ssdm: An r package to predict distribution of species richness and composition based on stacked species distribution models

Abstract

Abstract1. There is growing interest among conservationists in biodiversity mapping based on stacked species distribution models (SSDMs), a method that combines multiple individual species distribution models to produce a community-level model. However, no user-friendly interface specifically designed to provide the basic tools needed to fit such models was available until now.2. The "ssdm" package is a computer platform implemented in r providing a range of methodological approaches and parameterisation at each step in building the SSDM: e.g. pseudo-absence selection, variable contribution and model accuracy assessment, inter-model consensus forecasting, species assembly design, and calculation of weighted endemism.3. The object-oriented design of the package is such that: users can modify existing methods, extend the framework by implementing new methods, and share them to be reproduced by others.4. The package includes a graphical user interface to extend the use of SSDMs to a wide range of conservation scientists and practitioners.

A comparison of macroecological and stacked species distribution models to predict future global terrestrial vertebrate richness

“…SDMs are based on statistical correlations between species occurrences and environmental predictor variables (Elith & Leathwick, 2009;Guisan & Thuiller, 2005;Guisan & Zimmermann, 2000), which can then be transferred into future time periods to predict future species distributions under climate change. Recently there has been a rise in the number of tools available to facilitate the implementation of SDMs (Golding et al, 2018;Naimi & Araújo, 2016;Schmitt, Pouteau, Justeau, de Boissieu, & Birnbaum, 2017;Thuiller, Lafourcade, Engler, & Araújo, 2009). …”

How to predict biodiversity in space? An evaluation of modelling approaches in marine ecosystems

“…On the other hand, MEM relied on elaborate community‐level inventories along environmental gradients, which limited it to predict beyond known communities (Ferrier & Guisan, ). As such, SSDM and JSDM were more flexible for biodiversity extrapolation, when ecological niches of individual species were reliable in predictions (Schmitt, Pouteau, Justeau, Boissieu, & Birnbaum, ). It is worthy to note that their predictions tended to be biased without considering macroecological constraints (D'Amen et al, ), for example, overestimating α‐diversity (Guisan & Rahbek, ; Schmitt et al, ).…”

Shifted distribution baselines: neglecting long-term biodiversity records risks overlooking potentially suitable habitat for conservation management

“…We created ensemble HSM [51] for each species by assembling five statistical methods (GAM, MAXENT,192 MARS, RF and GBM) to account for inter-model variability, using the ssdm package [52]. We ran ten 193 repetitions for each of the algorithms and produced an average of the models' outputs, weighting each 194 model according to its predictive ability.…”

“…Stacked-species distribution models (SSDM) combine multiple individual HSMs to produce a community-208 level model and predictive maps of potential species richness [54]. We used the ssdm package [52] to 209 compute maps of local species richness by summing the probabilities from continuous habitat suitability 210 maps provided by the ensemble HSMs, a method that performs better than stacking methods based on 211 thresholding site-level occurrence probabilities [55]. To highlight areas where the potential species 212 diversity is under-or over-estimated because of neglecting long-term occurrence records in the models, 213…”