Design and Model-based Approaches for Estimating Abundance of American Horseshoe Crab

Abstract

The American horseshoe crab (HSC), Limulus polyphemus, is one of four species of horseshoe crabs found throughout the world, and the only one found in North America. It is an economically and ecologically important species throughout its native range from Maine to the Yucatan Peninsula. Harvested for fertilizer and livestock feed in the 19th century, the species is now harvested as bait for whelk and eel fisheries, and for their blood by the biomedical industry. The Atlantic States Marine Fisheries Commission (ASMFC) started to formally manage HSC in 1998 with its Interstate Fisheries Management Plan (IFMP). Unique emphasis and harvest limits have been placed on the Delaware Bay stock, as it is commercially exploited and a critical food source for the threatened red knot, Calidris canutus rufa. Previously, estimates of relative and total abundance of HSC in the Delaware Bay area were based on a design-based approach using a stratified random sampling design. In Chapter 1 of this work, I developed hurdle models for each of the six HSC demographic groups to standardize catch-per-unit-effort (CPUE) and estimate relative abundance using a model-based approach. It was determined that while the two approaches resulted in mostly convergent estimates of relative abundance, external factors such as month, time-of-day, and average depth have major effects on the observed CPUE of all demographic groups. Chapter 2 involved the development of hurdle models for the three species of bycatch frequently caught in our trawls, channeled whelk (Busysotypus canaliculatus), knobbed whelk (Busycon carica), and summer flounder (Paralichthys dentatus). It was found that channeled whelk relative abundance has been at a historical low since 2016, while summer flounder has been at a consistent high. Recent estimates of knobbed whelk relative abundance have been less variable than previously seen, with estimates since 2016 being similar to those seen before 2012. These results provide the first estimates for whelk population trends in the mid-Atlantic region and add to the growing knowledge of summer flounder relative abundance in the area. In Chapter 3, I applied the hurdle models developed in Chapter 1 to estimate the total abundance of HSC in the Delaware Bay area. For this work, I developed two spatio-temporal variograms to estimate bottom temperature and bottom salinity at unmeasured cells per month in the time series. The results showed that night estimates of total abundance were consistently higher than daytime estimates, and estimates from September or November resulted in the highest estimated catch for all demographic groups. The results suggest that when comparing September model-based estimates at night to those of the design-based approach, nearly a third of all previous design-based estimates significantly underestimated the total abundance of HSC in the Delaware Bay area. This result suggests that the ASMFC can recommend increased harvest limits for mature individuals if that action aligns with the goals of their adaptive resource management (ARM) framework.