Why are there still large pelagic predators in the oceans? Evidence of severe hyper-depletion in longline catch-per-effort

Industrial catch-per-effort (cpue) data are used as an indicator of population abundance and, contrary to strong cautions of potential biases, are often used without regard for spatial interactions as the only data source from which inferences are made. Recent controversy surrounding the status of large predatory pelagic communities has resulted from varying interpretations of the Japanese longline cpue data. Contrary to many stock assessments on the major tuna species, ratio and fished-area-only cpue estimators for specific regions of the world oceans indicate that large pelagic predator biomass has declined to 10% of pre-industrialized fishing levels, with large declines occurring in the first few years of fishing. We have re-examined the SPC public domain longline 5°x5° global data set, correcting for errors that result from utilizing ratio cpue estimators, to see if such spatially corrected cpue data provide a more reliable indicator of population abundance. Although spatially corrected cpue time series indicate depletion more in line with current stock assessments, there is evidence of severe hyper-depletion even in the corrected data. For several species, estimates of recruitment (to size classes fished by longlining) based on catch and cpue would indicate linear stock recruitment relationships. Such relationships are contrary to current assessments and are expected under declining catchability (q) over time. When recruitment is assumed stable (utilizing compensation observed over most species) q is calculated to decline rapidly over the initial years of the fishery. Other, more complex assessments utilizing recruitment and abundance estimates from catch composition data also estimate declining q even after the period of early cpue decline. Apparent fishing mortality rates (F) required to produce the initial declines in ratio cpue with 20-30% of the maximum effort observed later, imply substantially higher F in later periods if q is assumed constant. Such an effect is also observed in the analysis of 5°x5° cell specific depletion models. Such high fishing mortality rates are inconsistent with current estimates from stock assessment, and with size composition of catches; yield per recruit analysis indicates that such Fs would have resulted in greater declines in mean weights than observed in the catch. Such observations can be explained by hyperdepletion in the early cpue data and further investigation into the early spatial distribution of fishing effort is required to determine if effort was initially targeted at localized spatial aggregations or alternately if longline effort initially removed more active and susceptible components of the population.