Integrating multivariate conditionally autoregressive spatial priors into recursive bivariate models for analyzing environmental sensitivity of mussels

Understanding ecological interrelationships between species and their environment is an important part of biological research and in case of productive livestock also economic research.We therefore investigate how adult mussels and mussel seeds (of Mytilus galloprovincialis) interact while simultaneously accounting for environmental factors. For that purpose, we develop a recursive bivariate simultaneous equation model which considers the environmental endogeneity of the mussels. In order to address the potentially highly complex ecological patterns, we include environmental factors via integrating multivariate autoregressive discrete spatial effect priors into the framework of simultaneous equation models. This allows for general correlation structures of the spatial effects and thus avoids the restrictive (implicit) assumption that adult mussels and mussel seeds of the same species react independently to their environment.