Correlation and path coefficient analysis of yield and yield
attributing character of maize genotype
Anka Kumari Limbu
Gokuleshwor Agriculture and Animal Science College, Baitadi, Nepal
Abstract:
A comprehensive study in yield potential is crucial within the selection among maize
genotypes. Grain weight alone doesn’t decide the portion of economic yield during a harvest
index as grain yield is that the subtractions of the many yield components. Information of
associations among traits and yield implies their relative importance for selection in yield
improvement. Correlation study may be a widely used measure to work out the degree of
relationship between various traits. The trail analysis may be a powerful measure which
partitions the coefficient of correlation and deals with the relative importance of yield
contributing traits on yield. The correlation and path coefficient analysis of grain yield in
maize is reviewed within the paper.
Keywords: Maize, grain yield, correlation, path coefficient analysis.
1.0 Introduction
Maize (Zea mays L.) is globally important as a favored staple food for many people in
Sub-Saharan Africa, Latin America and Asia (Selvaraj & Nagarajan, 2011). It is the third most
important cereal crop after wheat and rice as it provides raw materials for agriculture based
industries in most growing regions of the word (Anees, et al., 2016). The crop is a reliable source
of nourishment for humans, poultry, animals and livestock and its demand is increasing due to its
versatile uses (extraction of starch, ethanol etc). The increase in demand of maize could partially
be addressed either by bringing more area under maize cultivation or by increasing the
productivity of the crop through the development and adoption of high yielding hybrids (Jambagi
& Wali, 2016). Maize grain yield is a complex quantitative trait that depends on a number of
interacting factors: environmental conditions and various growth and physiological processes
throughout the crops life cycle. Interrelationships exist between yield and its contributing
components. An insight into such interrelationships can significantly improve the efficiency of
breeding program (Pavlov, et al., 2015) by providing appropriate selection indices. Direct
selection for yield is often misleading as it is highly influenced by unpredictable environmental
components (Talebi et al., 2007). In this regard, correlation coefficient analysis is useful in
providing knowledge for the selection of several traits simultaneously influencing yield (Menkir,
2008).
The correlation studies simply measure the associations between yield and other traits.
Correlation between various characters is of great value as it indicates the degree to which
various characters of a plant are associated with the economic productivity (Ahsan et al., 2008).
The association between two characters can be directly observed as phenotypic correlation, while
genotypic correlation expresses the extent to which two traits are genetically associated (Pavlov,
et al., 2015). Both genotypic and phenotypic correlations among and between pairs of agronomic
traits provide scope for indirect selection in a crop breeding program (Muhammad &
Muhammad, 2001).
2.0 Path Coefficient Analysis
Path coefficient analysis has been widely used in crop breeding to determine the nature of
relationships between grain yield (response variable) and its contributing components (predictor
variables), and to identify those components with significant effect on yield for potential use as
selection criteria (Mohammadi et al., 2003). The major advantage of path analysis is that it
permits the partitioning of the correlation coefficient into its components. One component being
the path coefficient that measure the direct effect of a predictor variable upon its response
variable; the second component being the indirect effect (s) of a predictor variable on the
response variable through another predictor variable (Dewey & Lu, 1959).
Knowledge of correlation between yield and its contributing characters is a basic and fore most
endeavor to find out guidelines for plant selection. Partitioning of total correlation into direct and
indirect effects by path coefficient analysis helps to make the selection more effective (Priya &
Joel, 2009). In view of the above, studies were conducted to estimate direct and indirect effects
of four agronomic traits on maize grain yield by the application of the simple coefficient
correlation and path coefficient analysis.
Fig. 1. Causal relationships of the path-coefficient analysis for birdsfoot trefoil seed production;
six irrigation treatments were used. Doubled arrowed lines indicate mutual associations that are
measured by correlation coefficients (rij), and single-arrowed lines represent direct influence as
measured by path-coefficients (Pij).
In the path diagram the double-arrowed lines indicate mutual association as measured by
correlation coefficients, rij, and the single-arrowed lines represent direct influence as measured
by path coefficients, Pij.
3.0 Biometric models used in maize hybrids
Several statistical models have been use to evaluate the performance of maize hybrids.
Models that allow the partition of genotype x environment interaction into environmental and
genetic components are useful to evaluate the adaptability and stability of hybrids, especially in
assessment of value for cultivation and use. Mixed models with fixed and random variance
components has also proved efficient to identify promising hybrids in breeding programs
(Baretta et al., 2016). Knowledge of association degree between traits is of fundamental
importance in plant breeding programs. This importance increases, especially if some desirable
trait present difficulty in assessment, or low heritability (Cruz et al., 2014). The Pearson product-
moment correlation coefficient (Pearson, 1920), has been widely used for this purpose. Although
this correlation reveals the direction and degree of linear association between a pair of traits, it
does not show interrelationships of cause and effect. Thus, Sewall Wright in his work entitled
'Correlation and causation' (Wright, 1921) proposed a method known as 'path analysis' allowing
this understanding. The method is based on partitioning of the linear correlation coefficient into
direct and indirect effects of a group of explanatory traits on response of a dependent trait. Path
analysis has been highlighted in breeding area because the selection aiming improving a
desirable trait that has difficulty- measure and low heritability, can be indirectly carried out by
another trait, directly associated with desirable trait, but that shows high heritability and is easy
to measure. In maize, as well as in several world-important crops, studies using path analysis has
been successful in the sense of revealing the interrelationships between traits, be them yielding,
grain quality or the effects of interaction genotype x environment or management of cultivation
(Adesoji et al, 2015, 2015; Jadhav et al, 2014; Ma et al, 2015; Nardino et al, 2016). Studies with
path analysis in maize were succeeded in revealing the interrelationships between yield
components. In summary, the results converge to a common conclusion: the number of kernels
per ear and thousand-kernel weight are the traitswith greater direct association with grain yield
(Adesoji et al., 2015; Khameneh et al., 2012; Mohammadi et al., 2003; Reddy et al., 2012). As
the heritability in the broad sense of these trait is high (> 0.90), indirect selection from these
traits aiming at increasing grain yield (trait highly influenced by the environment) can be
effective (Ojo et al., 2006).
4.0 Path analysis conception
Path analysis is originally based on ideas developed by Sewall Wright (Wright, 1921),
however from its conception to the method's consolidation, some disagreement about the
reliability of the mathematical method that explains the relationships of cause and effect were
observed. In 1922, Henry E. Niles, in his paper entitled “Correlation, Causation and Wright's
theory of path coefficients”, made a criticism of the method proposed by Wright, claiming that
the philosophical basis of the path coefficients method was doubtful. Niles, testing Wright's
method, had observed in some of its results correlations exceeding | 1 |, saying “these results are
ridiculous” and that Wright would have to provide much more convincing evidences than he was
presenting (Niles, 1922). In the following year, Sewall Wright in his paper entitled “The theory
of path coefficients: a reply to Niles's criticism”, consolidates his method concluding that Niles
seemed to be based on incorrect mathematical concepts, result of a failure to recognize that path
coefficient it is not a symmetric function of two traits, but it necessarily has direction. Wright
concludes his work by stating that the path analysis does not provide a formula to infer causal
relationships from knowledge of the correlations; it is, however, within certain limitations, a
method of evaluating the logical consequences of a causal hypothesis relationship in a system of
correlated traits. It adds that the criticism offered by Niles nothing invalidates the theory or
application of path coefficient (Wright, 1923). Currently, the statistical method of path
coefficient is consolidated and worldwide used in several areas of science. In order to estimate
path coefficients, normal equations models are used to partition the linear coefficients into direct
and indirect effects of a set of explanatory traits on a dependent trait.
5.0 Path analysis estimation
After obtaining linear correlation estimates (r), partitioning of linear correlations into
direct and indirect effects of an explanatory dataset with p-traits can be performed by derivation
of the set of normal equations (X'Xβ = X'Y) in order to estimate parameters of multiple
regression using OLS. Thus, β estimate is given by: β = X’X-1 X’Y, where β is the partial
regression coefficient (β1, β2, β3,...,βp) to p + 1 rows; X’X-1is the inverse of linear correlation
matrix among explanatory traits; and X’Y is the correlation matrix between each explanatory
trait with the dependent trait. After estimating the regression coefficients (βp), the direct and
indirect effects of a set of p- explanatory trait towards the dependent trait can be estimated.
Consider the following example, where a set of explanatory traits (a, b, c and d) are used to
explain the relationship of cause and effect on the response of dependent variable (y). After
partial regression estimations (β1, β2, β3 eβ4), direct and indirect effects of ‘a’ on ‘y’ are given
by: ra:y = β1+ β2ra:b +β3ra:c +β4ra:d, where ra:y is the linear correlation between ‘a’ e ‘y’, β1 is
the direct effect of ‘a’ on ‘y’; β2a:bis the indirect effect of ‘a’ on ‘y’ via ‘b’, β3a:c is the indirect
effect of ‘a’ on ‘y’ via ‘c’ andβ4a:dis the indirect effect of ‘a’ on ‘y’ via ‘d’. Similar equations
are used in order to estimate direct and indirect effects of b, c, and d. The coefficient of
determination of the model, i.e., how much of the variance in the dependent trait is explained by
the interrelationship on explanatory traits, is given by R2 = β1ra:y + β2rb:y +β3rc:y+β4rc:y.
6.0 Methodology
Information related to the correlation and path coefficient analysis was collected using different
national and international open access journals. Furthermore, articles, conference papers,
bulletins, newspaper articles, websites were also used to collect the information regarding the
yield and yield attributing character of maize genotype.
Yield and yield attributing traits
1000- kernel weight
The weight of 1000- kernel weight of samples was recorded along with their moisture meter and
then it was converted to 15% moisture content by using formula:
1000- kernel weight (12.5% moisture adjusted) = Kernel weight × (100 – moisture%)
100-12.5
Number of ears per hectare
Number of ear per hectare was calculated by counting the entire ear number in the row
within plot and converting it into hectare by using following formula:
NOEH = No. of ears per plot ×10000 (m2)
Plot size (m2)
Number of rows per cob
Number of kernel rows from a randomly selected ear from each of five randomly selected
cobs was counted and average value was taken.
Cob length
Length of the ear was measured and recorded in centimeters at the time of harvest as its
total length (from the base to the tip of the ear) for five selected ear and average values was
taken.
Cob diameter
From five randomly selected plant ear (cob) diameter was recorded and average value
was taken.
Quantitative traits
Days to 50% anthesis
The number of days taken from the days of sowing to the day on which 50% of the plants
in a treatment plot showed full tassel emergence with shedding of pollens was recorded as days
to 50% tasseling.
Days to 50% silking
The number of days taken from the days of sowing to the day on which 50% of the plants
in a treatment plot showed complete emergence of silk was recorded as days to 50% silking.
Anthesis Silking Interval (ASI)
This is very important parameter to be determined for the appropriate pollination of
maize. The interval between days to 50% tasseling and days to 50% silking is recorded as the
anthesis silking interval (ASI).
Reproductive growth period
The number of days from complete tasseling to the days on which maturity is gained for
cobs was recorded as days to reproductive growth period.
Plant architectural traits
Plant height
Plant height was expressed in centimeters by measuring the plant stalk from the ground
level to the base of tassel branch of the matured plant. The plant height was observed from the
reproductive five tagged plants.
Ear height
The height of the plant from ground level upto the base of the upper most bearing
internode from where cobs emerge (major cob) was recorded as ear height in centimeters.
Acknowledgements
I express my deepest sense of gratitude and indebtedness from the core of my heart
to Assistant Professor Dr. Mukti Ram Poudel for his guidance and support for understanding
the research methodology and preparation of the manuscript.
Conclusion
Effective selection is important for yield improvement in plant breeding. Information of
associations among traits and yield clarifies their relative importance for selection in yield
improvement. Test weight, days to maturity, biomass and harvest index are correlated with yield
so can be considered as the good estimates for effective selection of high yielding maize
genotypes. Path analysis showed that plant height, number of rows per cob, cob length, cob
diameter, flag leaf area, days to flowering, days to maturity, peduncle length, grain- filling
periods, biological yield, and harvest index had positive direct effect on grain yield. These traits
could be considered as good contributors to grain yield.
References
Adesoji, A.G., Abubakar, I.U., and Labe, D.A. 2015. Character Association and Path Coefficient
Analysis of Maize (Zea mays L.) Grown under Incorporated Legumes and Nitrogen. Journal of
Agronomy, 14:158–163.
Ahsan, M., Hader, M. Z., Saleem, M. and M., Islam (2008): Contribution of various leaf
morpho-physiological parameters towards grain yield in maize. International Journal of
Agricultural and Biology: 546-550.
Anees, M. U., Khan, H. Z., Ahmad, Z., Akhtar, M. J., Ahmad, A., Choudhary, F. A. and Ahmad,
N (2016): Role of organic amendments and micronutrients in aize (Zea mays L.) sown on
calcareous soils. American-Eurasian Journal of Agriculture and Environmental Science,
16(4):795-800
Baretta, D., Nardino, M., Carvalho, I.R., Oliveira, A.C. de, Souza, V.Q. de, and Maia, L.C. da,
2016. Performance of maize genotypes of Rio Grande do Sul using mixed models. Científica,
44:403-411.
Cruz, C.D., Carneiro, P.C.S. and Regazzi, A.J., 2014. Modelos Biométricos Aplicados ao
Melhoramento Genético, 3rd ed. UFV, Viçosa, MG.
Dewey, D. R. and K. H. Lu (1959): A correlation and path coefficient analysis of components of
crested wheat grass seed production. Agronomy Journal, 51:518
Jadhav, N.H., Kashid, D.N., and Kulkarni, S.R. 2014. Subset selection in multiple linear
regression in the presence of outlier and multicollinearity. Statistical Methodology, 19:44–59.
Jambagi B. P. and M. C. Wali (2016): Heritability, correlation and path coefficient analysis in
maize germplasm for starch and oil content. Journal of Farm Science 29(2):257-260
Khameneh, M.M., Bahraminejad, S., Sadeghi, F., Honarmand, S.J., and Maniee, M. 2012. Path
analysis and multivariate factorial analyses for determining interrelationships between grain
yield and related characters in maize hybrids. African Journal of Agricultural Research, 7:6437–
6446.
Ma, Z., Qin, Y., Wang, Y., Zhao, X., Zhang, F., Tang, J., and Fu, Z., 2015. Proteomic analysis of
silk viability in maize inbred lines and their corresponding hybrids. PLoS One 10: e0144050.
Menkir, A., (2008): Genetic variation for grain mineral content in tropical-adapted maize inbred
lines, Food Chemistry, 110:454-464
Mohammadi, S. A., Prasanna, B. M. and N. N. Singh (2003): Sequential path model for
determining interrelationships among grain yield and related characters in Maize. Crop Science,
43:1690-1697.
Mohammadi, S.A., Prasanna, B.M., and Singh, N.N. 2003. Sequential Path Model for
Determining Interrelationships among Grain Yield and Related Characters in Maize. Crop
Science 43:1690–1697.
Muhammad, Y. and S. Muhammad (2001): Correlation analysis of S1 families of maize for grain
yield and its components. International Journal of Agriculture and Biology:387-388.
Nardino, M., Souza, V.Q. de, Baretta, D., Konflanz, V.A., Carvalho, I.R., Follmann, D.N., and
Caron, B.O., 2016. Association of secondary traits with yield in maize F1’s. Ciência Rural
46:776–782.
Niles, H.E. 1922. Correlation, Causation and Wright’s Theory of “Path Coefficients.” Genetics,
7:258–273.
Ojo, D.K., Omikunle, O.A., Oduwaye, O.A., Ajala, M.O., Ogunbayo, S.A., 2006. Heritability,
character correlation and path coefficient analysis among six inbred-lines of maize (Zea mays
L.). World Journal of Agricultural Sciences, 2:352–358.
Pavlov, J., Delic, N., Markovic, K., Crevar, M., Camdzija, and M. Stevanovic (2015): Path
analysis for morphological traits in maize (Zea mays L.). Genetika, 47(1):295-301
Pearson, K. 1920. Notes on the history of correlation. Biometrika, 13:25–45.
Priya, A. A. and A. J. Joel, (2009): Grain yield response of rice cultivars under upland condition.
Elect. Journal of Plant Breed., 1:6-11
Reddy, V.R., Jabeen, F., Sudarshan, M.R., and Rao, A.S. 2012. Studies on genetic variability,
heritability, correlation and path analysis in maize (Zea mays L.) Over locations. International
Journal of Applied Biology and Pharmaceutical Technology, 4:196–199.
Selvaraj, C. I. and P. Nagarajan (2011): Interrelationship and path- coefficient studies for
qualitative traits, grain yield and other yield attributes among Maize (Zea mays L.). International
Journal of Plant Breeding and Genetics 5(3):209-223
Talebi, R., Fayaz, F. and N. A. B. Jeloder (2007): Correlation and path coefficient analysis of
yield and yield components of chickpea (Cicer arietinum L.) under dry condition in the west of
Iran. Asian Journal of Plant Science, 6:1151-1154
Wright, S. 1921. Correlation and causation. Journal of agricultural research, 20:557–585.
Wright, S. 1923. The Theory of Path Coefficients a Reply to Niles’s Criticism. Genetics, 8:239–
255.
