Ensemble modelling of species distribution: the effects of geographical and environmental ranges

Abstract

The aim of this study was to analyse the effects of species geographical and environmental ranges on the predictive performances of species distribution models (SDMs). We explored the usefulness of ensemble modelling approaches and tested whether species attributes influenced the outcomes of such approaches. Eight SDMs were used to model the current distribution of 35 fish species at 1110 stream sections in France. We first quantified the consensus among the resulting set of predictions for each fish species. Next, we created an average model by taking the average of the individual model predictions and tested whether the average model improved the predictive performances of single SDMs. Lastly, we described the ranges of fish species along four gradients: latitudinal, thermal, stream gradient (i.e. upstream‐downstream) and elevation. After accounting for the effects of phylogenetic relatedness and species prevalence, these four species attributes were related to the observed variations in both consensus among SDMs and predictive performances by using generalized estimation equations. Our results highlight the usefulness of ensemble approaches for identifying geographical areas of agreement among predictions. Although the geographical extent of species had no effect on the performances of SDMs, we demonstrated that more consensual and accurate predictions were obtained for species with low thermal and elevation ranges, validating the hypothesis that specialist species yield models with higher accuracy than generalist ones. We emphasized that significant improvements in the accuracy of SDMs can be achieved by using an average model. Furthermore, these improvements were higher for species with smaller ranges along the four gradients studied. The geographical extent and ranges of species along environmental gradients provide promising insights into our understanding of uncertainties in species distribution modelling.

Uncertainty in ensemble modelling of large-scale species distribution: Effects from species characteristics and model techniques

“…Nevertheless, it was not so surprise cause RF model gives the predictions by generating thousands of trees and aggregated with an average (Breiman, 2001), and the algorithm allow the model to avoid over-fit, this procedure could improve the predictive performance and reduce the variance (Elith et al, 2008). Thus, RF could be a robust technical modelling for species distribution prediction (He et al, 2010;Cheng et al, 2012;Grenouillet et al, 2011). Actually, plenty of publications have noted the algorithm which Random Forest relied on, they thus present the ensemble modelling framework which aggregated several single models and given the average or consensus results (Araújo and New, 2007).…”

“…Several former studies have verified that among plenty of mathematical models, only RF could show the equal performance with the average outcomes of several model techniques. Therefore ensemble modelling was also regarded as the best solutions to reduce the single model uncertainties and bias (Grenouillet et al, 2011;Buisson et al, 2010).…”

“…However, very few researches quantify both the uncertainties from species characteristic and model techniques synthetically, especially in aquatic ecosystems (but see Grenouillet et al, 2011). In addition, former research only conducted in a small geographical scale, which would be constrained by the country border and thus hide the effects of geographical and environmental ranges (Grenouillet et al, 2011). While it is worth noting that the factors determine species distributions vary according to the scale of analysis (Whittaker et al, 2007).…”

“…Up until now, model techniques, data characteristics, species prevalence, latitudinal range, spatial autocorrelation, rarity and species environmental ranges have all considered to be the potential uncertainty source which could affect the model performance (Manel et al, 2001;Segurado and Araujo, 2004;Boone and Krohn, 1999;Marmion et al, 2009a;Marmion et al, 2009b). However, very few researches quantify both the uncertainties from species characteristic and model techniques synthetically, especially in aquatic ecosystems (but see Grenouillet et al, 2011). In addition, former research only conducted in a small geographical scale, which would be constrained by the country border and thus hide the effects of geographical and environmental ranges (Grenouillet et al, 2011).…”

Predicting fish species richness and assemblages with climatic, geographic and morphometric factors: A broad-scale study in Chinese lakes

“…In fact, this phenomenon is in accordance with some previews studies (Manel et al, 2001;Liu et al, 2005;Gevery et al, 2009). The species' environmental range could also be one of the main factors which drive the uncertainty of the prediction of the three species in this study, as numerous studies validate that species with a smaller range can be better predicted than species with a larger environmental range (Hernandez et al, 2006;Grenouillet et al, 2011).…”