*Article* **A Hotspot of Arid Zone Subterranean Biodiversity: The Robe Valley in Western Australia**

**Huon L. Clark 1, Bruno A. Buzatto 1,2,3 and Stuart A. Halse 1,\***


**\*** Correspondence: stuart.halse@bennelongia.com.au

**Abstract:** Knowledge of subterranean fauna has mostly been derived from caves and streambeds, which are relatively easily accessed. In contrast, subterranean fauna inhabiting regional groundwater aquifers or the vadose zone (between surface soil layers and the watertable) is difficult to sample. Here we provide species lists for a globally significant subterranean fauna hotspot in the Robe Valley of the Pilbara region, Western Australia. This fauna was collected from up to 50 m below ground level using mining exploration drill holes and monitoring wells. Altogether, 123 subterranean species were collected over a distance of 17 km, comprising 65 troglofauna and 58 stygofauna species. Of these, 61 species were troglobionts and 48 stygobionts. The troglofauna occurs in small voids and fissures in mesas comprised mostly of an iron ore formation, while the stygofauna occurs in the alluvium of a river floodplain. The richness of the Robe Valley is not a localized aberration, but rather reflects the richness of the arid Pilbara region. While legislation in Western Australia has recognized the importance of subterranean fauna, mining is occurring in the Robe Valley hotspot with conditions of environmental approval that are designed to ensure species persistence.

**Keywords:** stygobiont; troglobiont; conservation; groundwater; arid zone; mining

#### **1. Introduction**

Subterranean habitats have little to no light, scarce organic matter, constantly high humidity, and a much more stable temperature than the surface [1]. Unsurprisingly, the vast majority of studies of troglofauna and stygofauna have been in caves that people can access or in streambeds (the hyporheic zone). Both types of habitats allow researchers relatively easy access to explore and collect animals [2,3]. It is often overlooked, however, that much more extensive habitats likely to support subterranean fauna occur worldwide. For stygofauna, these habitats occur in unconfined regional groundwater aquifers. For troglofauna, they occur in the vadose zone that extends from c. 2 m below the ground surface to the watertable. Abundant small spaces for subterranean animals are present in geologies ranging from detritals (e.g., colluvium, alluvium, scree, etc.) to sedimentary rocks, often at tens of meters depth [4–6]. Sampling troglofauna and stygofauna from these environments, which have no connection at the human scale to the surface, is obviously much more difficult than exploring large cave systems or digging in streambeds and the documentation of their biodiversity has lagged behind the general study of subterranean fauna.

Recognition that Western Australia supports rich stygofauna communities in arid zone aquifers began in the 1990s, with sampling at Cape Range on the mid-west coast and in the Pilbara and Yilgarn regions [7,8] (Figure 1). The richness of stygofauna across most Pilbara landscapes has been confirmed by a large regional survey [9,10] and many surveys associated with the environmental impact assessment of mining projects [11,12]. Recognition that equally rich troglofauna communities are present did not begin until

**Citation:** Clark, H.L.; Buzatto, B.A.; Halse, S.A. A Hotspot of Arid Zone Subterranean Biodiversity: The Robe Valley in Western Australia. *Diversity* **2021**, *13*, 482. https://doi.org/ 10.3390/d13100482

Academic Editors: Tanja Pipan, David C. Culver and Louis Deharveng

Received: 9 September 2021 Accepted: 29 September 2021 Published: 30 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

stygofauna surveys in the Robe Valley during 2004 and 2005 also collected troglofauna, mostly schizomids [13,14]. Knowledge of both troglofauna and stygofauna in the Pilbara has continued to grow in the past 15 years, and it was recently estimated that 1511 troglofauna and 1329 stygofauna species occur in the region [5], with the caveat that the precision of these estimates is limited by a weak taxonomic framework for certain fauna groups.

**Figure 1.** The Pilbara region of Western Australia, showing subterranean fauna communities recognized as significant for conservation. The Robe Valley hotspot is at the western end of the Robe Valley Pisolitic Hills.

The richness of subterranean fauna in the Pilbara has been explained as partly a result of many aquatic and mesic-adapted terrestrial invertebrate species escaping the harshness of surface habitats during the Miocene aridification by moving underground, sometimes with dramatic subsequent radiation [15,16]. Schizomids are a well-documented case; these typically forest litter-dwelling animals moved into subterranean habitats as forest habitat was being lost, and then radiated in these habitats [17,18]. Equally impressive radiations have occurred, however, in stygofauna groups such as candonid ostracods [19], syncarids [20] and amphipods [21,22], which are essentially subterranean groups and would not have colonized from the surface. Information from a regional survey suggests that, as a result of these radiations, there are about 30% more stygofauna than surface aquatic invertebrate species in the Pilbara [23]. As far as we are aware, this greater richness of subterranean fauna is unique, but comparative data for other regions are scarce because of the rarity of published regional inventories.

While the Pilbara can be regarded as an area of globally significant radiation of subterranean fauna, here we focus on the richness of troglofauna and stygofauna in the western part of the Robe River catchment (Figure 2). We call this the Robe Valley, although it is a mostly flat landscape. The area has previously been identified as a hotspot for troglofauna [24] and stygofauna [25]. Cape Range, to the south-west of the Pilbara, the

offshore Barrow Island, and the inland Ethel Gorge have also been identified as globally significant hotspots in, or near, the Pilbara ([24], Figure 1).

**Figure 2.** Components of the Robe Valley subterranean fauna hotspot.

In addition to providing species lists for sites in the Robe Valley, we have three goals in this manuscript. One is to highlight the importance of vadose zone and groundwater aquifers for troglofauna and stygofauna. The second is to illustrate that, in many parts of the world, huge numbers of subterranean species remain to be discovered and described. The third is to point out that water extraction and mining, particularly for iron ore but also gold, base metals, lithium, potash and rare earth metals, may sometimes threaten the conservation of subterranean communities. At the same time, we recognize that it is the wells and exploration drill holes of these developments that provide the portal that has enabled the discovery of the otherwise inaccessible and hidden subterranean fauna of the Pilbara region.

#### **2. Materials and Methods**

We confirm previous identification of the Robe Valley as a subterranean fauna hotspot by compiling lists of species for discrete areas or sites within this hotspot. For troglofauna, we provide lists for three small mesas (Mesas A, B and C) on the west flank of the Robe River, approximately 47 km from the river mouth (Figure 2). The mesas represent topographically distinct features in the landscape (Figure 3), and their outlines as defined here can be seen in Figure 2. For stygofauna, we provide lists of species collected from two wells (Robe 2A and 1A) near the Robe River as it crosses the coastal plain, approximately 8 km north of the mesas. The wells are portals to a stygofauna community for which the geographical boundary is not obvious at the surface.

**Figure 3.** Upper panel, Mesa A, showing hardcap on top of the mesa; inset A, a scrape net; inset B, a troglofauna trap. Lower panel, Robe 2A well showing steel collar set in concrete; inset A, a stygofauna net. Photos by Claire Stephenson, Mike Scanlon and Melanie Fulcher.

Troglofauna were sampled at the mesas for the purposes of environmental impact assessments and monitoring compliance with the conditions for mining. Most of the sampling was undertaken by Biota Environmental Sciences, although in some years other consultants (including Bennelongia) or mining company staff in conjunction with consultants, undertook the work. The predominant sampling method was trapping, whereby

between one and four cylinders of PCV containing moist leaf material were suspended at varying depths in a series of vertical exploration drill holes. The drill holes (diameter 130 mm) had 2–3 m of casing at the surface (called a collar) to prevent the collapse of sand and loose rock into the hole, but the remainder of the hole was in direct contact with the surrounding substrate. Traps were left in place for 6–8 weeks. A smaller number of samples were collected by a technique called scraping, whereby a net was lowered to the bottom of the exploration hole and scraped along the substrate surface while being retrieved [26]. Approximately 1100 trap or scrape samples were collected from Mesa A (21.66072 ◦S, 115.89547 ◦E; hole depth range: 8–88 m) in 11 years of sampling between 2004 and 2017. This equates to about 450 sampling events (i.e., site/date combinations) because multiple traps were set in most holes, and from 2014 onwards holes were also scraped. Sampling effort was lower at the other two mesas, with approximately 400 samples and 180 sampling events during seven years between 2005 and 2016 at Mesa B (21.65894◦S, 115.94547◦E; depth range: 15–70 m) and 300 samples and 180 sampling events in three years of sampling between 2005 and 2017 at Mesa C (21.68928◦S, 115.96206◦E; depth range: 15–125 m). Troglofauna traps were set in an iron ore formation called channel iron deposit (CID) at depths between 5 m and 70 m below ground level, with the median depth being 20 m. Animals were recovered from depths up to 50 m.

Stygofauna were collected mainly by net hauling at Robe 2A (well G70730102, 21.58123S◦S, 115.87047◦E) and Robe 1A (well G70730101, 21.57542◦S, 115.88266◦E). These wells are 14 m and 23 m deep, with standing water level fluctuating between 4–9 m and 6–10 m below ground level, respectively. This level of groundwater fluctuation is characteristic of Pilbara floodplains. The wells had a steel collar extending approximately 1 m above and below the ground surface and a 100 mm diameter PVC casing inside the collar that extended to the end of the hole. The casing was slotted continuously below the watertable. A small, weighted plankton net, with a diameter only slightly smaller than the cased hole, was lowered to the base of the hole, agitated vigorously to stir 'benthic' fauna into the water column, and then retrieved. Two sizes of mesh (50 μm and 150 μm) were used in different hauls. Pump samples were also collected on two dates, whereby three times the bore volume was pumped through a 50 μm-mesh net [9]. Twelve sampling events occurred at Robe 2A in five years of sampling between 2002 and 2007, while eight sampling events occurred in the three years between 2002 and 2004 at Robe 1A.

#### *Determining Species Numbers and Subterranean Affinity*

The taxonomic framework for the identification of most groups of subterranean fauna in the Pilbara is poor. As a result, even when there was competent species-level discrimination by different survey teams, the compilation of species lists for the hotspots was difficult. Species-level discrimination was mostly achieved using morphology, often coupled with molecular sequencing, but also by sequencing alone. Very few species have been described to date, so most species were identified by codes linked to voucher specimens or molecular sequences. In a few cases, when Western Australian Museum staff work on a group and verified identifications, a museum voucher name was used. More often codes were generated by individual consultancies. Representative specimens of most putative species are lodged in the Western Australian Museum.

Another part of the process of compiling the species lists was determining the subterranean affinity of specimens. This is more important for troglofauna, because aquatic species collected from groundwater beneath a completely dry terrestrial surface rely entirely on subterranean habitat for persistence in that setting. On the other hand, scrapes and traps usually collect an order of magnitude more surface animals than troglofauna. While most surface drop-ins can be easily identified as such, some surface soil species lack pigmentation and eyes as a result of their specialization to life in topsoil and leaf litter, and it can difficult to distinguish between them and troglofauna. We considered species to be troglofauna if they are known only from scrapes or traps and possess some troglomorphic traits in addition to lack of pigmentation and eyes. Pauropods and symphylans are the

groups with greatest potential for surface species to be treated as troglofauna but soil and litter sampling in the Pilbara has collected very few species of either group [26; Bennelongia, unpublished data], suggesting few species in these groups are drop-ins. Mites and collembolans were omitted from the lists, reflecting advice from the Western Australian Museum to the environmental impact assessment process many years ago that few species in these groups are subterranean. We recognise that this is incorrect for some species.

#### **3. Environment and Geology**

The Pilbara Craton is one of the oldest and most stable geological regions in the world, with basement rocks dating as far back as 3.5 billion years [27]. Uniquely, this basement rock in the form of granite and greenstone terrains remains largely exposed [28], particularly in the north of the craton [5]. Throughout the south, and to a lesser extent in the north, the basement rock is overlain by depositional sedimentary rocks from the Archaean to Proterozoic time periods [29]. The major ranges of hills in the Pilbara (i.e., the Hamersley and Chichester Ranges) consist largely of iron-rich sedimentary rocks, mostly banded iron formations. Less resistant rock formations (such as the dolomitic Wittenoom formation) have been highly weathered, with geologies such as CID being deposited in the resultant drainage lines [29].

Despite geological stability, the Pilbara climate has changed substantially over the past 20 million years from cool and wet to hot and arid as Australia separated from Antarctica and began moving northwards [16]. The Pilbara now has very hot summers, moderate winters, and low but highly variable annual rainfall. Mean daily maximum temperatures exceed 35 ◦C from October to March, and the hottest recorded Pilbara day is 50.5 ◦C (at Mardie near the mouth of the Robe River in February 1998). Mean annual rainfall at Mardie is 275 mm, with 54% of this rain falling between January and March, mostly from cyclones or monsoonal rain. A further 33% falls between April and June, mostly from late cyclones or large cold fronts that extend to the Pilbara from the south at the start of winter. The main source of groundwater recharge is cyclones [30].

The prospective habitats for troglofauna in the Robe Valley, and the Pilbara generally, mostly comprise centimeter to millimeter-scale fissures and vugs. In some ways, these Pilbara habitats are similar to the small microcaverns radiating out from limestone caves, although limestone microcaverns often become blocked with detritus and are unsuitable for troglofauna [31]. The Pilbara habitats are perhaps more directly comparable with canga and other ferruginous formations in South America, although caves are more common in South American formations and are the focus of most studies [32].

The major troglofauna habitat in the Robe Valley lies within the flat-topped mesas that typically stand 30–50 m above the surrounding landscape. The flat top consists of laterised hardcap, with a thickness of up to 10 m (Figure 3). This cap has a near vertical face at the edges of the mesas, below which is a sloping face of detritals. The core geology of the mesas is CID. Details of the genesis of this CID in the palaeochannel of the Robe River are contentious, but it is agreed CID began to be laid down after sediment in-filled the Robe River palaeovalley during the Palaeogene, then larger areas of CID formed in the Miocene [33,34]. Subsequent surface flow events eroded the in-fill and left the mesas observed today [33]. The typical stratigraphy of the mesas consists of a weathered hardcap (loosely analogous to Brazilian canga) above layers of CID. The CID is predominately made up of iron-rich pisoids, which are 2–10 mm rounded bodies that are bonded together by a geothitised secondary matrix. Over time, this sedimentary rock has become weathered, principally through groundwater leaching, leaving vugs and other voids in the matrix. Some of the voids have gradually combined to form larger spaces, resulting in limited connectivity of spaces throughout the matrix. Most of these subterranean spaces now sit above the watertable and have been colonised by troglofauna. Mesa A has an approximate area of 5.7 km2, Mesa B 1.7 km2 and Mesa C 2.3 km2. Relative humidity in the subterranean habitats of Robe River mesas is 100% [5].

Near the surface, the troglofauna habitat in sections of the Robe River mesas, and in the Pilbara more generally, may be considered to have similarities with Milieu Souterrain Superficial (MSS) habitat, especially on valley slopes [35]. However, this is not the case at depth, where the troglofauna sampling was undertaken. The mesas, and Pilbara as a whole, mostly consist of consolidated rock habitat, including bedrock, rather than scree and loose rock.

The habitat for stygofauna in the Robe Valley is provided mainly by aquifers in CID or alluvium. Robe 2A and 1A lie in an alluvial aquifer associated with the active channel of the Robe River as it crosses the Pilbara coastal plain; CID occurs in the vicinity if not at the wells themselves. The surface provides little indication of the rich stygofauna community below (Figure 3). Depth to the watertable varies from 4–10 m, depending on season and year, with salinity in the top metre of groundwater being 2400–5500 μS/cm and temperature ranging from 30.5–32.4 ◦C during two years. The Robe River flows, on average, less than once a year and most of the riverbed is dry the remainder of the time. Flood volumes can be large (mean annual flow is 108 GL [36]).

#### **4. Results**

Altogether, 65 species of troglofauna have been collected from Mesas A, B and C, with 35 species in Mesa A, 30 species in Mesa B and 17 species in Mesa C (Table 1, Figure 4). Two species, the millipede *Lophoturus madecassus* and the pseudoscorpion *Tyrannochthonius aridus*, are also known from surface habitats and are troglophiles [37,38]. Two other species, the hemipteran *Cixiidae* sp. B02 and the dipteran *Allopnyxia* sp. B01, are treated as troglophiles. It has been shown genetically that they are widespread across the Pilbara. Adult *Allopnyxia* sp. B01 has a vestigial eye, consisting of a dark crescent with a few variable-sized ommatidia and, therefore, may have some surface dispersal. *Cixiidae* sp. B02 has obligate subterranean nymphs with eyes that become larger as they approach the adult stage. Adults are capable of surface dispersal (Bennelongia, unpublished data). All other species are troglobionts known only from Mesas A, B and C, and usually from only one of these mesas. They are extreme short-range endemics.


**Table 1.** Troglofauna species collected from Mesas A, B and C. Records with an asterisk represent identifications above species level, but they are usually aligned with a related species to reduce table size. Species in bold have special conservation protection. Comments show species delimited using CO1 results, with divergence values from other species indicated where available. # indicates species are troglophiles, whereas all other species are likely troglobionts.


**Table 1.** *Cont.*

Only 12 of the 65 troglofauna species at the mesas are described, but assignments to at least 34 other species have been supported by sequencing, and another three have been through the process of receiving museum codes. Many additional species, such as the cockroach *Nocticola* 'OES11', the spider *Trichocyclus* sp. 'MesaA', various beetles, the diplurans *Japygidae* sp. and *Projapygidae* sp., and the centipedes *Cormocephalus* sp. and *Geophilidae* sp., are the single occurrence of their families in a mesa (Table 1), which ensures it is valid to recognize each of these taxa as representing at least one unit in the species list.

**Figure 4.** Troglofauna at the mesas. (**A**) *Trinemura* 'MesaA1'; (**B**) *Ptinella* sp. B01; (**C**) *Japygidae* sp.; (**D**) *Troglarmadillo* 'Helix-ISA005'; (**E**) *Allopnyxia* sp. B01; (**F**) *Tyrannochthonius aridus*; (**G**) *Draculoides* sp.; (**H**) *Prethopalpus scanloni*. Photos by Jane McRae.

The most difficult issue when compiling species lists was reconciling the multiple species names used in different surveys. Some problems were caused by different levels of identification (earlier surveys tended to have less taxonomic information available and hence less rigorous identifications). Others resulted from lack of clear diagnoses associated with each voucher name, leading to conflicting views about which species animals belonged to. For example, delimitation of the five species of *Ideoblothrus* at Mesa A in an area <6 km2 has been confused, particularly in relation to *Ideoblothrus* 'MesaA', which was described but not formally named [39]. The 'type' animal could not be sequenced, but some other animals identified by the Western Australian Museum as *Ideoblothrus* 'MesaA' appear genetically to be *Ideoblothrus linnaei*, which was described in the same paper. Whether some museum identifications were wrong, or whether *Ideoblothrus* 'MesaA' should be treated as *Ideoblothrus linnaei,* remains unclear. Delimitation is further complicated by co-occurrence with *Ideoblothrus* 'MesaA2', 'MesaA3' and 'MesaA4'. *Ideoblothrus* 'MesaA2' and 'MesaA3' are separated from other species by a borderline 9% sequence divergence in CO1 (Table 1).

Five troglofauna species at Mesa A and one at Mesa B are listed in Western Australian species conservation legislation or associated mechanisms. These are two schizomids (*Draculoides anachoretus* and *D. bythius*) that are listed as Vulnerable under the Western Australian *Biodiversity Conservation Act 2016* and three pseudoscorpions (*Lagynochthonius asema*, *Tyrannochthonius* 'MesaA' and *Ideoblothrus* 'MesaA') that are informally listed by the government conservation agency as Priority 1 species (defined as poorly known from

few locations and potentially under threat). In addition, the troglofauna communities and their habitats at Mesas A, B and C, as well as nearby mesas in the Robe Valley, are listed informally by the conservation agency as a priority ecological community, namely the Priority 1 *Subterranean invertebrate communities of mesas in the Robe valley region*. Another Priority 1 troglofauna community, the *Subterranean invertebrate community of pisolitic hills in the Pilbara,* occurs immediately west of the mesa community (Figure 2). Priority 1 communities are poorly known but considered to have restricted occurrence and to be potentially threatened.

At least 58 species of stygofauna have been collected from the Robe 2A and 1A wells, with 46 species at Robe 2A and 31 species at Robe 1A (Table 2, Figure 5). All species collected are reliant on groundwater for their persistence at the site, although the cyclopoid copepod *Apocyclops dengizicus* is cosmopolitan and would usually be classed as a stygophile [40]. Twenty-nine of the species are described, largely through taxonomy funded in association with a large biodiversity survey of the Pilbara [41]. An additional species, the thermosbaenid *Halosbaena* sp. PL, is undescribed, but has been well studied genetically [42]. A lack of diagnostic information for the multiple names used (in different surveys) for syncarid taxa resulted in them being reduced to two species, Bathynellidae sp. and *Atopobathynella* 'A'. This probably underestimates the number of syncarid species present. There is also taxonomic and ecological uncertainty associated with oligochaete records. The *Enchytraeus*, Phreodrilidae and Tubificidae species names in Table 2 represent morphological clades to which immature animals are assigned, and each clade potentially contains multiple species. The groundwater dependence of enchytraeids is uncertain, and they perhaps should be regarded as amphibious species [43], although collected commonly in stygofauna sampling in Western Australia and apparently subterranean. Similarly, some tubificids are stygophiles, and the groundwater dependence of rotifers, nematodes and the leech is unknown. In total, 48 species are recognised as stygobionts.


**Table 2.** Stygofauna species collected from wells Robe 2A and Robe 1A. Records with an asterisk represent identifications above species level. # indicates species are stygophiles, ‡ indicates uncertain status; all other species are likely stygobionts.

**Table 2.** *Cont.*


Unlike the mesas, which have unique assemblages of species associated with easily delineated topographic features, the spatial extent of the stygofauna hotspot is unclear. Of the four wells within 2 km of Robe 2A, only Robe 1A was sampled by the team that sampled Robe 2A. The other wells yielded few species, but Robe 2A and 1A appear to have the same rich stygofauna community, with 61% of the species collected at Robe 1A also found at the better sampled Robe 2A. In addition, five described ostracod and copepod species, three undescribed ostracods, two undescribed amphipod species, a rotifer and a nematode not found at Robe 2A were collected from Robe 1A. The amphipods include a new and distinctive lineage of Eriopisidae that is widespread in coastal parts of the Pilbara

and the low-lying Fortescue Valley, as well as a species of *Wesniphargus,* representing a family (Neoniphargidae) that is relatively uncommon in the Pilbara (Table 2, Figure 5).

**Figure 5.** Stygofauna at Robe 2A and 1A. (**A**) *Humphreyscandona fovea*; (**B**) *Wesniphargus* 'Helix-AMN004'; (**C**) *Namanereis pilbarensis*; (**D**) *Kagalana tonde*; (**E**) *Microcerberidae* sp.; (**F**) Eriopisidae gen. nov. 'Helix AMM006'; (**G**) *Nedsia* 'mcraei' King & Cooper in press; (**H**) *Halosbaena* sp. PL.; (**I**) *Peza* sp. Photos by Jane McRae.

Some described stygobiont species have relatively large linear ranges, e.g., the polychaete *Namanereis pilbarensis* (c. 500 km), the calanoid copepod *Stygoridgewayia trispinosa* (>300 km), the cyclopid copepods *Halicyclops rochai* (>300 km), *Diacyclops* species (400–600 km) and *Orbuscyclops westaustraliensis* (>800 km), as well as the harpacticoid copepods *Pseudectinosoma galassiae* (>300 km), *Schizopera roberiverensis* (c. 300 km), *Megastygonitocrella* species (600–700 km) and *Parastenocaris jane* (600 km, Bennelongia unpublished data). Ostracods tend to have smaller ranges, e.g., *Deminutiocandona aenigma* (c. 200 km), *Humphreyscandona* species (80–190 km), *Pierrecandona posteriorrecta* (100 km), *Pilbaracandona rosa* (120 km) and *Areacandona* species (50–60 km). Most other species can be considered short-range endemics [9].

The relatively small amount of comprehensive genetic work done on stygofauna (rather than comparisons of individual specimens a few kilometres apart to determine whether they are conspecific) suggests it is likely that some, if not all, of the stygofauna species currently regarded as widespread are, in fact, either cryptic species complexes or have very high intraspecific variability (or both). For example, what was previously regarded as *Halosbaena tulki* is now regarded as comprising up to five different species, including *Halosbaena* sp. PL [42].

No stygofauna species from Robe 2A and 1A are listed for protection under conservation legislation, and neither is the community. However, the community of an area to the southeast of the mesas, namely the Priority 1 *Stygofauna community of the Bungaroo aquifer,* has been listed (Figure 2). It contains at least 17 (30%) of the species present in the Robe 2A and 1A wells.

#### **5. Discussion**

The voids in groundwater aquifers and the vadose zone are amongst the least accessible continental habitats for fauna. Hence, it is unsurprising they harbour a subterranean fauna that is not as well characterized as that of streambeds and caves. Sampling yields are low, especially for troglofauna, where the combination of scraping and trapping often yields less than one animal per site and one species for every four sites sampled [26]. Net hauling for stygofauna usually yields about 50 animals and three species per site in areas where stygofauna is abundant [10]. A major contributing factor to these low yields is the disconnection between the sampler on the surface and the animals many metres below in a hidden matrix. In this situation, acquiring fine scale habitat and ecological information to improve sampling yields is difficult. Use of packers and optical viewers [44] can provide information on habitat preferences in small areas but is too expensive for large-scale deployment. Despite these difficulties, sampling of wells and drill holes has enabled documentation of a globally significant number of subterranean fauna species in the Robe Valley. Altogether, 123 species have been recorded over a distance of 17 km. Troglofauna sampling was concentrated in an area of approximately 16 km2. The spatial extent of the stygofauna community documented at Robe 2A and 1A is unknown but probably <4 km2. Of the 123 subterranean species collected, 61 species are considered troglobionts and at least 48 species are stygobionts.

When comparing species richness of the Robe Valley with other subterranean fauna hotspots, some of our assignments of species as troglobionts may seem controversial to readers who think of subterranean animals outside caves as deep soil fauna. However, all troglofauna sampling occurred in CID, which is consolidated rock. Unlike caves, where animals can position themselves along a subterranean gradient according to their ecological requirements, nearly all species collected in the Robe Valley mesas spend their full life cycles in environmental conditions typical of the deep sections of caves [5]. This is also seen in stygofauna, for which there is relatively little overlap in the assemblages of the hyporheic zone and regional aquifers [45]. We treat the stygofauna species that occur in both habitats as stygophiles.

Recognition of the Robe Valley as a hotspot for subterranean fauna is not new. The troglofauna community at Mesa A has previously been identified as a hotspot with 24 troglofauna species [24]. Additional data for Mesa A and the inclusion of Mesas B and C species explains the much larger number of species recognised in Table 1. The stygofauna community at Robe 2A has also previously been recognized as a hotspot with 32 stygobiont [25] or 54 stygofauna species [24], depending on how survey data were interpreted [10]. The stygobiont list [25] excluded rotifers, worms and mites, and the stygofauna list [24] included some taxa that are treated here as duplicate names. Table 2 includes data from well Robe 1A.

The landscape of the Pilbara region as a whole (approximately 250,000 km2) is estimated to support 1511 troglofauna and 1329 stygofauna species [5], nearly all of which we classify as troglobionts or stygobionts. The subterranean fauna richness of the Robe Valley does not stand out in a major way from many other parts of the Pilbara, with a compilation of impact assessment surveys for troglofauna between 2007 and 2016 showing that the two major ranges of hills in the Pilbara to the east of the Robe Valley (Hamersley and Chichester)

are also important for troglofauna [12]. In fact, survey data suggest the richest parts of the Hamersley Range contain a greater density of troglofauna species than the Robe Valley (Bennelongia, unpublished data). Broadscale stygofauna survey of the Pilbara between 2001 and 2005 identified nine areas, including the Robe Valley, as important for stygofauna [10]. Subsequent compilation of impact assessment surveys for stygofauna between 2007 and 2016 identified additional areas in the eastern Pilbara as important [12]. Overall, the Pilbara region appears to have a significance for subterranean fauna similar to that of the Dinaric karst in Europe, which has historically been regarded as the most speciose region globally [46,47]. The Pilbara has an estimated 1511 mostly troglobiont species [5] compared with 995 troglobionts in the Dinaric karst [46], and 1329 mostly stygobiont species [10] compared with 680 stygobionts in the Dinaric karst [46]. The Postojna-Planina cave system, the richest hotspot in the Dinaric karst, supports 71 stygobiont and 45 troglobiont species [48], which is similar to the numbers for the Robe Valley hotspot.

As is the case in most areas outside Western Europe and North America [49], especially when considering fauna outside caves, the majority of troglofauna and stygofauna species in the Pilbara are undescribed [10,26]. This is less pronounced for stygofauna because of extensive taxonomic work on ostracods [19] and copepods [50]. With the current state of taxonomy and the complexities of species delimitation for subterranean species, some described Pilbara species are known, or likely, to be species complexes in which several cryptically different species hide under a single name [42,51,52]. On the other hand, the disjunct occurrence of the well-studied cave eel *Ophisternon candidum* in the Pilbara and surrounds [53] highlights that care should be taken when assuming large-range taxa are species complexes.

The lack of formal species description holds back conservation in two ways. Firstly, accurately determining the number of species in an area is difficult without sound taxonomy. A confusion of names arises when relatively little time can be spent during inventory surveys delimiting near-cryptic species, and when various surveys independently apply their own codes for the same undescribed species. This confusion can create doubts about the validity of species counts, which in turn reduces the credibility of calls to conserve areas or communities. Molecular sequencing is very useful for assigning individual animals, especially if juvenile or physically damaged, to described or vouchered species for which sequence information is available. It can also assist with species delimitation, but it does not replace formal taxonomy or the use of morphology when recognising new species, especially among subterranean animals where intraspecific genetic distances are highly variable and often large [18,52,54]. In situations where a species is encountered only once in an area and is subsequently being compared to a single animal from another area, it can be difficult to determine genetically whether the animals represent different species, different evolutionarily significant units, or just a genetically variable species [55]. Only additional sampling between the two areas or across the possible range of the species can clarify the issue. Furthermore, different genes seem to work better for different taxa and over-splitting into different genetic species, rather than lumping of morphologically cryptic species, seems to be a risk if the 'wrong' gene is selected for sequencing [56]. In this context, we point out that COI sequencing without morphological examination of animals has produced more taxa than expected for pseudoscorpions at Mesa A (*Ideoblothrus*, 5 species) and Mesa B (Chthoniidae, 3 species). While this may have led to a slight overestimate of troglofauna species richness in the Robe Valley, other factors such as sampling adequacy are likely to have caused underestimation.

The second way in which lack of species description obstructs conservation is by hampering the accumulation of ecological knowledge. Species descriptions provide a framework for the collation of information about the ecology and occurrence of different subterranean groups. Synthesis of this knowledge enables more informed decision-making in the environmental impact assessment process and supports future management and conservation of subterranean fauna [57].

Recognition of conservation values of subterranean fauna has strongly influenced land management at Mesas A, B and C, where high conservation values coincide with high economic value of the subterranean habitat. A compromise between fauna conservation and economic development means that Mesa A is currently being mined for iron ore, while future mining of Mesas B and C has been approved by government [58]. However, at all three mesas an outer 'shell' approximately 80–200 m wide will remain un-mined the full way around the mesa in recognition of the importance of protecting troglofauna values. These shells should provide a sufficient volume of mesa to ensure persistence of the troglofauna species restricted to each mesa. There has been extensive monitoring at Mesa A since mining began (reported to the Environmental Protection Authority) to ensure that all troglofauna species known from that mesa are persisting in the un-mined shell. The documentation of stygofauna values has also been important, with early consideration of a borefield in the vicinity of Mesa 2a and 1A abandoned. These trade-offs are an example of a process that is likely to become increasingly important worldwide as awareness grows of the occurrence of subterranean fauna outside caves and, at the same time, there is increased demand for the products of mining.

**Author Contributions:** Conceptualization, H.L.C., S.A.H., B.A.B.; data curation, H.L.C.; writing—original draft preparation, S.A.H., B.A.B., H.L.C.; writing—review and editing, S.A.H., B.A.B.; visualization, H.L.C., S.A.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** Funding for manuscript preparation was provided by Bennelongia Environmental Consultants.

**Institutional Review Board Statement:** Not applicable.

**Data Availability Statement:** Data are provided in Tables 1 and 2.

**Acknowledgments:** We thank the editors for inviting us to contribute to this special issue and accommodating our late contribution. We also thank the consultants who collected most of the data on which this paper is based. Vitor Marques assisted with mapping; Claire Stephenson, Mike Scanlon, Melanie Fulcher and Jane McRae provided photographs.

**Conflicts of Interest:** Bennelongia Environmental Consultants has sometimes undertaken monitoring of stygofauna and troglofauna in the Robe Valley for the Robe River Mining Company, including at Mesas A and B.

#### **References**


### *Editorial* **An Overview of Subterranean Biodiversity Hotspots**

**David C. Culver 1,\*, Louis Deharveng 2, Tanja Pipan <sup>3</sup> and Anne Bedos <sup>2</sup>**


#### **1. Introduction**

Riding a wave of interest in biodiversity patterns in surface-dwelling communities, in 2000, Culver and Sket [1] published a paper listing 20 caves and karst wells with 20 or more known species. At the time of their study, it was widely recognized that because of the extremely narrow distribution of most cave specialists, the overall species richness of a region or country was the result of the accumulation of species from a number of different caves, and that average single-site species richness (α-diversity) was a minor component of overall species richness (γ-diversity) [2]. This has since been confirmed in a number of largescale studies of subterranean species diversity, especially in Europe [3–6]. Although the number of studies measuring both α-diversity (local diversity) and β-diversity (betweensite diversity) has grown in the intervening 20+ years, they have largely been confined to Europe and, to a lesser extent, North America [3–7], mainly because it requires extensive datasets for about 100 sites for these to be estimated [6,8]. However, it is also well known that considerable subterranean species α-diversity lies outside these regions, including in Australia [9,10], Brazil [11,12], China [13], South East Asia [14], and Mexico [15], in part as a result of a search for additions to the list of hotspot caves reported by Sket and Culver.

The collection of papers in this Special Issue of *Diversity*—"Hotspots of Subterranean Biodiversity"—expands and enriches previous hotspot lists [2,14,16,17]. These hotspots may not capture all regions of high subterranean species richness; all of these hotspot caves are worthy of study and protection in their own right, and any protection strategy should include these exceptional sites as well as regional areas of species richness. In total, papers in this Special Issue cover 13 of 22 sites known to harbor at least 25 obligate subterranean species. Culver and Sket's original list of 20 has become too lengthy to be of much use; thus, we raised the bar to 25. One other cave is included in this Special Issue because of its special geographical position distant from any other hotspot site—Ganxiao Dong Cave in China [18]. However, we did not include it in our summary of hotspot caves.

#### **2. Overview of Hotspot Subterranean Sites**

In Table 1, we list the 22 known sites with at least 25 obligate subterranean species, and they are also shown on the map in Figure 1. Two of these sites (Ojo Guareña in Spain and Sistema Huautla in Mexico) have not been previously noted in any list as hotspot caves. There may be additional non-cave subterranean sites, particularly in hyporheic habitats, but these data have for the most part not been assembled into species lists for individual sites. In addition, there is the difficulty in deciding what constitutes a single site—is it just one Bou-Rouch pump site or a stream reach (e.g., a riffle)?

**Citation:** Culver, D.C.; Deharveng, L.; Pipan, T.; Bedos, A. An Overview of Subterranean Biodiversity Hotspots. *Diversity* **2021**, *13*, 487. https://doi.org/10.3390/d13100487

Received: 22 August 2021 Accepted: 23 August 2021 Published: 6 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** Map of hotspot sites listed in Table 1. Map prepared by Magda Aljanˇciˇc, used with permission.

**Table 1.** All sites with 25 or more stygobionts plus troglobionts. Caves assessed in this special issue are marked with an asterisk.


<sup>1</sup> Unpublished records of L. Deharveng show 30 stygobionts and 11 troglobionts from the Baget Basin, including Grotte de Sainte-Catherine. Two species are undescribed. <sup>2</sup> Numbers are for the Undara Basalt Flow as a whole. Bayliss Cave, the largest cave, has 23 troglobionts and 1 stygobiont. <sup>3</sup> Clark et al. [34] describe a nearby terrestrial area but it includes both deep soil and MSS (milieu souterrain superficiel) components, and is not directly comparable to the other sites.

> Several additional sites are worth mentioning because it is likely that if further collecting and identification is performed, they will have 25 stygobionts plus troglobionts. Trajano et al. [12] report a highly diverse troglobiotic fauna in a Brazilian karst region, with 29 troglobionts reported from 11 caves in a 25 km<sup>2</sup> region. This record is noteworthy both for its geographic position and the fact that the caves are in limestone, even though no one cave system has 25 troglobionts. No hotspot caves are reported from Africa, but

Deharveng and Bedos, in an ongoing study, have reported 30 troglobionts and 3 stygobionts known from an Algerian karst area—Djurdjura—in an area no more than 50 km in length. Especially noteworthy is that there are many species restricted to karstic holes with permanent snow above 2000 m, especially beetles and Collembola, which exhibit eye and pigment reduction, most living in caves as well. In Ganxiao Dong in China, a total of 19 troglobionts and 1 stygobiont have been reported [18], and it is likely that many more species will be found, especially because multi-taxa collection is in its infancy. The fauna is noteworthy because of the high level of morphological specialization, especially in Coleoptera and Myriapoda.

Outside of the Dinaric mountains, none of the sites were highly ranked (i.e., in the top ten) with respect to both their aquatic fauna and terrestrial fauna. Postojna Planina Cave system is by far the richest known subterranean site with respect to species richness; it ranks first with respect to both aquatic and terrestrial species richness [19]. Some sites had no terrestrial fauna (San Marcos well in Texas, Jameos del Agua in the Canary Islands, Walsingham Cave in Bermuda, Triadou Aquifer, and Robe Valley wells in Australia) and others had no aquatic fauna (Huautla in Mexico, Cueva de Felipe Reventón and Cueva del Viento-Sobrado in the Canary Islands). All continents are represented by at least one site, except for Africa.

When the aquatic and terrestrial fauna are considered separately, a different pattern emerges for the two habitats. For aquatic species, the top ten sites include six anchialine or deep phreatic sites, three sites in the Dinaric mountains, one site in the Cantabrian Mountains of Spain. The Dinaric and Cantabrian sites are within the ridge of high-diversity regions in Europe [4,6]. All of these sites are likely to be enhanced in available organic carbon and nutrients relative to other sites, but historical reasons in the case of the Dinaric sites may be important as well [4]. In particular, their proximity to the Mediterranean Sea, a source of colonists, especially during the Messinian salinity crisis, may be important.

For terrestrial species, the sites are more scattered outside the Dinaric mountains, which is the location of three of the five richest terrestrial caves (Table 1). The lava tubes on Tenerife in the Canary Islands are the site of two of the ten species-rich terrestrial caves, as are two caves in Mexico; the other top ten are scattered in Australia, Brazil, Sulawesi, and the United States. The mesas near the Robe Valley in Australia are very species rich, but the reported subterranean fauna includes the soil fauna [34], making comparisons impossible. It is difficult to determine what unites these terrestrial sites, because they are a combination of lava tubes and karst caves, with very different levels of organic carbon and nutrients. It is, however, noteworthy that tropical or subtropical sites are relatively more represented than for aquatic fauna.

The latitudinal distribution of hotspot sites shows an unexpectedly bimodal distribution (Figure 2). The majority of sites are in the temperate zone, between 40 and 50 degrees north or south of the equator. This corresponds, at least in the northern hemisphere, to the region traditionally held to be the area of highest subterranean species richness, a result often attributed to the effects of repeated Pleistocene glaciations [36]. The second mode is in the sub-tropical and sub-temperate regions (between 20 and 30 degrees north and south of the equator). These sites are mostly lava tubes and wells connected to chemoautotrophic zones (Table 1). None of the sites are located in the arid tropics, also suggesting that food availability may be an important driver of species richness.

**Figure 2.** Distribution distances from the equator of the subterranean hotspot sites listed in Table 1.

#### **3. Caveats and Challenges**

#### *3.1. Differences in Working Definitions of Troglobionts and Stygobionts*

Although there are many nuances [14,37–39], troglobionts and stygobionts are considered to be species exclusively found in subterranean habitats, which usually, but not always, show convergent morphological modification to subterranean life, often called troglomorphies [40]. Troglomorphies include, but are not limited to, reduced or absent eyes and pigment. For areas and groups with a well-studied surface fauna, it is possible to detect species occurring in subterranean habitats with well-developed eyes and pigment, but absent from the surface. In many, perhaps most cases, these are recent isolates in subterranean habitats. The difficulty is that in areas where the surface fauna is not well studied, such as most of the tropics, it is impossible to ascertain that such non-troglomorphic species are troglobiotic (or stygobiotic) species. Therefore, numbers of troglobiotic and stygobiotic species may be underestimated or overestimated in these sites. Such species may be common and the uncertainty of the ecological assignment (e.g., troglobiont or troglophile) is a bias. Deharveng and Bedos [41] report that soil-inhabiting species (especially Collembola and Diplura) are commonly listed as troglobionts.

Some studies of poorly known fauna may be biased by the inclusion of too many species as troglobionts and stygobionts if the operational criteria for inclusion are eyelessness and lack of pigment. Many litter- and soil-dwelling species are eyeless and depigmented, and some major taxonomic groups, e.g., Symphyla or the highly diversified Onychiuridae (Collembola) species, are entirely eyeless and depigmented regardless of habitat [29,42]. Additionally, some troglophilic species have populations that have reduced eyes and pigment. Sometimes these are recognized as subspecies [43], but if not, they are usually not included in the lists.

A strong case could be made that the relevant communities to compare are the permanent, sustaining populations (troglobionts [stygobionts] and eutroglophiles [eustygophiles]) but such lists are rarely available. Eutroglophiles and eustygophiles (see [38] for definitions) may be hard to delineate from other non-troglobionts and non-stygobionts.

#### *3.2. Differences in Taxonomic Groups, Habitats Sampled, and Sampling Techniques*

Differences in taxonomic composition can be because of regional differences in the composition of the subterranean fauna or because of incomplete sampling and/or taxonomic description. The tropics and arid areas tend to be richer in Arachnida, with some minor arachnid orders entirely missing at higher latitudes [14,41]. Other cases are clearly ones of inadequate collection. The waters of epikarst are usually dominated by Harpacticoidea crustaceans, but they have not been sampled or inadequately sampled outside of Europe (but see [44]), and the fauna can be quite rich, with more than ten species in some Slovenian caves [45,46]. Among the species listed in Table 1, only Postojna Planina Cave System [19] and Ojo Guareña [23] have been thoroughly sampled for epikarst fauna. Other groups may be under-represented due to their small size. Among Collembola, the Neelidae have several species less than 0.5 mm long [41,47] and are probably much more common than existing records would indicate. Ostracoda have rarely been sampled, with only the Postojna Planina Cave System [19], Ojo Guareña [23] and San Marcos well [28] among the sites in Table 1 having been surveyed. Finally, there are often groups that are under-collected because they require special preservation techniques or there are no taxonomists to study them (the Racovitzan and Linnean shortfalls [48]). Planarians are a good example. Although many records of planarians are known, many, if not most, have not been identified to species, and few subterranean species have been described recently.

#### *3.3. Described versus Undescribed Species*

A particularly vexing problem in the analysis of species numbers in subterranean sites is the nearly universal problem of undescribed species. Except for Cueva de Felipe Reventón, Cueva del Viento-Sobrado, Walsingham Caves, and Mammoth Cave, all the hotspot caves listed in Table 1 include undescribed species, as do Bayliss Cave, Sistema Huautla, Towakkalak system and Ganxiao Dong included in this Special Issue. The percentage of undescribed species tends to be highest in tropical countries, with a rich fauna and a shorter tradition of taxonomic description, such as Brazil, but it is also high in countries such as Australia (Table 1).

The difficulty is that not all records reported as new species turn out to be new species. The West Virginia cave fauna provides an example of the problem. In 1976, Holsinger, Baroody, and Culver [49], listed 27 undescribed stygobionts and troglobionts. In a reexamination of the West Virginia cave fauna in 2007, Fong et al. [50] found that of the 27 undescribed species listed in 1976:


This suggests a "discount rate" for undescribed species of 68% (13 of 19, because 6 were not new species after all), but this rate is likely to vary from region to region, and if it were applied to the data in Table 1, it would result in a bias against tropical regions and any other region with large numbers of undescribed species. Perhaps the best solution is to list both groups of species in any compilation of species richness but to indicate the number that remain undescribed, as we have done in Table 1.

Finally, there are questions about the quality of the taxonomic descriptions. This may be especially problematic for descriptions more than one century old, where the descriptions were very brief, and the range of diversity of different groups was largely unknown. This is, for example, a problem with mites described from Mammoth Cave in the 19th century, some of which have not been recollected [24].

#### **4. Protection of Hotspot Caves**

It is difficult to generalize, either about the threats or the level of protection for the hotspot caves listed in Table 1, at least beyond the existential threat that global warming poses for subterranean ecosystems [51,52]. In addition, a few generalities are possible. First, all sites are affected by surface processes. For caves, these inputs include infiltration through soil and epikarst, direct inputs from sinking streams, and direct human actions, including tourism.

Secondly, the area of the Earth's surface that impacts a subterranean fauna is greater for the aquatic fauna than for the terrestrial fauna. At one extreme is the artesian well at San Marcos, Texas, which accesses groundwater in the Edwards Aquifer at a depth of 59.5 m, and the Edwards Aquifer itself, which covers an area of over 10,000 km<sup>2</sup> [28]: water quality and quantity in the San Marcos well are potentially affected by events occurring through the Edwards Aquifer. At another extreme are the lava tubes in the Canary Islands. Most, if not all, of the organic input comes from tree roots as well as organic matter deposited from small cracks and crevices in the lava tube. In this case, the vulnerable zone is the projection of the cave onto the surface. In many cases, the limit of the vulnerable zone is unknown.

Thirdly, cave tourism is an important aspect of the state of affairs for a number of hotspot caves. Two caves in Table 1 have in excess of 100,000 visitors per year (Postojna Planina Cave System and Mammoth Cave). Others have smaller numbers of visitors, including wild cave tours (Walsingham Caves, Križna jama, Jameos del Agua, and Cueva del Viento Sobrado). Although cave tourism is often thought of in negative terms because of the attendant problems of light pollution, increases in CO2, structural modifications of the cave, etc., it can have a positive impact in that the economic value of a tourist cave makes major disruption unlikely. Additionally, the tourist part of the cave is typically a small fraction of the entire cave.

Furthermore, many of the hotspot caves are protected by some form of designation as park land, or protected area. Even if there is no official designation, there are often laws protecting caves in general [53,54]. However, the efficacy of these regulations and designations is never absolute, and at some level all sites face threats. For example, even Mammoth Cave, a U.S. National Park, faces a number of problems, ones due to actions taken outside the park, as well as past actions by the NPS itself [55]. On the obverse side, some caves are largely protected by their difficulty of access, by their length, as well as by the secrecy of some of their entrances, such as Sistema Huautla, one of the deepest caves (>1 km) in the world [56] or Coume Ouarnède [25].

Finally, there is not one template for threats or one prescription for the protection of subterranean hotspots. What works for a site in the United States probably will not work for a site in Brazil. Each site must be considered separately and carefully, and upon threat evaluation. Threat levels are the basic criteria to be used to derive protection measures. All the sites listed in Table 1 are worthy of protection as well as a source of regional and national heritage.

**Author Contributions:** Data compilation: D.C.C., L.D., T.P., and A.B.; Manuscript writing: D.C.C.; Editing: L.D., T.P., and A.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Diversity* Editorial Office E-mail: diversity@mdpi.com www.mdpi.com/journal/diversity

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com

ISBN 978-3-0365-2360-6