Partitioning prediction uncertainty in climate-dependent population models

Abstract

The science of complex systems is increasingly asked to forecast the consequences of climate change. As a result, scientists are now engaged in making predictions about an uncertain future, which entails the efficient communication of this uncertainty. Here we show the benefits of hierarchically decomposing the uncertainty in predicted changes in animal population size into its components due to structural uncertainty in climate scenarios (greenhouse gas emissions and global circulation models), structural uncertainty in the demographic model, climatic stochasticity, environmental stochasticity unexplained by climate-demographic trait relationships, and sampling variance in demographic parameter estimates. We quantify components of uncertainty surrounding the future abundance of a migratory bird, the greater snow goose (Chen caeruslescens atlantica), using a process-based demographic model covering their full annual cycle. Our model predicts a slow population increase but with a large prediction uncertainty. As expected from theoretical variance decomposition rules, the contribution of sampling variance to prediction uncertainty rapidly overcomes that of process variance and dominates. Among the sources of process variance, uncertainty in the climate scenarios contributed less than 3% of the total prediction variance over a 40-year period, much less than environmental stochasticity. Our study exemplifies opportunities to improve the forecasting of complex systems using long-term studies and the challenges inherent to predicting the future of stochastic systems.

Reproductive success of a keystone herbivore is more variable and responsive to climate in habitats with lower resource diversity

“…Our results are consistent with a number of other studies that have reported positive effects of seasonal warming on reproductive success and overall population growth rate in geese (e.g. Alisauskas, ; Gauthier, Péron, Lebreton, Grenier, & van Oudenhove, ; Jensen, Madsen, Johnson, & Tamstorf, ; Morrissette et al., ). However, our study also suggests these effects will not be uniform across the breeding grounds.…”

“…Further studies that integrate these effects across the entire life cycle (e.g. Gauthier et al., ; Jenouvrier et al., ) will be crucial for forecasting population trends and spatial distributions and for prioritizing management in rapidly changing climates.…”

Survival and breeding interval of an endangered marine vertebrate, the leatherback turtle Dermochelys coriacea, in French Guiana

“…A priority will be to search for environmental correlates of this variation, although our demographic parameter estimates are based on a mix of individuals at sea and close to shore, and with different lengths of time since their last reproductive season. Links with local co variates such as indicators of beach erosion and with fishing and oceanographic covariates such as the North Atlantic Oscillation index have to be investigated, with the hope of being able to project demographic results based on climate models (Gauthier et al 2016). The choice of such covariates will benefit from the knowledge of the geographical trajectories of individuals marked with satellite transmitters (Fossette et al 2010, Galli et al 2012, Baudouin et al 2015, 2016a,b, 2017, Treasure et al 2017, Nivière et al 2018.…”

“…As a consequence, recent decades have seen the development of a number of long-term capture-mark-recapture (CMR) monitoring programs (Clutton-Brock & Sheldon 2010) and the joint development of increasingly sophisticated statistical models to analyze the resulting data (Lebreton et al 2009. In this context, the assessment of relationships between demographic parameters and environmental covariates, in particular those related to climate, must be based on a large number of years to be useful for population projections (Gauthier et al 2016), en hancing the value of long-term capture-recapture programs. The leatherback turtle Dermochelys coriacea (leatherback hereafter) is particularly challenging: the females come ashore to breed only intermittently, with several clutches within a season followed by one or usually several seasons without reproduction.…”

A mechanistic–statistical species distribution model to explain and forecast wolf (Canis lupus) colonization in South-Eastern France

“…Mechanistic spatio-temporal models have been developed to offer an alternative to regression-based SDMs that encounter difficulties associated colonization as a consequence of dispersal processes (Hefley et al 2017b). Mechanistic models are based on biological processes, such as survival or dispersal, describing processes through which environmental factors affect a biological system of interest (Morin & Thuiller 2009;Mouquet et al 2015;Gauthier et al 2016). SDMs accounting for dynamic mechanisms are relevant tools to assess ecological colonization, because they improve our ability to get predictions in space and time and at the same time include reliable measures of prediction errors (Williams et al 2017).…”