APPENDIX 1

A Statistical Procedure for Continuous Collection of Fishery Data Using a Restricted Number of Recorders

In this Appendix we discuss the operational and technical aspects of a simple survey system for the collection of catch and effort data on an ongoing basis. There are four main stages in the implementation of a Catch Assessment Survey (CAS):

1. Planning the survey
2. Designing the survey
3. Field operations for data collection
4. Processing and presentation of the results

Another stage which should be added up is the one dealing with the analysis and reporting on the estimates calculated.

1. Planning the Survey

At the planning stage of a CAS the questions that require consideration are:

(a)  The scope of the survey must be clearly specified, i.e., should the survey be aimed at one objective (by resources or fish consumption) or should it be multi-purpose in character aimed at satisfying the needs of a variety of users such as planners, economists, fishery scientists? If the latter, so, their individual requirements be taken into account to the extent possible within manpower and financial constraints.

(b)  The geographical coverage of the survey (traditional-artisanal fisheries, industrial fisheries) must be defined.

(c)  The hierarchy of units in the population must be defined. In the case of traditional and artisanal fisheries the following hierarchy of units can usually be identified:

(d)  Decision must be made on the nature of information to be collected, and an efficient coding system for species, vessels, gear, fishermen and effort units must be prepared.

(e)  The required source documents for data collection, i.e., lists, questionnaires, mapping material and other supporting documents must be prepared.

(f)  A decision must be made on the method of data collection, e.g., census survey, sample survey, and of the method of obtaining the information, i.e., by interview, by objective measurements made by the officer at landing places, etc.

(g)  A frame survey must be conducted and the appropriate sampling frame must be constructed providing information of the basic characteristics by sampling units.

2. Sampling Design for Small-Scale Fisheries

Catch Assessment Surveys of traditional and artisanal fisheries are based on sample surveys “sampling in space and time”. As far as space is concerned the decision to be taken is on the number of landing places to be selected within each island for data collection. As far as time is concerned, the decision to be taken is on the number of survey-days to be allocated to each sample landing place. It should be noted that CAS are “cost-oriented” surveys and that decisions concerning the sizes of the samples (space, time) should be made, taking into account available resources.

(a) Calculation of sample sizes

Assume that within an island transport is not a problem and that recorders can move from one landing place to another, and assume as a simple example, that there are 24 landing places on the island, of which 4 are the most important ones. A “working group” of two recorders is assigned for data collection on a monthly basis. Steps to be taken for the calculation of the sample sizes are:

(i) Calculation of the time available Z, which is given by:

Z = number of recorders (2) × working days per month (20) = 40 (recorder-days)

(ii) How many samples?

One reasonable approach is to use the census method in space (complete coverage) for the most important landing places (= 4) and the sample method in space for the remaining landing places (= 24 - 4 = 20). Also, it is reasonable to allocate more survey days to the most important landing places than to the remainder. For example, the choice might be six survey days to be allocated for each important landing place, and three survey days to be allocated for each landing place of secondary importance. In such a case, the sample size of secondary landing places is:

where:

40: total number of recorder-days per month
24: total number of recorder-days allocated to the important landing places
16: total number of recorder-days allocated to the secondary landing places
3: number of survey recorder-days allocated per secondary landing place.

The composition of the overall sample sizes are:

It should be noted that the above sampling scheme introduces a kind of stratification into the sampling process, i.e.:

Str. 1: large LP's (N₁ = 4)

Str. 2: other LP's (N₂ = 16)

Monthly estimates of the survey magnitudes are first calculated on a stratum basis. Total monthly estimates for the island as a whole are calculated by adding up the strata estimates. Total monthly estimates for the population as a whole, say, consisting of 3 islands, are calculated by adding up the monthly estimates of the individual islands.

(b) Sampling within the sample LP's

Within the sample landing places, items of information from the boat-landings are usually obtained on a sample basis. Specifically, a few boat-landings are selected from the total landings within a survey day for data collection. If there is more than one main fishery operating, a number of sample survey days are allocated to each of them. Steps to be taken:

Three distinct magnitudes:

For each sample day, data must be recorded on to the following three distinct magnitudes:

(i)  total number of fishing boats operating by type of fishery (N_f);

(ii)  total number of boat-landings of the fishery surveyed (M_f);

(iii) sample size of selected landings of the survey fishery (m_f).

The list below (List 1) portrays the format of the document needed for the collection of the required information on N_f, M_f and m_f, on a daily basis within the selected landing places. Specifically, in the preparation of List 1, the assumption is made that two fisheries are operating at the sample landing place which belongs to Stratum 1.

List 1

(c) Completed questionnaires

For each selected boat-landing a questionnaire will be completed providing the required items of information. A code number consisting of 5 digits identifies the completed questionnaire plus the survey date:

(d) Raising sample data to the population total

For each item of information included in the completed questionnaires, monthly estimates can be calculated by fisheries using the proper raising factors:

Str: 1 - In our example, there are four landing places in Str. 1. The census method in space (complete enumeration of landing places) was used for data collection. As far as time is concerned, three days were allocated to each fishery within a sample landing place for obtaining the required information. The following steps should be taken for obtaining Stratum monthly estimates for a given characteristic, say total fish catch (variable - y) and for the given fishery (fishery - f).

Steps: Estimated monthly totals within each sample landing place

(i)  pool the completed questionnaires of the three sample days together;

(ii) add up the values of the characteristic total fish catch, which have been recorded in the individual questionnaires, say, sample total value - A;

(iii) multiply A by the following ratio:

Note: the required data for the calculation of R₁ value are given in List 1.

(iv) multiply B by the time-raising factor

i.e., C = R₂B

Magnitude C expresses total monthly catches within a given sample landing place and for the given fishery.

Monthly stratum estimates

Monthly stratum estimates are simply calculated by adding up the calculated monthly estimates of the individual landing places (remember, in Stratum 1 our survey system is based on the census method, i.e., a complete coverage of all landing places) i.e.:

1^C = C₁ + C₂ + C₃ + C₄

where: 1^C = total monthly catches in Str. 1.

Str. 2 - For the estimation of monthly stratum totals in Stratum 2 the following steps should be taken:

Estimate monthly totals within each sample landing place

Follow exactly the same operational steps (i) to (iv) described above, i.e., Stratum 1.

Monthly stratum estimates - Steps:

(i)  add up the monthly estimates which were calculated for each sample landing place:

2^C = C₁ + C₂ + C₃ + C₄ + C₅

(ii)  multiply ₂C by the stratum-raising factor, R₃ = total number of operating fishing boats in the stratum of the given fishery (frame data) divided by the total number of operating fishing boats in the sample landing places, i.e.:

₂E = R₃₂C

Where: ₂E = total monthly catches in Str. 2.

Overall monthly estimates for the island as a whole are calculated by adding the strata totals:

Y = ₁C + ₂E

Yearly estimates are calculated by adding up the monthly estimates.

If the survey population consists of more than one island, monthly estimates are first calculated on an island basis. Overall monthly estimates for the population as a whole are calculated by adding up the individual estimates.

(e) Estimations

Note: This section should be ommitted by the readers not having a mathematical background.

The operational steps described above for the calculation of estimates of the survey magnitudes can be presented in a concise form by using the proper symbols and formulae:

Suffixes

h: a given, island in the survey population (h = 1, 2, ..)
l: the strata of landing places
i: the i-th landing place (i = 1, 2, ..)
f: a given fishery (f = 1, 2, ..)
j: a given boat-landing (j = 1, 2, ..)
d: days (d = 1, 2, ..)
∑(d): survey days

Symbols

_hN₁: total number of landing places within the 1-th stratum in the h-th island
_hn₁: selected landing places within the 1-th stratum in the h-th island
h^Wlf: total number of fishing boats of f-fishery, in the l-th stratum in the h-th island
h^wlfi: total number of fishing boats of f-fishery in the i-th landing place, etc.
h^Mlfid: total landings on the d-th sample day, etc.
h^mlfid: selected landings on the d-th sample day, etc.
D: total fishing days in a given month
∑: scummator

Estimators (= formulae) - monthly estimates

(a) Estimated monthly total, say, catches in Stratum-1 = 1, in the h-th island:

(b) Estimated monthly total, say, catches in Stratum 1 = 2, in the h-th island:

(c) Estimated monthly total, say, catches for the island as a whole:

(d) Estimated monthly total, say, catches for the population as a whole:

(e) Yearly totals are estimated by adding up the monthly estimates.

3. Field Operations for Data Collection

The main task of the field recorders is the collection of the required information by following the instructions and procedures spelt out in the design of the survey. They keep to a detailed “timetable” of their activities and of the time allocated to each of them.

The landings of the sample units surveyed are sorted out by species or species groups and their weight is taken and recorded. Data fishing effort is obtained by inquiry, and expressed in units appropriate to the fishing unit.

Selection of boat-landings within the sample landing places is made by using systematic sample with a random starting number.

The quality of the raw material obtained and calculated estimates depends among other things, on the magnitude of coverage errors and listing errors, measurement errors, and other kinds of systematic errors inherent in the field operation.

4. Processing of Collected Raw Material

Because CAS of small-scale fisheries are based on sample surveys, processing of the sample data has to be undertaken while taking into account the estimation procedure of the survey.

Processing might be manual by following the simplified procedure portrayed in the previous sections or by using mechanical means, e.g., micro-computers. In the former case, it is advisable to transfer the data from the completed questionnaires to specifically designed “working sheets”. As an example, we present the format of a “working sheet” which was used for catch and fishing effort estimates at Tanganyika Lake.

If a decision were taken for the mechanical processing of data, the advice of a specialist (system analyst/programmer) should be requested.

5. Analysis of the Results

Calculated estimates over time can be analysed by using proper statistical techniques. For example, regression techniques can be used for assessing existing relationships between the survey magnitudes. Time series analysis methods can also be used for assessing existing trends underlying the empirical time series and for forecasting purposes.

6. At Sea Observer's Check List

As part of a more detailed monitoring of the fishery, and in order to collect more data on aspects such as discarding that do not show up in port sampling, trained observers should go to sea on commercial vessels on an occasional basis to collect in situ data. Form H is one type of record that can usefully be used for the purpose, after modification to suit local conditions.

FORM H: AT SEA OBSERVER'S CHECK SHEET

A. (Complete one for each trip)

7. Size Frequency and Age Composition Data

In survey programmes for estimating length composition of the longest species, samples may be purposive i.e., at wholesale markets on random i.e., sampling of commercial catches (see also section 6.8). Specifically, sampling of commercial catches might be based on:

(i) random sampling at landing places;

(ii) on-deck sampling (thus estimating discards)

Random sampling programmes at landing places are based on sub-samples of catches/ landing taken during the main catch assessment survey (CAS). In such a case, weighing factors can be worked out in order to raise the sample data to the population totals.

Examples of the formats used for data collection are given in Forms I and J.

Remarks

Form I
Length frequency recording sheet

	Vessal name														Species name
	Vessel code	Cruise number	Station number	Sample number	day	month	year	Minimum length	Maximum length		Length increment	Species code			Stratum code			Total catch kg.			Sample wt. kg.		Raising factor		Sheet number
	Length scale cm
	0.0	0.5	1.0	1.5	2.0	2.5	3.0	3.5	4.0	4.5	5.0	5.5	6.0	6.5	7.0	7.5	8.0	8.5	9.0	9.5	0.0	0.5	1.0	1.5	2.0
*	(1)0	1	2	3	4	5	6	7	8	9	0	1	2	3	4	5	6	7	8	9	0	1	2	3	4
+				111	1				etc
S	5			3	1	5			etc
C
	2.5	3.0	3.5	4.0	4.5	5.0	5.5	6.0	6.5	7.0	7.5	8.0	8.5	9.0	9.5	0.0	0.5	1.0	1.5	2.0	2.5	3.0	3.5	4.0	4.5
	5	6	7	8	9	0	1	2	3	4	5	6	7	8	9	0	1	2	3	4	5	6	7	8	9
S
C

key

* Add digit before length scale, to adjust the length scale to the size range encountered

+ Showing method of recording observations

S Sample total

C Catch total (after weighting up, using data form J)

8. Observations of Fishing Vessels at Sea

Information on foreign fishing, either legal licensed vessels, joint venture or chartered boat or illegal, may be collected by the Coast Guard or Navy. These arms of government may agree to carry a patrol vessel log, keeping data for use by the Fisheries Department. Form K is an example of a possible format.

FORM K: PATROL VESSEL SIGHTING LOG

(Carbon copy sent to department of fisheries for analysis, original retained by the National Defense Force Vessel)

	Date:
	Vessel:
	Observer:
	Time left port:
	Time returned:
	(Enter patrol track on map, showing vessels sighted with a cross and numbered in sequence)

Sightings (fill in one for each vessel)

9. Socio-Economic Statistical Surveys

(a) Cost and Earning Statistical Studies of Fishing Units (CE-S's)

In order to assess the techno-economic and socio-economic performance of various fishing systems, specific statistical surveys are designed and executed within the overall survey system of work using the principle of “integration”. Specifically, the survey system of a CE-S is fully integrated with the system of work of the main catch assessment survey (CAS). The integration principle ensures the improvement of the efficiency of the survey, reduction of the overall cost and the improvement of the utility of the results obtained. The main operational steps in which integration is achieved are:

1. Concepts, definitions and classifications are, to the extent possible, made compatible.
2. The organization and field execution of the various components of the surveys are integrated.
3. Sampling integration: Since CAS's are based on stratified two-stage sampling, CE-S's will be considered as partially integrated if they are based on the same Primary Sampling Units (PSU's).
4. The mechanical processing of data is integrated, (e.g., using common code or reference numbers).

CE-S's are multi-purpose in character, in the sense that the results obtained can be used by those responsible for planning and fisheries development programmes, in assessing their economic and social impact, and for management purposes.

(b) CE-S's of Non-Mechanized Fishery Units

In this section we discuss in a summary form, methodological and sampling aspects of a CE-S (small-scale fisheries). It must be emphasized at the very beginning that CE-S's provide only estimates of the average values of the survey magnitudes, i.e., overall mean estimates. This in turn means that the sample data obtained are only used for the estimation of the required mean values e.g., average catch values, average gross revenue values, etc. Consequently, CE-S's are free of the sampling complexities inherent in large-scale fisheries statistical surveys aiming at the estimation of population totals.

In CE-S's the survey subject is the “active” fishing unit. That is to say, the target population is a sub-population of the total population of fishing units. Because of the observed spatial and temporal variations in the number, structure and operational characteristics of the active fishing units, the results of the surveys should be treated with care. Because of the lengthy questionnaire used in the survey for obtaining information on the complete spectrum of the characteristics required, hence, the selection of a small sample of active fishing units for cost-oriented surveys, the number of samples, and hence the statistical power of the calculated estimates is not very high. Another problem which affects the results of the survey, is the “non-response” rate observed, which effects the generality of the data, and the level of quality of the information obtained.

The sample design of a CE-S is based on stratified two-stage sampling. As far as the PSU's is concerned, the sample of a CE-S is a sub-sample of a sample unit of the main (integrated) CAS. The method of post-stratification is used within the sample PSU's for the selection of the sample of fishing units. Specifically, active fishing units are grouped into domains by using a hierarchy of criteria for stratification, e.g.,

1. Fisheries
2. Kind and size of fishing craft and gear
3. Importance of fishing units expressed in terms of capital investment or number of units of gear used.
4. etc.

A sample of fishing units is selected within some (or all) domains of interest for data collection. The selected units sampled are kept fixed over time, and the required items of information are collected from these sample fishing units over a number of successive occasions e.g., monthly, for a duration of at least one year. Final decision on one sampling process requires a detailed study of the overall survey system.

Sampling in time could also be introduced in the process of data collection. For example, for dynamic characteristics, e.g., trends or fluctuations in levels of catches, effort, etc. data can either be collected on a daily basis for the whole reference period, of say, a month (census in time) or for a chosen sample period, say, a week within the reference period (sample in time).

A proper codification system should be developed, to economically describe port and area fished, date, time, fishing units and their characteristics, and this should be introduced during the design of the survey. This simplifies mechanical processing, and presentation and analysis of the sample data.

Usually the same statistical recorders carrying out the CAS are also used for data collection in the catch and earnings survey. Extra time is allocated to them for the additional field operations required.

Either the book-keeping method or interview method can be used for obtaining the required information. For small-scale fisheries the interview method is usual. The book-keeping approach needs literacy and good cooperation by the fishermen; it is associated with a high non-response rate.

The items of information which may be obtained by a CE-S of non-mechanized fishing units can be grouped under the following headings:

An example of the format of the questionnaire used for non-mechanized fishing units is shown in Form L, from Kurien and Willmann (1982).

(c) CE-S's of Mechanized Fishing Units

Catch and earnings surveys of mechanized fishing units, have often been concentrated particularly on trawl fisheries, which are not always of great importance to small island fisheries. However, as with the other sample forms shown, Form M should be modified in the light of local conditions while preserving, to the extent possible, the main categories of information. The main objective of the surveys is to estimate the productivity and profitability of selected fisheries, for management purposes.