Assessing benthos through predator stomach contents: spatiotemporal modeling of abundance and habitat use

Abstract

In shelf ecosystems, benthic invertebrates facilitate nutrient recycling and the transfer of energy to higher trophic levels. However, large‐scale monitoring through direct sampling (e.g. using benthic grabs or bottom trawls) can be costly in terms of time and labor. Here, we demonstrate a method for developing standardized abundance indices of forage groups (i.e. species or functional groups preyed upon by predators) based on predator stomach contents. The modeling approach is analogous to methods for estimating abundance indices from fisheries catch‐per‐unit‐effort data; accounts for predator species‐ and size‐specific differences in forage group selectivities, which may vary over space; and permits index estimation when diets are unevenly sampled over space, time, and across predator species. We apply the model to four decades of groundfish diet data from the eastern Bering Sea and estimate abundance indices for nine benthic forage groups. The fitted models were then used to generate: 1) time‐averaged maps of relative biomass density for forage groups and demarcation of potential core habitat areas; 2) region‐wide biomass index time series; and 3) an assessment of bioregions based on benthic forage community composition and change in bioregion area over time. Diet‐based biomass densities were on average correlated well with densities obtained from direct sampling (bottom trawl) for species of Chionoecetes crabs (0.61–0.69). Correlations for polychaetes and bivalves, which had fewer direct survey samples (benthic grabs) were also positive but weaker (0.45 and 0.20, respectively), potentially reflecting larger differences in selectivities between predators and sampling gear or sampling error. We argue that diet data can provide an additional and cost‐efficient lens through which abundances of forage species can be quantified and aid efforts to monitor change in marine ecosystems. Abundance indices can be used in subsequent whole‐of‐ecosystem models, and habitat utilization maps can be used to inform spatial management decisions.