Bayesian Hierarchical Modeling of Multi-Site Longitudinal Data: A Study of Horseshoe Crab Spawning Activity in the Delaware Bay from 1999 to 2012

• Activity Number: 608
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Penelope S. Pooler*+ and David R. Smith and Eric P. Smith
• Companies: SUNY Upstate Medical University and USGS and Virginia Tech
• Keywords: Bayesian hierarchical modeling ; longitudinal data ; horseshoe crabs

Abstract

Bayesian hierarchical modeling is an ideal tool for analyzing complex multi-site longitudinal data. As an example, we analyze data from the ongoing Delaware Bay horseshoe crab (Limulus polyphemus) spawning survey which originated in 1999. This survey uses volunteers to survey beaches throughout the bay during the spring spawning season each year. In 2012, twenty-five beaches were sampled on 12 nights around the new and full moons, although there were some missing values. We developed a Bayesian hierarchical model that examines trends in density and variability in female and male spawners. Our model correlates male and female spawning patterns with location in the bay, wind, and temperature. In addition, we estimate trend differences in spawning behavior between the two sexes as changes in fishery regulations have changed population dynamics. We present the details of our Bayesian model, how it addresses the complexity of this long term ongoing study, and our model findings. Lastly, we provide insights into how this model format can be applied to longitudinal multi-site questions in other disciplines, such as public health, and discuss its advantages over other modeling approaches.