A hierarchical population model for the estimation of latent prey abundance and demographic rates of a nomadic predator

Linking the demographic parameters underlying population change to environmental conditions is a central goal of population demography. However, multicollinearity among processes in ecological studies can complicate parameter estimation and inference. We sought to demonstrate the use of structural equation modelling, a technique for estimating hypothesized causal pathways among collinear observed and unobserved variables, in the context of integrated population models. We monitored a population of Tengmalm's owls(Aegolius funereus) breeding in the Jura Mountains of northwestern Switzerland and eastern France for 31 years (1990-2020) and collected data on captured prey items. We use concepts central to structural equation models (i.e., latent variables) and integrated population models to estimate the effects of latent prey abundance on Tengmalm's owl demographic parameters. We observed strong positive effects of latent prey abundance during time t on clutch size, fledging probability, and immigration rates into the breeding population, and strong effects of positive changes in latent prey abundance from time t to t+1 on first-year and adult survival. We also observed long-term declines in immigration into the study area. Our work provides a straightforward example of incorporating concepts central to structural equation models (e.g., latent variables) to model environmental processes underlying demographic rates in integrated population models, and has interesting implications for metapopulation ecology of Tengmalm's owl populations in Europe.