New Zealand waters support a diverse range of seabird species, which frequently interact with fisheries throughout the region. Some of these interactions result in the incidental capture and mortality of seabirds in different commercial fisheries. Fisheries observers onboard commercial fishing vessels record this bycatch of seabirds, documenting the number and species that are getting caught. These observer records allow the estimation of total seabird captures, forming the basis of bycatch assessments in New Zealand's Exclusive Economic Zone.

The current analysis provides the most recent annual assessment of seabird captures in New Zealand waters, including data from the 2017–18 fishing year. The analysis used statistical models to derive estimates of total seabird captures across all commercial trawl and longline fisheries, applying a unified modelling framework to estimate incidental captures of seabirds. It followed the same approach as recent bycatch assessments, but an overdispersion scaling parameter was included in the current model update. This parameter took a value between zero and two: with a value of zero, the distribution of captures was assumed to follow a negative binomial distribution (as previously); with a value of two, the variance in the distribution of captures was proportional to the mean (similar to a Poisson distribution). The mean value of the posterior distribution of the overdispersion scaling parameters in the fitted models ranged from 0.54 to 1.27. This new parameter resulted in marked decreases in the uncertainty for a number of total seabird capture estimates.

Included in the modelling were ten species and species groups: New Zealand white-capped albatross (Thalassarche steadi), Salvin's albatross (Thalassarche salvini), Buller's albatross (Thalassarche bulleri, combining both southern T. b. bulleri and northern T. b. platei subspecies), white-chinned petrel (Procellaria aequinoctialis), black petrel (Procellaria parkinsoni), grey petrel (Procellaria cinerea), sooty shearwater (Puffinus griseus), flesh-footed shearwater (Puffinus carneipes), and ''other albatrosses'' and “other birds”. The time periods of the current estimation were the fishing years from 2002–03 to 2017–18 for trawl fisheries, and from 1998–99 to 2017–18 for longline fisheries.

There was a total of 3328 (95% c.i.: 2971–3782) estimated seabird captures in trawl and longline fisheries (c.i., credible interval, the 95th quantile range of the posterior distribution) in 2017–18. The total estimate included 1506 (95% c.i.: 1344–1696) seabird captures in trawl fisheries, 1186 (95% c.i.: 913–1589) captures in bottom-longline fisheries, and 635 (95% c.i.: 536–758) captures in surface-longline fisheries. The highest number of total estimated captures in 2017–18 was of white-chinned petrel, with 681 (95% c.i.: 486–1040) estimated captures of this species. The second highest estimate was 462 (95% c.i.: 383– 565) captures of New Zealand white-capped albatross, followed by 426 (95% c.i.: 321–554) captures of flesh-footed shearwater. Capture estimates for other species included 350 (95% c.i.: 256–474) captures of Salvin's albatross, 242 (95% c.i.: 159–369) captures of black petrel, 231 (95% c.i.: 181–295) captures of Buller’s albatrosses, 213 (95% c.i.: 143–322) captures of sooty shearwater, and 139 (95% c.i.: 62– 293) captures of grey petrel. In addition to estimates for individual species, there were 331 (95% c.i.: 255–431) captures of other birds and 249 (95% c.i.: 181–340) captures of other albatrosses.

For eight of the ten modelled species groups, the total number of estimated captures decreased between 2002–03 and 2017–18. The exceptions were white-chinned petrel and grey petrel, for which the total number of estimated captures did not show a clear decrease over this time period. When estimated captures in 2017–18 were compared with estimates in 2006–07, after mandatory mitigation measures were introduced in trawl fisheries in January 2006, the former estimates were similar to those in 2006–07 for many species. Capture estimates for all species groups combined showed a clear decrease over the assessment period, with the lowest mean number of seabird captures in the reporting period in 2017–18. These decreases were largely determined by marked decreases in fishing effort.

There were sufficient numbers of captures in large-vessel fisheries to examine temporal trends in capture rates (number of captures per unit fishing effort). Patterns over time varied, depending on the fishing method and target fishery, and between albatrosses and petrels. These patterns included a decrease in albatross capture rates (number of captures per unit fishing effort) in squid and in hoki trawl fisheries in 2006, immediately following the introduction of mandatory warp mitigation; however, capture rates subsequently increased or showed no clear trend in more recent years. For petrels in the squid trawl fishery, capture rates exhibited a distinct pattern of higher captures in alternate years, but the reason for these fluctuations is unknown.

The estimation of seabird captures relies on data collected by onboard fisheries observers, but observer coverage and effort varied considerably across fishing methods and vessel sizes over the reporting period. Increasing observer coverage in the small-vessel fleets, and ensuring that all vessels have at least some observer coverage, would help to ensure that estimates based on observer data reliably reflect seabird bycatch in New Zealand's commercial trawl and longline fisheries.