Characterisation, CPUE and length-composition analyses of the New Zealand albacore tuna fishery

Albacore within the New Zealand fishery waters are part of the South Pacific Ocean stock which is distributed from the equator south to about 50°S. The broader stock is managed by two Regional Fisheries Management Organisations (RFMOs); the Inter-American Tropical Tuna Commission (IATTC) manages the eastern part of the stock, while the western part is managed under the Western and Central Pacific Fisheries Commission (WCPFC). In the WCPFC Convention Area, albacore are landed in a number of fisheries but primarily from longline and troll fisheries, with over 75% of the troll catch coming from New Zealand waters. This report updates the characterisation and CPUE analyses for the New Zealand albacore (Thunnus alalunga) fishery presented in 2021, and provides additional analyses into drivers of variability of albacore cohorts and length compositions in New Zealand waters. This issue is particularly relevant due to recent concerns about length composition data from New Zealand’s troll fishery contributing to potentially spurious signals of very low recruitment over 2015 and 2017, which drives a strong reduction in spawning biomass in projections from the 2021 South Pacific albacore stock assessment. The present analysis was undertaken to provide information for the regional stock assessment undertaken by the Pacific Community (SPC) and focuses primarily on the troll fishery which lands the bulk of the catch in New Zealand, but longline data are also included.

The present analysis describes the fisheries targeting and catching albacore in New Zealand, and undertakes a CPUE standardisation of the commercial troll fishery. Data for the fishery description included the catch and effort from 1990 to 2023 from albacore troll target sets, all other troll activity, and all trips where albacore was specified as the target or recorded as bycatch. In recent years, both unstandardised CPUE and landings have declined in the fishery. For the CPUE analysis, catch-effort data for albacore target trolling only were modelled using maximum-likelihood generalised linear models (GLMs) predicting the catch as a function of spatial, temporal and environmental predictors. The models without environmental predictors provided a minimal standardising effect, producing CPUE series that show little long-term trend. Overall, the series, which now extends 30 years, fluctuates without trend over the last two decades, but recent standardised CPUE is at an all-time low. Models with environmental predictors showed a strong standardising effect of the El-Niño Southern Oscillation on catch rates. The standardised CPUE index is likely a reflection of availability and/or catchability, rather than a representative signal of abundance.

Length compositions from New Zealand’s troll fisheries were analysed in two separate ways to gain insights into drivers of variability in the length composition of troll-caught albacore in New Zealand waters. First, a mixture model was constructed that aimed to partition the length data into three apparent length-cohorts by year. Although the model did not provide an good fit to the data, it suggested that the presence in the catch of the first cohort (thought to be one-year old fish) is highly variable in space and time, and may be related to vessel selectivity: there was little consistency in months or areas where these fish appeared in a given year, and only a small influence from sea-surface temperature on cohort abundance.

A more flexible model was fitted that avoided the assumption of specific cohorts. The model was used to investigate if standardising length compositions for environmental conditions and time- and area of fishing trips would lead to materially different predicted length compositions across the New Zealand troll fisheries from the compositions that have been reported based on scaled catch-sampling. All analyses pointed to fishing vessels contributing substantially to variability in observed compositions, suggesting that catch sampling effort should be spread across as many vessels as possible to ensure representative length compositions can be achieved. This analysis also confirmed findings from the mixture model, suggesting that spatio-temporal patterns in length compositions are significant, but cannot be explained by sea-surface temperature alone. Nevertheless, predicted length-compositions for the total catch effort of the NZ troll fishery suggested that standardising compositions has a minor impact. Significantly, the 2018 length-compositions remain largely unchanged by the standardisation approach, confirming abnormally high proportions of one-year-old fish in that year, with correspondingly lower proportions of older fish. However, both catch and CPUE were high in 2018, and compositions scaled to catch suggest that the composition in 2018 may be a reflection of abnormally high abundance of young fish in New Zealand waters in 2018, rather than abnormally low abundance of other cohorts. The result highlights the importance of interpreting the information of local length compositions in the light of local availability and catch trends.