Code and data for "Challenging and diagnosing structured population models by testing predictions from stochastic demography," Ellner et al., 2025

Structured population models are parameterized to accurately projectexpected population sizes, stage/state distributions, and population growth rates, but they also predict the variation in outcomes amongindividuals, such as the variance and skewness of reproductive output (LRO) and lifespan, the probability of never reproducing, and many other life history metrics. Testing these predictions about individual outcomes can serve as ``stress tests'': a fitted structured population model that mispredicts individual outcomes is a model with problems. Moreover, the ways in which predictions fail may inform us about what the problems are, help us decide whether the problems are important for the model’s intended applications, and guide efforts to fix them.We present case studies (including zooplankton, plants, and mammals) to demonstrate how structured population models can be tested by comparing individual-level predictions against individual-level data. Some general themes emerge: i) We often detect un-modeled individual heterogeneity, ii) Un-modeled senescence can affect higher moments of lifespan even when lower moments and LRO are predicted well. iii) Fitting one parametric model to multiple clones, species, locations, etc. can cause errors about populations for which the model is insufficiently flexible. Structured population models are ``workhorses'' for ecology: these methods can help ensure their reliability.