Anthropogenic impacts drive habitat suitability in South Asian bats

South Asia is host to 151 species of bats across nine families (Srinivasulu et al. 2023), yet there is very little information known about their distribution and ecology in this region (Bates and Harrison 1997). Despite their high diversity and importance as ecological indicators and ecosystem service providers (Jones et al. 2009; Kunz et al. 2011; Altringham et al. 2011), bats are relatively underrepresented in macroecological studies, especially in Asia (but see Srinivasulu et al. 2024). Subject to persecution and superstitions across many regions of Asia (Frembgen 2006), they are also given very little conservation value, an issue further complicated after the recent COVID-19 pandemic (MacFarlane and Rocha 2020).

A recent trend in ecology has been the study of abiotic and biotic factors on occurrence and distribution patterns across various scales, ranging from small communities to global populations (Leach et al. 2016; Lewis et al. 2017; Lopez et al. 2019). Most studies tend to focus on climate change as a major factor of species distributions (Araújo et al. 2019; Srinivasulu et al. 2021). However, analyses of biotic [including vegetation, land use and land cover (LULC)], topographic (elevation and hydrology), and anthropogenic factors are also vital to understanding distributions and ecological niches (Hughes et al. 2012). Due to effects of both climate-based and ecological cues (Bates and Harrison 1997) on bat behaviour, ecology, and habitat ‘selection’ and suitability, analyses must consider the influence of both to understand bat distributions and niches. Additionally, a deeper understanding of the specific impacts of ecogeographical factors on habitats and species can better inform local management and guide species- and site-specific conservation planning and policy.

In South Asia, bats are distributed across a wide range of habitats from forests to urban and suburban areas and previous work has offered insight into factors that affect habitat suitability for regions or groups. Wordley et al. (2015) analysed the association of bats with agricultural areas and riparian habitats in the Western Ghats, showing that structural diversity within agricultural habitats improves bat abundance and richness. Other studies have also shown that bat distribution in urban and suburban areas is affected by factors like artificial night-time light and distance from the nearest waterbody (Lewanzik et al. 2022), but the level of association with urban and suburban regions varies between species with some bats more likely to be distributed in ‘wilder’ areas (Gili et al. 2020). Forest bats have been found to depend on vegetation structure, and bats in arid and scrubland habitats are influenced by distance to wetlands and riparian vegetation (Razgour et al. 2018). While past studies show that bat distributions can be influenced by ecogeographical factors, there has been no large scale assessment to identify broader drivers and patterns of habitat suitability in South Asian bats.

Ecological niche modelling (ENM) is a method that analyses the known distribution of a species and measured conditions describing climate, geography, and ecology to extrapolate an envelope of spatial suitability approximating the species’ niche and quantify the importance of the different conditions in shaping that niche (Guisan & Thuiller 2005; Araújo and Guisan et al. 2006; Soberón & Arroyo-Peña, 2017). Importantly, ENM can be used to gain knowledge of cryptic, rare, or otherwise difficult to study species because these approaches can identify potential suitable habitat (which may guide monitoring efforts), as well as revealing the set of environmental conditions that influences potential presence (Rebelo & Jones 2010; Jeliazkov et al., 2022). Diverse types of environmental conditions can be considered, but often ENM studies focus on climate variables to project species’ current and future distributions and quantify potential climate change impacts (Guisan & Thuiller 2005). Some models do analyse ‘biotic’ ecogeographical factors, including biotic interactions and human activities, and these can offer critical insights and inform conservation planning and policy (Leach et al. 2016; Cosentino et al. 2023). For example, Hughes et al. (2012) conducted an ENM study combining abiotic climate and biotic ecogeographic variables that projected northward shifts in Southeast Asian bats. Combining multiple abiotic and biotic factors can create challenging model complexity, but separate models can be defined with the predicted climatic and ecogeographical suitability areas then compared and combined to provide a more comprehensive approximation of a species’ niche (Johnson et al. 2019). Additionally, there are many ENM algorithms available which can offer distinct results; thus, it is advisable to use various algorithms that are then aggregated into ensemble models for offer better performance and a clearer understanding of model reliability through a combination of goodness-of-fit metrics and inter-algorithm agreement (Thuiller et al. 2009). Ensemble ENMs can be highly reliable and interpretable even with presence-only data if robust approaches for generating pseudoabsences (Barbet-Massin et al. 2012) and standardised protocols for parameterisation (Feng et al. 2019) are applied.

In this study, we assess habitat suitability in South Asian bats using occurrence records and focusing on ecogeographic factors including land use and land cover, topography, hydrology, and anthropogenic impact variables. We generate ensemble ENMs incorporating multiple replicates of pseudo-absence datasets. Results identify the most important ecogeographic factors and present species habitat suitability maps we use to identify suitability hotspots. We then compare these hotspots with climatically suitable areas defined in a previous study (Srinivasulu et al. 2024) to offer a comprehensive understanding of suitability in the region. The results of this study form a foundation for site- and species-specific bat conservation prioritisation and planning.

Our study identifies Median Night-time Light as the ecogeographical factor with the highest importance to map habitat suitability in South Asian bats. Artificial night-time light is prevalent across many regions in South Asia due to the high human population in the region, increasing in brightness steadily with higher rates of urban and suburban expansion and growth (Kaushik et al. 2022). Night-time light impacts bat flight, behaviour, foraging, and roosting in varied but largely adverse ways (Stone et al. 2015; Spoelstra et al. 2017), yet to our knowledge, no study has been conducted in South Asia to specifically quantify the effects of night-time light on bat movement, feeding, and roosting. While generally important, it is noteworthy that the importance of night-time light varied among species. Species with differing ecologies and behaviours are likely to be impacted by light levels differently. Some bats are quite sensitive to light disturbance. For example, Median Night-time Light was a the most important variable for the fulvous roundleaf bat Hipposideros fulvus (variable contribution 52.80%), a species generally found in relatively darker areas (Median Night-time Light across its projected suitable area is 0.87 ± 3.89 DV). Cooler lighting with white and green components have been shown to cause significant decreases in the abundance of species like Rhinolophus (horseshoe bats), and Myotis (mouse-eared bats; Spoelstra et al. 2017), which are not usually found in or near human habitation. However, other bats like Pteropodids are more synanthropic and from our analyses appear to commonly occur in areas with high night-time light values. For example, Median Night-time Light was a very important variable for Kelaart’s pipistrelle Pipistrellus ceylonicus (51.32%) and the naked-rumped tomb bat Taphozous nudiventris (47.97%), and in contrast with Hipposideros fulvus both species were projected to occur in relatively lighter environments (Median Night-time Light in projected suitable areas 7.07 ± 10.03 DV, and 10.49 ± 9.98 DV respectively. For comparison human-inhabited areas in the region had values > 20 DN). These species may be more resilient to anthropogenic changes (able to cope with some artificial night-time light) and some may even benefit—some urban insectivore species are known to use streetlights in urban and suburban areas as feeding grounds (Hermans et al. 2024). Variation can also occur within groups of related species. Bats from families generally considered to be light-sensitive, like Pteropodidae which primarily use sight and smell rather than echolocation for navigation (Bates and Harrison 1997), can be associated with urban areas—e.g., greater short-nosed fruit bat Cynopterus sphinx is commonly found in city suburbs due to its association with fruiting trees (Bates and Harrison 1997), and Indian flying fox Pteropus medius has been observed in large colonies in various cities and towns (Pandian and Suresh 2021; Roy et al. 2024). Even for species that can tolerate higher levels of artificial night light, it is important to careful consider lighting practices to ensure natural communities and ecosystem services are not disrupted (Rowse et al. 2016; Voigt et al. 2021).

The suitability hotspots in the study area showed large contiguous clusters in northern India, the Indus River valley in Pakistan, the Himalayas of Bhutan, India, and Nepal, the Western Ghats in south India, and the highlands of Sri Lanka. This spatial distribution of hotspots broadly aligned with the Myers et al. (2000), with the highest amount of suitable area seen in the Indian Western Ghats and across the Sri Lanka hotspot, further emphasising the importance of this region as a South Asian biodiversity hotspot (Fig. 4). However, it is important to note that this hotspot also comprises large cities, towns, and complex infrastructure, and may be threatened by habitat destruction due to lateral expansion. While there is active conservation in place across the Western Ghats and Sri Lanka (Das et al. 2006; Bambaradeniya 2006), this tends to focus more on charismatic species and there is a need for bat-specific conservation efforts. Spatial suitability patterns were broadly consistent at all species suitability thresholds; regions suitable for ≥ 24 and ≥ 36 species were less contiguous than those suitable for ≥ 12 species, but the largest contiguous regions consistently remained in the Western Ghats and southwestern Sri Lanka (Fig. 4). Hotspots for half of the studied species revealed a large area of fragmented suitability in the Western Ghats of northern Karnataka, roughly situated between Sharavati Valley Wildlife Sanctuary in the south and Anashi National Park in the north, while the western coast was still projected to be suitable. The contiguity of suitability hotspots in the Nilgiri and Vindhya hill ranges, and the lower Himalayas—all regions with very specific and unique ecosystems, habitat structures, and vegetation (Olson et al. 2001)—was also lower when mapping suitability for more species (Fig. 4). River valleys, cities and towns, and large wilderness areas were consistently projected to be suitable for multiple species, and linear structures including forest corridors and urban and suburban structures such as roads also showed suitability, indicating the importance of such features as movement pathways. In some cases, these linear features may be used by forest- or scrubland-dominant species as corridors between suitable habitats, therefore making them vital for connectivity. While the higher suitability of conurbations and roads could be interpreted as an artefact of bias in the collection of occurrence data, this is not immediately apparent when viewing the occurrence data for our study species (Supplementary Material 3) and must be explored more deeply. It is important to note that these results are applicable only for a relatively small subset of the large diversity of bats in South Asia (~ 16% of the 151 species present in South Asia).

Suitability hotspots included relatively high proportions of Anthropogenic habitats, which likely captured common patterns from generalist species that can adapt to human impacts, but we found variation among groups and species. The lowest cover of anthropogenic habitats was found in the suitable areas of Rhinolophidae (horseshoe bats), a family distributed mostly in forests, caves in South Asia, had, while the highest were found in the Rhinopomatidae (mouse-tailed bats), a family associated with human habitation and features such as tombs, ruins, etc., and known anecdotally to be resilient to disturbance. Common species (e.g., greater short-nosed fruit bat Cynopterus sphinx, and Indian flying fox Pteropus medius) showed higher association with urban and suburban and human-inhabited areas than more specialist and more uncommon species (e.g., great roundleaf bat Hipposideros armiger, and intermediate horseshoe bat Rhinolophus affinis), likely due to the formers’ generalist ecological niches allowing more resiliency to anthropogenic disturbance.

Assessing habitat suitability can inform conservation planning and ENM is an especially effective approach (Jeliazkov et al. 2022). Suitability analyses tend to be climate-focused, occasionally incorporating topographic variables (Festa et al., 2023), but in regions of high habitat, species, and structural diversity such as South Asia (Myers et al. 2000; Srinivasulu and Srinivasulu 2012; Ramankutty et al. 2018; Raman et al. 2023), it is imperative that ecogeographic factors are also assessed. Regions shown to be climatically suitable may not be ecogeographically suitable or vice versa. Srinivasulu et al. (2024) defined climatic suitability for 110 species of South Asian bats revealing large contiguous suitability hotspots in the Himalayas, the Western Ghats, and Sri Lanka, similar to the hotspots described in this study. While broad area coincided, the overlap of climate and ecogeography suitability area for the 48 species reveal a much smaller suitable area. Moreover, these combined suitability areas face anthropogenic threats of habitat disturbance and destruction, and in some cases like the northern Western Ghats, appeared quite fragmented (Anand et al. 2010). Our results show that failing to consider both climate and ecogeography suitability can overestimate suitability, and we highlight the value of combined appraisal. Bats are likely to show large responses to climate change (Festa et al., 2023), these may outweigh the influence of ecogeographic factors in future distributions (Wani et al. 2021). However, ecogeographic factors will likely respond to climate change and socioeconomic development, while we did not consider projected changes to these factors in our study future studies could further evaluated these changes under projected shared socioeconomic pathways (O’Neill et al. 2017; Bukovsky et al. 2021).

The results of any ENM study are impacted greatly by various factors including the quality and filtering of the variables and occurrence data, how pseudoabsence are generated and results validated (Feng et al. 2019). In our analysis, we selected species with a minimum number of occurrences considering representation and model requirements. We acknowledge that by doing so, rarer species and those occurring in more remote area were more likely to be excluded. To identify ecogeographical variables we focused on proposed hypothesised relationships and then filtered to avoid high correlations that could affect inference. While we aimed to include a wide range of relevant variables, lack of available information at this scale prevented us from considering some likely important variables including those related to availability of food resources. We implemented a robust approach to define pseudo-absences that combines geographic constraints and replicated random sampling (Srinivasulu et al. 2024). This approach could be further improved using ecological filtering in niche space (Barbet-Massin et al. 2012; Iturbide et al. 2015), and a deeper analysis of survey and observation biases in the data. Finally, we used five-fold crossvalidation which is a widely used method but could be further improved by spatial block crossvalidation (Valavi et al. 2019). Ensemble ENMs are relatively new and quite powerful but can be computationally intensive and complex to interpret, thus often requiring a compromise between performance and feasibility.

Our study offers insight into the role of various ecogeographic factors on bat habitat suitability in South Asia, highlighting a role of anthropogenic factors, identifying suitable habitat hotspots, and revealing a worrying projected loss of ecogeographic and climatically suitable areas in the near future. We focused on describing broad patterns and effects, but to support conservation and policy we provide species-level results (Supplementary Material 2, 3) that can be used to consider effects within particular regions and for particular species. Bats in South Asia are a diverse group that faces some challenges. Some resilient generalist species, like pipistrelles, may be able to cope with human expansion, but others may be left with few suitable areas. Future work to further our understanding of bat ecological niches and distributions, including projected changes, would be needed and benefit from additional occurrence data and improved information on ecological variables and their projected changes.