Meeting Abstract

18.4  Saturday, Jan. 4 11:00  Using a Hidden Markov Model to find foraging areas for the gray smooth-hound shark (Mustelus californicus) in a Full Tidal Basin NEEMAN, N*; ESPINOZA, M; FARRUGIA, TJ; LOWE, CG; SOBEL, MJ; O’CONNOR, MP; Drexel University; California State University, Long Beach; California State University, Long Beach; California State University, Long Beach; Temple University; Drexel University nn72@drexel.edu

Analyzing movement pathways can allow unobservable, underlying, discrete behavioral states to be inferred from tracking data. This can be used to understand how animals use their habitats, for example where they tend to forage vs. what areas they use as transit corridors. This study analyzes acoustic telemetry data for the benthic, coastal predator Mustelus californicus (gray smooth-hound shark) to determine if and where they forage within the Full Tidal Basin of Bolsa Chica, Huntington Beach, California and whether or not this varies by individual shark. Preliminary analysis of the data for all the sharks showed that their speeds are a mixture of two log-normal distributions (log-mean -3.8 and -1.6) and that turning angles vary with speed such that at lower speeds the turning angles are uniformly distributed and at higher speeds the turning angles are a mixture of two normal distributions (centered around 0 and 180 degrees). Normally, analysis requires interpolating so the data is at regular time intervals. However, since there is abundant data and since the turn angles close to 180 degrees might be indicative of behavioral state, this analysis used the data as irregular time intervals and used a Hidden Markov model evaluated using particle filters to determine the behavioral state at each location. Preliminary results show that individual sharks have clustered foraging sites, more analysis is needed to determine how different these clustered sites are between individuals.