Patterns and Drivers of Subterranean Biodiversity Hotspots across the Globe

Abstract

Hotspots of cave biodiversity were mapped globally on the basis of recently published species lists of cave-limited invertebrates and vertebrates. Lists of sites with 20 aquatic or 20 terrestrial cave-limited taxa were mapped on a global scale, with special reference to Pleistocene glaciation, and present-day ecoregions. Twelve sites had 20 or more aquatic species; 17 had 20 or more terrestrial species. Only three sites from the Dinaric Karst had both 20 aquatic and 20 terrestrial species. Aquatic sites were either chemolithoautotrophic or along a ridge at about 45 (Europe) and at about 35 degrees north (in the United States). The terrestrial sites were chemolithoautotrophic or along the ridge but were also found in moist forests in tropics and subtropics. We describe emerging patterns of biodiversity hotspots and discuss the drivers of these patterns, including historical climate change, cave systems density, and resource availability, and highlight conservation challenges associated with protecting these unique ecosystems.

The extreme environments of caves and other subterranean habitats harbor a bizarre appearing fauna, specialized for these habitats. These aphotic, resource-poor environments that are more unchanging than any surface environments are home to thousands—perhaps tens of thousands—of aquatic (Gibert and Culver 2009) and terrestrial species (Deharveng and Bedos 2018), which lack eyes and pigment and which have elongated appendages and elaborated extraoptic sensory structures (Fenolio 2016). The known ranges of these specialized species are small, sometimes limited to a single cave, and they have a small population size (Bregović et al. 2019); if for no other reason, they are therefore at risk of extinction. This rarity combined with the distinctive morphology of subterranean species and their specialization to live exclusively in subterranean habitats makes the fauna an important conservation target.

Recent studies have shown that the pattern of species richness of the subterranean fauna is unlike the predominant pattern of the surface fauna and flora of tropical richness with a decline toward the poles (Hillebrand 2004) and unlike any pattern displayed by the surface fauna and flora (e.g., Mittermeier et al. 2004, Zagmajster et al. 2010). However, this pattern of species richness has remained elusive and incomplete. The consensus view until recently of the factors influencing the global pattern of subterranean species richness is that the climate changes brought about by Pleistocene glaciations were the main drivers of colonization and subsequent isolation of species in caves, at least for the terrestrial fauna (Culver et al. 2006, Anderson et al. 2013). Climate and other paleogeographic events further in the past played a role as well, such as drying in the Miocene in western Australia (Leys et al. 2003) and uplift of carbonate massifs in Europe during the Miocene (Borko et al. 2021). Notable early proponents of the Pleistocene hypothesis were Rene Jeannel (1943), a leading French speleobiologist and confirmed Lamarckian, and Thomas Barr (1968), a leading American speleobiologist and early neo-Darwinian student of cave fauna.

Cave-dwelling species are typically rare and challenging to collect because of the difficulties associated with collecting and sampling in caves. Furthermore, the ranges of most cave specialists are exceedingly small, as a result of the difficulties that cave-restricted species have in moving between caves, illustrated by the high level of β diversity relative to α diversity at a regional scale (Gibert and Deharveng 2002, Malard et al. 2009, Zagmajster et al. 2018). Getting a complete picture of subterranean biodiversity therefore requires intense sampling of different sites in a region (Culver et al. 2006). There are additional problems, especially with respect to undescribed species (Culver et al. 2012, Eme et al. 2018), but, compared with the rest of the world’s, European and to a lesser extent eastern United States subterranean biodiversity patterns have been well studied (Zagmajster et al. 2014, Eme et al. 2015, Bregović and Zagmajster 2016, Christman et al. 2016).

Because of inadequate sampling and a lack of formal taxonomic descriptions, especially in the tropics, reconstructing patterns in the rest of the world has proved difficult. Species lists by country or state are of limited use because these numbers are conflated by different areas and different numbers of caves sampled. Culver and Sket (2000) took a radically different approach, and rather than attempting estimates of regional diversity, they focused on caves with high numbers of species—20 or more per cave or karst well (see box 1 for definitions of key terms). This approach had several advantages. Most importantly, it allowed for global comparisons, because good species lists are available for caves throughout the world (see Deharveng and Bedos 2019). Hotspot caves (or regions) are defined as caves (or broader regions) with higher species diversity than others. The cutoff for defining a hotspot is necessarily arbitrary and is set to allow for consistent differentiation between low- and high-diversity sites. Hotspot caves may also reflect overall subterranean diversity because species accumulation curves infrequently cross so that α and β diversity may be correlated across sites (see Dole-Olivier et al. 2009). Finally, it may be the case that hotspot regions actually have only a few hotspot caves (Culver et al. 2004, Zagmajster et al. 2018), making individual hotspot caves a good predictor of regional diversity.

Box 1.

Glossary of terms.

The following terms were defined in Culver and Pipan (2019).

Chemolithoautotrophs: Organisms that produce their own food by extracting energy from inorganic chemical reactions rocks, e.g., sulfur bacteria.

Dinaric Karst: A limestone massif extending from extreme northeast Italy to Albania.

Epikarst: The highly porous uppermost zone of karst.

Hyporheic: Interstitial spaces within the sediments of a stream bed, a transition zone between surface water and groundwater.

Karst: Landscape in soluble rock where solution rather than erosion is the primary geomorphic agent, typically with caves, sinkholes, and springs.

Stygobiont (stygobite): Obligate, cave-restricted species of aquatic subterranean habitats.

Troglobiont (troglobite): Obligate, cave-restricted species of terrestrial subterranean habitats.

In this article, we evaluate and tentatively explain the patterns of subterranean hotspot occurrences for the aquatic and terrestrial cave-restricted species fauna. Previous studies (Deharveng et al. 2024) analyzed terrestrial and aquatic sites together and considered data quality in depth. They did not focus on causal explanations of diversity patterns. We analyze the terrestrial and aquatic fauna separately because they often have quite different evolutionary histories, and their range of subterranean habitats (and therefore the opportunities for colonization and dispersal) are very different. The ranges of aquatic subterranean species are larger (Zagmajster et al. 2014), presumably because of enhanced opportunities for dispersal. Among the primary noncave terrestrial subterranean habitats are interstitial habitats (such as the deep soil) and cracks and spaces between rocks (Mammola et al. 2016), whereas aquatic subterranean habitats include groundwater and interstitial spaces between rocks in stream beds (Ward and Palmer 1994, Culver and Pipan 2014).

Survey of environmental variables

For analyzing both aquatic and terrestrial patterns we considered the following environmental information: Pleistocene glaciation and the extent of the permafrost (extents of ice cap and permafrost at the last glacial maximum; 25,000–17,000 years ago, all mapped after Ehlers and Gibbard 2004a, 2004b, 2004c, Lindgren et al. 2016, and Petherick et al. 2022); distribution of karstified rocks, the limestones and dolomites where solutional caves occur (https://www.arcgis.com/apps/mapviewer/index.html?webmap=85e6f1b575a74ebbba09bcb380169cb7); higher than usual resource levels, as a result either of chemolithoautotrophy (Howarth and Stone 1990, Engel 2019) or of high surface net primary productivity (Eberhard and Howarth 2021); and world ecoregions and biomes distribution map (after Dinerstein et al. 2017). All maps were produced using ArcGIS Pro 3.1.3 (ESRI 2024).

The hotspots data used are those assembled by Deharveng and colleagues (2024) largely from a series of papers on hotspot caves published in the journal Diversity and described in more detail below.

Defining subterranean hotspots

The initial global hotspots list by Culver and Sket (2000), based on literature and researchers’ input, identified 20 caves and wells with 20 or more cave-restricted species. The impact of this initial publication led to subsequent claims that other caves had hotspot status (e.g., Souza Silva and Ferreira 2016, Ferreira et al. 2020). The thresholds used to define hotspots vary across studies and depend on the goals of the analysis, taxonomic scope, and data completeness. For example, Culver and Pipan (2013) applied a 25-species threshold to total troglobiotic richness (terrestrial and aquatic combined), whereas we use a threshold in the present study of 20 species for either aquatic or terrestrial obligate species to allow separate analysis of ecological groups.

Pipan and colleagues (2020) was a call for papers on hotspot caves for the journal Diversity. The lists of sites with 25 or more cave-restricted species generated by the 26 papers in Diversity were checked by the editors (DCC, LD, and TP) for consistency and accuracy, and small changes in numbers were made in the original lists; these numbers were published, along with a few sites not included in the Diversity special issues (Jourde et al. 2011, Souza-Silva and Ferreira 2016, Martínez and Gonzalez 2018) in table 2 in Deharveng and colleagues (2024). We used this table to generate the lists of sites with at least 20 aquatic or terrestrial cave-restricted species, because the patterns are obviously different for the two groups, and called for separate analyses (Culver and Pipan 2013, 2019, Bregović et al. 2019), counting a total of 26 hotspots and 25 references.

Cave-restricted species are often termed stygobionts (stygobites) if they are aquatic and troglobionts (troglobites) if they are terrestrial. This terminology and ancillary terms such as troglomorphic are confusing to the nonspecialist, especially because they are not always used in the same way and because they may be unnecessary (Culver et al. 2023). To promote clarity and accessibility for a broader audience, we sparingly use these terms in the text with the following meaning. Stygobionts are cave-restricted species in aquatic subterranean habitats, with or without obvious morphological modifications to subterranean life. Troglobionts are cave-restricted species in terrestrial subterranean habitats, regardless of whether they exhibit morphological modifications associated with subterranean life.

A case can be made that the best ecological comparison among caves would be the list of the permanent stream inhabitants, including not only the cave specialists but also the cave generalists. As a practical matter, this comparison is not possible because lists of cave generalists are infrequently available and because it is difficult to distinguish resident from visitor for nonspecialized species. For many species rich caves, such generalists are rare (e.g., Vjetrenica; Delić et al. 2023) or ignored.

Aquatic hotspots

The list of sites with 20 or more specialized aquatic species is shown in supplemental table S1; a total of 12 sites were identified. Sauve Spring, a site very near Cent Fonts in France, is also likely to be 20 or more specialized aquatic species (Prié et al. 2024), as is the cave Logarček, a site near the Postojna Planina Cave System, in Slovenia (Culver and Sket 2000). No complete lists are available for either of these sites, but their inclusion would not change the geographic distributions of aquatic subterranean hotspots. We know of no sites with 18 or 19 specialized aquatic species.

Of the 12 aquatic hotspots, 4 (Ojo Guareña, Baget, Robe River Well, and Križna Jama) are dominated by microscopic crustacea, oligochaetes, and flatworms. With the exception of the loss of Australia, the elimination of these sites because of inadequate sampling of these groups in other sites does not significantly change the geographic pattern of concentration of species in caves with chemolithoautotrophy or caves along a band in southern Europe (figure 1).

Figure 1.

Map of aquatic hotspots shown as dots, plotted against karst areas and extents of ice cap and permafrost at the last glacial maximum (25,000–17,000 years ago). Source: The map was produced using ArcGIS Pro 3.1.3 (ESRI 2024), with permafrost and ice cap data based on Ehlers and colleagues (2004a, 2004b, 2004c), Lindgren and colleagues (2016) and Petherick and colleagues (2022). The detail map shows the Mediterranean region.

The pattern of hotspots has two aspects. One are the sites in mainland Spain, France, Slovenia, and Bosnia and Hercegovina that correspond to a previously reported ridge of high aquatic and terrestrial diversity (Culver et al. 2006, Zagmajster et al. 2014). Their explanations for the ridge were high net primary productivity and high habitat availability (i.e., cave density). However, Eme and colleagues (2015) showed that productivity interacts with spatial heterogeneity and historical climate to produce the overall pattern of aquatic subterranean biodiversity in Europe. The other sites (Walsingham Caves, San Marcos Well, Robe River Well, Túnel de la Atlantida, and Comal Springs) are anchialine (Walsingham Caves and Túnel de la Atlantida) and chemolithoautotrophic (San Marcos Well, Robe River Well, Comal Springs). Anchialine habitats are typically high in organic matter relative to other subterranean habitats (Pohlman et al. 2000). Chemolithoautotrophy has not been directly demonstrated in the Robe River Well, but the presence of redox conditions strongly suggests the possibility (Humphreys et al. 2009, Engel 2019).

As was indicated in figure 1, 6 of the 12 sites—French, Bosnian, and Slovenian, lying along the 45° ridge of European hotspots—are at or close to the maximum extent of permafrost, which would block dispersal. The paradox is that blocking dispersal may reduce local diversity (Christman et al. 2005, Borko et al. 2021), but climate forcing may outweigh less dispersal, and dispersal blocking may be of relatively short duration. Of course, these northern regions would also be less hospitable because of lower productivity. There is no apparent relationship between sites in North America and the extent of glaciation. There are karst areas closer to the permafrost, such as northern Virginia and West Virginia (figure 1).

The European ridge of aquatic subterranean hotspots (a total of seven) is at the boundary between the temperate and broadleaf and mixed forest zone and the Mediterranean forest, wetlands, and scrub (figure 2). The other five (chemolithoautotrophic or lava tubes) are in four different biomes: deserts, Mediterranean forests, temperate grasslands, and coniferous forests (table S1).

Figure 2.

Aquatic hotspots shown as dots plotted against ecoregions and biomes (after Dinerstein et al. 2017), produced using ArcGIS Pro 3.1.3 (ESRI 2024). The detail map shows the Mediterranean region.

Terrestrial hotspots

The 17 terrestrial hotspot sites are listed in supplemental table S2. The only additional potential hotspots that we are aware of, with 19 known terrestrial subterranean specialists, are Ganxiao Dong, in China (Huang et al. 2021), and Fern Cave, in Alabama, in the United States (Niemiller et al. 2023). In addition, Moutaouakil and colleagues (2024) reported that Aziza Cave in Morocco has 12 species and an additional 10 (mostly Collembola) are potentially troglobionts but are insufficiently studied to make a definitive statement. It is noteworthy that three of the hotspot sites have more than 40 species: Vjetrenica (in Bosnia and Hercegovina), the Postojna Planina Cave System (in Slovenia), and Cueva del Viento (in the Canary Islands). There is a gap of seven species between these three sites and the next most diverse site. Only three sites have both 20 aquatic specialists and 20 terrestrial specialists—Vjetrenica, the Postojna Planina Cave System, and Križna Jama (in Slovenia)—all in the Dinaric Karst (figure 3). Although we mention these three sites in the present article because of their inclusion on the terrestrial hotspot list, their uniqueness in supporting high diversity in both faunal groups is noteworthy. However, we do not pursue a combined analysis of terrestrial and aquatic diversity in this article, because the two groups differ substantially in their ecology, biogeography, and evolutionary history.

Figure 3.

Map of terrestrial hotspots shown as dots plotted against karst areas and the extents of the ice cap and permafrost at the last glacial maximum (25,000–17,000 years ago). The map was produced using ArcGIS Pro 3.1.3 (ESRI 2024). The permafrost and ice cap extents are mapped after Ehlers and colleagues (2004a, 2004b, 2004c), Lindgren and colleagues (2016), and Petherick and colleagues (2022). The detail map shows the Mediterranean region.

Of the 28 sites reported by Deharveng and colleagues (2024) to have 25 or more combined aquatic and terrestrial cave specialists, only 2 fail to have 20 aquatic or terrestrial species—Fern Cave (in the United States) and the Coume Ouarnède System (in France). All in all, there is remarkably little overlap between the aquatic and terrestrial hotspot caves, although they often share the same region.

The geographic distribution of terrestrial hotspots is very different from that of the aquatic species. The ridge of terrestrial biodiversity in Europe is present but weaker than the ridge of aquatic biodiversity (figure 3) although other evidence for it is strong (Culver et al. 2006). A total of four caves, all in the Dinaric Karst, are within the ridge—Vjetrenica, the Postojna Planina Cave System, Križna Jama, and Lukina Jama or Trojama—and are at or close to the maximum permafrost extent. What is missing are sites in France, although there are two sites with 17 terrestrial species—Baget (Bréhier et al. 2024) and the Coume Ouarnède System (Faille and Deharveng 2021). The reason for this is unclear, but perhaps the total sampled habitat is less than at other sites. In the original description of the ridge (Culver et al. 2006), a corner of northeast Alabama (later extended to include an adjoining part of Tennessee; Niemiller and Zigler 2013), was included because it had high species diversity, high cave density, and high actual primary productivity. This part of the ridge is represented by Crystal-Wonder Cave (table S2, figure 2). Therefore, a total of five caves are in the ridge.

As with the aquatic hotspots, there is a site—but only one—with chemolithoautotrophy: Movile Cave. Two sites are lava tubes, which also have higher than typical organic matter, because of the presence of tree roots (e.g., Oromí and Socorro 2021): the Undara Lava Tube System (in Australia) and Cueva del Viento (in the Canary Islands). Therefore, 8 of 17 terrestrial hotspots follow the pattern of aquatic hotspots. However, the actual hotspot caves are different for aquatic and terrestrial hotspots, with the exception of three caves in the Dinaric Karst: Križna Jama, the Postojna Planina Cave System, and Vjetrenica (tables S1 and S2).

The other nine hotspots are a different story. Only one (Mammoth Cave) is north temperate, and the rest are tropical or subtropical, scattered through the tropics with the exception of Africa (figure 2). Mammoth Cave is somewhat of an anomaly because of its immense size (more than 650 kilometers [km]) and thorough study (Niemiller et al. 2021). Six of these nine terrestrial hotspot caves are not chemolithoautotrophic or in the Dinaric Karst, are longer than 10 km (the Água Clara Cave System, in Brazil; Mammoth Cave, in the United States; Sistema Huautla, in Mexico; the Towakkalak System, in Indonesia; Feihu Dong, in China; and the Areias Cave System, in Brazil). Tham Chiang Dao (Thailand), 5 km long, is a window on a much larger karstic system (Deharveng et al. 2023). Hang Mo So is in a relatively small karst area in Vietnam but one with high cave density (Deharveng et al. 2023). The final cave, the Igatu Cave System, in Brazil, is enigmatic, being less than 5 km in length and in quartzite (Gallão et al. 2023), not a classic soluble rock. This pattern suggests that a large cave or large system of interconnected caves is necessary for a hotspot to occur. On the other hand, worldwide, there are 679 caves known to be more than 10 km in length (Berger 2024), so other conditions are also required.

The distribution of terrestrial hotspot caves by ecoregion (figure 4) provides additional constraints. All of 15 sites not chemolithoautotrophic or in lava are only in four ecoregions (figure 5): Mediterranean forests, woodlands, and scrub (2); temperate broadleaf forests (5); tropical and subtropical dry broadleaf forests (2); tropical and subtropical moist broadleaf forests (5); and tropical and subtropical coniferous forests (1) at the contact.

Figure 4.

Terrestrial hotspots shown as dots plotted against ecoregions and biomes (after Dinerstein et al. 2017). The map was produced using ArcGIS Pro 3.1.3 (ESRI 2024). The detail map shows the Mediterranean region.

Figure 5.

Map of known terrestrial hotspots shown as dots plotted against moist forest ecoregions in tropical and temperate zones (after Dinerstein et al. 2017). The map was produced using ArcGIS Pro 3.1.3 (ESRI 2024). The detail map shows the Mediterranean region.

What best explains the aquatic global pattern?

The distribution of aquatic hotspots is relatively straightforward. Five are in high productivity caves and aquifers (chemolithoautotrophic sites: San Marcos Well, Comal Springs; anchialine sites: Robe River Well, Túnel de la Atlantida, Walsingham Caves). It is likely that all of the sites in the European ridge of high subterranean biodiversity are also high in secondary productivity (see Eme et al. 2015). Although it is tempting to attribute the distribution solely to the effects of high productivity sites in the overall low productivity habitat of caves, the ridge of high diversity is also an area of high cave density (Culver et al. 2006), as is the United States part of the ridge for terrestrial diversity.

One portion of the biodiversity ridge, the Dinaric Karst, which ranges from northeast Italy, in the vicinity of Trieste, to Montenegro, is also distinctive in its proximity to ancient marine basins such as the Paratethys, which likely served as a source of aquatic colonizers often indirectly through freshwater ancestors (Sket 1999, Borko et al. 2021). In addition, the impact of the Messinian Crisis, when the Adriatic dried up, goes beyond the Dinaric Karst, in a complex way, to the Pyrenees. The very high values of species richness in the Postojna Planina Cave System and Vjetrenica compared with the French and Spanish sites (table S1) suggest also that proximity to the Adriatic plays some role. Culver and colleagues (2009) argued that the high species richness in the Edwards Aquifer (including San Marcos Well) is due in part to proximity to seas during the Cretaceous or due to higher connectivity allowing for greater dispersal.

Finally, in Europe, but not North America, the ridge roughly corresponds to the maximum extent of the Pleistocene permafrost, which may be a barrier to northern sites. In North America, regions much closer to the maximum permafrost (e.g., southern Indiana and northern Virginia) are not hotspot sites.

We are left with several explanations for the ridge: high net primary productivity, which should reduce extinction rates; high cave density, which should increase subterranean dispersal (see Christman et al. 2005), as does the large homogeneous area of the Dinaric Karst; historical proximity to the Parathetys (Sket 1999) and other Cretaceous seas (Hutchins et al. 2021) as a source of colonists, a factor in explaining the especially high species richness of Dinaric and Texas sites. This multiplicity of explanation, at the smaller scale of European subterranean biodiversity patterns, was described in some detail by Zagmajster and colleagues (2018).

What best explains the terrestrial global pattern?

The pattern of terrestrial hotspots (figure 3) contains aspects of the aquatic pattern (figure 1). That is, there are four sites in the Dinaric Karst (part of the European ridge of hotspots); two are in lava with enhanced organic carbon, and one is chemolithoautotrophic. To this can be added Crystal-Wonder Cave, which is in the extension of the biodiversity ridge in North America (northeast Alabama and adjoining part of Tennessee; Culver et al. 2006, Niemiller and Zigler 2013). The much smaller ridge in the United States shares the features of high species richness, high actual evapotranspiration, and high cave density.

However, the other nine present a new pattern and not one anticipated by previous workers. Culver and Sket (2000) placed particular emphasis on midlatitude richness; however, improved data coverage—especially from tropical regions—has revealed novel hotspots that were previously undetected because of limited sampling. There are several important clues as to the distribution of hotspots. The first is that, with the exception of the enigmatic Ugatu Caves, in Brazil, all of the nine are long caves (more than10 km) or are in areas with high density of caves. Therefore, the extent of available habitat would appear to be a limiting factor, a conclusion reached by Christman and Culver (2001) with reference to predictions of terrestrial cave species richness for counties in southeastern United States. The second is that all nine of these sites are in forested ecoregions (table S2), rather than grasslands or deserts. The third clue is that ecoregions where hotspots are found are largely moist rather than dry forests. Because actual productivity is generally higher in areas of more precipitation (figure 3), once again, it would appear that hotspots occur in regions of higher productivity. Finally, eight of the nine sites are tropical or subtropical, and this suggests that some different mechanism is responsible for these high-diversity sites, unless climate fluctuations in the tropics, especially with respect to precipitation (Anderson et al. 2013), are more important than was previously recognized.

How complete and comparable are the data?

The data presented in the present article may be incomplete either because some caves and sites are unknown or because known sites are incompletely known. Deharveng and colleagues (2024) reviewed geographic coverage of the known cave fauna and pointed out that there are many more sampled areas than there are hotspot areas.

Nonetheless, sampling is certainly incomplete. For example, Berger (2024) listed more than 650 caves longer than 10 km and some of these caves (perhaps less than 100) are in areas where there is a potential for hotspot caves. We can also compare the list of 20 cave and well hotspots reported by Culver and Sket (2000) with the top 28 caves for overall biodiversity reported by Deharveng and colleagues (2024), but using only the top 20 sites. Of the 20 cave and well hotspots (caves with 20 or more specialized aquatic and terrestrial species) reported by Culver and Sket (2000), nine appear among the top 20 caves on the hotspot list (with 25 or more specialized aquatic and terrestrial species) reported by Deharveng and colleagues (2024). Two additional caves from the Culver and Sket (2000) list have 25 or more species but are not among the top 20. Finally, Culver and Sket (2000) listed Logarček (in Slovenia) as having 43 species, but Deharveng and colleagues (2024) did not include it because no detailed species list was available. Therefore, in the 25 years since the publication of Culver and Sket (2000), about half of the sites reported are new. The paper itself drew lots of attention (more than 400 citations), so perhaps it encouraged extensive research interest and most sites have been found afterward. In figure 5, we show the distribution of moist tropical and temperate forested sites, perhaps the most likely places to find new hotspots.

There are a number of problems associated with the species lists, the most important being sampling bias (the Racovitzan impediment; Ficetola et al. 2019) and taxonomic bias (the Linnean impediment). The Racovitzan impediment results from inconsistent sampling of some habitats, particularly epikarst. Epikarst and other interstitial habitats such as the hyporheic (Culver and Pipan 2011) often harbor a number of specialized microcrustacean species, especially copepods (Deharveng et al. 2024). However, not all sites have microcrustacean included, so some sites have uncollected species. The scope of this problem is displayed in table S1, where species numbers are displayed both as a total and as nonmicrocrustacean species. The difficulty with such subtraction is that we don’t really know whether and how many such species occur in undersampled sites. Sites may appear to be undersampled but may just not have many species in particular habitats. Not all epikarst and hyporheic habitats have a rich fauna, so the elimination of microcrustacean species may greatly reduce major faunal differences among sites. In any case, the elimination of microcrustaceans does not fundamentally change the geographic patterns (table S2).

Conservation challenges of subterranean hotspots

Subterranean ecosystems are among the least understood environments on Earth. The invertebrates inhabiting these ecosystems are unique, often narrowly distributed, and face a variety of threats, including habitat destruction, pollution, climate change, and unsustainable tourism. The conservation of subterranean biodiversity hotspots is particularly challenging because of their inaccessibility, their limited legal protection, and the difficulty of enforcing conservation measures because of the cryptic nature of subterranean fauna.

There is a large array of factors that may negatively affect subterranean hotspots. This is not surprising, given their occurrence in a number of countries and under a variety of geological and cultural settings. Some caves, such as the Postojna Planina Cave System, in Slovenia, receive up to a million visitors annually, whereas others such as Sistema Huautla, in Mexico, remain remote, inaccessible sites. The range of threats for the hotspots we discuss in the present article is wide, but mining has the most destructive and irreversible impact. Three sites—Robe River Well (in Australia), the Igatu Cave System (in Brazil), and Hang Mo So (in Vietnam)—are particularly at risk. The ongoing destruction of karst around Hang Mo So is a sobering reminder of how difficult it is to mitigate economic pressures—in this particular case, the cement industry and its international stakeholders (Deharveng et al. 2023).

Broadly speaking, several of the subterranean hotspots considered in the present article benefited from three types of protection:

Legal designation. Some sites may be designated as protected hotspots by virtue of their own interest or under the umbrella of larger areas under protection (national parks, natural reserves; tables S1 and S2). For example, Mammoth Cave National Park (in the United States) benefits from federal protection, whereas Comal Springs (in the United States) in Landa Park is situated within a city recreational park with more limited safeguards (figure 6).

Figure 6.

Current formal legal designation of the aquatic and terrestrial subterranean hotspots: international (Ramsar and or UNESCO site); national (Natura 2000 sites, national monuments, or part of the national parks or protected through national legislation); local (only partially protected, as part of the regional park, in local custody); and no legal designation (not protected by any of the entities above, some located in remote, inaccessible areas). The detail map shows the Mediterranean region.

Remoteness and inaccessibility. On at least a temporary basis, some sites, such as Sistema Huautla (in Mexico) or Lukina Jama (in Croatia, also a Natura 2000 site) are naturally protected because of their extreme inaccessibility. In fact, human-accessible cave passages represent only a small fraction of the subterranean voids available to invertebrate fauna in a karst (Gibert and Deharveng 2002), providing a natural and efficient protection for most subterranean invertebrate communities.

Cave tourism. Tourism is a double edged sword. In areas of the cave devoted to tourism, there typically remains little of the fauna, especially because of light contamination. On the other hand, tourist caves that are managed responsibly can serve as a bulwark against economic forces that tend to destroy and degrade the sites, especially mining. Nine of the hotspot sites are tourist caves: Cueva del Viento, Tunel del Atlantida, Feihu Dong, Tham Chiang Dao, Hang Mo So, Križna Jama, the Postojna Planina Cave System, Vjetrenica, and Walsingham Caves. For most of these caves, the touristic part is small compared with the total extent of the cave, and the positive aspects may largely exceed the negative ones. That is, however, not the case for Hang Mo So, where the touristic pressure is high, but the cave and the remaining karst outcrop that contains the cave are small.

Significant protection of several biodiversity hotspots is achieved at different extents through one of the three ways examined above or a combination of them (tables S1 and S2). However, another parameter has to be taken into account. Although the disturbances linked to terrestrial habitats can be mastered locally through territorial protection measures, those associated with water are much less easy to tackle. Water pollution and hydrological alterations are recognized as major concerns for water biodiversity, but socioeconomical considerations often contain the implementation of appropriate measures. For instance, the Lez aquifer (in France) is affected near its source by urbanization, land-use changes, and massive water pumping, which affect groundwater biodiversity (Malard et al. 1996, Le Coent et al. 2023). Implementing territorial protection strategies in such densely populated areas to mitigate these effects remains challenging. Similarly, underground rivers such as those in the Postojna-Planina Cave System (in Slovenia) and in the Towakkalak System (in Indonesia) originate from stream sinks that are flowing as epigean in highly populated areas outside the protected perimeter. They are, therefore, responding to any significant change affecting water that could happen in the unprotected part of the watershed. The most complex case is the Vjetrenica cave system (in Bosnia and Hercegovina), where the cave entrance has been gated but where the hydrological alterations from outside streams have a deleterious impact on cave endemics (Delić et al. 2023). Addressing such problems through the legal designation of protected areas would imply that the whole hydrological basin of the site is considered, which is an impractical measure in many cases. Consequently, aquatic biodiversity is at significantly greater risk than terrestrial biodiversity is (Englisch et al. 2024, Mammola et al. 2024).

In regions of exceptionally high cave density and species richness—such as the northwest Dinaric Karst—it may be unrealistic to protect every species-rich cave individually; conservation efforts should focus on securing representative cave systems and vulnerable habitats under immediate threat.

Designing a globally meaningful strategy for conserving subterranean biodiversity hotspots would therefore require case by case assessments and tailored conservation measures. These should account the biological value and composition of the fauna, the nature and severity of the threats, and the environmental and socioeconomical context. Global priorities can be set primarily by assessing threat severity and faunal uniqueness, but in any case, implementation largely depends on national policies, necessitating adaptive and opportunistic approaches.

New directions

Deharveng and colleagues (2024) provided a valuable global overview of subterranean biodiversity, and our study builds on this work by examining aquatic and terrestrial patterns separately and providing a preliminary interpretation of possible drivers influencing the distribution of hotspots.

There are both shared and unique features of the terrestrial and aquatic maps of hotspots of subterranean biodiversity. There are two shared features. First, there is a north temperate ridge of subterranean biodiversity roughly at 45 degrees north in Europe and 35 degrees north in the eastern United States, areas of high actual productivity and high cave density, providing greater opportunity for both survival and dispersal. Second, there are scattered sites in lava tubes and chemoautotrophic caves and wells, also with enhanced actual productivity. There is an additional pattern for terrestrial hotspots, one of scattered sites in the tropics and subtropics (apparently excluding Africa) of large caves with high species numbers. The explanation for this pattern is not yet clear, but moisture could be one of the key environmental parameters. Humidity is not the only factor; for example, Clearwater Cave, in Sarawak, is in a hyperhumid region but is not a hotspot.

Ultimately, more information about the region in which each of the hotspot caves occur would be highly beneficial, but such detailed regional analyses are difficult because they require data on upward of 100 caves and, in fact, are probably not feasible in many areas. To our knowledge, only one such analysis has been reported since the original analysis by Culver and colleagues (2006), and that is the study of Niemiller and Zigler (2013) of the cave fauna of Tennessee. Incidentally, it provides more support for the existence of a ridge of high subterranean biodiversity in the United States.

To date, our analysis highlights the importance of midlatitude karst regions and tropical large-cave systems in supporting subterranean biodiversity. Conservation efforts should focus on protecting these unique ecosystems, ensuring that subterranean biodiversity continues to be recognized and safeguarded as a critical component of global biodiversity. Strengthening conservation measures, expanding legal protections, and promoting sustainable management are critical steps toward ensuring the long-term survival of these vulnerable habitats.

Author Biography

Magdalena Nǎpǎruş-Aljančič (magdalena.aljancic@zrc-sazu.si) is affiliated with the Research Centre of the Slovenian Academy of Sciences and Arts, Karst Research Institute, in Postojna, Slovenia and with the Tular Institute, in Kranj, Slovenia. Tanja Pipan is affiliated with the Research Centre of the Slovenian Academy of Sciences and Arts, Karst Research Institute, in Postojna, Slovenia and is the UNESCO chair on karst education at the University of Nova Gorica, in Vipava, Slovenia. Louis Deharveng and Anne Bedos are affiliated with the Institut de Systématique, Évolution, Biodiversité at the Muséum National d’Histoire Naturelle, at Sorbonne Université, in Paris, France. David C. Culver is affiliated with the Department of Environmental Science at American University, in Washington, DC, in the United States

Funding

MNA and TP received funding from the following EU projects: eLTER Preparatory Phase Project, eLTER Advanced Community Project, LifeWatch ERIC, and the Implementation of the international infrastructure project LifeWatch + eLTER (I0-E016).