Limestone karst ecology and anthropogenic activities associated with cave-dwelling bats of Southern Shan State, Myanmar

Abstract

Limestone cave systems constitute an important ecological niche supporting numerous bat species in Southeast Asia, with Myanmar, having a particular high number of understudied caves. Investigating the natural and anthropogenic ecological drivers that influence higher densities and species diversity of bats co-roosting in these natural caverns is therefore vital for identifying regions of heightened conservation need. We examine characteristics of limestone caves and alternative sites of bat congregation/roosting associated with cave-dwelling bat abundance at 41 sites in Pinlaung Township, Southern Shan State, Myanmar. The characteristics evaluated included cave surface area and complexity, microclimate conditions, seasonality, and a modified biotic vulnerability (BV) index. Larger caves and caves with warmer temperatures were significantly more likely to support higher total numbers of bats. Further distance from the road was also associated with higher bat abundance; as the distance increased, human disturbance decreased, resulting in larger bat populations. The majority (63.2 %) of caves were classified as moderately vulnerable (BV class B), 18.4 % highly vulnerable (BV class A), and 18.4 % vulnerable (BV class C), indicating that environmental protections are essential to ensuring the long-term sustainability of cave-dwelling bat habitat within limestone karsts. Most cave sites evaluated faced multiple threats to bat populations, including bat hunting, guano harvesting, and cave development, which are known threats to cave-dwelling bats in other parts of Southeast Asia. This study identifies Southern Shan State, Myanmar as a critical habitat for limestone karst cave-dwelling bat species habitat that is in urgent need of conservation.

1. Introduction

Bats play a critical role in ecosystem health, including providing nutrients for cave ecosystems through guano deposition (Gnaspini and Trajano, 2000), and conducting insect control, plant pollination and seed dispersal (Kunz et al., 2011). Bats are also keystone species in some subterranean cave ecosystems (Ferreira, 2019) and may serve as an umbrella species in the protection of other species, including invertebrates and microorganisms (Frick et al., 2020; Phelps et al., 2016). Cave habitats support some of the largest and most diverse aggregations of bat species in the world providing protective refuges for roosting for roughly 40 % of the world’s bat species (Furey and Racey, 2016). With over a third of bat species considered threatened or data deficient (IUCN Red list of Threatened Species), preservation of cave habitats is a crucial element of bat conservation.

Limestone karst provides critical cave ecosystems for bats as caves occur in high densities in this porous substrate (Kingston, 2010; Furey and Racey, 2016, Struebig et al., 2009) with a high rate of species endemism (Tanalgo et al., 2022; Hutson et al., 2001). Limestone karsts are also important ecosystems for species diversity in general, with 143 species from karstic habitats known to be globally threatened, and of these, 31 occurring in Southeast Asia (Clements et al., 2006). The high species diversity and endemism in karsts arise from varied ecological niches afforded by complex terrains (e.g., fissured cliffs and extensive caves), diverse climatic conditions as well as relative geographic isolation (Clements et al., 2006).

Globally, cave habitats, and particularly limestone karst given its fragile geologic composition, face numerous threats. Climate change is altering microclimatic conditions in cave environments (Mammola et al., 2019) and habitat change at the surface, including urbanization and deforestation, affects cave systems (Couton et al., 2023). Overexploitation (e.g. tourism, guano harvesting, hunting for bushmeat and trade), and direct habitat destruction and alteration (e.g., mining) are also altering cave ecosystems (Chaber et al., 2021; Frick et al., 2020; Furey and Racey, 2016). Bats that roost in caves are especially vulnerable to disturbance because they form large, concentrated aggregations (Frick et al.2020) that can be easily disturbed by a single anthropogenic event.

In Southeast Asia, karst areas span approximately 400,000 square kilometers (km²), with Myanmar having the second largest karst area after Indonesia (Day and Urich, 2000). Located primarily on the Shan Plateau, karst limestone ridges stretch North to South in Myanmar over 1500 km (Oo et al., 2002), with the plateau itself covering an area of 150,000 km² at an average elevation of 1100 m above sea level (Dreybrodt and Steiner, 2015). Karsts are recognized as critical ecosystems and have long been included in national conservation plans within the region (MacKinnon and MacKinnon, 1986). Despite this, the karst and caves of Myanmar are one of the least studied in Asia, largely due to the country’s international isolation and poor infrastructure in cave regions (Dreybrodt and Nyi Nyi, 2019). Only a small number of speleologists and bat researchers have had the opportunity to explore Myanmar’s karst (Dreybrodt et al., 2021).

Cave bat fauna in Shan State, Myanmar includes both fruit bats and insectivorous bats (Dreybrodt and Steiner, 2015) and is likely comparable to neighboring Northern Thailand and Laos. Only a few studies have been published on the karst fauna in Myanmar, which have identified new species of reptiles (Cyrtodactylus pinlaungensis, Cyrtodactylus linnwayensis and Cyrtodactylus ywanganensis) and snails (Burmochlamys sp.) (Grismer et al., 2018, Grismer et al., 2019 and Grismer et al., 2020, Pholyotha et al., 2022). Intensive surveys of bats in this region have not been conducted, although cave-dwelling bats have been described in other parts of Myanmar including several new species to Myanmar and to science including Hipposideros lankadiva gyi, Rhinolophus malayanus, R. stheno, R. paradoxolophus, Craseonycteris thonglongyai, Myotis mystacinus, Ia io and Pipistrellus pulverstus (Bates et al., 2000, 2001, 2004, 2005, 2015, Struebig et al., 2005, Myo et al., 2020, Aung et al., 2020, Oo et al., 2017). Recent genomic analyses suggest there are dozens of undiscovered bats, with potentially up to 40 % of species in Southeast Asia incompletely described (Chornelia et al., 2022). Many karsts and cave-dwelling bats are yet to be discovered, particularly in the Shan State region of Myanmar. Karsts in Myanmar are threatened by many anthropogenic disturbances such as quarrying, tourism and recreation, renovation for religious purposes, guano collection and bat hunting, all of which threaten cave-dwelling biodiversity.

Human disturbance of bat cave ecosystems, particularly in Myanmar also has implications for public health. Anthropogenic activities in karst caves may introduce external stressors that alter viral shedding of individual bats and change the composition of species interactions, potentially increasing the likelihood of disease emergence. Moreover, human activity near dense bat colonies also provides interfaces for zoonotic disease transmission. Many viruses of public health consequence are of bat origin (Bonilla-Aldana, 2021), including ebolaviruses, rabies virus, and coronaviruses. Coronaviruses with evolutionary relationships to SARS-CoV-2 have been found in bats in Yunnan province, China (Wu et al., 2023; Zhou et al., 2021), as well as in northern Laos (Temmam et al., 2022), Cambodia (Delaune et al., 2021), and Thailand (Wacharapluesadee et al., 2021). Several novel bat coronaviruses have been found in Myanmar (Valitutto et al., 2020), and serological evidence of sarbecovirus (the subgenus of coronavirus containing SARS-CoV and SARS-COV-2) exposure prior to the SARS-COV-2 pandemic has been documented in communities engaged in bat guano harvesting and extractive industries in Myanmar (Evans et al., 2021). Community involvement in extractive industries and guano collection may heighten the risk of coronavirus spillover to humans (Yadana et al., 2023).

The goal of this study was therefore to investigate natural and anthropogenic ecological influences on bat abundance and roost species diversity to inform bat population management in Shan State, Myanmar and to better understand surrounding environmental factors that could be contributing to higher risk for disease emergence.

2. Methodology

2.1. Study area and study period

The study was conducted in caves and other bat congregation / roosting sites in Pinlaung Township, Taunggyi District, Myanmar. Karst landscape is situated in the southwestern region of the Shan Plateau, extending northward to the former Paung Laung river, now, the upper Paung Laung dam (Fig. 1). The plateau presents a diverse terrain characterized by a combination of karstic hills, parallel mountain ranges and intermontane basins, forming a complex geographical mosaic with elevations ranging from 300 m to more than 1700 m (Kitahara et al., 2019; Mon et al., 2009). Forest type is broadly classified as mixed deciduous forest in lower elevation areas and evergreen forest in higher elevation areas (Kitahara et al., 2019; Mon et al., 2009). Most of the local population in the study area were of the Pa-O ethnic group, while other ethnic groups such as Bamar, Shan, Kayin, Kareni were also present. Crop cultivation varied across the elevational gradient with potatoes, corn, cabbage, tea and other horticulture crops predominant in upland mountainous regions and paddy and turmeric in lower areas. A total of 41 sites (Table 1), consisting of 38 caves, two houses and one pagoda were visited. Fourteen caves, two houses and one pagoda were visited from September to October 2023 (rainy season) and 36 caves and one house were visited in December 2023 (dry season) to assess the ecology of limestone karsts and seasonal fluctuation of cave-dwelling bats in this area. Thirteen caves were visited and explored (Cave # 1–11, Cave # 14 and 16) during both dry and wet seasons.

Fig. 1.

(A) Locations of roost sites surveyed during the rainy and dry seasons (2023–2024) in Myanmar. Symbol shape indicates roost type (cave or building), and symbol color indicates the level of human disturbance based on the Human Disturbance Index (HDI). The HDI integrates measures of human activity and accessibility, including distance to the nearest road and village, disturbance within and around caves, guano collection, and cave development. (B–E) Photographs of representative roost sites across the human disturbance gradient. (B) Cave with Buddha statues, classified as high human disturbance (HDI = 5). (C) Htan Saung Ra Cave with bamboo poles used for guano collection, classified as moderate human disturbance (HDI = 3). (D) Limestone quarrying adjacent to “Big Bang Cave,” also classified as moderate human disturbance (HDI = 3). (E) “Paung Laung Cave 3,” classified as low human disturbance (HDI = 2).

Table 1.

Characteristics of sites surveyed in Paung Laung, Shan State, Myanmar including geographic locations, accessibility assessment, and anthropogenic activities.

Site	Name	Coordinates	Elevation (m)	No. of Entrances	Surface area (m2)	Total Length (m)	Feasibility*	HDI**
1.	NTY Cave 1	20.2924 ° N 96.7637 ° E	1495	1	306.18	21.64	2	5
2.	NTY Cave 2	20.2935° N 96.7367° E	1476	4	2438.09	63.99	2	4
3	Hti Phar Cave	20.3409° N 96.8598° E	1123	3	3086.99	50.40	2	5
4.	Buddha Head Cave	20.2264° N 96.8424° E	1307	1	280.71	6.00	1	5
5.	Hotel Cave 1	20.199° N 96.8252° E	1134	1	8632.72	194.70	2	4
6.	Lay Tet Cave	20.2906° N 96.7385° E	1493	1	301.96	18.90	2	3
7.	Naung Pi Home	20.2365 ° N 96.8524°	1310	1	239.26	10.00	1	3
8.	Dragon Cave	20.1952° N 96.7062° E	1431	6	2102.43	86.20	1	4
9.	Paung Laung Cave 3	20.2211° N 96.5641° E	617	1	329.11	2.80	3	2
10.	Paung Laung Cave 1	20.1433° N 96.5814° E	372	1	774.15	27.60	2	1
11.	Eden Cave	19.8917° N 96.4233° E	375	2	954.54	32.70	2	2
12.	NanPalin House	20.3203 ° N 96.7746° E	1470	1	596.27	14.60	1	4
13.	Lwe Kaung Pagoda	20.2920 ° N 96.7826° E	1610	1	NA	NA	1	3
14.	Hti Ri Cave 1	20.3090° N 96.8288° E	1490	2	448.45	20.20	1	4
15	Big Bang Cave 1	20.1719° N 96.7907° E	1333	2	NA	NA	3	3
16.	Paung Laung Cave 2	20.1455° N 96.5809° E	383	1	NA	NA	3	2
17.	Banana Cave	19.9822° N 96.6852° E	676	2	6503.07	63.00	3	2
18.	Saung Paung Cave 1	20.2573° N 96.7218° E	1444	1	403.83	11.33	1	4
19.	Saung Paung Cave 2	20.2558° N 96.7207° E	1451	4	2895.05	167.91	1	5
20	Phayar Phyu Cave	20.2671° N 96.7146° E	1401	2	1824.65	114.30	3	4
21.	Hti Ri Tayin Cave 2	20.2959° N 96.8088° E	1624	1	210.28	9.15	3	3
22	Hti Ri Tayin Cave 1	20.2939° N 96.8100° E	1546	5	1491.51	72.88	1	5
23.	Hti Ri Tayin Cave 3	20.2958° N 96.8088° E	1651	2	9409.59	73.83	3	3
24.	Hotel Cave 2	20.1949° N 96.8326° E	1217	3	3616.99	92.20	1	4
25.	Hotel Cave 3	20.1956° N 96.8310° E	1237	2	3801.86	75.00	1	3
26.	Hotel Cave 4	20.1949° N 96.8326° E	1218	2	19.87	2.00	1	3
27.	Pat Kaw Cave	20.2831° N 96.6523° E	1374	4	1922.30	60.96	1	3
28.	Tat Ohn Cave	20.2830° N 96.6518° E	1374	2	355.50	29.42	2	2
29.	Saung Paung Cave 3	20.2830°N 96.6518° E	1393	1	11,262.10	325.20	2	4
30.	NTY Cave 3	20.2934°N 96.7665° E	1508	1	33.22	2.13	1	4
31.	Htan Saung Ra Cave 1	20.2923°N 96.6786° E	1404	1	441.50	26.70	2	3
32.	Htan Saung Ra Cave 2	20.2923°N 96.6786° E	1404	1	548.91	36.70	2	3
33.	Htan Saung Ra Cave 3	20.2923°N 96.6786° E	1404	2	427.45	16.30	2	3
34.	Hti Ri Cave 2	20.3179°N 96.8111° E	1653	1	19.87	2.00	1	5
35.	Big Bang Cave 2	20.1716°N 96.7910° E	1421	1	1264.52	51.00	1	2
36.	Myauk Hti Pwar Cave	20.2119°N 96.6696° E	1347	1	10,557.55	118.00	3	4
37.	Saung Nein Cave	20.2568°N 96.6694° E	1443	1	9640.24	150.00	2	4
38.	Pot In Cave	96.7140° N 20.2410° E	1365	NA	2377.63	23.30	2	5
39.	Nyan Pin Gyi Cave	96.8120°N 20.2430° E	1435	NA	754.06	19.00	2	4
40.	Paw Yar Cave 1	20.3154°N 96.6682° E	1389	3	2230.39	54.89	1	4
41.	Paw Yar Cave 2	20.2621°N 96.7275° E	1421	1	94.59	10.68	1	4

^*Feasibility (Easy = 1, Moderate = 2, Difficult = 3)

^**HDI (Human disturbance index, High = 5 and 4 points, Medium = 3 points, low = 1 and 2 points)

NA = Not available

2.2. Cave surveys

A series of survey points, starting with the first point located at the main entrance, were established along the length of the cave. New points inside the cave were located based on variations in the height or width of the cave passage, as well as the presence of any side passages between the points (Phelps et al., 2016). All passages were thoroughly surveyed in each cave, except for some situations where safety concerns prevented the field team from doing so. At each survey point, we measured the distance to the previous point as well as the height and width of the cave passage using a tape measure to estimate cave surface area. Total cave surface area was calculated by using the elliptical cylinder surface area calculation method: $C=2\pi L{\left(\frac{\left[a^2+b^2\right]}{2}\right)}^{1/2}$, where “a” was the width of the cave, “b” was the height of the cave and “L” was the length of the cave between points (Brunet and Medellín, 2001). If the passage height was unreachable, we used a laser rangefinder (UNI-T LM600, 5–600 yd, speed 300 km/h) or visually estimated the height. The number of cave entrances, including vertical ones were also recorded. Cave complexity index was calculated as the ratio of the total length of all passages to the maximum length between the first and last surveyed point in the cave (Arita,1996, Brunet and Medellín, 2001). Additionally, at each point, surface temperature (°C) and ambient temperature (°C) were recorded using an Infrared Thermometer (UNI-T UT301D, −32–600 °C) and relative humidity (%) using a hand-held humidity and temperature datalogger (Extech-RHT20). Microclimate measurements were taken once during each cave visit. GPS coordinates and elevation of each cave were also recorded at the cave entrance.

Human activities were recorded included the presence of bamboo poles, remains of bonfires, torches, nets, quarrying, vegetation clearing, urban development, infrastructure near caves, garbage disposal, damage to cave structure, tourism activities, mineral extraction, guano collection and construction within caves. Human disturbance at each cave site was additionally assessed based on seven criteria: human activities (“1” point if a human sign was present and “0” if absent), accessibility (“1” point if cave could be accessed by car or motorcycle and “0” if not), distance to the nearest road (“1” point if the distance was within 1 km and “0” if greater than 1 km), distance to the nearest village (“1” point if the distance was within 1 km and “0” if greater than 1 km) (Kingston, 2013, Vargas-Mena et al., 2020), disturbance around cave (“1” point if human disturbance activities were inside the cave such as cave drawings or destroying the cave walls and “0” if absent), guano collection (“1” point if present and “0” if absent) and cave development (“1” point if caves were developed by humans for either religious or tourism purposes by construction and “0” if not). We adapted the Biotic Vulnerability (BV) component of the Bat Cave Vulnerability Index 3.0 (BCVI-S; Tanalgo and Hughes, 2025) to match our binary human-disturbance dataset. The disturbance variable (e.g., human encroachment, guano collection, vehicle access, cave development, distance to roads/villages) was scored as 1 = disturbance present (highest vulnerability) or 4 = absent (lowest vulnerability). We then averaging across threats to obtain BV (BV = ΣLandT/N°T). And caves were classified into vulnerability categories following Tanalgo and Hughes (2025): A (1.00–1.99), B (2.00–2.99), C (3.00–3.99), D (4.00).

Feasibility of cave exploration was categorized into three levels: “1” if entrances and pathways inside were relatively simple to access and traverse. These caves could be easily walked into and explored without major difficulties, “2” if occasional obstacles were present, resulting in difficulty walking or requiring some climbing, crawling, or jumping, and “3” if specialized equipment was required, such as ropes and rigging techniques for ascending and descending. These caves may have unstable substrates, loose rock formations, and wet or slippery conditions. Water inside each cave was evaluated using a binary scale: “0” for the absence of water, “1” for the presence of water, and “NA” for water presence was not assessed.Table 2

Table 2.

Cave ecological factors and bat genera present in Paung Laung, Shan State, Myanmar.

Site	Name	Estimated # of Individuals by Season		Species Genus Level		Complexity Index	Average humidity (%)		Average Surface Temp (°C)		Water Cover*
		Rainy	Dry	Rainy	Dry		Rainy	Dry	Rainy	Dry	Rainy	Dry
1.	NTY Cave 1	200	2	Aselliscus sp. Hipposideros sp. 1 Hipposideros sp. 2 Rhinolophus sp	Hipposideros sp. 1 Asellicus sp Rhinolophus sp	1.54	84.25	82.80	19.93	18.2	1	0
2.	NTY Cave 2	500	200	Hipposideros sp	Hipposideros sp Myotis sp	2.86	87.79	83.25	19.59	17.6	1	0
3.	Hti Phar Cave	200,000	1000	Hipposideros sp	Hipposideros sp	1.10	73.30	47.95	22.22	19.50	0	0
4.	Buddha Head Cave	50	0	NR	NA	1.00	83.50	NR	20.70	NR	0	0
5.	Hotel Cave 1	200	2	Rhinolophus sp Aselliscus sp	Rhinolophus sp Aselliscus sp	1.11	79.10	72.30	19.20	18.43	1	1
6.	Lay Tet Cave	20	0	NR	NA	1.21	83.53	89.00	20.35	19.5	0	0
7.	Naung Pi Home	25	10	NR	NR	1.00	82.70	NR	25.70	NR	0	0
8.	Dragon Cave	100	0	Rhinolophus sp Hipposideros sp	NA	1.64	79.07	NR	19.78	NR	0	0
9.	Paung Laung Cave 3	500	1000	NR	Asellicus sp Eonycteris sp	1.00	83.80	69.93	24.40	19.60	0	0
10.	Paung Laung Cave 1	800	600	Rhinolophus sp Hipposideros sp	Aselliscus sp	1.61	83.70	89.25	25.03	23.25	1	1
11.	Eden Cave	1000	150	Hipposideros sp	Hippposideros sp Aselliscus sp	1.00	96.23	91.28	23.87	19.58	1	1
12.	NanPalin House	0	NV	NA	NV	1.00	96.23	NV	19.20	NV	0	NV
13.	Lwe Kaung Pagoda	0	NV	NA	NV	1.00	73.30	NV	25.03	NV	0	NV
14.	Hti Ri Cave 1	4	0	Aselliscus sp Rhinolophus sp	NA	1.00	79.20	NR	17.60	NR	0	0
15.	Big Bang Cave 1	50	NV	NR	NV	NR	NR	NV	NR	NV	1	NV
16.	Paung Laung Cave 2	0	0	NA	NA	NR	NR	NR	NR	NR	0	0
17.	Banana Cave	150	NV	NR	NV	1.60	80.73	NV	24.17	NV	1	NV
18.	Saung Paung Cave 1	NV	3	NV	NR	2.00	NR	77.93	NR	14.98	NA	0
19.	Saung Paung Cave 2	NV	500	NV	Rhinolophus Sp. Aselliscus Sp. Hipposideros Sp. Myotis sp	2.00	NV	82.19	NV	18.54	NV	0
20.	Phayar Phyu Cave	NV	200	NV	Rhinolophus Sp.	NA	NV	71.06	NV	16.76	NV	0
21.	Hti Ri Tayin Cave 2	NV	0	NV	NR	2.00	NV	54.23	NV	19.27	NV	0
22.	Hti Ri Tayin Cave 1	NV	10	NV	Hipposideros sp Myotis sp	2.00	NV	56.82	NV	17.37	NV	1
23.	Hti Ri Tayin Cave 3	NV	0	NV	NA	1.66	NV	77.18	NV	21.23	NV	0
24.	Hotel Cave 2	NV	1	NV	Myotis sp	2.00	NV	71.08	NV	16.98	NV	1
25.	Hotel Cave 3	NV	2	NV	NR	1.00	NV	77.60	NV	19.40	NV	1
26.	Hotel Cave 4	NV	0	NV	NA	1.00	NV	76.40	NV	17.60	NV	0
27.	Pat Kaw Cave	NV	0	NV	NA	2.00	NV	64.47	NV	16.92	NV	0
28.	Tat Ohn Cave	NV	0	NV	NA	NR	NV	72.78	NV	15.53	NV	0
29.	Saung Paung Cave 3	NV	1500	NV	Hipposideros sp Asellicus sp Myotis sp	2.00	NV	73.07	NV	16.22	NV	1
30.	NTY Cave 3	NV	0	NV	NA	1.00	NV	NA	NV	NA	NV	0
31.	Htan Saung Ra Cave 1	NV	1	NV	Rhinolophus Sp.	2.00	NV	67.31	NV	16.71	NV	0
32.	Htan Saung Ra Cave 2	NV	0	NV	NA	2.00	NV	69.26	NV	16.74	NV	0
33.	Htan Saung Ra Cave 3	NV	2	NV	Rhinolophus Sp. Myotis sp	2.00	NV	66.90	NV	16.13	NV	0
34.	Hti Ri Cave 2	NV	0	NV	NA	NR	NV	NA	NV	NA	NV	0
35.	Big Bang Cave 2	NV	1	NV	Rhinolophus Sp.	2.00	NV	68.47	NV	17.00	NV	1
36.	Myauk Hti Pwar Cave	NV	10	NV	Rhinolophus Sp. Myotis Sp.	NA	NV	NA	NV	NA	NV	0
37.	Saung Nein Cave	NV	7	NV	Rhinolophus sp Aselliscus sp Myotis sp	2.00	NV	NA	NV	NA	NV	0
38.	Pot In Cave	NV	200	NV	NR	NR	NV	64.43	NV	17.50	NV	0
39.	Nyan Pin Gyi Cave	NV	3	NV	NR	NR	NV	66.44	NV	16.94	NV	0
40.	Paw Yar Cave 1	NV	5	NV	NR	2.00	NV	67.78	NV	16.15	NV	0
41.	Paw Yar Cave 2	NV	0	NV	NA	2.00	NV	50.15	NV	16.25	NV	0

NV = not visited NR = not recorded NA = not available

^*Water cover (1 = if present, 0 = if absence)

2.3. Land use and land cover characteristics

The landuse and land-cover variables at the landscape-level (Table 3) were extracted within a one-kilometer buffer zone of sampling sites. These data were acquired from the open-sourced web portal European Union’s Copernicus Land Monitoring Service Information (EUCGLS) which provides satellite images for the classification of land use and land cover (Buchhorn et al., 2020). The latest land cover data provided by EUCGLS is from 2018 at 100 m resolution and images were acquired in Google Earth Engine (GEE). Three land use types representing the most common land uses in the study site (forest, agriculture/crop cover, and all other landuse) were extracted. Proportion of forest and crop coverage were calculated relative to the three available landuse types for the areas surrounding each cave in a 1 km buffer area (Table 3). Distance to the nearest village, road and Paung Laung (the center of the forest) were calculated by using the Euclidean Distance tool in ArcGIS (10.4.1).

Table 3.

List of dependent and independent variables and their characteristics in rainy and dry season.

Category	Abbreviation	Variables descriptions (Unit)	Mean (SD) in rainy season (n = 17)	Mean (SD) in dry season (n = 39)
Bats	abund	Number of individuals of all bat species in each cave	11,976 (48,453)	145 (340)
Cave characteristics	surface area	Total surface area (m2) of each cave	1693 (2494)	2313 (3237)
	complexity	Cave complexity index	1.35 (0.64)	1.57 (0.49)
	elevation	Altitude above sea level (m)	1148 (464)	1300 (329)
Climate	humidity	Relative humidity of each cave (%)	83 (5)	71 (11)
	temperature	Surface temperature of cave (°C)	21.34 (2.39)	17.85 (1.82)
Landscape	crop	Proportion of the crop area in 1 km buffer area around each roosting site or cave (%)	34 (31)	31 (29)
	forest	Proportion of forested area in 1 km buffer area around each cave (%)	57 (37)	63 (33)
Human influence indexes	distance to road	Euclidean distance of each cave to the nearest road (km)	4.95 (6.34)	4.45 (4.65)
	distance to village	Euclidean distance of cave to the nearest village (km)	3.10 (3.26)	0.98 (2.08)
	distance to_PL town	Euclidean distance of cave to PL town, the center of the project location(km)	32.87 (10.91)	33.97 (6.99)

n = Number of caves

PL=Paung Laung

^*All data types numerical

2.4. Survey of cave-dwelling bats and bat abundance

The bat population size of each cave was measured by non-invasive methods (observation and photographic) to minimize the disturbance to bat populations. In large caves or caves with multiple entrances, different approaches and combined methods were employed. To minimize disturbance, only two trained observers conducted the bat counts using headlamps and torch lights. In the case of large groups, a fraction of the group was counted and multiplied by the size of the colony to measure the abundance of bats. While detailed keys were followed (Bates and Harrison, 1997; Francis, 2019), identification to the species level was not possible without further genetic identification in many cases. Photographs of the roosting bats were taken and later evaluated to identify bat taxonomic classification at the genus level. The identification process followed the detailed descriptions and keys provided by Bates and Harrison (1997) and Francis (2019), enabling accurate genus-level classification.

2.5. Assessing local knowledge and behaviors around cave and cave-dwelling bats

A semi-structured questionnaire was implemented to collect information on human demographics, livelihoods, local knowledge, and activities relevant to caves and cave-dwelling bats (e.g., hunting practices, consumption of bats, guano collection practices), as these factors can affect the status of bat species which are residing near human communities and risk for zoonotic disease transmission. Questions were open and closed-ended and focused on understanding how human activities can influence habitat occupancy and level of disturbance in bat populations (Supplementary file #2). This approach and questionnaire were evaluated by the University of California, Davis IRB and determined to not be human subjects research (UC Davis IRB determination 2091398). Adult participants were selected using a convenience sampling method of individuals who were residing or working close to bat caves; however, respondents consisted of a variety of demographic groups and occupations (Figure S1), therefore representing broad perspectives from the community. Consent from the respondents was acquired verbally before starting the questions. A total of 65 respondents from 24 villages who voluntarily participated were interviewed.

2.6. Statistical analysis

Multivariable generalized linear regression modeling with a negative binomial distribution was used to evaluate all potential environmental and anthropogenic variables (Table 3) association with bat abundance. We included the total surface area (m²) of each cave (i.e., surface area), the cave complexity index (i.e., complexity), the relative humidity of each cave (i.e., humidity), the surface temperature of the cave (°C) (i.e., temperature), the proportion of the crop area in the 1 km buffered around each roosting site or cave (i.e., crop) and distance to road as an index for human influence in the full model. The proportion of the forested area, elevation, distance to village, distance to Paung Laung town, and complexity were not included in the full model due to high correlations with other variables (Pearson correlation < 0.5; Supplementary Table S1) (Zuur et al., 2010). Season could not be included in the full model because not all caves were accessible during the rainy season and therefore including season as a variable would have decreased our sample size. Additionally, preliminary analyses showed a high correlation between season and environmental variables like temperature and humidity, risking multicollinearity if included directly (Supplementary Figure S7). To address skewness and normalize the distribution of the predictor variables, a natural logarithmic transformation was applied where appropriate. This transformation also helps in stabilizing the variance, improving the interpretability and numerical stability of the regression models. Additionally, all variables were standardized by subtracting the mean and dividing by the standard deviation by using “scale” function in R to ensure they were contributing equally to the analysis, facilitating a more straightforward comparison of regression coefficients. The full model was built using the “glmmTMB” package in R (Brooks et al., 2017). To identify the most appropriate statistical model for our study on bat populations, we performed stepwise model selection using backward elimination based on Akaike’s Information Criterion (AIC) (Burnham and Anderson, 2002) by using “dredge” function in “MuMIn” package (Bartoń 2022) in R version 4.5.0 (R Development Core Team, 2021). This process started with the full model, predictors were systematically removed based on their contribution to model fit, assessed by AIC, until the most parsimonious model was obtained. Incident rate ratios (IRRs) and their 95 % confidence intervals (CIs) are shown for the final model (Table 4).

Table 4.

Multivariable negative binomial regression model predicting the number of bats present in a cave site.

Variablea	IRRb	95 % confidence interval of IRR	p-value
Surface area of cave	3.38	1.28, 8.92	0.01
Temperature of cave	3.52	1.19, 10.43	0.02*
Distance to road	2.39	1.23, 4.66	0.01
Cropc	1.92	0.83, 4.46	0.13 ns

ns = not significant

^aResults shown are from the best model (i.e., with smallest AIC) using a negative binomial distribution.

^bThe incident rate ratio (IRR) reflects the relative influence on the expected number of bats in a given cave site for a given category compared to the reference category.

^cCrop was retained in the final multivariable model, despite not remaining significant, because it reduced AIC.

Descriptive statistics were used to evaluate the knowledge of caves and cave-dwelling bats, frequencies and percentages of socio-demographic characteristics and behaviors of 65 respondents living or working near cave sites (Figure S1, S2, S3, S4 & S5). The Chi-squared test was used to determine significant differences between bat related activities among different demographic groups (age (<30, ≥30), gender (male, female), educational status (≤middle school, ≥ high school) and occupation (farmers and others)) and cave locations.

3. Results

3.1. Caves characteristics

A total of 41 sites including 38 caves, 2 houses and 1 pagoda were surveyed (Table 1 & 2). Most of the caves were located at higher elevations (>1000 m) of the southernmost part of the Shan plateau while three were at lower elevations inside the Paung Laung Reserved Forest. The length of the caves ranged from 2 m (Hotel Cave 4) to 325.2 m (Saung Paung Cave 3). The estimated surface area of the caves ranged from 19.87 m² (Hotel Cave 4) to 11,262.1 m² (Saung Paung Cave 3). Out of the 38 caves surveyed, 19 had a single entrance. The cave with the maximum number of entrances was Dragon Cave, which had a total of 6 entrances, closely followed by Hti Ri Tayin Cave 1, which had a total of 5 entrances. The access feasibility of 41 studied sites showed that 8 caves (19 %) were classified as difficult to access, 15 caves (37 %) as moderately accessible, and 15 caves, 2 houses, 1 pagoda (44 %) as easy to access. The cave complexity index for the 38 studied caves ranged from 1.00 to 2.86. Ten caves had an index of 1.00, indicating they were single-tunnel caves, while the most complex cave, NTY Cave 2, had an index of 2.86. Both cave humidity and temperature were statistically significantly higher in rainy compared to dry season (P < 0.001) (Supplementary Figure 7).

3.2. Human disturbance

Among the 38 caves that were surveyed, 31 caves (82 %) were found to have visible human disturbance/disruption for anthropogenic activities. Guano collection tools such as bamboo ladders and propylene bags with sand were found in most caves (Fig. 1). Hotel cave 1 that is situated in a newly constructed resort campus was found to be modified by tourism infrastructure such as constructed steel ladders, lights and a concrete road inside the cave. Other limestone caves were found to be renovated for tourism purposes, including artificial lighting and structural modifications such as viewing platforms. Other disturbances such as noises of visiting and immediate quarrying sites were also explored.

Among the 38 caves surveyed, the majority were classified as moderately vulnerable (BV class B; 24 caves, 63.2 %), indicating that most sites experienced a moderate level of human disturbance. A smaller number of caves were categorized as highly vulnerable (BV class A; 7 caves, 18.4 %), while BV class C caves (n = 7, 18.4 %) were also considered vulnerable but showed comparatively lower levels of disturbance than classes A and B. No caves were classified as very low vulnerability (BV class D). Overall, the distribution suggests that human disturbance is widespread across the surveyed region (Fig. 1) and (Fig. 2).

Fig. 2.

Distribution of Bat Cave Vulnerability (BV) classes across 38 surveyed caves. BV classes were derived following the Bat Cave Vulnerability Index (BCVI) framework and categorized as A – High vulnerability, B – Moderate vulnerability, and C – Low vulnerability. Numbers in parentheses indicate the number of caves assigned to each class.

3.3. Predictors of bat abundance

Five total genera of bat were identified across Paung Laung cave sites including Aselliscus (9 sites), Eonycteris (1 site), Hipposideros (9 sites), Myotis (8 sites), and Rhinolophus (12 sites) species with number of individuals varying by site (Table 2). Fourteen out of 19 sites in which bats were identified at the genus level were mixed genus roosts and abundance ranged from 2 individuals to > 200,000 individuals. Our final multivariable regression model included surface area, surface temperature, distance to the road and crop cover (Table 4). Among these variables, cave surface temperature exhibited the strongest influence (IRR = 3.52; P = 0.02), showing a significant positive association with bat abundance—warmer caves supported higher bat populations. Similarly, cave surface area had a notable positive effect (IRR = 3.38; P = 0.01), with larger caves harboring greater numbers of bats. Distance to the nearest road was also positively associated with bat abundance (IRR = 2.39; P = 0.01), indicating that bat populations increased in caves located farther from roads. In contrast, the proportion of crop area surrounding the roosting site was retained in the final model because its inclusion reduced AIC but it was not significantly associated with bat abundance.

3.4. Local knowledge and human activities involving caves and bats

During this study, sixty-five participants from 24 villages were interviewed, the majority being male (80 %), aged ≥ 30 years (82 %), and farmers (89 %) (Table S1) (Supplementary tables).

Bat knowledge.

Nearly all respondents (97 %) observed bats, mainly roosting in caves (34 %), trees (29 %), and buildings (28 %). Most reported year-round presence (74 %), with higher numbers in the rainy season (87 %). Colony sizes ranged from < 10 to > 500 bats, with perceived changes linked to food availability (62 %) and weather (31 %).

General cave knowledge and human activities.

Most respondents (82 %) were aware of nearby caves, generally within 1–50 km. Few (11 %) reported changes in recent years, mainly tourism infrastructure or guano harvesting. Caves were primarily visited for guano collection (40 %) and general interest (37 %), with fewer visits for religious purposes (14 %), work (6 %), or bird hunting (3 %). Visitors were mainly local (78 %), with tourism peaking in dry season (72 %). Only 24 % noted cultural ties to caves (Table S2) (Supplementary tables).

Resource harvesting and bat consumption.

Over half (62 %) of the people interviewed were aware of guano collection, mostly for agriculture (89 %). Harvesting was infrequent (<5 times/year in 95 % of cases), seasonal (rainy season, 47 %), and low volume (11–100 kg annually, 56 %). Forty-two percent had heard of bat consumption, mainly as bushmeat (74 %) or for medicinal purposes (22 %). Men were more likely than women to report knowledge of bat consumption (p < 0.05).

4. Discussion

We have described a previously unexplored karst ecosystem in Pinlaung Township, Southern Shan State. The results of this study provide important insights into the key environmental factors that predict bat abundance in cave habitats with potential implications for better understanding high risk cave sites for disease emergence and prioritizing conservation. This study identified cave surface area, temperature, and distance to the road as significant predictors of bat abundance with cave surface area having the greatest relative influence. Larger caves were strongly associated with higher bat abundance, a finding that aligns with previous studies (Silva et al., 2020). These larger caves may offer more roosting sites, and better protection from predators and environmental extremes, better supporting larger bat populations (Furey and Racey, 2016).

Temperature had a significant positive effect on bat abundance. This result is consistent with the thermoregulatory needs of bats, as warmer caves can offer more favorable conditions for both roosting and reproduction (Rodríguez-Durán, 2009). Warmer temperatures might reduce the energetic costs of maintaining body heat, especially in tropical or subtropical environments, thereby allowing bats to conserve energy for other activities like foraging (Speakman, 2003). This finding also highlights the potential vulnerability of bat populations to shifts in microclimate, particularly in the context of climate change and habitat degradation. Over the past 60 years, Myanmar has seen increasing temperatures, more erratic rainfall, and a progressive alteration of its monsoon patterns with large variations across the country (Oo, 2023). Myanmar is consistently ranked among the world’s top most climate vulnerable countries.

Distance to the road was positively associated with bat abundance, with caves farther from roads supporting more bats. This trend suggests that human activity, as measured by proximity to roads, negatively impacts bat populations, likely due to disturbance, noise pollution, or habitat degradation (Berthinussen and Altringham, 2012). Roads can also act as barriers to bat movement, reducing access to critical foraging areas and increasing mortality due to vehicle collisions (Medinas et al., 2013). In addition, hunting was identified as a factor in this study area and caves closer to roads may be more accessible by hunters. Infrastructure development including road building was occurring at an accelerated rate between 2016 and 2021 during the brief economic opening of the country.

Bat abundance was also found to be positively correlated with proportion of land in crop production surrounding caves. Inclusion of this variable in our final multivariable model decreased AIC, although the variable did not remain significant. Cropland conversion is the main driver of deforestation in Myanmar (Yang et al., 2019). Plantations and crop production are generally considered to negatively impact bat biodiversity (Azhar et al., 2015; Frick et al., 2020), but certain generalist species can flourish (Tanaglo et al., 2021; Azhar et al., 2015). This has been demonstrated to be particularly true for specific species of fruit bats, as cropland may provide foraging grounds, help interconnect habitats and decrease the risk of predation (Cely-Gomez et al, 2013; Azhar et al., 2015). Insectivorous bats are thought to be negatively impacted by crop production as they are assumed to be highly sensitive to the use of pesticides on crops (Bayat et al., 2014), and potential loss of insect prey (Frick et al., 2020). However, in some settings the impacts of crop production on bat populations seem to be somewhat mitigated by landscape heterogeneity (Kahnonitch et al., 2018). Cave roosting, forest-interior species, like many of the bats studied here, may also be more resilient to cropland conversion to a certain extent (Kingston et al., 2013).

In this study, the number of bats varied seasonally and was particularly lower in dry season. This seasonal pattern is consistent with the interviews from community members who noted bats were more commonly seen during the rainy season. This year-round usage of karst caves by bats however emphasizes the importance of karst systems in the conservation of cave-dwelling bats as potential sites for roosting, courtship and mating, maternity, and hibernation (Culver and Pipan, 2019), although we couldn’t identify any specific maternity roosts in our study.

To evaluate disturbance consistently across sites, we used the Biotic Vulnerability (BV) component of the Bat Cave Vulnerability Index 3.0 (BCVI-S; Tanalgo and Hughes, 2025), which focuses on human-driven threats and is appropriate when species-level abundance data are unavailable. This simplified approach allows comparisons among caves despite data limitations, while remaining conceptually aligned with the broader BCVI methodology. Our results show that human disturbance is widespread across the surveyed region, with nearly one-fifth of all caves classified as highly vulnerable and none falling into the very low vulnerability category. This pattern highlights the growing pressure on karst ecosystems and the urgent need for conservation measures to protect cave-dwelling bats and their habitats. This information is essential for conservation management and planning within Myanmar. Setting national level priorities is crucial for developing effective policy and targeting conservation locations (Doi and Takahara, 2016; Rudd et al., 2011). Cave sites currently categorized as moderate vulnerable and vulnerable were found to be under increasing threats of further human activities, which will likely result in roost abandonment and population declines as found in previous studies (Deleva and Chaverri, 2018, Phelps et al., 2018, Quibod et al., 2019). We recommend non-visiting zones be demarcated for caves containing cave dwelling bat species and specifically during maternal roosting periods, allowing them to maintain at least a portion of the cave undisturbed.

In addition, bat hunting, although reported infrequently in this study, was identified in the region which can have direct impacts on bat populations. Moreover, bat hunting and guano harvesting have also been identified as risk behaviors for spillover of zoonotic diseases, particularly coronaviruses among rural communities involved in extractive industries, such as logging, hunting, and forest product harvesting, where close contact with wildlife is common (Evans et al., 2020; Evans et al., 2023). Southeast Asia, and particularly Myanmar is thought to be a hotspot for emerging and reemerging coronaviruses because of the high distribution and diversity of Rhinolophus bats as well as remaining high biodiversity of natural habitats. (Anthony et al., 2017, Coker et al., 2011, Sarah et al., 2022).

4.1. Conclusions and conservation implications

The present study indicates cave-dwelling bats in Southern Shan state, Myanmar are impacted by environmental factors and anthropogenic disturbances and highlights the conservation importance of this Karst landscape in the region. Most of the caves observed and the biodiversity closely tied to these caves were threatened by multiple factors, especially bat hunting, guano collection, and cave development. These activities pose immediate threat to limestone karst and cave-dwelling bats and bring risk for zoonotic disease transmission. Most caves were of priority status for increased conservation activity which emphasizes the need for policies to maintain healthy cave ecosystems for the conservation of critical habitats for bats in Myanmar. Given the significant influence of cave surface area and temperature on bat abundance, efforts should focus on preserving large cave systems and maintaining favorable microclimatic conditions within these habitats. Genetic analysis is recommended for systematic analysis of species-specific abundance, and diversity in each cave, and should be considered when prioritizing selection of caves for conservation efforts. Additionally, mitigating human disturbances, particularly in areas close to roads, could help to preserve bat populations and prevent potential spillover of zoonotic diseases. Creating buffer zones around caves and restricting human activity near critical bat habitats may help to sustain the bat populations and limit human-bat contact, especially for species that are particularly sensitive to habitat disturbance. In addition, public awareness should be increased to promote the protection of limestone karsts and cave-dwelling bats in Myanmar and throughout the region. This is for the benefit of bats, the surrounding environments they support and human public health.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.gecco.2026.e04071.

Data availability

Data will be made available on request.