Spatial bias in mark-recapture data: estimation and consequences on stock assessments of Patagonian toothfish in the Kerguelen EEZ (TAAF)

Following a general overview of the fishery in the Kerguelen EEZ and the methodological framework
of mark- recapture models, an exploratory analysis of spatial changes in fishing vessels in this area
revealed the main shifts in the distribution of fishing effort. Although the spatial footprints of different
fishing seasons are generally similar, the early years (2002-2005) stand out from the others. The
development of a computational simulation framework allowed for the design and validation of
corrective factors. These factors are based on the ratios between the area, or the abundance of fish,
within the annual fishing footprints and a global fishing footprint. It was demonstrated that corrective
factors considering the relative abundance of individuals are more relevant than those considering
only the area. These corrections were then applied to the real fishing data from Kerguelen. Several
statistical analysis and prediction tools, (including predictive GAM models and KDE), were used in this
spatial analysis framework. The corrected biomass estimates did not differ significantly from the
uncorrected estimates, suggesting that potential biases balance out.