Simulation modeling of a trawl-acoustic survey design for patchily distributed species

Abstract

Spatially patchy populations present challenges for precise estimation of abundance from surveys, which typically result in high estimation errors compared to surveys of more evenly distributed species. In this study, we used simulations to evaluate the performance of the Trawl and Acoustic Presence/Absence Survey design (TAPAS) in reducing the variability in estimated biomass. This approach is a double sampling design in which high-density patches observed in a first phase using hydroacoustics are subsequently more intensively sampled (relative to non-patch areas) in a second phase using trawls and area-swept methods. Information on Gulf of Alaska Pacific ocean perch (Sebastes alutus), a patchily distributed rockfish species, was used to parameterize the simulations. The performance of the TAPAS design depends upon the degree to which high acoustic backscatter represents “patch” areas of high density and variability, as the relationship between backscatter and abundance of a given species can be affected by areas unavailable for sampling (i.e., the “dead zone”) and the contribution of multiple species to the backscatter intensity. With a strong relationship between backscatter intensity and density, the TAPAS design resulted in improved precision compared to simple random sampling (SRS). Additionally, more intensive sampling of the patches occurred when areas of high backscatter intensity were randomly distributed over space rather than located in spatially correlated clusters. However, with a weak relationship between backscatter intensity and density, the precision of the TAPAS design was not substantially larger than SRS. The potential improvement in precision when a strong relationship exists between backscatter intensity and fish density offers motivation to continue to refine relationships between underlying fish density, acoustic backscatter, and trawl catches.