Data relating to threats to passion fruit production in the Neotropics due to agricultural area loss and pollinator mismatch as consequence of climate changes

The data in this article are associated with the research article ‘Agricultural area losses and pollinator mismatch due to climate changes endanger passion fruit production in the Neotropics’ (A.D. Bezerra et al. 2019). The data consists of the occurrence points, AUC scores models, presence and absence and co-occurrence maps of the passion fruit (Passiflora edulis) crop and its pollinators, Xylocopa bees (Xylocopa frontalis and Xylocopa grisescens), in current and future scenarios (RPC 4.5 and 8.5, in the years 2060 and 2080) in the Neotropics. Data was obtained though literature review (articles, systematic surveys, dissertation and thesis), as well as systematic searches in entomological collections available in data portals provided by the SpeciesLink and Global Biodiversity Information Facility – GIBF, and analyses by the MaxEnt algorithm and binary transformation. Occurrence error points that did not represent the actual spatial distribution of the species were removed to obtain the current occurrence points and data analyses proved good performance of models for all prediction scenarios. The data-generated maps of pollinators and crop occurrence and co-occurrence also show how climate change may impact the spatial distribution of pollinators and potential losses of this crop's agricultural areas.


a b s t r a c t
The data in this article are associated with the research article 'Agricultural area losses and pollinator mismatch due to climate changes endanger passion fruit production in the Neotropics' (A.D. Bezerra et al. 2019). The data consists of the occurrence points, AUC scores models, presence and absence and co-occurrence maps of the passion fruit (Passiflora edulis) crop and its pollinators, Xylocopa bees (Xylocopa frontalis and Xylocopa grisescens), in current and future scenarios (RPC 4.5 and 8.5, in the years 2060 and 2080) in the Neotropics. Data was obtained though literature review (articles, systematic surveys, dissertation and thesis), as well as systematic searches in entomological collections available in data portals provided by the SpeciesLink and Global Biodiversity Information Facility e GIBF, and analyses by the MaxEnt algorithm and binary transformation. Occurrence error points that did not represent the actual spatial distribution of the species were removed to obtain the current occurrence points and data analyses proved good performance of models for all prediction scenarios. The data-generated maps of pollinators and crop occurrence and co-occurrence also show how climate change may impact the spatial distribution of pollinators and potential losses of this crop's agricultural areas. © 2019 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons. org/licenses/by/4.0/).

Data
The data on records of occurrence points for the passion fruit pollinators, Xylocopa bees (X. frontalis and X. grisescens), were obtained through literature review, systematic surveys, articles, dissertations and theses, as well as any information available in data portals provided by the SpeciesLink and Global Biodiversity Information Facility e GIBF and the centroids of 1195 Brazilian counties where passion fruit has been grown for the past ten years (see spreadsheet in supplementary material). After removing duplicated data, incomplete information or incorrect coordinates, these data showed the actual occurrence points of Xylocopa bees and the passion fruit crop in the Neotropics (Fig. 1, Fig. 2 and Fig. 3). To the spatial distribution modelling, 19 layers of each scenario from WorldClim was used in our analyses, each layer having a spatial resolution of 2.5 arc minutes (Cells with size~4.5 km resolution at the equator). The spatial distribution modelling analyses of all three species produced AUC scores close to 1.0 indicating good performances of the models' prediction (Table 1). More information about the importance of bioclimatic variables is available in Ref. [1]. The maps of species occurrence and cooccurrence indicate relevant effects of climate change in species shift, potential area losses and crop-pollinator mismatch in the future scenarios (Figs. 4e9) (see Fig. 6) (see Fig. 7) (see Fig. 8) (see Fig. 5).

Value of the data
The data present AUC score models and maps of occurrence and co-occurrence for pollinators and the passion fruit crop in the Neotropics for the current and future scenarios of climate changes in 2060 and 2080. The data can be used by researchers and policy makers to assist studies and policies regarding the implications of croppollinator mismatch and losses of agricultural areas; These are the first data showing the current and future effects of climate change acting concomitantly on a tropical crop and its key pollinators. The data show losses of adequate area for the crop and the pollinators over time and increasing crop-pollinator mismatch.

Specifications table
Subject area Agriculture More specific subject area Climate change Type of data Table, figure How data was acquired Literature review (articles, systematic surveys, dissertation and thesis), as well as systematic searches in entomological collections available in data portals provided by the SpeciesLink and Global Biodiversity Information Facility e GIBF.

Data format Filtered and Analyzed Experimental factors
Cleaning and extraction of the occurrence errors points that did not represent the actual spatial distribution of these species Experimental features Data analyses of spatial distribution of the crop and bees using the MaxEnt algorithm Data source location The Neotropics Data accessibility All data are presented in the paper

Biotic and climate data
Data on sampling points of the most common pollinators of passion fruit flowers in Central and South Americas, the carpenter bee species X. frontalis Oliver, 1789 and X. grisescens Lepeletier, 1841 were analyzed for the occurrence and absence of these bee species in the Neotropical region [2e4].
Records of occurrence points for both bee species were obtained through literature reviews, as well as systematic searches in entomological collections available in data portals such as the Specieslink and    6. The presence and absence maps in the current, moderate (RCP 4.5) and pessimistic (RCP 8.5) scenarios in years 2060 and 2080 show the suitable area for growing passion fruit (Passiflora edulis Sims f. flavicarpa) in the Neotropics. All models show that climate change will be jeopardizing passion fruit crop due to the future climatic conditions, which will reduce the suitable cropping areas of passion fruit orchards. When comparing the current scenarios with the futures scenarios for 2060 and 2080 the suitable areas for the passion fruit crop will be reduced in all scenarios, mainly in the pessimistic one (RCP 8.5; 2080) with reductions up to 63.67%%. The potential area losses for cropping passion fruit may be between 42.90% and 64.86%.   Global Biodiversity Information Facility e GIBF. The data generated 371 occurrence points to X. frontalis and 188 occurrence points to X. grisescens (see spreadsheet in supplementary material). Occurrence points found in duplicated, incomplete information or incorrect coordinates were removed [5], and the remaining data showed the actual occurrence points of Xylocopa bees in the Neotropics. Passion fruit coordinate points were obtained in the County Agricultural Production database available from the Brazilian Institute of Geography and Statistics (IBGE) website. The centroids of 1195 counties where passion fruit was cultivated in the last ten years were used as the basis to point coordinates [6].
Layers of bioclimatic variable for current and futures scenarios were obtained through Worldclim [7], each layer having a spatial resolution of 2.5 arc minutes (Cells with size~4.5 km resolution at the equator). A total of 19 layers were used in all scenarios. We obtained forecasts for the future climatic conditions considering the scenarios RCP 4.5 and RCP 8.5 (Representative Concentration Pathways), in 2060 and 2080 developed by Hadley Center Global Environmental Model (HadGEM2-ES)) [8]. Pearson's correlation coefficient and Principal Components Analysis (PCA) were used for each pairwise comparison of the 19 climatic variables to estimate the correlation levels between bioclimatic variables. The most explanatory variable was selected according to PCA, only when the bioclimatic variable presented strong correlation (r!j0.75j) with two or more variables. The number of bioclimatic variables left in the models is available in Ref. [1].

Model building and current and future occurrence
The Species Distribution Modelling (SDM) was developed using the package Dismo v.1.1e4 [9], R language [10]. For this, we used the ecological niche modelling applying the MaxEnt (Maximum entropy) algorithm [11]. The high AUC scores indicated that the model is based on information, if compared to a random model without information, which the AUC score was 0.50. The AUC scores (close to 1.0) indicated a good performance of the models (Table 1).
A random set of pseudo-absence data to each species was required to make the current and futures scenarios maps to the whole study area (Latitude, 32 East to À60 West; Longitude, 24 North to À125 South). We determined the threshold (cutoff) with Dismo v1.1-4 package [9], and with Biomod2 v.3.3e7 package [12] in R language [10], and we perform a binary transformation for each models, the current and future scenario of each map was compared. We also compared the passion fruit map scenarios with X. frontalis and X. grisescens. It was possible to show the suitable areas for each species in the current scenario and the changes in the spatial distribution range of the species in the future scenarios, as well as the overlap area suitable for crop and for the bee species.