Assessment of dual selection in grid based selectivity systems
Herein we propose a method to assess dual selection in grid based selectivity systems. This method takes into account the parameter “grid contact likelihood” (Cgrid), which can be interpreted as the proportion of fish that actually makes an attempt to escape through the grid. In a case study of the...
| Published in: | Fisheries Research |
|---|---|
| Main Authors: | , , , |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
2010
|
| Subjects: | |
| Online Access: | https://orbit.dtu.dk/en/publications/6db6bde0-8518-4ea7-9c1c-fbd59398b605 https://doi.org/10.1016/j.fishres.2010.05.006 |
| Summary: | Herein we propose a method to assess dual selection in grid based selectivity systems. This method takes into account the parameter “grid contact likelihood” (Cgrid), which can be interpreted as the proportion of fish that actually makes an attempt to escape through the grid. In a case study of the Barents Sea cod and haddock trawl fishery, we demonstrate that our model describes the experimental data better than the models previously used to fit similar data. For both cod and haddock, Cgrid was significantly smaller than 1.0, which demonstrated the relevance of the proposed model. Cgrid was higher for haddock than for cod, which might be due to behavioral differences between the species. The Cgrid values for both species suggest that the grid functions well, as on average more than 75% of the cod and more than 94% of the haddock were predicted to able to attempt an escape through the device. For both species the contact L50 for the grid (L50grid) was significantly higher than the L50 for the codend (L50codend). These values agree with the experimental observations that most of the escaping fish use the grid to escape, whereas only a very few escape occurs through the codend. By parametric simulation and using the case study results as the baseline, we investigated and compared the precision of the selectivity parameters estimated with our model for two different experimental setups. The results show that except for some extreme situations, the data for such studies need to be collected with a three-compartment setup to avoid imprecise estimates of Cgrid, L50grid, SRgrid, L50codend, and SRcodend. In general, only the combined selectivity of the grid and the codend could be estimated with acceptable precision using a standard two-compartment sampling approach. |
|---|