Disentangling scale dependencies in species environmental niches and distributions

Abstract

Understanding species’ responses to environmental conditions, and how these ­species–environment associations shape spatial distributions, are longstanding goals in ecology and biogeography. However, an essential component of species–environment relationships – the spatial unit, or grain, at which they operate – remains unresolved. We identify three components of scale‐dependence in analyses of species–environment associations: 1) response grain, the grain at which species respond most strongly to their environment; 2) environment spatial structure, the pattern of spatial autocorrelation intrinsic to an environmental factor; and 3) analysis grain, the grain at which analyses are conducted and ecological inferences are made. We introduce a novel conceptual framework that defines these scale components in the context of analyzing species–environment relationships, and provide theoretical examples of their interactions for species with various ecological attributes. We then use a virtual species approach to investigate the impacts of each component on common methods of measuring and predicting species–environment relationships. We find that environment spatial structure has a substantial impact on the ability of even simple, univariate species distribution models (SDMs) to recover known species–­environment associations at coarse analysis grains. For simulated environments with ‘fine’ and ‘intermediate’ spatial structure, model explanatory power, and the frequency with which simple SDMs correctly estimated a virtual species’ response to the simulated environment, dramatically declined as analysis grain increased. Informed by these results, we use a scaling analysis to identify maximum analysis grains for individual environmental factors, and a scale optimization procedure to determine the grain of maximum predictive accuracy. Implementing these analysis grain thresholds and model performance standards in an example east African study system yields more accurate distribution predictions, compared to SDMs independently constructed at arbitrary analysis grains. Finally, we integrate our conceptual framework with virtual and empirical results to provide practical recommendations for researchers asking common questions about species–environment relationships.

Testing methods in species distribution modelling using virtual species: what have we learnt and what are we missing?

“…4a): when using a threshold simulation approach, presence-absence predictions are almost perfect at all scales, especially at the finest resolution, but when a probabilistic approach is used, AUC is actually poorer at the finest resolution and increases as we up-scale the data, although variability between runs also increases. Overall, however, when using a threshold simulation approach on this noisy gradient, it is always better to use the finer resolution dataset, as concluded by Mertes and Jetz (2018). When we apply the same methods to a noisy gradient (Fig.…”

“…Effects of resolution and extent, upscaling and downscaling Bombi and D'Amen 2012, Lauzeral et al 2013, Nakazawa and Peterson 2015, Fernandez et al 2017, Connor et al 2018, Mertes and Jetz 2018, Moudry et al 2018 • Predictions are best when the layers used for model calibration are at the same resolution than the species response, and when the whole extent of the species distribution is included. Species prevalence or rarity Real et al 2006, Albert and Thuiller 2008, Jimenez-Valverde et al 2009, Meynard and Kaplan 2012, Fukuda and De Baets 2016 • There is a strong effect of sample bias when sample prevalence is different from species prevalence.…”

“…1, stage 2). However, simulating the landscape itself can have some advantages when we are trying to separate the effects of a particular method from the effects of environmental structure (Thibaud et al 2014, Yackulic and Ginsberg 2016, Mertes and Jetz 2018. 1, stage 2).…”

“…This shows a disconnect between niche and metapopulation theories on the one hand, which usually assume bell-shaped environment-occurrence relationships and probabilistic occurrence patterns (Hanski 1994, Austin et al 2006, Austin 2007, Soberón 2010, and the VS simulation strategy on the other. If we look at threshold VS studies, they usually conclude that using finegrain datasets always results in increased classification rates and better model performance as compared to datasets that have been up-scaled to coarser resolutions (Fernandez et al 2017, Mertes andJetz 2018). VS studies dealing with scaling issues provide for a good example of how this simulation choice (threshold versus probabilistic simulation approach) can feed into scaling theories, resulting in opposite expectations and becoming ingrained in the current scientific paradigm.…”

Fine scale waterbody data improve prediction of waterbird occurrence despite coarse species data

“…recent data collected using modern technologies versus historical records), species characteristics (e.g. 1-2, Table 1), it is important to find the optimal spatial resolution of environmental predictors to match the available species data in order to achieve the highest possible HSM prediction accuracy (Guisan et al 2007, Gottschalk et al 2011, Lechner et al 2012a,b, Cord et al 2014, Mertes and Jetz 2018. Similarly, the spatial resolution of environmental predictors can range from centimeters to hundreds of kilometers, depending on the method of data acquisition (e.g.…”