumar, S. D., & Manonmani, V.
(2024). Effect of Moisture Content on Combine Harvested Seed Crop and its Quality. Journal
of Experimental Agriculture International, 46(3), 114-138.
171
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Ejara, E., Mohammed, W., & Amsalu, B. (2018). Genetic variability, heritability and expected
genetic advance of yield and yield related traits in common bean genotypes (Phaseolus
vulgaris L.) at Abaya and Yabello, Southern Ethiopia. African Journal of
Biotechnology, 17(31), 973-980.
Ghebrekidan, B. Z., Olaniyan, A. M., Wako, A., Tadesse, A. G., Alemu, D., & Lema, T. (2024).
Gravimetric characteristics and friction parameters of common bean (Phaseolus vulgaris
L.). Turkish Journal of Agricultural Engineering Research, 5(1), 76-93.
Güvercin, S., & Yıldız, A. (2018). Optimization of cutting parameters using the response surface
method. Sigma Journal of Engineering and Natural Sciences, 36(1), 113-121.
Huertas, R., Karpinska, B., Ngala, S., Mkandawire, B., Maling'a, J., Wajenkeche, E., & Foyer, C.
H. (2023). Biofortification of common bean (Phaseolus vulgaris L.) with iron and zinc:
Achievements and challenges. Food and Energy Security, 12(2),e406.
Jan, S., Rather, I. A., Sofi, P. A., Wani, M. A., Sheikh, F. A., Bhat, M. A., & Mir, R. R. (2021).
Characterization of common bean (Phaseolus vulgaris L.) germplasm for morphological and
seed nutrient traits from Western Himalayas. Legume Science, 3(2), e86.
Joshi-Saha, A., Sethy, S. K., Misra, G., Dixit, G. P., Srivastava, A. K., & Sarker, A. (2022).
Biofortified legumes: Present scenario, possibilities and challenges. Field Crops Research,
279,108467
Juraev, D. T., Dilmurodov, S. D., Kayumov, N. S., Xujakulova, S. R., & Karshiyeva, U. S. (2023).
Evaluating Genetic Variability and Biometric Indicators in Bread Wheat Varieties:
Implications for Modern Selection Methods. Asian Journal of Agricultural and Horticultural
Research, 10(4), 335-351.
Lee, G. H., Moon, B. E., Basak, J. K., Kim, N. E., Paudel, B., Jeon, S. W., ... & Kim, H. T. (2023).
Assessment of Load on Threshing Bar During Soybean Pod Threshing. Journal of
Biosystems Engineering, 48(4), 478-486.
Lisciani, S., Marconi, S., Le Donne, C., Camilli, E., Aguzzi, A., Gabrielli, P., & Ferrari, M. (2024).
Legumes and common beans in sustainable diets: nutritional quality, environmental benefits,
spread and use in food preparations. Frontiers in Nutrition, 11,1385232.
Parker, T. A., Gallegos, J. A., Beaver, J., Brick, M., Brown, J. K., Cichy, K., & Gepts, P. (2022).
Genetic resources and breeding priorities in Phaseolus beans: Vulnerability, resilience, and
future challenges. Plant breeding reviews, 46, 289-420.
Que, K., Tang, Z., Wang, T., Su, Z., & Ding, Z. (2024). Effects of Unbalanced Incentives on
Threshing Drum Stability during Rice Threshing. Agriculture, 14(5),777.
Savic, I. M., & Savic Gajic, I. M. (2021). Optimization study on extraction of antioxidants from
plum seeds (Prunus domestica L.). Optimization and Engineering, 22, 141-158.
Savic, I. M., Nikolic, I. L., Savic-Gajic, I. M., & Kundakovic, T. D. (2019). Modeling and
optimization of bioactive compounds from chickpea seeds (Cicer arietinum L). Separation
Science and Technology, 54(5), 837-846.
Strecker, K., Bitzer, V., & Kruijssen, F. (2022). Critical stages for post-harvest losses and nutrition
outcomes in the value chains of bush beans and nightshade in Uganda. Food Security, 14(2),
411-426.
Uebersax, M. A., Cichy, K. A., Gomez, F. E., Porch, T. G., Heitholt, J., Osorno, J. M., ... & Bales,
S. (2023). Dry beans (Phaseolus vulgaris L.) as a vital component of sustainable agriculture
and food security—A review. Legume science, 5(1), e155.
Umbataliyev, N., Smailova, G., Toilybayev, M., Sansyzbayev, K., Koshanova, S., &
Bekmukhanbetova, S. (2023). Optimization Of The Technological Process Of Threshing
Combine Harvester. Eastern-European Journal of Enterprise Technologies, 124(1).
172
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Wang, W., & Cichy, K. A. (2024). Genetic variability for susceptibility to seed coat mechanical
damage and relationship to end‐use quality in kidney beans. Crop Science, 64(1), 200-210.

173
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REMOTE AND PROXIMAL SENSING

174
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

HIGH-RESOLUTION MAPPING OF LONG-TERM SOIL ORGANIC CARBON

STOCKS AND CHANGES IN MOROCCO
#11287

1*Mohamed Bayad, 1Bruno Gerard, 1Abdellah Hamma, 2Malcolm J. Hawkesford, 3Eyal Ben-Dor,
4Henry Wai Chau
1College of Sustainable Agriculture and Environmental Sciences, Mohammed VI Polytechnic

University, Benguérir, Morocco; 2Rothamsted Research, Harpenden AL5 2JQ, UK; 3Remote
Sensing Laboratory, Geography Department, Porter School of the Environment and Earth
Sciences, Faculty of Exact Sciences, Tel Aviv University, Tel Aviv 699780, Israel; 4Lethbridge
Research and Development Centre, Agriculture and Agri-Food Canada, Lethbridge, Alberta,
Canada
* Corresponding author at: College of Sustainable Agriculture and Environmental Sciences,
Mohammed VI Polytechnic University, Benguérir, Morocco.
*
e-mail: [email protected]

ABSTRACT

Soil organic carbon (SOC) is an important attribute for soil productivity and climate change
mitigation. It stabilizes the soil structure and provides nutrients to the soil solution while playing a
major role in carbon sequestration processes. Current regional SOC maps are not detailed enough
and thus, do not support decision-making at farm and landscape and do not track long-term changes
of carbon. Using large soil dataset, multispectral satellite data, climate data and machine learning
approach, we created a topsoil (0-30 cm), 30m spatial resolution soil carbon stocks and temporal
changes map of Morocco over the last 32 years. Our results show a total topsoil SOC stock of 3.57
Pg C, with a median SOC density of 4.98 kg C m−2. The Moroccan biomes have acted as a net
carbon sink in the last 32 years and absorbed an average 3.11 Mt C yr -1, i.e., only 15.4% of the
current anthropogenic annual carbon emissions of Morocco. However, high losses are estimated in
niche areas such as the Acacia-Argania biosphere, parts of the coastal Mediterranean forest and
large cropland-dominated areas due to anthropogenic pressure. The strength of sequestration is
likely to diminish, if necessary, measures are not taken to protect these active carbon sinks. The
present SOC mapping approach uses the largest soil C database ever recorded in North Africa and
provides more accurate predictions compared to other regional studies. Our maps will help land
managers and decision-makers improve climate mitigation actions and help understand trade-offs
between soil carbon, biodiversity traits, and ecosystem management.

INTRODUCTION

Soil organic carbon (SOC) has received significant attention as a critical carbon pool of the
terrestrial biosphere as well as a crucial soil property that governs soil health. Globally, about 2300–
2500 Pg (1015g) C (60% organic and 40% inorganic) is sequestered in the top 2 m of soil, of which
approximately 30% is stored in the 0-20 cm topsoil (Batjes, 1998; Paustian et al., 2016). Whilst the
top 2m soil pool is hardly accessible to agroecosystem manipulation, the top 30 cm soil layer has
promise as the most manipulatable layer through agroecosystem changes as it represents the root
zone and the interface between the pedosphere and atmosphere. The global topsoil (0-30cm)
organic carbon stocks (1500 Pg C) represent more than three times as much carbon as either the

175
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

atmospheric CO2 or the above-ground biomass. This makes the terrestrial biosphere a potential sink
or source of atmospheric CO2. Since the anthropogenic exploitation of terrestrial ecosystems can
alter the SOC pool drastically (Deng et al., 2016), substantial efforts have been made to consolidate
its potential role as a net sink of atmospheric CO 2. It has long been understood that ecosystem
management and disturbance can affect the organic carbon stocks of the soil, and thus affect soil
quality and atmospheric CO2 emissions. Organic carbon plays a vital role in ecosystem
sustainability and species occurrence and survival, which in turn control organic carbon inputs and
cycling in the soil.

The dynamics of soil organic carbon are primarily influenced by the interplay of carbon inputs and
residence time in the soil, which are influenced by various processes including net primary
productivity, decomposition and factors such as fire and grazing, that can either facilitate or impede
SOC loss or retention (Lal, 2004). At the regional scale, climatic factors and elevation play
significant roles in determining soil C balance (Jobbágy & Jackson, 2000), whilst at farm and field
levels, soil texture, mineralogy and topography interact with climate to shape SOC dynamics
(Batjes, 1996; Bellamy et al., 2005). Temperature and precipitation regimes drive the occurrence
of plant species with analogous functional traits within conspicuous areas forming biomes
(Woodward et al., 2004). The species abundance, productivity, and functional traits are per se the
main drivers of soil carbon inputs. Still, species interactions may also play a role in carbon
dynamics (De Deyn, 2008).

Globally, there has been substantial interest in carbon sequestration in agricultural soils, not only
to reach CO2 mitigation targets, but also to enhance soil health (Frank et al., 2015; Lal, 63 2004).
Carbon dioxide emissions caused by land use changes include deforestation, conversion from
natural to farming ecosystems, biomass burning and drainage of wetlands for agriculture
development (Lal, 2006). Some cultivated soils have lost 50-60% of the initial SOC stocks causing
the release of up to 78 Pg C into the atmosphere. These losses are exacerbated by land misuse and
poor soil management (Lal, 2004). Previous research showed that soil 68 organic C potential for
CO2 sequestration can be improved dramatically through ecosystem restoration strategies, smart
cultivation, and improved management practices in agricultural lands. Lal (2004) recommended a
range of improved management practices to enhance C stocks in agricultural soils.

Different climatic zones exhibit distinct patterns of SOC accumulation. Cold and wet climates tend
to promote high primary productivity and low decomposition rates, resulting in the build-up of
SOC (Batjes, 1996; Jobbágy and Jackson, 2000). Arid regions, on the other hand, typically have
low SOC due to limited biomass production (Schlesinger, 1977). Tropical regions, however,
display intermediate SOC levels due to their high rates of primary productivity, which offsets rapid
decomposition (Houghton, 2007; Davidson et al., 2014). In temperate ecosystems, environmental
and biological factors determine the persistence of SOC (Schmidt 79 et al., 2011). Houghton (2007)
suggests that globally, high-latitude areas have the highest levels of SOC due to the slow
decomposition caused by low temperatures and are still serving as a net sink for CO 2. The Atlas
Mountains ecosystem might be potential carbon sink in North Africa as they were reported to have
high SOC stocks (Sabir et al., 2020). Apart from climate, the characteristics of parent material and
soil properties also influence SOC persistence. The association of SOC with minerals and the
formation of soil aggregates play important roles in SOC retention (Chenu et al., 2000).

176
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Multiple lines of evidence indicate that climate change is altering terrestrial SOC stocks, primarily
by accelerating the decomposition rate. Despite large uncertainties related to the magnitude of the
losses, climate-carbon cycle feedback has an undeniably significant impact on SOC (Walker et al.,
2018). Terrestrial air temperature increased by 1.03°C on average between 1919 and 2018, which
could have caused an average loss of 2.5 ± 5.5% of the agricultural topsoil (0-30 cm) SOC (Poeplau
& Dechow, 2023). Moreover, climate change can alter soil carbon indirectly through increasing
the occurrence of wildfires. The effect of wildfires on SOC depends on various factors such as fire
severity, fire frequency, vegetation type, climate, and soil properties. The immediate effect of
wildfires is the combustion of above-ground vegetation, which can lead to a substantial release of
CO2 into the atmosphere. The most intuitive impact soils undergo during a fire is the loss of organic
matter. Subject to fire severity, organic carbon can be volatilized, charred, or completely
mineralized. Up to 15% of the burned biomass is transformed to pyrogenic organic carbon (Santín
et al., 2015), whose residence time lasts from decades to millennia. In the last decade, Morocco has
lost nearly 77,000 ha of land to wildfire with 32,000 ha recorded in 2022 alone. However, the
impact of wildfires on SOC stock in forest ecosystems in Morocco has not been studied. The
recovery of SOC in burnt forests could occur rather quickly with the natural or artificial
resettlement of vegetation, due to the high productivity attributed to secondary ecological
successions (Certini, 2005). Baudena et al., (2020) suggested that recurring fires could transform
Mediterranean forests into shrublands, hosting flammable biomass that regrows rapidly after fire.
The authors theorized that this mechanism allegedly benefits shrubland persistence and may be
enhanced in the future, with an eventual aridity increase (Baudena et al., 2020). Johnson & Curtis,
(2001) revealed a post-fire time effect on soil organic carbon in forest ecosystems, using a meta-
analysis of 48 different studies.

Given the high importance of organic carbon as a soil health indicator and a potential global carbon
sink, accurate characterization is of utmost importance. A growing body of literature has shown
complementarity between remote sensing and ecosystem modelling in studying organic carbon in
the biosphere (Turner et al., 2004). Conventional approaches to soil organic carbon mapping
include geostatistical methods that depend greatly on soil sampling (e.g., regression kriging
(Somarathna et al., 2016)), or relate SOC status solely to land use and landcover (Minelli, 2018).
These methods have a major limitation as they do not allow monitoring of soil carbon status over
time, without recourse to new observations. Advances in cloud computing and remote sensing have
opened new horizons for spatiotemporal assessment of soil organic carbon mapping from farm to
global scale. Several studies have attempted machine learning, remote sensing, climate and
biological predictors for high-resolution of SOC mapping at the country scale. For example, Venter
et al., (2021) produced a low uncertainty prediction model of SOC stocks in South Africa’s natural
soils. The authors suggested a long-term carbon change map based on the high accuracy model.

In this study, we attempted the construction of a national long-term soil organic C stocks map for
Morocco. We also aimed to improve the prediction accuracy of organic carbon using a large soil
dataset, Landsat satellite imagery, climate and vegetation proxies in a machine-learning workflow.
This method also permitted the estimation of 32 years of SOC stocks dynamics at 30 m spatial
resolution mapping. These high-resolution maps are required to understand the national trends of
soil carbon stocks from landscape to national scale. The resulting maps of soil carbon stocks and
changes will inform future research on the drivers impacting potential active carbon sinks and will
guide restoration efforts to reverse losses while preserving ecosystem vital functions.

177
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

MATERIALS AND METHODS

Study area
Morocco comprises eight ecoregions with contrasting north-south primary productivity and
precipitation gradients. These ecoregions represent four different terrestrial biomes including 1)
Mediterranean Forests, Woodlands and Scrub, 2) Temperate Coniferous Forests 3) Mediterranean
Grasslands and Shrublands 4) Deserts and Xeric Shrublands. The Mediterranean woodlands and
forest in the north are characterized by hot and dry summers and pleasant and humid winters. North
Saharan Xeric Steppe and woodland and south Sahara Desert experience low rainfall (50-100mm)
in the winter and high temperatures (40-45 °C) during summer. Mediterranean Acacia-Argania dry
woodlands and succulent thickets cover the northwest of the country (Fig. 1).

Soil carbon data

Figure 1. Major soil types (Dewitte et al., 2013) and eco-regions of Morocco (Dinerstein et al,
2017).

The complexity of the ecosystem resulted in diverse soil genesis that produced variable soil types.
Moroccan soils are predominantly Calcisols, Luvisols, Cambisols, Leptosols and Kastanozems
(Fig. 1). Other under-represented typologies include Vertisols, Regosols, Planosols and Fluvisols.
The anthropogenic impact includes a wide range of land use going rom intensive cropping in plains
and plateaux to complex agroecosystems including tree cultivation and grazing in high altitudes.
Cultivated land represents around 12% of the total surface area of Morocco (8.7 M ha).

Over 52,000 soil samples were collected within Fertimap project the Al-Moutmir extension
program backed by Mohammed VI Polytechnic University (UM6P) and OCP Morocco. The soil
sampling campaigns occurred between 2011 and 2020. Topsoil (0-30cm) was sampled from

178
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

agriculture and natural ecosystems and soil organic carbon content was analyzed using the
Walkley-Black oxidation method (Walkley and Black, 1934). Soil stocks were estimated using 162
the following equation.

𝑆𝑂𝐶 𝑠𝑡𝑜𝑐𝑘 (𝑘𝑔 𝐶 𝑚−2) = 𝑆𝑂𝐶 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑔 𝑘𝑔−1) × 𝐵𝐷 (𝑔 𝑐𝑚−3)×𝑑 (𝑐𝑚)

Prediction covariables and modelling

Where 𝑆𝑂𝐶𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 is organic carbon concentration and BD is bulk density that was not
measured but estimated using a linear pedo-transfer function inferred from Ruehlmann &
Körschens (2009). The coarse elements percentage was not considered because of the lack of this
data.

The covariates used represent proxies for climate, surface biomass, and topography as determining
factors of SOC stocks. In natural ecosystems, SOC is more likely to be controlled by environmental
variables (e.g., climate, biomass, topography). In cultivated land, SOC dynamics are also strongly
impacted by anthropogenic factors, which include tillage, cropping rotation, irrigation, residue
management, etc. Temporal dynamics of Landsat data (e.g., NDVI) may inform on cropping
intensity and even crop classification. Surface reflectance time series from Landsat 5, 7, and 8 were
used from 1990 to 2022. Landsat surface reflectance (L2SR) data archives provided by USGS were
atmospherically corrected by the Land Surface Reflectance Code. Clouds, cloud shadow and snow
were masked using 'QA_PIXEL' band. Surface reflectance data from the tree sensors was
harmonized using the cross-calibration method from Roy et al., (2016). Annual median and
variance composites of NDVI and reflectance from all bands were calculated and tested as
predictors.

The environmental covariates include mean climate water deficit, precipitation, Palmer Drought
Severity Index, minimal temperature, and maximal temperature form the TerraClimate dataset
provided by the University of California Merced (Abatzoglou et al., 2018). Topographic used
predictors are data elevation, slope, and aspect are provided by NASA, USGS, and JPL-Caltech
(Farr et al., 2007). Topographic diversity index derived from ALOS provided by Conservation
Science Partners (Theobald et al., 2015). The Fraction of Absorbed Photosynthetically Active
Radiation (FAPAR) and leaf area index (LAI) derived from the AVHRR sensor onboard the NOAA
satellite (Claverie et al., 2014). Net primary productivity derived from MODIS and provided by
NASA's Land Processes Distributed Active Archive Center (LP DAAC). Two- and 5-year median
aggregates of all the predictors, prior sampling dates, were tested for predicting SOC stocks using
a random forest algorithm. The used predictors are summarized in Table 1. After several iterations,
the best model was adopted, and some variables were excluded because of their non-availability
for the whole of the studied 32 years period and their low impact on model accuracy. All the
predictors used for the final model extend from over the whole study period (1990-2022). The data
processing and modelling workflow is summarized in Figure 3.

The random forest model hyperparameters ntree and mtry were set to 500 and the square root of
the number of variables, respectively. A 30% sampling points subset was used for model validation.
Models’ prediction accuracy was evaluated using the coefficient of determination (R2) and root
mean scare error (RMSE). The variables' importance of the random forest model is derived from
the sum of decreases in the Gini impurity index, to see what predictors are more relevant in SOC
stocks prediction.
179
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 2. Distribution map of soil sampling across Morocco. Sampling points are colored by 170
SOC values (kg C m-2). Insert plots represent count of samples per years and distribution 171
histograms of SOC (kg C m-2) per biomes.

The validated random forest model was used to predict SOC stocks based on satellite and the
environmental predictors from 1990 to 2022. The predictor variables were aggregated in 5-year
median and used as inputs of random forest to predict annual SOC stocks (Fig. 3). The annual
predictions served as basis for estimation of long-term average stocks and changes. The carbon
stock changes were estimated using the Sen’s slope (Sen, 1968) of the predicted stock time series

180
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

at each pixel. This method has been used by Venter et., (2021) to estimate long-term SOC stocks
temporal dynamics and the changes were estimated as:

Δ𝑆𝑂𝐶 (%) = s÷𝑆𝑂𝐶𝐿𝑇𝐴×100216

where 𝑠 is the Sen’s slope and 𝑆𝑂𝐶𝐿𝑇𝐴 _is long-term average C stocks of the considered pixel.

Figure 3. Workflow diagram that summarizes the data preparation and modelling framework.

RESULTS

Results indicate a total SOC stock of 3.57 Pg C in the years 2018 and a long-term 32-year average
stock of 3.94 Pg C. This provides an estimate of the overall carbon storage capacity in the assessed
area (~700,000 km2) over all Moroccan biomes. The soil C stocks had an average 227 of 5.14 kg
C m-2 and a median of 5.12 kg C m-2 (Table 2, and Table S1). This metric provides insights into
the typical carbon content per unit area and helps in assessing the baseline SOC levels in Moroccan
ecosystems.

This study reveals significant variations in SOC stocks distribution between northern-west and
southern regions simulating the north-south climate gradient (Fig. 4 and Fig. S2). The northwestern
and High Atlas areas exhibited higher carbon stocks, compared to the eastern and southern Saharan
eco-regions. This highlights the strong impacts of climate and ecosystem characteristics in
determining SOC stocks at the regional level.

The analysis demonstrates variations in long-term average SOC stocks across different biomes.
The temperate conifer forest ecosystems show the highest SOC content per surface area, with a

181
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

median of 6.06 kg C m-2, followed by mountain grasslands and shrublands with 5.14 kg C m-2, and
Mediterranean forests woodlands & scrub areas with 5.31 kg C m-2, while the deserts and xeric
shrublands had the lowest SOC concentrations (4.76 kg C m-2).

Table 1. Climate, biomass and topographic variables that were used to model SOC stock in
Morocco.

Category Spatial resolution (meter) Predictors

Climate 4638 Mean annual precipitation
Annual climate water deficit
Palmer drought severity index
Minimal temperature
Maximal temperature
Biomass 5566 FAPAR Mean
LAI Mean
500 Net primary productivity
30 Red band reflectance
Green band reflectance
Blue band reflectance
Shortwave infrared band 1 reflectance
Shortwave infrared band 2 reflectance
NDVI median
NDVI variance
NDVI 10th percentile
NDVI 90th percentile
Topographic 30 Elevation
Slope
270 Topographic diversity index

Using climate and remote sensing time series data allowed to derive historic estimates of the
spatiotemporal dynamic changes of C stocks in Morocco. Model outputs indicated a 0.08% net
increase in SOC stocks over 32 years (1990-2022). Whilst the findings suggest an overall slim
increase in SOC stocks, the strongest changes were observed in the Mediterranean Forest,
Woodlands and Scrub biome. Losses were observed in large parts of the Mediterranean Acacia-
Argania dry woodlands and succulent thickets and Mediterranean woodlands and forests
ecoregions (Fig. 5 and Fig. S3). While Desert and Xeric Shrublands experienced the smallest
dynamics in SOC stocks, temperate conifer forest and montane grasslands and shrublands biomes
showed the most important net increase in Morocco, indicating sequestration of 1.2x10-2 kg C m-2
and 1.1x10-2 kg C m-2, respectively.

182
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 2. Random Forest SOC stocks model analytical metrics including R2 and RMSE (kg C m-2)
and number of observations (n) used in the training and validation. The first model uses the
predictors described in Table 1 (except green and blue bands) aggregated from 2017-2018. The
second model uses the same predictors as the first model from the 2014-2018 period but excluded
net primary productivity.

Model Year n RMSE R2 n RMSE R2

1 2017-2018 4473 1.086 0.734 2025 1.395 0.491
2 2014-2018 4478 1.035 0.729 2025 1.393 0.493

Figure 4. Long-term (1990-2022) average SOC stocks (kg C m-2) map. Insert plot represent SOC
stocks frequency distribution over biomes in Morocco.

183
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 5. Long-term (1990-2022) average SOC changes (%) map. Insert panel shows SOC
change frequency distribution in different biomes.

Model performance uncertainty

The variables importance values in the random forest prediction model showed that the drought
severity index was the most influential in determining SOC, followed by temperature (Min, Max),
elevation and primary productivity (Fig. 6 and Fig. S1). Vegetation dynamics captured by high-
resolution NDVI contributed equally to the prediction compared to LAI and precipitation proxies
(Fig. 6). The variables importance changed slightly when option for two instead of four-year
aggregation period of the predictors (Fig. S1).

184
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 6. Random forest model variables importance (%) from the first model, derived from the
sum of decrease in Gini impurity index.

SOC prediction performance was evaluated using the RMSE and R2 of the random forest model.
The model validation had an R2 of 0.49 and RMSE 1.39 kg C m-2. The model showed an R2 of
0.73 and RMSE 1.08 kg C m-2 with the training set (Table 2, Fig. 7). Long term change estimates
were not validated because of the lack of repeated records in the same sampling locations.
However, soil C stock times series estimates were compared with the measured values over 8 years
(Fig. 8).

Although the model uncertainty was highly variable over space and time, in general, the error
values showed unimodal distribution. When validated against measures from different years, the
random forest model had a median absolute error 0.13 kg C m-2 (Q1=-1.32; Q2= 1.47). Model error
was unevenly distributed over space, with the highest inaccuracy recorded in the montane
grasslands and shrublands. The model underestimated SOC stocks in these biomes, which include
the High Atlas and Mediterranean dry woodland and steppe eco-regions (Fig. 8). The model
uncertainty varied between years, with the highest inaccuracies recorded in 2012, 2014, and 2020
(0.99, 0.95, and 0.88 kg C m-2, respectively) (Fig. 8 and Fig. S4). The lowest uncertainties were
observed in 2013, 2017 and 2018, where median absolute error values were lower than 0.17 kg C
m-2.

185
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 7. Scatter plots of predicted versus measured SOC stocks (kg C m-2) with (a) training and
(b) test datasets plot). Predictors include all variables shown in Fig. 6 aggregate from 2017-2018
time period.

Table 3. Long-term (32 year) mean and median SOC stocks and net change estimations by
biomes in Morocco as estimated from the random forest model.

Biome Mean SOC Median SOC SOC stock (Pg Net change (%) Net change (kg
stocks stocks C) C m-2)
(kg C m-2) (kg C m-2)
Deserts & 4.77 4.76 1.91 0.012 5.7 10-4
Xeric
Shrublands
Temperate 6.07 6.06 0.08 0.196 1.18 10-2
Conifer Forest
Montane 5.14 5.14 0.04 0.215 1.1 10-2
Grasslands &
Shrublands
Mediterranean 5.19 5.17 1.91 0.056 2.9 10-3
Forests
Woodlands &
Scrub
Total 5.14 5.12 3.94 0.079 4 10-3

186
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 8. Spatial distribution map of the Random Forest error calculated as difference between
predicted and measured SOC stocks (kg C m-2).

187
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 4. Long-term (32 year) mean and median SOC stocks estimations by eco-regions in
Morocco as estimated from the random forest model.

Eco-region name Biome name Mean SOC (kg C Median SOC (kg SOC stock (Pg
m-2) C m-2) C)

Saharan Atlantic Deserts & Xeric 4.66 4.65 9.8 1013

coastal desert Shrublands
South Sahara Deserts & Xeric 4.82 4.82 5 1014
Desert Shrublands
Mediterranean Temperate 6.07 6.06 7.9 1013
conifer and Conifer Forests
mixed forests
Mediterranean Montane 5.13 5.14 3.8 1013
High Atlas Grasslands &
juniper steppe Shrublands
Mediterranean Mediterranean 5.19 5.09 5.7 1014
Acacia-Argania Forests,
dry woodlands Woodlands &
and succulent Scrub
thickets

DISCUSSION

Using climate and remote sensing predictors we estimated the soil C stock dynamics over a 32-
year period in Morocco. Our mapping method uses a large soil database, 30 m resolution satellite
data, climate and morphological data. This allowed the i) assessment of spatial dynamics of soil
carbon from paddock to national scale, ii) estimation of the magnitude of topsoil stocks in Morocco
at 0-30 cm, and iii) to step back on time to get historical estimates of soil C and thus, assess long-
term C changes. Accessing this amount of detail is impossible with lower resolution maps, due to
the substantial variability of soil properties from farm to pedon scale. While the SOC long term
changes map was based solely on the spatial changes of the environmental proxies, it informs on
change drivers, as well as potential increases in certain areas, as influenced by their inherent climate
and edaphic features. This will give land managers a useful tool to detect and reverse losses using
appropriate actions.

Previous attempts to estimate soil C magnitudes in different areas of Africa showed inaccuracies
due to the low resolution of the maps employed and insufficient soil data. For example, the
estimation of SOC stocks by Henry et al., 2009, who used DSMW, ISRIC and ETOPOS maps at a
1:5M scale, showed estimates of 2.87 and 2.23 Pg C (0-30cm) for South Africa and Morocco,
respectively. However, Venter et al., 2001 quantified 5.59 Pg C in South Africa, using a large
national soil database and a high-resolution mapping approach. The magnitude of our estimation
of topsoil C stocks was at 3.94 Pg C in Morocco (711,000 km2), which seems in accordance with
other studies in the Mediterranean region–e.g., Spain (505,000 km2), which is predominantly
Mediterranean Forests, Woodlands, and Scrub, has a SOC stock of 3.3 Pg C (Calvo de Anta et al.
2020). These high variabilities and inaccuracies in regional studies, show the inadequacy of small-
scale maps in quantifying the spatially highly dynamic soil carbon.

188
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

The level of accuracy of our model (RMSE = 1.36 kg C m-2) is higher than some regional maps
and national studies. Examples include subtropical maps such as South Africa (RMSE = 2.45 kg C
m-2, Venter et al.2021). Even when the model was tested against independent past data points, it
still yielded low uncertainty for most past years, except for 2012, 2014, and 2020, when absolute
errors were 0.99, 0.95, and 0.88 kg C m-2, respectively (Fig. 8). Coupled with high-resolution
multispectral data from Landsat, the large number of samples representing different ecosystems in
Morocco clearly strengthened the accuracy of the model.

SOC stocks by biomes and ecozones

Sabir et al. (2020) attempted to quantify soil C in different agro-ecozones in Morocco and estimated
high SOC levels in the Middle Atlas (mean of 4.55, min. 0.15, max. 9.81 kg C m-2), followed by
the Rif zone (mean of 4.13, min. 0.3, max. 7.71) that intersects with both Mediterranean conifer
and mixed forest and Mediterranean woodlands and forest eco-regions. The authors estimated the
lowest soil C concentrations in the Acacia-Argania dry woodlands (mean 2.14, min. 0.7, max. 3.93
kg C m-2), and the Sahara Desert (mean of 0.18, min 0.15, max. 0.24 kg C m-2). Boulmane et al.,
(2010) reported SOC stocks of 5.6-8 kg C m-2 in the forest green belt in the Middle Atlas
Mountains. Sabir et al. 2004 also reported high C (10.5 kg C m-2) values in the Quercus suber L.
forest in northern Morocco. These values agree with the high median (6 kg C m-2) found in the
present study, in the Mediterranean conifer and mixed forests.

Changes in soil carbon stocks

It has long been appreciated that management and land-cover changes can alter the amount of
organic carbon sequestered in the soil (Laganière et al., 2010), which subsequently affects both soil
quality and atmospheric CO2 fluxes (Powers et al., 2011). Land use change can cause a change in
surface biomass and an associated disturbance in soil C stocks. Ecosystem changes can occur
naturally or be the result of anthropogenic pressures. Each ecosystem has a potential carbon-
carrying capacity and an equilibrium carbon status defined by inherent climate and edaphic
characteristics. The soil carbon cycle is disturbed by land use changes until a new equilibrium is
eventually attained in the ecosystem. Throughout this procedure, alterations in soil C stocks might
have occurred, either as a sink or as a source of carbon. We estimated a 32-years average changes
in soil carbon stocks at 0.08% (3.11 Mt C) in Moroccan topsoil. This average annual gain represents
only 15.4% of the anthropogenic carbon emissions reported at 0.02 Pg C in 2021 in Morocco
(Crippa et al., 2022). Nevertheless, the current annual anthropogenic carbon emissions represent
only 5 per-mille (‰) of the topsoil stocks, indicating the substantial potential of the soil organic C
pool to offset CO2 emissions in Morocco. Over the 32-year period studied; Moroccan biomes
constituted a net carbon sink. Soil carbon change magnitude at the regional scale is limited by
climate and edaphic criteria. However, a net carbon sink is observed currently in the terrestrial
biosphere of the northern hemisphere. De Vries et al., (2006) reported that for European forests,
net carbon capture is in the range of 100 to 150 Mt C yr-1. Similarly, Heath et al., (1993) suggested
that temperate forest produces a net sink of 205 Mt C yr-1. Our estimation shows the highest soil
carbon increase in the Mediterranean conifer and mixed forests in the north and Mediterranean
High Atlas juniper steppe (including Anti-Atlas) eco-regions. The highest losses we estimated were
in the Acacia-Argania dry biosphere, the Gharb Forest in the northwest of Morocco and large parts
of the Prerif Mediterranean woodland and forest. Losses in the Gharb Forest were estimated at 21%
(3000 ha) in the last 20 years (Hansen et al., 2013). The Prerif areas have lost up to 8% of the forest
in the last 20 years. Net losses observed in the Prerif areas (Taounate region) are likely related to
climate and anthropogenic pressures. In the first decade of the century, more than half a million
189
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

olive and carob trees were planted in this area. Still, these efforts are still not enough to reverse
carbon losses.

For the Acacia-Argania woodlands, le Polain de Waroux et al., (2012) reported a net decrease of
tree density of 44.5% between 1970 and 2007. Consequently, this area will continue to act as a
carbon source until a new equilibrium is reached. Although this endemic species is well adapted to
the Mediterranean dry climate in Morocco, anthropogenic pressure presented by overgrazing and
use as fire fuel are the main causes of this decline (Le Potain de Waroux et al., 383 2012). Croplands
around the world are losing massive soil carbon stocks depending on their initial state and a high-
loss area was the cropland in the coastal plain in the Settate region. Similar net carbon sink patterns
were also observed at the country scale (Janssens et al., 2005), where areas with a high prevalence
of cultivated land tended to be a carbon source, whilst forest and grassland-dominated areas acted
as net terrestrial carbon sinks (Janssen et al., 2005). In the future, the Northern Hemisphere will
maintain a role as a carbon sink, although the upward trends are likely to be decreased (Canadell
et al., 2007; Zaehle et al., 2007). Although our estimates of carbon changes are consistent with the
theoretical dynamics of Soil C, given the land use changes and anthropogenic pressures, future
work should validate the change trend map using repeated measurements from the current sampling
locations.

CONCLUSION

The present work provides the first high-resolution dynamic map of topsoil carbon in Morocco.
This national map provide accurate and valuable insights onto the soil carbon magnitudes in north
African Biomes and an estimate of the C stock changes in the last 32 years. The map could be used
as a soil carbon stock watch that will support CO2 mitigation actions. Generally, Moroccan biomes
are still acting as net carbon sink. However, high losses were estimated in ecological niches such
as the dry Acacia-Argania ecoregion, which undergoes relentless anthropogenic pressure. Using
this high-resolution map, different stakeholders should take an important leap forward in
identifying carbon source areas and target appropriate remedial actions, whilst understanding trade-
offs between ecosystems management, biodiversity, and soil carbon. The extensive database has
the potential for future applications, including the modelling of how climate changes affect carbon
sequestration in Morocco.

REFERENCES

Abatzoglou, J. T., Dobrowski, S. Z., Parks, S. A., & Hegewisch, K. C. (2018). TerraClimate, a
high-resolution global dataset of monthly climate and climatic water balance from 1958–
2015. Scientific Data, 5(1), Article 1. https://doi.org/10.1038/sdata.2017.191
Batjes, N. h. (1996). Total carbon and nitrogen in the soils of the world. European Journal of Soil
Science, 47(2), 151–163. https://doi.org/10.1111/j.1365-2389.1996.tb01386.x
Batjes, N. H. (1998). Mitigation of atmospheric CO2 concentrations by increased carbon
sequestration in the soil. Biology and Fertility of Soils, 27(3), 230–235.
https://doi.org/10.1007/s003740050425
Baudena, M., Santana, V. M., Baeza, M. J., Bautista, S., Eppinga, M. B., Hemerik, L., Garcia
Mayor, A., Rodriguez, F., Valdecantos, A., Vallejo, V. R., Vasques, A., & Rietkerk, M.
(2020). Increased aridity drives post-fire recovery of Mediterranean forests towards open
shrublands. New Phytologist, 225(4), 1500–1515. https://doi.org/10.1111/nph.16252
190
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Bellamy, P. H., Loveland, P. J., Bradley, R. I., Lark, R. M., & Kirk, G. J. D. (2005). Carbon losses
from all soils across England and Wales 1978–2003. Nature, 437(7056), Article 7056.
https://doi.org/10.1038/nature04038 453
Boulmane, M., Makhloufi, M., Bouillet, J.-P., Saint-André, L., Satrani, B., Halim, M., & Elantry-
Tazi, S. (2010). Estimation du stock de carbone organique dans la chênaie verte du Moyen
Atlas marocain. Acta Botanica Gallica, 157(3), 451–467.
https://doi.org/10.1080/12538078.2010.10516222
Dinerstein, E., Olson, D., Joshi, A., Vynne, C., Burgess, N. D., Wikramanayake, E., Hahn, N.,
Palminteri, S., Hedao, P., Noss, R., Hansen, M., Locke, H., Ellis, E. C., Jones, B., Barber, C.
V., Hayes, R., Kormos, C., Martin, V., Crist, E., … Saleem, M. (2017). An Ecoregion-Based
Approach to Protecting Half the Terrestrial Realm. BioScience, 67(6), 534-545.
Calvo de Anta, R., Luís, E., Febrero-Bande, M., Galiñanes, J., Macías, F., Ortíz, R., & Casás, F.
(2020). Soil organic carbon in peninsular Spain: Influence of environmental factors and
spatial distribution. Geoderma, 370, 114365.
https://doi.org/10.1016/j.geoderma.2020.114365
Canadell, J. G., Kirschbaum, M. U. F., Kurz, W. A., Sanz, M.-J., Schlamadinger, B., & Yamagata,
Y. (2007). Factoring out natural and indirect human effects on terrestrial carbon sources and
sinks. Environmental Science & Policy, 10(4), 370–384.
https://doi.org/10.1016/j.envsci.2007.01.009 469
Certini, G. (2005). Effects of fire on properties of forest soils: A review. Oecologia, 143(1), 1–10.
https://doi.org/10.1007/s00442-004-1788-8
Chenu, C., Le Bissonnais, Y., & Arrouays, D. (2000). Organic Matter Influence on Clay
Wettability and Soil Aggregate Stability. Soil Science Society of America Journal, 64(4),
1479–1486. https://doi.org/10.2136/sssaj2000.6441479x
Crippa, M., Guizzardi, D., Banja, M., Solazzo, E., Muntean, M., Schaaf, E., ... & Vignati, E. (2022).
CO2 emissions of all world countries. JRC Science for Policy Report, European Commission,
EUR, 31182. 477
Davidson, E. A., Savage, K. E., & Finzi, A. C. (2014). A big-microsite framework for soil carbon
modeling. Global Change Biology, 20(12), 3610–3620. https://doi.org/10.1111/gcb.12718
480
De Deyn, G. B., Cornelissen, J. H., & Bardgett, R. D. (2008). Plant functional traits and soil carbon
sequestration in contrasting biomes. Ecology letters, 11(5), 516-531.
De Vries, W., Reinds, G. J., Gundersen, P., & Sterba, H. (2006). The impact of nitrogen deposition
on carbon sequestration in European forests and forest soils. Global Change Biology, 12(7),
1151–1173. https://doi.org/10.1111/j.1365-485 2486.2006.01151.x
Deng, L., Zhu, G., Tang, Z., & Shangguan, Z. (2016). Global patterns of the effects of land-use
changes on soil carbon stocks. Global Ecology and Conservation, 5, 127–138.
https://doi.org/10.1016/j.gecco.2015.12.004
Dewitte, O., Jones, A., Spaargaren, O., Breuning-Madsen, H., Brossard, M., Dampha, A., Deckers,
J., Gallali, T., Hallett, S., Jones, R., Kilasara, M., Le Roux, P., Michéli, E., Montanarella, L.,
Thiombiano, L., Van Ranst, E., Yemefack, M., & Zougmore, R. (2013).
Harmonisation of the soil map of Africa at the continental scale. Geoderma, 211-212,
138-153. https://doi.org/10.1016/j.geoderma.2013.07.007
Farr, T. G., Rosen, P. A., Caro, E., Crippen, R., Duren, R., Hensley, S., Kobrick, M., Paller, M.,
Rodriguez, E., Roth, L., Seal, D., Shaffer, S., Shimada, J., Umland, J., Werner, M., Oskin,
M., Burbank, D., & Alsdorf, D. (2007). The Shuttle Radar Topography Mission. Reviews of
Geophysics, 45(2). https://doi.org/10.1029/2005RG000183
191
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Frank, S., Schmid, E., Havlík, P., Schneider, U. A., Böttcher, H., Balkovič, J., & Obersteiner, M.
(2015). The dynamic soil organic carbon mitigation potential of European cropland. Global
Environmental Change, 35, 269–278. https://doi.org/10.1016/j.gloenvcha.2015.08.004
Hansen, M. C., Potapov, P. V., Moore, R., Hancher, M., Turubanova, S. A., Tyukavina, A., ... &
Townshend, J. (2013). High-resolution global maps of 21st-century forest cover change.
science, 342(6160), 850-853.
Heath, L. S., Kauppi, P. E., Burschel, P., Gregor, H.-D., Guderian, R., Kohlmaier, G. H., Lorenz,
S., Overdieck, D., Scholz, F., Thomasius, H., & Weber, M. (1993). Contribution of temperate
forests to the world’s carbon budget. Water, Air, and Soil Pollution, 70(1), 55–69.
https://doi.org/10.1007/BF01104988
Houghton, R. A. (2007). Balancing the Global Carbon Budget. Annual Review of Earth and
Planetary Sciences, 35(1), 313–347.
https://doi.org/10.1146/annurev.earth.35.031306.140057
Janssens, I. A., Freibauer, A., Schlamadinger, B., Ceulemans, R., Ciais, P., Dolman, A. J.,
Heimann, M., Nabuurs, G.-J., Smith, P., Valentini, R., & Schulze, E.-D. (2005). The carbon
budget of terrestrial ecosystems at country-scale – a European case study. Biogeosciences,
2(1), 15–26. https://doi.org/10.5194/bg-2-15-2005
Jobbágy, E. G., & Jackson, R. B. (2000). The Vertical Distribution of Soil Organic Carbon and Its
Relation to Climate and Vegetation. Ecological Applications, 10(2), 423–436.
https://doi.org/10.1890/1051-0761(2000)010[0423:TVDOSO]2.0.CO;2
Johnson, D. W., & Curtis, P. S. (2001). Effects of forest management on soil C and N storage:
Meta analysis. Forest Ecology and Management, 140(2), 227–238.
https://doi.org/10.1016/S0378-1127(00)00282-6
Laganière, J., Angers, D. A., & Paré, D. (2010). Carbon accumulation in agricultural soils after
afforestation: A meta-analysis. Global Change Biology, 16(1), 439–453.
https://doi.org/10.1111/j.1365-2486.2009.01930.x
Lal, R. (2004). Soil carbon sequestration to mitigate climate change. Geoderma, 123(1), 1–22.
https://doi.org/10.1016/j.geoderma.2004.01.032
Lal, R. (2006). Enhancing crop yields in the developing countries through restoration of the soil
organic carbon pool in agricultural lands. Land Degradation & Development, 17(2), 197–
209. https://doi.org/10.1002/ldr.696
le Polain de Waroux, Y., & Lambin, E. F. (2012). Monitoring degradation in arid and semi-arid
forests and woodlands: The case of the argan woodlands (Morocco). Applied Geography,
32(2), 777–786. https://doi.org/10.1016/j.apgeog.2011.08.005
Minelli, S. (2018). Default data: Methods and interpretation - A guidance document for 2018
UNCCD. https://policycommons.net/artifacts/1983874/default-data/2735639/
Paustian, K., Lehmann, J., Ogle, S., Reay, D., Robertson, G. P., & Smith, P. (2016). Climate-smart
soils. Nature, (7597), Article 7597. https://doi.org/10.1038/nature17174
Poeplau, C., & Dechow, R. (2023). The legacy of one hundred years of climate change for organic
carbon stocks in global agricultural topsoils. Scientific Reports, 13(1), Article 1.
https://doi.org/10.1038/s41598-023-34753-0
Powers, J. S., Corre, M. D., Twine, T. E., & Veldkamp, E. (2011). Geographic bias of field
observations of soil carbon stocks with tropical land-use changes precludes spatial
extrapolation. Proceedings of the National Academy of Sciences, 108(15), 6318–6322.
https://doi.org/10.1073/pnas.1016774108

192
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Ruehlmann, J., & Körschens, M. (2009). Calculating the Effect of Soil Organic Matter
Concentration on Soil Bulk Density. Soil Science Society of America Journal, 73(3), 876–
885. https://doi.org/10.2136/sssaj2007.0149
Sabir, M., Sagno, R., Tchintchin, Q., Zaher, H., & Benjelloun, H. (2020). Chapitre 3. Évaluation
des stocks de carbone organique dans les sols au Maroc. In M. Brossard, L. Chapuis-Lardy,
T. Chevallier, & T. M. Razafimbelo (Eds.), Carbone des sols en Afrique: Impacts des usages
des sols et des pratiques agricoles (pp. 71–90). IRD Éditions.
https://doi.org/10.4000/books.irdeditions.34902
Santín, C., Doerr, S. H., Preston, C. M., & González-Rodríguez, G. (2015). Pyrogenic organic
matter production from wildfires: A missing sink in the global carbon cycle. Global Change
Biology, 21(4), 1621–1633. https://doi.org/10.1111/gcb.12800
Schlesinger, W. H. (1977). Carbon Balance in Terrestrial Detritus. Annual Review of Ecology and
Systematics, 8(1), 51–81. https://doi.org/10.1146/annurev.es.08.110177.000411
Somarathna, P. D. S. N., Malone, B. P., & Minasny, B. (2016). Mapping soil organic carbon content
over New South Wales, Australia using local regression kriging. Geoderma Regional, 7(1),
38–48. https://doi.org/10.1016/j.geodrs.2015.12.002
Theobald, D. M., Harrison-Atlas, D., Monahan, W. B., & Albano, C. M. (2015). Ecologically-
Relevant Maps of Landforms and Physiographic Diversity for Climate Adaptation Planning.
PLOS ONE, 10(12), e0143619. https://doi.org/10.1371/journal.pone.0143619
Turner, D. P., Ollinger, S. V., & Kimball, J. S. (2004). Integrating remote sensing and ecosystem
process models for landscape-to regional-scale analysis of the carbon cycle. BioScience,
54(6), 573–584.
Venter, Z. S., Hawkins, H.-J., Cramer, M. D., & Mills, A. J. (2021). Mapping soil organic carbon
stocks and trends with satellite-driven high-resolution maps over South Africa. Science of
The Total Environment, 771, 145384. https://doi.org/10.1016/j.scitotenv.2021.145384
Walker, T. W. N., Kaiser, C., Strasser, F., Herbold, C. W., Leblans, N. I. W., Woebken, D.,
Janssens, I. A., Sigurdsson, B. D., & Richter, A. (2018). Microbial temperature sensitivity
and biomass change explain soil carbon loss with warming. Nature Climate Change, 8(10),
Article 10. https://doi.org/10.1038/s41558-018-0259-x
Woodward, F. I., Lomas, M. R., & Kelly, C. K. (2004). Global climate and the distribution of plant
biomes. Philosophical Transactions of the Royal Society of London. Series B: Biological
Sciences, 359(1450), 1465-1476.
Zaehle, S., Bondeau, A., Carter, T. R., Cramer, W., Erhard, M., Prentice, I. C., Reginster, I.,
Rounsevell, M. D. A., Sitch, S., Smith, B., Smith, P. C., & Sykes, M. (2007). Projected
Changes in Terrestrial Carbon Storage in Europe under Climate and Land-use Change, 1990–
2100. Ecosystems, 10(3), 380–401. https://doi.org/10.1007/s10021-007-9028-582 9

193
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

EVALUATING SOIL PARAMETERS USING PROXIMAL SOIL SENSORS (EM38,

MSP3)
#11090
1,2*H.
Hajjaj, 1K. Ibno Namr, 2C. El khalif, 2A. El-Aissaoui
1Laboratory of Geosciences and Environmental Techniques, Department of Geology,
Faculty of Sciences, Chouaïb Doukkali University, BP.20, El Jadida 24000, Morocco
2Laboratory of Agricultural Engineering and Precision Agriculture, National Institute of

Agronomic Research, CRRAS, PO 586, Settat,26000, Morocco

e-mail: [email protected]

INTRODUCTION

Site-specific farming requires an accurate information on soil spatial variability. In fact, by using
digital mapping of physical, biological and chemical soil parameters, some indicators and
pedotransfert functions can be developed for evaluating not only soil quality but also for monitoring
performances of agricultural system, natural resource management, climate modelling and
environmental science (Liu et al.2006; Robinson & Metternicht, 2006; Bhunia et al. 2018). Thus,
adequate information on the status behaviour of soil parameters is required for spatiotemporal
monitoring and assessment. However, direct measurements are precise but expensive, time-
consuming and labour-intensive (Bhunia et al.2018). Otherwise, better planning and management
of soil data cannot be systematically effective using interpolation at unsampled sites.

According to the technological progress, today’s spatial data analysis methods and tools allow the
monitoring of spatiotemporal changes in almost all soil attributes at various levels (Mabit and
Bernard 2010; Dai et al. 2014; Bhunia et al. 2018). In fact, proximal soil sensing refers to the use
of sensors in the field to obtain signals from the soil when the sensor detector is either in contact
or close to soil matrix (Viscarra Rossel et al., 2011).

This paper aims to review the proximal soil sensing technologies mainly the MSP3 and EM38
ECa meter and to evaluate both technologies for their ability to predict soil parameters under arid
conditions

On the go soil sensor (MSP3)

On-the-go soil sensors such as MSP3 (Fig 1), equipped with GPS, is designed to effectively
delineate soil differences across fields, capturing data that other technologies might miss
(Adamchuk et al., 2004). It utilizes dual wavelength optical sensors to measure organic matter and
pH, along with coulter electrodes for direct electrical conductivity readings at depths of 0-12 inches
and 0-36 inches. As the implement is pulled behind a tractor at speeds up to 6 mph, data is logged
on GPS maps, which can inform fertility management decisions such as split fertilizer applications
and variable rate seeding. The system records digital reflectance data and GNSS coordinates at a
rate of 1 Hz, averaging 260 reflectance data points per hectare while ensuring effective sensing
through self-cleaning optical module (Kweon & Maxton, 2013; Bönecke et al., 2020).

194
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 9. a) MSP3 sensor, b) pH,OM and EC coulters c) MSP3 prepared to map in the field d) MSP3 in the
field (Veris Staff; Mackowiaket al., 2016).

Electromagnetic induction sensor (EM38 SENSOR)

Non-contact electrical conductivity sensors use electromagnetic induction and do not need to come
into contact with the ground. These sensors measure the change in mutual impedance between the
coils (McNeill, 1980). This measurement is then converted into an estimate of the EC known as
the apparent EC (ECa). These units are capable of monitoring at greater depths than contact sensors.

The EM38 sensor (Fig 2), designed for versatile use on the ground, in the air, and in boreholes, is
a hand-held instrument that can also be mounted and directed using GPS for automated data logging
in a Geographic Information System (McDaniel et al., 2018) The standard EM38-MK2 model
measures 1.05 m in length and weighs 3.5 kg, powered by a 9-volt battery with a life of up to 20
hours. It records soil electrical conductivity (ECₐ) in both horizontal (H-mode) and vertical (V-
mode) dipole modes, providing measurements at depths of 0.375 m and 0.75 m in H-mode, and
0.75 m and 1.5 m in V-mode. ((Kitchen et al., 2005; Geonics, 2012).

195
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 10. The EM38 sensor in horizontal (H) and vertical (V) mode with the effective depth
range responses of ECa for coil separation of 1 m and 0.5 m when placed on the soil surface.
(Petsetidi & Kargas, 2023).

EM38 AND MSP3 USE IN ARID LAND CONDITIONS

The MSP3 sensor has been used and proved its effectiveness in analysing soil properties in arid
and semi-arid regions. Several studies (Kweon, 2012a, 2012b; Mackowiak et al., 2016; Novais et
al., 2019) showed that this sensor could be effective to distinguish soil textures and organic matter
variability, with moderate to strong correlations compared to laboratory results. Its potential use
for precise soil management in challenging environments has been emphasized, particularly for
detecting small-scale spatial variability and controlling variable rate inputs management with
reference to established prescription maps.

Similarly, EM38 sensor has been used in arid lands to produce valuable data for assessing soil
properties, particularly in understanding soil electrical conductivity (ECa) and its behaviour and
relationship with the soil parameters according to the occurring soil water content. Several Studies
(Rhoades et al., 1976,1989,1990,1997,1999; Corwin & Lesch, 2005b, 2013, 2017; Molinl et
al,2005) indicated that ECa measurements could effectively differentiate between dry and saturated
soils, which is crucial to approach soil parameters in arid regions having irregular soil water content
due to intermittent rainfall.

196
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 6. Some studies based on using MSP3 sensor in arid land context.

Study Objective Location Key Findings Recommendations

Novais et al. Calibrate and Guanacaste, Correlations established Alternative calibration methods
(2019) validate the MSP3 Costa Rica for soil texture and needed for deeper
in rainy conditions organic matter (R² ≥ 0.55) measurements (30-90 cm).
at 0-30 cm; varied
textures observed.
Mackowiak Improve soil Florida USA Organic matter (OM) Use MSP3 to delineate
et al. (2016) mapping for values help distinguish differences for variable rate
nutrient and water soil texture differences; management rather than
management using dark colors correlated achieving specific OM or EC
MSP3 with higher OM. targets.
Kweon Evaluate MSP3 Midwest Proximal sensor Multi-sensor approach
(2012a, performance USA measurements correlated increases chances of identifying
2012b) across multiple well with lab samples; soil property variations;
fields in the detected small-scale improve CEC predictions
Midwest USA variability not seen in through data integration.
traditional methods.

Table 7. Studies based on using EM38 in arid land context.

Years Objective Study Area Findings Recommendations

(Rhoades et Obtain empirical California, The authors stated that Predictions were found
al., 1976; coefficients used in USA electromagnetic to be more accurate
1989a; equations to predict measurements on soils with using the new
1990a; 1997; ECa by depth intervals less than 10% water by coefficients rather than
1999a; within the soil profile weight are not a reliable those previously
1999b) from EM readings taken indication of salinity, and available.
above ground. for very sandy soils, the
limiting value of moisture
content is probably higher.
Proximity to the water table
also
influences EM38 readings.
(Corwin & Applications of Arizona, It appeared to be a stronger Evaluates site-specific
Lesch, 2005a, ECa measurements in USA than normal water content management from a
2005b, 2013, agriculture, particularly influence on the EM38 holistic perspective of
2017) site-specific crop signal data, consistent with environmental, crop
management typical surveying conditions productivity, and
encountered because of the economic impacts.
prevalence of lighter
textured soils.
Molinl et al to perform spatial Brazil demonstrated the potential ECa is a qualitative
2005 monitoring of soil of ECa sensors for indicator in areas with
moisture in two different understanding soil high spatial variability in
experimental fields over characteristics and their soil texture. In the field,
two consecutive years impact on crop yields, where soil moisture
and evaluate the particularly in no-till fields. range was lower, ECa
influence of moisture on was not associated to
soil ECa. moisture levels.

197
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

CONCLUSION AND RECOMMENDATIONS

This paper outlines a comprehensive approach to assessing soil parameters using two proximal soil
sensor tools, mainly the MSP3 and EM38 tools.

This review showed several studies highlighting importance of using the Veris MSP3 mapping tool
in improving land management decisions. The MSP3’s electrical conductivity (EC) package could
effectively track nutrient variations, while its pH tool can show promise, though further calibration
is required to confirm its accuracy. Challenges such as technical glitches with inexperienced
operators and the underdevelopment of the organic matter (OM) tool emphasize the need for
enhanced data support and interface improvements to facilitate research.

Additionally, while the EM38 performs well under conventional irrigation systems, it faces
challenges with micro-irrigation due to localized salinity variations, highlighting the need for
updated research protocols. The growing adoption of inverse modelling for soil salinity profiling
is crucial for managing water resources in cash crops, especially in the context of climate change.
(Corwin & Scudiero, 2016, 2019)

Ultimately, while the MSP3 excels in data integration and the creation of management zones, the
EM38 remains valuable for rapid field assessments. The combined use of both tools could
significantly enhance soil management practices.

REFERENCES

Bhunia, G.S., Shit, P.K., Maiti, R. 2018. Comparison of GIS-based interpolation methods for
spatialdistribution of soil organic carbon (SOC). Journal of the Saudi Society of
Agricultural Sciences, 17(2),114–126. DOI: 10.1016/j.jssas.2016.02.001
(PDF) Modelling and Mapping of Soil Organic Matter in Doukkala Plain Morrocan Semi-
Arid Region.
Bönecke, E., Kling, C., Kramer, E., Lück, K., Philipp, G., Rühlmann, J., Schröter, I., & Gebbers,
R. (2022). Direct prediction of site-specific lime requirement of arable fields using the base
neutralizing capacity and a multi-sensor platform for on-the-go soil mapping. Precision
Agriculture, 23(1), 127–149. https://doi.org/10.1007/s11119-021-09830-x
Corwin, D. L., & Lesch, S. M. (2005a). Apparent soil electrical conductivity measurements in
agriculture. Computers and Electronics in Agriculture, 46, 11–43.
https://doi.org/10.1016/j.compag.2004.10.005
Corwin, D. L., & Lesch, S. M. (2013). Protocols and Guidelines for Field-scale Measurement of
Soil Salinity Distribution with EC a -Directed Soil Sampling . Journal of Environmental
and Engineering Geophysics, 18(1), 1–25. https://doi.org/10.2113/jeeg18.1.1
Corwin, D. L., & Lesch, S. M. (2017). Validation of the ANOCOVA model for regional-scale
ECa to ECe calibration. Soil Use and Management, 33(2), 178–190.
https://doi.org/10.1111/sum.12262
Corwin, D. L., & Scudiero, E. (2016). Field-Scale Apparent Soil Electrical Conductivity.
Methods of Soil Analysis, 1(1), 0. https://doi.org/10.2136/methods-soil.2015.0038
Corwin, D. L., & Scudiero, E. (2019). Review of soil salinity assessment for agriculture across
multiple scales using proximal and/or remote sensors. In Advances in Agronomy (Vol. 158,
pp. 1–130). Academic Press Inc. https://doi.org/10.1016/bs.agron.2019.07.001
198
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Kweon, G. (2012a). Delineation of site-specific productivity zones using soil properties and
topographic attributes with a fuzzy logic system. Biosystems Engineering, 112(4), 261–
277. https://doi.org/10.1016/j.biosystemseng.2012.04.009
Kweon, G., Lund, E., & Maxton, C. (2013). Soil organic matter and cation-exchange capacity
sensing with on-the-go electrical conductivity and optical sensors. Geoderma, 199, 80–89.
https://doi.org/10.1016/j.geoderma.2012.11.001
Kitchen, N. R., Sudduth, K. A., Myers, D. B., Drummond, S. T., & Hong, S. Y. (2005).
Delineating productivity zones on claypan soil fields using apparent soil electrical
conductivity. Computers and Electronics in Agriculture, 46, 285–308.
https://doi.org/10.1016/j.compag.2004.11.012
Mackowiak, C., Pi, L., & Wright, D. (2016). Soil Mapping Enhancements via Veris MSP3.
McDaniel, P., Ross, M., Jimenez, J., Strawn, D. G., Valerio, M., Kimsey, M. J., Campbell, S., &
Falen, A. (2018). Pedogenic Pathways in Andic Soils of the Northern Rocky Mountains
(USA). Soil Science Society of America Journal, 82(6), 1308–1318.
https://doi.org/10.2136/SSSAJ2018.05.0186
McNeill, J. D. (1980). Electromagnetic Terrain Conductivity Measurement at Low Induction
Numbers. In Technical note TN (Vol. 6, Issue TN--6, p. 13).
http://www.geonics.com/pdfs/technicalnotes/tn6.pdf
Novais, W., Rodríguez-Mejías, J. C., Perret, J., Soto, C., Villalobos, J. E., Fuentes, C. L., &
Abdalla, K. (2019). Calibration and validation of Veris MSP3 on two soils in Guanacaste,
Costa Rica. Agronomia Mesoamericana, 30(2), 535–551.
https://doi.org/10.15517/am.v30i2.33579
Petsetidi, P. A., & Kargas, G. (2023). Assessment and Mapping of Soil Salinity Using the EM38
and EM38MK2 Sensors: A Focus on the Modeling Approaches. In Land (Vol. 12, Issue
10). Multidisciplinary Digital Publishing Institute (MDPI).
https://doi.org/10.3390/land12101932
Rhoades, C. C., Brosi, S. L., Dattilo, A. J., & Vincelli, P. (n.d.). Effect of soil compaction and
moisture on incidence of phytophthora root rot on American chestnut (Castanea dentata)
seedlings. https://doi.org/10.1016/S0378-1127(03)00147-6
Robinson, T. P., & Metternicht, G. (2006). Testing the performance of spatial interpolation
techniques for mapping soil properties. Computers and Electronics in Agriculture, 50(2),
97–108. https://doi.org/10.1016/j.compag.2005.07.003
Veris Technologies Inc. (2016). Quantifying the Value of Precise Soil Mapping. 1–8.
Viscarra Rossel, R. A., A., K., & Schulze, P. (2011). Sensor Fusion for Precision Agriculture.
Sensor Fusion - Foundation and Applications, May 2014. https://doi.org/10.5772/19983

199
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

SMALL HOLDERS AND PRECISION AGRICULTURE

200
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

THE USE OF UNMANNED AERIAL VEHICLE (UAV) REMOTELY SENSED DATA

AND BIOPHYSICAL VARIABLES TO PREDICT MAIZE ABOVE-GROUND BIOMASS
(AGB) IN SMALL-SCALE FARMING SYSTEMS
#11068

1*Celuxolo M Dlamini, 1John Odindi, 1Onisimo Mutanga, 2Trylee N Matongera

1Discipline of Geography and Environmental Science, School of Agricultural Earth and
Environmental Sciences, University of KwaZulu-Natal, Scottsville, Pietermaritzburg 3209, South
Africa; 2Centre for Transformative Agricultural and Food Systems, School of Agricultural, Earth
and Environmental Sciences, University of KwaZulu-Natal, Scottsville, Pietermaritzburg 3209,
South Africa
*e-mail: [email protected]/[email protected]; +27780632013

ABSTRACT

Considering the current and projected increase in human population, approaches to optimize crop
productivity to meet the rising demand are paramount. Timely and accurate maize Above Ground
Biomass (AGB) measurements allow for development of models that can precisely predict yield
prior to harvesting, useful for food production management and sustenance. The development of
Unmanned Aerial Vehicles (UAVs) as a new generation of robust remote sensing platforms,
mounted with high-resolution sensors has allowed timely and accurate prediction of maize AGB
in pursuit of sustaining food security. This study aimed to predict maize crop AGB in small-scale
farming systems using UAV-remotely sensed data and landscape biophysical variables. The DJI
Matrice 300 UAV mounted with a MicaSense multispectral camera was used to acquire high-
resolution images at four phenological stages that covered the vegetative (V8 &V12) and
reproductive stages (R2 & R5). Furthermore, in-situ plant biophysical measurements and landscape
variables were acquired and combined with UAV-remote sensing derived vegetation indices to
model maize AGB using a Deep Neural Network (DNN) model. Results showed that the V12
phenological stage yielded a better overall prediction accuracy (R2 = 0.74) than the V8 (R2 = 0.65),
R2 (R2=0.71), and R5 (R2=0.67) phenological stages. The study concludes that the V12 and R2
phenological stages are optimum for estimating maize AGB. This study contributes to a better
understanding of maize crop health and crop monitoring efforts for improved food security.

Keywords: UAV-Remote Sensing; Above-Ground Biomass; Maize; Smallholder Farming; Deep

Neural Networks.

INTRODUCTION

Small-scale crop farming plays a critical role in the economies of developing countries and is
crucial for sustaining food security. However, productivity in smallholdings is often adversely
affected by unfavourable bioclimatic conditions, climate change, and lack of farming resources
(Mgbenka et al., 2016). Maize (Zea mays) is ranked as one of the most extensively cultivated crops
worldwide. In South Africa, maize is widely produced and consumed as a staple food by the
majority population and also used for livestock fodder (Luo et al., 2019; Ngoune Tandzi &
Mutengwa, 2019). Other uses of maize include the production of starch, ethanol, and fuels
(Mgbenka et al., 2016). Although the demand for maize has significantly increased in South Africa,

201
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

challenges related to production and yield remain prevalent (Haarhoff et al., 2020; Verschuur et
al., 2021). Hence, it is imperative to adopt prompt and robust techniques such as crop yield
prediction to accurately counteract these challenges.

Maize Above Ground Biomass (AGB) is an essential basis for crop yield formation as it indicates
plant growth and productivity (Meiyan et al., 2022; Tang et al., 2023). A higher maize AGB
signifies a superior crop performance in capturing and converting sunlight, nutrients, and water
into energy for grain development and increased yield (Luo et al., 2019). A direct positive
correlation between maize AGB and yield is well established in literature (Leroux et al., 2019;
Tollenaar & Lee, 2002; Zhang et al., 2021). Hence, timely and accurate maize AGB measurements
allow for development of models that can precisely predict yield prior to harvesting, useful for
strategic evaluations, financial planning, efficient irrigation, and food production management
(Yahui Guo et al., 2020). Furthermore, maize AGB serves as a crucial source of nutrition for
livestock during periods of limited forage availability, such as the dry season (Palacios‐Rojas et
al., 2020). Therefore, the assessment of maize AGB to optimise yield, particularly in small-scale
farming systems, is essential for optimising productivity and mitigating potential losses (Cheng et
al., 2020).

Traditionally, quantifying maize AGB involves in-situ measurements of foliar weight, which is
destructive and laborious, hence unsuitable for large spatial extents and repeated observations
(Gerke, 2019; Han et al., 2019b). Recently, satellite remote sensing has been widely adopted to
accurately monitor agricultural crops, with many studies showing a positive correlation between
remotely sensed variables and AGB (Battude et al., 2016; Kayad et al., 2019; Leroux et al., 2019).
For instance, Geng et al. (2021) estimated maize AGB using Moderate Resolution Imaging
Spectroradiometer (MODIS) reflectance data and machine learning, achieving a coefficient of
determination of 0.77 (R2 = 0.77). However, despite these successes, the application of satellite
remote sensing is limited by as among others cloud cover, which significantly restricts maize crop
monitoring requirements for small-scale farming systems (Zhang et al., 2021). Furthermore, small-
scale farming systems are characterized by small spatial extents of less than two hectares, hence
higher spatial resolution sensors are necessary for effective capture of crops spectral information
(Peter et al., 2020). In addition, the transition between phenological stages in maize crops occurs
rapidly, necessitating the use of high-temporal-resolution sensors and on demand dataset to
accurately monitor and capture the changes in AGB at each growth stage (B. Yang et al., 2022).

Recently, Unmanned Aerial Vehicles (UAVs), also known as drones, have demonstrated a
remarkable capability to bridge the gap between satellite remote sensing and ground-based
observations (Gargiulo et al., 2023). This is attributed to their ability to provide cloud-free, near-
real-time data at ultra-high spatial resolution (Z. Li et al., 2022; Sharma et al., 2022). UAVs offer
several benefits for agricultural crop monitoring that include the ability to hover over areas of
interest and fly beneath cloud cover at flexible altitudes, allowing for high resolution imagery and
precise monitoring of individual crops (Aasen et al., 2018). Additionally, their flexible flight
mission make them ideal for capturing data during optimal periods, such as the short-window peak
photosynthetic phase in maize crops (B. Yang et al., 2022). However, despite these advancements
and capabilities, studies on the use of UAV technology on small holder farms, particularly in the
global south, remain scarce. This underscores the need for studies that investigate the potential of
UAVs, equipped with high resolution sensors, in predicting maize AGB in small-scale farming
systems.
202
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

High resolution sensors mounted onto UAV platforms cover a wide range of the electromagnetic
bands including the visible, near-infrared, and red-edge sections that are useful in predicting maize
AGB and deriving vegetation indices to support yield estimations (Li et al., 2016). For instance,
vegetation indices derived from the near-infrared and red-edge wavelengths such as the Normalized
Difference Vegetation Index (NDVI), have demonstrated the ability to detect subtle changes in
crops properties such as canopy structure, photosynthetic activity, and crop health (Che et al., 2022;
Vélez et al., 2023). For example, Brewer et al. (2022) obtained satisfactory results by using various
multispectral derived vegetation indices such as NDVI and Soil Adjusted Vegetation Index (SAVI)
for estimating leaf chlorophyll content to determine crop health and vigour.

Typically, maize crops are characterised by variable stock height, density, and greenness, while
canopy vegetation index remains unchanged (Adewopo et al., 2020). Hence, vegetation index-
based empirical approaches alone cannot accurately estimate maize AGB. Consequently, to
account for these variations, biophysical variables such as leaf chlorophyll content and leaf area
index (LAI) can be combined with vegetation indices to accurately predict maize AGB (Meiyan et
al., 2022). Leaf chlorophyll content and LAI have been identified as strong crop health indicators
that positively correlate with maize AGB (Che et al., 2022; Liu et al., 2019; Luo et al., 2019).
However, measuring the aforementioned biophysical variables is only ideal for small spatial
extents (Liu et al., 2023). In addition, considering that most small-scale farmlands are often
characterized by challenging terrain featuring steep topography, it is essential to assess the
influence of landscape variability on maize AGB (Polzin & Hughes, 2023). Therefore, landscape
and landscape related variables that directly and indirectly influence crop growth such as soil
moisture, slope, aspect, and elevation can provide a precise maize AGB estimation (Fry & Guber,
2020; Goldenberg et al., 2022; Svedin et al., 2021). Consequently, integrating drone-derived
multispectral bands, with optimal vegetation indices, and biophysical landscape variables can
provide better and precise estimates of maize AGB in small-scale farming systems.

Numerous regression techniques have been proposed in literature for the prediction of crop
properties (Ali et al., 2022; Khan et al., 2022; Tripathi et al., 2022). Machine learning algorithms,
combined with spectral variables from remote sensing datasets have proven superior for data
analysis than other statistical approaches (Altaweel et al., 2022). Deep learning algorithms, such as
Deep Neural Networks (DNN), have particularly gained popularity over the past decades for their
ability to learn and discover patterns from large and complex datasets and generate accurate
predictions (X. Li et al., 2022; Muruganantham et al., 2022). DNN comprises a hierarchy of more
than two hidden neural network layers and are subsequently called ‘deep learning’ (Odebiri et al.,
2021a, 2021b). The primary limitation of this technique is its propensity to overfitting and
requirement of large datasets for optimal performance (Cao et al., 2022). However, features such
as regularization and dropout in neural networks can counteract these effects (Vojnov et al., 2022).
Numerous studies have successfully adopted DNN to predict maize agronomic variables and
obtained results surpassing other machine learning algorithms (Khaki & Wang, 2019; Lischeid et
al., 2022). Despite its potential, deep learning is the least used approach in agricultural monitoring
applications, particularly at small-scale extents due to small acquirable datasets. Therefore, there
is need for further research to explore the full potential of UAV remotely sensed data combined
with landscape and biophysical variables for estimating and mapping maize crop AGB using DNN
machine learning techniques.

203
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Studies have employed either plant biophysical, landscape variables or remotely sensed data to
estimate maize AGB (Liu et al., 2019; Luo et al., 2019; Meiyan et al., 2022). Generally, studies
have seldom integrated the two, with the landscape variables for precision agriculture. Therefore,
this study sought to evaluate the utility of UAV remotely sensed data combined with landscape and
biophysical variables in estimating maize AGB in small-scale farming systems using DNN
machine learning techniques. The main objective of this study was to predict maize AGB using a
combination of UAV remotely sensed data, landscape variables, and plant biophysical variables.
Additionally, this study sought to determine the optimum phenological stage for timely and
efficient maize AGB prediction in subsequent seasons. Finally, the study sought to assess the
performance of DNN algorithm to identify an optimal model for predicting maize AGB using small
spatial extent acquired dataset.

MATERIALS AND METHODS

Description of the study area

This study was conducted in Swayimane communal area (Latitude: -29.524444°, Longitude:
30.699846°) within the UMshwathi Municipality, in the KwaZulu-Natal province, South Africa
(Fig.1). The experimental field is approximately 1.4 hectares and exhibits distinct variations in
slope, aspect, and elevation. Average air temperature is 17 ℃, while the minimum and maximum
temperatures are 11.8 ℃ and 24 ℃, respectively (Ndlovu et al., 2021) . Annual rainfall ranges from
600 mm to 1100 mm, with most rainfall received in summer. Swayimane is characterized by wet-
hot summers and dry-cold winters. Since cropping activities in the study area are rain fed, crops
are grown during the summer season. The area has excellent bioclimatic and physical conditions,
that include loam soils with efficient nutrient and water holding capacity as well as optimum terrain
for efficient sunlight capture, making it suitable for crop farming. Farmers in the area mainly rely
on traditional farming methods such as use of kraal manure as fertilizers and animal draft
implements for ploughing and weeding. However, with recent socio-economic improvements in
the area, some farmers are adopting artificial fertilizers and mechanized farming, particularly in
larger fields. In addition to maize, legumes, sweet potatoes, taro and small holding sugarcane are
grown in the study area.

204
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 1. Location of the study site.

Maize phenotyping
The maize field was planted with the SC 701 hybrid (Pannar Seed Company, SA) on the 24th of
February 2023, and harvested on the 7th of July 2023. The SC 701 seed type was chosen because
of its high yield capacity estimated at more than 13 tons per hectare according to the seed producers.
The SC 701 hybrid is late maturing (140-148 days) and known to be heat and drought tolerant.
However, in such cases, irrigation is recommended for maximum yield. The maize was rain-fed
throughout the growing season, and no drought and extreme temperatures were recorded. The
maize was planted in rows perpendicular to the slope to minimize nutrient runoff and soil erosion
during rainfall. The distance between the crops and rows was at least 20 cm and 70 cm,
respectively, to avoid inter-competition within the crops and stunted growth. To eradicate weeds,
an affordable water-soluble Basagran herbicide with a mixability of 480 g/l was applied when the
maize was 30 days old, and a nitrogen-phosphorus-potassium [N: P: K (2:3:4=30)] fertilizers
applied when the maize was 50 days old to enhance growth.

Data acquisition
Ground data collection
Data for the study was collected at four phenological stages ranging from the vegetative to
reproductive growth phases i.e. V8 (32 days old), V12 (47 days old), R2 (96 days old) and R5 (123
days old) (Table 1). The vegetative stages were selected as they are characterized by fully
developed leaves, which is essential for field measurements and light reflectance. The R2 is full
canopy stage while R5 represents the end of mass gain in maize crop. Field measurements were

205
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

conducted at four-week intervals to capture the above-mentioned stages of the growth cycle. Using
a handheld Trimble Global Positioning System (GPS), 200 points were sampled using a stratified
random approach within the experimental plot. The experimental plot was divided into sub strata
based on slope, crop health, and crop size. Thereafter, random crops within the strata were sampled,
ensuring variability capturing and a comprehensive and representative sample of the maize
population. The approach was adopted to capture the size variability and representative crops for
the whole maize field. Each sample point was marked with red tape and labelled for consistent
monthly measurements. Field measurements were conducted on clear sunny days between 10 a.m.
and 14:00 p.m. to capture data at peak photosynthetic activity and maximum reflectance.

Table 1. Maize phenological stages used in the study.

Vegetative Stages
Growth Stage
V8 V12
Day after sowing 32 47

Maize Crop

Reproductive stages
Growth Stage
R2 R5
Day after sowing 96 123

Maize Crop

At each sampling point, LAI was obtained using a LiCOR 2200C plant canopy analyser (LI-COR
GmbH, Germany). The analyser uses 7°, 22°, 38°, 52°, and 68° zenith angles to measure light

206
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

interception and transmittance below and above the plant canopy and ultimately estimates the LAI
(Buthelezi et al., 2023). Soil moisture content was measured using HH2 moisture probe (Delta-T
soil moisture sensors, United states) at each sample point. The HH2 soil moisture probe is inserted
in the soil close to the root systems of the crop and records soil moisture volume with a 5% accuracy
based on standard calibration (Cheng et al., 2022). Leaf chlorophyll content was measured using a
Konica Minolta Soil Plant Analysis Development (SPAD) 502 chlorophyll meter (Minolta
corporation, Ltd., Osaka, Japan). The SPAD measures a unit less chlorophyll reflectance in the leaf
using the Red and Infrared portions of the electromagnetic spectrum (Brewer et al., 2022). Finally,
at the R6 phenological stage, marking the end of the growing season, the designated maize crops
underwent sampling, involving cutting the aboveground foliage, followed by weighing it using a
mass balance to determine the fresh AGB values at each sampling point. No mass correction was
performed on the maize crops, considering their crucial role in small-scale farming systems as a
source of both livestock fodder and human consumption. The decision to retain moisture in the
maize aligns with its practical use for easy swallowing, addressing the specific needs of both
animals and humans during this stage of maturity.

UAV platform and remotely sensed data acquisition

The digital multispectral images were collected over four phenological stages using a DJI Matrice
300 series (M300) UAV platform (SZ DJI Technology Co., Ltd, China) mounted with a MicaSense
Altum multispectral and thermal sensor (AgEagle Aerial Systems Inc, Kansas) (Fig.2a). The Altum
sensor is equipped with six spectral bands [red (668 nm), green (560 nm), blue (475 nm), red edge
(717 nm), near-infrared (840), and thermal band (8 to 14 nm)] (Fig.2c). The M300 is equipped with
Internet of Things (IoT) technology, such as obstacle avoidance sensors and a locational GPS
connected to the camera, making the drone safe to operate and capture automatically georectified
images. The UAV flights were conducted simultaneously with field measurements. A flight path
covering the experimental field was digitized from Google Earth Pro and imported into the drone
controller (Fig.2b). Before and after each flight, a whiteboard calibration panel was used to
calibrate the reflectance of the images (Fig.2d). The calibration panel was used to determine
illumination and atmospheric conditions during the flight for accurate vegetation indices retrieval.
The flights were conducted between 10:00 a.m. and 14:00 p.m. under open sky and suitable weather
conditions for optimum sunlight reflectance. The drone was operated at 15 m/s speed and 100 m
altitude with 80% forward and 70% side overlap. The images were collected at 6 cm per pixel
spatial resolution, based on 8mm focal length and 8o x37o field of view (FOV) angle.

207
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

A. DJI Matrice 300 B. UAV Controller and Flight Path

C. MicaSense Altum Sensor D. Calibration Panel

Figure 2. The UAV platform, controller with flight plan, image sensor, and calibration panel used
for remotely sensed data acquisition in this study.

Image pre-processing and retrieval of vegetation indices

A total of 480 images were collected during each flight at each sampled growth stage. During the
flights, the digital images were automatically georectified by the GPS payload mounted on the
M300 UAV platform. Subsequently, the Pix4D 4.6 Fields photogrammetry software (Pix4D Inc.
Denver, USA) was used to pre-process the images and generate an orthomosaic image and a digital
elevation model (DEM). In addition, the index calculator of the Pix4D photogrammetry software
was used to calculate optimal vegetation indices for estimating maize AGB (Table 2). The Pix4D
index calculator uses mathematical equations from the Index Data Base (IDB)
(https://www.indexdatabase.de/) to compute vegetation indices and provide a raster data showing
their spatial distribution. The maize sample points, orthomosaic, and vegetation index raster images
were imported into ArcGIS pro 10.7.1 software for data extraction using the ‘extract multi-values
to points’ in the Arc Toolbox. The extracted band reflectance and vegetation indices for each
sample point were then exported into Microsoft Excel for statistical analysis. Evidence from
literature has proven the efficiency of vegetation indices in predicting maize AGB (Han et al.,
2019b; Li et al., 2020; Li et al., 2016; Yue et al., 2023).

208
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 2. Selected optimum vegetation indices for predicting maize AGB.

Vegetation Index Formula Reference

𝑁𝐼𝑅 − 𝑅𝐸𝐷
NDVI (Shi & Xingguo, 2011)
𝑁𝐼𝑅 + 𝑅𝐸𝐷
𝑅𝐸𝐷
CVI NIR ( ) (Hunt Jr et al., 2011)
(𝐺𝑅𝐸𝐸𝑁)(𝐺𝑅𝐸𝐸𝑁)

𝑁𝐼𝑅 − 𝐵𝐿𝑈𝐸
BNDVI (Wang et al., 2007)
𝑁𝐼𝑅 + 𝐵𝐿𝑈𝐸
𝑅𝑒𝑑𝑒𝑑𝑔𝑒 − 𝑅𝐸𝐷
NDVI_Rededge (Ehammer et al., 2010)
𝑅𝑒𝑑𝑒𝑑𝑔𝑒 + 𝑅𝐸𝐷
𝑁𝐼𝑅 − (𝑅𝐸𝐷 + 𝐵𝐿𝑈𝐸)
RBNDVI (Wang et al., 2007)
𝑁𝐼𝑅 + (𝑅𝐸𝐷 + 𝐵𝐿𝑈𝐸)
((𝑁𝐼𝑅 + 𝐺𝑅𝐸𝐸𝑁) − (2 ∗ 𝐵𝐿𝑈𝐸))
ENDVI (Ahamed et al., 2011)
((𝑁𝐼𝑅 + 𝐺𝑅𝐸𝐸𝑁) + (2 ∗ 𝐵𝐿𝑈𝐸))
𝑁𝐼𝑅
CI_Rededge −1 (Hunt Jr et al., 2011)
𝑅𝑒𝑑 − 𝑒𝑑𝑔𝑒
2(𝐺𝑅𝐸𝐸𝑁 − 𝑅𝐸𝐷 − 𝐵𝐿𝑈𝐸)
GLI (Baroni et al., 2004)
2(𝐺𝑅𝐸𝐸𝑁 + 𝑅𝐸𝐷 + 𝐵𝐿𝑈𝐸)
(𝑁𝐼𝑅 − 𝑅𝐸𝐷)
EVI 2.5 ∗ (Glenn et al., 2010)
(𝑁𝐼𝑅 + 6𝑅𝐸𝐷 − 7.5𝐵𝐿𝑈𝐸) + 1
𝑁𝐼𝑅 − 𝑅𝐸𝐷
EVI2 2.4 ∗ (Miura et al., 2008)
𝑁𝐼𝑅 + 𝑅𝐸𝐷 + 1
𝑁𝐼𝑅
IPVI (𝑁𝐷𝑉𝐼 + 1) (Kooistra et al., 2003)
𝑁𝐼𝑅 + 𝑅𝐸𝐷
2
𝑁𝐼𝑅 − 𝑅𝐸𝐷
SAVI (1 + 0.5) (Heiskanen, 2006)
𝑁𝐼𝑅 + 𝑅𝐸𝐷 + 0.5
𝑁𝐼𝑅 − 𝑅𝐸𝐷
OSAVI (1 + 0.16) (Wu et al., 2008)
𝑁𝐼𝑅 + 𝑅𝐸𝐷 + 0.16
𝑁𝐼𝑅
SR (Malthus et al., 1993)
𝑅𝐸𝐷
𝑁𝐼𝑅
CI_Green −1 (Ahamed et al., 2011)
𝐺𝑅𝐸𝐸𝑁

GDVI 𝑁𝐼𝑅 − 𝐺𝑅𝐸𝐸𝑁 (Tucker et al., 1979)

Where, NDVI= Normalized Difference Vegetation Index, CVI= Chlorophyll Vegetation Index, BNDVI= Blue
Normalized Difference Vegetation Index, NDVI_Rededge =Normalized Difference Vegetation Index Red edge,
RBNDVI= Red Blue Normalized Difference Vegetation Index, ENDVI= Enhanced Normalized difference Vegetation
Index, CI_Rededge= Chlorophyll Index Red edge, GLI= Green Leaf Index, EVI= Enhanced Vegetation Index, IPVI=
Infrared Percentage Vegetation Index, SAVI= Soil Adjusted Vegetation Index, OSAVI= Optimised Soil Adjusted
Vegetation Index, SR= Simple Ratio, CI_Green= Chlorophyll Index Green, GDVI= Generalised Difference Vegetation
Index

209
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Retrieval of landscape variables

To complete the objective of this study, landscape variables that significantly influence maize
growth such as slope and aspect were acquired from the System for Automated Geoscientific
Analyses (SAGA) Geographic Information Systems (GIS) 7.8.2 software (University of Hamburg,
Germany). The digitized experimental field boundary and maize sample points were then used to
clip and extract the landscape variables to the extent of the study area using ArcGIS Pro. Even
though soil moisture was measured in-field together with biophysical variables, it was categorized
under landscape variables because it quantifies the amount of water held by the soil. In addition,
the DEM generated by Pix4D software was used to extract elevation data to the maize sample
points as a landscape variable. The UAV remotely sensed data was then combined with the
extracted landscape variables and field-measured biophysical variables in an Excel file for
statistical analysis (Table 3). The data was then split into training (70%), and testing (30%) datasets
using randomisation, thereby ensuring non-bias splitting and ensuring representative subsets for
model training and validation.

Table 3. Input variables.

Variable Data type Number of variables

Remotely sensed Spectral bands 21
Vegetation indices
Landscape variables Aspect 4
Elevation
Slope
Soil moisture
Biophysical variables Leaf chlorophyll content 2
LAI
Total 8 27

AGB prediction
Deep learning architecture
Jupyter notebook extended from Anaconda3 was used to build a fully connected DNN model
featuring 17 inputs, three hidden, and one output layer using python programming environment for
predicting maize AGB at four phenological stages (Fig.3). The combination of innovative
computational tools and sophisticated DNN architecture facilitates precise AGB predictions,
contributing to a deeper understanding of maize growth dynamics and potential applications in
agriculture (Coulibaly et al., 2022; Fuentes et al., 2017). DNN models are powerful in capturing
non-linear relationships by self-learning from large datasets and make precise predictions (Zhang
et al., 2022). DNN models use multiple layers with fully connected neurons that are similar to
human brain neurons and known to produce highly accurate results, surpassing human experts
(Saranya et al., 2023; Z. Zeng et al., 2022). Therefore, DNN have the potential to improve
prediction accuracy of maize AGB compared to other traditional machine learning and statistical
methods.

210
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 3. The diagrammatic illustration of the DNN model.

A good selection of hyperparameters based on the dataset is essential for building an optimum
model (Dominguez-Olmedo, 2019). Therefore, the rectified linear unit (ReLu) was used in the
input and hidden layers, respectively, to introduce non-linearity in the model. Linearity in DNN
imply that all hidden layers have the same power in predicting the output (Kapočiūtė-Dzikienė et
al., 2020). Due to the complexity and non-linearity within datasets, the hidden layers must have
different magnitude of power in predicting the output (Tsai & Fang, 2021). Therefore, it is essential
to introduce activation functions in the neural network to distinguish the hidden layers from each
other for better detection and learning of the non-liner relationship between the input and predictor
variables (Dubey et al., 2022; Jiang et al., 2022; X. Wang et al., 2022). The model was run over
500 epochs, implying that weights in the hidden layers were constantly adjusted five hundred times
to minimize error and improve the maize AGB prediction accuracy. The input data is forwardly
propagated to the hidden layers, where the weights and biases in the neurons predict the output by
self-learning non-linear patterns from the input dataset. The loss functions quantify the deviation
from the expected output and backwardly propagate the output to the hidden layers for adjustments
in pursuit of minimising the prediction error (Dubey et al., 2022)

The output layer was fed with a SoftMax activation function and “Adam’ optimizer for model
optimization and best results. Optimizers reduce the loss by selecting optimum weights in hidden
layers to determine an optimum model for accurate prediction (Cho et al., 2020). Adam is known
to surpass other optimizers such as stochastic gradient descent due to its ability of generalization
and convergence speed within new datasets (Gaddam et al., 2022; Salem et al., 2022; Y. Wang et
al., 2022). A batch size of 32 and normal initialization were also implemented in the model for best

211
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

results. Neural network models are well known for overfitting, which is explained as when the
training dataset yields significantly better results than the testing dataset (Frei et al., 2022). Such
model cannot be generalized and cannot accurately predict from an unknown dataset. Therefore,
the L2 regularization (0.001) and a dropout of 0.4 were implemented in the layers of the model to
minimise overfitting. The dropout and regularization features in DNN minimize loss between the
predicted output and observed input and nullify the contribution of “bad” neurons towards
subsequent layers, hence a better prediction accuracy.

Accuracy assessment
The Root Mean Square Error (RMSE) and coefficient of determination (R2) were used to evaluate
the metrics. The RMSE is the difference between the predicted and the observed output, while the
R2 reflects the percentage of the AGB variance that is explained by the model. The best performing
model is represented by a higher R2 value and a lower RMSE. The variable importance in predicting
maize AGB was evaluated using the SHapley Additive exPlanations (SHAP) approach. The SHAP
uses a theoretic approach that selects the top twenty variables of high magnitude impact in the
performance of the model (Ekanayake et al., 2022).

Data preparation, variables selection, and model validation

Data preparation and variables selection
The correlation coefficient (R) was calculated between the predictor variables using correlation
heat maps to choose significantly low correlated values for best results (Fig.4). Thereafter, highly
correlated variables within the dataset were identified and removed to ensure maximum prediction
accuracy as such variables have technically the same magnitude impact in the performance of the
model.

212
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 4. Pearson correlation (R) between the selected maize AGB predictor variables for all the
phenological stages.

DNN model validation

Fig.5 shows loss curves during the validation of the DNN model using the training and test dataset
over 500 epochs. Model validation is necessary for evaluating the performance of the DNN during
self-learning from the dataset. The data was separated into 70% training and 30% testing dataset,
and subsequently validated using the latter. The training and validation curves showed a uniform
function, implying a gradual decrease in the maize AGB prediction error across all the phenological
stages, hence the model was perfectly validated. An optimum model for predicting maize AGB
was established and tested using the training dataset.

213
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 5. Loss graphs for model validation during all the phenological stages.

RESULTS

Descriptive statistics
The variations in field-measured biophysical and landscape variables of maize crops are shown in
Table 4. On average, the recorded SPAD unit-less leaf chlorophyll content was 39.26, 37.38, 31.22,
and 41.36 during the V8, V12, R2, and R5 phenological stages, respectively. The R5 phenological
stage recorded the highest average chlorophyll content of 41.36. Soil moisture averages were
21.87%, 21.41%, 16.97%, and 19.1% during the V8, V12, R2, and R5 phenological stages,
respectively. It was observed that soil moisture content decreased with growth from the V8-R5
phenological stages. The averages for LAI were 3.64, 2.78, 3.25, and 3.16 during the V8, V12, R2,
and R5 phenological stages, respectively, with V8 recording the highest average.

Landscape variables are not subjected to rapid changes over a short time and were therefore
assumed to be the same throughout the duration of the study. The average slope, elevation, and
aspect were 9%, 856 m, and 2.73 degrees, respectively. The slope, elevation, and aspect ranged
from 2% to 14%, 847m to 862m, and 2.20 degrees to 3.42 degrees, respectively. The recorded
maize AGB was 1.19 kg/m2 on average and ranged from 0.4 kg/m2 to 1.81 kg/m2, with 2.03 kg/m2
and 2.11 kg/m2 recorded as outliers. The outliers were due to measurement errors in the field and
were therefore removed from the analysis for best results.

214
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 4. Descriptive statistics of field measured biophysical and landscape variables across all
phenological stages.

V8 Stage
Range (Min-Max) Mean Median Std.
Chlorophyll 31.36-49.61 39.26 38.99 2.59
LAI 13.99-27.01 21.87 22.16 2.21
Soil Moisture 2.48-4.64 3.64 3.64 0.25
(%)
V12 Stage
Range (Min-Max) Mean Median Std.
Field measured variables

Chlorophyll 29.70- 44.76 37.38 37.60 2.62

LAI 15.67- 30.41 21.41 21.06 1.57
Soil Moisture 1.62- 4.58 2.78 2.74 0.23
(%)
R2 Stage
Range (Min-Max) Mean Median Std.
Chlorophyll 21.36- 47.49 31.22 31.55 4.17
LAI 13.52- 23.67 16.97 16.64 1.36
Soil Moisture 1.92- 7.75 3.25 3.38 0.49
(%)
R5 Stage
Range (Min-Max) Mean Median Std.
Chlorophyll 21.6-59.4 41.36 41.7 6.27
LAI 10.6-31.3 19.68 19.4 4.13
Soil Moisture 1.41-6.8 3.16 3.03 4.85
(%)
Across all stages
Landscape

Range (Min-Max) Mean Median Std.

variables

Slope (%) 2-14 9 10 4

Elevation (m) 847-862 856 857.6 4.02
Aspect (degrees) 2.20-3.42 2.73 2.68 0.30

Deep Neural Network model evaluation in maize AGB prediction

Fig. 6 illustrates the maize AGB prediction results obtained when the most important and best
performing variables were combined for all phenological stages. The V8 (R 2=0.65, RMSE= 0.1
kg/m2, RMSE%=8.5%) and R5 (R2=0.67, RMSE= 0.091 kg/m2, RMSE%=7.6%) phenological
stages had a relatively lower prediction accuracy. However, the prediction error was within the 10
% accepted range. The V12 (R2=0.74, RMSE=0.07 kg/m2, RMSE% =5.9%) and R2 (R2=0.71,
RMSE=0.086 kg/m2, RMSE%=7.3%) phenological stages performed optimally, with relatively
high prediction accuracy.

215
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 6. Predicted maize AGB using DNN model for the V8, V12, R2, and R5 phenological
stages.

Variable importance assessment

Fig. 7 shows the most important variables in the prediction of maize AGB by the DNN model using
the SHAP approach. The SR vegetation index was most important during the V8 and R2, while
leaf chlorophyll content and elevation were most influential during the V12 and R5 phenological
stages. The figure shows that all landscape variables were important in the prediction of maize
AGB across all phenological stages. The biophysical variables (LAI and leaf chlorophyll content)
were among the top six important variables during the V8, V12, and R5 phenological stages, while
the Red spectral band was the least important variable in maize AGB prediction across all
phenological stages. EVI had an extremely low importance in predicting maize AGB during the
V12 and R5 phenological stages.

216
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 7. SHAP generated variable importance ranking of the model’s input variables for all the
phenological stages.

Mapping the spatial distribution of predicted AGB across the phenological stages
Fig.8 shows the spatial distribution of predicted maize AGB during all the phenological stages.
The spatial distribution map was generated utilizing the important predictor variables (Fig.7) for
maize AGB prediction and the equation of the line of best fit derived from scatter plots comparing
predicted and observed AGB at each phenological stage. Typically, a raster file of the most
important maize AGB predictor variable is generated using ArcMap, and the equation y=mx+c is
applied, substituting x with the raster file. The generated distribution maps show an increase in
maize AGB from the V8 to the R2 phenological stage. There was a slight decrease in the
concentration of AGB during the R5 stage. This distribution is also shown by the prediction

217
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

accuracy previously presented in Fig.6, which shows relatively higher prediction accuracy during
the V8 and the R1, and lower during the R5 and V8 phenological stages. Similarly, the distribution
maps show the same relationship in maize AGB concentration. During all phenological stages,
high AGB concentration was observed towards the edges and the field’s downslope. In addition,
during all the phenological stages, low AGB was observed in a middle of the experimental field.

Figure 8. Spatial distribution of predicted maize AGB across all the phonological stage.

DISCUSSION

In developing countries, small-scale farming systems typically lack crop monitoring resources and
knowledge on techniques to optimize yield (Onyango et al., 2021). Hence, this study bridged the
gap by implementing an affordable crop monitoring resources such as the in-situ instruments, the
UAV platform and sensor to accurately estimate maize AGB, which can serve as a proxy to yield.
Specifically, this study aimed to develop a model that can accurately predict maize AGB and
determine the optimal phenological stage for maize AGB estimation.

The potential of UAV-remotely sensed data in predicting maize AGB

Unmanned Aerial Vehicle-remotely sensed data offer a promising capability to effectively estimate
maize AGB in small spatial extents. This is attributed to the remarkable ability of the platform
mounted with sophisticated sensors to provide high spatial resolution dataset, enabling individual

218
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

sensing and assessment of maize crops for accurate AGB estimation (Khun et al., 2021; Niu et al.,
2019). Unmanned Aerial Vehicle-mounted cameras such as multispectral sensors offer a broad
range of the electromagnetic bands including the visible, NIR, Red-edge, and thermal sections,
allowing for efficient retrieval of vegetation indices capable of estimating maize AGB (Olson &
Anderson, 2021). This study successfully predicted AGB at various maize phenological stages
using UAV-remotely sensed data and deep learning approach. The results indicated that the V12
and R2 phenological stage reported relatively high accuracy in AGB predictions (R 2=0.74 and
RMSE=0.07 kg/m2) and (R2=0.71 and RMSE= 0.086 kg/m2), respectively. The V12-R2
phenological stages are the mid-stages of maize growth cycle and portray dark green leaves,
symbolizing a high concentration of leaf chlorophyll content (Herrmann et al., 2010). Hence, the
best results were obtained during the V12-R2 period due to optimum reflectance of maize leaves
and minimal soil background noise. The findings of our study concur with B. Yang et al. (2022)
who used multi-temporal and mono-temporal UAV-remotely sensed data and noted that R3 was
the most suitable phenological stage for maize AGB prediction. Similarly, Amanullah et al. (2009)
investigated maize yield using traditional methods, and established that the V12-R1 phenological
stages had relatively higher yield compared to other phenological stages. Therefore, based on our
results, we can deduce that V12-R2 is the optimum phenological stage for maize AGB estimation.

The V8 phenological stage and R5 phenological stages had lower maize AGB prediction
accuracies, i.e., R2 = 0.65 and R2 = 0.67, respectively. This was because the maize canopy was not
fully developed and soil background was more pronounced at V8 stage, hence, interfering with
maize reflectance signatures (Y. Zeng et al., 2022). The spatial distribution map shows a high maize
AGB downslope and some parts of the field where soils were thick and appeared rich in nutrients
(Fig.8). Thin soils were also observed upslope and in some parts of the field; low maize AGB was
observed in those areas. Considering that the study area is small, there was a significant variation
in soil thickness, which is why the predicted concentration of maize AGB is not uniform across the
experimental field. Thick soils have a better water retention and nutrient holding capacity for crop’s
use, hence higher maize AGB (Mu et al., 2018).

Brewer et al. (2022) noted that NIR derived vegetation indices can surpass variable background
effects compared to conventional bands. The soil-adjusted vegetation indices were selected to
eliminate soil background and accurately predict maize AGB. As expected, SAVI was among the
significantly influential variables in the estimation of maize AGB during all the phenological
stages, including the V8 where vegetation cover was minimal. The R5 phenological stage was
characterized by dry-denting leaves and marked the end of mass gain. We speculate that the dry
leaves significantly reduced the reflectance; hence remotely sensed variables were less important
and lower maize AGB prediction accuracy was observed during this stage. While Red-edge-based
vegetation indices were influential, they did not have a significant contribution to AGB prediction
as compared to NIR-derived indices. The findings of this study are supported by Gao et al. (2017)
who confirmed the efficacy of vegetation index-based biomass estimation in maize crops.

Plant biophysical variables in maize AGB prediction

The relationship between LAI, leaf chlorophyll content, and AGB is crucial in understanding the
physiological and agronomic aspects of maize growth and productivity (Ban et al., 2019). LAI
represents the total leaf area per unit ground area and is often indicates the canopy structure and
the light interception capacity of maize crop (Liu et al., 2022). It is positively correlated with
photosynthetic activity, as a higher LAI generally implies a larger surface area for light absorption,
219
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

hence high productivity (Li et al., 2023). Leaf chlorophyll content is a key factor influencing
photosynthesis, as chlorophyll is responsible for capturing light energy and transform it into
chemical energy (Y. Guo et al., 2020). Higher chlorophyll content is generally associated with
increased photosynthetic rates, contributing to greater biomass production (Meena et al., 2021).
Hence, optimal LAI and chlorophyll content contribute to enhanced photosynthesis, leading to
increased biomass accumulation in maize crops.

In this study, the recorded leaf chlorophyll content was higher in the early stage (V8) and the late
reproductive stage (R5). This is supported by Brewer et al. (2022) who noted that high chlorophyll
concentrations are associated with early vegetation and late reproduction stages when maize grows
rapidly and kernelling, respectively. Similarly, leaf chlorophyll content in the early and late
reproductive stages is associated with high LAI (H. Yang et al., 2022). As shown by the SHAP
variable importance approach, leaf chlorophyll content had a relatively high impact on maize AGB
prediction across all the phenological stages. Our results concur with Liu et al. (2019) who
established a positive co-relationship between maize AGB and leaf chlorophyll content. In
addition, LAI also had a relatively high impact on maize AGB prediction during the V8, V12, and
R5 phenological stages. Contrary to our results, Tang et al. (2023) also established a strong
relationship between LAI and maize yield after the R1 phenological stage in maize crops.

The potential of landscape variables on improving maize AGB prediction

Landscape variables significantly increased the maize AGB prediction accuracy and were all
important during all the phenological stages (Fig.7). In addition, the landscape variables were less
correlated to each other, hence the DNN model performed well with their inclusion. A study by
Sun et al. (2023) successfully combined topographic variables and vegetation and texture indices
to predict maize yield and obtained satisfactory results (R 2 = 0.81, RMSE = 0.297t/ha), which
confirms the value of landscape variables in maize AGB prediction. Similarly, Behera et al. (2023)
used elevation, slope, and aspect to model maize AGB, and obtained satisfactory results (R 2 = 0.72
and RMSE= 69.18 mg/ha). Salinas-Melgoza et al. (2018) modelled a relationship between
landscape variables and noted that landscape variables explained 21% of AGB in reforested areas.
Salinas-Melgoza et al. (2018) argued that human activities such as deforestation, land degradation,
improper irrigation methods, changing land uses, and pollution have a significant impact on
landscape alteration, while crops productivity heavily depend on landscape variables. These human
activities facilitate soil erosion, urban expansion, and alterations of soil productivity which
significantly affect the slope, aspect, elevation and soil water holding capacity (Mariye et al., 2022).

The performance of deep neural network model in maize AGB prediction

The deep learning approach in maize yield prediction was evaluated using UAV-remotely sensed
data combined with biophysical and landscape variables. Furthermore, with DNN requirements for
large datasets, the three variables used in this model were adequate to feed enough information to
the model for accurate maize AGB prediction. The main objective of this study was to evaluate
deep learning approach in maize AGB prediction, particularly with minimal dataset obtained from
small spatial extent. To obtain a reliable statistical relationship, DNN require large sample size to
effectively learn and discover patterns between the predictor and test variables (Zhang et al., 2021).
Therefore, we sampled 200 points to generate an effective model. Based on the overall RMSE and
RMSE% achieved in this study, our model had minimal prediction errors (< 10%) across all the
phenological stages. In addition, combining three different sources of datasets improved the
prediction accuracy of maize AGB. This was because DNN require a lot of complex and nonlinear
220
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

datasets to perform effectively. Han et al. (2019a) successfully modelled maize AGB in
commercial farming systems using DNN and other machine learning algorithms and achieved
satisfactory results. However, this study argues that DNNs require significant repeat training,
necessitating a lot of computational power to obtain an optimal model in minimal time.

Implications and recommendations

Unmanned Aerial Vehicle-mounted multispectral sensors provide high resolution dataset, allowing
for detection of crop agronomical characteristics facilitating maize AGB estimation (Zhai et al.,
2023). However, the spectral information of crops remains coarse due to multispectral wide bands
portraying lower spectral resolution. Therefore, hyperspectral remote sensing for precise spectral
information retrieval and effective maize AGB prediction is highly recommended. Additionally,
the study concluded that landscape variables have a significant impact on maize AGB prediction.
However, the analysis did not include an assessment of the magnitude of each landscape variable's
influence on maize AGB prediction. Hence, it is recommended that forthcoming research
endeavours explore the specific impact of individual landscape variables in the prediction of maize
AGB. This will contribute to the improvement of validated data availability for further yield
predictions. The DNN model requires extensive hyperparameters adjustments to obtain an optimal
model. Consequently, its suitability for tasks demanding rapid turnaround times is not
recommended. The acquired DNN model was trained using maize derived datasets, and validated
using unknown maize dataset, thereby enhancing its applicability in other locations with variability
in landscape variables. However, the performance of the acquired model remains limited to
adequate dataset and applicable to maize AGB prediction only. Despite the success of the DNN
model in adequately predicting maize AGB, more studies need to extensively explore the full
potential of this approach, considering its promising potential to make accurate predictions. Despite
the success of the DNN model in adequately predicting maize AGB, more studies need to
extensively explore the full potential of this approach, considering its promising potential to make
accurate predictions.

CONCLUSION

The study sought to assess the utility of landscape and biophysical variables combined with UAV-
remotely sensed data in predicting maize AGB using the DNN algorithm at four phenological
stages (V8, V12, R2, and R5). Based on the attained results, it can be concluded that:

• The V12-R1 phenological stages are optimal for maize AGB prediction when vegetation
reflectance is at peak.
• Landscape variables improve the prediction accuracy of maize AGB and can therefore be
used in maize AGB estimation.
• The Near infrared spectral bands were the most influential variables in predicting maize
AGB prediction.
• Landscape variables, biophysical variables, and UAV-remotely sensed vegetation indices
demonstrated significant importance in predicting maize AGB. Hence, the combination of
these variables has demonstrated the ability to improve maize AGB prediction, underscoring
the effectiveness achieved through their collaborative utilization in this study.
• Finally, the DNN algorithm yielded satisfactory results, attributable to the combined
dynamic and non-linear datasets in pursuit of a good model.

221
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

The results of this study have a significant contribution to precision agriculture particularly in
underprivileged small-scale farming systems. Furthermore, the findings of this study address gaps
in the current literature, notably by introducing smart agriculture concepts to the global south for
improved maize production and sustenance.

REFERENCES

Aasen, H., Honkavaara, E., Lucieer, A., & Zarco-Tejada, P. J. (2018). Quantitative remote sensing
at ultra-high resolution with UAV spectroscopy: a review of sensor technology, measurement
procedures, and data correction workflows. Remote sensing, 10(7), 1091.
Adewopo, J., Peter, H., Mohammed, I., Kamara, A., Craufurd, P., & Vanlauwe, B. (2020). Can a
combination of uav-derived vegetation indices with biophysical variables improve yield
variability assessment in smallholder farms? [Article]. Agronomy, 10(12), Article 1934.
https://doi.org/10.3390/agronomy10121934
Ahamed, T., Tian, L., Zhang, Y., & Ting, K. (2011). A review of remote sensing methods for
biomass feedstock production. Biomass and bioenergy, 35(7), 2455-2469.
Ali, A. M., Abouelghar, M., Belal, A., Saleh, N., Yones, M., Selim, A. I., Amin, M. E., Elwesemy,
A., Kucher, D. E., & Maginan, S. (2022). Crop yield prediction using multi sensors remote
sensing. The Egyptian Journal of Remote Sensing and Space Science, 25(3), 711-716.
Altaweel, M., Khelifi, A., Li, Z., Squitieri, A., Basmaji, T., & Ghazal, M. (2022). Automated
archaeological feature detection using deep learning on optical UAV imagery: Preliminary
results. Remote sensing, 14(3), 553.
Amanullah, Yasir, M., Khalil, S. K., Jan, M. T., & Khan, A. Z. (2009). Phenology, growth, and
grain yield of maize as influenced by foliar applied urea at different growth stages. Journal
of Plant Nutrition, 33(1), 71-79.
Ban, H.-Y., Ahn, J.-B., & Lee, B.-W. (2019). Assimilating MODIS data-derived minimum input
data set and water stress factors into CERES-Maize model improves regional corn yield
predictions. PLoS One, 14(2), e0211874.
Baroni, F., Boscagli, A., Di Lella, L., Protano, G., & Riccobono, F. (2004). Arsenic in soil and
vegetation of contaminated areas in southern Tuscany (Italy). Journal of Geochemical
Exploration, 81(1-3), 1-14.
Battude, M., Al Bitar, A., Morin, D., Cros, J., Huc, M., Sicre, C. M., Le Dantec, V., & Demarez,
V. (2016). Estimating maize biomass and yield over large areas using high spatial and
temporal resolution Sentinel-2 like remote sensing data. Remote Sensing of Environment,
184, 668-681.
Behera, D., Kumar, V. A., Rao, J. P., Padal, S., Ayyappan, N., & Reddy, C. S. (2023). Estimating
Aboveground Biomass of a Regional Forest Landscape by Integrating Textural and Spectral
Variables of Sentinel-2 Along with Ancillary Data. Journal of the Indian Society of Remote
Sensing, 1-13.
Brewer, K., Clulow, A., Sibanda, M., Gokool, S., Naiken, V., & Mabhaudhi, T. (2022). Predicting
the chlorophyll content of maize over phenotyping as a proxy for crop health in smallholder
farming systems. Remote sensing, 14(3), 518.
Buthelezi, S., Mutanga, O., Sibanda, M., Odindi, J., Clulow, A. D., Chimonyo, V. G., &
Mabhaudhi, T. (2023). Assessing the prospects of remote sensing maize leaf area index using
UAV-derived multi-spectral data in smallholder farms across the growing season. Remote
sensing, 15(6), 1597.
222
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Cao, Y., Chen, Z., Belkin, M., & Gu, Q. (2022). Benign overfitting in two-layer convolutional
neural networks. Advances in neural information processing systems, 35, 25237-25250.
Che, Y., Wang, Q., Zhou, L., Wang, X., Li, B., & Ma, Y. (2022). The effect of growth stage and
plant counting accuracy of maize inbred lines on LAI and biomass prediction. Precision
Agriculture, 23(6), 2159-2185.
Cheng, M., Jiao, X., Liu, Y., Shao, M., Yu, X., Bai, Y., Wang, Z., Wang, S., Tuohuti, N., & Liu,
S. (2022). Estimation of soil moisture content under high maize canopy coverage from UAV
multimodal data and machine learning. Agricultural Water Management, 264, 107530.
Cheng, Z., Meng, J., Ji, F., Wang, Y., Fang, H., & Yu, L. (2020). Aboveground biomass estimation
of late-stage maize based on the WOFOST model and UAV observations [Article]. Yaogan
Xuebao/Journal of Remote Sensing, 24(11), 1403-1418.
https://doi.org/10.11834/jrs.20200069
Cho, H., Kim, Y., Lee, E., Choi, D., Lee, Y., & Rhee, W. (2020). Basic enhancement strategies
when using Bayesian optimization for hyperparameter tuning of deep neural networks. IEEE
access, 8, 52588-52608.
Coulibaly, S., Kamsu-Foguem, B., Kamissoko, D., & Traore, D. (2022). Deep learning for
precision agriculture: A bibliometric analysis. Intelligent Systems with Applications, 16,
200102.
Dominguez-Olmedo, R. (2019). Sample-Efficient Deep Reinforcement Learning for On-The-Fly
Thermal Process Control in Laser Powder Bed Fusion University of Sheffield].
Dubey, S. R., Singh, S. K., & Chaudhuri, B. B. (2022). Activation functions in deep learning: A
comprehensive survey and benchmark. Neurocomputing.
Ehammer, A., Fritsch, S., Conrad, C., Lamers, J., & Dech, S. (2010). Statistical derivation of fPAR
and LAI for irrigated cotton and rice in arid Uzbekistan by combining multi-temporal
RapidEye data and ground measurements. Remote Sensing for Agriculture, Ecosystems, and
Hydrology XII,
Ekanayake, I., Meddage, D., & Rathnayake, U. (2022). A novel approach to explain the black-box
nature of machine learning in compressive strength predictions of concrete using Shapley
additive explanations (SHAP). Case Studies in Construction Materials, 16, e01059.
Frei, S., Chatterji, N. S., & Bartlett, P. (2022). Benign overfitting without linearity: Neural network
classifiers trained by gradient descent for noisy linear data. Conference on Learning Theory,
Fry, J. E., & Guber, A. K. (2020). Temporal stability of field-scale patterns in soil water content
across topographically diverse agricultural landscapes. Journal of Hydrology, 580, 124260.
Fuentes, A., Yoon, S., Kim, S. C., & Park, D. S. (2017). A robust deep-learning-based detector for
real-time tomato plant diseases and pests recognition. Sensors, 17(9), 2022.
Gaddam, D. K. R., Ansari, M. D., Vuppala, S., Gunjan, V. K., & Sati, M. M. (2022). A performance
comparison of optimization algorithms on a generated dataset. ICDSMLA 2020: Proceedings
of the 2nd International Conference on Data Science, Machine Learning and Applications,
Gao, Y., Gao, J., Wang, J., Wang, S., Li, Q., Zhai, S., & Zhou, Y. (2017). Estimating the biomass
of unevenly distributed aquatic vegetation in a lake using the normalized water-adjusted
vegetation index and scale transformation method. Science of the Total Environment, 601,
998-1007.
Gargiulo, J. I., Lyons, N. A., Masia, F., Beale, P., Insua, J. R., Correa-Luna, M., & Garcia, S. C.
(2023). Comparison of Ground-Based, Unmanned Aerial Vehicles and Satellite Remote
Sensing Technologies for Monitoring Pasture Biomass on Dairy Farms. Remote sensing,
15(11), 2752.

223
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Geng, L., Che, T., Ma, M., Tan, J., & Wang, H. (2021). Corn biomass estimation by integrating
remote sensing and long-term observation data based on machine learning techniques.
Remote sensing, 13(12), 2352.
Gerke, D. (2019). An Evaluation of Unmanned Aerial Systems and Structure-From-Motion for
Fluvial Large Wood Sensing and Risk Assessment.
Glenn, E. P., Nagler, P. L., & Huete, A. R. (2010). Vegetation index methods for estimating
evapotranspiration by remote sensing. Surveys in Geophysics, 31, 531-555.
Goldenberg, M. G., Burian, A., Seppelt, R., Ossa, F. A. S., Bagnato, C. E., Satorre, E. H., Martini,
G. D., & Garibaldi, L. A. (2022). Effects of natural habitat composition and configuration,
environment and agricultural input on soybean and maize yields in Argentina. Agriculture,
Ecosystems & Environment, 339, 108133.
Guo, Y., Wang, H., Wu, Z., Wang, S., Sun, H., Senthilnath, J., Wang, J., Bryant, C. R., & Fu, Y.
(2020). Modified red blue vegetation index for chlorophyll estimation and yield prediction
of maize from visible images captured by uav [Article]. Sensors (Switzerland), 20(18), 1-16,
Article 5055. https://doi.org/10.3390/s20185055
Guo, Y., Wang, H., Wu, Z., Wang, S., Sun, H., Senthilnath, J., Wang, J., Robin Bryant, C., & Fu,
Y. (2020). Modified red blue vegetation index for chlorophyll estimation and yield prediction
of maize from visible images captured by UAV. Sensors, 20(18), 5055.
Haarhoff, S. J., Kotzé, T. N., & Swanepoel, P. A. (2020). A prospectus for sustainability of rainfed
maize production systems in South Africa. Crop Science, 60(1), 14-28.
Han, L., Yang, G., Dai, H., Xu, B., Yang, H., Feng, H., Li, Z., & Yang, X. (2019a). Modeling
maize above-ground biomass based on machine learning approaches using UAV remote-
sensing data. Plant Methods, 15(1), 10. https://doi.org/10.1186/s13007-019-0394-z
Han, L., Yang, G., Dai, H., Xu, B., Yang, H., Feng, H., Li, Z., & Yang, X. (2019b). Modeling
maize above-ground biomass based on machine learning approaches using UAV remote-
sensing data. Plant methods, 15(1), 1-19.
Heiskanen, J. (2006). Estimating aboveground tree biomass and leaf area index in a mountain birch
forest using ASTER satellite data. International Journal of Remote Sensing, 27(6), 1135-
1158.
Herrmann, I., Karnieli, A., Bonfil, D., Cohen, Y., & Alchanatis, V. (2010). SWIR-based spectral
indices for assessing nitrogen content in potato fields. International Journal of Remote
Sensing, 31(19), 5127-5143.
Hunt Jr, E. R., Daughtry, C., Eitel, J. U., & Long, D. S. (2011). Remote sensing leaf chlorophyll
content using a visible band index. Agronomy Journal, 103(4), 1090-1099.
Jiang, F., Sun, H., Ma, K., Fu, L., & Tang, J. (2022). Improving aboveground biomass estimation
of natural forests on the Tibetan Plateau using spaceborne LiDAR and machine learning
algorithms. Ecological Indicators, 143, 109365.
Kapočiūtė-Dzikienė, J., Balodis, K., & Skadiņš, R. (2020). Intent detection problem solving via
automatic DNN hyperparameter optimization. Applied Sciences, 10(21), 7426.
Kayad, A., Sozzi, M., Gatto, S., Marinello, F., & Pirotti, F. (2019). Monitoring within-field
variability of corn yield using Sentinel-2 and machine learning techniques. Remote sensing,
11(23), 2873.
Khaki, S., & Wang, L. (2019). Crop yield prediction using deep neural networks. Frontiers in plant
science, 10, 621.
Khan, S. N., Li, D., & Maimaitijiang, M. (2022). A Geographically Weighted Random Forest
Approach to Predict Corn Yield in the US Corn Belt. Remote sensing, 14(12), 2843.

224
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Khun, K., Tremblay, N., Panneton, B., Vigneault, P., Lord, E., Cavayas, F., & Codjia, C. (2021).
Use of oblique RGB imagery and apparent surface area of plants for early estimation of
above-ground corn biomass [Article]. Remote Sensing, 13(20), Article 4032.
https://doi.org/10.3390/rs13204032
Kooistra, L., Leuven, R., Wehrens, R., Nienhuis, P., & Buydens, L. (2003). A comparison of
methods to relate grass reflectance to soil metal contamination. International Journal of
Remote Sensing, 24(24), 4995-5010.
Leroux, L., Castets, M., Baron, C., Escorihuela, M.-J., Bégué, A., & Seen, D. L. (2019). Maize
yield estimation in West Africa from crop process-induced combinations of multi-domain
remote sensing indices. European Journal of Agronomy, 108, 11-26.
Li, B., Xu, X., Zhang, L., Han, J., Bian, C., Li, G., Liu, J., & Jin, L. (2020). Above-ground biomass
estimation and yield prediction in potato by using UAV-based RGB and hyperspectral
imaging. ISPRS Journal of Photogrammetry and Remote Sensing, 162, 161-172.
Li, F., Hao, D., Zhu, Q., Yuan, K., Braghiere, R. K., He, L., Luo, X., Wei, S., Riley, W. J., & Zeng,
Y. (2023). Vegetation clumping modulates global photosynthesis through adjusting canopy
light environment. Global Change Biology, 29(3), 731-746.
Li, W., Niu, Z., Chen, H., Li, D., Wu, M., & Zhao, W. (2016). Remote estimation of canopy height
and aboveground biomass of maize using high-resolution stereo images from a low-cost
unmanned aerial vehicle system. Ecological Indicators, 67, 637-648.
Li, X., Geng, H., Zhang, L., Peng, S., Xin, Q., Huang, J., Li, X., Liu, S., & Wang, Y. (2022).
Improving maize yield prediction at the county level from 2002 to 2015 in China using a
novel deep learning approach. Computers and Electronics in Agriculture, 202, 107356.
Li, Z., Zhao, Y., Taylor, J., Gaulton, R., Jin, X., Song, X., Li, Z., Meng, Y., Chen, P., & Feng, H.
(2022). Comparison and transferability of thermal, temporal and phenological-based in-
season predictions of above-ground biomass in wheat crops from proximal crop reflectance
data. Remote Sensing of Environment, 273, 112967.
Lischeid, G., Webber, H., Sommer, M., Nendel, C., & Ewert, F. (2022). Machine learning in crop
yield modelling: A powerful tool, but no surrogate for science. Agricultural and Forest
Meteorology, 312, 108698.
Liu, C., Liu, Y., Lu, Y., Liao, Y., Nie, J., Yuan, X., & Chen, F. (2019). Use of a leaf chlorophyll
content index to improve the prediction of above-ground biomass and productivity. PeerJ, 6,
e6240.
Liu, G., Yang, Y., Liu, W., Guo, X., Xie, R., Ming, B., Xue, J., Zhang, G., Li, R., & Wang, K.
(2022). Optimized canopy structure improves maize grain yield and resource use efficiency.
Food and Energy Security, 11(2), e375.
Liu, J., Fang, Y., Wang, G., Liu, B., & Wang, R. (2023). The aging of farmers and its challenges
for labor-intensive agriculture in China: A perspective on farmland transfer plans for farmers'
retirement. Journal of Rural Studies, 100, 103013.
Luo, S., Wang, C., Xi, X., Nie, S., Fan, X., Chen, H., Yang, X., Peng, D., Lin, Y., & Zhou, G.
(2019). Combining hyperspectral imagery and LiDAR pseudo-waveform for predicting crop
LAI, canopy height and above-ground biomass. Ecological Indicators, 102, 801-812.
Malthus, T. J., Andrieu, B., Danson, F. M., Jaggard, K. W., & Steven, M. D. (1993). Candidate
high spectral resolution infrared indices for crop cover. Remote Sensing of Environment,
46(2), 204-212.
Mariye, M., Mariyo, M., Changming, Y., Teffera, Z. L., & Weldegebrial, B. (2022). Effects of land
use and land cover change on soil erosion potential in Berhe district: A case study of Legedadi
watershed, Ethiopia. International Journal of River Basin Management, 20(1), 79-91.
225
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Meena, R. K., Reddy, K. S., Gautam, R., Maddela, S., Reddy, A. R., & Gudipalli, P. (2021).
Improved photosynthetic characteristics correlated with enhanced biomass in a heterotic F 1
hybrid of maize (Zea mays L.). Photosynthesis Research, 147, 253-267.
Meiyan, S., Mengyuan, S., Qizhou, D., Xiaohong, Y., Baoguo, L., & Yuntao, M. (2022).
Estimating the maize above-ground biomass by constructing the tridimensional concept
model based on UAV-based digital and multi-spectral images. Field Crops Research, 282,
108491.
Mgbenka, R., Mbah, E., & Ezeano, C. (2016). A review of smallholder farming in Nigeria: Need
for transformation. International Journal of Agricultural Extension and Rural Development
Studies, 3(2), 43-54.
Miura, T., Yoshioka, H., Fujiwara, K., & Yamamoto, H. (2008). Inter-comparison of ASTER and
MODIS surface reflectance and vegetation index products for synergistic applications to
natural resource monitoring. Sensors, 8(4), 2480-2499.
Mu, C., Li, L., Zhang, F., Li, Y., Xiao, X., Zhao, Q., & Zhang, T. (2018). Impacts of permafrost
on above-and belowground biomass on the northern Qinghai-Tibetan Plateau. Arctic,
Antarctic, and Alpine Research, 50(1), e1447192.
Muruganantham, P., Wibowo, S., Grandhi, S., Samrat, N. H., & Islam, N. (2022). A Systematic
Literature Review on Crop Yield Prediction with Deep Learning and Remote Sensing.
Remote sensing, 14(9), 1990. https://www.mdpi.com/2072-4292/14/9/1990
Ndlovu, H. S., Odindi, J., Sibanda, M., Mutanga, O., Clulow, A., Chimonyo, V. G., & Mabhaudhi,
T. (2021). A comparative estimation of maize leaf water content using machine learning
techniques and unmanned aerial vehicle (UAV)-based proximal and remotely sensed data.
Remote sensing, 13(20), 4091.
Ngoune Tandzi, L., & Mutengwa, C. S. (2019). Estimation of maize (Zea mays L.) yield per harvest
area: Appropriate methods. Agronomy, 10(1), 29.
Niu, Y., Zhang, L., Zhang, H., Han, W., & Peng, X. (2019). Estimating above-ground biomass of
maize using features derived from UAV-based RGB imagery [Article]. Remote Sensing,
11(11), Article 1261. https://doi.org/10.3390/rs11111261
Odebiri, O., Mutanga, O., Odindi, J., Naicker, R., Masemola, C., & Sibanda, M. (2021a). Deep
learning approaches in remote sensing of soil organic carbon: A review of utility, challenges,
and prospects. Environmental Monitoring and Assessment, 193, 1-18.
Odebiri, O., Mutanga, O., Odindi, J., Naicker, R., Masemola, C., & Sibanda, M. (2021b). Deep
learning approaches in remote sensing of soil organic carbon: a review of utility, challenges,
and prospects. Environmental Monitoring and Assessment, 193(12), 802.
https://doi.org/10.1007/s10661-021-09561-6
Olson, D., & Anderson, J. (2021). Review on unmanned aerial vehicles, remote sensors, imagery
processing, and their applications in agriculture. Agronomy Journal, 113(2), 971-992.
Onyango, C. M., Nyaga, J. M., Wetterlind, J., Söderström, M., & Piikki, K. (2021). Precision
agriculture for resource use efficiency in smallholder farming systems in sub-saharan africa:
A systematic review. Sustainability (Switzerland), 13(3), 1158.
Palacios‐Rojas, N., McCulley, L., Kaeppler, M., Titcomb, T. J., Gunaratna, N. S., Lopez‐Ridaura,
S., & Tanumihardjo, S. A. (2020). Mining maize diversity and improving its nutritional
aspects within agro‐food systems. Comprehensive Reviews in Food Science and Food
Safety, 19(4), 1809-1834.
Peter, B. G., Messina, J. P., Carroll, J. W., Zhi, J., Chimonyo, V., Lin, S., & Snapp, S. S. (2020).
Multi-spatial resolution satellite and sUAS imagery for precision agriculture on smallholder
farms in Malawi. Photogrammetric Engineering & Remote Sensing, 86(2), 107-119.
226
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Polzin, M., & Hughes, J. (2023). Automation in steep terrain agriculture: an optimal controller to
prevent tipping and slipping of tethered robots on slopes. Advanced Robotics, 37(15), 987-
998.
Salem, H., Kabeel, A., El-Said, E. M., & Elzeki, O. M. (2022). Predictive modelling for solar
power-driven hybrid desalination system using artificial neural network regression with
Adam optimization. Desalination, 522, 115411.
Salinas-Melgoza, M. A., Skutsch, M., & Lovett, J. C. (2018). Predicting aboveground forest
biomass with topographic variables in human-impacted tropical dry forest landscapes.
Ecosphere, 9(1), e02063. https://doi.org/https://doi.org/10.1002/ecs2.2063
Saranya, T., Deisy, C., Sridevi, S., & Anbananthen, K. S. M. (2023). A comparative study of deep
learning and Internet of Things for precision agriculture. Engineering Applications of
Artificial Intelligence, 122, 106034.
Sharma, P., Leigh, L., Chang, J., Maimaitijiang, M., & Caffé, M. (2022). Above-ground biomass
estimation in oats using UAV remote sensing and machine learning. Sensors, 22(2), 601.
Shi, H., & Xingguo, M. (2011). Interpreting spatial heterogeneity of crop yield with a process
model and remote sensing. Ecological Modelling, 222(14), 2530-2541.
Sun, X., Yang, Z., Su, P., Wei, K., Wang, Z., Yang, C., Wang, C., Qin, M., Xiao, L., & Yang, W.
(2023). Non-destructive monitoring of maize LAI by fusing UAV spectral and textural
features. Frontiers in plant science, 14, 1158837.
Svedin, J. D., Kerry, R., Hansen, N. C., & Hopkins, B. G. (2021). Identifying within-field spatial
and temporal crop water stress to conserve irrigation resources with variable-rate irrigation.
Agronomy, 11(7), 1377.
Tang, Y., Zhou, R., He, P., Yu, M., Zheng, H., Yao, X., Cheng, T., Zhu, Y., Cao, W., & Tian, Y.
(2023). Estimating wheat grain yield by assimilating phenology and LAI with the
WheatGrow model based on theoretical uncertainty of remotely sensed observation.
Agricultural and Forest Meteorology, 339, 109574.
Tollenaar, M., & Lee, E. (2002). Yield potential, yield stability and stress tolerance in maize. Field
Crops Research, 75(2-3), 161-169.
Tripathi, A., Tiwari, R. K., & Tiwari, S. P. (2022). A deep learning multi-layer perceptron and
remote sensing approach for soil health based crop yield estimation. INTERNATIONAL
JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 113,
102959.
Tsai, C.-W., & Fang, Z.-Y. (2021). An effective hyperparameter optimization algorithm for DNN
to predict passengers at a metro station. ACM Transactions on Internet Technology (TOIT),
21(2), 1-24.
Tucker, C. J., Elgin Jr, J., McMurtrey Iii, J., & Fan, C. (1979). Monitoring corn and soybean crop
development with hand-held radiometer spectral data. Remote Sensing of Environment, 8(3),
237-248.
Vélez, S., Martínez-Peña, R., & Castrillo, D. (2023). Beyond Vegetation: A Review Unveiling
Additional Insights into Agriculture and Forestry through the Application of Vegetation
Indices. J, 6(3), 421-436.
Verschuur, J., Li, S., Wolski, P., & Otto, F. E. (2021). Climate change as a driver of food insecurity
in the 2007 Lesotho-South Africa drought. Scientific reports, 11(1), 3852.
Vojnov, B., Jaćimović, G., Šeremešić, S., Pezo, L., Lončar, B., Krstić, Đ., Vujić, S., & Ćupina, B.
(2022). The effects of winter cover crops on maize yield and crop performance in semiarid
conditions—artificial neural network approach. Agronomy, 12(11), 2670.

227
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Wang, F.-M., HUANG, J.-f., TANG, Y.-l., & WANG, X.-z. (2007). New vegetation index and its
application in estimating leaf area index of rice. Rice Science, 14(3), 195-203.
Wang, X., Ren, H., & Wang, A. (2022). Smish: A novel activation function for deep learning
methods. Electronics, 11(4), 540.
Wang, Y., Xiao, Z., & Cao, G. (2022). A convolutional neural network method based on Adam
optimizer with power-exponential learning rate for bearing fault diagnosis. Journal of
Vibroengineering, 24(4), 666-678.
Wu, C., Niu, Z., Tang, Q., & Huang, W. (2008). Estimating chlorophyll content from hyperspectral
vegetation indices: Modeling and validation. Agricultural and Forest Meteorology, 148(8-9),
1230-1241.
Yang, B., Zhu, W., Rezaei, E. E., Li, J., Sun, Z., & Zhang, J. (2022). The optimal phenological
phase of maize for yield prediction with high-frequency UAV remote sensing. Remote
sensing, 14(7), 1559.
Yang, H., Ming, B., Nie, C., Xue, B., Xin, J., Lu, X., Xue, J., Hou, P., Xie, R., & Wang, K. (2022).
Maize Canopy and Leaf Chlorophyll Content Assessment from Leaf Spectral Reflectance:
Estimation and Uncertainty Analysis across Growth Stages and Vertical Distribution.
Remote Sensing, 14(9), 2115.
Yue, J., Yang, H., Yang, G., Fu, Y., Wang, H., & Zhou, C. (2023). Estimating vertically growing
crop above-ground biomass based on UAV remote sensing. Computers and Electronics in
Agriculture, 205, 107627.
Zeng, Y., Hao, D., Huete, A., Dechant, B., Berry, J., Chen, J. M., Joiner, J., Frankenberg, C., Bond-
Lamberty, B., & Ryu, Y. (2022). Optical vegetation indices for monitoring terrestrial
ecosystems globally. Nature Reviews Earth & Environment, 3(7), 477-493.
Zeng, Z., Yao, Y., Liu, Z., & Sun, M. (2022). A deep-learning system bridging molecule structure
and biomedical text with comprehension comparable to human professionals. Nature
communications, 13(1), 862.
Zhai, W., Li, C., Cheng, Q., Mao, B., Li, Z., Li, Y., Ding, F., Qin, S., Fei, S., & Chen, Z. (2023).
Enhancing Wheat Above-Ground Biomass Estimation Using UAV RGB Images and
Machine Learning: Multi-Feature Combinations, Flight Height, and Algorithm Implications.
Remote Sensing, 15(14), 3653.
Zhang, H., Zhao, S., Song, Y., Ge, S., Liu, D., Yang, X., & Wu, K. (2022). A deep learning and
Grad-Cam-based approach for accurate identification of the fall armyworm (Spodoptera
frugiperda) in maize fields. Computers and Electronics in Agriculture, 202, 107440.
Zhang, Y., Xia, C., Zhang, X., Cheng, X., Feng, G., Wang, Y., & Gao, Q. (2021). Estimating the
maize biomass by crop height and narrowband vegetation indices derived from UAV-based
hyperspectral images. Ecological Indicators, 129, 107985.
https://doi.org/https://doi.org/10.1016/j.ecolind.2021.107985

228
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ENHANCING THE ESTIMATION OF EQUIVALENT WATER THICKNESS IN

NEGLECTED AND UNDERUTILIZED TARO CROPS USING UAV ACQUIRED
MULTISPECTRAL THERMAL IMAGE DATA AND INDEX-BASED IMAGE
SEGMENTATION
#11212

1*HelenS. Ndlovu, 1John Odindi, 2Mbulisi Sibanda, 1Onisimo Mutanga

1 Discipline of Geography and Environmental Science, School of Agricultural Earth and

Environmental Sciences, University of KwaZulu-Natal, Scottsville, Pietermaritzburg 3209, South

Africa; 2Department of Geography, Environmental Studies & Tourism, Faculty of Arts,
University of the Western Cape, Bellville, Cape Town 7535, South Africa.
*e-mail: [email protected]

ABSTRACT

Due to the impact of climate variability and change, smallholder farmers are increasingly faced
with the challenge of sustaining crop production. Taro, recognized as a future smart neglected and
underutilized crop due to its resilience to abiotic stresses, has emerged as valuable for diversifying
crop farming systems and sustaining local livelihoods. Nonetheless, a significant research gap
exists in spatially explicit information on the water status of taro, contributing to the paradox of its
ability to adapt to diverse agro-ecological conditions. Precision agriculture, including the use of
unmanned aerial vehicles (UAVs) equipped with high-resolution multispectral and thermal
imagery, has proven effective in farm-scale monitoring and provides near-real-time information on
crop water status. Hence, this study sought to evaluate the utility of multispectral and thermal
infrared UAV imagery in understanding taro’s water status. Leveraging deep learning techniques
to evaluate the use of thermal remote sensing and three index-based segmentation techniques in
predicting the canopy equivalent water thickness (EWT) of taro crops, this study sought to
determine EWT as a proxy to its water status in smallholder farmlands. The study findings illustrate
a significant difference in the prediction accuracies of taro EWT with and without the thermal band
(P < 0.05). Additionally, results (R2 = 0.92, RMSE = 8.04 g/m2, and rRMSE = 15.31% including
the thermal band and 0.91, 8.73 g/m2, and 16.64% excluding the thermal band) reveal the value of
the Excess Green minus Excess Red (ExGR) technique in accurately predicting EWTcanopy.
Furthermore, the near-infrared, red edge, and thermal sections of the electromagnetic spectrum,
together with their derived indices, were critical in estimating taro EWT. This study serves as a
foundation for a robust, efficient, and spatially explicit monitoring framework of neglected and
underutilized crops such as taro. Furthermore, this study offers valuable insights into neglected and
underutilized crop water use within smallholder farming systems, critical for optimizing crop
productivity and mitigating the effects of climatic variability and change.

INTRODUCTION

The world is challenged by the pressing need to sustain food supply and ensure food security due
to climate change and the increasing global population (Din et al., 2022; Hillary Mugiyo et al.,
2021). Recently, driven by water scarcity, there has been increasing interest in the potential use of
neglected and underutilised crop species (NUS) in addressing food and nutrition challenges
(Chivenge et al., 2015; Mabhaudhi et al., 2017; Hillary Mugiyo et al., 2021). NUS, characterised

229
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

by historical domestication with limited scientific research and predominantly confined to

smallholder farming systems, have emerged as key drought-tolerant crops for diversifying
communal cropping systems (Chivenge et al., 2015; Mabhaudhi et al., 2017). Taro (Colocasia
esculenta (L)) is one of the oldest and most widely cultivated NUS crops in the world’s tropical
and subtropical regions (Mabhaudhi et al., 2011; Mawoyo et al., 2017; Van Wyk, 2021). In South
Africa, taro, locally known as amadumbe, is known to be heat tolerant and primarily cultivated for
subsistence, especially within small and marginalised communities (Joshi et al., 2020; Mabhaudhi
et al., 2014; Oyeyinka & Amonsou, 2020; Van Wyk, 2021). Taro is identified as a future smart
food under the NUS category because of its edible tubers, which are rich in carbohydrates, protein,
and vitamins (Kapoor et al., 2022; Li & Siddique, 2018). Despite taro and indeed other NUS’s
value, literature shows that they have largely been ignored.

Thermal infrared remote sensing has emerged as a valuable tool for crop water assessment and
monitoring, offering a direct correlation with crop water biophysical and biochemical elements
(Khanal et al., 2017; Messina & Modica, 2020). The recent advancements in image acquisition,
like unmanned aerial vehicles (UAVs) mounted with light-weight multispectral sensors provide,
spatially explicit near-real-time information on crop water status (Hussain et al., 2020). In addition
to ultra-high spatial resolutions of UAV multispectral thermal imagery, image enhancement
techniques and robust algorithms have been demonstrated to improve model accuracies. For
instance, Index-Based Image Segmentation has been demonstrated to be effective in robustly
segmenting plants in colour images, enabling the extraction of vegetation cover and removing soil
background for enhanced crops spectral signatures (Hamuda et al., 2016). (Lu et al., 2022). The
Excess Green (ExG) and Excess Red (ExR) indices were proposed by Woebbecke et al. (1995) and
Meyer et al. (1999), respectively, to enhance plant segmentation accuracy by emphasizing plant
greenness by accounting for the relative proportions of red and physiological green. Additionally,
Meyer and Neto (2008) leveraged the strength of both ExG and ExR to develop the Excess Green
minus Excess Red (ExGR) index to improve crop water assessment and monitoring using thermal
remote sensing systems.

In this regard, leveraging the capabilities of deep learning, this study sought to assess the
performance of thermal remote sensing and index-based segmentation techniques in improving
canopy EWT estimation of smallholder taro crops using UAV multispectral thermal imagery.
Specifically, the study sought to: (1) assess the potential of the UAV thermal band in estimating
EWT of smallholder taro, (2) compare the performance of crop canopy images extracted using
the ExG, ExR, and ExGR color indices in improving EWT estimations of taro crop, and (3)
evaluate the potential of UAV multispectral thermal imagery in EWT estimations of taro crop in
smallholder farming systems.

MATERIALS AND METHODS

Study area description and the experimental field

This research was conducted in the rural community of Swayimana, situated within the uMshwathi
Municipality, northeast of Pietermaritzburg city, KwaZulu-Natal, South Africa (29°31′ 24’’ S;
30°41′ 37’’ E) (Figure 1).

The taro experimental plot was cultivated during the early rainy season, aligning with its optimal
growing conditions. The selected plot covered 2864.56 m2 and was rainfed. The taro crop was sown
230
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

in mid-October 2022 and was approximately 171 days old at the time of the experiment.
Specifically, the crop was intermediate between the late vegetative and early maturity growth
stages. The selection of this growth stage is crucial for capturing the developmental dynamics of
the crop during a period of heightened canopy growth, providing valuable insights for the research
objectives.

Figure 1. a) Location of experimental field in Swayimane and b) taro crops.

Field Sampling and in-situ measurements

A polygon delineating the taro field was created using Google Earth Pro and imported into ArcMap
10.6 to facilitate the generation of 100 stratified random sampling points. This approach was
adopted to ensure variability and accurate representation of all taro crops within the field. These
sampling points were subsequently uploaded into a Trimble handheld Global Positioning System
(GPS) with a sub-centimetre accuracy, enabling precise location of each sampling point within the
taro field. In-situ measurements were obtained at each sampling point to compute respective EWT
values.

A portable LiCOR-2200C Plant Canopy Analyzer was used to obtain the leaf area index (LAI) of
the crops. The LAI measurements were obtained using the 38° zenith angle with a 270° view cap
and the ABBBB sequence, where A corresponds to a reference reading ‘above’ the canopy and B
corresponds to a reading ‘below’ the canopy. Thereafter, the above ground biomass of each
sampled crop was obtained, and the fresh weight (FW) obtained using a calibrated scale with a 0.5
g measurement error. The sampled biomass was then placed in a labelled brown paper bag and
dried in an oven at 60 °C, until a constant dry weight (DW) was reached (approximately 72 hrs).

UAV platform and multispectral-thermal camera

The DJI Matrice 300 (M300) platform, mounted with a MicaSense Altum camera and
Downwelling Light Sensor 2 (DLS 2) was used to collect multispectral-thermal imagery. A total
of 1626 raw images of the experimental field were obtained and pre-processed in Pix4D
photogrammetry software. Ground reference points surveyed prior to fieldwork were then used to

231
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

improve the geometric accuracy of the acquired images in ArcGIS 10.6. Lastly, EWTcanopy in-situ
measurements and the locational of each sampled taro point were overlaid with UAV multispectral-
thermal image. The multispectral and thermal reflectance data of taro was extracted from the UAV
imagery and used to derive vegetation indices (VIs) for the development of the EWTcanopy
prediction model. These VIs were selected based on their optimal performance in literature and
relationship with crop water status (Baluja et al., 2012; Ozelkan, 2020; Zhang & Zhou, 2019).

Index-based image segmentation of taro crops' spectral signatures

To delineate the crop canopy and eliminate soil background from the multispectral thermal image,
an index-based segmentation technique was employed. Specifically, The Excess Green (ExG),
Excess Red (ExR), and Excess Green minus Excess Red (ExGR) color indices were computed
using the green, red, and blue bands of the UAV multispectral thermal imagery (Hamuda et al.,
2016; Meyer et al., 1999; Meyer & Neto, 2008; Woebbecke et al., 1995). Finally, the threshold
method was used to generate a binary image from the gray-level histograms obtained during the
index-based segmentation process (Shu et al., 2021).

Model development and statistical analysis

In this study, we employed a deep machine learning approach to estimate EWT canopy using UAV
derived multispectral optical and thermal datasets. The study utilised a three-layer neural network
model consisting of an input layer, a hidden layer, and an output layer. A rectified linear unit
(ReLU) was applied to stimulate the EWTcanopy prediction model with the maximum epochs set to
200 interactions, indicating that the weights in the hidden layers were iteratively adjusted 200 times
to reduce error and enhance EWTcanopy prediction accuracy. Thereafter, the SoftMax activation
function was used to transform the raw outputs of the neural network into a vector of probabilities,
and the Adaptive moment estimation (Adam) optimizer was used to optimise the results of the
output model. Furthermore, the dropout regularization technique was applied to avoid overfitting
and improve the generalization of the model (Deepan & Sudha, 2020). The hyperparameters of the
DNN model were tuned to a learning rate of 0.001, batch size of 32 and an input and hidden layer
dropout of 0.4 and 0.2, respectively.

RESULTS AND DISCUSSION

Performance of the thermal band in predicting EWTcanopy of taro crops

The performance of the thermal band in predicting EWTcanopy of taro crops revealed a consistent
trend across the various index-based segmentation techniques. It was observed that the exclusion
of the thermal band in EWTcanopy analysis resulted in lower estimation accuracies (P < 0.05),
emphasising the importance of the thermal band in characterising taro crop water status.
Surprisingly, our study found no significant difference in prediction accuracies when thermal data
was considered in comparison to its exclusion in the ExGR-based model. These results underscore
the effectiveness of the ExGR-based technique, particularly its robust performance irrespective of
the inclusion and exclusion of the thermal channel.

Additionally, it was observed that the thermal band was among the topmost predictor variable
across all EWTcanopy models. Literature confirms the invaluable role of thermal infrared remote
sensing in assessing and monitoring crop water status, establishing a direct correlation with crop
water biophysical and biochemical elements (Khanal et al. 2017, Messina and Modica 2020,
Krishna et al. 2021). The use of thermal remote sensing is based on the premise that thermal
232
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

characteristics of crop leaves are effected by leaf transpiration, which decreases in a state of water
deficit, resultantly reducing leaf and canopy temperatures (Maes and Steppe 2012, Gerhards et al.
2019). The findings of this study align with a recent study by Guan and Grote (2023), which
achieved an R2 of 0.74 when incorporating the thermal channel, compared to an R2 of 0.63 with
the thermal band excluded, highlighting the integration of multispectral and thermal data and its
combined value in understanding crop water status. The findings of this study are further
corroborated by those of García-Tejero et al. (2018) who concluded that the thermal band is feasible
for monitoring almond water stress for irrigation scheduling, and Cheng et al. (2023) who
highlighted the applicability of thermal imaging in assessing the crop water conditions of summer
maize crop.

Performance of index-based segmentation techniques for the estimation of taro EWTcanopy

This study shows that the inclusion of soil background reduces the accuracy of EWT canopy
predictions within taro crop (R2 of 0.61, RMSE of 25.35 g/m2, and rRMSE of 43.87%). It was
noted that the prediction accuracy of taro EWTcanopy improved significantly after the removal of
soil background through the ExGR-based image segmentation technique, yielding an optimal R2
of 0.92, RMSE of 8.04 g/m2 and rRMSE of 15.31. These results align with the broader consensus
in the literature. Xu et al. (2021) and Li et al. (2022) for instance emphasized the challenge posed
by soil background in influencing crop canopy spectra, particularly in UAV-derived imagery.
Notably, while the ExG and ExR techniques demonstrated acceptable accuracy in quantifying taro
EWTcanopy (R2 of 0.90, and R2 of 0.76, respectively), the ExGR method outperformed both these
techniques. This notable enhancement can be attributed to the inherent capabilities of the ExGR
technique in effectively mitigating soil background interference (Zhai et al. 2023). The
comprehensive nature of the ExGR method combines the advantages of both the ExG and ExR by
simultaneously leveraging ExG for extracting the crop canopy and ExR for eliminating background
noise (Meyer et al. 2004, Hamuda et al. 2016, Riehle et al. 2020, Upendar et al. 2021).

Overall, the removal of soil background has proven imperative for enhancing the accuracy of taro
EWTcanopy predictions. These findings are further supported by Shu et al. (2021) that reported a
significant increase in prediction accuracy from R2 of 0.45 and RMSE of 7.13 before to an R2 of
0.74 and RMSE of 3.68 after performing soil background removal in estimating the SPAD
chlorophyll content of a maize crop. These parallel findings underscore the significance of
addressing soil background interference for accurate and reliable estimations in crop water-related
assessments.

REFERENCES

Baluja, J., Diago, M. P., Balda, P., Zorer, R., Meggio, F., Morales, F., & Tardaguila, J. (2012).
Assessment of vineyard water status variability by thermal and multispectral imagery using
an unmanned aerial vehicle (UAV). Irrigation Science, 30, 511-522.
Chivenge, P., Mabhaudhi, T., Modi, A. T., & Mafongoya, P. (2015). The potential role of neglected
and underutilised crop species as future crops under water scarce conditions in Sub-Saharan
Africa. International journal of environmental research and public health, 12(6), 5685-5711.
https://mdpi-res.com/d_attachment/ijerph/ijerph-12-05685/article_deploy/ijerph-12-
05685.pdf?version=1432635125

233
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Deepan, P., & Sudha, L. (2020). Object classification of remote sensing image using deep
convolutional neural network. In The cognitive approach in cloud computing and internet of
things technologies for surveillance tracking systems (pp. 107-120). Elsevier.
Din, M. S. U., Mubeen, M., Hussain, S., Ahmad, A., Hussain, N., Ali, M. A., Sabagh, A. E.,
Elsabagh, M., Shah, G. M., & Qaisrani, S. A. (2022). World nations priorities on climate
change and food security. In Building Climate Resilience in Agriculture (pp. 365-384).
Springer.
Hamuda, E., Glavin, M., & Jones, E. (2016). A survey of image processing techniques for plant
extraction and segmentation in the field. Computers and Electronics in Agriculture, 125, 184-
199.
Hilary van Wyk, R., & Oscar Amonsou, E. (2021). Physiochemical and functional properties of
albumin and globulin from amadumbe (Colocasia esculenta) corms. Food Science and
Technology.
Hlophe-Ginindza, S. N., & Mpandeli, N. (2020). The role of small-scale farmers in ensuring food
security in Africa. Food Security in Africa, 1-12.
Hussain, S., Gao, K., Din, M., Gao, Y., Shi, Z., & Wang, S. (2020). Assessment of UAV-Onboard
Multispectral Sensor for non-destructive site-specific rapeseed crop phenotype variable at
different phenological stages and resolutions. Remote Sensing, 12(3), 397.
Joshi, B. K., Shrestha, R., Gauchan, D., & Shrestha, A. (2020). Neglected, underutilized, and future
smart crop species in Nepal. Journal of Crop Improvement, 34(3), 291-313.
Kapari, M., Hlophe-Ginindza, S., Nhamo, L., & Mpandeli, S. (2023). Contribution of smallholder
farmers to food security and opportunities for resilient farming systems. Frontiers in
Sustainable Food Systems, 7, 1149854.
Kapoor, B., Singh, S., & Kumar, P. (2022). Taro (Colocasia esculenta): Zero wastage orphan food
crop for food and nutritional security. South African Journal of Botany, 145, 157-169.
Khanal, S., Fulton, J., & Shearer, S. (2017). An overview of current and potential applications of
thermal remote sensing in precision agriculture. Computers and Electronics in Agriculture,
139, 22-32.
Li, X., & Siddique, K. H. (2018). Future smart food. Rediscovering hidden treasures of neglected
and underutilized species for Zero Hunger in Asia, Bangkok.
Lu, Y., Young, S., Wang, H., & Wijewardane, N. (2022). Robust plant segmentation of color
images based on image contrast optimization. Computers and Electronics in Agriculture, 193,
106711.
Mabhaudhi, T., Chimonyo, V. G., & Modi, A. T. (2017). Status of underutilised crops in South
Africa: Opportunities for developing research capacity. Sustainability, 9(9), 1569.
Mabhaudhi, T., Modi, A., & Beletse, Y. (2011). Growth response of selected taro [Colocasia
esculenta (L.) schott] landraces to water stress. II International Symposium on Underutilized
Plant Species: Crops for the Future-Beyond Food Security 979,
Mabhaudhi, T., Modi, A. T., & Beletse, Y. G. (2014). Parameterisation and evaluation of the FAO-
AquaCrop model for a South African taro (Colocasia esculenta L. Schott) landrace.
Agricultural and Forest Meteorology, 192, 132-139.
Mawoyo, B., Adebola, P., Gerrano, A. S., & Amonsou, E. (2017). Effect of genotypes and growth
locations on composition and functional properties of amadumbe flours. Journal of Food
Science and Technology, 54(11), 3577-3586.
Messina, G., & Modica, G. (2020). Applications of UAV thermal imagery in precision agriculture:
State of the art and future research outlook. Remote Sensing, 12(9), 1491.

234
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Meyer, G. E., Hindman, T. W., & Laksmi, K. (1999). Machine vision detection parameters for
plant species identification. Precision agriculture and biological quality,
Meyer, G. E., & Neto, J. C. (2008). Verification of color vegetation indices for automated crop
imaging applications. Computers and Electronics in Agriculture, 63(2), 282-293.
Mugiyo, H., Chimonyo, V., Sibanda, M., Kunz, R., Masemola, C., Modi, A., & Mabhaudhi, T.
(2021). Evaluation of land suitability methods with reference to neglected and underutilised
crop species: A scoping review. Land, 10(2), 125.
Mugiyo, H., Chimonyo, V. G., Sibanda, M., Kunz, R., Nhamo, L., Masemola, C. R., Dalin, C.,
Modi, A. T., & Mabhaudhi, T. (2021). Multi-criteria suitability analysis for neglected and
underutilised crop species in South Africa. PLoS One, 16(1), e0244734.
Oyeyinka, S. A., & Amonsou, E. O. (2020). Composition, pasting and thermal properties of flour
and starch derived from amadumbe with different corm sizes. Journal of Food Science and
Technology, 57(10), 3688-3695.
Ozelkan, E. (2020). Water body detection analysis using NDWI indices derived from landsat-8
OLI. Polish Journal of Environmental Studies, 29(2), 1759-1769.
Shu, M., Zuo, J., Shen, M., Yin, P., Wang, M., Yang, X., Tang, J., Li, B., & Ma, Y. (2021).
Improving the estimation accuracy of SPAD values for maize leaves by removing UAV
hyperspectral image backgrounds. International Journal of Remote Sensing, 42(15), 5862-
5881.
Van Wyk, H. R. A., O E. (2021). Physiochemical and functional properties of albumin and globulin
from amadumbe (Colocasia esculenta) corms. Food Science and Technology.
Woebbecke, D. M., Meyer, G. E., Von Bargen, K., & Mortensen, D. A. (1995). Color indices for
weed identification under various soil, residue, and lighting conditions. Transactions of the
ASAE, 38(1), 259-269.
Zhang, F., & Zhou, G. (2019). Estimation of vegetation water content using hyperspectral
vegetation indices: A comparison of crop water indicators in response to water stress
treatments for summer maize. BMC ecology, 19(1), 1-12.
Zulu, N. N. (2022). Water Scarcity and Household Food Security: A Case of Ulundi Local
Municipality in KwaZulu-Natal, South Africa. In Handbook of Research on Resource
Management and the Struggle for Water Sustainability in Africa (pp. 127-148). IGI Global.

235
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

SUSTAINABILITY OF MAIZE PRODUCTION WITH FARMERS’ PRACTICES AND

REDESIGNED CROP MANAGEMENT PRACTICES IN CRV AND JIMMA OF
ETHIOPIA
#11310

1*Workneh Bekere Kenea, 1Amsalu Nebiyu, 2Tesfaye Balemi

1Departmentof Horticulture and Plant Sciences, Jimma University, P.O. Box 307, Jimma,
Ethiopia; Crop Production Specialist, Food and Agriculture Organization, Semera Field Office,
2

Semera, Ethiopia
*e-mail: [email protected], +251-921235304

ABSTRACT

In this study, sustainability of maize production with farmer’s practice (FP), redesigned plant
density plus current fertilizer use (RDCF), current plant density plus redesigned fertilizer use
(CDRF), and redesigned plant density plus redesigned fertilizer use (RDRF) were assessed at
household level based on social, agronomic, economic and environmental principles. Farmers’
preference, farm household (maize) grain self-sufficiency, gross margin, and nitrogen use
efficiency were used as indicators for for the respective principles. The result revealed that the
preference of RDRF was 95% and 100% in CRV and Jimma respectively whereas the preference
of FP was 45% and 20% in the respective regions. With all production technologies, farmers in
both regions could achieve their family grain self-sufficiency but surplus production varied with
the technologies used for the production. In CRV, RDRF, RDCF, and CDRF were economically
viable to fertilizer use. However, in Jimma, less than 50% of farm households profited from RDRF
and CDRF production technologies. In CRV, maize production using all crop management
practices was associated with soil mining whereas in Jimma, the use of RDRF technology resulted
in 18% environmental sustainability. Based on (average) scores of the indicators, maize production
with RDRF showed higher social and economic sustainability in CRV and Jimma whereas maize
production with FP was the least sustainable maize system in both regions. We conclude that
further redesigning maize management technologies that meet environmental goal is of paramount
important for the current and future generations in CRV and Jimma, Ethiopia.

Keywords: Farm household, family grain self-sufficiency, weighted scores, N use efficiency

INTRODUCTION

The concept of sustainability was first used in Sweden in forest management sector (Wiersum,
1995) with the focus on sustainable timber production from sustainable tree production (Prins et
al., 2023). Since then, it has been applied in many disciplines across the globe. Sustainability in
agriculture refers to the ability of a farm or agricultural system to produce food, fiber, or other
products indefinitely without damaging or depleting the resources pool on which it depends. In
other words, it is fulfilling the need of the present generation without compromising the ability of
future generations to do the same (Rasmussen et al., 2017;Erbaugh et al., 2019). Several studies
have assessed sustainability of farming systems based on sustainability dimensions (Ezell et al.,
2021). This enables the selection of systems (practices) that fulfill social (Mandipaza, 2022),
agronomic, economic and environmental principles. This means that the technologies should be

236
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

socially acceptable, increase production, economically viable and environmentally safe (Pretty,
2008). Principles are universal ambitious commitments that are explained in terms of indicators.
As principles are not directly measurable parameters, indicators are often used. Whereas indicator
has been defined by different scholars, the definitions have similar meaning. Indicators are
variables that supply information about another variable that are difficult to measure (Büchi et al.,
2019;Heink and Kowarik, 2010). Alternatively, indicators are defined as variables for which a
quantitative value is determined and compared to a reference value (Girardin et al., 1999).
Indicators are very helpful in decision making for policy makers and managers.

Despite the inconsistency of the frameworks and tools, sustainability of farming system has been
well studied (Paas et al., 2021) in developed countries. However, such studies are rare in Ethiopian
agriculture, especially in staple food crops such as maize. Most studies are focusing on options of
yield increase (Srivastava et al., 2019) and other studies unraveled factors associated with yield
increase (Abate et al., 2015). This study is therefore conducted to (1) to assess the social,
agronomic, economic and environmental performance of technologies used for maize production,
(2) to assess the grain self-sufficiency at household level and estimate the land area that is required
for farm household grain self-sufficiency in CRV and Jimma, Ethiopia.

MATERIALS AND METHODS

Definitions of maize production technologies

1. Farmer practices (FP): It is the actual plant density and actual fertilizer amount used by
farmers.
2. Current density and redesigned fertilizer (CDRF): It is the average plant density used by
farmers whereas fertilizer use is re-designed based on 50% potential yield of the crop.
3. Redesigned density and current fertilizer (RDCF): In this production practice, plant density
is redesigned (53,333 plant ha-1 in CRV and 62,000 plant ha-1 in Jimma) but the fertilizer
use is the average rate used by farmers in the respective region.
4. Redesigned density and redesigned fertilizer (RDRF): This refers to a production practices
in which both plant density and fertilizer uses are redesigned.

Household grain self-sufficiency

Total maize production (t) at household level was obtained from the product of maize cultivation
area (ha) and maize yield (t/ha). Maize cultivation areas of the participated farms households were
measured using a handheld geographic positioning system essential (area measure) for the
participating farm households in 2017. The per capita maize requirement of the farm households
was based on an adult male equivalent (AME) of the 2017 family composition. Accordingly, a
female was equivalent to 0.82 AME whereas children (0–18 years old) were equivalent to 0.75
AME (FAO, 2001). The per capita maize requirement is 260 kg -1 adult-1 year-1 based on calorie
content of the crop.

Technology preference
Participating farmers were invited to evaluate the performance of maize crop grown with each
technology at crop maturity, moving from field to field. Simple questionnaires were prepared to
assess their preference of the tested technologies (Fertilizer use and plant density). A common
starting question was “Is this fertilizer use and plant density important to improve your livelihood
by improving maize productivity?”. The farmers answered as no, partly or yes. To evaluate this
237
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

response quantitatively and express in percentage (Equ. 1), the no, partly and yes options were
assigned with 0, 0.5 and 1 numerical values (Sannou et al., 2023) respectively.

∑𝑖1 𝑤𝑖 × 𝑥𝑖
𝑥̅ (%) = ( × 100) equ 1
∑𝑛𝑖 𝑋

Gross margin at farm level

To assess economic sustainability of maize production, input cost (fertilizer and seed cost) and
maize price during the growing period were documented separately. The amount of fertilizer (kg
farm-1) used at farm level was multiplied with fertilizer cost (ETB kg-1) during the study period.
Following similar procedure, the amount of maize seed (kg household-1) used for planting for was
multiplied with the seed cost (ETB kg-1). Similarly, the amount of maize yield (kg household-1)
was multiplied with market price of the maize grain (ETB kg-1). The value cost ratio was estimated
from the ratio of monetary value of maize from each technology to the cost of the technology
(fertilizer cost plus seed cost) for maize production by that technology.

Nitrogen use efficiency

The impact of each technology on the environment was assessed using nitrogen (N) use efficiency
of the maize production system as an indicator. Nitrogen use efficiency in the system is given by
the ratio of N output from the system to N input into the system (Marinus et al., 2023). The
assessment of N use efficiency was based on the EUNEP (2015) framework. Based on the
framework, the minimum and maximum N use efficiencies are 50% and 90% respectively and the
maximum N surplus per season is 80 kg ha-1. An N use efficiency below 50% or N surplus greater
than 80 kgha-1 indicates high risk of N loss to the environment whereas an N use efficiency greater
than 90% is associated with high risk of soil mining. According to this framework, the desired N
output is 80 kg ha-1 N and this value was adjusted to 50% of the maize yield potential in this study
(Marinus et al., 2023). For the FP yields, N output was estimated following (Njoroge et al., 2019)
and the N removed in this case is 1.54% of the maize grain yield. The N input was the amount of
N applied by the farmer for the FP and the N applied for the treatments for the on-farm experiments.

RESULTS AND DISCUSSION

Technology preference
The preference of FP, RDCF, CDRF and RDRF technologies by smallholder farmers in CRV and
Jimma was presented by Table 1. Almost all farms in both regions fully preferred RDRF
technology in both regions (Table 1). However, FP was the lowest preferred technology in CRV
(45%) and Jimma (20%). The preference of CDRF and RDCF was modest.

Family grain self-sufficiency

All farm households can attain their grain self-sufficiency and even produce surplus maize grain
under all production technologies in CRV and Jimma. Whereas the average maize production of
the farm household in CRV was 7594.5 kg per season, the production ranged from 420 kg under
FP to 22200 kg per household under redesigned planting density and redesigned fertilizer use
(RDRF). However, the grain requirement of the household in this region was found to be 1024
kg/household/year to 5120 kg/house/year based on the family size. On average, a farm household
in CRV requires 2444 kg of maize grain per year to be grain self-sufficient. The grain requirement

238
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

of farm households in Jimma ranged from 1024 kg to 2304 kg year-1 during 2017 and 2018.
However, the maize grain produced during 2017 and 2018 in Jimma ranged from 1313 kg year -1 to
30369 kg year-1.

Table 1. Preference of maize agronomic practices by smallholder farmers in CRV and Jimma of
Ethiopian.

Agronomic practices Regions

CRV Jimma
FP 45% 20%
RDCF 70% 65%
CDRF 65% 50%
RDRF 95% 100%

Figure 1. Proportion of maize grain for family self-sufficiency (blue bar) and extra production (red
bar) under various crop management technologies in CRV (a) and Jimma (b). FP, CDRF, RDCF
and RDRF represent farmer’s practices, current density plus redesigned fertilizer, redesigned
density plus current fertilizer and redesigned density plus redesigned fertilizer.

Marginal return
In Jimma, CDRF was economically viable for 54% farm households. In CRV, 91% of the farm
households were profited from CDRF production technology. In this region, using RDRF has the
same profitability (91%) with RDCF crop management technology. With the use of FP, however,
only 18% of the farm households were profited. Based on average value cost ratio, RDCF, CDRF
and RDRF were economically viable in CRV. In Jimma however, CDRF was not profitable
whereas RDCF and RDRF of maize management practices were profitable at farm household level.
In both regions, FP crop management was not economically viable at household level during.

Nitrogen use efficiency

Fig. 2 showed the N use efficiency in maze production with various maize crop management
practices in CRV and Jimma. The N output outweigh the N input in various crop management
technologies in both regions implying that all FP, CDRF, and RDCF technologies were associated

239
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

with soil mining that put sustainability of maize production at risk. In CRV, none of the N use
efficiency was in between a desirable range whereas in Jimma only 18% of the N use efficiency
was in a desirable range. Below 21 kg ha-1 N input in CRV and below 80 kg ha-1 N input in Jimma,
N use efficiency was associated with soil mining, putting the sustainability of maize production on
the knife edge. In addition, maize production system in CRV did not result in risk of environmental
pollution with surplus N whereas in Jimma, environmental risk from surplus N in maize production
could be as high as 20%.

Table 2. Value cost ratio and percentage of profited households from maize agronomic practices
in CRV and Jimma in 2017 and 2018.

Production technologies
Region parameters
FP CDRF RDCF RDRF
Average -1.9 3.8 5.0 4.3
CRV
Profitable households (%) 18 73 64 91
Average -0.62 0.56 1.82 1.34
Jimma
Profitable households (%) 23 23 54 54
FP, CDRF, RDCF and RDRF represent farmers practices, current density plus redesigned fertilizer,
redesigned density plus current fertilizer and redesigned density plus redesigned fertilizer.

Figure 2. Actual and desired N use efficiencies in maize production in CRV and Jimma, in Ethiopia. The
upper and the lower diagonal lines with a y-intercept of zero (red solid line) indicate an N use efficiency of
90% and 50% respectively. An N use efficiency above 90% shows a risk of soil N mining (pink color),
while an N use efficiency below 50% indicates a risk of N losses to the environment (orange color). The
space between these 50% and 90% N use efficiencies is further narrowed down (light yellow color); by a
dotted diagonal line (blue) indicating a maximum N surplus of 80 kg N ha −1, which, if exceeded, indicates
a risk of N losses to the environment and a horizontal dotted line (black) indicating a targeted N output,
which is equivalent to 50% of the water-limited yield potential per season (48 kg N ha−1 for CRV and 115
kg N ha−1 for Jimma) below this line indicate low productivity (light yellow). The remaining white area
indicates the desired range of N use efficiencies.

240
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

The performances of maize production technologies in CRV and Jimma with all indicators were
presented with Fig 3. Indicators with high scores were more sustainable except yield gap and land
area required for grain self-sufficiency. The high score (percentage) of the yield gap and land area
required for grain self-sufficiency was associated with low sustainability of the production system.
In this line, the score of existing yield gap and land area for FP was higher in CRV and Jimma
showing the low sustainability of the production system in this region.

Figure 3. Performance of maize production technologies with average scores per indicator for
farms in CRV and Jimma, Ethiopia. TP, Land_GSS, Yg, NUE stands for technology preference,
land required for grain self-sufficiency, yield gap and nitrogen use efficiency whereas M_return,
grain_SS and RN refer to marginal return grain self-sufficiency and return to N. All indicators
except yield gap were assessed at farm household level. Yield gap was assessed at field level. FP,
RDCF, CDRF and RDRF refer to farmer’s practice, redesigned density with current fertilizer,
current density with redesigned fertilizer and redesigned density with redesigned fertilizer.

REFERENCES

Abate, T., Shiferaw, B., Menkir, A., Wegary, D., Kebede, Y., Tesfaye, K., Kassie, M., Bogale, G.,
Tadesse, B. & Keno, T. 2015. Factors that transformed maize productivity in Ethiopia. Food
Security, 7, 965-981.
Büchi, L., Georges, F., Walder, F., Banerjee, S., Keller, T., Six, J., Van Der Heijden, M. & Charles,
R. 2019. Potential of indicators to unveil the hidden side of cropping system classification:

241
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Differences and similarities in cropping practices between conventional, no-till and organic
systems. European Journal of Agronomy, 109, 125920.
Erbaugh, J., Bierbaum, R., Castilleja, G., Da Fonseca, G. A. & Hansen, S. C. B. 2019. Toward
sustainable agriculture in the tropics. World Development, 121, 158-162.
EUNEP, 2015. Nitrogen Use Efficiency (Schut et al.) - an indicator for the utilization of nitrogen
in agriculture and food systems. Wageningen University, Alterra, Wageningen, the
Netherlands.
Ezell, B., Lynch, C. J. & Hester, P. T. 2021. Methods for weighting decisions to assist modelers
and decision analysts: A review of ratio assignment and approximate techniques. Applied
Sciences, 11, 10397.
FAO/WHO/UNU (2001). Human energy requirements. Report of a joint FAO/WHO/UNU expert
consultation, Rome
Girardin, P., Bockstaller, C. & Werf, H. V. D. 1999. Indicators: tools to evaluate the environmental
impacts of farming systems. Journal of sustainable agriculture, 13, 5-21.
Heink, U. & Kowarik, I. 2010. What are indicators? On the definition of indicators in ecology and
environmental planning. Ecological indicators, 10, 584-593.
Mandipaza, B. F. 2022. Socio-cultural dimensions of farming, small farm households and
conservation agriculture in Nyanga District, Zimbabwe. University of Pretoria.
Marinus, W., Van De Ven, G. W. J., Descheemaeker, K., Vanlauwe, B. & Giller, K. E. 2023.
Farmer responses to an input subsidy and co-learning program: intensification,
extensification, specialization, and diversification? Agronomy for Sustainable Development,
43, 40.
Njoroge, S., Schut, A. G. T., Giller, K. E. & Zingore, S. 2019. Learning from the soil’s memory:
Tailoring of fertilizer application based on past manure applications increases fertilizer use
efficiency and crop productivity on Kenyan smallholder farms. European Journal of
Agronomy, 105, 52-61.
Paas, W., San Martín, C., Soriano, B., Van Ittersum, M. K., Meuwissen, M. P. & Reidsma, P. 2021.
Assessing future sustainability and resilience of farming systems with a participatory
method: A case study on extensive sheep farming in Huesca, Spain. Ecological Indicators,
132, 108236.
Pretty, J. 2008. Agricultural sustainability: concepts, principles and evidence. Philosophical
Transactions of the Royal Society B: Biological Sciences, 363, 447-465.
Prins, K., Köhl, M. & Linser, S. 2023. Is the concept of sustainable forest management still fit for
purpose? Forest Policy and Economics, 157, 103072.
Rasmussen, L. V., Bierbaum, R., Oldekop, J. A. & Agrawal, A. 2017. Bridging the practitioner-
researcher divide: Indicators to track environmental, economic, and sociocultural
sustainability of agricultural commodity production. Global Environmental Change, 42, 33-
46.
Sannou, R. O., Kirschke, S. & Ghünther, E. 2023. Integrating the social perspective into the
sustainability assessment of agri-food systems: A review of indicators. Sustainable
Production and Consumption.
Srivastava, A. K., Mboh, C. M., Faye, B., Gaiser, T., Kuhn, A., Ermias, E. & Ewert, F. 2019.
Options for sustainable intensification of maize production in Ethiopia. Sustainability, 11,
1707.
Wiersum, K. F. 1995. 200 years of sustainability in forestry: Lessons from history. Environmental
Management, 19, 321-329.

242
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

OPTIMIZING MAIZE PRODUCTION THROUGH MINIMUM TILLAGE IN

CONSERVATION AGRICULTURE (CA) SYSTEMS: EVIDENCE FROM MALAWI'S
LOWLANDS
#11670

1Donald
Nyamayevuab, 2Isaiah Nyagumbob, 1Liang Wei-lia, 1Li Rui-qia
1Collegeof Agronomy, Hebei Agriculture University, 289 Lingyusi Street, Baoding, Hebei,
China; International Maize and Wheat Improvement Center (CIMMYT), Box MP163, Mt
2

Pleasant, 12.5 Km Peg, Mazowe Road, Harare, Zimbabwe

e-mail: [email protected]

ABSTRACT

Sustainable intensification in agricultural systems has been implemented and promoted across Sub-
Saharan Africa (SSA) as a strategy for addressing low crop productivity often resulting in
widespread food and nutritional insecurity. This study sought to assess the productivity potential
of conservation agriculture (CA) cropping systems and associated crop establishment techniques.
An on-station study was conducted at Chitala research station in Malawi. Maize grain productivity
varied with crop establishment technique and cropping systems. Planting basins showed better
performance during seasons with low to moderate wetness, and intervening less rainy months as
observed in 2014/15 and 2015/16 cropping seasons. Conversely, direct seeding techniques with
less soil surface disruption (dibble stick and Jab planter) performed better during seasons of high
and persistent rainfall (2016/17 and 2017/18), with totals exceeding 800 mm. Rotation systems,
particularly maize groundnut, outperformed other systems in maize grain yield, while intercropping
systems incurred higher grain yield penalties among the tested systems. These results confirm
previous findings on CA, indicating that rotating maize with legumes boosts maize grain yield,
while maize-legume intercropping may reduce it.

INTRODUCTION

In SSA rainfall anomalies often lead to water stress, low crop yields accompanied by large yield
gaps (Ligowe et al., 2017; Nyagumbo et al., 2020). This has resulted in widespread poverty, food
insecurity and malnutrition (Makate et al., 2018). In effort to reduce the negative impacts of these
challenges, conservation agriculture (CA) has gained significant attention across smallholder
farmers, it has been promoted as a potential sustainable agricultural intensification technology in
response to food insecurity and the adverse effects of climate ( Omulo et al., 2024). Its principles
hinge on reducing soil disturbance, crop diversification, and permanent soil cover (Mupangwa et
al., 2021).

Minimum tillage as part of CA has been implemented across the region using varied crop
establishment technologies such as manually prepared planting basins, jab planter and dibble sticks
(Ngoma et al., 2015; Kidane et al., 2019). CA planting basins have been found useful in coping
with rainfall variability and moisture deficits (Ngwira et al., 2013) as they improve conservation
of soil moisture in the root zone thereby mitigating in-season dry spells (Ngwira et al., 2014).
Alternative manual CA techniques, direct seeding using dibble stick or jab planter has also proved
to be more profitable, less risky and also deliver labor reductions ranging between 45 to 55%

243
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

relative to the traditional farmer practice (Mupangwa et al., 2019). Crop diversification through
legume inclusion into cereal based cropping systems has also been promoted as a solution to
counter yield losses, enhance stability, and ensure nutritional security in a sustainable manner (
Madembo et al., 2020).

The objective of this study was to evaluate the performance of maize cultivated as sole crop or
integrated with grain legumes either as intercropping or rotation and to determine the maize grain
yield performance of minimum tillage crop establishment techniques.

MATERIALS AND METHODS

The study was conducted, at Chitala research station in Malawi. The trial was laid out in a
Randomized Complete Block Design with three replications of the 12 cropping systems. Cropping
systems tested included conventional practice, CA sole maize, CA maize-legume intercrops and
CA maize- legume rotations. Crop establishment techniques involved (1) the conventional semi-
permanent hand hoed ridge and furrow system, (2) jab planter, (3) tapered wooden dibble sticks
and (4) hand hoe prepared CA planting basins.

Using R (version 4.3.1), linear mixed models were fitted to test for significant differences in maize
grain yield across treatments, seasons, cropping systems and to quantify the sources of residual
variance in the data.

RESULTS AND DISCUSSION

Response of maize grain yield and total biomass to crop establishment techniques in different
seasons
Crop establishment techniques significantly influenced maize grain and biomass yields across
different seasons (Figure 1). During the 2014/15 and 2015/16 seasons, characterized by medium
and low rainfall, planting basins and ridge-furrow systems yielded higher (4073 and 3907 kg ha-1
maize grain, respectively) compared to jab planter and dibble stick systems (3476 and 3213 kg ha-
1). Conversely, in the wetter 2016/17 and 2017/18 seasons, basin and ridge-furrow yields decreased

(2807 and 2836 kg ha-1), while dibble stick and jab planter yields improved (3915 and 3256 kg ha-
1). Biomass production showed a similar trend across these seasons. The CA basin system
performance could potentially be attributed to its higher water harvesting capacity that promotes
deeper water infiltration, better soil profile recharge and enhanced water retention capacity
compared to dibble stick and jab planter (Nyagumbo et al., 2016). These results support the notion
that CA basin systems can be an alternative and most preferred to drought prone regions of SSA
(Mupangwa et al., 2017). These findings also agree well with regional findings in on-farm studies
from southern Africa that put forward that CA basins can have negative impact on yields whenever
incessant rainfall events leading to water logging, occurred (Mupangwa et al., 2012; Nyagumbo et
al., 2020).

244
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 11. Interaction effects of season and crop establishment techniques on maize grain yield
and total biomass during experimentation in Malawi Chitala (2014-2017). Circles inside boxes
represent means, horizontal bar in the middle of each box represents the median, while lower and
upper box plot boundaries represent the 25th and 75th percentiles respectively. Lower and upper
whiskers represent the minimum and maximum values respectively. For each rainfall regime
different letters above boxes indicate significant differences at 5% significance level between
respective crop establishment techniques. Crop establishment techniques: BS = basin system, DS=
dibble stick, JP= jab planter and RF= Ridge-furrow.

Maize grain yield under different cropping systems

Significant differences in maize grain yields were observed among cropping systems over four
seasons (Figure 2, A-D). Maize-legume rotations consistently yielded the highest, while maize-
legume intercropping systems had the lowest yields. Maize-groundnut rotation outperformed
maize-cowpea intercropping by 1173, 878, 2700, and 987 kg ha-1 across the seasons. Within
intercropping systems, maize-pigeon pea consistently yielded 9%, 4%, 45%, and 7% more than
maize-cowpea. Similarly, maize-cowpea rotation surpassed maize-cowpea intercropping, with
yield increases ranging from 725 to 2700 kg ha-1 across seasons. Maize grain yield advantage of
the rotation system can be attributed to the high legume densities in rotation systems (i.e. the
legume phase of the rotation) which may result in biological nitrogen fixation (BNF) that can
supplement the applied mineral nitrogen thereby leading to high yield performance of rotation
systems (Mutsamba et al., 2020; Mupangwa et al., 2021). Also, high plant density in intercropping
systems combining maize and the associated legumes is usually 1.5–2 times the density of plants

245
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

in sole crops, thus, resulting in inter and intraspecific competition for essential growth resources
such as nutrients, water and light between maize and the companion legume and this can lead to
suppression of component crop yields compared to rotations and sole systems (Madembo et al.,
2020; Njira et al., 2021). In intercropping systems, maize-pigeon pea significantly outperformed
maize-cowpea. Pigeon pea develops much slower initially, and its greatest demand for water and
nutrients occurs after maize has been harvested and as such, there will be little competition with
the primary maize crop (Kimaro et al., 2009; Madembo et al., 2020).

Figure 12. Mean maize grain yield of the tested cropping systems over four consecutive growing
seasons (2014-15 to 2017-18) in Chitala, Malawi. For each season, different letters above bars
indicate significant differences between respective cropping systems at P < 0.05. Cropping systems
RotCwp = Maize-cowpea rotation, IntCwp = Maize-cowpea intercrop, RotGnt = Maize-groundnut
rotation, IntPp = Maize-pigeon pea intercrop, MzSole = Maize sole and Conv = Conventional.

REFERENCE

Cooper, P. J.M., J. Dimes, K. P.C. Rao, B. Shapiro, B. Shiferaw, and S. Twomlow. 2008. “Coping
Better with Current Climatic Variability in the Rain-Fed Farming Systems of Sub-Saharan
Africa: An Essential First Step in Adapting to Future Climate Change?” Agriculture,
Ecosystems and Environment 126(1–2): 24–35. doi:10.1016/j.agee.2008.01.007.
Kidane, Simon M., Dayton M. Lambert, Neal S. Eash, Roland K. Roberts, and Christian
Thierfelder. 2019. “Conservation Agriculture and Maize Production Risk: The Case of
Mozambique Smallholders.” Agronomy Journal 111(6): 2636–46.
doi:10.2134/agronj2018.05.0331.
Kimaro, A. A., V. R. Timmer, S. A. O. Chamshama, Y. N. Ngaga, and D. A. Kimaro. 2009.
“Competition between Maize and Pigeonpea in Semi-Arid Tanzania: Effect on Yields and
Nutrition of Crops.” Agriculture, ecosystems & environment 134(1–2): 115–25.
Ligowe, Cleoups Nalivata, Njoloma Joyce, Makumba Wilkson, and Thierfelder Christian. 2017.
“Medium-Term Effects of Conservation Agriculture on Soil Quality.” African Journal of
Agricultural Research 12(29): 2412–20. doi:10.5897/ajar2016.11092.

246
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Madembo, Connie, Blessing Mhlanga, and Christian Thierfelder. 2020. “Productivity or Stability ?
Exploring Maize-Legume Intercropping Strategies for Smallholder Conservation Agriculture
Farmers in Zimbabwe.” 185(April). doi:10.1016/j.agsy.2020.102921.
Makate, Clifton, Marshall Makate, and Nelson Mango. 2018. “Farm Household Typology and
Adoption of Climate-Smart Agriculture Practices in Smallholder Farming Systems of
Southern Africa.” African Journal of Science, Technology, Innovation and Development
10(4): 421–39. doi:10.1080/20421338.2018.1471027.
Mugiyo, Hillary, Tedious Mhizha, and Tafadzwanashe Mabhaudhi. 2018. “Effect of Rainfall
Variability on the Maize Varieties Grown in a Changing Climate: A Case of Smallholder
Farming in Hwedza, Zimbabwe.” (September). doi:10.20944/PREPRINTS201809.0152.V1.
Mupangwa, W., M. Mutenje, C. Thierfelder, M. Mwila, H. Malumo, A. Mujeyi, and P. Setimela.
2019. “Productivity and Profitability of Manual and Mechanized Conservation Agriculture
(CA) Systems in Eastern Zambia.” Renewable Agriculture and Food Systems 34(5): 380–94.
doi:10.1017/S1742170517000606.
Mupangwa, W., I. Nyagumbo, F. Liben, L. Chipindu, P. Craufurd, and S. Mkuhlani. 2021. “Maize
Yields from Rotation and Intercropping Systems with Different Legumes under Conservation
Agriculture in Contrasting Agro-Ecologies.” Agriculture, Ecosystems and Environment 306.
doi:10.1016/j.agee.2020.107170.
Mupangwa, W., C. Thierfelder, and A. Ngwira. 2017. “Fertilization Strategies in Conservation
Agriculture Systems with Maize-Legume Cover Crop Rotations in Southern Africa.”
Experimental Agriculture 53(2): 288–307. doi:10.1017/S0014479716000387.
Mupangwa, W., S. Twomlow, and S. Walker. 2012. “Reduced Tillage, Mulching and Rotational
Effects on Maize (Zea Mays L.), Cowpea (Vigna Unguiculata (Walp) L.) and Sorghum
(Sorghum Bicolor L. (Moench)) Yields under Semi-Arid Conditions.” Field Crops Research
132: 139–48. doi:10.1016/j.fcr.2012.02.020.
Mutsamba, EF, I Nyagumbo, and W Mupangwa. 2020. “Forage and Maize Yields in Mixed Crop-
Livestock Farming Systems: Enhancing Forage and Maize Yields in Mixed Crop-Livestock
Systems under Conservation Agriculture in Sub-Humid Zimbabwe.” NJAS-Wageningen
Journal of Life Sciences 92: 100317.
Ngoma, Hambulo, Nicole M. Mason, and Nicholas J. Sitko. 2015. “Does Minimum Tillage with
Planting Basins or Ripping Raise Maize Yields? Meso-Panel Data Evidence from Zambia.”
Agriculture, Ecosystems and Environment 212: 21–29. doi:10.1016/j.agee.2015.06.021.
Ngwira, Amos, Fred H. Johnsen, Jens B. Aune, Mulugetta Mekuria, and Christian Thierfelder.
2014. “Adoption and Extent of Conservation Agriculture Practices among Smallholder
Farmers in Malawi.” Journal of Soil and Water Conservation 69(2): 107–19.
doi:10.2489/jswc.69.2.107.
Njira, Keston O.W., Ernest Semu, Jerome P. Mrema, and Patson C. Nalivata. 2021. “Productivity
of Pigeon Pea, Cowpea and Maize under Sole Cropping, Legume–Legume and Legume–
Cereal Intercrops on Alfisols in Central Malawi.” Agroforestry Systems 95(2): 279–91.
doi:10.1007/s10457-020-00589-0.
Nyagumbo, I, S Mkuhlani, D Kamalongo, D Dias, and M Mekuria. 2016. “Maize Yield Effects of
Conservation Agriculture Based Maize–Legume Cropping Systems in Contrasting Agro-
Ecologies of Malawi and Mozambique.” Nutrient Cycling in Agro-ecosystems 105: 275–90.
Nyagumbo, Isaiah, Walter Mupangwa, Lovemore Chipindu, Leonard Rusinamhodzi, and Peter
Craufurd. 2020. “A Regional Synthesis of Seven-Year Maize Yield Responses to
Conservation Agriculture Technologies in Eastern and Southern Africa.” Agriculture,
Ecosystems and Environment 295(March): 106898. doi:10.1016/j.agee.2020.106898.
247
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Omulo, Godfrey, Thomas Daum, Karlheinz Köller, and Regina Birner. 2024. “Unpacking the
Behavioral Intentions of ‘Emergent Farmers’ towards Mechanized Conservation Agriculture
in Zambia.” Land Use Policy 136: 106979.
Steward, Peter R, Andrew J Dougill, Christian Thierfelder, Cameron M Pittelkow, Lindsay C
Stringer, Maxwell Kudzala, and Gorm E Shackelford. 2018. “Agriculture , Ecosystems and
Environment The Adaptive Capacity of Maize-Based Conservation Agriculture Systems to
Climate Stress in Tropical and Subtropical Environments : A Meta-Regression of Yields.”
Agriculture, Ecosystems and Environment 251(September 2017): 194–202.
doi:10.1016/j.agee.2017.09.019.
TerAvest, Dan, Lynne Carpenter-Boggs, Christian Thierfelder, and John P. Reganold. 2015. “Crop
Production and Soil Water Management in Conservation Agriculture, No-till, and
Conventional Tillage Systems in Malawi.” Agriculture, Ecosystems and Environment 212:
285–96. doi:10.1016/j.agee.2015.07.011.
Thierfelder, Christian, Rumbidzai Matemba-mutasa, W Trent Bunderson, Munyaradzi Mutenje,
Isaiah Nyagumbo, and Walter Mupangwa. 2016. “Agriculture , Ecosystems and Environment
Evaluating Manual Conservation Agriculture Systems in Southern Africa.” 222: 112–24.

248
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

USE OF DIGITAL EXTENSION TOOLS FOR AGRICULTURAL INFORMATION

MANAGEMENT AMONG CASSAVA VALUE CHAIN ACTORS IN IBADAN
METROPOLIS, OYO STATE, NIGERIA
#11756

1OlusholaFadairo, 2Adenike Durojaiye, 3Racheal Adegbayi

1Directorateof Entrepreneurship, Innovation and Sustainability Emmanuel Alayande University
of Education, Oyo, Nigeria; 2Department of Agricultural Science Education, Emmanuel
Alayande University of Education, Oyo, Nigeria; 3Department of Agricultural Extension Forestry
Research Institute, Nigeria
e-mail: [email protected]; [email protected]

ABSTRACT

Poor coverage of farmers by extension services and other limitations necessitates the need to adapt
the agricultural process to new opportunities, one of which is digital extension. This is moreso
important for cassava value chain actors given the recognition of the crop for food security in
Nigeria and its widespread promotion by several local and regional development interventions.
Therefore, the use of digital extension tools for agricultural information management among
cassava value chain actors in the Ibadan metropolis was investigated. Multi-stage sampling
procedure was used to select 195 cassava value chain actors comprising extension workers,
farmers, processors, and marketers. Data were collected using a structured questionnaire on their
enterprise characteristics, awareness, access, and use of digital extension tools for agricultural
information management. Data were analysed using frequency distribution, percentages, mean, and
ANOVA at α0.05. Respondents were mostly male (51.6%) and had 7.3±5.2 years of experience in
the cassava value chain. Awareness and access levels to digital communication tools were high
among the cassava value chain actors while the extent of use of the tools was still low. Agricultural
extension workers were the highest user (𝑥̅ =26.4) while the processors were least users (𝑥̅ =15.0)
of the digital extension tools. Awareness and use of digital tools for extension information
management is still below average among the cassava value chain actors. More emphasis is
required to be put on innovative digital information management systems in the traditional
extension approach to promote its wider use.

Keywords: Digital extension, Information management, Cassava value chain, Innovative

agriculture.

INTRODUCTION

Cassava is a major staple food and is widely grown across Nigeria owing to its wide adaptability,
economic importance and acceptance. Oyo State in the South-west Nigeria is a major producer of
cassava. In Nigeria, Cassava is increasingly becoming a raw material for food, feed and industrial
applications. Nigeria is the largest producer of cassava, and 12th largest for maize in the world,
amounting to annual production of about 60 million MT. Nevertheless, the nation is still known
for producing below its capacity with a national average of 9 t/ha (FAOSTAT, 2018). Worse still,
use of available improved technologies and innovations to address low cassava productivity in

249
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

most cases, have been limited by ineffective information transfer among the actors in the value
chain (Atser et al, 2024).

In Nigeria, agricultural extension services play critical roles in disseminating information on

improved farm technologies to farmers. Unfortunately, the performance of these service providers
has been largely disappointing in the recent past due to numerous intractable challenges which
bothers on their effectiveness, efficiency and reach (Atser et al, 2024). These challenges include
the significant financing shortfall for agricultural extension services, the low farmer-extension
workers ratio, the aging of the extension agents and the dearth of new employees. Davis et al (2019)
put the total workforce of public extension agents at roughly 7,000 and the ratio of extension agents
to farmers at between 1:5,000 and 1:10,000. Additionally, 60% of extension agents were
documented to be above the age of 40. Against this background, some authors have argued in
support of digitalization of extension services as a powerful tool to reach most smallholder farmers
(Adesope, 2021). This study was thus conducted to assess the existing digital extension landscape
and use among the cassava value chain actors in Ibadan Metropolis, Oyo State, Nigeria, and the
extension workers providing services to them.

METHODOLOGY

The study was carried out in Ibadan metropolis which is the capital of Oyo State, located in the
southwestern Nigeria. The city has 11 Local Government Areas (LGAs) and covering an area of
129.65km2 has the largest human population in the state and is also renowned for largest land size
in Nigeria. A multi-stage sampling procedure was used to select cassava farmers (44), processors
(12), marketers (28) [who were registered with their commodity association] and agricultural
extension workers (11) from three LGAs (Akinyele, Ido and Lagelu) in the study area. Data were
collected using a structured questionnaire on respondents’ enterprise characteristics, awareness,
access, and use of digital extension tools for agricultural information management. Awareness was
measures as yes and no; access was measured as always, sometimes and never; use was measured
as frequently, occasionally, just a trial and never. Scores were allotted to the options following the
Likert-type scale procedure. Data were analysed using descriptive statistics and ANOVA at α 0.05.

RESULTS AND DISCUSSION

Table 1 shows that 51.6% of the respondents were male indicating almost a parity level of male
and female involvement in the cassava value chain sub-sector in the study area. The respondents
were mostly educated at Diploma/Certificate level (57.9%), aged 39.4±8.1 years, had 7.3±5.2 years
of experience in the cassava value chain sub-sector and have attended 2.1±2.0 number of trainings
on digital agricultural extension. The respondent’s fair level of literacy, youthfulness and
experience in the value chain presents an obvious advantage for improving the cassava production
sub-sector in Nigeria if well harnessed.

250
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 1. Respondents’ personal characteristics.

Variables F %
Sex
Male 49 51.6
Female 46 48.4
Education
No education 1 1.1
Diploma/certificate 55 57.9
Bachelor and above 39 41.1
Years of experience Mean/SD = 7.3±5.2
Numbers of trainings attended Mean/SD = 2.1±2.0
Age in years Mean/SD = 39.4±8.1

Table 2 shows that virtual classroom (70.5%) and animated video clips (70.5%) enjoyed the most
popularity of all the digital tools among the value chain actors. This was followed by mobile
applications (66.3%) and skits (65.3%). Respondents (99.9%) were mostly unaware of mobile
phones. Video conferencing and IITA news App as forms of digital extension tools. In similar vein,
skits and virtual classroom (𝑥̅ =1.12) were mostly accessible to the respondents among the digital
extension tools while Mobile phone, video conferencing were least accessible (𝑥̅ =1.1). This trend
was also observed for use of the digital extension tools as virtual classroom and animated video
clips (𝑥̅ =2.69) were the mostly used of the tools in cassava value chain information exchange
among the respondents. Largely, the distribution suggests a plausible interrelationships among the
value chain actors’ awareness, access and use of the tools. Literature posits that awareness, and
access determines use of technology (Cui et al 2022).

Table 2. Awareness, access and use of digital extension tools among cassava value chain actors.

Digital tools Awareness Access Use

(%) index index
Mobile application e.g. e-diary, herbicide calculator 66.3 1.09 2.41
Interactive Voice Response (IVR) 63.2 1.06 2.44
Virtual Classroom e.g. Google meet 70.5 1.12 2.69
Web-based extension platform e.g. Farm Crowdy and 57.9 0.90 2.13
Agro Data
Geographical Information System (GIS) 61.1 0.83 1.87
Animated video clips 70.5 1.09 2.69
Satellite system 55.8 0.70 1.66
Drones 55.8 0.63 1.48
Internet of things/Remote sensing 54.7 0.82 1.88
Skit 65.3 1.12 2.08
Mobile phone 1.1 0.10 2.41
Video conferencing 1.1 0.10 2.44
IITA news App 1.1 0.10 2.15

251
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 3 shows that while awareness and access levels to digital communication tools were high
among 57.9% and 53.7% of the cassava value chain actors, respectively, the extent of use of the
tools for innovation dissemination and exchange was still low among most of the actors (51.6%).
The low usage despite a high awareness and access are traceable to the problem of underinvestment
and poor reach of the potential users by the e-digital extension service providers (Baconguis, 2022).
This is addition to the challenge of poor internet connectivity.

Table 3. Categorization of cassava value chain actors based on their levels of awareness, access
and use of digital tools.

Awareness level Access level Use extent

F % F % F %
Low 40 42.1 44 46.3 49 51.6
High 55 57.9 51 53.7 46 48.4
Min=0; Max =12; 6.2±4.1 Min=0; Max=22; 9.4±6.8 Min=10; Max=40; 20.8±8.7

Table 4 shows a significant difference in the use of digital extension tools among the cassava value
chain actors (F=3.598; p < 0.05) with agricultural extension workers having the highest level of
usage (𝑥̅ =26.36) while the processors had the least usage level (𝑥̅ =20.29).

Table 4. Difference in the use of agricultural digital extension tools among the cassava value chain
actors.

Sum of square df Mean square F Sig

Between groups 766.614 3 255.538 3.598 0.017
Within groups 6462.692 91 71.019
Total 7229.305 94
Mean Separation of value chain actors
Processors 𝑥̅ =15.00 Marketers Farmers Agricultural
𝑥̅ =20.29 𝑥̅ =21.39 extension
workers
𝑥̅ =26.36

REFERENCES

Adesope O.M. (2021). Agricultural Extension in Contemporary Nigeria. Optimizing Our

Development Potentials. University of Port Harcourt Inaugural Lecture Series No170.
Thursday June 17, 2021. 102p
Atser, G. L., Oluoch, M., Gambo, A., Diso, H. Z., Fadairo, O. S., & Jibrin, A. A. (2024). The
commodity association traders/trainers extension approach: smallholder productivity, input
and market linkages in Nigeria. African Journal of Food, Agriculture, Nutrition &
Development, 24(3) (Kenya).
Baconguis, R. T. (2022). Agricultural technology: Why does the level of agricultural production
remain low despite increased investments in research and extension? (No. 2022-06). PIDS
Discussion Paper Series.

252
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Cui, S., Li, Y., Jiao, X., & Zhang, D. (2022). Hierarchical linkage between the basic characteristics
of smallholders and technology awareness determines Small-Holders’ willingness to adopt
green production technology. Agriculture, 12(8), 1275.
Davis K., Lion K., and Arokoyo T. (2019). Organizational Capacities and Management of
Agricultural Extension Services in Nigeria: Current Status. South African Journal of
Agricultural Extension Volume 47, No 2, 2019:118-127. Http://dx.doi.org/10.17159/2413-
3221/2019/v47n2a508.
FAO Statistics (2013, June). Food and Agricultural Organisation, Data base results. Retrieved from
June, 2015, FAO.

253
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

EMPOWERING SMALLHOLDER FARMERS: PRECISION AGRICULTURE

MODELS IN AFRICA (A CASE STUDY OF NIGERIA)

Stephen Oluwaseun Popoola

Agro Optimizer Limited, Iseyin, Oyo State, Nigeria
e-mail: [email protected]; +2349064685088

ABSTRACT

This paper investigates the transformative potential of precision agriculture (PA) in enhancing the
productivity and sustainability of smallholder farmers across Nigeria. As the backbone of Africa's
agricultural sector, smallholder farmers face persistent challenges, including resource scarcity,
climate variability, and limited access to modern agricultural practices. By leveraging technologies
such as GPS, remote sensing, and data analytics, PA offers innovative strategies to optimize
resource use, improve crop yields, and support data-driven decision-making.

Utilizing a mixed-methods approach that includes field surveys, in-depth farmer interviews, and
empirical yield analysis across three regions—Southwest, Northern, and Middle Belt—this study
reveals significant regional disparities in PA adoption. The Southwest shows higher awareness and
adoption rates, while the North faces greater challenges due to inadequate infrastructure and limited
access to extension services. Key barriers identified include high implementation costs, insufficient
training, and infrastructural gaps.

Despite these obstacles, the study finds that PA technologies substantially improve productivity,
resource efficiency, and resilience in smallholder farming systems. The findings underscore the
need for robust policy frameworks, targeted training programs, and scalable technology solutions
to foster wider PA adoption. Ultimately, this research highlights actionable recommendations for
stakeholders, including policymakers, agricultural extension services, and technology providers, to
enhance food security and promote sustainable agricultural development in Nigeria and beyond.

INTRODUCTION

Agriculture remains the backbone of Africa's economy, supporting the livelihoods of millions, with
smallholder farmers making up nearly 80% of the continent’s agricultural workforce (AGRA,
2022). These smallholders are crucial to achieving food security and driving rural development,
yet they face persistent challenges that hinder their productivity. Limited access to modern farming
inputs, poor infrastructure, and increasing climate variability have led to stagnating agricultural
outputs, exacerbating food insecurity and slowing economic growth (FAO, 2023; Jayne et al.,
2019). As the world strives to meet the Sustainable Development Goals (SDGs), particularly Goal
2 (Zero Hunger) and Goal 13 (Climate Action), enhancing agricultural efficiency and resilience is
more urgent than ever.

In recent years, precision agriculture (PA) has emerged as a transformative approach to modern
farming. By leveraging technologies such as GPS navigation, remote sensing, drones, and data
analytics, PA offers innovative solutions to optimize farming practices, enabling efficient resource

254
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

use, better crop yields, and reduced environmental impact (Zhang et al., 2023; Gebbers &
Adamchuk, 2021). The adoption of PA technologies can particularly benefit Africa’s smallholder
farmers, who operate under resource constraints but have significant potential to boost productivity
through data-driven decision-making.

The critical question now is: why is it crucial to focus on precision agriculture for smallholder
farmers in Africa today? As the continent grapples with rapid population growth, projected to reach
2.5 billion by 2050 (UN, 2023), there is an urgent need to increase agricultural productivity to meet
rising food demands sustainably. Precision agriculture provides a pathway to enhance the
efficiency of smallholder farms, reduce wastage of inputs like water and fertilizers, and build
resilience against climate shocks (Mupangwa et al., 2024). However, despite its potential, the
adoption of PA in Africa remains low due to high costs, limited infrastructure, and a lack of
technical expertise among farmers (Makombe & Gachuiri, 2023).

This study delves into the application of precision agriculture models tailored for smallholder
farmers in Africa. By examining the tangible impacts of PA on productivity, resource management,
and sustainability, it seeks to identify both the opportunities and challenges associated with scaling
these technologies in rural contexts. The research emphasizes the socio-economic significance of
smallholder farmers, their operational constraints, and the barriers they face in adopting modern
agricultural practices.

Ultimately, this study aims to uncover practical strategies to enhance the adoption of PA in Africa,
contributing to broader global objectives, such as boosting food security, promoting sustainable
agricultural practices, and advancing rural economic development. By bridging the gap between
technology and smallholder needs, this paper underscores the transformative potential of precision
agriculture in empowering African farmers, thereby supporting the continent's progress toward the
SDGs.

METHODOLOGY AND RESULTS

Research Design
A descriptive research design was employed to assess awareness, challenges, and benefits of PA.
Data was collected using structured questionnaires and in-depth interviews to understand socio-
economic, infrastructural, and technological influences on PA adoption.

Sample Selection
The study focused on three regions in Nigeria—Southwest, Northern, and Middle Belt—chosen
for their distinct agro-ecological characteristics:

Southwest: Better infrastructure and higher tech adoption.

Northern: Harsh climate, emphasizing the need for resource-efficient PA.
Middle Belt: A mix of rain-fed and irrigated farming practices.

A stratified random sampling method was used to select 300 smallholder farmers (100 per region)
and conduct 30 interviews with agricultural stakeholders.

255
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Data Collection
Data sources included:

Primary data: Structured questionnaires and semi-structured interviews with farmers to gather
quantitative and qualitative insights on PA awareness and barriers.
Secondary data: Literature reviews, government reports, and agricultural databases to provide
context.

Data collection took three months, with enumerators using local languages to ensure accuracy.

Data Analysis
Quantitative data was analyzed using SPSS for descriptive statistics and correlations. Qualitative
data was transcribed and analyzed with NVivo for thematic insights.

Study Limitations
Geographical scope: Limited to three regions, potentially not reflecting national diversity.
Access to participants: Challenges in reaching remote areas may have affected data depth.
Technological literacy: Varying familiarity with technology among farmers might have impacted
their responses.

RESULTS

The results of this study present a comprehensive analysis of the adoption of precision agriculture
(PA) among smallholder farmers across the Southwest, Northern, and Middle Belt regions of
Nigeria. The findings highlight regional disparities in awareness, adoption rates, and the
effectiveness of extension services, which provide critical support for implementing PA
technologies.

Regional Analysis of PA Awareness and Adoption

Southwest Nigeria demonstrated the highest levels of awareness and adoption of PA technologies,
with 65% of surveyed farmers indicating familiarity with basic precision tools like GPS-enabled
devices and soil sensors. This region benefits from better access to agricultural extension services,
infrastructure, and proximity to urban centers, which facilitates exposure to modern farming
techniques.

In contrast, Northern Nigeria, particularly in the semi-arid zones, showed lower awareness (38%)
and adoption rates (20%). The harsh climatic conditions, coupled with limited infrastructure,
impede farmers' access to PA resources. Additionally, literacy rates in this region are lower, which
affects farmers' ability to adopt technologically advanced practices. However, those who did
implement PA technologies, such as drip irrigation and soil moisture sensors, reported a noticeable
increase in water-use efficiency, suggesting that targeted interventions could yield significant
benefits.

The Middle Belt region exhibited moderate awareness (55%) and adoption levels (45%). This area
benefits from a blend of rain-fed and irrigated farming, making it more adaptable to PA techniques
like remote sensing for weather predictions and variable rate application of inputs. However,

256
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

challenges such as inconsistent access to market information and limited financial resources
continue to hinder widespread adoption.
Key Findings and Trends
The study found that:

Access to Extension Services: Farmers in the Southwest region had better access to agricultural
extension officers (70%) compared to their counterparts in the Northern (30%) and Middle Belt
regions (50%). The presence of well-funded agricultural programs and proximity to research
institutes in the Southwest played a crucial role in this disparity.

Resource Availability: The Middle Belt region displayed a higher adoption of PA technologies
compared to the North due to relatively better access to inputs like fertilizers, seeds, and mobile-
based advisory platforms. This suggests that targeted resource allocation could significantly
improve PA adoption in resource-constrained areas.

Perceived Benefits of PA: Across all regions, farmers who adopted PA reported improved crop
yields (average increase of 25-30%) and resource efficiency, particularly in irrigation and fertilizer
usage. However, the initial cost and technical expertise required for PA adoption were cited as
major barriers.

Table 1. Awareness and Use of Precision Agriculture by Region.

Region Number of Aware of Understand Use Precision
Respondents Precision Precision Agriculture
Agriculture (%) Agriculture (%) (%)
Southwest 20 45% 35% 20%
(Iseyin and
Oyo)
Northern 20 20% 15% 5%
(Kano and
Kaduna)
Middle Belt 10 10% 5% 0%
(Benue State)
Total 50 28% 18% 12%
This table provides a breakdown of the respondents by region and shows their level of awareness,
understanding, and usage of precision agriculture technologies.

Table 2. Barriers to Precision Agriculture Adoption by Region.

Barrier Southwest Northern Middle Belt Total
(%) (%) (%) (%)
High Cost of Technology 65% 75% 70% 70%
Lack of Infrastructure 55% 70% 75% 65%
Limited Knowledge and 45% 55% 80% 58%
Training
Lack of Government Support 50% 50% 50% 50%
This table breaks down the specific barriers faced by farmers in each region, highlighting the
differences in challenges faced by Southwest, Northern, and Middle Belt farmers.

257
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 3. Perceived Benefits of Precision Agriculture.

Benefit Southwest Northern Middle Belt Total
(%) (%) (%) (%)
Improved Yield 60% 55% 40% 52%
Reduced Input Costs 45% 40% 35% 40%
Better Water Management 50% 65% 30% 48%
Increased Market 35% 20% 15% 24%
Competitiveness
This table summarizes the perceived benefits of precision agriculture as reported by the farmers,
showing the variation across different regions.

Table 4. Regional Distribution of Respondents.

Region Number of Respondents Percentage (%)
Southwest 20 40%
Northern 20 40%
Middle Belt 10 20%
Total 50 100%
This table shows the total number of respondents surveyed across the three regions, along with the
percentage distribution.

Table 5. Access to Extension Services.

Region Respondents with Access to Extension Services Percentage (%)
Southwest 8 40%
Northern 12 60%
Middle Belt 5 50%
Total 25 50%
This table shows the number of respondents in each region who have access to agricultural
extension services and the percentage of those respondents relative to the total number in each
region.

DISCUSSION AND RECOMMENDATIONS

Comparative Analysis with Other African Countries

The barriers to adopting PA observed in Nigeria mirror challenges faced in other parts of Africa,
particularly in countries like Kenya, Tanzania, and Ghana. Similar issues—such as limited access
to capital, inadequate infrastructure, and low levels of technological literacy—are common across
the continent. In Kenya, for instance, while PA holds promise for smallholders, the cost of
technology remains prohibitive for many, just as it does in Northern Nigeria. Meanwhile, Tanzania
has made strides in mobile-based agricultural advisory services, which could serve as a model for
improving PA adoption in Nigeria’s Middle Belt region.

Implications for Policy and Practice

The disparities identified in this study highlight the need for region-specific interventions to
promote PA adoption:

258
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Strengthening extension services: Policymakers should prioritize training extension officers in the
Northern region to enhance farmers' access to PA knowledge.

Subsidizing technology costs: Financial incentives, such as subsidies or low-interest loans, could
make PA tools more accessible, particularly in resource-poor areas.

Leveraging mobile platforms: Expanding mobile-based advisory services, especially in the Middle
Belt and Northern regions, could bridge the information gap and increase farmers' engagement with
precision agriculture.

Recommendations for Stakeholders

1. Farmers:
o Smallholder farmers should seek support from local cooperatives and agricultural
organizations to gain access to precision farming technologies and training. Pooling
resources can make technology more affordable.
o Farmers should adopt lower-cost, locally adapted precision farming tools where
available to help them transition into modern agricultural practices gradually.
2. Policymakers:
o Policymakers should focus on expanding agricultural extension services, especially
in underserved regions like the Southwest. Improved access will help bridge the gap
in technical knowledge, which is crucial for adopting precision agriculture.
o Governments need to provide financial incentives such as subsidies or grants to help
farmers afford precision agriculture tools. Investments in rural infrastructure—such
as reliable electricity and internet—are essential for enabling these technologies.
3. Researchers:
o Future research should explore the adoption of precision agriculture in a wider range
of regions, both in Nigeria and other African countries, to provide a more
comprehensive understanding of the barriers and drivers in different contexts.
o Researchers should also investigate the long-term economic impacts of precision
agriculture on smallholder farmers, particularly focusing on yield improvements and
sustainability, to inform more effective policy decisions

CONCLUSION

By addressing these barriers, Nigeria can better align with broader Sustainable Development Goals
(SDGs), particularly SDG 2 (Zero Hunger) and SDG 12 (Responsible Consumption and
Production), while fostering a more resilient agricultural sector capable of withstanding climate
variability and resource limitations.

REFERENCES

Akanbi, W. B., & Olayanju, T. M. (2021). Precision agriculture: A pathway for smallholder farmers
in Africa to enhance productivity and sustainability. African Journal of Agricultural
Research, 16(12), 140-154. https://doi.org/10.5897/AJAR2021.15699
Alabi, J., & Eze, P. (2023). The impact of precision agriculture on smallholder farmers' productivity
in Sub-Saharan Africa. African Journal of Agricultural Research, 18(5), 320-334.
doi:10.5897/AJAR2023.12345
259
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Bakhsh, A., & Khan, M. A. (2022). Assessing the socio-economic impacts of precision agriculture
on smallholder farmers in the Punjab region of Pakistan. Journal of Precision Agriculture,
23(3), 439-455. https://doi.org/10.1007/s11119-021-09712-9
BusinessDay. (2023). High costs limit adoption of precision agriculture in Nigeria. Retrieved from
https://businessday.ng
Choudhury, A., & Raghavendra, H. (2022). The role of technology in precision agriculture for
smallholder farmers: A systematic review. Agricultural Technology, 12(4), 290-305.
https://doi.org/10.1016/j.agt.2022.03.004
Davis, A. G., & McKenzie, H. (2021). The adoption of precision agriculture technologies by
smallholder farmers: A global perspective. Journal of Agricultural & Food Information,
22(3), 202-218. https://doi.org/10.1080/10496505.2021.1943765
Elham, A., & Ebrahimi, A. (2021). Precision agriculture and its implications for smallholder
farmers: Evidence from Iran. International Journal of Agricultural Sustainability, 19(1), 12-
26. https://doi.org/10.1080/14735903.2021.1888342
Kumar, A., & Gupta, A. (2022). Precision agriculture technologies: Enhancing productivity and
sustainability for smallholder farmers in India. Journal of Agricultural Science, 14(2), 112-
125. https://doi.org/10.5539/jas.v14n2p112
Moges, A., & Sweeney, J. (2021). Enhancing the resilience of smallholder farmers through
precision agriculture: Insights from Ethiopia. Agricultural Systems, 186, 102973.
https://doi.org/10.1016/j.agsy.2021.102973
Nwosu, C., & Adewale, T. (2024). Addressing barriers to technology adoption: A case study of
precision agriculture in West Africa. Journal of Sustainable Agricultural Technologies,
11(2), 85-97. doi:10.1016/j.jsat.2024.05.003
Nyangweso, P., & Obonyo, E. (2021). The impact of precision agriculture on smallholder farmers'
livelihoods in Kenya. International Journal of Agricultural Economics and Rural
Development, 14(1), 45-59. https://doi.org/10.1007/s40100-021-00401-6
Smith, R., & Owusu, D. (2023). Overcoming infrastructural challenges in the adoption of precision
agriculture among smallholder farmers. International Journal of Agricultural Innovation,
7(3), 154-169. doi:10.1016/j.ijai.2023.03.007
United Nations Development Programme (UNDP). (2024). Enhancing food security through
digital agriculture in Africa. Retrieved from https://www.undp.org
World Bank. (2023). Leveraging digital agriculture for sustainable development in Africa.
Retrieved from https://www.worldbank.org
Yadav, A., & Singh, S. (2022). Leveraging precision agriculture for sustainable development
among smallholder farmers in northern India. Sustainable Agriculture Research, 11(3), 89-
99. https://doi.org/10.5539/sar.v11n3p89

260
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

SOCIAL SCIENCE APPLICATIONS IN PRECISION AGRICULTURE

261
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

ASSESSMENT OF ACCESS TO AND UTILISATION OF TREADLE AND HIP PUMP

TECHNOLOGY BY FARMERS IN MACHAKOS COUNTY, KENYA
#11285

T. Gitau, G.B. Mose, and C. Masiga

Kenyatta University, Nairobi, Kenya
e-mail: [email protected] +254 728 805 205

ABSTRACT

The aim of this study was to assess the access to and utilization of the treadle and hip pump
technology by farmers in Machakos County, Kenya so that gender responsive strategies can be
suggested to improve the uptake of the technology among the farmers. This was due to low
adoption of the two pumps in Machakos county after previous KickStart International and
Washington State university (KSI/WSU) study that marketed and distributed the technology among
farmers in the county. The specific objective in this paper was to examine the challenges facing
farmers when accessing to and utilizing the pumps in the county that seemed to be in dire need of
the pumps due to the arid and semi-arid climate, yet the adoption was too low. The study was
guided by the social relations framework of analysis by Naila Kabeer which was complemented by
the diffusion of innovations theory. The study used experimental and descriptive research designs
for both qualitative and quantitative data. The study had 70 respondents comprising 42 farmers
who were pump buyers purposively selected for the study and 28 pump non-buyers identified via
snowball sampling method. Data were collected using interview schedules, key informant
interview guides and focus group discussion guides. Findings reported major challenges to access
to and utilization of the pump technology were at family level and market level. The study
concluded that the use the treadle and hip pump technology was still appropriate in compared to
the tools they were previously used to. However, the challenges faced by the farmers were both
internal and external, i.e., from buyers and marketers to the environment in which the operated.
The pump technology design hampered use by women since it required a lot of energy to use. The
study recommended putting gender responsiveness in design of pumps, consistent intergenerational
marketing and distribution of the pump technology with follow up by the innovation marketer for
adoption among women, men and youth.

INTRODUCTION

Globally, farming is perceived as a lucrative venture in regions with sustained fertility and water
resources. For arid and semi-arid region, farming has been a difficult and expensive venture as
irrigation projects are required to enable food security for both domestic and commercial use. For
this reason, the need to develop more approaches that will improve farming in arid and semi-arid
areas became a necessity for non-governmental organizations (NGOs) and government
stakeholders (Ivers & Cullen, 2011). Regardless of this provision, the need to develop small-scale
farmers required cheap farming technology that they could use in smallholder farms. This led to
embracing of technology in the agriculture sector. One of the technologies is the treadle pump,
which was developed in 1979 by a team working with the Rangpur Dinajpur Rural Service (RDRS).
The treadle pump extracted larger volumes of water more than the existing hand-operated pumps.
The design of the pumps targeted the poor and those in remote areas with limited access to diesel
and technical support (Purcell, 1997).
262
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Asia was the birthplace of the treadle pump (Orr et al. 1991), for this reason, the demand for the
technology in rural Bangladesh was significant. Its popularity then increased its supply across
neighboring nations including India and Pakistan. In these regions, the number of rural farmers was
massive as they solely depended on agriculture for their livelihood. Both men and women in the
regions were provided with treadle water pumps. NGOs’ involved in the programme wanted to
increase the buying power of the consumer thus launching a purchasing plan that would fit the
financial comfort of the farmers. The NGOs’ provided pumps for irrigation projects on credit to
farmers prior to agreeing to a payment plan (Prabhu, 1999). Since the creation of the treadle pump
in Bangladesh, 1.4 million pumps had been sold to the local farmers by 1991 (Alistair et al. 1991).
This depicted the importance of the affordable irrigation option.

The hip pump is a KSI innovation of the small-scale irrigation pump in terms of the body parts
used to operate it, lighter weight and lower cost than the treadle pump. From these efforts, the
treadle and hip pump technology is being used mainly in Kenya, Tanzania, Malawi, Zambia, Mali,
Burundi, Sudan, Burkina Faso, Uganda and Rwanda (Sijali & Mwago, 2009). The popularity of
the technology in these countries was mainly influenced by Kick Start International (KSI). The
organization was born in Kenya, which increased the use of pumps in the country as well in East
Africa.

Machakos County is one of the 47 counties in Kenya and formerly part of the Eastern province
where the pump technology has been adopted, thus of interest to this study. An introduction of
small-scale irrigation technology by KSI in conjunction with Washington State University (WSU)
targeting women began in 2014 and ended in 2016 in the county. The goal of KSI/WSU was to
offer the technology and purchasing plan fit to the financial comfort of women farmers and through
comprehensive marketing of the pumps to assure as many sales as possible. The farmers were to
purchase the pumps either in cash, on credit or savings options. The treadle pump cost KES 14,950
while the hip pump cost KES 9,500 which farmers were expected to pay back within six months.
Pumps were provided to individuals but based on a group loan and repayment. Once pumps were
distributed to the group members, the farmers were further taught how to use and maintain the
technology KSI (2015).

This study sought to assess how the women farmers were accessing the KSI/WSU technology as
well as utilizing the treadle and hip pump technology in their farms in Machakos County since the
former study was not scaled up as envisioned. The researcher engaged with women farmers who
took up the KSI/WSU pump technology and those who did not take up the technology, yet they
had water available to irrigate their crops, i.e., the non-buyers in the study.

MATERIALS AND METHODS

The study used experimental and descriptive research designs for both qualitative and quantitative
data. The study had 70 respondents comprising 42 farmers who were pump buyers purposively
selected for the study and 28 pump non-buyers identified via snowball sampling method.
Convenience sampling was used to sample the key informants i.e. chiefs, extension officers from
various wards and KSI/WSU representative in the county. Data were collected using interview
schedules, key informant interview guides and focus group discussion guides. The study’s
quantitative data was analyzed using SPSS version 23 to give descriptive statistics. Data were
263
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

presented as percentages, frequencies, averages tabulations, histograms, and pie charts. Qualitative
data were analyzed using content analysis, organized into themes and patterns formed, and
presented in a narrative form and verbatim quotations.

RESULTS AND DISCUSSION

Challenges faced by farmers when accessing and utilizing the treadle and hip pump
This section covers the results and discussion that sought to find out the challenges that women
farmers faced while accessing and utilizing the treadle and hip pump irrigation technologies. In the
study 62(88.6%) stated that they had challenges while 8(11.4%) did not face challenges in the
process of acquiring their pump. The researcher categorized the list of challenges given from
individual responses and FGD discussions into themes under institutions in Naila Kabeer’s social
relations framework on how access to and utilization of the treadle and hip pump challenges were
reworked in the four institutions categories of the family, community, market and state.

Table 1. Challenges as perceived by study respondents.

Buyers Non buyers KSI Chiefs Extension

representative officers

Levels of Female Male Female Male

challenges
No % No % No % No % No % No % No %
Family 20 67 5 42 11 58 5 56 1 100 4 50 2 50
Community 21 70 3 25 6 32 7 78 1 100 6 75 1 25
Market 16 53 10 83 4 21 5 56 1 100 2 25 3 75
State 25 83 7 58 8 42 7 78 1 100 4 50 2 50
Average
Family 23(54.5%) 14(50%) 1(100%) 4(50%) 2(50%)
Community 14(47.5%) 15(55%) 1(100%) 6(75%) 1(25%)
State 15(50%) 14(50%) 1(100%) 2(25%) 3(75%)
Market 20(68%) 11(38.5%) 1(100%) 4(50%) 2(50%)

Major challenges to access to and utilization of the pump technology among pump buyers were at
family level and market level. At family level, the women willingness to acquire the pumps was
met by the inability to do so due to lack of finances while at market level, the pump design itself
required some gender responsive considerations as some women said it was tiresome for them to
use.

“The pump is tiresome and requires two people when using especially the pedalling part
requires a lot of energy which we women lack. Personally, I am involved in directing
the hose to the crops or tank while my husband or son pedals the pump” (Interview with
a woman pump buyer at Kithimani 15/4/2017)

Community level and state challenges were reported too, however, solution to their contribution to
the challenges would act as accessories to strengthen solutions at family and market level. Asked

264
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

about the disparities in representation, a female extension officer from Machakos town Sub County
had this to say,

“We have previously experienced situations where male officers were chased from
homes by some male homeowners especially when they appeared alone. Although this
was partly solved by meeting the farmers in groups, some women in groups may not
express themselves well in these groups which are also male dominated,” said Mueni.
(Key informant interview with Machakos sub county extension officer 15/2/2017)

Lack of water, finances, labor, time of pump repayment, priorities in a family, source of
information about the pumps and distance to point of purchase of the pump were also critical. The
following responses from the farmers informed part of the challenges.

“The time limit for repayment was too short” (Interviews with a savings buyer from
Machakos on 13/2/2017)

Group members from Kyangala also thought the distance to where they get the pumps was too far.

“The distance to the market especially to vendors is far and I am unable to replace my
rubber caps”. (FGD with a group from Kyangala on 14/2/2017)

The challenges faced by the farmers were both internal and external, i.e., from buyers and marketers
to the environment in which they operated.

The research recommended family level challenges to be addressed through education on

importance of support of farm chores by everyone in the family regardless of their gender. Market
level challenges to also be addressed by the marketer through design of pumps that are gender
responsive to allow participation in utilization by all members of the family. Another gender aspect
the marketer could utilize in marketing and distribution is through working with female opinion
leaders, establish demonstration farms with female headed households as well as appointing
enterprising lead farmers such as the group secretary and pump dealers to build supply networks
and create linkages with farmers. This secretary could earn steady income from a profit-sharing
model from sales. The marketer could also consider advertising and marketing that does not
conform to gender stereotypes during promotional campaigns through promoting multiple range of
products for women and men. The type of media chosen and content of message about the pump
to reach women should consider their low mobility outside their home. Community level
challenges could be adjusted through tapping on solutions at family level while state level
challenges to be addressed in a bottom-up manner by marketers together with other key players in
agriculture such as community-based organizations (CBO’s), NGOs, self-help groups and faith-
based organizations through lobbying for support at county and national level.

REFERENCES

Alistair, O., Islam, A. & Gunnar, B. (1991). The Treadle Pump: Manual Irrigation for Small
Farmers in Bangladesh. R.D.R.D. Pdf.
Ivers, L. and Cullen, K. (2011). “Food insecurity: special considerations for women. The American
Journal of Clinical Nutrition, 94, (6): 1740–1744.
265
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

KickStart International. Empowering women with technology. February 2015.

http://www.kickstart.org/about-us/Women's%20Empowerment%20Report%20-
%20KickStart%20International%20-%20February%202015.pdf
Orr, A., Islam, A. N., & Barnes, G. (1991). The treadle pump: manual irrigation for small farmers
in Bangladesh. Dhaka: Rangpur Dinajpur Rural Service.
Prabhu, M. (1999). Marketing treadle pumps to women farmers in India. Gender &
Development, 7(2), 25-33.
Purcell, R. (1997). Potential for small-scale irrigation in sub-Saharan Africa: the Kenyan example.
Sijali, V. & Mwago.M. (2009). Money Maker Pumps: Creating Wealth in Sub-Saharan Africa.
http://siteresources.worldbank.org/AFRICAEXT/Resources/258643-
1271798012256/YAC_chpt_18.pdf New York:United Nations

266
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

WATER MANAGEMENT FOR PRECISION AGRICULTURE

267
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REAL-TIME MOISTURE CONTROL IN IRRIGATION SYSTEMS FOR WATER USE

EFFICIENCY AND CLIMATE CHANGE RESILIENCE. A REVIEW
#11679

D. Kindikiza
West African Centre for Water, Irrigation and Sustainable Agriculture - University for
Development Studies (WACWISA-UDS), Tamale, Ghana
e-mail: [email protected]

ABSTRACT

Due to the increasing water scarcity and uncertainties of climate change, improving crop water use
efficiency and productivity, at the same time minimizing detrimental effects on the environment to
meet the world's rising food demand. Thus, is necessary to adopt innovative irrigation strategies,
such as drip irrigation. Smart irrigation has a potential of improving water use efficiency in
precision agriculture. Conventionally, irrigation systems rely on heuristic methods in order to
schedule irrigation which either leads to over-irrigation or under-irrigation which affects water use
efficiency. In this paper, we are describing and comparing different methods of irrigation water
control strategies for irrigation decisions with their impact on climate change resistance. Those are
open loop strategies that include manual, time and volume-based control against closed loop types
which are composed of optimal, artificial intelligence and linear controls. Furthermore, the paper
is reflecting on automation in different irrigation systems.

This review paper has important implications for farmers and agricultural stakeholders, to adapt
technologies, as it offers practical solutions to address the ongoing challenges emanated from water
scarcity and climate change.

Keywords: Soil moisture control, Irrigation systems, Water use efficiency, Climate change.

INTRODUCTION

Climate change and increasing population do impose additional pressure to Global water resources
use and scarcity, that are vital for agricultural production (Ungureanu et al., 2020). According to
the United Nations Department of Economic and Social Affairs (2019), the world population will
hit 9.7 billion by 2050 translating into increased demand for nutritious food and water resources.
The Food and Agricultural Organization (FAO) forecasts a more than 50 % increase in irrigated
food production by 2050, which will require a 10 % increase in water abstracted for agriculture,
provided water productivity improves (Food and Agriculture Organization, 2017). The land on
which food is cultivated does not expand, which means agricultural cropping systems need to
utilize the available water and land resources efficiently to feed the future population.
Understanding the mechanisms that can improve water use efficiency and result in significant water
savings and higher yield is therefore paramount (Saleem et al., 2013).

Smart irrigation control in precision agriculture is becoming popular due to water saving ability
buy providing water to the desirable location (root zone) and maximizing yield. Sensor-based
decision support and automation can reduce significantly manual intervention while operation

268
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

irrigation systems (Klein et al., 2018). Moreover, irrigation soil moisture control strategies, take
into consideration plants response to water stress, changing weather variables through Internet of
Things (IoT) monitoring (Abioye et al., 2020).

This paper implication provides valuable insights into the potential benefits of smart irrigation
control systems for water use efficiency and contribute to the development of more sustainable
solution for climate change resilience through agricultural irrigation for food security.

MATERIALS AND METHODS

This review applied methods of selecting the works of literature that are published on precision
irrigation and all its existing control strategies includes an extensive search through a different
multidisciplinary online database, such as Elsevier, Springer, Taylor & Francis, Google Scholar,
and other high ranked Scopus indexed journals. The emphases were placed on numerous research
articles and books related to irrigation water control and monitoring strategies were considered too.
The keyworks; soil moisture control strategies, irrigation systems, water use efficiency and climate
change were followed to relate the articles. Therefore, the papers were selected, read and
summarised to ensure the systematic flow of the ideas.

DISCUSSION

Irrigation Water Control Techniques in Precision Agriculture

Irrigation control strategies are divided into open-loop systems and closed-loop systems. While
open-loop systems apply a preset action like in simple irrigation timers, closed-loop systems
receive feedback from sensors, make decisions and apply the resulting decisions to the irrigation
system. Figure 1 presents a detailed classification of irrigation control strategies derived from
literature studies.

Climate Resilience Action

The data-driven models developed by the authors were able to estimate reference temperatures,
enabled automated calculation of the crop-water-stress index for effective assessment of crop water
stress (Vallejo-Gómez et al., 2023). There is a need to evaluate stochastic and Hybrid MPC
strategies for efficient irrigation scheduling in open-field agriculture. There are many uncertainties
and unpredictable disturbances from the environment unlike in a greenhouse where environmental
conditions are controlled.

Application of Real - time Moisture Control Strategies in Agriculture

Owing to the similarity of agricultural processes to industrial processes, Model Predictive Control
has been applied in product processing, agricultural production, greenhouses and irrigation systems
(Bwambale et al., 2023) . Model predictive control has been applied in canal flow control and
regulation, agricultural machinery, production, and processing and irrigation scheduling.

269
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 1. Classification of Irrigation Control Strategies. Adapted from (Abioye et al., 2020b; Erion
et al., 2023).

CONCLUSION

In this review, the systematic review conducted here shows that the literature on smart irrigation
control strategies shows a significant importance in agricultural water use efficiency as well as
climate change resilience. Therefore, this is considered a rising research niche to which we can
continue to contribute from many points of view, as mentioned throughout the text. Regarding the
technological aspects of the analysed works, it became evident that embedded systems are preferred
in the implementation of smart irrigation system prototypes, which use technologies considered to
be of interest for this work, like Model Predictive Control. However, smart irrigation systems
usually involve a significant cost, affordability must not be forgotten, and that implies the
involvement of different stakeholders is a must as the availability and accessibility is not an issue.

REFERENCES

Abioye, E. A., Abidin, M. S. Z., Mahmud, M. S. A., Buyamin, S., Ishak, M. H. I., Rahman, M. K.
I. A., Otuoze, A. O., Onotu, P., & Ramli, M. S. A. (2020). A review on monitoring and
advanced control strategies for precision irrigation. Computers and Electronics in

270
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Agriculture, 173(August 2019). https://doi.org/10.1016/j.compag.2020.105441

Bwambale, E., Abagale, F. K., & Anornu, G. K. (2023). Model-based smart irrigation control
strategy and its effect on water use efficiency in tomato production. Cogent Engineering,
10(2). https://doi.org/10.1080/23311916.2023.2259217
Food and Agriculture Organization, F. (2017). The future of food and agriculture Trends and
challenges.
Klein, L. J., Hamann, H. F., Hinds, N., Guha, S., Sanchez, L., Sams, B., & Dokoozlian, N. (2018).
Closed Loop Controlled Precision Irrigation Sensor Network. IEEE Internet of Things
Journal, 5(6), 4580–4588. https://doi.org/10.1109/JIOT.2018.2865527
Saleem, S. K., Delgoda, D. K., Ooi, S. K., Dassanayake, K. B., Liu, L., Halgamuge, M. N., &
Malano, H. (2013). Model Predictive Control for Real-Time Irrigation Scheduling. IFAC
Proceedings Volumes, 46(18), 299–304. https://doi.org/10.3182/20130828-2-SF-
3019.00062
Vallejo-Gómez, D., Osorio, M., & Hincapié, C. A. (2023). Smart Irrigation Systems in Agriculture:
A Systematic Review. Agronomy, 13(2), 1–25. https://doi.org/10.3390/agronomy13020342

271
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

UPLAND RICE YIELD RESPONSE TO SOIL MOISTURE VARIABILITY WITH

DEPTH ACROSS FERRALSOLS AND GLEYSOLS IN WESTERN UGANDA
#11659

1*MatilaTom Micah; 1B. Kiconco; 1L. Abasize; 1P. Tamale; 1E. Opolot; 1P. Ebanyat; 2T.
Oberthur, 1P. Musinguzi
1School of Agricultural Sciences, Makerere University, Kampala, Uganda; 2African Plant

Nutrition Institute, Benguérir, Morrocco

*e-mail: [email protected]

ABSTRACT

Soil moisture is a vital factor in boosting rice productivity by influencing the growth of healthy
plants. In mid-western districts like Kikuube where rainfall is unpredictable, maintaining optimal
soil moisture differs between a bountiful harvest and crop failure. Effective soil moisture
management leads to improved water use efficiency, allowing crops to withstand periods of
drought. This study assessed upland yield response to soil moisture variations with soil depth in
Ferralsols and Gleysols for two seasons: (August to December 2023 and March to June 2024).
Twenty-four (24) access tubes were installed in six fields, each field having four (4); three (3) fields
of Ferralsols and three (3) for Gleysols, from which soil moisture content was measured using the
Diviner 2000 at 10 cm intervals to a 1m depth. Calibrated rain gauges were also installed in each
field to measure the daily rainfall received. Soil physical properties such as texture, bulk density,
infiltration rates, soil water holding capacity, field capacity, and permanent wilting point (PWP)
were determined. In Ferralsols, the soil moisture decreased with an increase in depth whereas in
Gleysols, the soil moisture increased with an increase in depth. The increase in soil moisture with
an increase in depth is attributed to the contribution of capillary rise for Gleysols. Gleysols
registered higher yields of 5,840kg ha-1 compared to 3,527kg ha-1 in the Ferralsols during the
March-June 2024 season which had a high rainfall variability. However, high yields (6,375 kg ha-
1) in Ferralsols were registered in the August-December 2023 growing season characterized by less

rainfall variability. Both Ferralsols and Gleysols are suitable for upland rice production.
Nevertheless, in high variability of rainfall, the continuous supply of water by capillarity in
Gleysols meets the crop water requirements unlike in Ferralsols.

INTRODUCTION

Upland rice cultivation is a prevalent agricultural practice in many regions of the world (Gadal et
al., 2019), including western Uganda, where it plays a crucial role in food security and income
generation for smallholder farmers (Agric, 2023). This crop exhibits a complex yield response to
soil moisture variability, particularly when grown in Ferralsols and Gleysols (Niang, 2019). These
soil types are characterized by their unique properties and moisture dynamics and play a crucial
role in the growth and productivity of upland rice. Ferralsols, typically found in tropical regions
with high rainfall, are deeply weathered, leached soils with a high content of iron and aluminum
oxides, which can significantly influence the water-holding capacity and, consequently, the
moisture availability of rice plants (S. Michael, 2023). On the other hand, Gleysols are often located
in low-lying areas and are prone to waterlogging due to poor drainage, affecting root development
and nutrient uptake in rice crops (Bado et al., 2018). In western Uganda, the bimodal rainfall pattern

272
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

contributes to the variability of soil moisture with depth, impacting the yield of upland rice across
these soil types. During the rainy seasons, Ferralsols provide adequate moisture for rice growth,
but the challenge arises in managing excess water in Gleysols to prevent detrimental effects on the
crop (Ojara et al., 2024). Conversely, in the dry seasons, the retention of soil moisture becomes
critical, especially in Ferralsols, to sustain the rice during periods of water scarcity. A Study (Singh
et al., 2017) has shown that upland rice varieties exhibit different physiological responses to soil
moisture stress, with some genotypes demonstrating tolerance by maintaining growth and yield
under varying moisture regimes. The ability of rice plants to adapt to moisture stress is linked to
traits such as root depth and density, which determine the extent of water uptake from different soil
layers (Sandhu et al., 2016). In Ferralsols, deeper root systems can access moisture from lower soil
horizons, while in Gleysols, rice varieties with a higher tolerance to waterlogged conditions may
fare better.

MATERIALS AND METHODS

Two seasons were considered in this study (August to December 2023 and February to June 2024).
Six experimental plots on both Ferralsols (3) and Gleysols (3) were set up in Kikuube District. For
each plot, four fertilizer treatments were applied. These included full package fertilizer
recommendation (20kg N/ha + 30kg K/ha of DAP at planting, 20kg N/ha (Urea) + 40kg K/ha
(MoP) at tillering and 20 kg N/ha (Urea) + 40kg K/ha (MoP) at panicle formation), half package
(10kg N/ha + 15kg K/ha of DAP at planting, 10kg N/ha (Urea) + 20kg K/ha (MoP) at tillering and
10 kg N/ha (Urea) + 20kg K/ha (MOP) at panicle formation), quarter package (5kg N/ha + 7.5kg
K/ha of DAP at planting, 5kg N/ha (Urea) + 10kg K/ha (MoP) at tillering and 5 kg N/ha (Urea) +
10kg K/ha (MoP) at panicle formation), and the Farmer management practices (Control). In each
plot, four access tubes and one fabricated rain gauge were installed for reading soil moisture data
by the Diviner 2000 at an interval of 10 cm to 100cm depth and recording on-site daily rainfall
respectively. The Diviner 2000 was first calibrated to local soil conditions, likewise, fabricated rain
gauges were also calibrated using the standard rain gauge at Makerere University Weather Station
(the ratio of the value of the standard rain gauge to the value of the fabricated rain gauge) forming
the calibration factor. Soil moisture data was read at least four times a week at an interval of one
day. Weed management was done effectively from the time of planting by application of Butanil-
S as a pre-emergence herbicide at the time of sowing and Butanil N70 + Butanil 2-4D as post-
emergence herbicides. Pest management was also done using Roket type of pesticide only where
there were cases of pests. The general rice yield was considered for this study as presented in the
results below.

RESULTS AND DISCUSSION

Yield
There is a variation of upland rice production in the two soil types in Kikuube district. In both
seasons 2023B, Ferralsols exhibited a higher yield (6,375kg/ha) than Gleysols (5,714kg/ha)
whereas, in season 2024A, Gleysols had a higher yield (5,840kg/ha) than Ferralsols (3,527kg/ha).
However, the variation in yields in both soil types and seasons is insignificant at p<0.05 (Figure
1).

273
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 1. Rice yield for two seasons.

Soil moisture variation

Generally, soil moisture decreased with an increase in depth in Ferralsols whereas in Gleysols, the
soil moisture content increased with an increase in soil depth (Figures 2 and 3 respectively).

DISCUSSION

Ferralsols and Gleysols are two distinct soil types that exhibit unique moisture variation profiles
due to their inherent properties. The presence of micro-aggregates in Ferralsols enhances moisture
storage at field capacity, which is crucial for crops like upland rice that rely on consistent moisture
availability for optimal growth. The variation in soil moisture with depth (Increase with depth)
Figure 2(A) in Gleysols significantly impacts the root development of upland rice, as the roots are
not able to penetrate deeply enough to access the moisture in lower layers during drier periods. The
effect of soil moisture on upland rice yield is profound since it is typically grown in rain-fed
conditions, highly sensitive to soil moisture variability. Studies have shown that soil moisture stress
during critical growth stages, such as panicle development, severely impacts the growth and yield
of upland rice. For instance, improved upland rice varieties like NAMCHE 5 have demonstrated
excellent performance under limited soil moisture conditions through early heading and maturity,
contributing to higher grain yield. This is mainly due to their ability to produce heavier straw yield,
an abundant number of productive tillers, higher filled spikelets, and heavier weight of seeds, which
collectively enhance the harvest index and adjust to the soil moisture conditions. Furthermore,
climate variability has been found to influence soil moisture levels and, consequently, rice
production. Rain-fed upland rice systems are more vulnerable to these variations than irrigated
paddy rice, with about 10% of the variance in rice production anomalies on a national level co-
varying with soil moisture changes.

274
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

(A)
Moisture Content (vol/vol)
-10 0 10 20 30 40 50
0

20
Soil Depth (cm)

40

60 August
September 2024A
80 October
November
December
100
(B)
Moisture Content (vol/vol)
0 10 20 30 40
0

20
Soil Depth (cm)

40
August
60 September
October
2023B
80

100

Figure 2. Variation of soil moisture with depth; (A) Gleysols (B) Ferralsols.

CONCLUSION

Understanding the dynamics of soil moisture in Ferralsols and Gleysols is essential for predicting
and managing the yield of upland rice, which is a staple food for millions of people worldwide.
Effective water management strategies, including the selection of rice varieties with drought
tolerance and the timing of planting to coincide with optimal soil moisture conditions, are critical
for sustaining rice production in the face of changing climate patterns and soil moisture regimes.

REFERENCES

Agric, F. (2023). https://doi.org/10.18697/ajfand.124.23140. 23(9), 24507–24522.

Bado, V. B., Djaman, K., & Mel, V. C. (2018). Developing fertilizer recommendations for rice in
Sub-Saharan Africa, achievements and opportunities. Paddy and Water Environment, 16(3),
571–586. https://doi.org/10.1007/s10333-018-0649-8
Gadal, N., Shrestha, J., & Pokharel, B. B. (2019). A review on production status and growing
environments of rice in Nepal and in the world A review on production status and growing
environments of rice in Nepal and in the world. March.
https://doi.org/10.26832/24566632.2019.0401013
275
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Niang, A. (2019). Rice yield gaps in West Africa Abibou Niang.

Ojara, M. A., Namumbya, S., Ogwang, B. A., & Limbu, P. T. S. (2024). Regional and farm level
droughts characteristics for southwestern Uganda during 1981–2017. Soil Security, 15(June),
100148. https://doi.org/10.1016/j.soisec.2024.100148
S. Michael, P. (2023). the Importance of Sustainable Management of Acid Soils in the Humid
Tropics Under Climate Change and the Future Research Directions. In Ecofeminism and
Climate Change (Vol. 4, Issue 1). https://doi.org/10.26480/efcc.01.2023.29.40
Sandhu, N., Anitha Raman, K., Torres, R. O., Audebert, A., Dardou, A., Kumar, A., & Henry, A.
(2016). Rice root architectural plasticity traits and genetic regions for adaptability to variable
cultivation and stress conditions. Plant Physiology, 171(4), 2562–2576.
https://doi.org/10.1104/pp.16.00705
Singh, B., Reddy, K. R., Redoña, E. D., & Walker, T. (2017). Screening of Rice Cultivars for
Morpho-Physiological Responses to Early-Season Soil Moisture Stress. Rice Science, 24(6),
322–335. https://doi.org/10.1016/j.rsci.2017.10.001

276
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

RAINWATER HARVESTING AND NUTRIENT INTENSIFICATION IN MAIZE-

LEGUME FARMING SYSTEMS IN SEMI-ARID ZIMBABWE
#11639

*E.Mutsamba-Magwaza, D. Nyamayevu, I. Nyagumbo, R. Mandumbu, G. Nyamadzawo

International Maize and Wheat Improvement Centre (CIMMYT)-Zimbabwe, 12.5 km Peg
Mazowe Road, Harare, Zimbabwe
*e-mail: [email protected]; +263773047109

ABSTRACT

Agricultural productivity in Zimbabwe is declining mainly due to climate change, high cost of
fertilizers and inherently poor soil fertility. In response to these challenges, most smallholder
farmers are implementing either rainwater harvesting (RWH) or integrated soil fertility
management (ISFM). This study sought to investigate the role of integrating the tied-contour RWH
(TC-RWH) technique and ISFM on soil moisture, soil fertility, crop growth, and subsequent crop
yields in semi-arid areas of Zimbabwe. A split-split-split plot design was established where water
harvesting technologies namely tied contour (TC) and standard contour (SC) were considered as
the main plots, cropping system (sole maize, sole cowpea and maize-cowpea intercrop) as sub-
plots, different N-levels as sub-sub plots and manure as sub-sub-sub plots. Manure application
demonstrated a higher grain yield advantage by 297 kg ha-1 over treatments without manure.
Similarly, top dressing yielded more grain yield by 1146 kg ha-1 than untopdressed plots.
Intercropping had a total systems’ biomass yield 4397 kg ha-1 while cowpea sole and maize sole
yielded 3863 and 3247 kg ha-1 respectively. Standard contours without manure had the least total
biomass output. With respect to soil moisture, sole cowpea under SC retained more moisture than
intercropped plots and sole maize during first season. The Land Equivalent Ratio of 1.3 kg ha -1
obtained signifies a greater land productivity and efficiency realized through intercropping
compared to sole cropping systems. The study shows that when rainfall is abundant, as in the first
season, sole cowpea conserves more moisture under standard contours than intercrops and sole
maize. However, when conditions are dry, as in the case of the second season, sole and intercrops
performed similarly. On the other hand, under excessively dry environments, sole maize systems
with lower plant populations conserve moisture than sole cowpeas and maize-cowpea intercrop
while in wet conditions, intercrops conserve more moisture than sole crops. Overall, in the sandy
soils, failure to apply top dressing and manure resulted in serious yield penalties.

INTRODUCTION

The majority of smallholder farmers in Zimbabwe are located in semi-arid areas where rainfall is
low (Mupangwa, Makanza, et al. 2021), and soils are inherently poor. Of the little rain that is
received in the semi-arid regions, most is lost as runoff, and very little water is harvested for plant
growth or future use (Nyamadzawo et al. 2012). Expansion of cultivatable areas and the practice
of crop rotations have been ways of increasing crop productivity (Mupangwa et al., 2021;
Nyagumbo et al., 2016) in Sub-Saharan Africa (SSA), but is no longer sustainable with SSA’s
population expected to increase by 2050 (Trisos et al., 2022). Additionally, exorbitant prices of
inorganic fertilizers resulted in many farmers applying no to low rates of inorganic fertilizers,
thus leading to poor yields.

277
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Some smallholder farmers are already practicing water harvesting (Madamombe et al. 2024), and
maize legume intercropping/rotations (Thierfelder et al. 2024), whilst adding organic and or
inorganic fertilizers (Mutsamba, Nyagumbo, and Mupangwa 2019). To increase agricultural
productivity in these areas, there may be a need to combine various techniques such as in-field
water harvesting techniques, nutrient intensification and integrated nutrient management in semi-
arid conditions. Specifically, this study investigated the role of integrating tied-contour rainwater
harvesting (TC-RWH) technique and ISFM on soil moisture, soil fertility, crop growth and
subsequent crop yields in semi-arid areas of Zimbabwe.

MATERIALS AND METHODS

The study was carried out in Mutoko District located in the semi-arid region of Zimbabwe. Mutoko
receives annual rainfall of 450-600 mm and mean temperatures of about 35oC. The soils (Table 1)
are mainly predominated by low fertile inherent granitic sandy soil (Lixisols: (FAO 2014). The
suitable farming systems in these regions are semi-intensive farming systems with a mixture of
crops and livestock.

Table 8. Soil characteristics of the experimental site before trial establishment.

pH N P K Ca Mg
% Carbon Colour Texture
(CaCl2) (ppm) (ppm) (meq/100g) (meq/100g) (meq/100g)
1.19 PB Mgs 5.05 29.5 37.5 0.15 1.35 0.49
Where PB =Pale brown colour and Mgs = Medium grained sand

The experiment was laid out in a split-split-split plot design, replicated three times on one on-farm
field from 2020/21 to 2021/22 agricultural seasons. Water harvesting structures (tied and standard
contours) were the main plots, cropping systems (maize-cowpea intercrops and the respective sole
crops) were the sub-plots, N-fertiliser rates (top dressed and untopdressed) were the sub-sub plots,
and the manure application (with or without) were the sub-sub-sub plots (Fig 1b).

Tied contours (Fig 1a) measuring 0.3 m deep and 1 m wide were prepared at a slope of 1:250. The
cross ties were placed after every 5 m with a breadth of 0.5 m and a small opening was made at the
upper side of the tie to allow water to flow from one compartment to another when it was full. SCs
were made at a gradient of 1:250 with the same measurements as of tied contours. SCs are existing
water channels designed at 1:250 gradient to dispose of field water. Both TCs and SCs were spaced
at 8 m. Top dressing of ammonium nitrate (34.5% N) was applied to maize only at the different
specified rates 3-4 weeks after planting.

Crop height, chlorophyll, NDVI and soil moisture were measured using a meter rule, chlorophyll
meter (Apogee Instruments, MC 10), handheld green seeker (SPL technologies and a Field Scout TM
TDR 300 soil Moisture Meter (Spectrum Technologies, Inc.) respectively. Daily rainfall was
recorded using a rain gauge mounted at the experimental site. Analysis of variance (ANOVA)
across treatments was conducted using R-software to determine the effects of tied contours,
fertiliser and manure use on chlorophyll, NDVI, plant height and grain yield. For significant
treatment*season interactions, each season was the analysed separately.

278
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

a b

Figure 1. Picture of a) tied ridges b) cropping systems (maize-cowpea intercrop and sole cowpeas)
and maize with and without top dressing.

RESULTS AND DISCUSION

The 2020-21 and 2021-22 seasons received a total annual rainfall of 684 mm and 461 mm
respectively. There 4-way interaction between the cropping systems, water harvesting techniques,
manure application and top-dressing application was insignificant. Across the two cropping
seasons, it was observed that manure application demonstrated a higher grain yield advantage by
297 kg ha-1 over treatments without manure. Similarly, top dressing yielded more grain yield by
1146 kg ha-1 than untopdressed plots. In addition, the application of top dressing significantly
increased total biomass yield compared to untopdressed plots. However, there was no significant
difference in total above-ground biomass in plots applied manure and those without manure. The
results clearly indicate that despite the importance of inorganic fertilisers, it is crucial to acquire
nutrients from diverse sources, including organic materials like livestock manure and nitrogen-
fixing legumes (Sanginga and Woomer 2009). The use of both organic and inorganic inputs is
important because both resources fulfil different functions towards plant growth and neither of
them is available or affordable in sufficient quantities (Vanlauwe et al. 2015).

Total biomass yield was significantly affected by cropping system, where maize-cowpea intercrop
had significantly higher output than sole maize, concurring with (Mutsamba, Nyagumbo, and

279
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Mupangwa 2019). Intercropping yielded 4397 kg ha-1 while cowpea sole and maize sole yielded
3863 and 3247 kg ha-1 respectively. Standard contours without manure had the least total biomass
output. In other studies, tied ridging and rip & potholing yielded 25% more grain yield than
conventional mouldboard ploughing (Nyagumbo and Bationo, 2011). The Land Equivalent Ratio
(LER) of 1.3 kg ha-1 obtained signifies a greater land productivity and efficiency realized through
intercropping compared to sole cropping systems, concurring with (Bitew and Abera 2019;
Mutsamba, Nyagumbo, and Mupangwa 2019).

On soil moisture, sole cowpea under SC retained more moisture than intercropped plots and sole
maize during first season. This can be attributed to the ability of cowpea reducing evaporation from
the soil surface. As sole cowpeas’ plant populations doubled that of sole crops, it translated to
higher ground coverage and there was no competition for moisture due to adequate rains received.
However, during the second season under SC, all cropping systems performed similarly due to very
low rainfall amounts received. Conversely, under TC, intercrop systems had higher soil moisture
content compared to the sole cowpea and maize during first season, while sole maize had higher
soil moisture compared to intercrops during the second season. Given the dry conditions
experienced during the second season, sole maize retained more moisture compared to sole cowpea
and intercrop systems. This may be due to high plant densities in intercrops and sole cowpea
competing for soil moisture in the root zone. This can be attributed to high plant densities in
intercrops and sole cowpea competing for soil moisture in the root zone, as studies have reported
that intercrops often have lower soil moisture than sole crops due to greater root moisture extraction
(Eskandari and Kazemi 2011). This contradicts with (Ghanbari et al. 2010) who showed that maize
monocrop tend to have a lower soil moisture content compared to sole cowpea due to high soil
water losses through evapotranspiration..

Top dressed plants under TC displayed greater heights compared to plants under SC. This is
attributed to the water harvesting under TCs compared to SCs which disposed water due to
gradient. Manure application led to taller plants during second seasons. Intercropped plots had
higher NDVI than sole maize indicating the benefits of biological nitrogen fixation (Franke et al.
2018). NDVI was measured during the second season only when the green seeker was made
available. In conclusion, the benefits of water harvesting are determined by rainfall received.

REFERENCES

Bitew, Yayeh, and Merkuz Abera. 2019. “Conservation Agriculture Based Annual Intercropping
System for Sustainable Crop Production: A Review.” Indian Journal of Ecology 46(2): 235–
49.
Eskandari, Hamdollah, and Kamyar Kazemi. 2011. “Weed Control in Maize-Cowpea
Intercropping System Related to Environmental Resources Consumption.” Notulae Scientia
Biologicae 3(1): 57–60.
FAO. 2014. World Reference Base for Soil Resources’ International Soil Classification System.
Franke, A. C., G. J. van den Brand, B. Vanlauwe, and K. E. Giller. 2018. “Sustainable
Intensification through Rotations with Grain Legumes in Sub-Saharan Africa: A Review.”
Agriculture, Ecosystems and Environment 261(September 2017): 172–85.
doi:10.1016/j.agee.2017.09.029.
Fredriksen, Birger, and Ruth Kagia. 2013. “Attaining the 2050 Vision for Africa: Breaking the
Human Capital Barrier.” Global Journal of Emerging Market Economies 5(3): 269–328.
280
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Ghanbari, Ahmad, Mehdi Dahmardeh, Barat Ali Siahsar, and Mahmoud Ramroudi. 2010. “Effect
of Maize (Zea Mays L.)-Cowpea (Vigna Unguiculata L.) Intercropping on Light Distribution,
Soil Temperature and Soil Moisture in Arid Environment.” Journal of Food, Agriculture &
Environment 8(1): 102–8.
Madamombe, Sandra Makaita, Stanley Karanja Ng’ang’a, Ingrid Öborn, George Nyamadzawo,
Ngonidzashe Chirinda, Job Kihara, and Libère Nkurunziza. 2024. “Climate Change
Awareness and Adaptation Strategies by Smallholder Farmers in Semi-Arid Areas of
Zimbabwe.” International Journal of Agricultural Sustainability 22(1): 2293588.
Mupangwa, W., R. Makanza, L. Chipindu, M. Moeletsi, S. Mkuhlani, F. Liben, I. Nyagumbo, and
M. Mutenje. 2021. “Temporal Rainfall Trend Analysis in Different Agro-Ecological Regions
of Southern Africa.” Water SA 47(4): 466–79. doi:10.17159/WSA/2021.V47.I4.3844.
Mupangwa, W., I. Nyagumbo, F. Liben, L. Chipindu, P. Craufurd, and S. Mkuhlani. 2021. “Maize
Yields from Rotation and Intercropping Systems with Different Legumes under Conservation
Agriculture in Contrasting Agro-Ecologies.” Agriculture, Ecosystems and Environment 306.
doi:10.1016/j.agee.2020.107170.
Mutsamba, E. F., I. Nyagumbo, and W. Mupangwa. 2019. “Forage and Maize Yields in Mixed
Crop-Livestock Farming Systems: Enhancing Forage and Maize Yields in Mixed Crop-
Livestock Systems under Conservation Agriculture in Sub-Humid Zimbabwe.” NJAS -
Wageningen Journal of Life Sciences (November). doi:10.1016/j.njas.2019.100317.
Nyagumbo, Isaiah, and André Bationo. “Exploring Crop Yield Benefits of Integrated Water and
Nutrient Management Technologies in the Desert Margins of Africa : Experiences from
Semi-Arid Zimbabwe.” : 1–23.
Nyagumbo, Isaiah, Siyabusa Mkuhlani, Charity Pisa, Donwell Kamalongo, Domingos Dias, and
Mulugetta Mekuria. 2016. “Maize Yield Effects of Conservation Agriculture Based
Maize???Legume Cropping Systems in Contrasting Agro-Ecologies of Malawi and
Mozambique.” Nutrient Cycling in Agroecosystems 105: 275–90. doi:10.1007/s10705-015-
9733-2.
Nyamadzawo, G., P. Nyamugafata, M. Wuta, J. Nyamangara, and R. Chikowo. 2012. “Infiltration
and Runoff Losses under Fallowing and Conservation Agriculture Practices on Contrasting
Soils, Zimbabwe.” Water SA 38(2): 233–40. doi:10.4314/wsa.v38i2.8.
Ofori, Samuel Appiah, Samuel Jerry Cobbina, and Samuel Obiri. 2021. “Climate Change, Land,
Water, and Food Security: Perspectives From Sub-Saharan Africa.” Frontiers in Sustainable
Food Systems 5: 680924. doi:10.3389/fsufs.2021.680924.
Sanginga, N., and P.L. (eds.). Woomer. 2009. Integrated Soil Fertility Management in Africa:
Principles, Practices and Developmental Process. Nairobi: Tropical Soil Biology and Fertility
Institute of the International Centre for Tropical Agriculture.
Thierfelder, Christian, Blessing Mhlanga, Isaiah Nyagumbo, Kelvin Kalala, Esau Simutowe,
Mazvita Chiduwa, Chloe MacLaren, João Vasco Silva, and Hambulo Ngoma. 2024. “Two
Crops Are Better than One for Nutritional and Economic Outcomes of Zambian Smallholder
Farms, but Require More Labour.” Agriculture, Ecosystems & Environment 361: 108819.
Trisos, Christopher, Ibedun Adelekan, Edmund Totin, A Ayanlade, J Efitre, A Gemeda, K Kalaba,
et al. 2022. “Africa. In: Climate Change 2022: Impacts, Adaptation and Vulnerability.
Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental
Panel on Climate Change.”
Vanlauwe, B., K. Descheemaeker, K. E. Giller, J. Huising, R. Merckx, G. Nziguheba, J. Wendt,
and S. Zingore. 2015. “Integrated Soil Fertility Management in Sub-Saharan Africa:
Unravelling Local Adaptation.” Soil 1(1): 491–508. doi:10.5194/soil-1-491-2015.
281
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

FACTORS INFLUENCING FARMERS’ DECISIONS ON TIMING AND APPLICATION

RATES OF IRRIGATION WATER IN MWALA, MACHAKOS COUNTY
#11284

D.M. Onyancha, C.K.K. Gachene, and G. Kironchi

University of Nairobi, Nairobi, Kenya
e-mail: deborahmongí[email protected]; +254 713 377 983

ABSTRACT

This study examines the factors influencing farmers' water application decisions in arid and semi-
arid regions, where water scarcity is a significant challenge due to unreliable rainfall. The research
involved interviews with 41 farmers from the Equity Group Foundation extension scheme,
focusing on their irrigation practices, including scheduling, timing, application rates, and
considerations of plant and soil conditions. Additionally, soil samples from farms using furrow,
drip, sprinkler, and hosepipe irrigation systems were analyzed, with field measurements taken to
estimate water application rates. The findings reveal that farmers prioritize plant conditions (97%
degree of importance) when deciding when to irrigate, followed by soil conditions and days since
the last irrigation (both 95%), with recent rainfall also playing a key role (87%). For determining
the amount of water to apply, soil conditions were deemed most important (80% importance),
followed by plant conditions (75%) and recent temperature (70%). The study also found that water
application amounts often exceeded the crops’ requirements across all irrigation methods, with
furrow irrigation having the highest flow rate (0.479 litres per second). The soils were found to
have high coarse-textured contents, resulting in low moisture retention capacities and high
hydraulic conductivity. The study concludes that proper irrigation scheduling is essential for
effective water use and recommends enhancing extension services to provide farmers with training
on soil water holding capacity, crop water needs, and plant characteristics to guide irrigation
decisions.

Key words: Irrigation scheduling, Plant conditions, Soil conditions, Irrigation requirement

INTRODUCTION

Water scarcity and the growing competition for resources, especially in arid and semi-arid regions,
have made efficient water use in agriculture increasingly critical (Priyan, 2021). To address this
challenge, proper irrigation scheduling has become essential for minimizing water wastage. For
farmers to make informed decisions on how often and how much water to apply, they must have
accurate information about crop water use and soil moisture content (Ara et al., 2021; Lakhiar et
al., 2024). Several key factors influence irrigation decisions, including the properties of the soil,
characteristics of the crops, availability of water, and climatic conditions such as rainfall and
temperature (Gu et al., 2020; Zinkernagel et al., 2020). Understanding soil type and its water-
holding capacity is vital for determining the frequency of irrigation, as different crops require
varying amounts of water throughout their growth cycles, depending on environmental conditions
(Pereira et al, 2021, Plett et al., 2020).

282
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Effective irrigation scheduling not only optimizes the use of water but also ensures the efficient
use of energy and other agricultural inputs, leading to improved crop yields, enhanced crop quality,
and lower production costs (Ray and Majumder, 2024, Lakhiar et al., 2024). Despite these benefits,
research has shown that many farmers fail to practice proper irrigation scheduling, often due to a
lack of understanding of crop water needs or because water is perceived as being inexpensive (Sun
et al., 2022; Fernández García et al., 2020). This study explores the factors that influence farmers'
decisions regarding the timing and application rates of irrigation water in Mwala, Machakos
County, Kenya.

MATERIALS AND METHODS

Study design and data collection

The study was realized from survey statistics and field measurements done at the study area,
Mwala, Machakos County. The specific irrigation scheduling technique adopted by farmers in
irrigating crop(s), the factors influencing their timing and application rates, and any additional
information including the plant and soil conditions, planting dates and the farmers’ source of
technical advice was obtained in form of an interview questionnaire. This was administered to the
forty-one purposefully selected farmers who are registered by the Equity Group Foundation
extension scheme.

Laboratory soil analyses and field assessments of irrigation timing and application rates
Baseline soil characterization was done through laboratory analyses of disturbed and undisturbed
samples collected from farms using furrow, drip, sprinkler, and hosepipe irrigation systems,
focusing on soil texture, organic carbon, hydraulic conductivity, moisture retention, and bulk
density. The data collected to estimate the irrigation timing and application rates for the various
irrigation methods in the field measurements included: flow rate, the volumes of water applied and
dimensions of the irrigation units. The flow rate (Q) was assessed by measuring the time (T)
required to fill containers of a known volume (V) for drip, sprinkler and hosepipe irrigation systems
using Equation 1 as described by Trimmer (1994):

Δ𝑉 Equation 1
𝑄=
Δ𝑡

where: Q= flow rate in liter per second (l/s), V= volume in liters (l), t= time in seconds (s)

Furrow irrigation timing and application rates were estimated by measuring furrow lengths,
recording water flow times at various stations, and periodically measuring inflow and outflow rates.
The final inflow-outflow measurements and the maximum depth of flow were then recorded
according to the methods as outlined by Vázquez et al. (2005) (Equation 2)

Q×t=d×A Equation 2

where: Q= flow rate, in liters per second (l/s); t= set time or total time of irrigation (s); d= depth of
water applied (mm) and A= area irrigated (m2).

Soil samples were collected at different depths along furrows to estimate the wetting front by
measuring moisture content differences, with this process replicated three times on three farmers’
283
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

fields. Measurement of the amount of irrigation water applied (d) in mm is deduced by (Equation
3):

𝑄(𝑙/𝑠) Equation 3
𝑑(𝑚𝑚) =
𝐴(𝑚2 )

where: Q= flow rate, in liters per second (l/s); d= depth of water applied (mm) and A= area irrigated
(m2).

The collected data was analyzed using ANOVA in Genstat 15th version, with significance
determined at P ≤ 0.05, and further analysis was conducted using SPSS (Ver 21)

RESULTS AND DISCUSSION

Timing of irrigation application rates

The results of the analysis of farmers’ responses showed that majority ranked plant conditions
(97%), soil conditions (95%), days since last irrigation (95%), and recent rainfall (88%) as the most
critical factors in determining when to irrigate using a Likert scale. Key soil conditions included
moisture, soil type, and topography, while plant factors included crop type, drought sensitivity,
development stage, and fruiting. Farmers varied their irrigation schedules based on weather
conditions and relied on personal experience, extension personnel, and consultants for guidance.
These findings are consistent with recent studies by Dong (2022) and Owino and Söffker (2022)
which highlighted the importance of soil and plant conditions, weather, and evapotranspiration in
efficient irrigation scheduling.

Factors Influencing Irrigation Application Amounts

Regarding the amount of water applied, farmers identified soil moisture, water-holding capacity,
soil texture, topography, and plant conditions (such as stress, growth stage, rooting depth, and crop
yield) as the most important factors. However, while all farmers timed their irrigation, 60% did not
monitor the exact amount of water applied, often relying on rough estimates. Only 40% carefully
checked water application by visually assessing soil wetness or timing the emptying of tanks. This
lack of precise monitoring is concerning in water-scarce regions like Mwala, Machakos County, as
it could lead to either under- or over-irrigation, thereby failing to meet crop needs or wasting water.
These findings align with studies by Singh et al. (2024) and Nesamvuni et al. (2022), which
reported that farmers often apply excess water due to perceived yield benefits or the inconvenience
of irrigation scheduling. However, careful water management is crucial for maximizing irrigation
efficiency, conserving water, and reducing energy use (Ray and Majumder, 2024, and Kilemo,
2022).

Measurement of Irrigation Water

The results as illustrated in Figure 1 show that furrow irrigation exhibited the highest quantities of
irrigation water application per irrigation event as compared to the other systems. Drip had the least
but required more applications especially for mangoes and oranges. This data collection on
irrigation monitoring was done in the mid growth stage of the crops during the dry season. The
application amounts were then compared to the CROPWAT computed data on irrigation water
requirements to assess whether the farmers were practicing efficient irrigation application. The

284
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

lowest average differences between applied water and the irrigation requirements were observed
in vegetables, which was 8 mm and bananas 3 mm, and the largest in mangoes, 75 mm and oranges,
64 mm; followed by French beans, 36 mm, maize, 25 mm, and tomatoes, 18 mm. From these
comparisons, it is revealed that farmers were irrigating more than was required on the crops
especially on oranges and mangoes.

Field measurements revealed significant differences in flow rates among irrigation systems, with
furrow irrigation exhibiting the highest flow rate (0.479 l/s), use of hosepipe, 0.232 l/s, sprinkler,
0.074 l/s and drip irrigation the lowest (0.003 l/s). The lower flow rates in drip and sprinkler
systems are attributed to their more controlled water application methods, making them more
efficient compared to furrow and hosepipe systems. These findings are supported by studies such
as Asif et al. (2024) and Olamide et al. (2022), which demonstrated that drip irrigation systems use
less water and are more efficient than furrow systems.

Irr. Req. Drip Sprinkler Hosepipe Furrow

120
Applied water in mm/day

100
80
60
40
20
0
Maize French beans Tomatoes Vegetables Bananas Mangoes Oranges

Figure 13. Comparison of applied water and irrigation requirements of various crops.

Soil-Water Relationships
Soil properties under different irrigation systems were also examined, revealing significant
differences in sand content, bulk density, hydraulic conductivity, and soil moisture retention (Table
1). Soils under hosepipe irrigation exhibited the highest sand content and hydraulic conductivity,
leading to faster water infiltration and lower water retention capacity. This contrasts with drip
irrigation, which showed lower sand content and higher water retention. These findings align with
studies by Rivier et al. (2022) and Wang et al. (2021), which highlighted the impact of soil texture
on hydraulic properties and irrigation efficiency. Additionally, the low organic carbon content in
soils (ranging from 0.68% to 0.91%) was attributed to the arid and semi-arid conditions of the study
area, as noted by Lei et al. (2022) and Hag Husein et al. (2021). The lack of organic matter
application by farmers further exacerbated this issue, emphasizing the need for improved soil
management practices to enhance irrigation efficiency.

285
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 1. Soil properties under different irrigation methods.

Irrigation methods
L.S. D p
Soil properties Drip Furrow Hosepipe Sprinkler Value
Hydraulic conductivity (cm/hr) 0.48a 0.69b 3.09d 2.36c 0.09 <.001
Soil moisture retention (vol%) 95.80c 92.67c 81.00a 87.89b 3.54 <.001
Bulk density (g/cm3) 1.31a 1.47b 1.65c 1.54b 0.08 <.001
Organic carbon (%) 0.76ab 0.91b 0.68a 0.84b 0.15 0.037
Sand (%) 64a 72bc 77c 71b 6.10 0.01
Clay (%) 29b 22ab 20a 23a 4.77 0.011
Silt (%) 7a 6a 3a 6a 3.65 0.105

CONCLUSIONS AND RECOMMENDATIONS

The study assessed factors influencing farmers' decisions on irrigation timing and water
application, revealing that plant and soil conditions, recent rainfall, and days since last irrigation
were key determinants. Results showed significant differences in water application rates and net
irrigation requirements across irrigation methods, with furrow and hosepipe methods applying the
most water due to higher flow rates. Farmers' limited knowledge on proper irrigation scheduling
often led to excessive water use, which could be mitigated by improving extension services,
training on soil and crop characteristics, and encouraging better water management practices, such
as reducing irrigation time and applying manure to enhance soil water retention.

REFERENCES

Ara, I., Turner, L., Harrison, M. T., Monjardino, M., DeVoil, P., & Rodriguez, D. 2021.
Application, adoption and opportunities for improving decision support systems in irrigated
agriculture: A review. Agric. Water Manag. 257. 107161.
Dong, Y. 2022. Irrigation Scheduling Methods: Overview and Recent Advances. In Irrig. Drain.
—Recent Advances. IntechOpen. https://doi.org/10.5772/intechopen.107386
Fernández García, I., Lecina, S., Ruiz-Sánchez, M. C., Vera, J., Conejero, W., Conesa, M. R., ... &
Montesinos, P. 2020. Trends and challenges in irrigation scheduling in the semi-arid area of
Spain. Water. 12(3), 785.
Gu, Z., Qi, Z., Burghate, R., Yuan, S., Jiao, X., & Xu, J. 2020. Irrigation scheduling approaches
and applications: A review. J. Irrig. Drain. Eng. 146(6), 04020007.
Hag Husein, H., Lucke, B., Bäumler, R., & Sahwan, W. 2021. A contribution to soil fertility
assessment for arid and semi-arid lands. Soil Syst. 5(3), 42.
Lakhiar, I. A., Yan, H., Zhang, C., Wang, G., He, B., Hao, B., ... & Rakibuzzaman, M. 2024. A
Review of Precision Irrigation Water-Saving Technology under Changing Climate for
Enhancing Water Use Efficiency, Crop Yield, and Environmental
Footprints. Agriculture. 14(7), 1141.
Lei, Z. H. O. U., Xu, S. T., Monreal, C. M., Mclaughlin, N. B., Zhao, B. P., Liu, J. H., & Hao, G.
C. 2022. Bentonite-humic acid improves soil organic carbon, microbial biomass, enzyme
activities and grain quality in a sandy soil cropped to maize (Zea mays L.) in a semi-arid
region. J. Integr. Agric. 21(1), 208-221.

286
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Nesamvuni, A. E., Tshikolomo, K. A., Mpandeli, N. S., de Bruyn, M., Hlophe-Ginindza, S., & van
Niekerk, J. 2022. Perceptions on irrigation water supply and utilisation by smallholder
agricultural enterprises in Vhembe district of Limpopo Province, South Africa. Technium
Soc. Sci. J., 27, 968.
Olamide, F. O., Olalekan, B. A., Tobi, S. U., Adeyemi, M. A., Julius, J. O., Oluwaseyi, F. K.,
Olamide, F. O., Olalekan, B. A., Tobi, S. U., Adeyemi, M. A., Julius, J. O., & Oluwaseyi, F.
K. 2022. Fundamentals of Irrigation Methods and Their Impact on Crop Production. In Irrig.
Drain. —Recent Advances. IntechOpen. https://doi.org/10.5772/intechopen.105501
Owino, L., & Söffker, D. 2022. How much is enough in watering plants? State-of-the-art in
irrigation control: Advances, challenges, and opportunities with respect to precision
irrigation. Front. Control Eng. 3. https://doi.org/10.3389/fcteg.2022.982463
Pereira, L. S., Paredes, P., Hunsaker, D. J., López-Urrea, R., & Shad, Z. M. 2021. Standard single
and basal crop coefficients for field crops. Updates and advances to the FAO56 crop water
requirements method. Agric. Water Manag. 243, 106466.
Plett, D. C., Ranathunge, K., Melino, V. J., Kuya, N., Uga, Y., & Kronzucker, H. J. 2020. The
intersection of nitrogen nutrition and water use in plants: new paths toward improved crop
productivity. J. Exp. Bot. 71(15), 4452-4468.
Priyan, K. 2021. Issues and challenges of groundwater and surface water management in semi-arid
regions. Groundwater Resour. Dev. Plan. 1-17.
Ray, S., & Majumder, S. 2024. Water management in agriculture: Innovations for efficient
irrigation. Modern Agronomy; Sil, P., Chhetri, P., Majumder, S., Santosh, DT, Eds, 169-185.
Rivier, P. A., Jamniczky, D., Nemes, A., Makó, A., Barna, G., Uzinger, N., ... & Farkas, C. 2022.
Short-term effects of compost amendments to soil on soil structure, hydraulic properties, and
water regime. J. Hydrol. Hydromech. 70(1), 74-88.
Singh, K., Jhorar, R. K., Sidhpuria, M. S., Kumar, M., & Mehla, M. K. 2024. Impacts of improved
irrigation application efficiency and groundwater levels on energy use for crop production in
a semi-arid region of India. Environ. Dev. Sustain. 26(4), 10145–10162.
https://doi.org/10.1007/s10668-023-03139-7
Sun, X., Ritzema, H., Huang, X., Bai, X., & Hellegers, P. 2022. Assessment of farmers' water and
fertilizer practices and perceptions in the North China Plain. Irrig. Drain. 71(4), 980-996.
Wang, D., Wang, Z., Zhang, J., Zhou, B., Lv, T., & Li, W. 2021. Effects of soil texture on soil
leaching and cotton (Gossypium hirsutum l.) growth under combined irrigation and drainage.
Water, 13(24), 3614.
Zinkernagel, J., Maestre-Valero, J. F., Seresti, S. Y., & Intrigliolo, D. S. 2020. New technologies
and practical approaches to improve irrigation management of open field vegetable
crops. Agric. Water Manag. 242. 106404.

287
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

COMPARATIVE ASSESSMENT OF LANDUSE LANDCOVER CHANGES ON WATER

QUALITY OF RIVER KADUNA FROM 2012-2020 AT WUYA, NIGER STATE,
NIGERIA
#11238

1S.U.Ibrahim, 2M.O. Dania, 3F.O Yusufu, 4R.J Kolo, 5S.O.E. Sadiku

Department of Water Resources, Aquaculture and Fisheries Technology, School of Agriculture
and Agricultural Technology, Federal University of Technology, Minna, Niger State Nigeria.
e-mail: [email protected]; +234(0)7039025846, +234(0)9066425223

ABSTRACT

The study investigates the effects of land use land cover changes on water quality of River Kaduna
from 2012-2020 at Wuya, Niger state, Nigeria using Landsat 7 imagery. Five classes of LULC
types where selected and used as basis for classification. Also, five (5) sampling stations selected
on the water body for water quality analysis which were collected once monthly for a period of six
months from February 2020 to July 2020. The results of LULC classification depicts an increase
in water body from 2.241km2 in 2012 to 3.029km2 in 2020 while Agricultural areas increased
drastically from 4.718km2 in 2012 to 22.862km2 in 2020. Physical and Chemical parameters
showed range values of Total dissolved solids varying from (15.54±18.00 - 61.00±21.38) in 2012
while (3.67±1.05 - 5.67±2.57) in 2020, Alkalinity was between (31.33±8.08 - 62.33±17.79) in 2012
whereas (21.33±11.55 - 29.33±6.11) in 2020, Hardness ranged from (38.67±4.16 - 51.33±10.26)
in 2012 whereas (15.00±1.73 - 22.33±7.77) in 2020. The result showed no significant difference
(p > 0.05) except Total dissolved solids, Alkalinity and Hardness which recorded higher values on
both seasons across stations and months. In general, the study revealed increase in agricultural area
drastically, so there is a need to constantly monitor and update the check list of land use land cover,
Physical and Chemical parameters changes in and around the River to control anthropogenic
pollution from residence, on the water and from nearby farmlands.

Keywords: Land Use Land Cover, Water Quality, River, Wuya, Physical and Chemical

INTRODUTION

Land use is simply human activities that explores the usage of land, and Land cover on the other
hand can be seen as the amount of vegetation on land surfaces. Water quality could be referred to
as a measure of water use for different purposes (drinking, industrial, agricultural, recreational and
habitat) using various parameters such as physical, chemical and biological parameters which
varies according to location, time, weather and sources of pollution (Giri and Qiu, 2016).

The impact of Land use is high on water quality of rivers and inland water bodies found in Nigeria
due to the high rate of human interaction, this also result to change in land cover patterns within
the watershed of rivers. As human activities increase in an area, there is change in Land use Land
cover (LULC) hence affects the management of that area, the water quality of rivers and runoffs.
Remote sensing techniques have been recognized as a powerful means to obtain information on
Earth’s surface (Schneider, et, al. 2010). Through supervised classification, the relationship

288
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

between land use/land cover was analyzed using QGIS tool for change detections and how it relates
to water quality around River Kaduna downstream at Wuya for the years 2012 and 2020.

MATERIALS AND METHODS

The Study Area

Wuya located in Niger State. It is just some few kilometres from Bida, at a latitude of 09 o08’. 622”
N and longitude of 05o50’. 258” E with an elevation of 75.1M. The prominent feature which can
serve as a landmark in identifying the river is an MTN network mask mounted few meters to the
river and an overhead bridge that connect both ends of the River together to allow for easy passage
of travellers, cars, goods and services to and from across both ends of the river. The River is a
major source of water for domestic activities for the community.

Figure 1. Nigeria indicating Niger state and the study area.

The study was carried out in six month (6) from February and July 2020.Water samples were
collected once monthly for the period from 5 sampling stations. Five classes of LULC types where
selected (Water bodies, Agricultural areas, bare surfaces, Natural vegetation, Settlements) and used
as basis for classification of the LULC around the River using QGIS tool.

Site Selection
Station 1 had its land use classification as natural vegetation. This area consists of naturally
growing plants with little interference with human activities, (where pipes for irrigation purpose
were mounted). On coordinates point of 09o08’.061” N, 05o49’.858” E, and elevation as 71.1m.

Station 2 had its land use classification as agricultural area. This consists of where farming
activities and irrigation take place on coordinates 09o08’.141”N, 05o50’094”E and elevation as
70.6m.

289
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Station 3 had its land use classification as settlements on coordinates as 09 o08’.686” N,

05o50’.070” E and elevation as 69.0m.

Station 4 had its land use classification as water body. This area consists of a large volume of water
towards the middle of the river. The area recorded coordinates as 09o08’.853”N, 05o50’109” E
and elevation as 67.0m.

Station 5 had its land use classification as bare surface. This area is a plain ground with sharp sand
around and no grasses growing. The area recorded coordinates as 09o08’.784” N, 05o50’.302” E
and elevation as 65.9m.

Figure 2. Google Earth Image of River Wuya showing selected sites.

Landsat 7 data was gotten for the study area of 2012 and 2020 from earth explorer which collected
a panchromatic (black and white) imagery and multispectral imagery which was then imported into
remote sensing image processing software (QGIS tool) for analysis.

RESULTS AND DISCUSSION

The result in Figure 3 depicts changes in classified land use land cover imagery of 2012 and 2020
with a rise in agricultural areas, lesser bare surfaces and natural vegetation in 2020 while Figure 4
compares the percentage changes in land use land cover around the study area.

290
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 3. Classified Land sat image of the Land Use Land Cover around the study area in (2012
and 2020) with Comparison of their Percentage changes.

Figure 4. Digital Elevation Model of the Study Area.

291
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 2. Magnitudes, percentage and average rates of change in Land Use Land Cover around
study area (2012 and 2020).

LULC Types 2012 Area 2020 Area Magnitude of % Average Rate of Remark
(Km2) (Km2) Change (km2) Change Change (Km2)
Water Bodies 2.241 3.0285 0.7875 2.24 0.3514 Increase
Agricultural 4.7178 22.8618 18.144 48.42 3.8458 Increase
Area
Bare Surfaces 13.6728 3.5532 -10.1196 27.07 -0.7401 Decrease
Natural 8.7471 1.0098 -7.1937 19.27 -0.8845 Decrease
Vegetation
Settlements 8.2035 7.1289 -1.0746 3.0 -0.1309 Decrease
Total 37.5822 37.5822 0.5436 100 2.4417 Increase

Table 2 above depicts that there was a decrease in the area covered by bare surfaces from
13.673km2 in 2012 to 3.553km2 in 2020 which relates directly with natural vegetation which also
decreased in area covered from 8.747km2 in 2012 to 1.010km2 in 2020 due to anthropogenic
activities. An increase in water body of the area was recorded from 2.241km2 in 2012 to 3.029km2
in 2020 this could be attributed to the change in vegetation type as part of land preparation, increase
in runoff, uncontrolled human activities, demand for firewood and so on. There was a decrease in
Settlements of the area from 8.204km2 in 2012 to 7.129km2 in 2020 which could be related to the
increase in water body of the area that may have caused flooding in settlements along the water
shed and made settlers relocate. A drastic increase in agricultural areas was recorded from
4.718km2 in 2012 to 22.862km2 in 2020. The highest increase maybe because of conversion bare
surfaces into farmlands as agriculture is one of the main sources of income for the riparian
communities and it is practiced all year round such irrigation. With these changes in land use land
cover effects, water quality is highly affected and needs to be controlled. Relationships between
land use land cover and surface water quality are relevant topics for discussion as human activities
increase in a watershed (Ding, et al. 2015).

The physico-chemical parameters results vary across the sampling stations in both seasons (Table
3 and 4 below).

Hydrogen ion Concentration: pH recorded its highest value across stations for 2012 in the dry
season (7.22±0.18a) at station 1 with its lowest value in the wet season (6.86±0.77 a) at station
2.While 2020 had its highest recorded value across stations in the wet season (6.83±0.02 a) at
station 4 with its lowest value in the dry season (6.20±0.43a) at station 5.There was no significant
difference (p>0.05) for pH across means. Ideal range for pH 6.5 – 7.5 (Ding et al., 2015). pH
recorded negative correlation with Phosphate and positive correlation with total suspended solids,
alkalinity, hardness and nitrate.

Temperature: Biological and chemical changes of water are greatly influenced by temperature,
with the ideal range being 26oC – 32oC (Ding et al., 2015). Highest value across stations was
recorded for 2012 in the dry season (34.77±4.13oC) at station 2 with its lowest value in the wet
season (27.67±1.15oC) at station 4. highest recorded value across stations in the dry season
(31.66±0.75oC) at station 3 with its lowest value in the wet season (27.63±1.66oC) at station 4.
There was no significant difference (p>0.05) across means. A slightly negative correlation was
recorded against dissolve oxygen.

292
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 3. Dry Season Stations Variation (February - April 2012 and 2020).

SAMPLE STATIONS
Parameters Years
1 2 3 4 5
2012 7.03±0.70a 6.86±0.77 a 6.90±0.56 a 7.00±0.64a 7.00±0.45a
pH
2020 6.82±0.04a 6.82±0.04 a 6.81±0.02 a 6.83±0.02a 6.83±0.05a
2012 29.33±0.57a 29.00±1.00a 28.00±0.00a 27.67±1.15a 28.00±1.73a
T(0C)
2020 27.67±1.50a 27.80±1.65a 27.77±1.55a 27.63±1.66a 27.77±1.66a
2012 92.33±30.62a 83.00±64.86a 67.33±33.32a 69.67±42.19a 67.00±44.44a
EC
2020 76.67±18.90a 65.67±5.86a 64.33±5.51a 67.33±5.51a 58.00±16.82a
2012 12.67±3.06 a
12.00±2.00 a
10.67±4.16 a
11.33±5.03 a
14.00±5.29a
DO (Mg/l)
2020 10.67±3.06a 9.33±1.16a 10.67±2.31a 13.33±3.05a 10.00±3.46a
2012 61.00±21.38a 55.00±44.23a 44.33±22.81a 45.67±28.68a 44.33±29.87a
TDS (Mg/l)
2020 4.91±1.21a 4.20±0.37a 4.12±0.35a 4.30±0.34a 3.67±1.05a
2012 62.33±17.79 a
56.66±5.77 a
57.00±18.36 a
51.66±7.64a 53.00± 28.05a
Alkalinity (Mg/l)
2020 28.00±4.00a 28.00±10.58a 29.33±6.11a 29.33±2.31a 30.67±6.11a
2012 38.67±4.16 a
42.00±15.87 a
37.33±8.08 a
51.33±10.26 a
39.33±18.58a
Hardness (Mg/l)
2020 15.00±1.73a 18.00±5.29a 20.67±3.06a 21.00±2.00a 17.33±5.03a
2012 0.70±0.47a 0.62±0.48a 0.66±0.50a 0.66±0.47a 0.80±0.63a
PO4 (Mg/l)
2020 2.94±0.51a 2.40±0.37a 2.96±0.49a 2.49±0.21a 2.81±0.18a
2012 2.88±0.93a 2.90±0.76a 2.84±.89a 3.04±0.72a 2.95±0.86a
NO3 (Mg/l)
2020 0.22±0.02a 0.33±0.19a 0.24±0.04a 0.24±0.04a 0.24±0.03a
Means in the same row having the same superscript are not significantly different from other means (p>0.05)

Table 4. Wet Season Stations Variation (May - July 2012 and 2020).

SAMPLE STATIONS
Parameters Years
1 2 3 4 5
2012 7.22±0.18a 7.11±0.07a 7.04±0.19a 7.11±0.19a 7.07±0.15a
pH
2020 6.51±0.76a 6.41±0.41a 6.27±0.54a 6.21±0.53a 6.20±0.43a
2012 32.17±4.65a 34.77±4.13a 32.87±5.22a 29.33±1.15a 30.17±0.28a
T(0C)
2020 31.46±0.71a 31.50±0.75a 31.66±0.75a 31.43±0.80a 31.43±0.77a
2012 76.00±27.84a 72.67±25.42a 78.00±27.06a 83.00±36.01a 79.67±32.87a
EC
2020 87.00±37.32a 77.00±24.27a 78.33±27.47a 85.00±39.00a 93.67±28.75a
2012 8.00±3.46a 7.83±0.29a 6.33±0.58a 8.00±2.00a 8.47±1.50a
DO (Mg/l)
2020 11.33±2.31a 11.33±4.16a 12.67±1.16a 11.33±3.05a 10.00±0.00a
2012 48.88±16.92a 46.60±15.54a 15.54±18.00a 52.70±23.40a 50.53±21.17a
TDS (Mg/l)
2020 5.67±2.57 a
5.24±2.08 a
5.19±2.06 a
4.79±1.37 a
5.45±1.23a
2012 33.33±11.55a 39.33±13.61a 41.33±18.90a 31.33±8.08a 34.67±12.86a
Alkalinity (Mg/l)
2020 21.33±11.55a 24.00±12.00a 26.67±2.31a 26.67±6.11a 32.67±6.43a
2012 38.00±7.21a 45.33±9.87a 46.67±10.07a 41.33±6.11a 43.33±3.06a
Hardness (Mg/l)
2020 23.67±14.15a 19.33±8.08a 22.33±7.77a 31.00±12.29a 28.67±4.16a
2012 0.80±0.14a 0.88±0.23a 0.94±0.16a 0.77±0.06a 0.88±0.10a
PO4 (Mg/l)
2020 3.15±0.29a 2.68±0.14a 3.16±1.28a 2.70±0.32a 2.54±0.24a
2012 2.59±0.94a 2.64±1.25a 3.27±1.75a 2.57±1.48a 2.65±1.57a
NO3 (Mg/l)
2020 0.21±0.05a 0.22±0.04a 0.21±0.04a 0.24±0.07a 0.23±0.09a
Means in the same row having the same superscript are not significantly different from other means (p>0.05).

293
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

CONCLUSION

The study on the comparative analysis of land use and land cover changes on water quality of
River Wuya using QGIS, exposed causes of the land use and variation in physicochemical
parameters. The use of QGIS allowed a more graphical representation on the study area to see
trends and how land use land cover changed over time. The variations of land use land cover
changes & physico-chemical parameters may be due to climate change, land use types, human
activities, change in vegetation types, erosion due to the cut down of trees and more land
rendered plain and open.

REFERENCES

Ding J, Jiang Y, Fu L, Liu Q, Peng Q, Kang M. 2015. Impacts of land use on surface water
quality in a subtropical river basin: a case study of the Dongjiang River Basin,
Southeastern China. Water, 7(8), 4427-4445.
Giri C, Zhu Z, Reed B (2005) A comparative analysis of the global land cover2000 and MODIS
land cover data sets. Remote Sens Environ 94:123–132
Schneider, A.; Friedl, M.A.; Potere, D. Mapping global urban areas using MODIS 500-m data:
New methods and datasets based on ‘urban ecoregions’. Remote Sens. Environ. 2010,
114, 1733–1746.

294
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

DESIGN AND DEVELOPMENT OF A LORA COMMUNICATION SYSTEM FOR

SCALABLE SMART IRRIGATION SYSTEMS
#11694
1*J. Ikabat, 2P. Hess, 3J. Wanyama, 4E. Bwambale, 5P. Nakawuka, 6I. Kabenge, 7S. Felicioni,
8A. Bühlmann, 9T. Anken, 10J.H. Muyonga
1,3,4,5,6,10Makerere University, Kampala, Uganda; 2,7,8,9Agroscope, Wädenswil and Tänikon,

Switzerland
*e-mail: [email protected], +256750377244

ABSTRACT

The adoption of smart irrigation in Sub-Saharan countries such as Uganda is still hindered by
several challenges, primarily related to the high initial costs, infrastructure dependence, and
limited hardware availability. This study aimed to address these challenges by developing and
testing an affordable smart irrigation system based on LoRa radio technology. The system
design incorporates readily available hardware components and Chirpstack open-source
software, which serves as both the LoRaWAN Network and Application server for scalable
and license-free radio operation. A real-world scenario functionality test was conducted at the
Makerere University Agricultural Research Institute Kabanyolo (MUARIK) in which a smart
valve was connected to the LoRa gateway at two different locations. Two connection
parameters, the Signal to Noise Ratio (SNR) and Received Signal Strength Indicator (RSSI),
were compared to analyze the range and reliability of the system. Results demonstrated robust
communication, particularly in areas with fewer obstructions, highlighting the importance of
optimal antenna placement to ensure efficient communication to automate irrigation systems
as well as agricultural systems. The LoRa radio enhances smart irrigation by offering low
power consumption, long-range capabilities, and collaborative farmer access, which helps
reduce initial costs. This system improves water efficiency and crop yields, aiding smallholder
farmers in adapting to climate variability and promoting sustainable agriculture in Uganda.

INTRODUCTION

Agriculture remains a fundamental pillar of the economy in many Sub-Saharan African

countries, including Uganda, where most farmers are smallholders residing in rural areas.
These farmers frequently encounter significant challenges, such as food insecurity exacerbated
by a rapidly growing population [1], [2]. Traditionally, Uganda's agriculture has relied heavily
on natural climatic patterns, particularly the two annual wet seasons [3]. Furthermore,
anthropogenic climate change has increasingly disrupted these patterns, leading to more
frequent and severe dry spells, floods, and rising temperatures, all of which threaten food
security [4], [5].

In light of these challenges, policymakers have emphasized the critical role of irrigation in
enhancing food security amidst climate variability [6]. Initiatives such as the Micro-scale
Irrigation Program supported by the World Bank and administered by Uganda's Ministry of
Agriculture, Animal Industry, and Fisheries (MAAIF), are advancing sustainable agricultural
practices by providing subsidized irrigation equipment and training to smallholder farmers.
The adoption of technology-enabled smart irrigation has emerged as a vital strategy for
adapting to climate change due to the potential benefits of these technologies which include
increased agricultural productivity and more efficient use of resources, thereby contributing to
the sustainability and resilience of food production systems. This is achieved through the

295
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

deployment of extensive sensor networks that monitor crop water needs, coupled with
automated actuators like pumps and valves.

Despite these advantages, the adoption of smart irrigation technologies in Uganda faces
significant obstacles such as financial constraints that are a major barrier, as the average annual
income of farming households in Uganda is only USD 222 [7]. Additionally, infrastructure
challenges such as frequent power outages and limited internet availability hinder the effective
operation of IoT-powered smart irrigation systems. Wanyama et al. [8] also identified the lack
of material and service supply as a burden to irrigation adoption in many regions of Uganda.
LoRa radio technology is particularly well-suited for connecting smart components due to its
low power consumption and long-range coverage, making it ideal for IoT and smart farming
applications [9], [10]. This article aims to propose a smart irrigation system designed to
overcome these challenges, focusing on affordability, resilience to infrastructure limitations,
and the use of readily available materials to ensure successful broad adoption in Uganda.

MATERIALS AND METHODS

Study Site
The study was carried out at the Makerere University Agricultural Research Institute
Kabanyolo (MUARIK), located 21km north of Kampala city center, Uganda. The institute is
located at coordinates 00°27'06.000" N, 32°03'60.240" E, with an altitude varying between
1250 and 1320 meters above sea level in the Wakiso district.

System Design and Assembly

In alignment with the principles of affordability and accessibility, the components for the
system were selected based on their low cost and availability, either through international
delivery or local sourcing in Uganda. To ensure independence from internet infrastructure, a
local processing approach was adopted. The system includes battery backups to mitigate the
effects of power outages. An overview of the proposed smart irrigation system is illustrated in
Figure 1(a).

(a) (b)

Location B

Smart Valve

Location A

Figure 14. (a) Design overview of proposed LoRa Smart irrigation system, (b) Map of the
locations and distances of the functionality test.

296
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

The system features a proposed sensor node positioned at the field level, which connects
common soil moisture sensors, such as the commercial Watermark sensor [WATERMARK
200SS, Irrometer Company, Riverside, USA] or the more cost-effective Chameleon sensors
[Chameleon Sensor Array, VIA Ltd, Melbourne, Australia], to an enclosed LoRa module
consisting of an interface module [SMX, EME Systems, Berkeley, USA], a microcontroller
[HUZZAH32, Adafruit Industries, New York, USA] and a LoRa transmitter [E220-900T22D
LoRa Wireless UART Module, Ebyte, China]. These components are powered by a waterproof
battery and solar panel [EESBAO-35W, Shenzen Tengyunfei Technology Co., Shenzen City,
China]. The collected data is transmitted to a LoRa gateway unit, which is typically located in
an office, farmer's house, or community center.

At the core of the gateway unit is a microcontroller [Raspberry Pi 4B, Raspberry Pi Ltd,
Cambridge, United Kingdom], which is connected to a LoRa concentrator board
[GPMLx9332-PX V3, Greenpalm, Hangzhou, China] and a 0.5-dB antenna. The gateway
operates on open-source software [Chirpstack V4, Orne Brocaar, Amsterdam, The
Netherlands], which includes network and application server functionality. To safeguard
against power interruptions, a small uninterruptible power supply [UPS Module 3s, Waveshare,
Shenzen, China] is integrated into the system. When irrigation is required, a command is sent
to activate a wireless motorized shut-off smart valve [STREGA, Ohain, Belgium] that is
programmed to open for a specific period depending on the soil’s antecedent soil moisture and
its field capacity. The valve is set to operate as a Class A LoRa device.

Functionality Test
The system was tested for its functionality by studying the signal strength and connectivity
using the Received Signal Strength Indicator (RSSI) and Signal to Noise Ratio (SNR) of the
periodic upload messages sent by the smart valve to the gateway communicating its status
information (e.g. temperature and battery level) to the gateway. The valve was installed at an
existing drip irrigation system at the Department Demonstration plot located at the study site
and gateway set to operate in the 868 Mhz frequency. Two different locations were chosen:
Location A had a much more unobstructed line of sight to the valve at a distance and elevation
difference of 402 m and 7.6 m, and location B was obstructed by several buildings and
vegetation. It also had a greater distance and elevation difference to the valve of 660 m and
8.5 m (Figure 1(b)). For each site, 10 of these messages were analyzed for their RSSI and SNR
values to assess the quality of the connection.

RESULTS AND DISCUSSION

The test of the system's functionality confirmed full operation between the gateway unit and
the smart valve. The signal strength data showed that connectivity was stronger in areas with
a clearer line of sight (Location A) than with more obstacles (Location B).

When analyzing the data packets sent to the gateway (Figure 2), Location B exhibited generally
worse connectivity as compared to Location A with the SNR values here ranging between -
12.5 to -18.8 dB and averaging -15.7 dB. RSSI values ranged from -93 to -98 dBm, with an
average of -96.3 dBm as the valve communicated at a Spreading Factor of 12. On the other
hand, the SNR values at Location A ranged from -2 to 6 dB, with an average of 2 dB. RSSI
values ranged from -96 to -102 dBm, averaging -98.7 dBm as the valve communicated at a
Spreading Factor of 7 after applying Adaptive Data Rate (ADR) protocol at both locations.
This explains the lower RSSI values for Location B as the valve used more battery power to
ensure the same communication. Additionally, the presence of blockages from buildings and

297
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

vegetation, along with the effects of ambient factors like temperature and relative humidity,
significantly impact signal strength, as corroborated by studies from Antoine-Santoni [11] and
Iova [12]. The function test results highlight the critical role of antenna positioning on signal
strength. For optimal performance, the LoRa antenna should be installed above all obstructions
to ensure a clear line of sight to the smart valve.

This study introduces a cost-effective smart irrigation system for small-scale Ugandan farmers,
using affordable, locally sourced components and open-source ChirpStack software to reduce
expenses and reliance on commercial services. The system is adaptable to different Raspberry
Pi models and supports advanced irrigation algorithms [8] to improve water efficiency and crop
yields. It’s designed to withstand Uganda’s infrastructure challenges with a UPS and local
computing, ensuring operation during power outages and without depending on unreliable rural
internet. While the chosen motorized valve is costly, it is crucial to highlight the need for a
more affordable low-pressure option such as latched solenoid valves as suggested by
Maksudjon [13]. The system’s use of LoRaWAN technology allows a single gateway to
support an entire community, reducing individual costs and promoting community
collaboration as demonstrated in the work of Dongore [14].

Figure 15. Visualization of the signal strength parameters RSSI (a) and SNR (b) received
during the system functionality test.
CONCLUSION AND RECOMMENDATION

The study reviews challenges in smart irrigation in Uganda and proposes a LoRa-based system
to address them. While the system shows promise, it requires field testing with sensor nodes to
evaluate its scalability and effectiveness on smallholder farms. Successful implementation
could lead to nationwide adoption, helping farmers increase profitability, sustainability, and
resilience to climate change.

ACKNOWLEDGEMENT

Authors are grateful to the European Union’s Horizon 2020 research and innovation
programme which funded this work through FoodLAND project under grant agreement No
862802.

298
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

REFERENCES

[1] B. Barasa, “Local and regional variations in conditions for agriculture and food security
in Uganda,” AgriFoSe2030, 2018. Accessed: Aug. 19, 2024. [Online]. Available:
https://nru.uncst.go.ug/handle/123456789/5781
[2] D. N. Mubiru et al., “Climate trends, risks and coping strategies in smallholder farming
systems in Uganda,” Clim. Risk Manag., vol. 22, pp. 4–21, Jan. 2018, doi:
10.1016/j.crm.2018.08.004.
[3] M. R. Jury, “Uganda rainfall variability and prediction,” Theor. Appl. Climatol., vol.
132, no. 3, pp. 905–919, May 2018, doi: 10.1007/s00704-017-2135-4.
[4] M. A. Islam, “Agricultural Adaptation to Climate Change: Issues for Developing
Countries,” Glob. Discl. Econ. Bus., vol. 2, no. 2, Art. no. 2, Dec. 2013, doi:
10.18034/gdeb.v2i2.178.
[5] U. Nuwagira and I. Yasin, “Review of the Past, Current, and the Future Trend of the
Climate Change and its Impact in Uganda,” East Afr. J. Environ. Nat. Resour., vol.
5, no. 1, Art. no. 1, Apr. 2022, doi: 10.37284/eajenr.5.1.605.
[6] J. Wanyama et al., “Irrigation Development in Uganda: Constraints, Lessons Learned,
and Future Perspectives,” J. Irrig. Drain. Eng., vol. 143, no. 5, p. 04017003, May
2017, doi: 10.1061/(ASCE)IR.1943-4774.0001159.
[7] F. Atube, G. M. Malinga, M. Nyeko, D. M. Okello, S. P. Alarakol, and I. Okello-Uma,
“Determinants of smallholder farmers’ adaptation strategies to the effects of climate
change: Evidence from northern Uganda,” Agric. Food Secur., vol. 10, no. 1, p. 6,
Feb. 2021, doi: 10.1186/s40066-020-00279-1.
[8] J. Wanyama et al., “A systematic review of fourth industrial revolution technologies in
smart irrigation: Constraints, opportunities, and future prospects for sub-Saharan
Africa,” Smart Agric. Technol., vol. 7, p. 100412, Mar. 2024, doi:
10.1016/j.atech.2024.100412.
[9] “LoRa PHY | Semtech.” Accessed: Aug. 19, 2024. [Online]. Available:
https://www.semtech.com/lora/what-is-lora
[10] A. Pagano, D. Croce, I. Tinnirello, and G. Vitale, “A Survey on LoRa for Smart
Agriculture: Current Trends and Future Perspectives,” IEEE Internet Things J., vol.
10, no. 4, pp. 3664–3679, Feb. 2023, doi: 10.1109/JIOT.2022.3230505.
[11] T. Antoine-Santoni, B. Poggi, D. Araujo, and C. Babatounde, “Factors Influencing
LoRa Communication in IoT Deployment: Overview and Experience Analysis,”
Apr. 2022. doi: 10.5220/0011102600003194.
[12] O. Iova et al., “LoRa from the city to the mountains: Exploration of hardware and
environmental factors,” in International Conference on Embedded Wireless Systems
and Networks (EWSN) 2017, Uppsala, sweden, 20-22 February 2017, Uppsala
University, 2017, pp. 317–322. Accessed: Apr. 30, 2024. [Online]. Available:
https://openpub.fmach.it/handle/10449/42132
[13] M. Usmonov and F. Gregoretti, Design and implementation of a LoRa based wireless
control for drip irrigation systems. 2017, p. 253. doi:
10.1109/ICRAE.2017.8291389.
[14] A. Dongare et al., “OpenChirp: A Low-Power Wide-Area Networking architecture,” in
2017 IEEE International Conference on Pervasive Computing and Communications
Workshops (PerCom Workshops), Mar. 2017, pp. 569–574. doi:
10.1109/PERCOMW.2017.7917625.

299
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

EVALUATING THE IMPACT OF SEASONAL WEATHER VARIABILITY ON

SOIL MOISTURE CONSERVATION UNDER MULCHING SYSTEMS FOR DATE
PALM PRODUCTION IN OASES
#11301
1,3U.Safi, 2E.J. Mustapha, 1O. Abdallah, 3S. Ahmed
1Mohammed VI Polytechnic University, Benguérir, Morocco; 2Agronomic and Veterinary
Institute Hassan II, Rabat, Morocco; 3National Institute for Agricultural Research, Rabat,
Morocco
e-mail: [email protected] +212708362464

ABSTRACT

Efficient soil moisture conservation is crucial for sustaining date palm production in arid
Moroccan oases, where water scarcity is a significant challenge. This study evaluates how
different mulching materials affect soil moisture retention in these regions. The research
focuses on three mulching materials: polyethylene plastic (PP), polypropylene woven ground
cover (PWGC), and date palm residues (DPR), examining their effectiveness under varying
seasonal climates. Results demonstrate that the PP+DPR combination was the most effective
in conserving soil moisture at shallower depths during high temperatures, while DPR and
PWGC+DPR provided consistent moisture retention at greater depths. The findings highlight
the importance of mulching systems in mitigating weather variability impacts, enhancing soil
moisture conservation that can sustain the date palm production in arid regions.

INTRODUCTION

The date palm (Phoenix dactylifera) holds considerable importance in Moroccan oases. It
primarily contributes to creating a favorable agricultural microclimate for other crops, which
helps safeguard the land against desertification (Abdelaaziz et al., 2024). Furthermore,
scientific research has demonstrated that date fruits to be highly nutritious and possess
medicinal properties (Ouamnina et al., 2024). Consequently, date palm production is
economically valuable in this region.

However, due to the climatic conditions in arid regions where precipitation is inadequate to
offset evapotranspiration, irrigation becomes necessary to meet the water needs for date
production (Yan et al., 2022). Though, with climate change, water resources in arid regions are
becoming increasingly scarce over time. Thus, implementing solutions for reducing water
usage and conservation are crucial in these areas (Morante-Carballo et al., 2022).

Mulching, a well-known water conservation practice, has demonstrated potential in enhancing

soil moisture retention, reducing evaporation, and stabilizing soil temperature (Ramakrishna et
al., 2006). Considering the interaction between the land and atmosphere, as weather transitions
from winter to summer, fluctuations in atmospheric temperature and precipitation impact soil
moisture content (Cho et al., 2016). Hence, mulching serves as a buffer by mitigating the
adverse effects of these interactions.

This study aims to evaluate the effectiveness of three mulching materials—polyethylene plastic
(PP), polypropylene woven ground cover (PWGC), and date palm residues (DPR)—in
conserving soil moisture in date palm production under varying seasonal climate conditions.
The implementation of mulching systems may mitigate the effects of weather variability on

300
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

soil moisture and enhance date palm production by stabilizing soil temperature, reducing
evaporation, improving moisture retention and promoting the transpiration of the plants,
particularly during periods of extreme seasonal changes.

MATERIALS AND METHODS

Experimental Site
The experiment began in early April 2024 and ended August 2024 at the experimental farm of
the National Institute of Agricultural Research (INRA), located in Errachidia Province in the
South-East of Morocco. The experiment occupies an area of a half hectare.

Experimental Design
This experiment was conducted on a date palm (Phoenix dactylifera) on the variety called
Nadja, that is known for its resistance to the famous date palm disease Bayoud and other
important agronomic traits. Drip irrigation was employed throughout the experiment, with the
same amount of water applied across all treatments. However, the irrigation schedule differed,
where from April to the end of May irrigation was done three times a week, and from June till
August, irrigation was done four times a week.

Three mulching materials were tested: polyethylene plastic (PP), polypropylene woven ground
cover (PWGC), and date palm residues (DPR). The synthetic mulches (PP and PWGC) were
combined with the organic mulch (DPR). The experiment was set up using a randomized
complete design, comprising 72 palm trees divided into 36 experimental units, each containing
2 trees, with 9 replications. Each experimental unit represented a different mulching treatment.
The mulching treatments were distributed as follows:

• T0: Control (non-mulch) treatment (NM)

• T1: Organic mulch made from ground dry date palm leaves (DPR)
• T2: Polyethylene plastic (PP) combined with date palm residues (DPR)
• T3: Polypropylene woven ground cover (PWGC) combined with DPR

Data Collection
To measure soil moisture content as a representation of the entire field, 24 date palm trees with
different mulching systems were selected. For each tree, a profile probe tube was installed at
30 cm from the irrigation dripper. Soil moisture data were collected consecutively on irrigation
days, one hour before, using the Profile Probe (PR2) sensor that uses the gravimetric method.
This sensor measured and recorded soil moisture data at depths of 0-100 mm, 100-200 mm,
200-300 mm, 300-400 mm, 400-600 mm, and 600-1000 mm.

Data analysis
The experimental data collected were analyzed with d software tools, specifically Python's
Matplotlib and R's ggplot2. These tools facilitated the precise visualization and comparison of
soil moisture content across various experimental conditions, enhancing the clarity and
interpretability of the data. Microsoft excel was used to analyze the weather data (temperature
and precipitation)

301
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

RESULTS AND DISCUSSION

1. Weather variability (average temperature and precipitation)

Figure 1 illustrated the variations in average temperature and precipitation from April 1st to
August 21st at the experimental farm in the Errachidia region. The graph indicated that average
temperatures gradually increase from April, reaching a peak during the mid-summer months
(June to August) and remaining consistently high. Although there were minor fluctuations, the
mean temperatures generally fell between 20°C and 35°C. In contrast, the precipitation pattern
remains relatively low throughout this period, with occasional spikes representing rainfall
events, but most days experience little to no precipitation.

Figure 1. Temporal Trends in Average Temperature and Precipitation from March to August.

2. Mulching materials on soil moisture retention at different depths under seasonal

temperature fluctuations.
Figures 2 and 3 demonstrated that different mulching systems (DPR, PP+DPR, and
PWGC+DPR) had a significant impact on soil moisture retention across various soil depths
and seasonal temperature fluctuations, compared to no mulch (NM). At depths of 100mm,
200mm, and 300mm, Figure 2 showed that the PP+DPR mulching system was the most
effective in conserving soil moisture, particularly as temperatures increase, outperforming
other mulching type, though there was an increase in irrigation schedule. Conversely, the NM
system consistently proved to be the least effective, especially under conditions of high
temperature variability. Additionally, at a depth of 400mm, the PP+DPR system markedly
increased soil moisture levels with rising temperatures, while both DPR and PWGC+DPR
systems also exhibited strong moisture retention capabilities, demonstrating their ability to
maintain consistent soil moisture in deeper soil profiles under varying temperature conditions.

302
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 2. Soil moisture variability at 100mm (A), 200mm (B), 300mm(C) to 40mm (D)soil
depth at date palm experiment with seasonal temperature variation from April (Spring) to
August (summer) under different mulching systems (DPR (Dry palm residue), PP
(Polyethylene plastic) + DPR, PWGC ((Polypropylene woven ground cover) +DPR, NM (Non-
mulch)).

A study conducted by Ma et al. (2009) on corn plastic mulching in the East area of Jilin
Province clearly reported that the corn soil moisture increased under plastic mulching
particularly in the dry and rainless days, which quite like our study especially for the topsoil
profile.

Figure 3. Soil moisture variability at 600mm (A) and 1000mm (B) soil depth at date palm
experiment with seasonal temperature variation from April (Spring) to August (summer) under
different mulching systems (DPR (Dry palm residue), PP (Polyethylene plastic) + DPR, PWGC
((Polypropylene woven ground cover) +DPR, NM (Non-mulch)).

303
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

At a depth of 600mm (Figure 3(A)), the DPR system was the most effective in conserving soil
moisture as temperatures increase. The PWGC+DPR system also performed well, closely
matching the effectiveness of the DPR system, and both outperform the NM system. At a depth
of 1000mm (Figure 3(B)), soil moisture retention varied among the different mulching systems,
with the DPR and PWGC+DPR systems providing the most consistent moisture retention as
temperatures rise. The NM system exhibited minimal variation, suggesting that deeper soils
retain moisture more naturally. Meanwhile, the PP+DPR system, effective at shallower depths,
showed reduced effectiveness at both 600mm and1000mm soil depths. These results matches
with the findings in the study that Yin et al. (2022) conducted on the effect of plastic film
mulching system on deep soil moisture where they found that plastic mulching reduced soil
water storage (SWS) in the 0–100 cm- 200cm and 300 cm soil profile. This shows that less
water is accumulated in the deeper soil profile compared to other mulching systems.

304
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

INFLUENCE OF MULCHED-DRIP IRRIGATION SYSTEM ON YIELD AND

PHYSIOLOGICAL ATTRIBUTES OF PEPPER VARIETIES IN SOUTHWEST,
NIGERIA

1I.A.
Kehinde, 1*O.O. Fawibe, 1A.S. Oyelakin
1Department of Pure and Applied Botany, Federal University of Agriculture Abeokuta, PMB
2240, Alabata, Abeokuta, Nigeria
*e-mail: [email protected]

ABSTRACT

The unpredictable rainfall pattern due to climate change pose a major challenge to pepper
production and its yield outcome in Southwest, Nigeria; therefore, prompting the need for
supplementary irrigation practice. The field study assessed the influence of mulched-drip
irrigation on yield and agronomic characteristics of three varieties of pepper. The experiment
was arranged in a split-plot design with three replicates. The irrigation type; rainfed (RF), drip
irrigation with plastic -film (DIP), and drip irrigation with biodegradable film (DIB) served as
the main plot while the subplots included three Capsicum frutescens varieties: var. baccatum
(var. A) var. abbreviatum (var. B), and var. acuminatum (var. C Although the total water
supplied to the DIP and DIB increased by 51.7% and 60.3%, respectively compared with RF,
the yield of Capsicum also significantly increased by 53.6% and 55.7% under DIP and DIB,
respectively. The increased yield outcomes under mulched-drip irrigation were ascribed to the
significant increase in chlorophyll content, leaf area and total biomass of Capsicum. Notably,
DIP and DIB similarly influenced yield and agronomic attributes of Capsicum; however, DIP
saved 13.3% of irrigated water more than DIB. The yield of the Capsicum frutescens varieties
were comparable. In conclusion, mulched-drip irrigation irrespective of mulch type
significantly increased the yield of pepper in Southwest, Nigeria and could serve as
supplementary irrigation technique for the promotion of pepper.

Keywords: water-saving irrigation, climate-smart, pepper, intermittent drought

INTRODUCTION

Pepper (Capscicum frutescens) is a popular vegetable in West Africa, with economic, food and
medicinal importance (Dagnoko et al., 2013). However, its production is heavily dependent on
the availability of water, especially during the flowering and fruiting stages (Mackic et al.,
2023). Nigeria produces about 50% of the pepper consumed in Africa (Idowu-Agida et al.,
2010). Although pepper is widely cultivated throughout Nigeria, yields obtained by farmers
are often very low (Idowu-Agida et al., 2010), as a result of the rainfed agricultural practice
with unpredictable precipitation patterns. Due to high sensitivity of pepper to shortfall in water
supply, plant growth and biomass production are restricted causing severe yield losses (Shama
et al., 2019).

Mulched-drip irrigation has become increasingly acceptable and practiced as a water-saving

technique that involves the use of drip pipes for water application under a protective film cover
(Fawibe et al., 2020; Wang et al., 2023). The technique has been reported as a climate-smart
agricultural practice in different regions of the world- China (Wang et al., 2024); Italy (Sifola
et al., 2024); India (Job et al., 2018); and Japan (Fawibe et al., 2019). Mulch directly prevents
the exchange of water vapour between the soil surface and atmosphere thereby suppress weed

305
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

growth, reduce soil water evaporation, optimize soil temperature and promote plant growth and
yield of high-quality crops (Lamont, 2017; Zhang et al., 2019). In recent years, varying mulch
types have been developed with variation in colour and composition (Nithisha et al., 2022).
Despite the diverse benefits of plastic films; its removal from the soil after harvest remains
difficult and laborious due to the adhesion of the soil particles to the film; hence, resulting in
fragmentation of the plastics (Qi et al., 2022). To ensure food security in a sustainable
environment, biodegradable mulch (BDM) has been developed. Biodegradable film has shown
to disintegrate into CO2 and H2O through the process of oxidation and microbial degradation
(Amanna et al., 2021). However, it remains unknown if it could produce comparable results
with the plastic film mulch. Hence, this study aimed to determine the yield and agronomic
characteristics of pepper varieties under mulched drip irrigation techniques compared with
conventional cropping practices.

MATERIALS AND METHODS

A field experiment was conducted at the experimental farm of the Department of Pure and
Applied Botany, Federal University of Agriculture, Abeokuta (7.0728oN, 3,3367oE) from April
to September 2023.

Experimental Design and field management

The experimental plots were arranged in a split-plot design (main plot: irrigation and mulching
system; subplot: Capsicum varieties) with three replicates. The main plot consisted of three
irrigation and mulching systems: rain-fed (RF), drip irrigation with plastic-film mulching (DIP)
and drip irrigation with biodegradable film mulch (DIB). The subplots included three
Capsicum frutescens varieties: var. baccatum (var A) var. abbreviatum (var. B), and var.
acuminatum (var C). The RF plots depended on rainfall and the total amount of rainfall during
the experimental period was recorded.

In both DIB and DIP plots, drip tubes, each of 6 m long were laid beneath a black polyethylene
plastic film and a biodegradable film, respectively. The drip pipes were connected to a source
of water (reservoir) and irrigation water supplied to DIP and DIB was measured with an
installed flow meter at 75% field capacity (Fawibe et al., 2020). Basal application of 100 kg/ha
of NPK (15:15:15) was carried out, while weeding was done at intervals in all experimental
plots when necessary.

Plant Sampling and Analyses

Chlorophyll content and leaf area were measured at vegetative and flowering stages using
SPAD-502 (Konica Minolta Co., Ltd.) and leaf area meter (Portable Leaf Area Meter,
Hangzhou Mindfull Technology Co. Ltd, China), respectively. Pepper plants covering an area
of 1m2 were harvested at the maturity stage and oven-dried at 80℃ for 72 hours to determine
the total biomass. At the fruiting stage, fruit of each variety covering an area of 2 m2 were
harvested weekly for 4 weeks to determine the average yield per hectare. The harvest index
and water use efficiency were also calculated.

Statistical analysis
Data were analyzed using Analysis of variance (ANOVA) and means were separated by
Duncan’s multiple range test when the effects are significant at p<0.05.

306
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

RESULTS AND DISCUSSION

The total water supplied to the Capsicum species through DIB and DIP systems were 60.2%
and 51.7% higher than that of the RF (Table 1); however, commensurate quantities of yields
were produced with an increased rate of 55.7% and 53.6%, respectively (Table 2). The
increased yield outcomes could be attributed to the significant increase in photosynthetic-
associated parameters such as chlorophyll content and leaf area under both DIB and DIP. The
significant increase in chlorophyll concentration and leaf area of Capsicum varieties under DIB
and DIP compared to CF enabled both photosystem II and photosystem I to harvest light over
a larger surface area thereby producing more assimilates for increased growth and development
(Feng et al., 2015). The non-significant variation in the growth and yield parameters of the
Capsicum varieties under varying treatments could be attributed to the similarity in their ability
to adapt to mild water stress caused by intermittent rainfall during the period of plant growth.
Our study shows that the use of either plastic-film or biodegradable film mulch similarly
influenced the growth and yield of capsicum varieties; however, DIP saved 19% of irrigated
water more than DIB. This is attributable to the texture, thickness, and composition of the
materials used. The biodegradable film used in this study was composed of starch-based
materials, cellulose acetate, and cellulose nitrate that gradually disintegrated into CO 2 and H2O
through the process of oxidation and microbial degradation (Amanna et al., 2021).

CONCLUSION

The use of drip irrigation with mulching practices (DIB and DIP) significantly increased the
yield and agronomic attributes of capsicum varieties compared with the conventional rainfed
method by alleviating mild drought stress because of dwindling rainfall pattern especially
during the flowering and fruit formation stages.

Table 1. Irrigation precipitation and total water input under rainfed and drip irrigation with
different mulch types.

Irrigation (mm) Precipitation (mm) Total water supplied (mm)

RF 35 720 755
DIB 490 720 1210
DIP 425 720 1145
Total water supplied = Irrigation water + precipitation; RF, DIB, and DIP indicate rainfed, drip irrigation
with biodegradable mulch, and drip irrigation with plastic film mulch, respectively.

307
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Table 2. Yield, harvest index, and water-use efficiency of Capsicum varieties under varying
irrigation and mulching practices.

Irrigation Varieties Yield Fruit Fruit Fruit Total dry HI WUE

(t/ha) length width length- weight (kg/m3)
(cm) (cm) width ratio (t/ha)
RF A 2.25a 2.70a 1.60a 1.66a 3.98b 0.56a 0.31a
B 2.32a 2.76a 1.70a 1.57a 4.06b 0.50b 0.32a
C 2.49a 2.83a 1.73a 1.63a 4.59a 0.61a 0.34a
DIB A 3.54a 3.56a 2.30a 1.55a 6.11a 0.57a 0.29a
B 3.75a 3.60a 2.26a 1.58a 6.88a 0.54a 0.31a
C 3.69a 3.60a 2.43a 1.48a 6.48a 0.57a 0.30a
DIP A 3.57a 3.46a 2.36a 1.46a 6.12b 0.58a 0.31a
B 3.61a 3.46a 2.43a 1.42a 6.73a 0.53ab 0.31a
C 3.66a 3.56a 2.53a 1.40a 6.99a 0.52b 0.31a
Irrigation *** *** *** *** *** ns ns
(I)
Varieties ns ns ns ns * * ns
(V)
IXV ns ns ns ns ns ns ns
Values within a column for each irrigation type followed by different superscripts letters are significantly
different at p<0.05 by Duncan’s multiple range test. * and *** are significant differences at p < 0.05 and p
< 0.001 respectively; ns means non-significant by ANOVA. var. baccatum (var. A) var. abbreviatum (var.
B), and var. acuminatum (var. C).

Table 3. Chlorophyll content and leaf area of Capsicum varieties under varying irrigation and
mulching practices
Irrigation Varieties Chlorophyll content Leaf Area (cm2)
Vegetative stage Flowering stage Vegetative stage Flowering stage
RF A 26.83a 25.96a 10.26a 27.26b
B 27.80 a 26.93 a 11.43 a 30.03a
C 27.63 a
27.73 a
12.03 a
30.53a
DIB A 38.23 a 38.23 a 19.03 b 60.36b
B 39.50 a 38.33 a 19.26 b 69.63a
C 37.66 a 38.73 a 21.63 a 74.23a
DIP A 38.56 a 39.40 a 13.70 b 53.53b
B 39.30 a 38.00 a 15.63 b 58.73b
C 38.86 a 40.80 a 18.46 a 68.96a
Irrigation (I) *** *** *** ***
Varieties (V) ns ns *** ***
IXV ns ns ns ns
Values within a column for each irrigation type followed by different superscripts letters are significantly
different at p<0.05 by Duncan’s multiple range test. *** indicates significant differences at p < 0.001 while
ns means non-significant by ANOVA. var. baccatum (var. A) var. abbreviatum (var. B), and var.
acuminatum (var. C).

REFERENCES

Amanna, R., Mahal, Z. Vieira, E.C.S., Samavi, M., Rakshit, S.K. (2021). Plastics: Toward a
Circular Bioeconomy. In: Circular Bioeconomy-Current Status and Future Outlook.
Biomass, Biofuels, Biochemicals, pg 781-811
Dagnoko, S., Yaro-Diarisso, N., Sanogo, P.N., Adetula, O., Dolo-Nantoume, A.,Gamby-
Toure, K., Traore, A., Katile, S. and Diallo-Ba, D. (2013). Overview of pepper
(Capsicum spp) in West Africa. African Journal of Agricultural Research, 8(13): 1108-
1114

308
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Fawibe, O.O., Honda, K., Taguchi, Y., Park, S. and Isoda, A. (2019). Greenhouse gas
emissions from rice field cultivation with drip irrigation and plastic film mulch. Nutrient
Cycling in Agroecosystems, 113, 51-62.
Fawibe, O.O., Hiramatsu, M., Taguchi, Y., Junfa, W. and Isoda, A. (2020). Grain yield, water-
use efficiency and physiological characteristics of rice cultivars under drip irrigation with
plastic-film-mulch. Journal of crop improvement, 34 (3): 414-436.
Feng, Y., Chen, R., Hu, J., Zhao, F., Wang, J., Chu, H. and Lin, X. (2015). Bacillus asahii
comes to the fore in organic manure-fertilized alkaline soils. Soil Biology and
Biochemistry, 81:186-194.
Idowu-Agida, O.O., Ogunniyan, D.J. and Ajayi, E.O. (2010). Flowering and fruiting behavior
of long cayenne pepper (Capsicum frutescens L.) collected from the south eastern
Nigeria. Proceedings of the 27th HORTISON, conference, Kano, 11-16th October, 2009,
127-132
Job, M., Rusia, D.K., Dinmani, and Singh, V.K. (2018). Effect of drip irrigation and plastic
mulch on yield, water use efficiency of Cauliflower. Journal of Pharmacognosy and
Phytochemistry, 2018: 2556-2560
Lamont, W.J. (2017). Plastic mulches for the production of vegetable crops. In: A guide to the
manufacture, performance, and potential of plastics in Agriculture. Plastic Design
Library pg 45-60
Mackic, K., Bajic, I., Pejic, B., Vlajic, S., Adamovic, B., Popov, O. and Simic, D. (2023). Yield
and Water Use Efficiency of Drip irrigation of pepper. Water, 15(16) 2891
Nithisha, A., Bokado, K., Charitha, K.S. (2022). Mulches: Their impact on the crop production.
The Pharma Innovation Journal, 11(7): 3597-3603
Qi, Y., Ossowicki, A., Yergeau, E., Vigani, G., Geissen, V. (2022). Plastic mulch film residues
in agriculture: impact on soil suppressiveness, plant growth and microbial communities.
FEMS Microbiology Ecology, 98 (2)
Shama, S., Leskovar, D., Crosby K. (2019). Genotypic differences in leaf, gas exchange and
growth responses to deficit irrigation in reticular and indonus melons (Curcumis melo
L). Photosynthetica, 57: 237-247.
Sifola, M.I., Cozzolino, E., Ciancolini, A., Falce, M., Raimo, F., Enotrio, T., Sicignamo, M.,
Baiano, S. and Paino, L. (2024). Crop management system consisting of biodegradable
mulching film + drip irrigation increases yield and quality of flue-cured Tobacco.
Sustainability, 16(16): 7089
Wang, J., Fawibe, O.O., Fawibe, K.O. and Isoda, A. (2023). Water productivity, sink
production, and varietal differences in panicle structure of rice (Oryza sativa L.) under
drip irrigation with plastic-film mulch. Field Crops Research, 291: 1-11.
Wang, C., Li, S., Huang, S. and Feng, X. (2024). A review of the application and impact of
drip irrigation under plastic mulch in agricultural ecosystems. Agronomy, 2024 14(8):
1752
Zhang, G., Wu, H., Jiang, K., & Li, Z. (2019). Microplastics in soils: A review of possible
sources, analytical methods and ecological impacts. Journal of Hazardous Materials,
400, 123328.

309
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

SMARTSPROUT: LEVERAGING IOT FOR SUSTAINABLE AND EFFICIENT

SMART IRRIGATION SYSTEMS IN AFRICA

Oyedepo John Adebayo and Oluyege, Damilola Emmanuel

Institute of Food Security, Environmental Resources and Agricultural Research, Federal
University of Agriculture, P.M.B. 2240, Abeokuta. Nigeria
e-mail: [email protected]; +2348032485583

ABSTRACT

Water scarcity and inefficient irrigation practices are critical challenges in agriculture, these
has been exacerbated by climate change and a growing global population. This paper
introduces SmartSprout, an IoT-based smart irrigation system designed to optimize water
usage and improve crop yields sustainably. The system integrates soil moisture sensors, RTC
modules, solenoid valves, and wireless connectivity to automate and optimize irrigation based
on real-time data. This study details the hardware and software design, prototype testing,
market potential, and socio-environmental impacts of SmartSprout. The results demonstrate its
potential to revolutionize irrigation practices for smallholder and commercial farmers globally.

INTRODUCTION

Efficient water management is a cornerstone of sustainable agriculture, especially in the

context of escalating water scarcity and climate change. Traditional irrigation practices often
result in resource inefficiency and suboptimal crop health, a challenge that has catalyzed the
integration of smart technologies into agriculture [1]. IoT-based smart irrigation systems have
demonstrated the potential to address these issues by offering precise, automated water
management tailored to crop needs. Recent advancements in sensor technologies, such as soil
moisture and weather monitoring, have enhanced their applicability and reduced costs, making
them more accessible to farmers in resource-constrained settings.

Studies highlight that IoT-enabled systems can reduce water usage by 20-40% while improving
yields, contributing to global food security. This paper introduces SmartSprout, as an
affordable and scalable smart irrigation solution designed to empower both smallholder and
commercial farmers intensify off season farming. The prototype aims to revolutionize water
management in agriculture thus enhancing sustainable year-round crop production across
Africa [2]

MATERIALS AND METHODS

System Overview
Drip irrigation remains for example is one of the most innovative means of supplementary
water supply to crops today. A typical drip irrigation system consists of drip tapes or tubes,
sub-main pipes, connectors, filters and the fertigator connected to a source of water. With the
introduction of internet of things, it has been possible to automate operations thus making the
process fully computerized. In this paper, the performance of drip irrigation is presented as
improved in reducing labour requirements, efficient water use, remote management of farms,
real-time monitoring of soil, moisture and increased productivity. The schema in figure 1 is the
general architecture of the SmartSprout system.

310
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Figure 1. The layout of the SmartSprout connected with Drip Irrigation System.

Water from the tank is released to the field through the pilot-operated solenoid valves. The
opening or closing of the orifice in the solenoid valve body is determined by the instructions
of the actuator which are derived from the module of the Real Time Clock (RTC) and the soil
moisture sensor. In this case, water is allowed through the solenoid valve when the soil
moisture is low, and the time is in the morning or evening. Water is prevented to flow through
the valve where the two conditions are not met. At field capacity the solenoid valve locks water
from reaching the main irrigation line [3].

Hardware Components and Specifications

Soil Moisture Sensor
Model: Capacitive Soil Moisture Sensor v1.2
Specifications:
o Operating Voltage: 3.3V–5V.
o Analog Output Range: 0–3.3V, mapping precise soil moisture levels.
o Enhanced Feature: Corrosion-resistant coating ensures longevity, even in
high-salinity soils, while delivering accurate data critical for precise irrigation
scheduling.
RTC Module
Model: DS3231 Real-Time Clock Module
Specifications:
o Accuracy: ±2 ppm from 0°C to +40°C, maintaining high precision across
varied environmental conditions.
o Features: Built-in temperature-compensated crystal oscillator ensures
consistent timing, critical for automated irrigation cycles.
o Application: Allows synchronization with seasonal and diurnal crop water
requirements.
Solenoid Valve
Model: 12V DC Solenoid Valve
Specifications:
o Voltage: 12V DC
o Flow Control: Adjustable for both partial and full irrigation requirements.

311
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

o Durability: Engineered with anti-corrosive materials, ideal for long-term

deployment in diverse soil and water conditions.
o Improved Design: Enables precision in water delivery, reducing wastage and
optimizing soil saturation.
Microcontroller
Model: ESP32
Specifications:
o Dual-core processor: Xtensa 32-bit LX6, providing robust computational
capacity.
o Connectivity: Integrated Wi-Fi and Bluetooth allow seamless remote
operation and real-time updates.
o GPIO Pins: Supports multiple peripheral integrations, enabling scalability for
additional sensors or modules.
o Advancement: The combination of high-speed processing and low power
consumption makes the ESP32 ideal for sustainable, solar-powered systems.
Power Supply
o Solar Panel and Battery: 12V, 10W solar panel with a 12V, 7Ah
rechargeable battery.
Enclosure
o Material: Polycarbonate casing with IP65 rating for water and dust resistance.

The SmartSprout solution is a very portable but intricate tool. Figure 2 shows the system in a
polycarbonate enclosure and the battery.

Figure 2. Polycarbonate casing (from all sides) enclosing all components with the exception
of the solar panel battery.

Software Architecture
Programming Environment: Arduino IDE for microcontroller programming.
Libraries Used:

312
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

o WiFi.h and HTTPClient.h for wireless communication.

o DS3231.h for RTC integration.
o AnalogRead.h for soil moisture sensor calibration.

Data Processing:
o Logic for threshold-based irrigation control implemented in C++.
o MQTT protocol for real-time data transmission to the cloud.

Visuals and Flow Diagrams

System Architecture Diagram
A flow diagram depicting the SmartSprout system:
o Input Sensors: Soil moisture sensor and RTC module.
o Processing Unit: ESP32 microcontroller.
o Actuation: Solenoid valve controlling water flow.
o Power Source: Solar panel and battery system.
o Output: Mobile app displaying real-time data and enabling remote control.

The System Architecture Flow Diagram for the SmartSprout system as requested. It
visually represents the connections between the key components:

Figure 3. A flow diagram depicting the SmartSprout system [4].

The flow clearly shows how data and power flow through the system for optimal operation.
You can download and integrate the diagram into your work [].

313
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Field Setup Illustration

A labelled image of the prototype deployed in a test farm, showing the placement of
components:
• Sensor embedded in soil.
• Protective enclosure housing the microcontroller and other electronics.
• Water delivery via pipes controlled by the solenoid valve.

Prototype Testing and Results

The SmartSprout prototype was field-tested in various agricultural settings. Key performance
metrics included:
• Irrigation Efficiency: Reduced water usage by 30% compared to traditional methods.
• Soil Moisture Maintenance: Maintained optimal soil moisture levels (20–40%) for
maize and tomato crops.
• Power Efficiency: Solar-powered system provided uninterrupted operation for 72
hours during cloudy conditions.

Case Study 1 from Prototype Testing: Small mother trials plot at the Institute of Food
Security, Environmental Resources and Agricultural Research Federal University of
agriculture Abeokuta in Southwest

Crops: Pepper and tomatoes.

Setup: 0.1-hectare plot with one soil moisture sensor.
Results:
1. Water Savings: 25% reduction in water usage compared to manual methods.
2. Yield Improvement: 15% increase in crop yield due to consistent moisture levels.
3. User Feedback: Farmers reported improved ease of irrigation management.

Case Study 2: Demonstration at the maiden edition of the TETFUND National

Research Fair/Exhibition at the Eagle Square Abuja Nigeria.
Setup: Micro demonstration-plot with a network of sensors.
Results:
1. Energy Efficiency: Solar power sustained operations during peak sunny periods.
Real-Time Clock (RTC) works effectively to control when to irrigate and when not.
Regardless of aridity signals from the Soil moisture Sensor, the RTC ensured that the
solenoid valve does not open until the cool of the day. This proofs efficiency in
preventing irrigating at hot periods and scorching plant roots.
2. Scalability: Modular design allowed for easy expansion across additional plots.
3. Impact: Significant labor cost reductions and better crop uniformity.

Market Potential
Target Audience
SmartSprout is designed for:
• Smallholder farmers in water-scarce regions.
• Commercial farmers focusing on large-scale efficiency.
• NGOs promoting sustainable agricultural practices.

314
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

Market Growth
The global smart irrigation market, estimated at $1.5 billion in 2022, is expected to grow to
$3.2 billion by 2030, with a CAGR of 10%.

Impact Assessment
Social Impact
SmartSprout empowers smallholder farmers by providing them access to advanced
technology, improving crop yields, and enhancing food security in resource-constrained
regions.

Economic Impact
Efficient water management reduces operational costs, increases profitability, and fosters
sustainable economic growth in rural communities.

Environmental Impact
By optimizing water usage, SmartSprout significantly reduces water waste and contributes to
biodiversity conservation.

Scaling and Future Work

Scaling Strategies
• Production: Establishing partnerships with manufacturers to streamline component
procurement.
• Market Outreach: Tailored marketing campaigns targeting water-scarce regions.
• User Support: Comprehensive training programs for farmers.

Future Enhancements
• Machine Learning Integration: Predictive algorithms for irrigation scheduling
based on weather forecasts.
• Advanced Sensors: Integration of nitrate and pH sensors for nutrient monitoring.
• Mobile Application: Real-time monitoring and control via an intuitive user interface.

CONCLUSION

SmartSprout offers a transformative solution for addressing water scarcity and improving
irrigation practices. By combining IoT technology with sustainable design principles, it
empowers farmers, enhances productivity, and reduces environmental impact.

REFERENCES

1. Bahn, R. A., Yehya, A. A. K., & Zurayk, R. (2021). Digitalization for sustainable agri-food
systems: potential, status, and risks for the MENA region. Sustainability, 13(6), 3223.
2. García L, Parra L, Jimenez JM, Lloret J, Lorenz P. IoT-Based Smart Irrigation Systems: An
Overview on the Recent Trends on Sensors and IoT Systems for Irrigation in Precision
Agriculture. Sensors. 2020; 20(4):1042. https://doi.org/10.3390/s20041042
3. Obaideen, K., Yousef, B. A., AlMallahi, M. N., Tan, Y. C., Mahmoud, M., Jaber, H., &
Ramadan, M. (2022). An overview of smart irrigation systems using IoT. Energy
Nexus, 7, 100124.
4. Shamsudeen, S. (2021). Automated Soil Moisture Sensor Irrigation System using
Microcontroller. An unpublished undergraduate research report at the department of
computer Science, Federal University, Dutsima. Pg 78

315
 3rd African Conference on Precision Agriculture | 3-5 December | 2024

316
 www.PAafrica.org | #AfCPA2024

www.ispag.org
www.apni.net um6p.ma www.paafrica.org/AAPA

PRECISION AGRICULTURE in ACTION for AFRICA