FAO

ISSN 2070-7010
FISHERIES AND
AQUACULTURE
TECHNICAL
PAPER

669

Handbook for data collection

on recreational fisheries in the
Mediterranean and the Black Sea
Handbook for data collection
FAO
FISHERIES AND
AQUACULTURE

on recreational fisheries in
TECHNICAL
PAPER

the Mediterranean and the 669

Black Sea

Fabio Grati
Recreational Fisheries Specialist
General Fisheries Commission for the Mediterranean
Food and Agriculture Organization of the United Nations
Rome, Italy

Anna Carlson
Fishery Officer
General Fisheries Commission for the Mediterranean
Food and Agriculture Organization of the United Nations
Rome, Italy

Paolo Carpentieri
Fishery Resources Monitoring Specialist
General Fisheries Commission for the Mediterranean
Food and Agriculture Organization of the United Nations
Rome, Italy

Jacopo Cerri
Research Associate
Faculty of Mathematics, Natural Sciences and Information Technologies
University of Primorska
Koper, Slovenia

FOOD AND AGRICULTURE ORGANIZATION OF THE UNITED NATIONS

Rome, 2021
Required citation:
Grati, F., Carlson, A., Carpentieri, P. & Cerri, J. 2021. Handbook for data collection on recreational fisheries in the
Mediterranean and the Black Sea. FAO Fisheries and Aquaculture Technical Paper No. 669. Rome, FAO.
https://doi.org/10.4060/cb5403en

The designations employed and the presentation of material in this information product do not imply the expression of any
opinion whatsoever on the part of the Food and Agriculture Organization of the United Nations (FAO) concerning the legal
or development status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers
or boundaries. The mention of specific companies or products of manufacturers, whether or not these have been patented,
does not imply that these have been endorsed or recommended by FAO in preference to others of a similar nature that are
not mentioned.

The views expressed in this information product are those of the author(s) and do not necessarily reflect the views or policies
of FAO.

ISSN 2070-7010 [Print]

ISSN 2664-5408 [Online]

ISBN 978-92-5-134630-3
© FAO, 2021, last updated 18/10/2021

Some rights reserved. This work is made available under the Creative Commons Attribution-NonCommercial-ShareAlike
3.0 IGO licence (CC BY-NC-SA 3.0 IGO; https://creativecommons.org/licenses/by-nc-sa/3.0/igo/legalcode).

Under the terms of this licence, this work may be copied, redistributed and adapted for non-commercial purposes, provided
that the work is appropriately cited. In any use of this work, there should be no suggestion that FAO endorses any specific
organization, products or services. The use of the FAO logo is not permitted. If the work is adapted, then it must be licensed
under the same or equivalent Creative Commons licence. If a translation of this work is created, it must include the following
disclaimer along with the required citation: “This translation was not created by the Food and Agriculture Organization of the
United Nations (FAO). FAO is not responsible for the content or accuracy of this translation. The original [Language] edition
shall be the authoritative edition.”

Disputes arising under the licence that cannot be settled amicably will be resolved by mediation and arbitration as described
in Article 8 of the licence except as otherwise provided herein. The applicable mediation rules will be the mediation rules of
the World Intellectual Property Organization http://www.wipo.int/amc/en/mediation/rules and any arbitration will be conducted
in accordance with the Arbitration Rules of the United Nations Commission on International Trade Law (UNCITRAL).

Third-party materials. Users wishing to reuse material from this work that is attributed to a third party, such as tables, figures
or images, are responsible for determining whether permission is needed for that reuse and for obtaining permission from the
copyright holder. The risk of claims resulting from infringement of any third-party-owned component in the work rests solely
with the user.

Sales, rights and licensing. FAO information products are available on the FAO website (www.fao.org/publications) and
can be purchased through [email protected]. Requests for commercial use should be submitted via: www.fao.org/
contact-us/licence-request. Queries regarding rights and licensing should be submitted to: [email protected].

Cover photograph: Ancona Italy, Conero Promontory Adriatic Sea. ©FAO/Davide Fagioli
 iii

Preparation of this document

This handbook has been prepared by the General Fisheries Commission for the
Mediterranean (GFCM) of the Food and Agriculture Organization of the United
Nations (FAO) to address the priorities identified by Mediterranean and Black Sea
countries in the context of existing international commitments and regional strategies.
The impetus for the handbook development originated with the GFCM’s mid-term
strategy (2017–2020) towards the sustainability of Mediterranean and Black Sea fisheries
(mid-term strategy), namely its Target 2 which aims at supporting livelihoods for coastal
communities through sustainable small-scale fisheries. In particular, one output of this
target was the collection of robust and timely information on the impacts of small-scale
fisheries and recreational fisheries on marine living resources and on their interactions
with other human activities in coastal communities. This document addresses this
output by providing a clear methodological framework to allow Mediterranean and
Black Sea countries to implement suitably harmonized sampling and survey monitoring
schemes for recreational fisheries and is foreseen to support continued improvement
in recreational fisheries data collection, as foreseen in the GFCM 2030 Strategy for
sustainable fisheries and aquaculture in the Mediterranean and the Black Sea.
The collection of recreational fisheries data is limited in many countries and the
lack of reliable estimates of catch, effort and socio-economic data has led to the
exclusion of recreational fisheries data from stock assessments, with implications
for fisheries management. The GFCM Scientific Advisory Committee on Fisheries
(SAC) has highlighted the potential issues posed by this lack of data, particularly
for stocks which are overexploited by commercial fisheries and for which
recreational fisheries might be an additional component of fishing mortality. At the
same time, it has been observed that the data-poor nature of recreational fisheries
also undermines the sustainable development of this sector, in light of its potential
for positive socio-economic contributions to coastal communities. In light of this,
this handbook provides information on the basic set of information necessary for
monitoring recreational fisheries and presents a framework for the implementation
of harmonized regional data collection, based on a standard methodology, in order
to facilitate the comparison of results in the Mediterranean and Black Sea region.
The handbook was elaborated under the expert guidance and overall coordination
of Fabio Grati (GFCM Recreational Fisheries Specialist), who prepared the first
draft, ensuring consistency with GFCM priorities and existing methodologies in
place, as well as applicability across the region. Anna Carlson (GFCM Fishery
Officer for Socio-Economic Issues), Paolo Carpentieri (GFCM Fishery Resources
Monitoring Specialist) and Jacopo Cerri (Consultant) also provided expert inputs
to the handbook’s preparation and revision.
At the suggestion of the twentieth session of the SAC (FAO, 2018a), the experience
from select pilot studies in the Mediterranean and the Black Sea was incorporated
into the handbook. The handbook also benefitted from extensive revisions from
the experts of the GFCM Working Group on Recreational fisheries (WGRF) in
2020. These consultations contributed to fine-tune the methodology and adapt the
handbook to the different characteristics and recreational fisheries scenarios found
in the Mediterranean and the Black Sea, so that it could be useful and replicable in
different countries and areas. The handbook was endorsed on the occasion of the
first meeting of the WGRF (GFCM, 2021).
iv

The publication of this work was coordinated by Dominique Bourdenet (GFCM

Scientific Editor) who ensured editing, layout and publishing, with the assistance of
Ysé Bendjeddou (GFCM Publication and Documentation Specialist) and Matthew
Kleiner (GFCM Editing and Communications Intern). The overall graphic design
and layout of the publication was managed by Chorouk Benkabbour.
This document was produced with the financial support of the European Union.
 v

Contents

Preparation of this document iii

Abstract viii
Acknowledgements ix
Abbreviations and acronyms x
Definitions xi
Background xiii

1. Introduction 1
1.1 Objectives of the handbook 1
1.2 Definition of recreational fisheries 1
1.3 Status quo 2

2. Data collection 5
2.1 Defining the target population 7
2.1.1 National license system 8
2.1.2 General population screening survey 9
2.1.3 Mandatory fee-free online registration 11
2.2 Sampling strategy 11
2.2.1 Non-probability sampling 12
2.2.2 Probability sampling 12
2.3 Stratifying the population 14
2.4 Estimating the sample size 15
2.5 Selecting the sample 16
2.6 Additional considerations 17

3. Methodology 19
3.1 Off-site surveys 19
3.1.1 Logbook surveys 19
3.1.2 Recall surveys 25
3.2 On-site surveys 26
3.2.1 Other on-site methods 28
3.3 Type of information to be collected 30
3.3.1 Fishing effort 30
3.3.2 Catch data 30
3.3.3 Economic data 33

4. Data analysis 35
4.1 Data quality check 35
4.2 Response and completion rates 36
4.3 Measuring central tendency and data dispersion within a sample 38
4.4 Estimators: estimating population mean, totals and variance 39
4.4.1 Simple estimations 40
4.4.2 In-depth estimations 40
vi

5. Stakeholder engagement 45

References 47

Annexes 53
Annex 1. GFCM area of application, subregions and geographical subareas 53
Annex 2. Codes for recreational fishing techniques 54
Annex 3. Template for screening survey and enrollment of fishers
in data collection panel 55
Annex 4. Templates for mandatory fee-free online registration of marine
recreational fishers 56
Annex 5. Template for logbook and/or recall survey 59
5.a. General information for logbook and/or recall survey 59
5.b. Catch information for logbook and/or recall survey 60
5.c. Released species information for logbook and/or recall survey 61
5.d. Expenditure information per fishing trip for logbook survey 62
5.e. Expenditure information per month for recall survey 63
Annex 6. Template for on-site surveys 64
6.a. General information for on-site survey 64
6.b. Catch information for on-site survey 65
6.c. Released species information for on-site survey 66
Annex 7. Fishing effort measurement 67

Tables

Table 1. Example of sample sizes needed by population size and level

of sampling error 15
 vii

Figures

Figure 1. Distribution of the main taxa targeted by recreational fisheries

across the GFCM subregions 3
Figure 2. Total survey error components linked to the steps in the
measurement and representational inference process 5
Figure 3. Process for the identification of a sample of recreational fishers
to participate in data collection 7
Figure 4. Extraction of a sample from the target population 7
Figure 5. Process for the identification of recreational fishers from a
general population screening 10
Figure 6. Process for the identification of recreational fishers from a
general population screening 12
Figure 7. Stratification of the target population 14
Figure 8. Example of computerized routine to select random samples 17
Figure 9a. Example of how to compile a logbook (general information) 22
Figure 9b. Example of how to compile a logbook
(retained species information) 23
Figure 9c. Example of how to compile a logbook
(released species information) 24

Plates

Plate 1. Measurement of total length in bony fish 32

Plate 2. Measurement of total length and carapace length in crustacean’s
decapoda 32
Plate 3. Measurement of dorsal mantle length in cephalopods 33
viii

Abstract

Marine recreational fisheries are an integral part of Mediterranean and Black Sea coastal
life and are commonly practiced throughout the region. However, despite their ubiquity
and potential socio-economic contribution, recreational fisheries are a data-poor sector.
Data collection programmes to monitor their impact are limited and can vary widely
from one country to another, thus impairing proper consideration of the recreational
fisheries sector in policy-making and undermining efforts towards sustainable fisheries
management at the regional level.
The main goal of this handbook is therefore to provide a clear methodological
framework to allow Mediterranean and Black Sea countries to implement suitably
harmonized sampling and survey monitoring schemes for recreational fisheries. The
handbook consists of five parts. A first section provides an introduction to the recreational
fisheries sector in the Mediterranean and Black Sea region, including the rationale for
improving data collection. The second section provides guidance on how to set up a
data collection programme, including how to define the target population – particularly
in the absence of an up-to-date census or complete licensing system – as well as how
to select a sample for data collection. Next, in section three, the handbook provides a
comprehensive explanation of the harmonized regional methodology for carrying out
data collection, including through a combination of on-site and off-site techniques. This
section also presents a minimum set of necessary information allowing for monitoring
recreational fisheries (namely, fishing effort data, catch data and economic data), while,
at the same time, allowing for flexibility to accommodate national specificities and data
collection needs. Section four then provides a short primer to guide readers through the
data analysis process. A final section highlights the importance of engaging stakeholders
in the data collection process and provides advice on how to do so.
 ix

Acknowledgements

This document was prepared under the overall direction of Abdellah Srour
(GFCM Executive Secretary) and benefitted from the expert guidance of
Miguel Bernal (GFCM Fishery Resources Officer) and from the support of
Margherita Sessa (GFCM Fishery Liaison Officer), as well as the staff of the
GFCM Secretariat. Special appreciation goes to the national focal points involved
in the recreational fisheries pilot studies for taking the time to provide highly
detailed feedback, share their experience and help improve the methodology
proposed so as to enhance its practicality and usefulness. In particular, Marouene
Bdioui, Luca Bolognini, Murat Erbay, Naciba Labidi Neghli and Imad Lahoud
are warmly thanked. Valuable comments were also received from numerous
experts, in particular Jorge Baro, Valerio Crespi, Arnau Luk Dedeu Dunton,
Teresa García, Bruce Hartill, Matías Lozano, Sofiane Mahjoub, Ariadna Purroy
Albet, Valerio Sbragaglia, the members of the Mediterranean Advisory Council, as
well as other members of the GFCM Working Group on Recreational Fisheries;
their contribution to enhance the quality of this product is fully recognized.
The guidance of the members of the GFCM Scientific Advisory Committee on
Fisheries and of the GFCM Working Group on the Black Sea is also gratefully
acknowledged.
x

Abbreviations and acronyms

ABS address-based sampling

CPCs contracting parties and cooperating non-contracting parties
(GFCM)
CPUE catch per unit effort
FAO Food and Agriculture Organization of the United Nations
GFCM General Fisheries Commission for the Mediterranean
GSA geographical subarea (GFCM)
ICES International Council for the Exploration of the Sea
MEDAC Mediterranean Advisory Council
Mid-term mid-term strategy (2017–2020) towards the sustainability of
stategy Mediterranean and Black Sea fisheries (GFCM)
RDD random digit dialing
RPAS remotely piloted aircraft systems
SAC Scientific Advisory Committee on Fisheries (GFCM)
SCESS Sub-Committee on Economic and Social Sciences (GFCM)
WGRF Working Group on Recreational fisheries (GFCM)
 xi

Definitions

Angling Fishing with handlines, fishing rods and/or poles using

natural and/or artificial baits.
Avidity The frequency of fishing trips undertaken over a
commonly defined period.
Catch The total number or weight of individuals caught
during fishing operations, including fish that were
caught and released.
Catch-and-release The process of capturing a fish, usually by angling,
and releasing it alive. Catch-and-release procedures
range from legally required to mandatory release of
protected sizes and species to voluntary catch-and-
release of fish that could have been retained.
Fishing effort The amount of fishing gear of a specific type used
on the fishing grounds over a given unit of time (e.g.
total number of fishing days by fleet segment).
Geographical Geographical subareas were established by the GFCM
subareas (GSAs) within its area of application (Mediterranean and
Black Sea) in order to compile data, monitor fisheries
and assess fisheries resources in a georeferenced
manner (see Annex 1).
Harvest The part of the catch that is kept, not released.
Jurisdiction A province or territory with recreational fisheries
management responsibility.
Logbook survey A survey of recruited fishers who are asked to record
their effort and/or catch in supplied logbooks.
Mail survey Data collected through questionnaires sent to
recipients by post asking for information about
previous fishing activity, catch or expenses.
Non-resident Someone that fishes in a particular area, but is
excluded from the resident sampling frame for
surveys in that area.
Off-site sampling Selecting respondents outside of areas where fishing
activity takes place or can be observed, e.g. household
and/or over the phone.
Online survey Questionnaire that can be completed over the
internet. Online surveys are usually created as web
forms with a database to store the answers and a
statistical software to provide analytics.
On-site sampling Selecting respondents at principal areas of activity,
e.g. fishing sites.
xii

Panel survey An ongoing survey of a group of fishers who have

been enrolled in a panel for a fixed period.
Random digit dialing A method of selecting people for telephone statistical
(RDD) surveys by generating telephone numbers at random.
Recreational fisheries A non-commercial fishing activity exploiting marine
living resources for recreation, tourism or sport.
Screening survey A survey to identify the target population of
recreational fishers and their fishing characteristics.
Sport fishing An organized activity involving free competition
between fishers to catch the largest fish of certain
species, the largest number of specimens or the
largest total weight, depending on the rules of each
particular competition.
Survey A method of gathering information from a number
of individuals, known as a sample, in order to learn
something about the larger population from which
the sample is drawn.
 xiii

Background

Marine recreational fisheries, including sport fishing, is an integral part of Mediterranean

and Black Sea coastal life and communities. It has high cultural importance in the
region and represents an important economic component of coastal tourism, which
constitutes one of the region’s main maritime sectors in terms of gross value added and
employment. Nevertheless, despite the perceived socio-economic benefits, the lack of
reliable catch estimates has resulted in recreational fisheries being excluded from stock
assessments. This can be challenging for assessing stocks that are already overexploited
by commercial fisheries and for which recreational fisheries might present an additional
component of fishing mortality. Such a shortage of catch data, coupled with the limited
availability of data on the socio-economic impact of recreational fisheries, impairs
proper consideration of this sector in policy-making and undermines efforts toward
the sustainable management of fish stocks (Hyder et al., 2014). The data-poor nature of
the recreational fisheries sector also compromises the potential of its development for
positive socio-economic contributions to coastal communities (Arlinghaus et al., 2019).
Considering that the main objective of the General Fisheries Commission for the
Mediterranean (GFCM) of the Food and Agriculture Organization of the United
Nations (FAO) is to ensure the conservation and the sustainable use, at the biological,
social, economic and environmental levels, of marine living resources, as well as the
sustainable development of aquaculture in the Mediterranean and in the Black Sea,
recreational fisheries activity needs to be duly considered. Catch mortality should
include all reported or estimated commercial fishing landings, plus landings from
recreational fisheries and subsistence fisheries and, ideally, estimates of post-release
mortality too. Such data boasts a wide range of existing or potential end users, including
national governments, the scientific community, as well as the GFCM.
Improved information on this sector will help to design effective and enforceable
control measures and to support the development of long-term regional management
plans and marine spatial planning. These are crucial issues that should be urgently
addressed in order to foster better management of marine living resources in the
Mediterranean and the Black Sea.
 1

1. Introduction

1.1 OBJECTIVES OF THE HANDBOOK

Mediterranean and Black Sea fisheries face serious challenges, with approximately
75 percent of the scientifically assessed stocks considered to be fished outside of
safe biological limits (FAO, 2020). To take concerted action towards improving
this situation, the General Fisheries Commission for the Mediterranean (GFCM)
of the Food and Agriculture Organization of the United Nations (FAO)
developed a programmatic multiannual mid-term strategy (2017–2020) towards
the sustainability of Mediterranean and Black Sea fisheries (mid-term strategy)
(GFCM, 2017a).
The implementation of the mid-term strategy sought to reverse the trend
in the status of commercially exploited stocks by means of a series of targets,
outputs and activities. In this context, Output 2.1 of Target 2, “Robust and timely
information on the impacts of small-scale fisheries and recreational fisheries on
living marine resources and on their interactions with other human activities in
coastal communities,” foresaw the establishment of a permanent working group
on recreational fisheries and the assessment of the impacts of recreational fisheries,
providing the impetus for preparing this handbook. The collection of recreational
fisheries data is still a recent development in many countries and there exists no
clear framework for the integration of these data in stock assessments or fishery
management.
The main goal of this handbook is therefore to provide a clear methodological
framework to allow Mediterranean and Black Sea countries to implement suitably
harmonized sampling and survey monitoring schemes for recreational fisheries.
The information that this handbook suggests collecting is considered as the
basic set necessary for monitoring recreational fisheries. At the same time, it is
important to take into account national specificities and data collection needs when
implementing a recreational fisheries monitoring programme, including whether
the collection of additional information, such as social data or data on interactions
with vulnerable species, would be necessary.

1.2 DEFINITION OF RECREATIONAL FISHERIES

In order to understand each other and communicate using a common language,
recreational fishers, managers, politicians and scientists need a proper definition of
recreational fisheries for research, management and legal purposes. Past discussions
on recreational fisheries within the context of GFCM statutory and technical
meetings have primarily focused on the identification of a harmonized definition
for recreational fisheries. Deliberations from the Transversal Workshop on the
Monitoring of Recreational Fisheries in the GFCM Area (GFCM, 2010a) and the
eleventh session of the former Sub-Committee on Economic and Social Sciences
(SCESS) (GFCM, 2010b) agreed on the following definition for recreational fishing:
“Fishing activities exploiting marine living aquatic resources for leisure or sport
purposes, from which it is prohibited to sell or trade the catches obtained”. It was
further specified that “leisure purposes” refers to “fishing practiced for pleasure”,
whereas “sport purposes” refers to “fishing contests practiced within an established
institutional framework which sets rules, collects data on catches and informs on the
outcomes of the event” (GFCM, 2010a). Building on these discussions, the following
2 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

definition has been adopted within two GFCM recommendations (Recommendation

GFCM/43/2019/2 on a management plan for the sustainable exploitation of blackspot
seabream in the Alboran Sea (geographical subareas 1 to 3) and Recommendation
GFCM/42/2018/1 on a multiannual management plan for European eel in the
Mediterranean Sea) (GFCM, 2019):
“Recreational fishing means a non-commercial fishing activity exploiting marine
living resources for recreation, tourism or sport”

This definition is considered to be the working definition of the GFCM, barring

further decision-making by the GFCM.
It must be noted, however, that there exists an array of definitions in the literature and
within national legislations pertaining to recreational fisheries and its constituent parts
and related sectors (Pawson, Glenn and Padda, 2008), with subsequent implications for
the regulation of these sectors at the national level. For example, in general, there are
some discrepancies between national legislations over the term “sport fishing”. In some
countries, “recreational” and “sport” fishing have different meanings, while in others,
they are used interchangeably (EAA, 2004). However, as in the GFCM definition,
some definitions interpret “sport fishing” as a type of recreational fishery that is more
sportive, competition-oriented and technically complex than general recreational or
leisure fishing (Pawson, Glenn and Padda, 2008).
Furthermore, national definitions differ over the role of subsistence fishing within
recreational fisheries. In reality, not all non-commercial fishing can be described as
purely recreational. In the Mediterranean and Black Sea region, it is common for fishing
activity to meet both recreational and personal consumption needs, with the catch
directly consumed by the fisher and/or his/her family. The FAO Technical Guidelines
for Responsible Fisheries touch on this issue by defining recreational fishing as “fishing
of aquatic animals (mainly fish) that do not constitute the individual’s primary resource
to meet basic nutritional needs and are not generally sold or otherwise traded on
export, domestic or black markets” (FAO, 2012).
Nevertheless, there is overwhelming consensus among various definitions at the
regional level that recreational fisheries have a non-commercial, non-profit purpose
and expressly exclude the sale of the catch (Hyder et al., 2017a).

1.3 STATUS QUO

In the Mediterranean and the Black Sea, some countries already collect specific types
of data, including estimates of recreational catch and release of bluefin tuna (Thunnus
thynnus), European eel (Anguilla anguilla) and elasmobranchs (European Commission,
2016). However, standard and harmonized monitoring programmes for recreational
fisheries, with statistically robust sampling designs, are not yet regularly implemented
in most countries. Therefore, with a view to moving towards an assessment of
recreational fisheries in the GFCM area of application, the GFCM proposed a roadmap
to pilot recreational fisheries assessments towards the development of a harmonized
regional methodology (GFCM, 2017b).
As a first step, in 2017, the GFCM circulated a questionnaire on national
marine recreational fisheries among its contracting parties and cooperating
non-contracting parties (CPCs). Preliminary information collected within the context
of this questionnaire shows that marine recreational fisheries in the Mediterranean
and the Black Sea involve many different techniques such as rod and line, speargun,
traps, longlines and hand-gathering (see Annex 2) that can be employed from different
locations (i.e. shore, boat, underwater) and target a broad range of taxa (e.g. finfish,
shellfish, crustaceans, etc.).
In the Black Sea, recreational fishers primarily target four taxa: Scombridae,
Gobidae, Mugilidae and Pomatomidae (primarily bluefish (Pomatomus saltatrix)).
1. Introduction 3

In the Mediterranean, however, the catch composition includes a higher number of

taxa, and slight variations in the target species are observed among the four GFCM
Mediterranean subregions (Annex 1). The following are targeted in all Mediterranean
subregions: bluefin tuna; small pelagics, particularly Scombridae such as Atlantic
mackerel (Scomber scombrus) and Atlantic bonito (Sarda sarda); large pelagics,
particularly Carangidae such as greater amberjack (Seriola dumerili) and leerfish (Lichia
amia); Coryphaenidae, particularly dolphinfish (Coryphaena hippurus); Sparidae,
particularly gilthead seabream (Sparus aurata) and common dentex (Dentex dentex);
and Cephalopoda, particularly European squid (Loligo vulgaris), common cuttlefish
(Sepia officinalis) and common octopus (Octopus vulgaris).
As noted above, subregional variations do occur, for example: Serranidae are
mostly represented by different species of grouper, which are targeted along the
western coast of the Adriatic Sea and on the rocky bottoms of the western, central
and eastern Mediterranean; Mugilidae and bluefish are mainly exploited in the eastern
Mediterranean and the Adriatic Sea; and Moronidae, which are represented exclusively
by the European seabass (Dicentrarchus labrax), are targeted in all countries bordering
the Adriatic, as well as in Egypt, Libya, Spain and Turkey. A summary of the main
nekton taxa targeted by recreational fisheries in the GFCM area of application is
provided in Figure 1. Contracting parties and cooperating non-contracting parties
for which national licensing systems for marine recreational fisheries are in place are
highlighted in dark gray.

FIGURE 1
Distribution of the main taxa targeted by recreational fisheries across the GFCM subregions

GFCM SUBREGIONS

Western Mediterranean Licence system

Central Mediterranean
Adriatic Sea
Eastern Mediterranean
Black Sea

Note: based on responses to the GFCM questionnaire on national marine recreational fisheries (2017).
Source: FAO, 2018b.
 5

2. Data collection

Decisions on how to monitor recreational fisheries depend on various factors,

including the goals of the survey, its geographical scale, available sampling frames,
the spatial distribution of fishing effort and the types of fishing methods used
(Hartill et al., 2012). Multiple methods exist to perform this task, each one with its
advantages and limitations, and various designs are available to obtain representative
estimates. As measuring the entire study area is impossible, survey sampling, in its
various forms (e.g. catch analysis, questionnaires) is usually the favoured approach;
by collecting a sample of observations, researchers try to obtain a comprehensive
representation of the phenomenon of interest.
A good conceptual framework for understanding how to design a survey, and
where eventual problems can arise, is the total survey error framework (Groves
and Lyberg, 2010) (Figure 2). This framework can be divided in two components:
representation and measurement.
Representation refers to the study’s potential for generalization: how well do the
interviewed fishers represent the whole fishing community in the study area? This
question presents two different approaches:
• Census surveys collect information from all the statistical units in the target
population (e.g. from all the recreational fishers that exist, at a certain time,
in the Mediterranean and the Black sea).
• Sample surveys collect information from a small group of statistical units
in the target population (e.g. only from some of the recreational fishers
that exist in the Mediterranean and the Black sea). When certain conditions
characterize data collection, findings from sample surveys can be generalized
to the whole population of statistical units.

FIGURE 2
Total survey error components linked to the steps in the measurement
and representational inference process

Measurement Representation

Inferential
Construct
population
Validity
Target
population
Measurement Coverage
error
Measurement Sampling
error frame
Sampling
Response error
Sample
Processing
Non-response
error
error
Edited data Respondents

Survey statistic

Source: Groves and Lyberg, 2010.

6 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

While census surveys paint an exhaustive and representative picture of a certain

phenomenon, sample surveys are far more common, for many different reasons:
• selecting a sample is less time-consuming than selecting every item in the
population;
• selecting a sample is less expensive than performing a census;
• census surveys are often unfeasible in practice and sometimes they are
unethical; and
• sample surveys can be easily repeated over time to track changes in the
phenomena they investigate, while censuses cannot be easily repeated.
On the other hand, the measurement component of the total survey error framework
(Figure 2) refers to the ability to adequately measure the relevant phenomenon, in
this case recreational fishing effort, catch and economic data. Various methods are
available to perform this task, ranging from in-depth qualitative interviews to simple
questionnaires (Vaske, 2008). These methods will be described in further detail in
Section 3.
In recreational fisheries, effort, catch and economic data are frequently collected
by means of sample surveys all around the world (Sparrevohn and Storr-Paulsen,
2012; Bellanger and Levrel, 2017). This handbook will guide readers through the
implementation of such a survey, as it is considered the most relevant approach
for harmonized data collection by all countries in the Mediterranean and the Black
Sea region. It is important to note that in the implementation of a sample survey,
errors may be introduced at different stages. As such, it is useful to consider the
total survey error framework (Figure 2) when conceptualizing the survey design in
order to minimize error to the greatest possible extent. A combination of multiple
data collection methodologies and different sampling approaches can help minimize
the total survey error, while providing researchers with considerable flexibility
in monitoring recreational fisheries. To this end, it may be useful to consider
complementing a traditional sample survey with on-site data collection and/or any
number of new monitoring technologies being used for recreational fisheries, such as
mobile applications for data collection (Venturelli, Hyder and Skov, 2017) and social
media data mining (Sbragaglia et al., 2019).
In the Mediterranean and the Black Sea, this methodological flexibility is important in
order to adapt to the different characteristics and recreational fisheries scenarios found.
This handbook presents a harmonized framework for data collection in the
Mediterranean and the Black Sea, while facilitating the necessary flexibility to adapt
to the different specificities of the region. The subsequent sections of this handbook
will guide readers through the process of defining a sample of recreational fishers to
participate in data collection, as outlined in Figure 3, as well as provide guidance on the
data to be collected and analysed.

Adaptability and flexibility

During the early years of data collection, it is best to focus on developing a complete
understanding of the methodology and being flexible enough to make customizations as
required. Setting up a simple but effective method will allow a country to move to more
advanced survey techniques in due course.
2. Data collection 7

FIGURE 3
Process for the identification of a sample of recreational fishers
to participate in data collection

TARGET POPULATION
SAMPLE
(marine recreational fishers)
(Panel of
TOTAL possible strategies


recreational
POPULATION
Database fishers for
Database Estimated from
of population data
mandatory
national from screening fee-free online collection)
licenses survey registration

Identification Random Selected fishers

contacted by mail,

of strata sampling
(if necessary)
and sample
 (simple or
stratified)
email or telephone
to see if agree
size to participate

Fishers
decline to
participate:
substitute

2.1 DEFINING THE TARGET POPULATION

In order to set up a sample survey, a sample should be selected by isolating some
statistical units from the target population, also known as the statistical universe
(Figure 4). The ultimate goal of sampling is to obtain an overview of a certain target
population from a subset of units.
The first step of any sampling strategy is therefore defining the target population to
which the results of the survey will be generalized. The population is the full list of units
for which the survey will be conducted and about which conclusions will be drawn, in
this case, the full population of marine recreational fishers. Sometimes, a complete list of
all the units comprising the population is available, though this is not always the case:
sampling methods therefore differ between populations with and without lists.

FIGURE 4
Extraction of a sample from the target population
8 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

Data sources for target populations may vary across Mediterranean and Black
Sea countries, and some methods for identifying the target population that may be
practical for some countries may not be feasible or cost-effective for others. Many
countries do not have licensing programmes and databases that can provide a complete
list of all recreational fishers. In fact, most of the compulsory recreational fisheries
license systems in force either grant registration exceptions for some participants or
do not ensure that all participants actually register or renew their licenses when they
expire. Similarly, while some countries boast active recreational fisheries federations
or associations which include a high number of fishers, the membership of these
organizations should only be considered as completely representing the target
population when membership is obligatory for all recreational fishers. With that said,
recreational fisheries federations and associations can serve as valuable partners in
engaging stakeholders in data collection (see Section 5).
Where recreational fisheries license programmes do exist, are obligatory and
cover all types of recreational fisheries, it is still worthwhile to consider the level of
compliance with the relevant license regulations. If non-compliance is high and fishing
without a permit is common, then alternative data sources for the target population
may need to be found to account for this higher overall number of fishers. In general, it
is important that data collection accounts for the peculiarities of each country’s sector,
while at the same time ensuring that national datasets are organized in a way so as to
eventually allow for their combination at the desired level in a statistically valid way.
On the other hand, sampling from populations without a list is more complex and
less straightforward, as it requires careful designs to estimate inclusion probabilities
through time-consuming methods of field sampling, such as aerial surveys, point-
counts or capture-recapture models (Zischke and Griffiths, 2014). It is therefore
suggested that, where complete national licensing systems or similar registries do not
exist, a simple sampling frame should be adopted, such as the general population or
all national households, for which lists are typically readily available. This approach
is considered more effective and more easily tailored to the specificities of the
Mediterranean and the Black Sea region than approaches based on sampling without a
list. In this light, the following sections outline three possible strategies that have been
identified as appropriate for defining the target population of recreational fishers in
Mediterranean and Black Sea countries, each one with its advantages and limitations.

2.1.1 National license system

The identification of recreational fisher populations is much easier and cost-effective
when information can be obtained from national marine recreational fisheries license
systems and registration databases. Direct list frames of fishers, or fishing vessel
operators, can be constructed from fishing license programmes, fishing permit
programmes or fishing club memberships (when registration with these programmes/
clubs is obligatory). Some fishers may participate in more than one list frame by being,
for example, both a license holder and a fishing club member. It may also be the case
that a list frame includes fishers who have not fished during the survey reference
period; this issue can, however, be accounted for at later stages in the study. A list
frame of fishers should identify license holders, when appropriate, by postal mailing
address, e-mail address, telephone number, mobile telephone number and, possibly, a
national identity (ID) or social security number. Ideally, fishing licenses should cover
all possible recreational fisheries categories and should identify the fishing category
(or categories) practiced by each license holder, namely fishing from the coast, a boat
and/or underwater fishing.
As of 2017, based on the data collected through the GFCM questionnaire on national
marine recreational fisheries, most license systems in force in the Mediterranean and
the Black Sea were dedicated to boat fishing, while coastal and underwater fishing,
2. Data collection 9

in many cases, did not require a license. Such data sources face potential limitations,
however, in the form of national confidentiality protection requirements, which
might impede the use of contact lists for survey purposes. Researchers should make
all attempts to avoid potential pitfalls related to these limitations, including through
familiarization with existing legal frameworks for data collection. For countries that
do not have a complete license system in place, alternative options are described
below. Suggested options include performing a screening survey or a mandatory fee-
free online registration. A screening survey could also be valid for those countries
without a complete license system in place (e.g. licenses are mandatory only for boat
fishing) to cover the missing portion of the recreational fisher population (e.g. shore
and underwater fishing).

2.1.2 General population screening survey

When a list of recreational fishers from a license system is not available or is
incomplete, it is possible to conduct a screening survey, sampling from a broad
coverage system, such as a complete frame of resident households. It may not be
necessary to conduct a screening survey every year; every two to three years would
be sufficient. It is preferable to use a screening survey only as a means to identify
recreational fishers for a more detailed follow-up survey. A flow chart outlining this
process is provided below (Figure 5). Therefore, the first correspondence could be
limited to determining if any household residents participate in recreational fisheries,
collecting their contact information and recruiting them for participation in a more
detailed follow-up survey. The survey should collect the minimal data needed to
define and profile the fishing population.
A template for the enrollment of fishers for data collection from a screening
survey is shown in Annex 3. First of all, there is a need to collect information on the
gender and age of all members of the household. The second question concerns who
went fishing at sea during the last year and how many times they went per fishing
mode (a rough estimate is sufficient). The last question is perhaps the most important
and concerns the respondent’s availability to enroll in a panel that will be contacted
by phone (mobile phone number would be ideal) every month for data collection.
Respondents agreeing to participate in this panel would then be provided with a
logbook (Annex 5) in order to keep records of requested information (as described
in Section 3.1.1).
When requesting this information, privacy concerns must be considered and so
it is recommended to consult national privacy laws prior to initiating this work.
Common principles for data sharing and dissemination should always be respected
when carrying out data collection, in line with the concept of privacy constituting a
basic human right as recognized by the United Nations.1
At the global level, both address-based sampling (ABS), i.e. complete lists of
residential mailing addresses for mail or face-to-face surveys of recreational fisheries,
and random-digit-dialing (RDD), i.e. directory-based telephone surveys that provide
access to a majority of a resident fishing population, have been widely used. For
over 30 years, RDD telephone surveys have served as the mainstay of the survey
research industry (Link et al., 2008). Over the past decade, however, participation
in most RDD telephone surveys has declined due, most likely, to factors such as
the growth of call-screening technologies, heightened privacy concerns in the face
of frequent telemarketing calls and the proliferation of non-household telephone
numbers, which are typically non-voice and unassigned numbers (Link et al., 2008).
1
Article 12 of the Universal Declaration of Human Rights (General Assembly of the United Nations,
1948) states “No one shall be subjected to arbitrary interference with his privacy, family, home or
correspondence, nor to attacks upon his honour and reputation. Everyone has the right to the protection
of the law against such interference or attacks.”
10 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

Additionally, RDD frames may exclude households without a landline telephone

(e.g. due to increased use of cellular telephones). Increasingly, though, these RDD
surveys are conducted using computer-assisted telephone interviewing technology
or, where georeferenced mobile phone information is available, computer-assisted
mobile interviewing, which help circumvent the problems associated with fewer
and fewer people having landline telephones. However, probability sample design
alternatives to RDD that are comparable in speed, efficiency and cost are scarce.
Address-based sampling is one such alternative that may provide survey research
with a cost-effective alternative to RDD, as the growth of database technology
has allowed for the development and maintenance of large, computerized dwelling
address databases. In New Zealand, an advanced face-to-face survey from a dwelling
list is already performed, although this may not offer an optimal solution for all
Mediterranean and Black Sea countries due to the high budgetary requirements for
its implementation.
In the event that both approaches are possible, ABS and RDD directory frames
should be compared and evaluated to determine which provides the most complete
coverage for an effective screening of resident recreational fishers in each country.
Ideally, in order to reduce biases that may result from the undercoverage of any one
list frame, it would be best to use a dual frame approach. In this way, it is possible
to assess the coverage of the list frame by comparing recreational fishers occurring
in both frames to those appearing only in one. Furthermore, one could consider
stratifying coastal and non-coastal municipalities and applying design weights
(e.g. 70 percent coastal, 30 percent non-coastal) in order to oversample coastal
municipalities, where a higher number of marine recreational fishers are expected to
be found.
The screening approach described in this section would therefore provide access
to the resident population, excluding non-resident (i.e. tourist) marine fishers. In
countries where tourists represent an important component of recreational fisheries,
it is necessary to implement a supplementary survey frame dedicated to non-
residents. A possible solution for creating a list of non-resident marine recreational
fishers could be to enforce a mandatory fee-free online registration as described in
the following section.

FIGURE 5
Process for the identification of recreational fishers from a general population screening

Recreational
Sample of residents contacted fishers
(mail, telephone, face-to-face)
and asked whether members
Screening of the household are
recreational fishers
survey of
resident Non-fishers
households

Non-coverage and
non-responses
2. Data collection 11

2.1.3 Mandatory fee-free online registration

When a complete list of recreational fishers is not available, a list of non-resident
fishers does not exist, and a screening survey is not feasible, then the third solution
would be to enforce the registration of participants through the implementation
of an online fee-free registration programme, which collects a valid name, address,
e-mail address and telephone number for each participant. This approach has been
recently endorsed by the Mediterranean Advisory Council (MEDAC) as a valid
method for the assessment of recreational fisheries in the Mediterranean (MEDAC,
2016). Such registration should be mandatory for both residents and tourists,
regardless of age and of whether recreational fishing takes place from the shore,
from a boat or underwater. The word “license” should not be used in this case in
order to avoid conflict and refusal from the part of the population (ICES, 2010);
the word “census” could be suggested instead. The use of an online registration
offers many advantages, including ease of access, time saving and efficient data
management. However, a possible source of bias could arise from the internet not
being user-friendly for certain groups of recreational fishers, such as the elderly –
although this was not found to be the case in a study in Spain by Gordoa, Dedeu
and Boada (2019). To avoid such bias, it is recommended that fishing shops assist
fishers in online registration and print a copy of the document for the fishers,
certifying their registration.
The first step in implementing such an approach would be to create a dedicated
online platform, which should be endorsed by the national administration in charge
of managing fishing activity, such as a ministry for example. This step entails minor
costs, as internet domains are relatively low-cost. Recreational fishers planning
to pursue their fishing activities in marine national waters should register online
by completing a number of mandatory fields with some general details (name,
e-mail, place and date of birth, nationality, etc.). Once the general profile has been
filled in, an ID number, valid for a lifetime, should be assigned to each fisher.
Afterwards, fishers should be required to complete a second form requesting a list
of compulsory supplementary data: type and avidity for every type of recreational
fishing practiced and name of an eventual affiliation to a marine recreational
fisheries association. In some cases, such as the Balearic Islands registration system,
users are also required to specify the main areas in which they fish. Once the
fishers have completed the compulsory data entry, a certificate should be delivered,
either directly through the registration website or sent via e-mail. This certificate
should be fee-free but mandatory, in order to perform any kind of marine fishing
in national waters. Fishers should be requested to print this certificate and keep it
with them whenever carrying out marine recreational fishing activities. A template
for a mandatory fee-free online registration is shown in Annex 4. It is desirable
that the online registration be linked to a national database, where all information
collected is organized and stored.

2.2 SAMPLING STRATEGY

Once the target population has been defined, observations (i.e. recreational fishers)
can be sampled according to two criteria: probability and non-probability sampling
(Figure 6).
12 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

FIGURE 6
Process for the identification of recreational fishers from a general population screening

Simple random
Probability sampling
samples
Sampling
methodologies Stratified random
Non-probability sampling
samples

2.2.1 Non-probability sampling

Non-probability sampling, also known as purposive sampling, describes a family
of sampling techniques (e.g. convenience sampling, haphazard sampling, purposive
sampling, expert sampling, diversity sampling, modal instance sampling, quota
sampling, etc.) where the odds of any member being selected for a sample cannot
be calculated and sampling relies on the subjective judgement of the researcher
(Sabatella and Franquesa, 2003). These methods present some advantages, such as
convenience, speed and low cost. However, with these surveys, it is impossible to
know how well the population is represented, as the results cannot be generalized.
A further bias comes from the fact that confidence intervals and margins of error
cannot be calculated, rendering the results meaningless (Cochran, 1977; Lohr, 1999;
Levine et al., 2008). This is the main reason why non-probability sampling should
not be considered in the quantification of recreational fisheries in the Mediterranean
and the Black Sea. Non-probability sampling should be considered only when some
particular conditions apply. For example, the use of online registration constitutes a
form of non-probability sampling, which does not allow for any formal inference.
However, on some occasions where the sampling frame is unavailable, it may be the
only feasible approach.

2.2.2 Probability sampling

Within probability sampling, sample unit selection is based on known probabilities
calculated from demographic data collected during the initial screening survey and
from data provided by the most recent national census. This approach allows the
researcher to make unbiased and mathematically sound inferences about the population
of interest (Levine et al., 2008). In sampling designs for populations with a list, the
two most common forms of probability sampling are simple random sampling and
stratified random sampling. These sampling methods have two features in common:
i) every element of the population has a known non-zero probability of being sampled;
and ii) random selection of the sample is applied (Pinello, Gee and Dimech, 2017).

Simple random sampling

In simple random sampling, all the units from the target population have the same
probability of being chosen. For example, if there is a list of all the recreational anglers
who fish in a certain coastal area and the annual number of gilthead seabream caught
by an angler within a fishing season must be determined, the procedure is as follows:
extract a random sample of anglers, ask them the number of gilthead seabreams they
landed over the course of the fishing season, calculate an estimator (Hankin, Mohr
and Newman, 2019) expressing the total or the average number of gilthead seabreams
that were landed and then calculate the associated variance of the estimate. Provided
the sample is large enough, it can be reasonably claimed that simple random sampling
2. Data collection 13

offers an adequate picture of the recreational gilthead seabream fishery in the coastal
area being investigated.

Stratified random sampling

Stratified random sampling, on the other hand, represents a suitable choice when
the target measurements vary between the units being sampled. Using the example
of a situation where a list of recreational fishers exists at the national level and the
goal is to estimate how many people are exclusive sea fishers (i.e. they do not fish in
freshwater), if simple random sampling is adopted, the sample of fishers might contain
respondents from only inland areas. This sample is therefore likely to be biased, as it
would underestimate the total number of sea fishers, as sea fishing is almost certainly
more common in coastal areas. Instead, respondents could be divided on the basis of
their geographical provenance, for example by creating two subgroups of respondents,
one from inland and one from coastal areas. Then fishers can be randomly sampled
from each one of these two groups: the estimates will be correct, as the observations
are correctly weighted. The two groups of respondents, from inland and from coastal
areas, are called strata and they have to be mutually exclusive: a fisher cannot be a
resident, at the same time, of a coastal area and an inland area.
Both simple random sampling and stratified random sampling are correct, meaning
that, if sampling is well designed, they provide researchers and managers with unbiased
estimates of the phenomenon of interest, and that the standard error of the estimates
can be calculated correctly (Hankin, Mohr and Newman, 2019). If the variable of
interest – say the probability of being an exclusive sea fisher – is strongly associated to
the strata, then stratified random sampling can provide researchers with more accurate
estimates. On the other hand, if there is no strong variation between strata, simple
random sampling is preferred, as inference from stratified random sampling might be
inaccurate. Choosing each one of these approaches should be carefully predicated on
the basis of evidence at hand.

Statistical weighting of survey data

A final approach is called weighting. Weighting offers a way to account for unbalanced
sampling, once data has been collected. This procedure is particularly useful when simple
random sampling is adopted but can also be applied to stratified random sampling
schemes. Weights are calculated for target populations catalogued in lists; they cannot
be calculated for populations without a list. As an example, imagine a random sample
of recreational boats, from which the total seasonal catch of common cuttlefish is to be
estimated. Random sampling is carried out, collecting a sample of 410 boats. It is realized,
however, that the sample is unbalanced in terms of fishing activity: while 40 percent of
fishing boats in the whole study area use specialized gear to catch common cuttlefish,
only 10 percent of boats in the sample use specialized gear to target common cuttlefish.
Recreational boats targeting common cuttlefish are therefore under-represented within
the sample, while boats targeting other species are over-represented, with consequent
errors (i.e. a deflated value) in estimated harvests of cuttlefish.
Weights could be estimated as (Vaske, 2008):
Population percentage
Weight=
Sample percentage

Therefore, weights for boats that target common cuttlefish correspond to:
40/10 = 4, while weights for boats that are targeting other species are equal to
60/90 = 0.66. By multiplying each boat’s reported catch by its corresponding weight,
the estimates are adjusted.
Weighting is a powerful tool to correct estimates but requires accurate knowledge
of the target population and is not always feasible. Multiple approaches are available,
14 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

including the use of multiple variables, or the use of weighting to correct non-response
bias. Relevant references, including survey method texts, can provide further details
(Vaske, 2008; Groves et al., 2009; Dillman, Smyth and Christian, 2014).

2.3 STRATIFYING THE POPULATION

Once the target population has been defined, either through a national license system,
a screening survey or a mandatory fee-free online registration, the sample size can
be estimated (i.e. number of observations) and the sample of recreational fishers can
be selected. However, in the case of a stratified random sampling, one further step is
required: there is a need to identify the strata. As previously mentioned, an important
principle is that strata should be mutually exclusive, as simply illustrated in Figure 7.
For example, if the respondents were divided equally between residents of coastal and
inland areas and a sample of 640 units was needed, 320 respondents (50 percent) would
therefore have to be sampled at random from the inland stratum and 320 respondents
(50 percent) from the coastal stratum.
Therefore, stratification could be based on residents’ spatial provenance, grouping
by criteria such as area of residence or specific jurisdiction (e.g. GFCM geographical
subarea (GSA) – see Annex 1, subnational, region, port, etc.) rather than on fishing
habits (e.g. boat fishing, shore fishing or underwater fishing), as the same recreational
fisher may engage in various different types of fishing activities. Fishing habits could
be taken into account to weight observations, but it is recommended to make sure that
this choice is motivated by: i) a good rationale for why recreationists with different
fishing habits might differ; and ii) evidence on the soundness of existing data on
recreational fisheries. Any bias in estimated proportions of recreationists will further
affect estimates made through weighting.

FIGURE 7
Stratification of the target population

target population strata sample

2. Data collection 15

2.4 ESTIMATING THE SAMPLE SIZE

The biggest advantage of probabilistic survey sampling lies in its capacity to provide
accurate representations of large populations from surveys of small groups of units; a
minimum number of units, however, is required to make inferences about the target
populations. The minimum number of observations necessary is usually defined on the
basis of the desired sampling error, the size of the target population and the variability
of the trait of interest: a higher number of units is needed to make inference about a
large and heterogeneous population than for a tiny and homogeneous one. Similarly, to
obtain highly accurate estimates, a greater number of units is needed than for coarser
estimates. Table 1 provides an example from Salant and Dillman (1994).

TABLE 1
Example of sample sizes needed by population size and level of sampling error
± 3% Sampling error ± 5% Sampling error ± 10% Sampling error
Population size 50/50 split 80/20 split 50/50 split 80/20 split 50/50 split 80/20 split

100 92 87 80 71 49 38
250 203 183 152 124 70 49
500 341 289 217 165 81 55
750 441 358 254 185 85 57
1 000 516 406 278 198 88 58
2 500 748 537 333 224 93 60
5 000 880 601 357 234 94 61
10 000 964 639 370 240 95 61
25 000 1 023 665 378 234 96 61
50 000 1 045 674 381 245 96 61
100 000 1 056 678 383 245 96 61
1 000 000 1 066 682 384 246 96 61
100 000 000 1 067 683 384 246 96 61
Source: Salant and Dillman, 1994.

Sample size for random sampling may be easily estimated in cases of populations
with lists of units. Here follows an example of the formula provided by Vaske (2008):
(Np ) × ( p )× (1 − p )
Ns=
[( Np− 1 )× (B /C )2 ]+[( p ) × (1 − p )]

where:
Ns = the sample size
Np = the size of the target population (e.g. the number of recreational fishers
reported on a list)
p = the prevalence of the target variable (e.g. the number of recreational fishers who
are exclusive sea fishers, who do not fish in freshwater)
B = the desired level of sampling error which can be accepted (e.g. 5 percent = 0.05)
C = the Z-statistic associated with the confidence interval (e.g. for a 95 percent
confidence interval, Z = 1.96)
For stratified random sampling, the number of observations within each stratum
can be obtained through proportionate stratification. The procedure requires the
following steps:
1. Compute the desired sample size (see formula above).
2. Calculate the proportion of each stratum in the target population.
3. Assign the number of observational units proportionally to each stratum.
16 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

This procedure is called proportionate stratification and takes the following formula:

𝑁𝑁�
𝑛𝑛� � � � ⋅ 𝑛𝑛
𝑁𝑁

where:
nh = the number of observations in each stratum of the sample
Nh = the number of observations in each stratum of the population
N = the total number of observations in the population
n = the total number of observations collected with sampling

2.5 SELECTING THE SAMPLE

Once the target population has been defined (see Section 2.1) and stratified (see
Section 2.3) and the sample size determined (see Section 2.4), the sample of recreational
fishers to be enrolled in subsequent data collection can be selected. In order to initiate
this process, each fisher must possess a unique ID, identifying him or her from all other
fishers. In the case of a licensing system, this unique ID could be the license number,
whereas in the case of screening surveys or mandatory fee-free online registration,
each member of the target population should be assigned a unique ID. Following
the methodology of random sampling, the basic condition required for the selection
of the sample is randomness. To avoid human error, this process should be carried
out by a computerized routine, ensuring that all members of the population share
an equal chance of appearing in the sample, thereby guaranteeing randomness. This
computerized routine can be performed simply in Microsoft Excel, following these
steps (Figure 8):
1. Enter the complete target population list frame in the Excel file, ensuring
that each fisher is identified by a unique ID.
2. Assign a random number to each ID by means of the RAND function
in Microsoft Excel by typing =RAND() and hitting enter. A randomly
generated number will appear in the cell.
3. To ensure the RAND function does not continue to change the value of
the randomly generated numbers, copy and paste all random numbers
generated using “paste special – value” to insert the value only in the
column with the random number.
4. Sort the list of IDs by their random number, from smallest to largest.
5. According to the chosen sample size (n), select the first n rows of the
list: they constitute the randomly selected sample units.
This simple and straightforward procedure guarantees the perfect randomness of the
sample (Pinello, Gee and Dimech, 2017).
Once the sample population has been selected, the fishers should be contacted
(by e-mail or telephone) in order to determine whether or not they are willing to
participate in the data collection. If they agree to participate in the data collection,
then they should be enrolled in the panel survey; fishers who decline to participate
shall be substituted by other fishers randomly selected from the database. All
attempts should be made to encourage participation and to avoid replacement
whenever possible, as replacement can result in a less representative sample. When
feasible, it is useful to collect demographic and fishing avidity data from those who
refuse to participate, as this information can facilitate adjusting statistical weights to
account for non-response error.
2. Data collection 17

FIGURE 8
Example of computerized routine to select random samples

STEP 1 STEP 2 STEP 3 STEP 4

A B C D A B C D A B C D A B C D
1 ID 1 ID Random 1 ID Random 1 ID Random
numbers numbers numbers
2 1
3 2 2 1 0.40795484121 2 1 0.4079548421 2 1 0.3849454491

4 3 3 2 0.12845466897 3 2 0.5409865742 3 2 0.756328737

5 4 4 3 0.65770634522 4 3 0.02439870 4 3 0.09634201

6 5 5 4 0.07759856331 5 4 0.396948965 5 4 0.483920188

7 6 6 5 0.46540075623 6 5 0.659354280 6 5 0.295824718

8 7 7 6 0.87541332138 7 6 0.196493582 7 6 0.981839539

9 8 8 7 0.97566432311 8 7 0.6573209382 8 7 0.739291029

10 9 9 8 0.73242430743 9 8 0.7692759202 9 8 0.392093754

11 10 10 9 0.54786965421 10 9 0.874027539 10 9 0.285386549

12 11 11 10 0.56359821065 11 10 0.657098365 11 10 0.387652915

13 12 12 11 0.37664809832 12 11 0.39672892 12 11 0.498286423

14 13 13 12 0.89467563297 13 12 0.496492174 13 12 0.73954201

15 14 14 13 0.48675094327 14 13 0.002975662 14 13 0.297163209

16 15 15 14 0.98006574344 15 14 0.856743020 15 14 0.875927542

17 16 16 15 0.04567895 16 15 0.376950281 16 15 0.2954310865

18 17 17 16 0.18657965643 17 16 0.948677433 17 16 0.021783974

19 18 18 17 0.3678954321 18 17 0.759207512 18 17 0.297493214

19 18 = Rand () 19 18 0.645839734 19 18 0.536839202

STEP 1: the total population includes 18 recreational fishers (each assigned an ID number and entered into the
Excel file) and the objective is to randomly select 50 percent of them (nine fishers). STEP 2: with the RAND function,
create 18 random numbers. STEP 3: copy and paste as values these numbers. STEP 4: sort the fisher ID by the random
numbers from smallest to largest and select the first nine fisher IDs (i.e. 8, 2, 12, 7, 1, 6, 16, 18 and 11). These nine
fisher IDs constitute the sample.

2.6 ADDITIONAL CONSIDERATIONS

Selecting a sample goes beyond sampling design and the random extraction of
statistical units. In practice, many other decisions are involved in the process,
governing coverage error, sampling error and non-response error (see Figure 2).
Notably:
• The sampling frame might not fully overlap with the range of statistical
units in the population. Some units, given certain sampling mechanisms,
may not be covered by the survey, thereby biasing the estimates. A famous
case involves the use of online surveys: not every person has access to the
internet, meaning not every person can be recruited in an online survey
and, therefore, online surveys are often biased compared to other survey
administration modes (Vaske, 2011). Considering that internet usage may
be limited in rural areas and developing countries and among the elderly,
estimates from online surveys risk strong biases for recreational fisheries
surveys in the Mediterranean and the Black Sea where these three groups
make up a significant portion of the target population.
• Sampling error: as discussed in Section 2.2, sampling might be biased.
For example, simple random sampling may fail to be balanced in terms of
relevant groups of units (strata), biasing subsequent inferences.
• Non-response error: certain mechanisms adopted for unit selection
might produce problems due to people not responding to the survey;
self-administered surveys, if overly time-consuming and cognitively
demanding, may be rejected by less motivated respondents, or by
respondents with a lower level of literacy. In turn, non-respondents might
differ from respondents in relation to the target variable that researchers
aim to estimate, thereby biasing final estimates. A simplified example
is a self-administered mail survey, delivered to a random sample of
fishers, which asks them many different questions about seasonal catch.
The questionnaire is well-designed and protects privacy, but it is too long
and difficult to understand. Therefore, those fishers with low literacy
18 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

levels (and – for the sake of this simplified example – correspondingly

lower income levels) do not respond. As a result, responses come from
those fishers with higher literacy rates (and correspondingly higher
income levels). Owing to this latter group’s higher income, they likely
use more expensive and more efficient fishing gear, resulting in an
overestimation of the average seasonal catch.
Defining a sample frame and an administration mode are two practical aspects
of survey implementation that affect the estimation of recreational fishers in the
Mediterranean and the Black Sea.
 19

3. Methodology

Once the sampling frame of recreational fishers has been identified – regardless of
the data source (i.e. license system, screening survey or mandatory fee-free online
registration) – there are a number of different methods for contacting recreationists and
collecting effort, catch and economic data.
Each method comes with its advantages and disadvantages in terms of species and
geographical coverage, measurement accuracy and scalability of results (Wynne-Jones
et al., 2014). Ideally, data collection procedures should minimize coverage, sampling,
and non-response bias. Moreover, data collection should refrain from asking sensitive
questions and should avoid making respondents feel uncomfortable about their
answers (Krumpal, 2013). If these two conditions are met, data collection can provide
catch statistics that are unbiased and sufficiently precise for use in stock assessments
and for informing fisheries management.
There are two broad types of approaches to data collection:
• off-site surveys; and
• on-site surveys.
Off-site surveys are characterized by researchers drawing observational units
without going into the field. This context implies that they are inevitably conducted
for target populations whose lists are known and available and that they collect mostly
self-reported measurements.
On-site surveys, on the other hand, involve sampling fishers by going into the field
and approaching and interviewing them.
As a general recommendation, both off-site and on-site surveys should aim to ask as
few questions as possible in order to minimize the cognitive burden for respondents.
Furthermore, sensitive questions should be avoided and all efforts should be made to
build a trust-based relationship with respondents, particularly in the case of economic
data collection (see Section 3.3.4). Available evidence indicates that sharing detailed
information about the scope of the questionnaire and providing feedback on the
scientific findings to the respondents is useful in promoting trust (Vaske, 2008).

3.1 OFF-SITE SURVEYS

Off-site surveys offer a means of measuring all forms of fishing activity across large
spatial areas to produce total harvest estimates. There are certain potential advantages
to such methods, particularly in terms of geographical coverage and their ability to
reach all the various types of recreational fishers, even those who are the hardest to
recruit in on-site surveys. Respondents can be asked about fishing over defined periods
(e.g. day by day or over an extended period), especially when enrolled in a panel-
type survey (Wynne-Jones et al., 2014). However, it is important to note that off-site
surveys always rely on self-reported information. Off-site surveys can take two forms:
• logbook surveys; and
• recall surveys.

3.1.1 Logbook surveys

Logbooks provide a very cost-effective means of collecting both fishing effort, catch
and economic data. A template of a logbook is included in Annex 5. The logbook could
be delivered to selected recreational fishers as a paper book/diary at the beginning of
the survey period. Alternatively, online logbooks or a dedicated application for mobile
phones could be developed. As a first step, delivering paper logbooks is suggested as
20 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

they ensure maximum coverage. Each page of the logbook should correspond to one
fishing trip. Should a fisher engage in multiple fishing modes (e.g. from a boat, from
the shore or underwater) within the same day, each fishing mode should be considered
a separate fishing trip and a separate logbook should be completed. Fishers should be
asked to complete the logbook with:
• general information (Annex 5.a), including:
– name and surname of the panel participant;
– whether the logbook information is reported for a single fisher (the panel
participant) or multiple fishers (in the case that the panel participant
pools his/her catch with other fishers during the fishing trip and it is not
possible to determine the panel participant’s individual catch). In the case
of multiple fishers, the number of fishers (gender disaggregated) and their
ages should be reported;
– location of the fishing ground, such as the GFCM GSA, the city or distance
from the coast: this can be reported through geographical coordinates
(if available through GPS or mobile phone data) and/or by describing the
location (e.g. by reporting the basin and distance from the nearest harbor);
– total fishing time: the date and time of the fishing trip’s start and the date
and time of the fishing trip’s end;
– fishing mode: whether fishing took place from a boat, from the shore or
underwater;
– information about the fishing effort: fishing gear used, time spent fishing
per gear (fishing time), number of units used for each fishing gear (e.g.
number of rods, hooks, etc.). In case “multiple fishers” was selected at the
top of the logbook, then the cumulative fishing effort for all fishers should
be reported;
– catch by gear code: in case “multiple fishers” was selected at the top of the
logbook, then the cumulative catch for all fishers should be reported;
• retained species information (Annex 5.b), including:
– biological data of the retained catch, including length, weight and sex (if
known);
• released species information (Annex 5.c), including:
– information on the released catch, including the length and post-release
status; and
• expenditures (Annex 5.d), including:
– the value of all expenditures made in relation to the fishing trip, including
any expenditures incurred prior to the fishing trip (e.g. the purchase of new
equipment) since the last logbook was completed.
Fishing effort will be estimated taking into account the total fishing time of the
trip (ending time minus starting time, including travel to/from the port in the case
of boat fishing). In the example shown in Figure 9.a, the total fishing time is eight
hours. Data on fishing effort must be reported for each gear/technique used during
the trip. The effective fishing (soak) time per gear should be differentiated from total
fishing time since the catch should be standardized using effective fishing time. In
this example, five hours were dedicated to fishing with hooks (three hooks in total),
and three hours for traps (two traps in total). With regard to hook fishing, it is
important to know how many hooks were used; if, for example, a total of three rods/
handlines were used and each rod/handline had a tackle with three hooks, then the
total number of hooks will be nine.
When more than one person participates in the fishing trip and the individual effort
and/or catch of each person cannot be determined (e.g. when several people are fishing
on the same boat, collectively using the same gear and the catch is pooled together),
then fishing effort should reflect the cumulative effort of all participants and the total
3. Methodology 21

cumulative catch should be reported. During the data analysis phase, the catch and
effort of the logbook owner can be estimated as the mean of the effort and the mean
of the catch of all fishers participating in the fishing trip. For this reason, the number
of fishers is requested.
The catch must be recorded by gear typology. A list of gear codes is recorded in
Annex 2 as well as in Annex 5.a and Annex 6.a. The gear code is needed to ensure that the
respondent is referring to the correct gear and to facilitate the work of the researcher in
identifying the gear without errors. In the first column, the gear code must be reported
(see Annexes 5.a, 5.b and 5.c), while in the column titled “Species,” a valid name of
the species should be written. The scientific name would be the ideal way to report a
catch, but recreational fishers do not usually know the scientific name of each species.
Therefore, it would be better to ask for the common name and, in case such a name is
ambiguous, then it would be advised to contact the fisher and ask for an explanation.
Following the example in Figure 9.b, the first species recorded is the common pandora
(Pagellus erythrinus), one specimen has been kept (total length = 25 cm, corresponding
to a weight of 0.3 kg) and one specimen has been released. In the template for released
catch (see Annex 5.c) it would be important to report whether the released fish was
alive, almost dead, dead or not known, when released back into the sea. For example, in
Figure 9.c, in the case of the Mediterranean horse mackerel (Trachurus mediterraneus)
under the “catch information” logbook template (see Annex 5.b), it is noted that three
specimens were caught, with total lengths indicated for each one, followed by their
three respective weights. In the case of the abundant catch of black gobies (Gobius
niger) reported in Figure 9.a, it is sufficient to write the total number of fish (40) and
the total weight (1.2 kg) in the “general information” logbook template (see Annex 5.a).
For cephalopods, the mantle length in cm must be recorded in the “catch information”
logbook template (see Annex 5.b), as in the case of the common cuttlefish recorded in
Figure 9. Crustaceans must be measured for carapace length in mm. For other taxa (i.e.
echinoderms), it would be sufficient to report number and total weight in the “general
information” logbook template (see Annex 5.a). For further details on measuring catch,
see Section “3.3.2 Catch”.
Some fishers might not update their logbooks on a regular basis, which could
ultimately bias the study. In this case, follow-up by the researcher would be necessary
to determine why the relevant fishers did not fill out their logbook every month.
Regular communication and follow-up with panel participants could help increase the
proportion of completed logbooks. Another source of bias is the so-called “prestige
bias”, which involves fishers exaggerating catch size or numbers and providing
deliberately false information to make a better impression on others. On the other hand,
certain political or cultural contexts may lead fishers to understate their catch to avoid
management repercussions or due to superstitions believing in bad luck derived from
sharing information about the size of the catch. Both forms of bias might be reduced
by emphasizing that data will be reported anonymously, that it will be combined with
other means of data collection (e.g. on-site surveys), that honesty is important for the
ethics of fishing and that exaggerating data might have negative consequences for the
management of fish stocks (Ayal et al., 2015).
It would be useful to train recreational fishers in filling out the logbooks by means
of training courses (e.g. online tutorials, seminars, etc.). Within such training courses, it
is important to emphasize that logbooks should be completed on a regular basis, rather
than just before they are to be collected by researchers, as the latter routine might
introduce recall bias and present negative consequences for fisheries management.
Logbooks should be collected regularly, for instance every month, and data should be
entered into a database for subsequent analysis.
22 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

FIGURE 9.a
Example of how to compile a logbook (general information)

Logbook and recall template – general information

Logbook x Recall Reference month and year ______________________
Mario Rossi
Name and surname of panel participant ________________________

Information reported for:

Multiple fishers (in case the panel participant’s catch is
Only the panel participant pooled with other fishers on the same trip) x
If multiple fishers:
2
_____ age 45 51
No. Fishers:
1
_____ age 35

Fishing location Total fishing time

Geographical subarea (GSA) 17

___________ Start End

City Venice
___________ Date 13 May 19 13 May 19
Hour 06:00 14:00
Distance from the coast (in nm) 3
___________

Fishing mode* Boat x Shore Underwater

Gear Gear code Fishing time per gear (in hours) Number of units used per gear
Hand implements MHI
Harpoons HAR
Diving MDV
Diving (speargun) MDS
Diving (hand) MDH
Cast nets FCN
Boat seines SV
Beach seines SB
Hooks and lines (not specified) LX 5 9
Handlines and hand-operated pole and lines LHP 5 3
Traps (not specified) FIX 3 2
Pots FPO
Gillnets and entangling nets (nei) GEN
Gillnets GNS
Trammel nets GTR
Longlines (not specified) LL
Lift nets (not specified) LN
Scoop nets MSP
Gear not known NK
Gear not specified MIS

Catches
No. Gear code Species No. Retained Weight (kg) Retained No. Released
1 LX LHP common pandora 1 0.3 1
2 LX LHP horse mackerel 3 1.1
3 LX LHP gilthead seabream 1 1 1
4 LX LHP black goby 40 1.2
5 FIX cuttlefish 2 1.2
6
7
8
9
10

Comments:

* Complete one logbook/recall template per fishing mode

** Provide a description of the fishing gear in the comments section
3. Methodology 23

FIGURE 9.b
Example of how to compile a logbook (retained species information)

Logbook and recall template – retained species information

Logbook x 13 May 19
Date ____________ Recall Reference month and year ______________

Gear Weight Fishing mode***

No. Species (retained) Length* Sex**
code (kg) Boat Shore Under water
1 LX LHP common pandora 25 0.3 nd x
2 LX LHP horse mackerel 25 0.3 nd x
3 LX LHP horse mackerel 30 0.4 nd x
4 LX LHP horse mackerel 30 0.4 nd x
5 LX LHP gilthead seabream 40 1 nd x
6 FIX cuttlefish 14 0.5 male x
7 FIX cuttlefish 16 0.7 female x
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
* Total length for fish (cm), mantle length for cephalopods (cm), carapace length for crustaceans (mm)
** If known (M: male; F: female; ND: not determined)
*** Select only one fishing mode (boat, shore, underwater) per row
24 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

FIGURE 9.c
Example of how to compile a logbook (released species information)

Logbook and recall template – released species information

Logbook x 13 May 19
Date ____________ Recall Reference month and year ______________

Post-released status** Fishing mode***

Gear
No. Species (released) Length* Almost Not Under
code Alive Dead Boat Shore
dead known water
1 LX LHP common pandora 12 x x
2 LX LHP gilthead seabream 15 x x
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
* Total length for fish (cm), mantle length for cephalopods (cm), carapace length for crustaceans (mm)
** Mark the corresponding cell
*** Select only one fishing mode (boat, shore, underwater) per row
3. Methodology 25

Logbook surveys are also an effective way to measure economic expenditures.

Once reliable economic baseline data have been established, less frequent economic
data collection could be foreseen (e.g. every two to five years, rather than annually),
in order to simplify data collection and to avoid overburdening respondents. Selected
recreational fishers should be asked to register the money they spent to carry out
their fishing activity during each fishing trip, including for fishing equipment (e.g.
rods, reels, hooks, lines, swivels, spearguns, underwater accessories, traps, etc.), bait
(e.g. natural or artificial bait), travel and accommodation (e.g. train, plane, car, hotels,
etc.), boat expenses (e.g. charter, rental, boat ownership expenses such as fuel costs,
mooring fees and taxes, boat maintenance, etc.), electronics (GPS, echo sounder,
radar, etc.), license fees and other costs. In the case of underwater fishing, boat
expenditures should also be included if the fishing is performed using a boat. A more
detailed description of the information to be collected is recorded in Section 3.3.3,
and a template for collecting this information through the logbook can be found in
Annex 5.d. The monetary value to be inserted in each cell should be indicated in
the local currency. To facilitate regional comparison, the survey coordinator should
convert those values into a common currency, such as EUR or USD, by applying
current conversion rates.

3.1.2 Recall surveys

An alternative to logbook surveys is the recall survey, which relies on contacting,
via e-mail and/or telephone, selected recreational fishers and asking them to recall
information about their catch, effort and expenditures over a specific timeframe.
Extended timeframes (e.g. one-time surveys with a six- or 12-month recall period)
can significantly overestimate total recreational fishing effort. Typically, an average
catch per trip is recalled and then multiplied by the assumed number of trips. This can
potentially lead to a severe overestimation of the catch, as there is a general tendency
to exaggerate participation rates in recreational events (Tarrant and Manfredo, 1993;
Connelly and Brown, 1995; Vaske, Huan and Beaman, 2003). This is not always
the case, however, as noted in Connelly and Brown (2011); therefore, anglers and
recreational fishers should, in general, be treated as a heterogeneous group (Arlinghaus,
Bork and Fladung, 2008; Johnston, Arlinghaus and Dieckmann, 2010). Respondents
generally prefer to recall the catch in numbers, in which case converting those numbers
into weight can present issues. A specific problem with recall surveys is that the longer
the timeframe over which respondents are supposed to recall, the more the results tend
to be biased (Tarrant and Manfredo, 1993), so a short recall period would be preferred
to minimize possible recall errors. A one- to two- month recall period is suggested as
it is feasible yet not too long. At a more advanced stage in implementing recreational
fisheries monitoring programmes, it could be worthwhile to contact more avid fishers
more often during the peak season, although that may not be necessary in the early
stages of trialing these methods in a country.
The same information is required for the recall survey and for the logbook, meaning
that interviewers can use the same template as the logbook survey for catch and effort
data (see Annexes 5.a, 5.b, 5.c), filling in one template per fishing trip. As is the case
with the logbook, when a fisher engages in more than one fishing mode (e.g. fishing by
boat, shore or underwater) in the same day, each fishing mode should be considered
a separate fishing trip and therefore a separate logbook should be completed. For
economic expenditures, the information to be collected is identical between the
logbook and the recall survey, however, the reference period differs. The logbook (see
Annex 5.d) should include all expenditures made in relation to the specific fishing trip,
including any expenditures since the last fishing trip. On the other hand, the recall
survey (see Annex 5.e) should collect information in relation to all expenditures made
during the recall survey’s reference period (e.g. the previous one to two months). In all
26 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

cases, it is helpful to supply the fisher with a copy of the logbook template in advance,
so that he or she may keep notes and to facilitate jogging his or her memory at the time
of the recall survey interview.
Recall surveys can also be used as a complement to logbook surveys. Selected
fishers who are involved in the logbook programme should be contacted on a monthly
basis by telephone in order to verify the information reported in the logbook over the
previous month. Logbook information requiring verification could include the fishing
areas (e.g. wrong or questionable geographical coordinates, doubtful locations, etc.);
the number of gear used (e.g. verifying that the number of hooks is reported rather than
the number of rods, etc.); the common name of the target species (e.g. matching the
correct scientific names to the species); eventual peculiar numbers or weights of catch
(e.g. very high number of fishes, wrong correlation between length and weight, etc.);
and other eventual anomalies observed in the logbook.

3.2 ON-SITE SURVEYS

On-site surveys entail sampling fishers by going directly into the field and interviewing
them. On-site methods potentially represent a more accurate and direct approach
as fishery-independent staff members follow randomized probabilistic designs to
collect the data, usually soon after any fishing effort has taken place. Detection of, and
correction for, any bias are also potentially more feasible given the direct and verifiable
nature of the data collected. Unfortunately, on-site methods tend to be comparatively
expensive and logistically onerous, thus limiting the scale at which they can be applied
(Hartill, Watson and Bian, 2011). This type of survey could therefore be useful as
a means of validating and integrating the data acquired through off-site surveys
(e.g. logbook, recall), providing additional data on catch size and species composition.
In this way, an off-site logbook or recall-based survey method could function as
the primary means of estimating mean catch rates and effort, with on-site sampling
conducted by trained interviewers only used to validate the self-reported off-site data.
On-site surveys can therefore contribute to the detection of discrepancies between
self-reported data and data measured in the field.
In some countries, such as those with limited coastlines or a limited number of
access sites, it may be possible to conduct on-site surveys as the primary means of
collecting fishing effort and length data directly from fishers, in view of estimating
catch per unit of effort (CPUE) (see Section 3.3.1). In other countries this may not be
a feasible or cost-effective option.
Whether using an on-site survey as the primary means of data collection or simply
to validate off-site surveys, the main purpose of an on-site survey is to collect data
on as much recreational catch as possible, for as many species as possible. Engaging
recreational fisheries stakeholders through federations and associations is one way to
reach a high number of fishers (see Section 5). Interviews can be carried out at harbors,
beaches, ramp sites, slip ways, etc. The locations can vary, so it is important to include
all specific locations with fishing participants in the sample frame. The catch data to be
collected should include the species of fish, the number of fish caught of that species,
the number of fish kept and the number of fish released. In addition, interviewers
should attempt to obtain length and weight measurements from a random sample of
the kept fish that the angler is willing to make available.
Biases may arise within the on-site survey when fishers are selected for sampling
based on accessibility or convenience (e.g. by sampling only vessels that arrive in port
within certain hours). This selection would not constitute a random sample of the
population because the probability of selection would be unknown, thereby invalidating
the interpretation of the data (Grafton et al., 2006). The probabilities associated with
sampling at different times of day should be controlled consulting expert knowledge
of fishing patterns in different areas and seasons. See Section 3.2.1 for information on
other on-site data collection methods, beyond this traditional approach.
3. Methodology 27

How the interviewer introduces himself or herself to the fisher is one of the most
important considerations and can frequently determine the success of the interview.
It is important to establish a relationship of trust with the interviewee in order to
promote honest responses. It is therefore recommended to use the following approach
when introducing oneself to a potential interviewee on-site:

“Hello, my name is ____ and I am doing a recreational fisheries scientific research

survey for _____ (institution) on behalf of _______ (e.g. Ministry of Fisheries). Can
I ask you a few questions about your fishing today?”

If the fishers wish to know the objective of the study, it should be clearly explained
that the main aim of the survey is to collect information on local recreational fisheries
in order to foster their sustainable management and that the anonymity of the
participant is ensured.
The information listed below should be annotated during in the interview.

General information (Annex 6.a)

• Date of interview
• Whether information is being reported for a single fisher or for a group
of fishers (in the case that gear/catch are pooled and it is not possible to
determine one fisher’s individual catch). In this latter case, the number of
fishers (gender disaggregated) and their age should be reported.
• Fishing location: the location of the fishing ground should be requested.
Questions can be formulated in the following manner: “Roughly, where did
you fish today? Could you please estimate the distance from the coast?” It can
indeed be useful to bring a map and ask fishers to indicate directly on the map
where the fishing ground is located. In the event that fishers are particularly
collaborative, they could be asked to provide the exact fishing location by
geographic coordinates (latitude and longitude).
• Total fishing time: the time spent during the whole fishing session should
be recorded. For example, in the case of a boat trip, this also includes the
navigation time. The time should be clearly written in order to understand
whether it refers to before (a.m.) or after (p.m.) noon. The date should be
reported for both starting time and ending time of the boat trip to avoid errors
for fishing sessions taking place over multiple calendar days.
• Fishing time and number of gear: for each gear used, it is necessary to ask the
fishing time (how long the gear was in the water) and the number of gear (e.g.
number of rods and total number of hooks).
• Number and weight of retained species, as well as the number of released
species, by gear. Each species must be recorded using the local name or
the scientific name (if possible). In case there are doubts about the correct
identification, it is advisable to take a picture, using the timestamp on the
photo to associate the pictures with the interviews. For each species, register
the number of specimens and their total weight, as well as the number of
released individuals.
• Fishing trips performed during the previous year – in order to roughly
estimate the avidity of the fisher, the fisher should be asked to guess how
many fishing trips they performed over the previous year. This can sometimes
be a difficult question for a fisher to answer and it may be necessary to prompt
the fisher with potential responses (e.g. “was it five, 20 or 50 times?”). This
question should be asked to all fishers of the party and should refer to the
fishing mode (boat, shore or underwater fishing).
28 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

• Willingness to participate in a panel survey: as it is important to collect

contacts for subsequent diary/logbook or recall surveys, interviewed fishers
should be asked whether they are willing to be contacted in the future. If
their answer is in the affirmative, then contact information, including name
and mobile phone number (preferable to a landline phone number), should be
collected. This information should be requested at the end of the interview,
once fish have been measured and a rapport established with the fisher.
• Comments: any comments the interviewer may have regarding the interview
should be noted here. These annotations can help to understand eventual
oddities that may emerge during the survey and could include, for example,
the bait used or whether each fisher is listed in the national list of household
telephone numbers.

Retained species information (Annex 6.b)

• Total length, weight (and sex if possible) of retained species by gear: to
facilitate completing this task, it is useful to ask: “Can I please measure your
fish?” If the fisher agrees, then every retained species must be measured,
rounding down to the closest half cm for total length. Although useful, weight
and sex are not mandatory, as weight can be estimated subsequently by means
of the length/weight relationship, and it is usually required to open the belly
of the fish in order to determine the sex.

Released species information (Annex 6.c)

• The length of each released specimen should be requested, as well as
information on the post-release status (e.g. alive, almost dead, dead, unknown).

3.2.1 Other on-site methods

In countries where more extensive on-site data collection is considered feasible (e.g. in
countries with limited coastlines), two alternative survey methods, in addition to the
traditional on-site data collection approach described above, could be considered to
estimate catch and effort: the bus route method and aerial-access surveys.

Bus route method

Robson and Jones (1989) developed a procedure for collecting recreational fisheries
catch and effort, which is analogous to a “bus route” and allows for a limited number
of interviewers to sample a high number of access sites. Instead of visiting just one or
two access sites a day (the traditional approach), each interviewer makes a complete
circuit of all access sites over the course of each sampling day (Jones et al., 1990). The
agents have a precise schedule to follow each day and they arrive and depart from
each site on a predetermined timetable. Because the starting point along the circuit is
chosen randomly each day, each site is visited randomly throughout the day over the
survey period. This method is particularly appropriate when there are many access
sites to be sampled. For example, if the study area consists of 12 access sites, it would
be unreasonable to spend a full day at only one of so many access sites as each site
would then be sampled infrequently over the whole survey period. With the bus route
method, one interviewer (or one crew of interviewers) covers all 12 sites within a single
day. Following a traditional approach, the same number of interviewers would visit
only one to two sites per day.

Aerial-access surveys
The use of observers in aircraft flying at low altitudes (150 to 300 m, depending on
the minimum-permissible altitude under civil aviation regulations) offers an additional
means of counting recreational fisheries vessels or fishers from the shore. There are two
3. Methodology 29

forms of aerial-access design: the random-count design (described by Pollock, Jones

and Brown, 1994 and used by English, Shardlow and Webb, 1986; English, Searing and
Nagtegaal, 2002; Coutin, Conron and MacDonald, 1995; and Soupir et al., 2006) and
the less commonly reported maximum-count design (Parker, 1956; Dauk, 2000; Dauk
and Schwarz, 2001; Lockwood, Peck and Oelfke, 2001).
The random-count design divides the day into two or more time bins, and flights
are scheduled to take place at a random time within one or more diurnal strata on
each survey day. The estimated number of hours fished in a given time bin is defined
as the product of the number of hours occurring within that time bin and the aerial
count. This estimate is then multiplied by a catch rate estimate for the same period
to provide a catch estimate for this time interval (Hartill, Watson and Bian, 2011).
Flights can be scheduled to take place during all time bins within a day, and the
estimates of catch and effort obtained for each time bin can then be summed to
provide total estimates for that day. Estimates from a random subsample of available
days can then be averaged and generalized to provide catch end effort estimates for
a larger temporal stratum, such as an entire summer season. Alternatively, time bins
can be randomly sampled at a lower intensity across all survey days over a larger
temporal stratum. Care must be taken, however, to ensure that at least one time bin
is selected from each survey day and that sufficient replicates are sampled for each
time bin across all surveyed days. Regardless of which random-count design is used,
the high number of flights required to adequately estimate the total level of effort
occurring on a sample day and across all sample days is potentially prohibitive, yet
unavoidable, as these flights offer the sole means of estimating levels of effort when
this design is used (Hartill, Watson and Bian, 2011).
Maximum-count aerial-access designs, such as that described in Hartill, Watson
and Bian (2011), represent a more cost-effective option as only a single flight is
required per survey day. A count of fishing vessels made during this flight is used in
conjunction with creel survey data to describe the distribution of effort throughout
the day. The substantial reduction in flight costs is offset to some extent by the
need to station creel survey clerks at selected access points throughout the day.
Nevertheless, results from this study (Hartill, Watson and Bian, 2011) suggest that
catch rates vary throughout the day, and the best means of correctly accounting for
these changes is to interview fishers throughout the day. The same study showed
that it would be preferable to combine aerial count and fisher interview data together
at the level of the primary sampling unit, i.e. the day. Estimates of total effort and
catch were calculated for each randomly selected survey day and then averaged
over their respective temporal strata. Hartill, Watson and Bian (2011) observed that
the advantages of linking data from the aerial survey and fisher interviews on each
survey day to estimate levels of effort are twofold: i) fewer flights are required to
assess levels of effort, which can significantly reduce aircraft operating costs; and ii)
a defined relationship between these two data sources can be used to estimate levels
of effort on those days when flights are cancelled, which is a common problem with
aerial-access surveys.
Both the above-mentioned forms of aerial-access surveys are rather high-cost
data collection methods. With rapid advances and decreasing prices, however,
in the field of remotely piloted aircraft systems (RPAS) – colloquially known as
drones – researchers have access to a potentially innovative and cost-effective tool
for implementing this kind of survey. In addition to improved cost-efficiency over
existing techniques and the potential for highly replicable flight routes and for
accessing remote or inaccessible locations, RPAS also boast the added advantage
of being able to produce high-resolution mapping and capture footage beyond the
visible spectrum, as well as providing non-invasive survey techniques for marine
fauna. However, this technology is hindered by several important limitations,
30 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

including range, logistical considerations when operating over water, regulatory

requirements and battery life (Desfosses et al., 2019). Furthermore, it must be noted
that only limited studies have evaluated the suitability of RPAS as a recreational
fisheries data collection tool to date (Desfosses et al., 2019). While these tools offer
a number of potential benefits, it is important that they be adequately evaluated in
order to provide researchers with a more complete understanding of the potential
biases they may introduce (Beckmann et al., 2019) and their eventual suitability for
the sustainable management of marine living resources (Desfosses et al., 2019).

3.3 TYPE OF INFORMATION TO BE COLLECTED

Independent of the chosen survey method (logbook, recall or on-site), in order to
define the relationship between the sample and the statistical universe, it is necessary
to collect basic personal data, such as gender, age and residence. Place of residence is
required to spatially allocate the fisher within the sampled population. For off-site
surveys (logbook and recall) it is also recommended to collect names and the mobile
phone numbers. Other personal information is not relevant for the specific aims of
this study (e.g. profession, education), unless the study has specific socio-economic
objectives.

3.3.1 Fishing effort

Fishing effort is a measure of the fishing activity deployed by a certain fishing segment
and can be useful to calculate CPUE, which is needed to analyse changes in catch
quantities. This information is crucial for developing multiannual management plans.
Fishing effort can be calculated through a combination of inputs related to capacity,
gear and time.
In particular, it is useful to collect the following data:
• Number of fishing trips: the number of fishing trips conducted during the
interview period. A fishing trip is defined as a single fishing session, either
performed from the shore, from a boat or underwater (i.e. starting from the
shore or from a boat).
• Total fishing time (hours): the total duration (in hours) of a fishing trip
(including navigation in the case of boat fishing).
• Fishing time (hours): the number of hours using a specific gear (e.g. for set
nets, longlines and traps, the time from setting to pulling in; for hooks and
spearguns, the fishing time, etc.).
• Number of gear used: the number of nets (e.g. scoop net, cast net, beach seine,
etc.). This also refers to the number of panels for gillnets (or length of total set
nets used), the number of hooks used with rods or handlines and the number
of traps.
Guidance on how to measure fishing effort by fishing gear is provided in Annex 7.

3.3.2 Catch data

The objective of collecting catch data is to monitor and investigate the population
dynamics of the most important species in the area of study. Knowledge of the biomass
(by species) removed from the ecosystem by fishing operations is fundamental to
monitoring the status of stocks, as well as the impact of fishing on fish populations,
gear selectivity and catch at age.
In particular, it is useful to collect the following data:
• Species caught: identify the valid common name in order to define the
scientific name of each species caught.
• Number of specimens kept: the number of specimens caught and retained by
species (including all taxa, such as molluscs, crustaceans, echinoderms, etc.).
3. Methodology 31

• Number of specimens released and their post-release status: the number of

specimens caught and released by species (including all taxa, such as molluscs,
crustaceans, echinoderms, etc.).
• Status of specimens after release: i) “alive” – strong body movements and none
or only minor injuries; ii) “almost dead” – weak body movements and major
injuries; iii) “dead”; and iv) “not known” – when the status was not observed.
• Length (cm): length measurements are easy to make but require a well-defined
and standardized notation to allow for comparison of results. The length
measurements to be made depend on the group of species under study. The
lengths of fish and cephalopods should be generally measured, whenever
possible, with graduated fish measuring instruments, called ichthyometers,
while calipers are used for crustaceans (see below).
• Weight (kg): the weight of each single individual. If it is not possible to collect
this information, it is possible to convert length into weight by consulting the
length-weight relationship.
• Sex: determining the sex of caught individuals ranges from easy to extremely
difficult. For most fish, it would be necessary to open its belly and check the
gonads, and this operation should be authorized by the recreational fisher.
Macroscopic observations can distinguish four sex categories: male (M);
female (F); undetermined (U) – when, after dissection, it was not possible to
determine the species’ sex with the naked eye; and not determined (ND) – for
individuals that have not been examined. For some fish taxa (e.g. some gobies,
elasmobranchs, etc.), it is possible to determine the sex by observing some
external morphological features (e.g. fins, claspers, etc.).
Data on catch can be combined with effort data to estimate the CPUE, which is a
relative measure of fish stock abundance. Catch per unit effort can be used to estimate
absolute abundance and could be an indicator of fishing efficiency (GFCM, 2018). In
its basic form, the CPUE could be expressed as the captured biomass per each unit of
effort applied to a species/stock (e.g. total catch of a species divided by the total fishing:
kg/number of fish per longline hook days, or numbers retained or caught per trip).
Declining trends of this estimator could indicate overexploitation, while steady values
could indicate sustainable fishing.
Further consideration must also be made for the role of catch-and-release – when
fish are unhooked or set free from a trap or net and returned to the water alive – as
a considerable portion of fish caught by recreational fisheries can be released (Ferter
et al., 2013). The rates of released specimens, including species- and fishery-specific
catch-and-release mortality rates, are unknown for most recreational hook fisheries,
and therefore there is a need to estimate these mortality rates for use in stock
assessments. A mixture of desk-based study and experimental work is required to
compile data on the mortality of hook-and-line-caught fish and to bolster the evidence
base in order to account for survival. Such studies should consist of reviewing existing
literature, assessing the potential for extrapolation between species and fisheries, setting
up generic mortality profiles, and conducting species-specific mortality studies to fill
existing data gaps (ICES, 2014). This information is lacking for most target species in
the Mediterranean and the Black Sea and, until such information becomes available, a
precautionary approach could be adopted, assuming a survival rate of zero for those
released species with no survival estimate.
32 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

How to measure fish, crustaceans and cephalopods

Bony fish and elasmobranchs: for bony fish, sharks, skates and rays, the length should
be considered as the total length. The fish is measured, rounding down to the closest
half cm, from the tip of the snout to the end of the caudal fin (Plate 1).
The length classes should be reported in cm (as a whole number, or half cm, e.g. 0.5,
1.0, 1.5 etc.).
Crustaceans: for crustaceans (lobsters, crawfish, shrimps, prawns, stomatopods), the
standard measurement is the minimum carapace length. The length classes should be
reported in mm (as a whole number, e.g. 1, 2, 3, 4 etc.). The crustacean is measured,
rounding down to the nearest mm, from the back border of the eye orbit (inside of
the eye socket) to the posterior margin of the carapace (Plate 2). All measurements are
taken with calipers.

PLATE 1
Measurement of total length in bony fish

Common dentex (Dentex dentex) © A. Lucchetti

PLATE 2
Measurement of total length and carapace length in crustacean’s decapoda

European lobster (Homarus gammarus) © A. Lucchetti

3. Methodology 33

PLATE 3
Measurement of dorsal mantle length in cephalopods

Broadtail shortfin squid (Illex coindetii) © A. Lucchetti

Cephalopods: for cephalopods, the length is defined as the dorsal mantle length. The
length classes should be reported in cm. The cephalopod is measured rounding down
to the nearest half cm. The size should be reported in cm (as a whole number, or half
cm, e.g. 0.5, 1.0, 1.5 etc.). For decapoda, measurement is made along the dorsal midline
from the mantle margin to the posterior tip of the body, excluding long tails (Plate 3).

3.3.3 Economic data

Although recreational fisheries do not generate a direct commercial output, it has
been shown that these fisheries generate significant economic value, through, for
example, their contribution to the tourism sector (Gaudin and De Young, 2007).
For this reason, the assessment of the economic impact of this sector is essential
and economic data are an important component of any recreational fisheries data
collection programme. Considering that recreational fisheries are, by definition,
non-commercial, meaning that it is prohibited to sell or trade the catch obtained,
non-market valuation techniques must therefore be applied. While both revealed and
stated preference methods can be used to assess the value of recreational fisheries,
revealed preference methods, such as the travel cost method and the hedonic pricing
method, are most commonly used. These methods assess expenditures made as a
proxy for economic value. Data on the costs recreational fishers incur help to explain
their behavior and are useful in understanding the wider economic impact of this
fishing activity. A simple method for calculating recreational fisheries expenditures is
through a logbook or recall survey, by asking recreational fishers to report or recall
the expenses incurred to carry out their leisure activity over the reference period.
In the case of a logbook, fishers should include all expenditures in relation to the
current fishing trip, as well as any expenditures made since their last fishing trip (e.g.
purchase of a new rod, etc.). In the case of a recall survey, all expenditures within the
recall period should be reported. Templates are provided in Annex 5.d, for logbooks,
and Annex 5.e, for recall surveys. A description of the variables to be collected for
which expenditures should be calculated is listed below:
34 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

• Equipment: the costs incurred for the purchase of equipment. For shore
fishing and boat fishing, these may include the purchase of rods, hooks, reels,
cast nets, etc., whereas for underwater fishing, these may include the purchase
of a speargun, fins, mask, wetsuit, etc.
• Bait: the expenditures for both artificial baits (jigs, lures, spinner baits, etc.)
and natural baits (worms, sardines, anchovies, shrimps, etc.).
• Travel and accommodation: the travel costs to/from the fishing site. These
may include the costs of staying in a hotel (for the days spent fishing),
roundtrip expenses to/from the fishing site, such as train or airplane tickets
or expenses for travel by car (fuel costs, highway and parking fees, rental car
expenses, etc.).
• Fishing license fees: it should be indicated whether the license is an annual,
semi-annual, quarterly, monthly, weekly or daily license.
• Boat expenses: these may include, the purchase of a boat, boat rental or
charter fishing fees, fuel costs (including two strokes lubrication oil), boat
taxes (mooring, ramp, etc.), boat maintenance costs (engine maintenance,
antifouling, etc.), as well as electronics (echo sounder, GPS, radar, etc.).
 35

4. Data analysis

4.1 DATA QUALITY CHECK

Once data have been collected, they should be analysed and generalized to describe
the total population. Before these actions can be completed, however, a critical step is
to carry out a data quality check and the necessary data treatments. The accuracy of a
survey estimate refers to the proximity of the estimate to the true population value and
the difference between the two is referred to as the error of the survey estimate. This
latter value is a fundamental component in the following steps for making estimates.
Unfortunately, in practice, a true measure of sampling error can never be obtained,
only an estimate (Pinello, Gee and Dimech, 2017).
Sampling errors refer to those errors encountered in the estimate of a particular
parameter of the universe resulting from the fact that not all of the population, but only
a subset (the sample), is the object of observation.
Non-sampling errors can be defined simply as all other errors in the estimate arising
during the course of any survey activities other than sampling (e.g. the way you run
the survey). Unlike sampling errors, they can be present in both sample surveys and
censuses and are extremely difficult, if not impossible, to measure mathematically.
With this in mind, both survey designers and data quality evaluators must ensure
that non-sampling error is avoided to as great an extent as possible, or at least either
randomly distributed in order to eliminate its effect on the calculation of population
estimates or brought under statistical control.
The most common non-sampling errors result from poor coverage and selection
bias, low response rates, non-responses, interviewer errors and data entry errors.
Non-sampling errors are systematic errors, which tend to accumulate over the course
of the entire survey and these types of errors often lead to a bias in final results. While
sampling errors decline with an increase in sample size (disappearing completely for
censuses), the same is generally not true for non-sampling errors.
It is worth noting that, even for well-designed and well-implemented surveys,
non-response represents a serious threat to the validity of estimates. It is fundamental
to ensure that non-respondents do not belong to a specific segment of the target
population, thereby limiting the validity of the inferences. This point is of utmost
importance and non-responses must be investigated to ensure that they have the
same characteristics as respondents. The likely reason for each non-response should
be recorded for each non-respondent, so that appropriate weighting and calibration
methods are applied to correct for non-response.
Prior to producing estimates for end-users, a certain amount of data checking and
monitoring must be performed to confirm the completeness and quality of primary
data (FAO, 2002). Such control functions involve:
• Monitoring: providing summary lists and reports offering quick indications
as to the availability of samples on boat activities and catch in each estimation
context.
• Data range check: providing lists that record “extreme” values (the range of
values) for catch, sample effort and prices. Values that appear too high or too
low should be verified.
• Sample size check: providing lists showing expected sample size and accuracy
level for boat activities and landings.
36 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

To guarantee quality assurance of recreational catch estimates from national

surveys and to document bias in data collection, the Working Group on Recreational
Fisheries Surveys of the International Council for the Exploration of the Sea (ICES)
has developed a quality assurance toolkit for evaluation (ICES, 2013). The aim of
this evaluation is to provide statements of quality of recreational data for end-users,
including stock assessment scientists, and to identify potential improvements to
survey design. The quality assurance toolkit consists of three modules – sampling
designs, implementation and data analysis – with the objective of minimizing bias and
supporting an accurate estimate of precision, in order to make the most efficient use of
sampling resources.

4.2 RESPONSE AND COMPLETION RATES

Among the most important rates is the so-called “response rate”. When data are
collected through a nationwide screening survey, the first step of data analysis is to
identify the fraction of recreational fishers out of the total population who responded
to the survey. The percentage of this active fraction is defined as the response rate
(Arlinghaus, Tillner and Bork, 2014; Hyder et al., 2017b). For self-administered surveys,
like mail surveys, a response rate can be calculated by dividing the number of fishers
who took part in the survey by the total number of fishers contacted. The rate can then
be converted to a percentage, by multiplying it by 100. For example, when 200 persons
answer a telephone survey out of a sample of 1 000 individuals, the response rate is
0.2 (i.e. 20 percent). The response rate can likely be increased by sending out multiple
reminders, especially if units are surveyed through self-administered questionnaires.
In this case, it is important to record the rates for each wave of reminders, to obtain a
more nuanced overview of the survey effectiveness. For example, a survey could have
a 0.8 response rate in the first wave, a 0.6 rate in the second wave (following the first
reminder), a 0.3 rate in the third wave (following the second reminder), and so on.
Response rates generally decrease over time, wave after wave, and reminders might
draw on a considerable proportion of the budget. It is highly recommended to account
for reminders when planning the financial resources for a survey.
When the response rate is below one, a certain amount of non-response has
occurred. Non-response can be easily imagined, in self-administered questionnaires,
as fishers who received a questionnaire and never sent it back. Non-response can
represent a severe bias affecting estimates. Accounting for non-response is complex and
can be achieved through one of four available approaches (Fox, Negrete-Yankelevich
and Sosa, eds., 2015):
• Resampling: it addresses non-response by replacing non-respondents with a
corresponding number of randomly re-sampled units. In stratified random
sampling, replacements are taken from the same strata as the missing
observations.
• Data imputation: it is based on model fitting and “fills” missing observations
with predicted data from a model, but only when observations are missing
completely at random, which is rarely the case in practice.
• Calibration: it incorporates information from auxiliary variables associated
with non-response into estimators.
• Weighting: it assigns varying importance to collected observations on the basis
of the proportion of non-respondents in the sample. Weighting is particularly
common in survey studies (Vaske, 2008) and merits a short explanation. For
example, a self-administered mail survey of 1 000 households, asking about
the presence of recreational fishers in the household and collecting data
about seasonal catch, could have a response rate of only 30 percent and find
that 80 percent of households contain at least one recreational fisher. In this
4. Data analysis 37

case, a non-response check should be carried out by knocking at the door of

90 percent of non-respondents: in this non-response check, only 40 percent
of households host recreationists. This finding suggests that the likelihood
of answering the questionnaire was linked to the presence of recreational
fishers in the household, as they were interested in the survey and motivated
to respond. For example, a previous census revealed that only 50 percent of
households contain recreational fishers. So, the sample probably includes too
many households containing recreationists. Observations might therefore be
weighted on the basis of pre-existing information on the statistical population.
Weights can be expressed as:
weight = population percentage ⁄ sample percentage
In this case, the catch from households containing fishers would be weighted by
a factor of 0.63 (0.5 / 0.8 = 0.63), while the catch from households which do not
contain fishers would be weighted by a factor of 2.5 (0.5 / 0.2 = 2.5). The potential
of weighting the data to adjust for non-response makes clear the importance
of non-response checks, to appreciate differences between respondents and
non-respondents. It is important that strata that are represented and under-
represented are identified carefully, just as when weighting observations:
weighting for strata that are not relevant will further bias findings. Moreover,
researchers are encouraged to pay attention to the quality and the sample size
of non-response checks.
Another fundamental rate is the completion rate. In most quantitative surveys,
respondents answer structured questionnaires, each containing a fixed number of
questions. The completion rate represents the proportion of respondents who answered
each question and can also be calculated as a geometric mean for all the questions in the
survey. Low completion rates usually indicate that a questionnaire is too cognitively
demanding for respondents, that perceived privacy protection is low or that the reporting
burden is too high on the fishers. Piloting might provide researchers with valuable
insights, which can contribute to increasing completion rates. The completion rate should
be calculated for each question as the fraction of the total number of questionnaires
administered in which the question was answered. In addition, multiple completion
rates can be averaged by calculating their geometric mean, which summarizes the overall
extent to which questionnaires were completed.
For example, a questionnaire could be designed to measure three common forms of
non-compliance affecting recreational fisheries, each of a different sensitivity level and
with different potential sanctions: throwing away leftover fishing lines into the sea (low
sanctions), catching undersized fish (medium sanctions) and fishing within a commercial
harbour, which is often forbidden for safety reasons but is also quite common (high
sanctions). In the case of this example, from over 1 000 questionnaires collected,
not surprisingly, it is discovered that 896 respondents answered the first question,
451 answered the second question and only 86 respondents answered the third question.
Completion rates for the three questions are, respectively, 0.89, 0.45 and 0.09.

In this case, the average completion rate is 0.32. Of course, completion rates are
usually equal to 1.0 when surveys do not include self-reported information but
instead simply measure certain traits of the observational units. In a field survey where
technicians count fishing boats, there is no such thing as a response rate. Response rate
is a common issue in self-administered surveys, like mail, online or telephone surveys.
38 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

4.3 MEASURING CENTRAL TENDENCY AND DATA DISPERSION WITHIN A

SAMPLE
Once data have been collected, it is fundamental to characterize the sample in terms of
its centrality measures and in terms of its variability.
Centrality measures provide values around which observations are organized. The
most well-known measure is the arithmetic � mean, defined as the sum of all
1
measurements divided by the number of observations in the data set:
𝑥𝑥� � ⋅ � 𝑥𝑥�
𝑛𝑛 �
1 ���
𝑥𝑥� � ⋅ � 𝑥𝑥�
𝑛𝑛
���

The arithmetic mean can also be calculated for a quantitative character x, divided
into K classes, as:
�
1
𝑥𝑥¯� � ⋅ � �� ⋅ 𝑛𝑛�
𝑛𝑛 �
1 ���
𝑥𝑥¯� � ⋅ � �� ⋅ 𝑛𝑛�
where K is the number of classes in the𝑛𝑛 frequency distribution, cj is the central value of
���
each class and n is the absolute frequency of the character in the class. This procedure
correctly estimates the mean if each class’s central value corresponds to the mean
of that class’s values. This situation
∑����
occurs
𝑥𝑥� ⋅ the
when
𝑝𝑝� character is equally distributed
𝑥𝑥¯� � �
among classes. ∑��� 𝑝𝑝�
In case it is desirable to assign
∑���� 𝑥𝑥weights
different � ⋅ 𝑝𝑝� to the various observations, it is
𝑥𝑥¯� � �
possible to compute the weighted arithmetic ∑���mean:
𝑝𝑝�
In this case, xi represents values of the character within each class and pi the weights
to be assigned to each class. For example, three groups of fishers that might buy fishing
licenses to go fishing around a protected area are surveyed: exclusive recreational shore
anglers (n = 1 200), exclusive recreational spearfishers (n = 500) and exclusive boat
anglers (n = 100). Each of these three groups pays a different fee for a seasonal fishing
license: EUR 20, EUR 50 and EUR 100 respectively. If the goal is to estimate average
expenditures, the different sizes of the various classes must be taken into account:
(EUR 20 × 1 200 + EUR 50 × 500 + EUR 100 × 100) / (1 200 + 500 + 100) = EUR 32.7.
Two alternative measures of central tendency are:
• the mode, the value of the distribution � which appears with the highest
frequency; and 1
• the median, the middle value
𝑥𝑥� �that
⋅ � 𝑥𝑥� distribution of the measurements
𝑛𝑛 splits the
into two equal halves. ���
It is worth noticing that the median and the mode are the only measures of central
tendency that can be used for ordered variables, where values are ranked relative to
each other but not measured absolutely.
For a series of quantitative measures, with an uneven number of elements, the
median can be calculated as (n + 1) / 2, where n is the number of observations. On the
other hand, the median for a series with1 an � even number of elements can be calculated
as the semi-sum of the two central 𝑥𝑥¯� units,
� ⋅ n/2 � and�� ⋅ 𝑛𝑛
(n� + 1) / 2. The median is far more
robust than the arithmetic mean against 𝑛𝑛 extreme values. If the sample of recreational
���
fishers is highly heterogeneous, with few respondents boasting extremely high/low
catch, using the median will provide a more accurate measure of the data centrality,
compared to the arithmetic mean.
∑���� 𝑥𝑥� ⋅ 𝑝𝑝�
𝑥𝑥¯� �
∑���� 𝑝𝑝�
4. Data analysis 39

Variability indexes, on the other hand, represent the tendency of observational units
to take different values of the same measure. Typically, variability indexes have two
characteristics:
• their minimum value occurs when all observations have the same value of a
certain measure; and
• they increase with higher diversity of observations in the sample.
The most common indexes are based on differences between values for observational
units and their respective arithmetic means. The variance, for example, can be expressed
as the average squared difference between units and the mean:
�
1
�����𝑛𝑛�� � � ��𝑥𝑥� � 𝑥𝑥¯ ��
𝑛𝑛
���

The variance is always positive, and it can be converted to the original scale, by
a square root. This procedure generates another measure of variability: the standard
deviation. As it is obtained through a square root, the standard deviation can be
both positive and negative: if a sample of seasonal fish catch has a standard deviation
of 40 kg, this means that measured fish catch are distributed within 40 kg above or
below the arithmetic mean. Sharing the same measurement scale as the mean, standard
deviation is usually preferred to the variance. It is possible to measure the variability of
observations in terms of percentage, through the coefficient of variation (CV):
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
�� � ⋅ 100
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚

The coefficient of variation represents the ratio of the standard deviation to

the mean, and it is useful for comparing the degree of variation between two or
more distributions, even if their means are drastically different from one another.
For example, if a distribution has a coefficient of variation of 37.3 and another
distribution has a coefficient of 61.3, the second distribution is more heterogeneous
than the first one.
It is a common feature of recreational catch that a select few people catch most of the
fish while others catch no fish at all, and consequently catch distributions are generally
skewed highly positive. The sampling distribution of mean catch rate estimated from
such a distribution becomes more normal with increasing sample size, and when the
sample size is large enough, the standard error can be used to define the confidence
interval around the estimated parameter. However, in many surveys, sample sizes are
too small for normality to be assumed. In these situations, the bootstrapping technique
provides an appropriate alternative to parametric methods (Efron and Tibshirani,
1993). The basic idea of bootstrapping is that inferences about a population from
sample data can be modelled by resampling the sample data and performing inferences
about a sample from resampled data. As the population is unknown, the true error in
a sample statistic against its population value is unknown. In bootstrap-resamples, the
population is in fact the sample, and this is known; hence the quality of inference of
the “true” sample from resampled data is measurable. A comparison of bootstrapping
methods for calculating confidence intervals on catch estimates from recreational
fisheries surveys is explained in further detail in Hoyle and Cameron (2003).

4.4 ESTIMATORS: ESTIMATING POPULATION MEAN, TOTALS AND

VARIANCE
For most applications, researchers and practitioners need to make minimal
adjustments to proceed with the information obtained from their samples. Two routine
operations that are performed in every form of survey research are the estimation of
participation rates, which should always be reported, and the use of raising 58
factors
in order to shift from sample totals to population totals. Calculating these two
40 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

forms of information is straightforward and does not pose any particular problem to
practitioners and researchers.

4.4.1 Simple estimations

Participation rate
In case the data source comes from a nationwide screening survey, the first step of data
analysis is to identify the fraction of recreational fishers from the total population. The
percentage of the active fraction is called participation rate (Arlinghaus, Tillner and
Bork, 2014; Hyder et al., 2017b). The participation rate is calculated by dividing the
number of active recreational fishers by the total number of people constituting the
population, then multiplying the resulting quotient by 100 to get the percentage.
Example: if by means of a telephone survey, 1 000 people are randomly contacted
and it is obtained that 200 of them perform marine recreational fishing, then the
participation rate is 20 percent.

Raising factor
The raising factor is the factor by which the numbers in the sample have to be
multiplied to give the total numbers in the population sampled (FAO, 1966). This is a
vital step in combining and analysing sample data.
Example: assume that n catch units∑(or 𝑦𝑦 fishing effort) are sampled randomly from N
made by a segment/stratum (e.g. boat𝑦𝑦� �fishing) during a quarter of a year, and total numbers
𝑛𝑛
of fish (or fishing days) is denoted by∑y.𝑦𝑦The mean number (or mean weight) per trip is:
𝑦𝑦� �
∑𝑛𝑛𝑦𝑦
𝑦𝑦� �
𝑛𝑛
� � 𝑦𝑦�𝑁𝑁
and the estimated total caught number (or weight) Y for the segment/stratum is:
� � 𝑦𝑦�𝑁𝑁
The raising factor is thus: � � 𝑦𝑦�𝑁𝑁
𝑁𝑁
𝑛𝑛
𝑁𝑁
This approach could be used also
𝑁𝑁 for raising length frequency distribution of catches
𝑛𝑛
(ICCAT, 2016).
𝑛𝑛
4.4.2 In-depth estimations
For most types of data collection procedures, such as non-probabilistic sampling, working
with sample statistics is enough: they are easy to calculate and highly informative about
the data at hand. However, for making rigorous inference from probabilistic sampling,
considering information in the sample is not enough, for two reasons.
First, a practitioner or researcher needs to understand if, and how, collected
information must be treated, accounting for those units that were not observed. While
sample means may coincide with population means in simple random sampling, this
does not hold for other forms of sampling designs.
Moreover, measuring the uncertainty associated with a certain estimate presents
another daunting task: this calculation differs among different sampling designs.
A complete overview of statistical estimators is needed to address these two issues.
The following section is more technically demanding than the previous one and briefly
introduces how statistical estimators can be constructed. This text refers to a design-
based paradigm, which is covered in detail in Hankin, Mohr and Newman (2019).
Complete understanding of this paradigm is not necessary if using non-probabilistic
sampling or if simply aiming to measure means and totals in simple random sampling.
This short section introduces statistical estimation of population parameters,
following design-based inference. It serves three purposes: first, it shows how it
4. Data analysis 41

is possible to move from sample statistics to population-level estimates; second, it

explains the properties characterizing good estimation; and finally, it explains why
estimates always come with uncertainty. Until now, it has always been explained how
to calculate population measures, such as the mean, without focusing on their accuracy.
Well-designed random sampling, such as simple random sampling, enables one to
do this: the mean of the sample corresponds to the population mean. However, this
approach ignores the fact that when moving from samples to populations, estimates
are also characterized by uncertainty. Ignoring uncertainty is dangerous and it is
encouraged to better understand estimators, to better interpret information at hand.
An estimator is a statistic used to estimate a population parameter and incorporated
in a formula that can be applied to sample data to generate a numerical estimate of a
chosen population parameter. For example, estimating the average seasonal catch of a
certain fish species by recreational fishers in a certain area is based on the arithmetic
mean, which is then calculated over the sample:
^ � � �� �𝑛𝑛
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀
���
^ � � �� �𝑛𝑛
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀
However, it is important to note that population mean is an estimated value. It is
different from sample mean, which comes with no ��� uncertainty, because its calculation

is exclusively based on observed data only. Estimated population mean is uncertain,

because its value depends on S, which is the overall sample space containing all the
samples that can be extracted. Given𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀
a^realized �� �𝑛𝑛 selection S = s, the population
� �sample
mean will be calculated as: ���
^ � � �� �𝑛𝑛
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀
���

�𝑝𝑝𝑀𝑀� � ��𝑝𝑝𝑀𝑀�𝑡𝑡𝑀𝑀�
Therefore, it will𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 𝑀𝑀�𝑡𝑡𝑝𝑝𝑝𝑝𝑀𝑀𝑡𝑡𝑜𝑜𝑝𝑝
be calculated using the�sample �𝑀𝑀𝑀𝑀𝑣𝑣 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
s and𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝
y values of its units. 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝
While
y values are fixed, the population mean is a random variable, because several different
�𝑝𝑝𝑀𝑀� �
samples ��𝑝𝑝𝑀𝑀�𝑡𝑡𝑀𝑀�
could 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑜𝑜𝑜𝑜from
be extracted 𝑡𝑡𝑡𝑡𝑡 𝑀𝑀�𝑡𝑡𝑝𝑝𝑝𝑝𝑀𝑀𝑡𝑡𝑜𝑜𝑝𝑝
the target� population.
�𝑀𝑀𝑀𝑀𝑣𝑣 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑜𝑜𝑜𝑜
If 𝑡𝑡𝑡𝑡𝑡 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝
different 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝
recreational
fishers are sampled and their average catches are calculated, these will differ slightly:
estimators account for this variability. ^ The
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 � �� �𝑛𝑛
� probability distribution of the estimator, as
a random variable, is the distribution generated ��� by all the possible samples that could

be extracted, the sampling distribution. The sampling distribution of the population

parameter is essential to assessing the performance of a certain estimator in estimating
a population parameter.
The sampling distribution of an estimator depends on, at least, three elements:
the distribution of the population𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 variable,
^ � �the �� �𝑛𝑛sampling design and the estimator
itself. Just as the location and spread of the ��� distribution of the observed values can be
characterized, they can also be characterized for a sampling distribution, in terms of
expectation and sampling variance. Expectation is a measure of the average value of the
�𝑝𝑝𝑀𝑀� � ��𝑝𝑝𝑀𝑀�𝑡𝑡𝑀𝑀� 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 𝑀𝑀�𝑡𝑡𝑝𝑝𝑝𝑝𝑀𝑀𝑡𝑡𝑜𝑜𝑝𝑝 � �𝑀𝑀𝑀𝑀𝑣𝑣 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝

estimator, and variance is inversely related to its precision: the higher the variance, the
lower the precision of the estimate. A good estimator has a low, or nonexistent, bias:
Bias is the difference between the expected value of the estimator and the real
value of the population parameter: if bias is zero, the estimator is unbiased, and its
expected value is centered on the real value of the population parameter. This does not
mean that the estimation will be precise, but rather that its distribution will always be
sampled on the real value which it is attempting to estimate. For example, in a simple
random sampling survey estimating average seasonal catch from recreational fishers,
the estimator is unbiased if its distribution has an expected value centered on the
real average catch of fish that all the recreational fishers in the study area attain. This
may seem like an obvious concept, but it is not: only few statistical designs guarantee
 42 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

�
unbiased estimates and an���
analytical, � �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵�estimation
not approximated,
� 𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉 of the variance.
��� � 𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉 � �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵��

�𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵� �
Another important metric��� � �𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉
is���
the 𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉
mean squared �� �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵� �
error (MSE), which corresponds to:
The mean squared error is��� a sum�of𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉
the variance and the�squared
� �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵� ���value of the bias � �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵��
� 𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉
and it provides an overall measure of estimator precision. It is possible to calculate the
standard error (SE) and the ��
coefficient of variation
� �𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 (CV) of the estimator:
𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
�� � �𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉
��� � 𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉 � �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵�� 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
�����𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉�
�𝑉𝑉 �𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡
𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉� 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
�𝑉𝑉�� � �𝑉𝑉
�𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉� 𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉 ���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉� 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
� ��𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉
�𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉� 𝑜𝑜𝑜𝑜���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉�
𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉 𝑡𝑡𝑡𝑡𝑡 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣

Finally, because of the ��central

� �𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 ���
��� � 𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉 𝑜𝑜𝑜𝑜��𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵�
𝑡𝑡𝑡𝑡𝑡
� 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
�𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵�
𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉 �
��that
� �𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵��� the
�𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝑜𝑜𝑜𝑜of𝑡𝑡𝑡𝑡𝑡 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
���limit theorem,
� 𝑉𝑉𝐵𝐵𝑉𝑉𝐵𝐵𝐵𝐵𝑉𝑉𝑉𝑉𝑉𝑉
 � which � states distribution
�𝑉𝑉 � �𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉�
a sample mean converges to a Gaussian 𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉 ���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉� 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
distribution when n �→ ∞, regardless of the
∑� � ^ � ∑��� ��
shape of the sampled�𝑉𝑉 �� �𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉�
distribution, � 𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉
� ��� � it is possible ���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉�
to compute �𝑉𝑉𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
�� �𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉�
confidence 𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉
intervals for���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉� 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
the�� � �𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉
estimator. 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡to
It is important ∑�
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
���note
�
��� that it is not always possible
∑ �
^ �
��� � � to obtain an exact
expression of the variance∑�of �an� estimator; in many situations, �
variance
∑� can only be
��� � ^
�� � �𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 ��� ��
�𝑉𝑉 � �𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉� �
approximated with the Delta �
𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉 
∑ ��
� ���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉�
��method, 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
based
���𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝑜𝑜𝑜𝑜on
𝑡𝑡𝑡𝑡𝑡Taylor series.
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 �
𝜋𝜋 � ����� �𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝑜𝑜𝑜𝑜 𝑡𝑡𝑡𝑡𝑡 ^𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
� ∑ � �
𝜋𝜋��� � ��� �
For simple random sampling � without replacement, where  n
� observations are extracted
�𝑉𝑉 � �𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉� 𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉 ���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉� 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
�
�𝑉𝑉
∑��� ��� �𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉� 𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉� 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 ∑��� ��
from a population of𝜋𝜋N�units, ∑�
�𝑉𝑉 �estimating
��𝑡𝑡𝐵𝐵𝑉𝑉�𝐵𝐵𝑉𝑉� sample
𝑉𝑉𝑉𝑉𝑉𝑉𝑜𝑜𝑉𝑉mean ^ (μ,
𝜋𝜋���
���𝑉𝑉𝑉𝑉𝑡𝑡𝑉𝑉� � e.g. the^ average∑� ��� � seasonal catch
�𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣
� �
� � � ���
of a certain species among anglers), proportion (π, e.g. the
�
proportion of recreational
�∑� � � �
�� ^∑��� �∑� ��� �� ^
fishers among the ∑inhabitants
� � � ���
∑�
𝜏𝜏 �in a
��� 𝑦𝑦� � ��area, for yi which is 1
certain � or� 0)
∑��� �� �
�
and total (τ, e.g. the
��� �� � �
total number𝜋𝜋of�recreationists which fish in a certain 𝜋𝜋
area) ^
��� is �
straightforward through the
𝜏𝜏 ^ � � ⋅ ∑��� 𝑦𝑦� 𝑛𝑛�� ����� 𝜏𝜏^� ∑� ��� 𝑦𝑦� � �� ∑� �
�
use ���
of mean
� per
∑��� �� unit estimators (mpu): � �
∑��� �� ^ ∑� ��
�� ^� � �
��� ���
� �
𝜏𝜏����
^� � ��⋅ ∑��� 𝑦𝑦� ∑𝑛𝑛�
∑��� �� ^ � ∑∑��� �� �
�� �� ^� � �
� ��� �𝜋𝜋 ^ � ���
���
𝜋𝜋
From population �
� ���
mean ��
∑��� ��
sample mean is ��� estimated
� ^ � as:
∑ �
��� �
� �
∑ ��� �� � ^ ∑� �� ∑������ ∑��
𝜋𝜋 � � 𝑉𝑉 ^ � 𝑉𝑉� ^ � �1 � 𝑜𝑜� ⋅ 𝜋𝜋 �
𝜎𝜎��� �𝜋𝜋���^ ����
𝜋𝜋 ^ �
� ∑
𝜏𝜏 � ���� �𝑦𝑦��� � 𝜋𝜋�
^ ��
∑ � ^� � � ∑ �
��� �
From population proportion ∑��� �� � ������� sample 𝑛𝑛proportion �^ is���
∑��� ��𝜋𝜋 ^estimated
� ��� � as:
𝜋𝜋 � � �1 � 𝑜𝑜�𝜋𝜋⋅��� ^𝜎𝜎��� � �
� 𝑉𝑉�^^ � 𝑉𝑉�^^ �
∑�
��� �� 𝑦𝑦 𝑛𝑛 � �� ^ ∑��� �� 𝑛𝑛 �
^ �
𝜏𝜏��� 𝜋𝜋 � ⋅ ∑
��� ���
��� � ���𝜏𝜏 �
𝜋𝜋
∑� �
� ∑������ 𝑦𝑦� ����
� � 𝜋𝜋 ^ �� � 𝜋𝜋��� ^ � ��� �
∑ �
� �
� ^ � ^ �
� 𝑉𝑉���� ^ � � ⋅ 𝑉𝑉 ⋅ ����� �
� ∑�
From
𝜏𝜏 �𝜏𝜏 ∑ ���� 𝑦𝑦 � � 𝜏𝜏 �� ��� 𝑦𝑦� � ��
𝑦𝑦� � ��� sample total is estimated���
𝜏𝜏 � ∑�as: 𝑦𝑦� � ��
^ � � ⋅ ∑���
𝜏𝜏��� 𝑦𝑦� population
𝑛𝑛 � ����� ^total � ∑��� ^
𝑉𝑉���� ^
^ � � ⋅ 𝑉𝑉 ⋅ ����� �
^ � � ⋅ ∑��� 𝑦𝑦� 𝑛𝑛 �
𝜏𝜏��� ����� ^
^ � � ⋅ ���
𝜏𝜏��� ∑����𝑦𝑦� 𝑛𝑛����� �����
∑  ^ ��� 𝜏𝜏 � ∑��� 𝑦𝑦� �⋅ ∑
^ � � � 𝑛𝑛
�1𝑦𝑦� 𝑜𝑜��⋅ 𝜎𝜎�� ^
𝜏𝜏 ^ � � ⋅ 𝑦𝑦 𝑛𝑛 � �� ^ � �^ ���
𝜏𝜏 � ∑��� 𝑦𝑦� � �� 𝜏𝜏 ��� ����� ���
^
𝑉𝑉����^ � 𝑉𝑉����
^^ �
^ Similarly,
� � ⋅ ∑���it𝑦𝑦�is𝑛𝑛possible ^ to estimate∑��� the 𝑛𝑛 �
𝜏𝜏��� � ����� �𝑦𝑦�sampling
� ���� ^ � variance �1 � 𝑜𝑜� ⋅ 𝜎𝜎for
�^ the averages (μ),
^
��� ��⋅ ∑��� 𝑦𝑦� 𝑛𝑛 � ��proportions
^
��� (π) and totals (τ). In this 𝜎𝜎 �^ �
��� case,𝑉𝑉the ^ �1 �
parameter
�1
^𝑉𝑉 �� 𝑉𝑉
𝑜𝑜� 𝑜𝑜�
^⋅ 𝜎𝜎 �⋅^
� 𝜎𝜎
of�^interest��� will be denoted as θ:
���� � 1 ������� ��� 𝑛𝑛
� ^
𝑉𝑉�^ ^
𝑉𝑉�^
��� 𝜎𝜎 ^ �
�^ 𝑉𝑉 �^
�
𝑉𝑉�∑
^
^^ �
���
� � � ����
�𝑦𝑦 ^𝑛𝑛 �
�^ �^
�1 �𝑛𝑛𝑜𝑜� ⋅ 𝜎𝜎���
��� ���
�
���
���
�1 � 𝑜𝑜� ⋅ 𝜎𝜎���
𝑉𝑉�^ ^ � 𝑉𝑉 ^
����^
𝑉𝑉 � 1
� 𝑉𝑉 ^^
���� � 𝑉𝑉 ^^
���� �
𝑉𝑉���^^ � �
����
� ^
⋅ 𝑉𝑉 ⋅�1 ������ 𝑜𝑜�� 𝑛𝑛 �^
�⋅ 𝜎𝜎��� 𝑛𝑛
𝑉𝑉�^ � 𝑉𝑉 ^ �
𝑉𝑉 ^ � � ⋅ ^
𝑉𝑉 ⋅ �� �
or �1 � 𝑜𝑜�
^
��� ^⋅^ 𝜎𝜎�
𝑉𝑉���� �^� � ⋅ 𝑉𝑉
� ^
��� � ^
^ ⋅ ����� �𝑛𝑛
��� ���
���
𝑉𝑉�^ ^ � 𝑉𝑉 ^
����^ � ^^ 𝑛𝑛 � � � ⋅ 𝑉𝑉 ^ ⋅ ����� �
��� 𝑉𝑉����
𝑉𝑉�the^^ � � � ⋅ 𝑉𝑉 ^ ⋅ �� � that appear ^^ � � � ⋅ 𝑉𝑉
𝑉𝑉���� ^
where f is the sampling fraction, ��� ^ fraction
𝑉𝑉���� � of ^ the N��� units in the sample⋅ ����� �
^ � � ⋅ 𝑉𝑉 ⋅ ����� � 64
of size n. ∑���
�
�^ � �𝑦𝑦� � ����
^ �
𝜎𝜎����
∑����𝑦𝑦 � ��^ � �
^^ � � � ⋅ 𝑉𝑉
𝑉𝑉���� ^ ⋅ ��𝜎𝜎���
�^ � ∑
� ����𝑦𝑦� � ���� ^ ���
� 𝑉𝑉64
� 1
�^ ���
𝜎𝜎��� � 𝑉𝑉 � 1
𝑉𝑉 � 1
�
When using stratified random sampling, ∑����𝑦𝑦� the
� �situation
^ � � is slightly more complex.
���
Units are divided into L strata
^
�
𝜎𝜎���of�size N∑h���
, h=1,
�𝑦𝑦� �2,�…,^ L �
^ � �, such that the sum∑���
of their �
𝜎𝜎 �^ �∑��� �𝑦𝑦
𝑉𝑉 � � � ����
1 ��� �𝑦𝑦 � � ^ �
observations equals N, the size𝜎𝜎��� ^ the�population.
of
� ��� �^ independently
𝑉𝑉 � 1Samples are selected
𝜎𝜎��� �
� ���
from each of the L strata, usually through𝑉𝑉simple � 1 random sampling. 𝑉𝑉 � 1
With any
particular stratum,∑it is possible to � unbiased estimates of means, proportions and
obtain
�^ ����𝑦𝑦� � ���� ^ �
𝜎𝜎��� � by using properly weighted stratified estimators it is also possible to
totals. Moreover, 64
𝑉𝑉 � 1 64

64
4. Data analysis 43

obtain unbiased estimates of the overall parameters, across strata. The main advantage
of stratified random sampling lies in its capacity to significantly reduce the variance
of estimated parameters, compared to simple random sampling. However, if strata are
not correctly identified, estimation will be biased. In the next few lines there will be
references to stratified estimators of a population parameter, denoted by the lowercase
“st” (e.g. μst) and combining information from multiple strata, and to stratum-specific
estimators, denoted by the lowercase “h” (e.g. μh).
The overall population mean (μ) can be expressed as:
� ��
𝜇𝜇 � � � 𝑊𝑊�� ⋅ 𝜇𝜇𝜇𝜇 � � � 𝑊𝑊� ⋅ 𝜇𝜇�
� 𝜇𝜇 � � � 𝑊𝑊� ⋅ 𝜇𝜇�
���
𝜇𝜇 � � 𝜇𝜇 �
��𝑊𝑊
�� ⋅ 𝜇𝜇
𝑊𝑊 ⋅ 𝜇𝜇��� �
It corresponds to the weighted average 𝜇𝜇 �of�� �
the
�
⋅𝜇𝜇𝜇𝜇�
𝑊𝑊�stratum
� �� 𝑊𝑊� ⋅ 𝜇𝜇�
���
means,
𝑊𝑊� ⋅weighted on the basis
�
𝜇𝜇𝜇𝜇 ��
𝜇𝜇
𝜇𝜇 � � �
����� � 𝑊𝑊
𝑊𝑊 𝑊𝑊
� ⋅⋅ 𝜇𝜇
𝜇𝜇⋅ �𝜇𝜇
𝑊𝑊 �⋅ 𝜇𝜇 � ���
��� � � 𝜇𝜇 � � 𝜇𝜇�
���
of the stratum weights, the fraction of the ���
total
������
number
� � of ���
units which are contained in
���
a certain stratum (Wh = Nh/N). Then the stratified estimator
� � � of the population mean is:
𝜇𝜇^�� � � � 𝑊𝑊�� ⋅𝜇𝜇𝜇𝜇^^��
� 𝜇𝜇�^����� � 𝑊𝑊𝑊𝑊
� � �⋅ 𝜇𝜇
⋅^�𝜇𝜇^�
���
𝜇𝜇^�� �𝜇𝜇^�� �� ��𝑊𝑊
�� ⋅
� 𝑊𝑊 𝜇𝜇
^ ���
⋅ 𝜇𝜇⋅^�𝜇𝜇^�� �
𝜇𝜇^�� � �� 𝑊𝑊�𝜇𝜇�^��
�
�
���𝑊𝑊� ⋅ 𝜇𝜇^�
𝜇𝜇𝜇𝜇^��
𝜇𝜇^��
�
�
^𝜇𝜇^��
�� �
����
� � 𝑊𝑊�𝑊𝑊
𝑊𝑊
𝑊𝑊
��� ⋅𝜇𝜇𝜇𝜇
� ⋅�
⋅ ^⋅^
𝜇𝜇 ��𝜇𝜇^
^ 𝜇𝜇
^
� �� � � 𝑊𝑊� ⋅ 𝜇𝜇^�
���
�� ��� � �
���
���
Individual stratum means are estimated using��� the mean-per-unit
��� ��� estimator, obtained
from Sh which is a random set of sample units selected
𝜇𝜇^�� � � ��� �𝜇𝜇^ from stratum
𝑛𝑛� � � � �𝑛𝑛 h.
𝜇𝜇^�� � � ��
�� ��� ��𝑛𝑛�
����
𝜇𝜇^�� �𝜇𝜇^�� � � ���� �𝑛𝑛
�
𝜇𝜇^�� � ����
𝑛𝑛
𝜇𝜇^�� � � ��� �
�𝑛𝑛� �� ��� �𝑛𝑛�
�� ����
𝜇𝜇𝜇𝜇^�� � � �� ��� 𝑛𝑛 ��������
𝜇𝜇^�� �
���
𝜇𝜇^�� � �𝑛𝑛𝑛𝑛����𝑛𝑛𝜇𝜇^ � � ��� �𝑛𝑛�
���� ��
^�� � ���
����
���� ����
����
����
���� ����
����
For simple random sampling within𝐸𝐸�𝜇𝜇 ^� � �the
strata, 𝜇𝜇� expected 𝐸𝐸�𝜇𝜇 ^�^� ��� value
𝜇𝜇�𝜇𝜇 of the mean for each
𝐸𝐸�𝜇𝜇 � �
stratum (μh) corresponds to the mean of 𝐸𝐸�𝜇𝜇 sampled
� �^𝜇𝜇���variables
^�𝐸𝐸�𝜇𝜇 𝜇𝜇 𝐸𝐸�𝜇𝜇 ^ � �in𝜇𝜇the stratum (μ):
𝐸𝐸�𝜇𝜇^� �� � 𝜇𝜇� � � �
𝐸𝐸�𝜇𝜇
𝐸𝐸�𝜇𝜇 ^^ �
� ��� 𝜇𝜇
^���^���
𝐸𝐸�𝜇𝜇
𝐸𝐸�𝜇𝜇 � 𝜇𝜇𝜇𝜇���𝜇𝜇� 𝐸𝐸�𝜇𝜇^� � � 𝜇𝜇�

Therefore, the expected value of the𝜇𝜇^�� � � ���

overall �𝜇𝜇^
mean 𝑛𝑛� �(μ�
𝜇𝜇^�� �st)�
�� is�also
���
�� unbiased:
�𝑛𝑛��𝑛𝑛
�
����
𝜇𝜇^�� �𝜇𝜇^�� �� �� �
�� 𝜇𝜇 𝑛𝑛
� ^� �� ����
𝑛𝑛 � ���� � � 𝑛𝑛
𝜇𝜇^�� � � ������𝑛𝑛� �� � �� �
𝜇𝜇^��
𝜇𝜇𝜇𝜇^�� �� � �� ��� 𝑛𝑛 �������
𝜇𝜇^����� �� �𝑛𝑛𝑛𝑛����𝑛𝑛𝜇𝜇^ �� � � ��� �𝑛𝑛�
���� ��
^�� � ���
����
���� ����
����
���� �
𝑉𝑉�𝜇𝜇^�� � � � 𝑊𝑊
����
𝑉𝑉�𝜇𝜇 ^��^�^�����is:
𝑉𝑉�𝜇𝜇 �
����
��
^�^� �
� 𝑊𝑊𝑊𝑊� �𝑉𝑉�𝜇𝜇
����
� �� mean
The stratified estimator of the variance of the � 𝑉𝑉�𝜇𝜇 �� � 𝑉𝑉�𝜇𝜇 �
� ��� � �
𝑉𝑉�𝜇𝜇^��𝑉𝑉�𝜇𝜇 �� ��𝑊𝑊� � 𝑉𝑉�𝜇𝜇 � �^
�
� ����� � � 𝑊𝑊� 𝑉𝑉�𝜇𝜇^� �
� ���
𝑉𝑉�𝜇𝜇^��^�� 𝑊𝑊 ��𝑉𝑉�𝜇𝜇 ^���� � �
𝑉𝑉�𝜇𝜇
𝑊𝑊 ^�� 𝑉𝑉�𝜇𝜇 � ^ �
𝑉𝑉�𝜇𝜇
𝑉𝑉�𝜇𝜇 ^��^
^�� ��� �
�� � � 𝑊𝑊𝑊𝑊�𝑊𝑊 � ^ ��� �
���𝑉𝑉�𝜇𝜇
𝑉𝑉�𝜇𝜇 �� � 𝑊𝑊��𝑉𝑉�𝜇𝜇
𝑉𝑉�𝜇𝜇𝑉𝑉�𝜇𝜇 ^ �^ ������� � 𝑊𝑊�� 𝑉𝑉�𝜇𝜇^� �
���
𝑉𝑉�𝜇𝜇 ^ �� � ��� 𝑉𝑉�𝜇𝜇
� ^��� ����� � �
���
��� ��� ���
���
And considering that the stratum-specific 𝑁𝑁 � 𝑛𝑛
� estimator 𝜎𝜎² �𝑁𝑁of the
𝑛𝑛� variance
𝜎𝜎²� of the mean is:
� ��
𝑉𝑉�𝜇𝜇^� � � � 𝑉𝑉�𝜇𝜇^��^����𝑛𝑛�� ��𝑁𝑁� � 𝑛𝑛�� �� �𝜎𝜎²�� �
𝑁𝑁� 𝑁𝑁��𝑁𝑁𝑛𝑛𝑉𝑉�𝜇𝜇 � 𝜎𝜎²
� � 𝑛𝑛� � 𝑁𝑁 � 𝑁𝑁 𝑛𝑛�𝑛𝑛
� �^�� � 𝑁𝑁
𝑉𝑉�𝜇𝜇^�𝑉𝑉�𝜇𝜇 � � �
�
𝑉𝑉�𝜇𝜇 �𝑛𝑛�
^ � ���𝜎𝜎²
� �𝜎𝜎²
� � �
� 𝑛𝑛
𝑁𝑁
� 𝜎𝜎²
� �� � � � ���
�
𝑉𝑉�𝜇𝜇^� � 𝑁𝑁
� 𝑁𝑁
𝑁𝑁
�𝑁𝑁 ��
𝑁𝑁���𝑛𝑛 𝑛𝑛
𝑛𝑛 �𝑛𝑛 �
� 𝜎𝜎²
𝜎𝜎²
� 𝜎𝜎²
𝑛𝑛�𝜎𝜎² �𝑁𝑁� � 𝑛𝑛�𝑛𝑛 𝜎𝜎²�
𝑁𝑁𝑁𝑁���������
�� ��𝑛𝑛 𝑁𝑁
^^
𝑉𝑉�𝜇𝜇 �� ��� �� ��
��
𝑉𝑉�𝜇𝜇 ^���^���� ��� � 𝑁𝑁
𝑉𝑉�𝜇𝜇 ^��𝑛𝑛 ��
� 𝑛𝑛�
𝑁𝑁��𝑁𝑁��𝑉𝑉�𝜇𝜇 ��� 𝑁𝑁 ��� �
�
𝑉𝑉�𝜇𝜇 𝑛𝑛 �𝑛𝑛 𝑛𝑛�
Then the stratified estimator of the variance of the mean is: 𝑁𝑁� 𝑛𝑛� �
� �
� �� �² � ���� ��� �²
𝑉𝑉�𝜇𝜇^�� � � ∑����𝑉𝑉�𝜇𝜇 𝑊𝑊��^ �� � � �
∑����� �� 𝑊𝑊��� �where � ��𝜎𝜎
������� � ^²�² �� ∑
� ��� ���� �� �� � �²
�� ���� ���where
�� ��� �²
� ���∑∑
�² 𝜎𝜎^²
𝑉𝑉�𝜇𝜇
��^
�� ���� �� ∑ ���
��𝑊𝑊 � �� � � where
��� � �𝜎𝜎
^²
�� ���
�
� ��
� �² �
�� �� �� �� �² ��� �
� � ∑����� � � � �� �� �²� �� �� � ��� �² �� ��� ��
���where
��� ���� �²�
𝑉𝑉�𝜇𝜇^��𝑉𝑉�𝜇𝜇 � ∑
�𝑊𝑊 ���� 𝑊𝑊 ����² �𝜎𝜎
�^² �
�� �� � � ������ �² �� �� � � �²
𝑉𝑉�𝜇𝜇^��^��� ���∑ ∑ � ^𝑊𝑊
𝑉𝑉�𝜇𝜇
���𝑊𝑊 �∑��
������������ ��² �� �
� where
where ��𝜎𝜎^² 𝜎𝜎^²
���²� �∑
�where
��� ∑
�� �𝜎𝜎 ^² ���∑
�
� ��� ��
� � � �� ����
���
�� ��² �² � � � �� � � �� � ���
���
��� ����
�� ��
��
�� �����²
�²���²��� � ���� ���� ��� �²
�where
� �� �
𝑉𝑉�𝜇𝜇
𝑉𝑉�𝜇𝜇 ^��
𝑉𝑉�𝜇𝜇
For ���estimating
^��^ ��
� ∑∑∑
��∑���
� 𝑊𝑊�𝑊𝑊
𝑊𝑊 ���^
𝑊𝑊𝑉𝑉�𝜇𝜇
�
��population
�
����
������
��
����∑
��������
��² � � �
where
����𝑊𝑊�
� 𝜎𝜎^²
�where
�
����proportions � � 𝜎𝜎^²𝜎𝜎^²
�
�� ��
� ��
(π), ∑∑
�
∑ ����
�
�where
����
the� ��
�
���same
���
��
�� ��
� �²� � ∑ �
��
�𝜎𝜎
^²procedure ��� is ��applied, assuming
�
𝑉𝑉�𝜇𝜇 ^���� ���
��� � �� �� � where ��𝜎𝜎
𝜋𝜋^ ^²�
�
�� �� ∑𝑊𝑊
���
⋅ 𝜋𝜋^ �������
�� �
��� ����� ����� � ���
�� 𝜋𝜋^��
� �� ��
𝜋𝜋�^���
� � � 𝑊𝑊𝑊𝑊� �⋅ 𝜋𝜋
⋅^�𝜋𝜋^�
�� � � � ��
that yj is always 1 or 0: � ��
���
𝜋𝜋^�� �𝜋𝜋^�� � ��𝑊𝑊�� � ⋅ 𝑊𝑊𝜋𝜋^� ⋅�𝜋𝜋^���
𝜋𝜋^�� ����
�
𝑊𝑊�𝜋𝜋^���⋅ 𝜋𝜋^� �� ���𝑊𝑊 � ⋅ 𝜋𝜋^�
𝜋𝜋𝜋𝜋
^^𝜋𝜋
�� ^�
𝜋𝜋^�� � �
�� ��
������� � 𝑊𝑊
𝑊𝑊 𝑊𝑊
� ⋅ ⋅
𝑊𝑊� ⋅ 𝜋𝜋^�
��� � �
𝜋𝜋
𝜋𝜋
^⋅^
�𝜋𝜋
�^
𝜋𝜋^
��� �
��� � 𝑊𝑊� ⋅ 𝜋𝜋^�
���
��������� ���
���
� � � �� �² ��
𝑉𝑉�𝜋𝜋^�� � ���𝑉𝑉�𝜋𝜋� ∑ 𝑊𝑊�^ �� � ∑ � �� 𝑊𝑊� ��where
� �
�
�
� � �� ����𝜎𝜎
�� ^² �²� �
�² � 𝜋𝜋 �1 � 𝜋𝜋������� 𝜋𝜋 �1 � 𝜋𝜋 �
�� ���� ���where
� � �
𝑉𝑉�𝜋𝜋��^�� ��� � ∑ ��� ��𝑊𝑊 � � �� �where
� ��
𝜎𝜎^²�𝜎𝜎^² � ����� ���� � �𝜋𝜋� �1 � �𝜋𝜋� �
� �� ���
�² �� � � �� ��
𝑉𝑉�𝜋𝜋^��𝑉𝑉�𝜋𝜋 � �^ ∑����� ∑ 𝑊𝑊 � � � � ������ �
�����
� �
� � ��²where
��� ���� 𝜎𝜎
where ^² �²�� �
� �
� �
𝜋𝜋 � �1 � 𝜋𝜋 �
�1 � � 𝜋𝜋� 𝜋𝜋
� 𝑊𝑊 � � � � 𝜎𝜎
^² �where� � � 𝜋𝜋 �� � � � �1 � 𝜋𝜋� �
𝑉𝑉�𝜋𝜋 ^ � � ∑ 𝑊𝑊 �² � � � � � 𝜎𝜎
^² �
� ∑ � � � � �
where � � � � ��
�1
𝑉𝑉�𝜋𝜋 ^ � 𝑊𝑊 � � � � 𝜎𝜎
^² � � � 𝜋𝜋 𝜋𝜋�� ��
�� ���
� �� �� �� ��� � �² � � ��� �
𝜏𝜏^𝑉𝑉�𝜋𝜋
𝑉𝑉�𝜋𝜋 �
𝑉𝑉�𝜋𝜋^ �� ^ ���∑
𝑁𝑁𝜇𝜇��
^��^��� ��
��� ∑for
∑ ∑
���
�
𝜏𝜏
^ 𝑊𝑊𝑊𝑊
� �
𝑊𝑊
�𝜏𝜏
^�
�
𝑉𝑉�𝜋𝜋 �
�
𝑁𝑁𝜇𝜇
�
�^and
�
�
^
��
����
�
�� � �also
��
�
�
� �� �
����∑��
�²
�
��𝐸𝐸�𝜏𝜏
�²� � �^
�²
���
�
�𝜏𝜏^
𝑊𝑊
� �
� where
where �
and
� �
where
�� � 𝐸𝐸�𝑁𝑁𝑁𝑁
��
also
� 𝜎𝜎
^²
𝜎𝜎
^²��
� ^
𝐸𝐸�𝜏𝜏
𝜎𝜎
^²
�
� � ��
��²
^ �
��
���
�
�
�
���
�
𝑁𝑁𝐸𝐸�𝜇𝜇
�����
where
� � ^ ��� 𝑁𝑁𝜇𝜇
�
��
��
𝜋𝜋
�𝐸𝐸�𝑁𝑁𝑁𝑁 𝜋𝜋
� 𝜋𝜋�1
�1
^𝜎𝜎
^² �
�1� � �
�
𝜋𝜋
�
� � ��
𝜋𝜋 �
𝜋𝜋
�� �
𝑁𝑁𝐸𝐸�𝜇𝜇𝜏𝜏�^��𝜋𝜋��1 𝑁𝑁𝜇𝜇��𝜋𝜋�𝜏𝜏�𝜏𝜏
�� Finally,
𝑉𝑉�𝜋𝜋^�� � � ∑���
�� �
���
��� totals
���
𝜏𝜏^𝑊𝑊
��
��� ��� �
�� ����𝑁𝑁𝜇𝜇 (τ),
���
���^
�� ���
��
�
the
� �∑ procedures
���
���
�
� �
���
����� � �
��
^� and��also
�𝜏𝜏where are
� ���
𝜎𝜎^²�𝐸𝐸�𝜏𝜏 ��almost
��� �^��������
� ��
�
�
�� identical:
� �� �
��
�𝐸𝐸�𝑁𝑁𝑁𝑁
𝜋𝜋 �1
� � ���^��� �� 𝜋𝜋
�� �𝑁𝑁𝐸𝐸�𝜇𝜇
�� �
�� �� ��^�� ��� 𝑁𝑁𝜇𝜇 �
� � � � ��
𝜏𝜏^�� �𝜏𝜏^𝑁𝑁𝜇𝜇 ^ � ∑ ∑𝜏𝜏
^ � and � also
and 𝐸𝐸�𝜏𝜏
� ^
also𝐸𝐸�𝜏𝜏 � � 𝐸𝐸�𝑁𝑁𝑁𝑁 ^ � � 𝑁𝑁𝐸𝐸�𝜇𝜇
� ^ � � 𝑁𝑁𝜇𝜇 � 𝜏𝜏
𝜏𝜏^���������� 𝑁𝑁𝜇𝜇 ^��𝜏𝜏���
^� � 𝜏𝜏^�^𝜏𝜏��^�and 𝜏𝜏^𝐸𝐸�𝜏𝜏 � ��
�^ ��𝐸𝐸�𝑁𝑁𝑁𝑁 ��^ �������𝑁𝑁𝐸𝐸�𝜇𝜇𝑁𝑁𝐸𝐸�𝜇𝜇
�� ^^ ��
�� 𝑁𝑁𝜇𝜇�� 𝜏𝜏 �𝜏𝜏 � 𝑁𝑁𝜇𝜇 � 𝜏𝜏
�
�� 𝑁𝑁𝜇𝜇 � ∑also �� and also 𝐸𝐸�𝜏𝜏
�� � ^ 𝐸𝐸�𝑁𝑁𝑁𝑁 ������ 𝑁𝑁𝐸𝐸�𝜇𝜇 ^��
𝑁𝑁𝜇𝜇 �^�� � 𝑉𝑉∑���∑ ^ 𝑁𝑁𝐸𝐸�𝑁𝑁𝑁𝑁
�^�^�^ � ^
�𝑁𝑁𝜇𝜇�²
��� �∑ �� �� �� �² ����
𝑉𝑉�𝜏𝜏 ^��𝜏𝜏^��
𝜏𝜏𝜏𝜏𝜏𝜏^�� ^��� 𝑁𝑁𝜇𝜇
𝑁𝑁𝜇𝜇 ∑
^^�����
𝑁𝑁𝜇𝜇 ^ � �∑
�𝑉𝑉�𝜏𝜏 �𝜏𝜏
�^
�^
∑ ����
���
𝜏𝜏 𝜏𝜏�^
𝜏𝜏
^ �then
�𝜏𝜏^ and
and
∑ � �𝑉𝑉�𝜏𝜏
and
𝑁𝑁𝜇𝜇 ^also
also
𝑉𝑉
���
^��
also
� �𝜏𝜏�^𝐸𝐸�𝜏𝜏
𝐸𝐸�𝜏𝜏
∑ �
𝐸𝐸�𝜏𝜏
�� ^ �^��
then
^�� �𝜏𝜏^
��and
�
��� � 𝐸𝐸�𝑁𝑁𝑁𝑁
𝐸𝐸�𝑁𝑁𝑁𝑁
𝑉𝑉�𝜏𝜏 �^also
𝐸𝐸�𝑁𝑁𝑁𝑁 �
^���
�𝐸𝐸�𝜏𝜏
�
��
∑ ��
^ ���𝑁𝑁𝐸𝐸�𝜇𝜇
�𝑁𝑁𝐸𝐸�𝜇𝜇
𝑁𝑁𝐸𝐸�𝜇𝜇 𝑁𝑁
� �^�^𝐸𝐸�𝑁𝑁𝑁𝑁
�
�^ ����
��� ���� 𝑁𝑁𝜇𝜇
𝑁𝑁𝜇𝜇
^��
�����
𝑁𝑁𝜇𝜇��� 𝜏𝜏���𝜏𝜏� ^�� � � 𝑁𝑁𝜇𝜇 �
𝜏𝜏
�²
𝑁𝑁𝐸𝐸�𝜇𝜇 𝜏𝜏
^�� � 𝑁𝑁𝜇𝜇^��
�� �� �� 𝑉𝑉�𝜏𝜏
���
�����^
� �����𝜏𝜏^� and���
∑ ����� � � � � ∑
��� �� 𝑉𝑉 �𝜏𝜏
� ^
��� �� �
also 𝐸𝐸�𝜏𝜏�^�� � � 𝐸𝐸�𝑁𝑁𝑁𝑁 then
�
�� � 𝑉𝑉�𝜏𝜏 �� ^
���
��
�� �
� �
��
� � ∑
^������ 𝑁𝑁𝐸𝐸�𝜇𝜇
���
�� ���
�
�²� ^
� 𝑁𝑁
� ��
��� �
�� �
�� �
� 𝑁𝑁𝜇𝜇 ���𝜏𝜏
��
𝑉𝑉�𝜏𝜏^ And � �^the ∑��variance ��𝜏𝜏^� � �𝜏𝜏 is:
then � ^�^ ∑ �� � �� �� ������ �² ��� ��
��𝑉𝑉�𝜏𝜏
� ����� ∑ �𝑉𝑉
∑ 𝑉𝑉�𝜏𝜏 ^ 𝑉𝑉�𝜏𝜏
�^ � ^��∑ �then 𝑉𝑉�𝜏𝜏𝑉𝑉
�then ^��𝑉𝑉�𝜏𝜏
�𝜏𝜏 ���
� �����
then �∑∑ �𝑁𝑁 � ^𝑁𝑁
𝑉𝑉�𝜏𝜏 ��� � ��� ∑���� � � 𝑁𝑁��² �������������� ��²��
𝑉𝑉�𝜏𝜏 ^���� ��� 𝑉𝑉�� ��� 𝑉𝑉�𝜏𝜏 ^� ��� 𝑁𝑁 �� ��
��
� �� ������� �� ������ �
� �²
� �
�² �² �
𝑉𝑉�𝜏𝜏
𝑉𝑉�𝜏𝜏 � ���
^���^��
^�� ∑∑�∑� ���
𝑉𝑉𝑉𝑉 �𝜏𝜏
�𝜏𝜏^^���^ ��^��then
�then ∑𝑉𝑉�𝜏𝜏
��
��
^���𝑉𝑉 ��
� ���
∑∑����� 𝑁𝑁���𝑁𝑁 ���� �^��� �� � ��
��������∑
� � � ����� ����� �²�
��� 65
^�� � � � 𝑉𝑉�𝜏𝜏 ^��
�� � 𝑁𝑁
𝑉𝑉�𝜏𝜏 � ∑��� 𝑉𝑉𝑉𝑉�𝜏𝜏 �𝜏𝜏
𝑉𝑉�𝜏𝜏 �then
then
�𝑉𝑉�𝜏𝜏
� �𝜏𝜏
^��^�� �^� ∑�then
𝑁𝑁��𝑉𝑉�𝜏𝜏
�
�������� ���
��² ���� � � 𝑁𝑁� � �� � 6565
𝑉𝑉�𝜏𝜏 ^ � �� 𝑉𝑉�𝜏𝜏 �∑ �� �
������ ��� ������
��� ��� �� ����� �� ��
65 65 65
65
6565
65 65
65
 45

5. Stakeholder engagement

Recreational fisheries stakeholders include all parties with an interest in the development
of sustainable recreational fisheries. The term “stakeholder” is most often employed to
refer to the recreational fishers themselves, including the federations and associations
of recreational fishers and charters (e.g. Federazione Italiana Pesca Sportiva e Attività
Subacquee in Italy, Federación Española de Pesca y Casting in Spain, etc.). However,
the term “stakeholder” can also include the public authorities at both the local and
national levels (e.g. port authorities and ministries in charge of fisheries management,
respectively), environmental associations, non-governmental organizations and research
institutes. This list is by no means exhaustive and other organisms/stakeholders, such
as other users of the aquatic resources and representatives from secondary industries
(e.g. the gear and tourism industries) could be included (Gaudin and De Young, 2007).
In this context, the relevant advisory councils in European Union countries (European
Commission, 2013) that also work on recreational fishery issues (e.g. MEDAC for the
Mediterranean) play an important role, since their opinion includes mediation efforts
with recreational fishers and other fisheries sectors sharing and exploiting the same
fishing resources.
Engaging stakeholders is vital for delivering a successful survey and, ultimately,
for the sustainable management of recreational fisheries. When properly achieved,
stakeholder engagement can help develop credibility and trust between researchers,
decision-makers and fishers. This trust is essential to ensuring robust participation in
studies, facilitating accurate data reporting, building a healthy platform for decision-
making discussions and securing buy-in for eventual management measures. The
overall objective of stakeholder engagement should be to close the gap between
decision-making and practice.
Stakeholders can be engaged at all stages of the survey process. During the planning
and development of the survey, the views of the recreational fishing community should
be considered, as they know far more about recreational fisheries than most scientists,
while scientists know much more about scientific methods than the recreational
fishing community. By involving stakeholders in the planning of surveys, clear
communication can be established regarding survey objectives and how surveys are
designed to produce reliable results, helping to develop credibility and trust. During
the data collection phase of the survey, stakeholder engagement is even more crucial.
Stakeholder engagement could be promoted by means of panels for data collection,
reference groups and committees, distribution of leaflets (via mail, websites or
meetings), websites, journals/newspapers and other media (ICES, 2011). It is important
to engage stakeholders as early as possible in data collection and monitoring initiatives
in order to build trust through open discussions and transparent processes.
Working together leads to all parties’ experiences and knowledge being incorporated
in the design and implementation of recreational fisheries surveys. This involvement
enhances the quality of the data collected, leading to greater utility for scientists
and the recreational fishing community alike (ICES, 2012). Finally, efforts should
be made to ensure survey results are reported back to stakeholders at the end of
the survey. Communicating results empowers stakeholders to actively participate in
management and decision-making processes. Recreational fisheries clubs, federations
and associations can be particularly useful partners in this regard. In this way, the data
collected is useful not only for public authorities, but also for angling organizations
that may wish to develop their own policies and regulations (ICES, 2012).
46 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

There are many successful examples of such stakeholder engagement in the

context of recreational fisheries. These include the Marine Recreational Information
Program of the United States, which applied new communication methods in order
to re-establish trust in their recreational fishery estimates. The programme achieved
this success by providing fact sheets, videos and background information on a website
(NOAA, 2021). In this case, a communications team was established to provide expert
advice in order to effectively communicate with stakeholders. To improve messaging,
videos were chosen as a new communication method. Similarly, the experience from
co-management committees (e.g. the case of the Roses Bay in Spain’s Catalonia region)
has showed that when recreational fishers were included in fisheries co-management
committees, providing them with a forum to share their perspectives and engage in
decision-making, fishers were surprisingly willing to self-regulate.
 47

References

Arlinghaus, R., Bork, M. & Fladung, E. 2008. Understanding the heterogeneity of

recreational anglers across an urban-rural gradient in a metropolitan area (Berlin,
Germany), with implications for fisheries management. Fisheries Research, 92(1): 53–62.
Arlinghaus, R., Tillner, R. & Bork, M. 2014. Explaining participation rates in recreational
fishing across industrialised countries. Fisheries Management and Ecology, 22(1): 45–55.
Arlinghaus, R., Abbott, J.K., Fenichel, E.P., Carpenter, S.R., Hunt, L.M., Alós, J.,
Klefoth, T., et al. 2019. Opinion: Governing the recreational dimension of global fisheries.
Proceedings of the National Academy of Sciences of the United States of America,
116(12): 5209–5213.
Ayal, S., Gino, F., Barkan, R. & Ariely, D. 2015. Three principles to REVISE people’s
unethical behavior. Perspectives on Psychological Science, 10(6): 738–741.
Beckmann, C., Tracey, S., Murphy, J., Moore, A., Cleary, B. & Steer, M. 2019. Assessing
new technologies and techniques that could improve the cost-effectiveness and robustness
of recreational fishing surveys. Proceedings of the national workshop, Adelaide, South
Australia, 10-12 July 2018. Adelaide, South Australia, South Australian Research and
Development Institute.
Bellanger, M. & Levrel, H. 2017. A cost-effectiveness analysis of alternative survey
methods used for the monitoring of marine recreational fishing in France. Ocean &
Coastal Management, 138: 19–28.
Cochran, W.G. 1977. Sampling techniques. Third edition. United States, John Wiley &
Sons.
Connelly, N.A. & Brown, T.L. 1995. Use of angler diaries to examine biases associated
with 12-month recall on mail questionnaires. Transactions of the American Fisheries
Society, 124(3): 413–422.
Connelly, N.A. & Brown, T.L. 2011. Effect of recall period on annual freshwater fishing
effort estimates in New York. Fisheries Management and Ecology, 18: 83–87.
Coutin, P., Conron, S.D. & MacDonald, M. 1995. The daytime recreational fishery in Port
Phillip Bay, 1989-94. Queenscliff, Australia, Victorian Fisheries Research Institute.
Dauk, P.C. 2000. Estimation in creel surveys under non-standard conditions. Department of
Mathematics and Statistics, Simon Fraser University. (PhD dissertation).
Dauk, P.C. & Schwarz, C.J. 2001. Catch estimation with restricted randomization in the
effort survey. Biometrics, 57(2):461–468.
Desfosses, C., Adams, P., Blight, S., Smallwood, C. & Taylor, S. 2019. The feasibility of
using remotely piloted aircraft systems (RPAS) for recreational fishing surveys in Western
Australia. Fisheries Occasional Publication No. 137. Western Australia, Department of
Primary Industries and Regional Development.
Dillman, D.A., Smyth, J.D. & Christian, L.M. 2014. Internet, phone, mail, and mixed
mode surveys: the tailored design method. Fourth edition. Hoboken, New Jersey, John
Wiley & Sons.
EAA. 2004. Recreational angling. Definition. A definition on recreational angling agreed
by the European Anglers Alliance at the General Assembly 2004 in Dinant, Belgium.
Document presented at the tenth General Assembly of the European Anglers Alliance,
25–28 March 2004, Dinant, Belgium, European Anglers Alliance.
Efron, B. & Tibshirani, R.J. 1993. An Introduction to the Bootstrap. London, Chapman
& Hall.
48 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

English, K.K., Shardlow, T.F & Webb, T.M. 1986. Assessment of Strait of Georgia sport
fishing statistics, sport fishing regulations and trends in Chinook catch using creel survey
data. Canadian Technical Report of Fisheries and Aquatic Sciences 1375. Nanaimo,
British Columbia.
English, K.K., Searing, G.F. & Nagtegaal, D.A. 2002. Review of the Strait of Georgia
recreational creel survey, 1983-1999. Canadian Technical Report of Fisheries and Aquatic
Sciences 2414. Nanaimo, British Columbia.
European Commission. 2013. Regulation (EU) No 1380/2013 of the European Parliament
and of the Council of 11 December 2013 on the Common Fisheries Policy, amending
Council Regulations (EC) No 1954/2003 and (EC) No 1224/2009 and repealing Council
Regulations (EC) No. 2371/2002. Official Journal of the European Union L354: 22–61.
European Commission. 2016. Commission Implementing Decision (EU) 2016/1251 of
12 July 2016 adopting a multiannual Union programme for the collection, management
and use of data in the fisheries and aquaculture sectors for the period 2017-2019 (notified
under document C(2016) 4329). Official Journal of the European Union, L207: 116–177.
FAO. 1966. Manual of sampling and statistical methods for fisheries biology. Part 1.
Sampling methods. FAO Manuals in Fisheries Science No. 3. Rome, FAO. (also available
at http://www.fao.org/3/X5684E/x5684e00.htm).
FAO. 2002. Sample-based fishery surveys: A technical handbook. FAO Fisheries Technical
Paper No. 425. Rome, FAO. 144 pp. (also available at http://www.fao.org/3/y2790e/
y2790e.pdf).
FAO. 2012. Recreational fisheries. FAO technical guidelines for responsible fisheries No.
13. Rome, FAO. 176 pp. (also available at http://www.fao.org/3/i2708e/i2708e00.pdf)
FAO. 2016. International Standard Statistical Classification of Fishing Gear [online].
Coordinating Working Party on Fishery Statistics (CWP). [Cited 22 April 2021]. http://
www.fao.org/3/bt988e/bt988e.pdf
FAO. 2018a. General Fisheries Commission for the Mediterranean. Report of the twentieth
session of the Scientific Advisory Committee on Fisheries, Tangiers, Morocco, 26-29 June
2018/Rapport de la vingtième session du Comité scientifique consultatif des pêches,
Tanger,Maroc, 26-29 juin 2018. FAO Fisheries and Aquaculture Report No. R1245.
Rome, FAO. 225 pp. (also available at http://www.fao.org/gfcm/statutory-meetings/
detail/en/c/1158491/)
FAO. 2018b. The State of Mediterranean and Black Sea Fisheries 2018. General Fisheries
Commission for the Mediterranean. Rome. http://www.fao.org/3/ca2702en/CA2702EN.pdf
FAO. 2020. The State of Mediterranean and Black Sea Fisheries 2020. General Fisheries
Commission for the Mediterranean. Rome. https://doi.org/10.4060/cb2429en
Ferter, K., Weltersbach, M.S., Strehlow, H.V., Vølstad, J.H., Alós, J., Arlinghaus, R.,
Armstrong, M., et al. 2013. Unexpectedly high catch-and-release rates in European
marine recreational fisheries: implications for science and management. ICES Journal of
Marine Science, 70(7): 1319–1329.
Fox, G.A., Negrete-Yankelevich, S. & Sosa, V.J., eds. 2015. Ecological statistics:
contemporary theory and application. Oxford, United Kingdom, Oxford University
Press.
Gaudin, C. & De Young, C. 2007. Recreational fisheries in the Mediterranean countries: a
review of existing legal frameworks. GFCM Studies and Reviews No. 81. Rome, FAO.
85 pp. (also available at http://www.fao.org/3/a1500e/a1500e00.htm)
GFCM. 2010a. Report of the Transversal Workshop on the Monitoring of Recreational
Fisheries in the GFCM Area. Palma de Mallorca, Spain, 20–22 October 2010 [online].
[Cited 3 February 2021]. http://gfcmsitestorage.blob.core.windows.net/documents/
web/SAC/13/GFCM_SAC13_2011_Inf.18-e.pdf
GFCM. 2010b. Draft report of the eleventh session of the SAC Sub-Committee on economic
and social sciences (SCESS). Saint George’s Bay, Malta, 29 November–2 December
2010. [online]. [Cited 3 February 2021]. http://gfcmsitestorage.blob.core.windows.net/
documents/web/SAC/13/GFCM_SAC13_2011_Inf.7-e.pdf
References 49

GFCM. 2017a. Mid-term strategy (2017–2020) towards the sustainability of Mediterranean

and Black Sea fisheries. Rome, FAO. 24 pp. (also available at http://www.fao.org/3/
i7340e/i7340e.pdf)
GFCM. 2017b. Report of the first meeting of the Working Group on Small-Scale and
Recreational fisheries (WGSSF). FAO headquarters, Rome, Italy, 12–13 September 2017.
[online]. [Cited 3 February 2021]. http://www.fao.org/gfcm/technical-meetings/detail/
en/c/1061844/
GFCM. 2018. GFCM Data Collection Reference Framework (DCRF). Version: 20.1.
[online]. Data Collection Reference Framework [Cited 7 December 2020]. http://www.
fao.org/gfcm/data/dcrf/
GFCM. 2019. Compendium of GFCM decisions [online]. [Cited 4 March 2021]. www.fao.
org/gfcm/decisions/en.
GFCM. 2021. Report of the first meeting of the Working Group on Recreational Fisheries
(WGRF). Online, 25–26 February 2021. [online]. [Cited 21 April 2021]. http://www.fao.
org/gfcm/technical-meetings/detail/fr/c/1402637/
Gordoa, A., Dedeu, A.L. & Boada, J. 2019. Recreational fishing in Spain: first national
estimates of fisher population size, fishing activity and fisher social profile. Fisheries
Research, 211: 1–12.
Grafton, R.Q., Kirkley, J., Kompas, T. & Squires, D. 2006. Economics for fisheries
management. Hampshire, UK, Ashgate Publishing.
Groves, R.M. & Lyberg, L. 2010. Total survey error: past, present, and future. Public
opinion quarterly, 74(5): 849–879.
Groves, R.M, Fowler, F.J., Couper, M.P., Lepkowski, J.M., Singer, E. & Tourangeau, R.
2009. Survey Methodology. Hoboken, New Jersey, Wiley.
Hankin, D., Mohr, M.S. & Newman, K.B. 2019. Sampling theory: for the ecological and
natural resource sciences. United Kingdom, Oxford University Press.
Hartill, B.W., Watson, T.G. & Bian, R. 2011. Refining and applying a maximum-count
aerial-access survey design to estimate the harvest taken from New Zealand’s largest
recreational fishery. North American Journal of Fisheries Management, 31(6):1197–1210.
Hartill, B.W., Cryer, M., Lyle, J.M., Rees, E.B., Ryan, K.L., Steffe, A.S., Taylor, S.M.,
West, L. & Wise, B. 2012. Scale- and context-dependent selection of recreational harvest
estimation methods: the Australasian experience. North American Journal of Fisheries
Management, 32(1): 109–123.
Hoyle, S.D. & Cameron, D.S. 2003. Confidence intervals on catch estimates from a
recreational fishing survey: a comparison of bootstrap methods. Fisheries Management
and Ecology, 10:97–108.
Hyder, K., Armstrong, M., Ferter, K. & Strehlow, H.V. 2014. Recreational sea fishing –
the high value forgotten catch. ICES Insight, 51:8–15.
Hyder, K., Radford, Z., Prellezo, R., Weltersbach, M.S., Lewin, W.C., Zarauz, L.,
Ferter, K., et al. 2017a. Research for PECH Committee – Marine recreational and semi-
subsistence fishing – its value and its impact on fish stocks. Brussels, European Parliament,
Policy Department for Structural and Cohesion Policies.
Hyder, K., Weltersbach, M.S., Armstrong M., Ferter, K., Townhill, B., Ahvonen, A.,
Arlinghaus, R., et al. 2017b. Recreational sea fishing in Europe in a global context –
participation rates, fishing effort, expenditure, and implications for monitoring and
assessment. Fish and Fisheries, 19: 225–243.
ICCAT. 2016. ICCAT Manual. In: ICCAT [online]. Madrid, Spain. [Cited 2 February
2021]. https://www.iccat.int/en/iccatmanual.html
ICES. 2010. Report of the Planning Group on Recreational Fisheries Surveys (PGRFS),
7–11 June 2010, Bergen, Norway. [online]. [Cited 3 February 2021]. http://ices.dk/
sites/pub/Publication%20Reports/Expert%20Group%20Report/acom/2010/PGRFS/
PGRFS%202010.pdf
50 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

ICES. 2011. Report of the Planning Group on Recreational Fisheries Surveys (PGRFS), 2–6
May 2011, Esporles, Spain. [online]. [Cited 3 February 2021]. http://www.ices.dk/sites/
pub/Publication%20Reports/Expert%20Group%20Report/acom/2011/PGRFS%20
2011/PGRFS%202011.pdf
ICES. 2012. Report of the Working Group on Recreational Fisheries Surveys (WGRFS),
7–11 May 2012, Esporles, Spain. [online]. [Cited 3 February 2021]. http://www.ices.dk/
sites/pub/Publication%20Reports/Expert%20Group%20Report/acom/2012/WGRFS/
WGRFS%202012.pdf
ICES. 2013. Report of the ICES Working Group on Recreational Fisheries Surveys
2013 (WGRFS), 22–26 April 2013, Esporles, Spain. [online] [Cited 5 February 2021].
http://www.ices.dk/sites/pub/Publication%20Reports/Expert%20Group%20Report/
acom/2013/WGRFS%202013/WGRFS%20Report%202013.pdf
ICES. 2014. Report of the Working Group on Recreational Fisheries Surveys (WGRFS),
2–6 June 2014, Sukarrieta, Spain. [online] [Cited 5 February 2021]. http://www.ices.dk/
sites/pub/Publication%20Reports/Expert%20Group%20Report/acom/2014/WGRFS/
WGRFS%20Final%20Report%202014.pdf
Johnston, F.D., Arlinghaus, R. & Dieckmann, U. 2010. Diversity and complexity of
angler behaviour drive socially optimal input and output regulations in a bioeconomic
recreational-fisheries model. Canadian Journal of Fisheries and Aquatic Sciences, 67(9):
1507–1531.
Jones, C.M., Robson, D.S., Otis, D. & Gloss, S. 1990. Use of a computer simulation
model to determine the behaviour of a new survey estimator of recreational angling.
Transactions of the American Fisheries Society, 119(1):41–54.
Krumpal, I. 2013. Determinants of social desirability bias in sensitive surveys: a literature
review. Quality & Quantity, 47(4): 2025–2047.
Levine, D.M, Stephan, D.F., Krehbiel, T.C. & Berenson, M.L. 2008. Statistics for
managers: using Microsoft Excel. Fifth Edition. Upper Saddle River, New Jersey, Prentice
Hall, Pearson Education, Inc.
Link, M.W., Battaglia, M.P., Frankel, M.R., Osborn, L. & Mokdad, A.H. 2008. A
comparison of address-based sampling (ABS) versus random-digit dialing (RDD) for
general population surveys. Public Opinion Quarterly, 72(1): 6–27.
Lockwood, R.N., Peck, J. & Oelfke, J. 2001. Survey of angling in Lake Superior waters
at Isle Royale National Park, 1998. North American Journal of Fisheries Management,
21(3):471–481.
Lohr, S.L. 1999. Sampling: design and analysis. Second edition. Boston, USA, Brooks Cole.
MEDAC. 2016. Recreational fisheries in the Mediterranean. MEDAC Advice for a
regulatory framework and efficient management for recreational fisheries in the
Mediterranean based on “FAO Technical Guidelines on Responsible Recreational
Fisheries”. Document ref. 155/2016 presented at the Executive Committee, 20 April
2016, Split, Croatia, Mediterranean Advisory Council.
NOAA. 2021. Recreational Fishing Data. In: NOAA [online]. Washington, DC. [Cited 22
April 2021]. https://www.fisheries.noaa.gov/topic/recreational-fishing-data
Parker, N.A. 1956. Discussion. In K. D. Carlander, eds. Proceedings of Iowa state creel
survey symposium, pp. 59–62. Ames, United States, Iowa Cooperative Fisheries Unit.
Pawson, M.G., Glenn, H. & Padda, G. 2008. The definition of marine recreational fishing
in Europe. Marine Policy, 32: 339–350.
Pinello, D., Gee, J. & Dimech, M. 2017. Handbook for fisheries socio-economic sample
survey – principles and practice. FAO Fisheries and Aquaculture Technical Paper No.
613. Rome, FAO. 136 pp. (also available at http://www.fao.org/3/i6970e/i6970e.pdf).
Pollock, K.H., Jones, C.M. & Brown, T.L. 1994. Angler survey methods and their
applications in fisheries management. Bethesda, Maryland, American Fisheries Society,
Special Publication No. 25.
References 51

Robson, D.S. & Jones, C.M. 1989. The theoretical basis of an access site angler survey
design. Biometrics, 45(1):83–98.
Sabatella, E. & Franquesa, R. 2003. Manual of fisheries sampling surveys: methodologies
for estimations of socio-economic indicators in the Mediterranean Sea. Studies and
Reviews. General Fisheries Commission for the Mediterranean No. 73. Rome, FAO. 37
pp. (also available at http://www.fao.org/3/y5228e/y5228e00.htm#Contents).
Salant, P. & Dillman, D.A. 1994. How to conduct your own survey. Hoboken, Wiley &
Sons.
Sbragaglia, V., Correia, R.A., Coco, S. & Arlinghaus, R. 2019. Data mining on YouTube
reveals fisher group-specific harvesting patterns and social engagement in recreational
anglers and spearfishers. ICES Journal of Marine Science, 77(6): 2234–2244.
Soupir, C.A., Brown, M.L., Stone, C.C. & Lott, J.P. 2006. Comparison of creel survey
methods on Missouri River reservoirs. North American Journal of Fisheries Management,
26(2):338–350.
Sparrevohn, C.R. & Storr-Paulsen, M. 2012. Using interview-based recall surveys to
estimate cod Gadus morhua and eel Anguilla anguilla harvest in Danish recreational
fishing. ICES Journal of Marine Science, 69(2): 323–330.
Tarrant, M.A. & Manfredo, M.J. 1993. Digit preference, recall bias, and nonresponse bias
in self reports of angling participation. Leisure Sciences, 15(3): 231–238.
Vaske, J.J. 2008. Survey research and analysis: applications in parks, recreation and human
dimensions. State College, Pennsylvania, Venture Publishing.
Vaske, J.J. 2011. Advantages and disadvantages of internet surveys: introduction to the
special issue. Human Dimensions of Wildlife, 16(3): 149–153.
Vaske, J.J., Huan, T.C. & Beaman, J. 2003. The use of multiples in anglers’ recall of
participation and harvest estimates: some results and implications. Leisure Sciences, 25(4):
399–409.
Venturelli, P.A., Hyder, K. & Skov, C. 2017. Angler apps as a source of recreational
fisheries data: opportunities, challenges and proposed standards. Fish and Fisheries, 18(3):
578–595.
Zischke, M.T. & Griffiths, S.P. 2014. Time-location sampling with capture-recapture to
assess specialised recreational fisheries. Fisheries research, 157: 136–146.
Wynne-Jones, J., Gray, A., Hill, L. & Heinemann, A. 2014. National panel survey
of marine recreational fishers 2011–12: harvest estimates. New Zealand Fisheries
Assessment Report 2014/67. [online]. [Cited 3 February 2021]. https://www.mpi.govt.nz/
dmsdocument/4719/direct
 53

Annexes

ANNEX 1. GFCM AREA OF APPLICATION, SUBREGIONS AND

GEOGRAPHICAL SUBAREAS
FIGURE 1. GFCM area of application, subregions and geographical subareas

GFCM subregions
Western Mediterranean 30
Central Mediterranean
Adriatic Sea
Eastern Mediterranean
17 29
Black Sea 8
7
9
6 18
28
11.1 10 22
5 11.2
12
4
16 19
1
2 13 20
3
15
24
23 25
14
21 27
26

Source: FAO, 2020.

FAO statistical divisions GFCM geographical subareas (GSAs) Source: GFCM, 2018.

01. Northern Alboran Sea 07. Gulf of Lion 13. Gulf of Hammamet 19. Western Ionian Sea 25. Cyprus

02. Alboran island 08. Corsica 14. Gulf of Gabès 20. Eastern Ionian Sea 26. South Levant

09. Ligurian Sea and northern

03. Southern Alboran Sea 15. Malta 21. Southern Ionian Sea 27. Eastern Levant Sea
Tyrrhenian Sea
10. South and central
04. Algeria 16. South of Sicily 22. Aegean Sea 28. Marmara Sea
Tyrrhenian Sea
11.1. Sardinia (west)
05. Balearic islands 17. Northern Adriatic Sea 23. Crete 29. Black Sea
11.2. Sardinia (east)

06. Northern Spain 12. Northern Tunisia 18. Southern Adriatic Sea 24. North Levant Sea 30. Azov Sea
54 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

ANNEX 2. CODES FOR RECREATIONAL FISHING TECHNIQUES

Main recreational fishing gear and codes

Gear name Code Notes

Wrenching gear, clamps, tongs, rakes,

Hand implements MHI
spears
Harpoons HAR Knife, harpoon
Diving MDV
Diving (speargun) MDS*
Diving (hand) MDH*
Cast nets FCN
Boat seines SV
Beach seines SB
Hooks and lines (not specified) LX
Handlines and hand-operated pole and lines LHP Rod, handline
Traps (not specified) FIX
Pots FPO
Gillnets and entangling nets (nei) GEN
Gillnets GNS
Trammel nets GTR
Longlines (not specified) LL
Lift nets (not specified) LN
Scoop nets MSP
Gear not known NK
Gear not specified MIS
Adapted from FAO, 2016.
Annexes 55

ANNEX 3. TEMPLATE FOR SCREENING SURVEY AND ENROLLMENT OF

FISHERS IN DATA COLLECTION PANEL

Template for screening survey and enrollment of fishers in data collection panel

Phone number __________________________ City ________________________________

Number of members of the household ______________________________________

Marine Number of fishing trips in the Panel

Age Gender recreational previous year
Willingness to If YES. Contact
fisher? Boat Shore Underwater participate? info (phone/email)

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No

Yes No Yes No
56 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

ANNEX 4. TEMPLATES FOR MANDATORY FEE-FREE ONLINE REGISTRATION

OF MARINE RECREATIONAL FISHERS

Template for online registration

STEP 1 – general information

Name _________________________________________

Surname _________________________________________

Date of birth _________________________________________

Place of birth _________________________________________

Nationality _________________________________________

Address _________________________________________

E-mail _________________________________________

Telephone No. _________________________________________

Gender
Annexes 57

Template for online registration

STEP 2 – avidity
ID No. ____________________
How many fishing trips did you perform
Fishing mode Gear last year?
Boat Hand implements 1-5 6-10 11-25 26-50 >50
Harpoons 1-5 6-10 11-25 26-50 >50
Diving 1-5 6-10 11-25 26-50 >50
Diving (speargun) 1-5 6-10 11-25 26-50 >50
Diving (hand) 1-5 6-10 11-25 26-50 >50
Cast nets 1-5 6-10 11-25 26-50 >50
Boat seines 1-5 6-10 11-25 26-50 >50
Hooks and lines (not specified) 1-5 6-10 11-25 26-50 >50
Handlines and hand-operated pole and lines 1-5 6-10 11-25 26-50 >50
Traps (not specified) 1-5 6-10 11-25 26-50 >50
Pots 1-5 6-10 11-25 26-50 >50
Gillnets and entangling nets (nei) 1-5 6-10 11-25 26-50 >50
Gillnets 1-5 6-10 11-25 26-50 >50
Trammel nets 1-5 6-10 11-25 26-50 >50
Longlines (not specified) 1-5 6-10 11-25 26-50 >50
Lift nets (not specified) 1-5 6-10 11-25 26-50 >50
Scoop nets 1-5 6-10 11-25 26-50 >50
Gear not known 1-5 6-10 11-25 26-50 >50
Gear not specified 1-5 6-10 11-25 26-50 >50
Shore Hand implements 1-5 6-10 11-25 26-50 >50
Harpoons 1-5 6-10 11-25 26-50 >50
Diving 1-5 6-10 11-25 26-50 >50
Diving (speargun) 1-5 6-10 11-25 26-50 >50
Diving (hand) 1-5 6-10 11-25 26-50 >50
Cast nets 1-5 6-10 11-25 26-50 >50
Beach seines 1-5 6-10 11-25 26-50 >50
Hooks and lines (not specified) 1-5 6-10 11-25 26-50 >50
Handlines and hand-operated pole and lines 1-5 6-10 11-25 26-50 >50
Traps (not specified) 1-5 6-10 11-25 26-50 >50
Pots 1-5 6-10 11-25 26-50 >50
Gillnets and entangling nets (nei) 1-5 6-10 11-25 26-50 >50
Gillnets 1-5 6-10 11-25 26-50 >50
Trammel nets 1-5 6-10 11-25 26-50 >50
Longlines (not specified) 1-5 6-10 11-25 26-50 >50
Lift nets (not specified) 1-5 6-10 11-25 26-50 >50
Scoop nets 1-5 6-10 11-25 26-50 >50
Gear not known 1-5 6-10 11-25 26-50 >50
Gear not specified 1-5 6-10 11-25 26-50 >50
Underwater Hand implements 1-5 6-10 11-25 26-50 >50
Harpoons 1-5 6-10 11-25 26-50 >50
Diving 1-5 6-10 11-25 26-50 >50
Diving (speargun) 1-5 6-10 11-25 26-50 >50
Diving (hand) 1-5 6-10 11-25 26-50 >50
Gear not known 1-5 6-10 11-25 26-50 >50
Gear not specified 1-5 6-10 11-25 26-50 >50
58 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

Template for online registration

STEP 3 – certificate
Name ___________________________________

Surname ___________________________________

Nationality ___________________________________

Address ___________________________________

ID No. ___________________________________

Issuing date ___________________________________

Expiration date ___________________________________

Annexes 59

ANNEX 5. TEMPLATE FOR LOGBOOK AND/OR RECALL SURVEY

5.a. General information for logbook and/or recall survey

Logbook and recall template – general information

Logbook Recall Reference month and year ______________________
Name and surname of panel participant _______________________________________

Information reported for:

Multiple fishers (in case the panel participant’s catch is
Only the panel participant
pooled with other fishers on the same trip)
If multiple fishers:
_____ age
No. Fishers:
_____ age

Fishing location Total fishing time

Geographical subarea (GSA) ___________ Start End

City ___________ Date

Hour
Distance from the coast (in nm) ___________

Fishing mode* Boat Shore Underwater

Gear Gear code Fishing time per gear (in hours) Number of units used per gear
Hand implements MHI
Harpoons HAR
Diving MDV
Diving (speargun) MDS
Diving (hand) MDH
Cast nets FCN
Boat seines SV
Beach seines SB
Hooks and lines (not specified) LX
Handlines and hand-operated pole and lines LHP
Traps (not specified) FIX
Pots FPO
Gillnets and entangling nets (nei) GEN
Gillnets GNS
Trammel nets GTR
Longlines (not specified) LL
Lift nets (not specified) LN
Scoop nets MSP
Gear not known NK
Gear not specified MIS

Catches
No. Gear code Species No. Retained Weight (kg) retained No. released
1
2
3
4
5
6
7
8
9
10

Comments:

* Complete one logbook/recall template per fishing mode

** Provide a description of the fishing gear in the comments section
60 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

5.b. Catch information for logbook and/or recall survey

Logbook and recall template – retained species information

Logbook Date ____________ Recall Reference month and year ______________

Fishing mode***
Gear Weight
No. Species (retained) Length* Sex** Under
code (kg) Boat Shore
water
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
* Total length for fish (cm), mantle length for cephalopods (cm), carapace length for crustaceans (mm)
** If known (M: male; F: female; ND: not determined)
*** Select only one fishing mode (boat, shore, underwater) per row
Annexes 61

5.c. Released species information for logbook and/or recall survey

Logbook and recall template – released species information

Logbook Date ____________ Recall Reference month and year ______________

Post-released status** Fishing mode***

Gear Species
No. Length* Under
code (released) Almost Not Boat Shore
Alive Dead water
dead known
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
* Total length for fish (cm), mantle length for cephalopods (cm), carapace length for crustaceans (mm)
** Mark the corresponding cell
*** Select only one fishing mode (boat, shore, underwater) per row
62 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

5.d. Expenditure information per fishing trip for logbook survey

Logbook template - expenditures

Expenditures per fishing trip
(Any expenditures since the last logbook date, such as purchases of new
equipment, should be reported)

Start date of fishing trip: __________________________________________

Type of expenditure Value Currency

Rods and reels

Nets (set nets, lift, scoop,
etc.)
Accessories (hooks, lines,
etc.)

Speargun

Underwater accessories (fins, mask, etc.)

Traps

Artificial baits (jigs, lures, etc.)

Natural baits (worms, sardines, etc.)

Travel and accommodation

License fee

Boat rental

Charter

Fuel

Taxes

Electronics (GPS, radar, etc.)

Boat maintenance

Others __________________________

Others __________________________

Others __________________________

Others __________________________

Comments:
Annexes 63

5.e. Expenditure information per month for recall survey

Recall template - expenditures

Expenditures per month
Reference month and year: _________________________________________
Type of expenditure Value Currency

Rods and reels

Nets (set nets, lift, scoop,
etc.)
Accessories (hooks, lines,
etc.)

Speargun

Underwater accessories (fins, mask, etc.)

Traps

Artificial baits (jigs, lures, etc.)

Natural baits (worms, sardines, etc.)

Travel and accommodation

License fee

Boat rental

Charter

Fuel

Taxes

Electronics (GPS, radar, etc.)

Boat maintenance

Others __________________________

Others __________________________

Others __________________________

Others __________________________

Comments:
64 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

ANNEX 6. TEMPLATE FOR ON-SITE SURVEYS

6.a. General information for on-site survey
On-site survey template – general information
Date of interview _______________________________________
Information by:
Single fisher Multiple fishers (in the case that catches are pooled)

_____ age
No. Fishers:
_____ age

Fishing location Total fishing time

Geographical subarea (GSA) _______________________
City ______________________________________________ Start End
Distance from the coast (in nm) __________________ Date
Latitude __________________________________________ Hour
Longitude ________________________________________

Fishing mode*
Gear Fishing time per Number of units
Gear Under-
code Boat Shore gear (in hours) used per gear
water
Hand implements MHI
Harpoons HAR
Diving MDV
Diving (speargun) MDS
Diving (hand) MDH
Cast nets FCN
Boat seines SV
Beach seines SB
Hooks and lines (not specified) LX
Handlines and hand-operated pole and lines LHP
Traps (not specified) FIX
Pots FPO
Gillnets and entangling nets (nei) GEN
Gillnets GNS
Trammel nets GTR
Longlines (not specified) LL
Lift nets (not specified) LN
Scoop nets MSP
Gear not known NK
Gear not specified MIS
Catches
No. Gear code Species No. retain Weight (kg) retain No. release
1
2
3
4
5
6
7
8
9
10

Number of fishing trips performed by the Boat

interviewee during the previous year:
Shore
Underwater

Willingness of interviewee to participate in a follow-up panel? If yes, name and (mobile) telephone number:

Comments:

* Complete one logbook/recall template per fishing mode

** Provide a description of the fishing gear in the comments section
Annexes 65

6.b. Catch information for on-site survey

On-site survey template – retained species information

Date ___________________________

Fishing mode***
Gear Weight
No. Species (retained) Length* Sex** Under
code (kg) Boat Shore
water
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
* Total length for fish (cm), mantle length for cephalopods (cm), carapace length for crustaceans (mm)
** If known (M: male; F: female; ND: not determined)
*** Select only one fishing mode (boat, shore, underwater) per row
66 Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

6.c. Released species information for on-site survey

On-site survey template – released species information

Date ________________

Post-released status** Fishing mode***

Gear Species
No. Length* Under
code (released) Almost Not Boat Shore
Alive Dead water
dead known
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
* Total length for fish (cm), mantle length for cephalopods (cm), carapace length for crustaceans (mm)
** Mark the corresponding cell
*** Select only one fishing mode (boat, shore, underwater) per row
Annexes 67

ANNEX 7. FISHING EFFORT MEASUREMENT

Effort measurement by fishing gear

Gear name Code Unit of capacity Unit of activity Nominal effort

Hand implements MHI Number Fishing days Number x fishing days

Harpoons HAR Number Fishing days Number x fishing days
Diving MDV Number Fishing days Number x fishing days
Diving (speargun) MDS** Number Fishing days Number x fishing days
Diving (hand) MDH** Number Fishing days Number x fishing days
Cast nets FCN Number Fishing days Number x fishing days
Beach seines SB Net length* Fishing days Net length x fishing days
Number of
Hooks and lines (not specified) LX
hooks Fishing days Number of hooks x fishing days
Handlines and hand-operated pole
LHP
and lines Number Fishing days Number x fishing days
Number of
Traps (not specified) FIX
traps Fishing days Number of traps x fishing days
Number of
Pots FPO
pots Fishing days Number of pots x fishing days
Gillnets and entangling nets (nei) GEN Net length* Fishing days Net length x fishing days
Gillnets GNS Net length* Fishing days Net length x fishing days
Trammel nets GTR Net length* Fishing days Net length x fishing days
Number of
Longlines (not specified) LL
hooks Fishing days Number of hooks x fishing days
Lift nets (not specified) LN Number Fishing days Number x fishing days
Scoop nets MSP Number Fishing days Number x fishing days
* Length of net expressed in 100-m units
** Adapted from FAO, 2016.
 Office of Communications – November 2020

Office of Communications – November 2020

Handbook for data collection on recreational fisheries in the Mediterranean and the Black Sea

Corrigendum
Updated on 18 October 2021

The following corrections were made to the PDF after it went to print.

Page Location Text in printed PDF Text in corrected PDF

Cover Logos The GFCM logo was added.

page
Title Authors Authors’affiliations were added.
page
60 Annex 5.b The table was updated.

61 Annex 5.c The table was updated.

66 Annex 6.c The table was updated.

Contact: [email protected]
Marine recreational fisheries are an integral part of Mediterranean and Black Sea
coastal life and are commonly practiced throughout the region. Recreational
fisheries also represent an important driver of coastal tourism, which constitutes
one of the region’s most important maritime sectors in terms of gross value added
and employment. However, despite their ubiquity and potential socio-economic
contribution, recreational fisheries are a data-poor sector and can vary widely
from one country to another, thus impairing proper consideration of the
recreational fisheries sector in policy-making and undermining efforts towards
sustainable fisheries management at the regional level. The main goal of this
handbook is therefore to provide a clear methodological framework to allow
Mediterranean and Black Sea countries to implement suitably harmonized
sampling and survey monitoring schemes for recreational fisheries. This handbook
establishes a minimum set of necessary information for monitoring recreational
fisheries, while, at the same time, allowing for flexibility to accommodate
national specificities and data collection needs. It also provides guidance on the
data analysis process as well as advice to successfully engage stakeholders in the
data collection process.

ISBN 978-92-5-134630-3 ISSN 2070-7010

9 789251 346303
CB5403EN/1/07.21