A database of common vampire bat reports

The common vampire bat (Desmodus rotundus) is a sanguivorous (i.e., blood-eating) bat species distributed in the Americas from northern Mexico southwards to central Chile and Argentina. Desmodus rotundus is one of only three mammal species known to feed exclusively on blood, mainly from domestic mammals, although large wildlife and occasionally humans can also serve as a food source. Blood feeding makes D. rotundus an effective transmissor of pathogens to its prey. Consequently, this species is a common target of culling efforts by various individuals and organizations. Nevertheless, little is known about the historical distribution of D. rotundus. Detailed occurrence data are critical for the accurate assessment of past and current distributions of D. rotundus as part of ecological, biogeographical, and epidemiological research. This article presents a dataset of D. rotundus historical occurrence reports, including >39,000 locality reports across the Americas to facilitate the development of spatiotemporal studies of the species. Data are available at 10.6084/m9.figshare.15025296.

undisturbed forests, but also in disturbed areas such as agroforestry plots, silvopastoral systems, pastures, and secondary forests 13,16,17 . Desmodus rotundus usually roosts in small groups, from as few as 10 to a few hundred individuals, but can also be found roosting in groups of up to a few thousand individuals 2 . Desmodus rotundus also uses a variety of roosts, including tree holes, crevices, caves, and abandoned mines and houses 2 . The conservation status of D. rotundus was defined as of "Least Concern" in 2015 by the International Union for Conservation of Nature and Natural Resources (IUCN) Red List of Threatened Species, as it is presumed to be a common species with large and stable populations 18 .
Desmodus rotundus can act as a natural reservoir for various microorganisms with zoonotic potential, such as bacteria, including Bartonella spp. 19 , coronaviruses 20,21 , and rabies virus 12,22 . Bartonella spp. bacteria are globally distributed and have been known to cause endocarditis in humans and other animals 19,23,24 . Endocarditis is an infection of the inner lining of the heart, and can potentially be lethal 23,24 . Furthermore, several variations of coronaviruses have been identified in D. rotundus 20,25 . In the Americas, bats are considered to be a key reservoir of the rabies virus 26,27 , with D. rotundus being the main species responsible for transmitting rabies to livestock 28 . It has been estimated that bovine rabies transmitted by vampire bats causes the death of thousands of cattle annually, resulting in economic losses of hundreds of millions of dollars in Latin America 29,30 . Indirect costs associated with D. rotundus related rabies include the vaccination of millions of cattle as a preventative measure, and post-exposure treatments (rabies immunoglobulin serums and vaccination) for people exposed to D. rotundus bites 29,30 . The perpetuation of rabies in livestock may also be associated with the abundance and distribution of D. rotundus 28,31,32 , as sex-related (male) dispersal may contribute to the expansion of rabies virus into new areas 33 . Thus, the addition of livestock to the landscape promotes suitable conditions for D. rotundus breeding and feeding 28,[34][35][36] .
Due to its reservoir status for potential pathogens, D. rotundus is considered to be a major public health problem in the tropical and subtropical regions of the Americas. Public health concerns are particularly prevalent in Amazonian regions, where many people live in vulnerable housing 37,38 and human diseases transmitted by D. rotundus remain high 39 . In fact, since 2020, rabies has been included in the World Health Organization 2021-2030 road map as a zoonotic disease, and now requires coordination of mitigation strategies at the regional, national, and global levels 40,41 . Numerous outbreaks in rural human communities have been reported in Amazonian regions, including Peru 42 , Brazil 6,43 , and French Guiana 7,44 .
Several Latin American countries have developed programs to reduce the number of D. rotundus bites to humans and livestock 45 . Culling campaigns to reduce D. rotundus populations; however, have not proven useful in reducing the seroprevalence of rabies within vampire bat colonies 46 . It has been suggested that D. rotundus geographic distributional expansion is linked to landscape heterogeneity, degradation, and agricultural aggregations 13,37,47 . Nevertheless, an increase of suitable areas under future climatic scenarios may contribute to the increased risk of rabies in some regions of the Americas 34 . The study and analysis of D. rotundus occurrence data are, therefore, critical for the development of preventive measures for vampire-transmitted rabies 48

Methods
Data gathering for this dataset began in January 2020 and ended in December 2021. Occurrence reports of D. rotundus were collected from a variety of publicly available resources and databases, from a network of natural history museums across North, Central, and South America, from official repositories in ministries of agriculture and health, from published scientific literature across Latin America, and from privately held databases from individual contributors (Fig. 1). The final dataset includes 39120 individual occurrence reports (i.e., recorded instances where one or more D. rotundus individuals were recorded or observed) (Fig. 2) and 7576 unique geographic locations of D. rotundus existence. All data were collected in Darwin Core Archive format 51 . The Darwin Core Archive is a biodiversity and taxonomy based data definition format that makes use of standardized terms and file structures 51 . The use of the Darwin Core Archive allows for better data accessibility and mobilizations, as well as facilitates the data's compliance with intercommunity standars 51,52 . Desmodus rotundus occurrence reports were geo-referenced using the World Geodetic System 1984 coordinate system in decimal degree units. Inclusion criteria for this dataset were: A) That the report consisted of the modern species Desmodus rotundus 1 . B) The report consisted of at least one individual. C) The report had a recorded geographic coordinate (e.g., latitude and longitude), or a detailed locality description from which the occurrence could be geolocated (i.e., at finer detail beyond municipality level). D) The report was from a validatable database, museum record, published piece of literature, machine recording (e.g., acoustic monitor or camera), human observation, preserved specimen, or live specimen.
Metadata such as individual count (i.e., number of individuals recorded at each occurrence location, which may vary based on how the original report was collected), specimen age or life stage, basis of record, and date of capture were collected for each report whenever possible (Fig. 3). For occurrence reports where full metadata were not available or unable to be confirmed, the information was left blank in the final file (Supplementary Materials). Definitions for the database and metadata can be found in Online-only www.nature.com/scientificdata www.nature.com/scientificdata/

Data Records
To collect occurrence reports from published literature, a review was conducted of all publications available in the Web of Science literary repository on August 28, 2020 (Clarivate TM , 2020. available from: https://apps. webofknowledge.com/Search). We conducted a keyword search of topics in journal manuscripts, proceedings papers, and official reports. Keywords included "Desmodus rotundus", "vampire bat", and "common vampire bat", and resulted in 315 manuscripts. The resulting manuscripts in Spanish, English, and Portuguese were then  www.nature.com/scientificdata www.nature.com/scientificdata/ screened for associated D. rotundus occurrence data. A summary of these literature data sources can be found in Online-only Table 2. Additional reports were obtained from 37 institutions or researchers with privately held data. These contributors are summarized in Online-only Table 3. Data curation and validation followed the standardized protocol used for other data sources (see Technical Validation).
Occurrence reports were also collected from publicly available data repositories or databanks (i.e., web-based sources which centrally house data from other sources) 53 . These repositories included the Global Biodiversity Information Facility (GBIF) 54 , Biodiversidata 55 , and speciesLink 56 . Occurrence reports for D. rotundus were downloaded from the GBIF on October 30, 2020 54 . GBIF occurrence reports with coordinates which were located in the western hemisphere (n = 12865) were downloaded from the database in Darwin Core Archive format 51 . Occurrence reports based on fossil specimens (n = 3) were removed. After the cleaning and validation process (see Technical Validation), the final number of occurrences from GBIF was 12736. Originating datasets which contributed to the GBIF download are summarized in the Supplementary Materials. Occurrence reports from Biodiversidata (Uruguayan Consortium of Biodiversity Data Repository) 55 were downloaded in December of 2020 from the vertebrate mammal sub dataset (n = 67) 55 . Occurrence reports from speciesLink (Centro de Referência em Informação Ambiental) were downloaded in August of 2021 (n = 2578). Of these reports, 918 were found to be already present in the GBIF database. In total, 1660 occurrences from speciesLink were added to the final dataset. A total of 298 occurrence reports without recorded coordinates were also downloaded from the publicly available data repositories and 48 of these reports were able to be georeferenced based on their locality descriptions. The other 250 could not be located due to a lack of detail in their locality descriptions and were therefore excluded. Georeferencing was completed using the tidygeocoder package in R 57 . All data are stored in the finalized dataset in the Figshare data repository for public access 50 .

technical Validation
To validate the collected data, we identified "redundant" reports (i.e., unique reports present in more than one dataset repository). Occurrence reports were flagged as redundant when the occurrence geolocation information (i.e., latitude, longitude, locality, and elevation) and institutional information (e.g., institutional identification number, originating dataset, institutional code, etc.) were identical. Other metadata such as date of occurrence, individual count, sex, life stage, and basis of record were used to confirm or reject redundancy. Reports where these variables matched were flagged and manually investigated to confirm redundancy. When redundant reports were found and confirmed, the original source occurrence report was retained. This process was completed using the dyplr package and base functions such as duplicate and unique in R 58,59 .
All occurrence reports were also investigated to eliminate occurrences with errors in geolocation using the coordinateCleaner package 60 . Using the functions cc_cap, cc_cen, cc_gbif, and cc_inst, we identified and removed occurrence reports which were erroneously assigned to country capitals, country centroids, or the