Invasive alien species introduced to ports through cargo containers have destroyed the biodiversity worldwide. The introduction risk at ports must be estimated to control the early stages of invasion. However, limited data are available for this estimation in the introduction stage. Spatial statistical models have been used to address the lack of information by considering the observations of neighbors or integrating multiple data sources based on the assumption of spatial correlation. Unlike natural dispersal, methods to address these issues have not yet been established, because the spatial correlation between ports based on the geographical distance is not assumed for human-mediated species introduction through container cargo. Herein, we propose a multivariate conditional autoregressive model that considers a logistic network in order to integrate multiple data sources and estimate introduction risk. A relationship between locations based on logistics connectivity is assumed rather than the spatial correlation based on the geographical distance used in the past. Hierarchical Bayesian models integrating data through the network were implemented for two fire ant species (Solenopsis invicta and Solenopsis geminata) observed in Japanese ports. We observed that the proposed joint models improved the fit compared to conventional models estimated from a single dataset. This finding suggests that integrating data from multiple species or data types based on a network helps to address the lack of observations. This is one of the first studies to demonstrate the effectiveness of multivariate conditional autoregressive model in considering biological invasion networks and contributes to the development of reliable biosecurity strategies.
