Search

Forecasting distribution shifts under novel environmental conditions is a major task for ecologists and conservationists. Researchers forecast distribution shifts using several tools including: predicting from an empirical relationship between a summary of distribution (population centroid) and annual time series ("annual regression," AR); or fitting a habitat‐envelope model to historical distribution and forecasting given predictions of future environmental conditions ("habitat envelope," HE). However, surprisingly little research has estimated forecast skill by fitting to historical data, forecasting distribution shifts and comparing forecasts with subsequent observations of distribution shifts. I demonstrate the important role of retrospective skill testing by forecasting poleward movement over 1‐, 2‐ or 3‐year periods for 20 fish and crab species in the Eastern Bering Sea and comparing forecasts with observed shifts. I specifically introduce an alternative vector‐autoregressive spatio‐temporal (VAST) forecasting model, which can include species temperature responses, and compare skill for AR, HE and VAST forecasts. Results show that the HE forecast has 30%–43% greater variance than predicting that future distribution is identical to the estimated distribution in the final year (a "persistence" forecast). Meanwhile, the AR explains 2%–6% and VAST explains 8%–25% of variance in poleward movement, and both have better performance than a persistence forecast. HE and AR both generate forecast intervals that are too narrow, while VAST models with or without temperature have appropriate width for forecast intervals. Retrospective skill testing for more regions and taxa should be used as a test bed to guide future improvements in methods for forecasting distribution shifts. [ABSTRACT FROM AUTHOR]

Research has estimated associations between water temperature and the spatial distribution of marine fishes based upon correlations between temperature and the centroid of fish distribution (centre of gravity, COG). Analysts have then projected future water temperatures to forecast shifts in COG, but often neglected to demonstrate that temperature explains a substantial portion of historical distribution shifts. We argue that estimating the proportion of observed distributional shifts that can be attributed to temperature vs. other factors is a critical first step in forecasting future changes. We illustrate this approach using Gadus chalcogrammus (Walleye pollock) in the Eastern Bering Sea, and use a vector-autoregressive spatiotemporal model to attribute variation in COG from 1982 to 2015 to three factors: local or regional changes in surface and bottom temperature ('temperature effects'), fluctuations in size-structure that cause COG to be skewed towards juvenile or adult habitats ('size-structured effects') or otherwise unexplained spatiotemporal variation in distribution ('unexplained effects'). We find that the majority of variation in COG (including the north-west trend since 1982) is largely unexplained by temperature or size-structured effects. Temperature alone generates a small portion of primarily north-south variation in COG, while size-structured effects generate a small portion of east-west variation. We therefore conclude that projections of future distribution based on temperature alone are likely to miss a substantial portion of both the interannual variation and interdecadal trends in COG for this species. More generally, we suggest that decomposing variation in COG into multiple causal factors is a vital first step for projecting likely impacts of temperature change. [ABSTRACT FROM AUTHOR]

Aim Spatial analysis of the distribution and density of species is of continuing interest within theoretical and applied ecology. Species distribution models (SDMs) are being increasingly used to analyse count, presence-absence and presence-only data sets. There is a growing literature on dynamic SDMs (which incorporate temporal variation in species distribution), joint SDMs (which simultaneously analyse the correlated distribution of multiple species) and geostatistical models (which account for similarity between nearby sites caused by unobserved covariates). However, no previous study has combined all three attributes within a single framework. Innovation We develop spatial dynamic factor analysis for use as a 'joint, dynamic SDM' (JDSDM), which uses geostatistical methods to account for spatial similarity when estimating one or more 'factors'. Each factor evolves over time following a density-dependent (Gompertz) process, and the log-density of each species is approximated as a linear combination of different factors. We demonstrate a JDSDM using two multispecies case studies (an annual survey of bottom-associated species in the Bering Sea and a seasonal survey of butterfly density in the continental USA), and also provide our code publicly as an R package. Main conclusions Case study applications show that that JDSDMs can be used for species ordination, i.e. showing that dynamics for butterfly species within the same genus are significantly more correlated than for species from different genera. We also demonstrate how JDSDMs can rapidly identify dominant patterns in community dynamics, including the decline and recovery of several Bering Sea fishes since 2008, and the 'flight curves' typical of early or late-emerging butterflies. We conclude by suggesting future research that could incorporate phylogenetic relatedness or functional similarity, and propose that our approach could be used to monitor community dynamics at large spatial and temporal scales. [ABSTRACT FROM AUTHOR]