Dealing with area-to-point spatial misalignment in species distribution models

Bastien Mourguiart, Mathieu Chevalier, Martin Marzloff, Nathalie Caill-Milly, Kerrie Mengersen, Benoit Liquet

Research output: Working paperPreprint


This repository contains the R-scripts to reproduce the analysis presented in the paper: B. Mourguiart, M. Chevalier, M. Marzloff, N. Caill-Milly, K. Mengersen, B. Liquet. Dealing with area-to-point spatial misalignment in species distribution models. Ecography, (in press), doi: 10.1111/ecog.07104 The study uses a virtual species approach to investigate the effects of area-to-point spatial misalignment (i.e. a resolution or a grain mismatch between coarse-grain covariates and fine-grain response variables) on the explanatory and predictive performance of three species distribution models (SDMs). Specifically the code: 01_simulate_data.R generates virtual environmental and species presence-absence data under different scenarios with area-to-point spatial misalignment (i.e., environmental data with coarser resolution than presence-absence data); 02_run_xxx.R fits the three models: the GLM (generalised linear model) that does not account for spatial misalignment, the spGLM (spatial GLM) that estimates residual spatial autocorrelation not captured by the environmental covariates, and the BEM (Berkson error model) that accounts for area-to-point spatial misalignment; 03_compute_perfmetrics.R calculates performance metrics (e.g., AUC) to assess the explanatory and predictive performance of the three models over different levels of misalignment and spatial heterogeneity; 04_analyze_perfmetrics.R computes summary statistics of the performance metrics; 05_make_outputs.R generates the figures presented in the manuscript and supplementary materials. We recommend to first save the scripts in a folder named "analysis" within an Rproject, and then run the "make.R" file from within the Rproject. This will allow to run the workflow from simulating the data to producing the results presented in the manuscript, and keep the file path names as they are in the scripts.
Original languageEnglish
Publication statusSubmitted - 2024

Publication series



Dive into the research topics of 'Dealing with area-to-point spatial misalignment in species distribution models'. Together they form a unique fingerprint.

Cite this