Data and code for: Present-day and future global maps of non-native species richness across 17,343 islands worldwide

This repository contains the data and code necessary to reproduce the analyses described in the paper titled "Present-day and future global maps of non-native species richness across 17,343 islands worldwide." The primary objective of this paper was to train a model that enables the prediction of the current (2020) and future (2050) non-native species (NNS) richness for global islands. NNS richness prediction was based on a set of 11 biogeographic and anthropogenic predictors collected at the island level. We assessed the performance of four modeling strategies based on several algorithms and methods for accounting for data biases, and used the best model to predict NNS richness for the present day as well as for two future scenarios. The data set used for building the four modeling strategies and for predicting the NNS richness of all islands for the present and the two future scenarios is available from the file: Observed_NNS_richness_Present_Future_predictors.xlsx. The first sheet, "METADATA," contains the column names and descriptions of each variable from the subsequent sheets. The sheet "Obs_NNS_PRESENT_PRED" contains the identifier (ID column), name, archipelago, the 11 anthropogenic and biogeographic predictors, the sampling effort, and the observed NNS richness of each island for the present day (2020). The sheets "FUTURE_PRED_SSP1" and "FUTURE_PRED_SSP3" contain the identifier (ID column) and the 11 anthropogenic and biogeographic predictors of each island for the SSP1 and SSP3 future scenarios, respectively. Raw data of the predictors were obtained from the sources listed in Table 1 of the paper. They were freely accessible, and we followed the conditions to access them. The scripts to perform the NNS richness prediction are based on this dataset. However, the .rds files can be used directly in the scripts: Observed_NNS_richness_predictors.rds for the present predictors and observed NNS richness, 07_Data_isl_info_SSP1.RDS and 07_Data_isl_info_SSP3.RDS for the predictors for SSP1 and SSP3, respectively. One needs to run the scripts Model_building.R and Predicting_for_all_islands.R first, and then all the other scripts can be run in any order. The output dataset (Predicted_NNS_richness_17343_islands.xlsx) contains the identifier, name, mean predicted NNS richness across the 10 cross-validation folds for the present day and the two future scenarios, along with the associated standard deviation, for the hurdle random forest model, with sampling effort included as an observation weight during the calibration. Column names and descriptions are detailed in the METADATA sheet, and data are displayed in the "Output_Hurdle_RF_weighted_by_SE" sheet.