Evaluating the impacts of reduced longline fishing effort on the standardization of longline catch-per-unit-effort for bigeye tuna in the eastern Pacific Ocean

Indices of relative abundance directly inform how population abundance changes over time, providing one of the most important pieces of information for a stock assessment. Ideally, indices of abundance should be calculated based on fishery-independent survey data. Survey data are characterized by a spatially random or fixed sampling design, and consistent employment of the same fishing gear and fishing operation across time. However, the unavailability of survey data for most tuna species means that the derivation of abundance indices for these species comes solely from fishery-dependent catch-per-unit-effort (CPUE). We conduct two simulation experiments, based on real fishery-dependent longline data, to quantitatively evaluate the impacts of reduced fishing effort on the standardized longline CPUE for bigeye tuna in the eastern Pacific Ocean. The key findings of the two simulation experiments are 1) a reduced spatial coverage of CPUE data leads to increased bias in the abundance index; 2) the index bias has a minor long-term trend if the reduced spatial coverage of CPUE data is not caused by local depletion; and 3) that bias has a positive long-term trend (i.e., hyper-stable abundance index) if the reduced spatial coverage of CPUE data co-occurs with a local depletion in the abandoned area. This bias, however, can be significantly reduced if the CPUE standardization model includes a temporal correlation structure in spatiotemporal random fields. In addition, the CPUE standardization model provides more realistic estimates of the coefficient of variation of fish abundance when its spatiotemporal random fields are assumed to be correlated in time. This study underscores the necessity of accounting for the temporal correlation structure in spatiotemporal random fields in cases where local depletion and depletion-driven fishery contraction co-occur.