Integrated integral population models

Abstract

SummaryData integration allows obtaining better descriptions and forecasting of a population’s behaviour by incorporating data at both the individual and population levels. Structured population models include the matrix population models (MPMs), which structure a population through a discrete state variable, and the integral population models (IPMs), which use a continuous variable. Two decades ago, the integrated version of MPMs appeared, but their corresponding version for IPMs is still missing.Here, we propose the integrated integral population model (IIPM). This model takes up the ideas behind existing models used to describe and forecast the dynamics of continuously structured populations: IPMs, which use individual data, and inverse IPMs, which use population data. Particularly, we emphasise the construction and fitting of the IIPM under a Bayesian framework and use the Soay sheep database to compare the population dynamics generated by the IIPM and these existing models.The IIPM constructed with the Soay sheep data had a good performance both at the individual (vital rates) and population (size and structure) levels, because, as they are constrained to fit both sets of data, they produce a balanced population dynamics. In turn, the IPM produced the best individual estimates and the worst population estimates, whilst the inverse IPM produced the worst individual estimates and the best population estimates.The objective of a structured population model should be to correctly describe population patterns. IPMs, by not using population data, fail in this objective. An IIPM solves this problem.