Groundfish Forum’s Experiment to Test Observer Basket Sampling Methods

Reason for an EFP

Accurate estimates of catch by species: The North Pacific Council and NMFS have developed regulations and are in the process of considering additional regulatory initiatives to improve catch weight estimation in fisheries involving at-sea processing. These requirements have focused on improving the estimates of overall catch rather than improving the estimates of individual species. Knowing that a haul caught 10.03 MT rather than somewhere between 8 and 12 MT has a certain value, but if there is a plus or minus 40% error (hypothetically) surrounding the estimation of what species comprise the 10.03 MT haul, then the gain in accuracy in total catch estimation is overshadowed.

  • IR/IU: During the first year of the full retention regime for pollock and cod, some vessels found that they were unable to reconcile product case weight counts converted to round weight equivalents with estimates of catch of these species from observer data, despite the fact that they are retaining all cod and pollock. This has occurred even when they are putting up cod and pollock in round product form where product recovery rates are not an explanation for the difference.
  • Multispecies Management: Groundfish Forum is also aware that the environmental community is focusing on multi-species management regimes. This will put considerable focus on the management of species complexes and fishing effects on non-target species. This approach will require accuracy in estimates of removals by species and estimates of age composition of catch.
  • VBA’s: As management turns to individual vessel incentive programs such as vessel bycatch accounts (VBA) and community development quotas (CDQs), accuracy at the level of the individual haul will become more of an issue than it has been in the past.
  • Reduce conflict: Lastly, we are aware that a common source of conflict on fishing vessels (right or wrong, legitimate or not) is the accuracy and potential bias of observer sampling methods for prohibited species. We see benefit in working to reduce potential for bias in the data collection system and the ensuing conflict that can occur between industry, the Observer Program, and observers. We believe the development of an automated sampling technique, such as the one slated for testing in this EFP, will help to accomplish this objective.

Three Parts of the EFP

  • Identification and quantification of potential inaccuracies of basket sampling practices

Six 100 KG samples will be taken from each haul using a systematic design that observers must attempt to apply consistently and without the aid of the flow scale to gauge how much of the fish in the haul has passed from the live tank into the processing area. Observers, with the assistance of crew members (see responsibilities section below), will perform species composition analysis of each of the six samples and record data separately.

The time and flow scale readings will be recorded when each of the six samples are taken (the decision to sample must be made without reference to flow scale readings) and when the first and last fish cross the flow scale. Interval data and associated species composition samples (when compared to the actual composition of major species in the haul) will be used to provide an assessment of the accuracy of sampling and the consistency of systematic sampling techniques. Data derived from this portion of the project will be analyzed for inaccuracies from systematic sampling and sampling that, in effect, is not systematic. Given the expectation that stratification of fish does occur in fisheries with mixed catches, it is hoped that the EFP will help improve sampling for multi-species fisheries.

  • Development of automated sampling methods for species composition sampling

Groundfish Forum will provide software that will generate six sampling “trigger weights” or “sampling points” spread throughout the fish comprising the haul. These points will be fed into the flow scale and the flow scale software will “trip” a mechanical “diverter” mechanism which diverts the specified quantity of fish into a tote placed to catch the diverted fish at the prescribed weight intervals.

As with the test of the observer systematic sampling method above, species composition sampling will be performed on each of the six 100 KG samples and data on the weight and time intervals of the samples will be recorded for each sample. As with the test of the observer systematic sampling methodology, the true species composition of the hauls where automated sampling is used will be evaluate the relative accuracy of the sampling methodology. Data generated in the experiment will also be used to evaluate potential for improvements or simplification of the methodology through adjustments in sample size and intervals.

  • Exploration of the degree of stratification of fish by size in trawls and evaluation of potential effects of current size composition sampling practices on size composition estimates

In addition to the tows used to test the accuracy of species composition, a series of tows will be dedicated to a pilot test of the accuracy of size composition sampling. These cannot be done in the same tows as for the species composition sampling because the work necessary for testing species composition will require all of the available sampling resources.

At least six samples of twenty fish each will be drawn systematically for each of at least two target species. These samples will be measured and recorded separately. As with the species composition test, the separate sub-samples will be compared and recombined to test for the relation between errors and sample size or sequence in the catch. Reduced species composition sampling will be carried out during these tows to track the total catch.

How the Real Composition of a Hall will be Determined

  • Estimate of retained catch: In the first phase of the experiment, the participating vessel will make a tow in an area where a mixed catch of the species of interest for the target fishery can be expected. A quantity of fish (probably 100-200 KG) of a species will be set aside for each product form that will be produced during the experiment. The mixture of fish sizes comprising the 100-200 KG quantity should approximate the range of sizes targeted for the catch. Fish set aside in the 100-200 KG batch will be run through cutting machines and total remaining weight of fish will be determined to obtain a species and product-form specific estimate of recovery rate. This process will be repeated for each species and product form to see how much variability there is in the estimated recovery rate, and to develop an average with an associated variance that will be used for the statistical analysis. Several tests of the recovery rates established at the outset of the experiment will be done during the experiment to ensure that recoveries are consistent during the experiment. Cases of finished product will be marked by tow and the total weight will be “back-calculated” to round weight.
  • Estimate of discards: The discard line would have to be routed such that crew members responsible for sorting discards would have sufficient space to remove the major discard and PSC species. Those species would be placed in totes or other holding bins and an approved platform or flow scale would be used to record weights by species.

All fish will be passed over the flow scale so that the weight of total catch per tow is an accurate basis for further calculation. The total weight of other discards will be determined by subtraction of total retained weight per tow (as calculated above) from the flow scale weight of the tow. The weight of fish that are too large to be placed in the live tanks will be estimated on deck.

According to fishery data obtained from Sea State for the flathead sole target fishery in July and August of 1998, arrowtooth flounder may be a major component of discard and can exceed the volume of target species in some tows. Applicants will have to specify by which method they will determine arrowtooth weight. This could be achieved by separating out the arrowtooth and weighing it, or weighing all species but arrowtooth and arriving at the weight by subtractions.


The experimental fishery was conducted with the Groundfish Forum and with the NMFS Observer Program. The experiment fishery examined how catch stratification influenced observer catch composition estimates. The project was carried out by the captain and crew of the F/T American No.1 with field supervision from NMFS and Groundfish Forum. The observers collected six random samples from every tow. The EFP personnel then quantified all crab, halibut, skates, and five groundfish species in order to compare the actual catch composition with the basket sample estimates. The data are still being analyzed, but the study will prove useful to NMFS and the industry in assessing the accuracy of the current sampling regime.