User menu
Environmental, spatial, and temporal variability could impact the relative abundance of 8 highly migratory species. It becomes especially problematic when the variability affect the 9 standardization of CPUE (catch-per-unit-effort) used to assess the status of fish stocks. This 10 paper presents CPUE standardization and model comparison procedures for bigeye tuna 11 (Thunnus obesus) in the Indian Ocean based on multi-scale fisheries data and environment 12 data from 2008 to 2015. We used the fisheries datasets from two sources for comparison: (1) 13 the statistical longline datasets published by IOTC Secretariat with monthly catch-and-effort 14 of the 5ºor 1ºgrid; and (2) the survey datasets from the Chinese longline fishery with set by 15 set catch-and-effort data. We calculated multiple marine environmental factors for CPUE 16 standardization models. Beside those frequently used factors, such as sea surface temperature 17 (SST), sea surface height (SSH), concentration of sea surface chlorophyll a (Chla), we also 18 calculated factors that could possibly affect the fish distribution habitat but were rarely used in 19 previous CPUE standardization study, such as the vertical ocean temperature and salinity 20 factors based on 15 profiles of the ARGO buoys, the nearest distance between CPUE positions 21 and the SST fronts, and the eddy kinetic energy (EKE) derived from geostrophic velocities. 22 We applied cluster analysis methods to identify suitable environmental locations for the target 23 species, using the approaches developed in the 2015 IOTC CPUE standardization workshop. 24 The fisheries data were aggregated by the 5ºor 1ºgrid, and then clustered on species 25 composition in the catch, using the Ward hclust and kmeans method. The cluster group 26 parameter was then included as a categorical factor in models. We used generalized linear 27 model (GLM) with lognormal constant analyses for CPUE standardization. We generally built 28 three types of models based on the fishery dataset sources and the inclusion of potential 29 factors. The results showed that regions with high catch rate of bigeye tuna were identified based on the cluster analysis. The numbers of clusters selected varied among regions of 31 different spatial resolution, but in all cases were either 4 or 5. Basically, in the condition of the 32 same set of covariates, models with larger quantity of fishery datasets or at higher spatial resolution 33 had better performance. The residual patterns of GLMs with survey datasets from the Chinese longline 34 fishery showed a more uniform and symmetric distribution than GLMs with IOTC datasets of 1ºgrid, 35 which indicated that a finer spatial and temporal scale would be useful for subarea standardization. The 36 inclusion of more environmental variables and habitat clusters aided the standardization process as 37 well. Our results highlighted the importance of choosing the meaningful explanatory 38 environmental variables and the appropriate fishery datasets scales for CPUE standardization. 39 Conclusions of this study could be served as recommendations for further practices for CPUE
