Integrated survey methods to estimate harvest by marine recreational fishers in New Zealand

Marine recreational fishing is a popular pastime in a growing number of countries. Obtaining reliable harvests estimates is important to produce more accurate stock assessments and more certain management decisions, however, accurate measurement of marine recreational harvest is challenging. Previous national fisher diary surveys undertaken in New Zealand during the 1990s gave inconstant estimates of marine recreational harvests. Landline telephone listings and interviews were used to estimate the proportion of New Zealand residents who had fished during the previous 12 months and to recruit diarists. Slight changes in survey method produced variable and at times implausible results. After three years of planning and pre-testing a large-scale project was undertaken to develop a robust off-site harvest survey method and corroborate the results using with two concurrent on-site survey methods. For the off-site survey, the method was based on a national population proportionate sample of dwellings to recruit a panel of 7000 fishers and 3000 non-fishers using a face-to-face household survey. Panellists were contacted regularly by SMS and telephone for a year with a 94% completion rate. Computer assisted telephone interviews collected details of all species of fish harvested by fishing method. The second was a regional aerial-access survey that collected peak period vessel counts from the air to scale up boat-based harvest from concurrent all-day creel surveys on 45 days. Harvest estimates were generated for the most commonly encountered species, snapper, kahawai, trevally, tarakihi and red gurnard. The third and smallest survey was a combined access point survey in a sub-region using fixed and bus route creel surveys covering all significant access points on different set of random stratified days to the areal access survey. The main objective was to estimate the boat-based harvest by specialist fishers targeting scallop and rock lobster. The three concurrent surveys were designed to generate harvest estimates by fishing platform (boat or land based) at overlapping spatial scales. Harvest, in numbers of fish, were estimated independently for recreational fishers using boats. However, the on-site surveys relied on the proportion of harvest from land-based platforms provided by the off-site survey to derive total regional harvest estimates for all methods. The off-site panel survey relied on average weight data for each fish stock provided by the on-site surveys to convert harvest numbers to weight for management purposes. Choosing a sample frame and survey method that is reliable and repeatable into the future is critical to providing comparable estimates and the ability to monitor trends over time. Harvest estimates for the most common species in Fisheries Management Area 1, snapper and kahawai, were very similar. The estimates for snapper ranged from 3754 t (cv 0.06) to 3981 t (cv 0.08) and for kahawai 983 t (cv 0.32) to 942 t (cv 0.08). There were greater differences in estimates between surveys for secondary species. Each survey had independent error structures and this multi-method approach has provided valuable insight into likely sources of bias. High quality recreational harvest estimates are important to support management changes in high profile fisheries.