Climate change induced complex shifts in snake distributions expose people to snakebite and threaten biodiversity

Figures Abstract Snakes play pivotal roles in many ecosystems. While some species, including medically important ones, are considered threatened by the IUCN, snakebite takes a heavy toll on rural agricultural populations in the developing world. Approximately 138,000 deaths and 400,000 disabilities result from snakebite annually and WHO has pledged to reduce the resulting health burden by 50% by 2030. Among a plethora of reasons for insufficient snakebite mitigation, one is limited explicit knowledge of how, where, and when humans and snakes interact, which limits the timely, accurate, and efficient deployment of resources. Here, we revise the list of medically important snakes based on recent taxonomic updates and use high-resolution data from a broad range of published and unpublished resources to compare expert-derived ranges with statistical geographical models of habitat suitability for all 508 most medically important snake species globally. Our study is the first to model every single medically important snake species including data deficient ones, at the highest resolution to date, and with the largest supporting occurrence dataset. We generate geographically explicit estimates of how much human and snake populations overlap (snake-human-overlap-index; SHOI), which is the most fundamental prerequisite for human-snake conflict to occur. Finally, we model the effects of climate change on snake distributions. We predict substantial, short- and long-term shifts in snake distributions, including range contractions for many threatened species and increased human exposure to species of major public health concern. In combination with other drivers of increased snake-human conflict, such as human behaviours and snake traits, our predictions can be used to decide where to stockpile which antivenom, how to ensure adequate capacity of individual health facilities, how to improve health care accessibility of remote at-risk communities, and where to focus conservation efforts for threatened snake species. Hence, we highlight the need for geographically targeted efforts to benefit both vulnerable human and snake populations, as part of a One-Health strategy. Author summary Snakebite is a neglected tropical disease that affects millions every year and primarily results from conflicts in how humans and snakes use environments where they co-exist. Unfortunately, our understanding of where conflict is most pronounced is limited because of surprisingly sparse data on snakebite numbers, their locations, and snake distributions, especially in regions of the world where snakebites are most prevalent. Here, we collate unprecedented amounts of data on where medically important snake species occur and combine them with expert knowledge and statistical models of environmental suitability for snakes at a global 1km resolution now and under conditions of predicted climate change. This approach gives us detailed maps of snake-human population overlap, and allows stakeholders to implement targeted, future-proof efforts to improve human-snake co-existence as well as making snakebite treatments available in the most appropriate locations. It also enables us to detect knowledge gaps that still need to be addressed, based on disagreements of expert knowledge and statistical models. Our results can be used as a basis to study factors that increase human susceptibility to snakebite in areas where they co-exist, such as agricultural practices, housing conditions, lifestyle traits of different snakes, or weather events. Citation: Pintor AFV, Kanankege KST, Turner M, Abela B, de Castañeda RR, Moos B, et al. (2026) Climate change induced complex shifts in snake distributions expose people to snakebite and threaten biodiversity. PLoS Negl Trop Dis 20(5): e0014030. https://doi.org/10.1371/journal.pntd.0014030 Editor: Wuelton Monteiro, Fundação de Medicina Tropical Doutor Heitor Vieira Dourado: Fundacao de Medicina Tropical Doutor Heitor Vieira Dourado, BRAZIL Received: May 1, 2025; Accepted: February 11, 2026; Published: May 21, 2026 Copyright: © 2026 Pintor et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The data for this manuscript and a short user guide are stored in the Harvard Dataverse at https://doi.org/10.7910/DVN/QVWB4E. The data is also accessible through OPHIDS on the WHO Snakebite Information and Data Platform and updated versions will be uploaded when applicable [27]. Funding: This work was supported by Wellcome Trust (grant reference number 222215/Z/20/Z to DJW, MT, and AFVP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. AP received a salary as part of the funding received from Wellcome Trust (222215/Z/20/Z). Competing interests: The authors have declared that no competing interests exist. Introduction Snakebite envenoming (snakebite hereafter) is a neglected tropical disease that affects millions, kills over 130,000 people and leaves 400,000 suffering from long-term physical and psychological medical conditions annually [1]. It is largely a disease of poor rural communities [2,3] in low and middle income countries (LMICs) [1,4] in tropical and sub-tropical regions. Consequently, the burden of snakebite has historically been inadequately addressed in global disease mitigation strategies [5]. Snakes also play pivotal roles in many ecosystems [6], yet their conservation status is concerning: over 30% of species in the most medically important venomous snake families (Viperidae and Elapidae) are threatened, near threatened or data deficient [7], many are geographically restricted, have declining populations [8], and are negatively affected by landscape conversion for human activities [9], illegal collection for the pet [10] or food trade [11], and indiscriminate culling out of safety concerns [12]. Unlike many infectious diseases, eradicating the disease-transmitting (snakes) or causing (venom) agent is not an option; venomous snakes and people must co-exist and share common ecosystems. With the Global Strategy for Prevention and Control of Snakebite Envenoming [13], the World Health Organization (WHO) aims to reduce global snakebite deaths and disabilities by 50% by 2030. Understanding the geographic distribution of snakes, their overlap with human populations, and snakebite patterns across the world is, therefore, a crucial first step for both conservation and public health strategies and would support efficient positioning of mitigation efforts, such as risk-based placement of antivenoms, training healthcare staff to handle cases, and educating communities [14]. However, high-resolution data on these factors are sparse [15]. Inadequate resources, challenging field conditions, and complex political situations hinder data collection. Furthermore, future-proofing snakebite mitigation and conservation strategies involves assessing the likely impact of climate change on human-snake interactions [14], because the distributions of ectotherms such as snakes are highly dependent on climatic conditions [16]. WHO maintains a list of medically important venomous snakes (MIVS), categorized into two groups: 1. Primary MIVS that are highly venomous and common or widespread and cause many cases of bites, morbidity, disability or death; 2. Secondary MIVS that are highly venomous and capable of causing morbidity, disability or death, but either lack epidemiological data or bite less frequently because of their behavioural or ecological traits [17]. This list is the fundamental basis for attributing any spatial information to the correct species. However, the taxonomy of MIVS and information on the attributes that qualify them for listing are revised constantly, while updates of the MIVS list have been much less frequent. Traditionally, species’ distributions are recorded by collation of point occurrences and creation of expert derived range maps (EDRs) [18]. Recent advancement of online databases such as the Global Biodiversity information Facility (GBIF) [19], citizen science platforms such as iNaturalist [20] and extraction of location data from social media platforms (Facebook, Flicker etc.) [21] have greatly improved the accessibility of location information on MIVS. Despite these advances, data for many, even common species are still surprisingly sparse and even the most advanced studies on snake distributions only cover those MIVS with sufficient data [22]. Omission of data-sparse taxa negatively affects conservation outcomes [23], and snakebite mitigation. Additionally, the process of creating EDRs for field guides from these disparate data sources remains slow, repetitive, resource-intensive, and even contentious amongst experts. More recently, Species Distribution Models (SDMs), and Environmental Niche Models (ENMs), have become widely used in ecology and conservation to estimate species’ habitat suitability. They can offer detailed insights into poorly sampled species’ distributions even from sparse occurrence data [24,25] combined with high resolution environmental data and provide much more fine-scale estimates of species’ habitat suitability than coarsely ‘drawn’ EDRs [18]. Similar to EDRs, SDMs aim to predict the realized distribution of species, while ENMs model conditions suitable for species to thrive and project them to those available across the landscape (only some of which are part of the realized distribution due to access restrictions or biotic interactions) [26]. ENMs can also predict future changes in habitat suitability based on climate projections. In combination, EDRs and ENMs can provide more robust information on species’ distributions than either in isolation [18]. Here, we use a global, iterative, intensive data mining process across public and private databases, citizen science platforms, scientific literature, books, and social media to establish a constantly updated, dynamic database of all MIVS globally, their taxonomy, and their distributions. Resulting datasets include occurrence localities, Geographical Information Systems (GIS)-enabled EDRs and high-resolution ENMs, as well as estimates of human-snake overlap. ENMs and Snake-Human overlap estimates are available for current and future conditions (2050 and 2090). Data collation included rigorous vetting by an international expert panel comprised of leading researchers in the fields of snake distributions and taxonomy to guarantee a trustworthy one-stop-shop for researchers, governments, NGOs, and the general public. The ‘Occurrence Point Hub and Information Database for Snakes’ (OPHIDS) is publicly accessible through the WHO Snakebite Information and Data Platform [27] and forms a fundamental basis for future studies on snakebite prevalence across different social, behavioural, economic, and environmental contexts. Our main objective is to provide an overview of the potential contribution of one of several factors facilitating snakebite incidence, i.e., the overlap of human and snake populations, and provide a baseline consensus, dynamic dataset for researchers and governments to use for snake-human conflict mitigation. These data are of potential utility in guiding better adaptation strategies in public health and clinical management of snakebites by, for example, indicating which geographies need antivenoms supplies, as well as in conservation of ecologically important or rare snake species. Methods For a more detailed methodology please refer to S1 Text. The authors welcome feedback on all parts of the methodology, to improve future versions of the database. Study species We used the original listed WHO recognized MIVS from the 2018 version [1,17] and revised their taxonomy using published literature, and expert feedback. We included any currently recognized species that were either (i) already explicitly WHO listed [1], (ii) new species implicitly listed as part of a former parent species, or (iii) part of an unresolved species complex including listed species: we include 508 individual species in the updated list. Both, category 1 and 2 species were included. Because some species are considered different categories in different countries, it is impossible to separate the two groups. Many category 2 species may have significant impacts on people but are too poorly studied to quantify bite incidence. Including these species provides valuable information on potential underreporting or to identify knowledge gaps. See S1 Table for a complete species list. Note that many MIVS taxa are under ongoing revision and these will be reflected in future versions of the list of WHO recognized MIVS, which will be published iteratively online on the WHO Snakebite Information and Data Platform [27]. Occurrence data Occurrence records were collated from public, private, and citizen science databases (e.g., GBIF [19], iNaturalist [20]), museum records, books, a broad array of scientific literature, and verifiable personal observations from experts and social media (see S1 Text for more details). Descriptive localities were georeferenced, and their spatial uncertainty estimated using Google Maps [28] (S2 Text). Coordinates of records only displayed as maps in publications were extracted using DataThief [29] and map symbol radius was used as uncertainty. All occurrence records were vetted for taxonomic and location accuracy by an expert panel of >30 experts from around the world. For ENMs, records were reduced to 75% of those with highest location accuracy, except for species that were data deficient (0.8 removed based on biological relevance [25]. Only two sets of collinear variables were maintained based on expert experience of complementarity of those variables in other analyses despite collinearity. Collinearity has been shown to be of minor importance to Maxent model performance [31], and our variable selection process eliminates any redundancy in final model predictors. Therefore, including some collinear variables was considered preferably to excluding potentially biological important variables. The resulting set of variables included Temperature Seasonality, Maximum Temperature, Minimum Temperature, Annual Precipitation, Precipitation Seasonality, Precipitation of Driest Quarter, Precipitation of Warmest Quarter, Minimum Radiation, Maximum Radiation, Minimum Relative Humidity, Topographic Ruggedness Index, Degree of Orientation Towards the Equator, Mean Dry Matter, Range of Dry Matter Productivity, Mean Fraction Photosynthetic Active Radiation, Range of Fraction Photosynthetic Active Radiation, Landcover Type in 2018, Topsoil Bulk Density, Fraction of Coarse Fragment, Percent Clay, Percent Organic Carbon, Soil Type, Distance to Permanent Water, and People per Grid Cell in 2020 (as