Work

Report of the Workshop on Survey Design and Data Analysis (WKSAD) [21- 25 June, 2004, Aberdeen, UK]

Year: 2004
Type: report
Abstract: The Workshop on Survey Design and Analysis [WKSAD] met in Aberdeen, Scotland, UK, from 21-25 June 2004 to: a) review methods of designing and analysing fisheries surveys; b) summarise the current methods used for survey design and analysis; c) investigate where there are similar design and analysis problems; d) identify areas of agreement and specific areas of work where progress could be made; e) prepare work plans for identified areas of development; and f) investigate methods to deal with intercalibration studies of fishing gears and survey vessels. 2) A REVIEW OF METHODS OF DESIGNING AND ANALYSING FISH SURVEYS.An account of previous efforts to examine survey design and analysis is given.The essential statistical elements to survey planning are described and an analytical framework for survey analysis is proposed.Some surveys perform rather well and stock assessments may be improved using alternative models which use survey data more explicitly.3) SUMMARY OF CURRENT METHODS.A number of specific examples are given describing current survey practise in most of the ICES member states.Survey designs, estimation of abundance and variance, and use in assessments are covered for trawl, acoustic and other (ichthyoplankton, visual, drag or dredge) surveys. 4) SURVEY-SPECIFIC DESIGN AND ANALYSIS ISSUES.Tow duration in trawl surveys needs to be investigated in more detail because there are advantages to taking a short tow.Visual surveys are becoming popular and distance sampling methods should be investigated.Adaptive sampling can improve precision in cases where the target species is sedentary, but may be less efficient when more mobile patches of organisms are hard to find (small relative to the area).Biological sampling methods in acoustic surveys need improvement. 5) GENERAL DESIGN AND ANALYSIS ISSUES.There is an increasing array of model-based procedures.Geostatistics, for example, enable the precision of a survey to be estimated using the global estimation variance, providing the autocorrelation function can be determined.Simulations and model-based procedures which have non-linear approaches to dealing with extreme values can be informative and improve the estimation process, although the methods still need to be evaluated.Survey reports should include: a comprehensive description of estimation procedures, survey precision as the relative standard error, and measures of design efficiency.6) CHOICE OF SURVEY DESIGN.In the presence of positive local autocorrelation (common in most fish surveys), a more precise estimate of the population mean will usually be obtained by systematic sampling or stratified random sampling than by simple random sampling.The optimal sampling design depends on the population and the relative importance of getting the most precise estimate of the population mean and to getting a good estimate of that precision.Fixed survey designs are common in multispecies surveys (e.g., IBTS), and can be effective for detecting trends when the spatial distribution is persistent.They are also practical in areas with significant un-trawlable seabed.However, fixed designs cannot provide unbiased estimates of the variance.7) INCORPORATING ADDITIONAL INFORMATION.Information additional to that of fish density should be collected on surveys, particularly when that information is related (covariate) and can be collected more extensively.Incorporation of covariates (habitat, environment) can lead to improved precision of the abundance estimate, provided that a good relationship exists, and that the covariate is known at more sample locations than the fish density.Ideally, the covariate should be known at all locations (i.e., the whole survey area).8) INVOLVING STAKEHOLDERS IN FISHERIES SURVEYS.Information from the commercial fishing industry should be considered, where appropriate, to provide guidance on survey design.A range of other options were considered and guidelines for the conduct of cooperative research surveys are given.9) FUTURE WORKPLANS / 2005 ToR.The Terms of Reference for the next meeting are: a) Evaluate analyses of estimates of the abundance, associated variance, and density maps, from surveys of a simulated fish population whose abundance is known; b) Evaluate alternative analyses of seven survey datasets; c) Review the state of knowledge regarding the effect of trawl duration on fish catch rate with a view to considering a reduction in sample trawl duration; d) Evaluate analyses of covariate data which could provide improved precision of abundance estimates; e) Review methods for combining surveys of the same resource using different methods; f) Evaluate the sensitivity of methods to estimate biological parameters in terms of analytical assumptions and measurement error.10) INTERCALIBRATION STUDIES OF FISHING GEARS AND SURVEY VESSELS.A number of intercalibration studies of trawl surveys and acoustic surveys were presented.If calibration factors are estimated with poor precision (as is often the case
Author: ICES
Language: English
Subfield: Aquatic Science
Open Access status: green
OpenAlex ID: W7042798322
Cites: 226
Cited by: 3
Related to: 10