9. BOAT ACTIVITY SURVEYS

9.1 OBJECTIVES OF BOAT ACTIVITY SURVEYS
9.2 TARGET DATA POPULATION AND ACTIVITY STATUS
9.3 SAMPLING REQUIREMENTS
9.4 BACS RELYING ON FRAME SURVEYS
9.5 SAMPLING FOR BOAT ACTIVITIES
9.6 COMBINATION WITH LANDING SURVEYS
9.7 BRIEFING OF DATA COLLECTORS
9.8 IMPLEMENTATION ASPECTS
9.9 FREQUENT PROBLEMS

9.1 OBJECTIVES OF BOAT ACTIVITY SURVEYS

The primary objective of a Boat Activity Survey is the formulation of Boat Activity Coefficients (BAC), which represent the probability that a fishing unit of a given boat/gear type will be active on any day during a month. BACs are then combined with raising factors resulting from Frame Surveys and Active Days Surveys for estimating total fishing effort within a minor stratum, month and boat/gear category.

Boat Activity Surveys also assist in assessing the general accuracy of previous Frame Surveys through sampling (see section 8.10), as well as for recognizing significant changes in the fisheries.

9.2 TARGET DATA POPULATION AND ACTIVITY STATUS

The target data population of a Boat Activity Survey is the activity status of all operating boats on all days of a reference month. The activity status is set to one if a boat has been found fishing on a given day, or to zero if it has not.

Example: In a minor stratum during April (30 days) there are 100 gillnets. The target data population consists of 30 × 100 = 3,000 status elements for the month that are either 1 or 0. If the number of “active” (= 1) status indicators is 1,500, the BAC will be computed as 1,500/3,000 = 0.5.

9.3 SAMPLING REQUIREMENTS

The target data population is thus the number of fishing boats (as estimated by a Frame Survey) multiplied by the days in the month. Since the Boat Activity Survey is a sampling approach, the question is, How many of the target data population should be sampled to ensure that the estimate is representative of the entire data population? Or: What is the safe sample size for estimation of the BAC? (see also section 4. General Sampling Considerations).

Table 9.3 BAC sampling requirements at varying accuracy levels and data population size

Accuracy (%)	90	91	92	93	94	95	96	97	98	99
Data Population size	Safe sample size for BACs
300	73	85	100	119	141	168	200	234	267	291
400	77	91	109	132	160	196	240	291	343	384
500	81	96	115	141	174	217	273	340	414	475
600	83	99	120	148	185	234	300	384	480	565
700	84	101	124	153	193	248	323	423	542	652
800	86	103	126	157	200	260	343	457	600	738
900	87	105	129	161	206	269	360	488	655	823
1000	88	106	130	164	211	278	375	516	706	906
2000	92	112	140	179	235	322	462	696	1091	1655
*see notes 3000	93	114	143	184	245	341	500	787	1334	2286
4000	94	115	145	187	250	350	522	842	1500	2824
5000	94	116	146	189	253	357	536	879	1622	3288
6000	95	116	146	190	255	361	546	906	1715	3693
7000	95	117	147	191	257	364	553	926	1788	4049
8000	95	117	147	191	258	367	558	942	1847	4364
9000	95	117	148	192	259	368	563	954	1895	4646
10000	95	117	148	192	260	370	566	964	1936	4899
15000	95	118	149	193	262	375	577	996	2070	5855
20000	96	118	149	194	263	377	583	1013	2144	6488
25000	96	118	149	194	264	378	586	1023	2191	6939
30000	96	118	149	195	264	379	588	1030	2223	7275
35000	96	118	149	195	265	380	590	1036	2247	7536
40000	96	118	150	195	265	381	591	1039	2265	7745
45000	96	118	150	195	265	381	592	1042	2279	7915
50000	96	118	150	195	265	381	593	1045	2291	8057
> 50000	96	119	150	196	267	384	600	1067	2401	9602

The above table indicates recommended sample sizes for estimating BAC at a desired level of accuracy. In a sample-based survey for basic fishery data, the minimum accuracy level of an estimate is empirically set to 90%.

Safe sample sizes are a function of the population size but for populations with more than 50,000 elements (equivalent in one month to about 1800 boats in the stratum) their differences are practically negligible. For a graphic interpretation of the table see also Figure 4.3.

Notes on the table:

Sample sizes are interpreted as follows:

In the example given in 9.2 the data population BAC was 0.5. The data population size is 3,000 (100 boats x 30 days) and related safe sample sizes at varying degrees of accuracy are indicated in the highlighted line.

At a desired accuracy level of 95%,

• by the end of the month 341 boats should have been examined for activity status. Using 341 samples of examined boats, the sample BAC would be formulated by finding the number of active boats and dividing it by 341.
• The safe sample size of 341 corresponding to an accuracy level of 95% will assure that, at worst, the resulting BAC estimates would be as high as 0.55 or as low as 0.45 (in the example case).
• Assuming that sampling occurs during 10 days then about 35 boats should be examined on each sampling day in the minor stratum.

At a lower level of accuracy, to 90%,

• the corresponding sample size by the end of the month would be 93.
• At worst the data population BAC would be then estimated between 0.4 or 0.6.
• Working with 10 sampling days would mean that about 10 boats should be examined on each sampling day in the minor stratum.

9.4 BACS RELYING ON FRAME SURVEYS

This is a commonly used approach for formulating BACs and has been addressed in detail in section 8.3. On each sampling day data collectors visit a number of pre-selected homeports and record the total number of boats that were found active at these sites. This total number is then divided by the total number of boats indicated by the Frame Survey. An example data collection form for this approach is given by Table 9.4.

9.5 SAMPLING FOR BOAT ACTIVITIES

In this approach no use is made of frame survey information. BACs are calculated on the basis of representative samples of boats that are interviewed to determine their state of activity on a sampling day. Frame survey data are only used as raising factors. An example data collection form for this approach is given by Table 9.5.

Table 9.4 Boat Activity Survey - data form with Frame Survey data

Boat Activity Survey		Stratum: SW Coast Homeport: Channel		Recorder: Samuelson
Active boats
	Trawlers	Gillnets	Beach seines	Castnets	Traps
Frame data	10	30	9	12	11
Day
1
2
3
4	5	12	3	1	4
5
6
7
8	4	14	2	5	2
9
10
11
12
13
14	6	20	3	4	7
15
16
17
18
19
20	5	9	3	5	6
21
22
23
24
25	1	5	0	2	3
26
27
28
29	7	18	4	6	8
30

Table 9.5 Boat Activity Survey-form showing sample numbers

Boat Activity Survey		Stratum: SW Coast Homeport: Channel		Recorder: Samuelson
Active boats
Day	Trawlers	Gillnets	Beach seines	Castnets	Traps
1
2
3
4	5/8	12/19	3/6	1/4	4/8
5
6
7
8	4/9	14/22	2/5	5/8	2/6
9
10
11
12
13
14	6/9	20/24	3/5	4/9	7/11
15
16
17
18
19
20	5/12	9/16	3/8	5/6	6/13
21
22
23
24
25	1/3	5/12	0/5	2/5	3/8
26
27
28
29	7/15	18/19	4/9	6/12	8/10
30

In the example 9.5 data recorders indicate the number of boats found active out of the number of boats sampled. For instance, 5/8 for trawlers means that 8 fishers were asked and 5 answered that they were fishing, while three specified that they were not. Therefore, the BAC for trawlers is:

BAC = (5 + 4 + 6 + 5 + 1 + 7)/(8 + 9 + 9 + 12 + 3 + 15) = 28/56 = 0.5

• It is recommended for large ports, when it is difficult for the recorders to identify all boats that were active, or in cases of frequent migration of fishers from one place to another.
• Boats should be sampled without prior knowledge on their activity. It would be wrong to approach fishers that are known to have been fishing on the sampling day, since they would all be found active.
• It is good practice to pre-select boats or fishers prior to visiting a site and then track down the activities of the pre-selected fishing units or fishers.

9.6 COMBINATION WITH LANDING SURVEYS

At times it is convenient to combine Boat Activity Surveys with Landings Surveys. This can be done using a single form as follows:

In the example given above the form used for the recording of landings is also used to capture boat activity data.

Landings of a boat were sampled on 17 March 2001. In addition to landings information for effort parameters and species composition, the fisher was asked to specify if he went fishing on the three previous days. This was indicated by a 0 or 1 in the boxes printed on the right part of the form. The following three assumptions have been made:

• The fisher remembers his activities over the last three days.
• The same gear and boat were used (in theory this is not essential but data recording and transcription would be too complex).
• His homeport is also the landing site (again, to avoid complex data transcription).

The characteristics of the approach are as follows:

• It usually applies to boats that operate on a “one-trip-per-day” basis.
• Current day should not be included since all fishers will specify “YES-fishing” for that day.
• It can provide good time coverage for effort. If the Landing survey is conducted 10 times during a month, this approach will cover 30 days (three days per sampling day for landings).
• It requires that the 0-1 answers are further elaborated and summed for other fishers to produce an inputting form similar to that provided in table 9.5.

9.7 BRIEFING OF DATA COLLECTORS

Providing data collectors with precise and unambiguous instructions is fundamental for the reliability of the data obtained through Boat Activity surveys.

The following points are important in briefing data collectors:

• Explaining in full detail the recording forms, their use in data collection and their purpose and utility in the overall sampling programme.
• Clarifying the nature of boat/gear types, sequential and concurrent gear use, and whether new boat/gear types are significant or can be ignored, etc.
• Planning the visits to homeports.
• Methods for approaching and interviewing fishers and village authorities to obtain complete and reliable information on numbers of boat/gear types operating from homeports.
• Ways of cross-checking the obtained information and the appropriate course of action in cases of serious discrepancies.

9.8 IMPLEMENTATION ASPECTS

Implementation of successful Boat Activity Surveys needs careful planning including:

Pilot phase: small scale testing (e.g. within one or two landing sites) to identify possible design and operational drawbacks and to assess likely timeframes for full-scale operations.

Testing of the entire sampling programme: combine a pilot implementation of the Boat Activity Survey with the entire sampling programme for catch/effort assessment (normally 6-12 months) to ensure both meet their requirements.

Evaluation and revision phase: forms and classifications are revised and new instructions issued to data collectors.

9.9 FREQUENT PROBLEMS

9.9.1 Timing of field activities

• When BACs are based on Frame Survey data a quick way of finding out active boats is to visit a homeport before boats have started landing and count the boats that are present. The difference (frame boats) - (boats present) will provide an indication of “boats active”.
• When pre-selected boats or fishers are used for sampling it would be better to visit homeports when most boats have returned.
• When pre-selected boats or fishers are used for sampling and trips are longer than one day, some boats may not be possible to trace. Information may be obtained from other fishers.

9.9.2 Multiple use of fishing gear

BACs must be formulated according to survey standards and for each boat/gear element of the related classification. A frequent problem is that of boats operating different gears, which might be sequential or concurrent. Sections 8.3 and 8.4 provide some guidelines on these aspects.

9.9.3 Migration of fishing units

The migration of fishing units distorts the presence/absence of boats at homeports and for this reason the sampling approach should be used for formulating BACs.

• If migration occurs within a minor stratum there will be no implications in the estimation of total fishing effort within that context.
• If migration occurs across strata, for some strata effort will be under-estimated while in others it will be over-estimated.
• If migration is seasonal and can be anticipated, frame surveys should reflect such boat movements.

9.9.4 Outdated frame surveys

Sampling at homeports for BACs should take place on an equal number of days, so as to rely on the relative accuracy rather than the absolute accuracy of Frame Surveys.

If no significant migration of boats has occurred and sampled homeports show a consistent increase or decrease of boats not accounted by Frame Survey information, this might indicate that overall increases or decreases have occurred to the fishery and a new Frame Survey should be implemented, or its present contents adjusted to reflect such changes.

SUMMARY This section on Boat Activity Surveys completed the discussion on methodological and operational aspects of surveys that are involved in the estimation of total fishing effort using the Boat Activity Coefficient, including: Objectives of Boat Activity Surveys. Target population. Sampling requirements and safe sample sizes achieving a desired level of accuracy in the estimation of BACs. Commonly used sampling approaches. Need for training and effective briefing of data recorders. Pilot implementation, feedback and survey expansion. Frequent problems in Boat Activity surveys.