N‐SDM: a high‐performance computing pipeline for Nested Species Distribution Modelling

Abstract

Predicting contemporary and future species distributions is relevant for science and decision making, yet the development of high-resolution spatial predictions for numerous taxonomic groups and regions is limited by the scalability of available modelling tools. Uniting species distribution modelling (SDM) techniques into one high-performance computing (HPC) pipeline, we developed N-SDM, an SDM platform aimed at delivering reproducible outputs for standard biodiversity assessments. N-SDM was built around a spatially-nested framework, intended at facilitating the combined use of species occurrence data retrieved from multiple sources and at various spatial scales. N-SDM allows combining two models fitted with species and covariate data retrieved from global to regional scales, which is useful for addressing the issue of spatial niche truncation. The set of state-of-the-art SDM features embodied in N-SDM includes a newly devised covariate selection procedure, five modelling algorithms, an algorithmspecific hyperparameter grid search, and the ensemble of small-models approach. N-SDM is designed to be run on HPC environments, allowing the parallel processing of thousands of species at the same time. All the information required for installing and running N-SDM is openly available on the GitHub repository https://github. com/N-SDM/N-SDM.

A tale of three vines: current and future threats to wild Eurasian grapevine by vineyards and invasive rootstocks

“…Because they are more limited in availability, such fine-scale information can pose challenges related to the 'niche truncation' when modelling species distribution, as it may limit the generalizability of the modelling across the species' entire distribution (Chevalier et al, 2021(Chevalier et al, , 2022.To better incorporate the influence of scale-dependent drivers on species distribution, our comprehensive coarse-scale dataset and predictions can now be combined with fine-grained regional environmental factors using hierarchical data integration methods. Such approaches can enhance the accuracy and robustness of the predictions by capturing the complexity of species-environment relationships across different scales (Adde et al, 2023;Fletcher et al, 2019;Mateo et al, 2019).…”

Global distribution modelling of a conspicuous Gondwanian soil protist reveals latitudinal dispersal limitation and range contraction in response to climate warming

“…Our modelling approach aimed at predicting the past, present and future potential distributions of A. vas based on high-resolution climatic and topographic variables by using N-SDM v1.0.1 (Adde et al, 2023), an end-to-end high-performance computing pipeline for species distribution modelling. Modelling building and analyses were reported in the ODMAP protocol (Zurell et al, 2020) (Appendix S1).…”

“…Since these soil temperature data were available for the current climate solely, past and future spatial projections were made using air data only. To select the best subset of variables to model A. vas potential distribution among the 20 candidates, we used the automated procedure included in the N-SDM workflow with default setting (see Adde et al, 2023 and ODMAP for more details).…”

“…To account for class imbalance, occurrences and background pseudo-absences were equally weighted in the models. For each algorithm, the variable importance was calculated using algorithm-specific measures (see Adde et al, 2023 for more details). Response curves were drawn for each variable and for each algorithm to show how predicted values changed along each variable gradient while keeping all other variable at their mean value (Elith et al, 2005).…”