A restructured Bayesian approach to estimate the abundance of a rare and invasive fish

Quantifying invasive species abundance informs management and control strategies. However, estimating abundance can be challenging, particularly when dealing with rare species early in the invasion process. Data generated from control strategies, such as removing invasive species, are usually not suited to conventional statistical modelling approaches. Hence, we developed a Bayesian model using data generated by a grass carp ( Ctenopharyngodon idella ) control program in the Sandusky River, Ohio (USA) for estimating the abundance of rare, invasive species. The model is a restructured N-mixture model modified to incorporate the data generating process (i.e., setting a trammel net to isolate a sampling area followed by boat-mounted electrofishing). Allowing the estimation of grass carp abundance from the species removal data, which had very few detections relative to the sampling effort. Our results indicated that the average number of grass carp present in the river at any one time did not change substantially from 2018 to 2020. The highest abundance estimates were in the lower and upper-middle segments of the river, suggesting possible recolonization from Lake Erie, and possibly other tributaries. Ultimately, the ability to use species-control data to estimate abundance provides important information for management, particularly for invasive ‘sleeper’ species in freshwater ecosystems.