A novel ensemble framework for background selection to improve species distribution models

Global occurrence coordinates of the taro caterpillar (Spodoptera litura Fabricius, 1775), the spotted lanternfly (Lycorma delicatula White, 1845), and the red imported fire ant (Solenopsis invicta Buren, 1972) were obtained from the Global Biodiversity Information Facility(GBIF) to predict the potential distribution under climate change. Spatial filtering was performed using ArcGIS to reduce sampling bias and improve model performance. Meteorological were obtained from 1981 to 2010 at a 2.5 minute resolution, were obtained from CHLESA (www.chelsa-climate.org). After extracting the data related to land climate, the units were changed to the International System of Units using a raster calculator in ArcGIS. Then, Then, the climate data was converted into 19 bioclimatic variables by using the “biovars” function in the “dismo” package of R software. Additinally, the meteorological data were processed to have a .mm file format for CLIMEX model using MetManager, a tool provided by CLIMEX. We used ENMeval in the R package to determine the model features and regularization multiplier (RM). We also additionally used R for random forest. Files and variables File: occ_and_bg.zip: Occurrence and absence data for MaxEnt or random forest models. File: code_R.zip: Example code for selecting optimal model structure of MaxEnt and operating Random forest. File: Bioclimatic variables.zip: Selected bioclimatic variables for the models of three species. File: Model results.zip: The species distribution model results of three species Access information Other publicly accessible locations of the data: GBIF.org (3 May 2023) GBIF Occurrence Download https://doi.org/10.15468/dl.38d8u2 GBIF.org (31 July 2023) GBIF Occurrence Download https://doi.org/10.15468/dl.dryurf GBIF.org (9 August 2023) GBIF Occurrence Download https://doi.org/10.15468/dl.ps6k67