Springs can provide a perennial water-source for wetlands, and when they occur in arid landscapes represent tiny islands of habitat for aquatic species (Davis et al., 2017). It has already been recognised that spring wetlands can provide habitat for endemic species across the globe (Stevens and Meretsky, 2008; Rossini et al., 2018; Souza and Eguiarte, 2022). Endemism in this context refers to species that are only known from habitats associated with springs, and in many cases are only known from single locations. Wetlands with a source from artesian springs have greater biodiversity values than other wetlands in arid-lands. The diversity and habitat specialisation of plants, molluscs and fish in four permanent wetland types: riverine waterholes, rockholes, outcrop springs, and discharge (or artesian) springs from the arid-lands of Australia were compared (Fensham et al., 2011). Outcrop springs are fed by gravity and occur in unconfined geological substrata, generally close to where their water source is recharged. They contrast with artesian springs, which occur in confined aquifers and are fed by pressurised groundwater with sufficient residence time to be affected by the aquifer substrate and are often in areas remote from where recharge is occurring. The riverine waterholes, rockholes, outcrop springs had low levels of endemism compared to artesian springs, which were shown to have a high diversity of specialised endemics across all three taxonomic groups. This study identified three factors that may be important determinants of endemism in artesian spring wetlands: isolation, distinctive habitat, and long-term continuity of habitat.

Aridity (a surrogate of isolation) was a predictor of increasing endemic diversity in the artesian springs of the Great Artesian Basin (GAB) in Australia (Rossini et al., 2018). In contrast riverine waterholes have low concentrations of endemics because the perpetuation of unique genotypes is precluded by flooding disturbance and connection to more mesic environments during floods (Fensham et al., 2011; Stevens et al., 2020).

Specialisation to spring habitats relates to environmental constancy, for example, in water level, low turbidity, and geochemistry (Fensham et al., 2011). Habitat permanency is an essential pre-requisite for endemism in spring wetlands by providing sufficient time for occurrence and fixation of unique genotypes. Supporting this contention, permanent wetlands support more distinctive biota compared to ephemeral wetlands (Erman and Erman, 1995; Wood et al., 2005). The lack of endemism in the outcrop springs and rockholes of Australia may relate to the impermanence of their aquatic environments over evolutionary timescales (e.g., Ponder and Slatyer, 2007).

The occurrence of endemic species is evidence that some springs have fed permanent wetlands over very long timescales. They may function as paleo-refugia (Nekola, 1999), where lineages that evolved in ancestral mesic conditions survive despite aridification of environment. Species that have persisted in spring wetlands as relics of previous climate regimes have been referred to as “paleo-endemic” species (Stebbins and Major, 1965). Alternatively, organisms may have colonized springs recently from more mesic environments and then evolved in these habitats as “neo-endemic” species. Paleo-endemic species are relics of a formerly more widespread matrix community, while neo-endemic species occur in habitats that are younger than the matrix community (Nekola, 1999).

Around the periphery of spring wetlands where the water chemistry is affected by the aquifer substrate there is a distinctive ‘scalded’ habitat. These habitats are not wetlands despite being characterised by salts which are concentrated by the evaporation of groundwater. In Australia, the soils in these habitats typically have pH values greater than 9 supporting a low cover of succulent plants (Fensham et al., 2021). These habitats also support endemic species with very local distributions compared to other species specialised to non-alkaline saline environments.

Three complexes of spring wetlands on two continents have reasonably comprehensive biological surveys and reveal high concentrations of both aquatic and terrestrial endemic species: Ash Meadows, Nevada, United States (Stevens and Meretsky, 2008), Byarri (also known as Edgbaston Springs) in Queensland (Rossini et al., 2018), Australia, and Cuatro Ciénegas Basin, Coahuila, Mexico (Souza and Eguiarte, 2022). While the macro-biota of these three sites is relatively well known, not all species have been formally described and collation of comprehensive, comparative species lists has not previously been attempted.

Here we present lists of the endemic species occurring in the three spring complexes and use the data to 1) highlight their overall extraordinary unique biodiversity; 2) compare their endemic diversity across taxonomic groups; 3) compare common features of the three regions to further understand the critical features of arid-land springs related to endemic diversity; and 4) and establish that both ‘paleo-endemics’ and ‘neo endemics’ occur in these habitats. We also comment on the possibilities for new discoveries of unique life in these specialised habitats.

This compilation of macro-organisms from these three locations on two continents demonstrates that an extraordinary concentration of local (fine-scale) endemism occurs in habitats associated with arid-land springs. The largest of the three locations, Cuatro Ciénegas, is only 399 km² in area but provides habitat for 60 species known only from spring-associated habitats at that location. Ash Meadows and Byarri are even smaller in area (Table 1) and both have 24 species restricted to those locations. There are an additional five species at Ash Meadows and 12 species at Byarri that are only known from habitats associated with springs, but also occur further afield than those two individual locations (see Supplementary Material). Eukaryote endemism at these sites is associated with a broad range of phyla (plants, platyhelminths, annelids, molluscs, arthropods and chordates), but these levels of endemism may even be greater when comprehensive surveys are complete and taxonomy resolved. There is also likely a vast but largely undescribed array of microbial taxa and eDNA will be an important tool to reveal that diversity (Souza et al., 2006; Moreno-Letelier et al., 2012; Paulson and Martin, 2019). A species of cyanobacteria that forms macro-colonies at Byarri is the single described endemic prokaryote from any of three sites (McGregor and Sendall, 2017). Endemic species are not only associated with the aquatic environments of the spring wetlands but also with the scalded saline areas around the wetlands.

Many of these endemic species are related to taxa that are distributed regionally, particularly in the terrestrial environments (see Supplementary Material). However, five aquatic and semi-aquatic endemic species at Ash Meadows, five at Byarri and six at Cuatro Ciénegas are from genera without congeneric relatives in non-spring habitats in the major catchments where they currently occur. These genera have either evolved in springs as paleo-endemics, been isolated after the extinction of ancestors, or have radiated after unlikely long-distance dispersal across major catchment divides.

Spring-associated endemic aquatic plant species are diverse at Byarri but are relatively few in North America (Table 1). Endemic beetles and Hemiptera have only been described from Ash Meadows. Endemic snails, particularly those in the caenogastropod superfamily Truncatelloidea occur at all locations. Scaturiginichthys (Byarri) and Empetrichthyes (Ash Meadows) are fish genera entirely restricted to spring habitats. A lizard and three species of semi-aquatic turtle are confined to Cuatro Ciénegas and are the only other non-fish chordates at any of the locations. The last association is probably dependent on the especially large size of the wetlands at Cuatro Ciénegas, which have twice the discharge of Ash Meadows, and about 30 times greater discharge than Byarri.

The high diversity of endemic aquatic plant species at Byarri and their scarcity at the other two locations is difficult to explain. These plant species are perennial and form short, rhizomatous mat-like colonies, a growth form that occurs widely throughout the springs of arid and semi-arid Australia (Fensham et al., 2004). This growth form may be particularly well adapted to small wetlands with a constant flow, but like grasses are also adapted to grazing. Such islands of green forage in a barren landscape are occasionally subject to extreme grazing pressure from livestock and feral and native herbivores. Populations of endemic plants, such as Eriocaulon carsonii can be greatly diminished by intense grazing and pugging by feral pigs, only to recover when these pressures are reduced (R. Fensham pers. obs.). The springs-dependent endemic plants of Australia have been subject to intense herbivory pressure from marsupials including extinct Pleistocene megafauna (Johnson, 2006) over evolutionary time; thus, the perennial mat-forming growth form may partly be an adaptation to withstand herbivory.

Spring wetlands represent tiny aquatic islands surrounded by a vast “sea” of harsh desert habitat and are extremely isolated from other aquatic habitats. Thus, a species that has dispersed or become trapped in a spring-fed wetland has evolved into a new form under reduced competition with its ancestral relatives. For this allopatric evolution to occur, the spring habitats must be relatively persistent and isolated over at least recent evolutionary time. The magnitude of these timescales is partly informed by the age of spring wetland complexes. For example, the spring deposits at Ash Meadows lie in the Great Basin, which was subjected to rain shadow aridification beginning in Miocene time. Ash Meadows contains spring mound deposits dating to 3.2 my ago. Surface water connections existed with Owens Valley and through the Pleistocene advances and retreats of Lakes Tecopa and Manly in the lower Amargosa River drainage (Phillips, 2008). Thus, while largely surrounded by Paleozoic limestones and dolomites (Denny and Drewes, 1965), its Neogene sedimentological history indicates that groundwater expression has been commonplace in Ash Meadow for several million years. Also, periodic but intensifying aridification, has reduced spring density and discharge over time (Denny and Drewes, 1965; Hay et al., 1986; Menges, 2008).

In Australia, the aridification of the interior of the continent commenced about 4 Ma, and formation of the major dune fields, that are still in place today, commenced about 1 Ma (Fujioka et al., 2009). This time-scale coincides with the evolutionary lineages that represent much of the contemporary arid-zone flora (Byrne et al., 2008). The hydrological gradient of the Great Artesian Basin is the result of large synclinal structure uplifted and exposed from the development of the Great Dividing Range in the vicinity of its eastern margin (Habermehl, 2020). The timing of this orogeny remains controversial, but Ollier (1982) opined that it mostly occurred between 90 and 30 Ma. The suggested age of the springs towards the western margin in the Lake Eyre region is late Tertiary (Jessup and Norris, 1971), and a <2 Ma time frame is consistent with the uranium series disequilibrium dating of travertine deposits that are <1 my old (Priestley et al., 2018).

Deep phylogenetic divergence of species lineages of snails (Ponder et al., 1995; Perez et al., 2005; Murphy et al., 2012; Ponder et al., 2019) and isopods (Guzik et al., 2012) in isolated springs complexes of the Great Artesian Basin in Australia suggest they are derived from ancestors much older than 2 Ma. At Cuatro Ciénegas sub-populations of the endemic snail Mexipyrgus churinceanus occupy isolated drainage areas within the springs complex and appear to represent distinctive lineages that shared a common ancestor 2.5 Ma (Johnson, 2005). These timescales support the understanding that at least some of the species represented in desert springs are paleo-endemics representing relictual lineages from previous, more mesic environments.

There is also clear evidence that some of the endemics in spring-fed wetland habitats are neo-endemics. The endangered devils hole pupfish (Cyprinodon diabolis) is characterised by its small population size and its lack of pelvic fins. It has a population less than 1000 individuals and breeds exclusively on a shallow 3.5 by 5 m ledge near the surface of a subaqueous cavern, which is its sole habitat. Long thought to be a Pleistocene relic (Echelle, 2008), recent genetic studies (subject to considerable debate) indicate that the species diverged from its sister taxon Cyprinodon nevadenesis less than 10,000 ya (Sağlam et al., 2017; Sağlam et al., 2018) and possibly less than 1,000 ya (Martin et al., 2016; Martin et al., 2017). At Ash Meadows, lineages of hydrobiid snails in the genus Pyrgulopis exhibit both deep structure, suggesting divergence during the Pliocene, and also shallow structuring indicating divergence younger than 0.7 Ma (Hershler and Liu, 2008). The latter in some cases appear to have involved cross-catchment dispersal.

To what extent is the endemism of the springs associated with their unusual geochemistry? The terrestrial endemic plant species are associated with unusual alkaline soda rich (NaHCO₃) or neutral gypsum-rich (CaSO₄) saline environments. These environments are a product of the evaporites of the groundwater and the endemic halophytes have a specific adaptation to these specialised environments (Fensham et al., 2021).

Aquatic conditions in the springs themselves are generally more benign than the surrounding ‘scalds’, with relatively neutral or slightly alkaline pH at all three sites (Table 1). A simple experiment where the spring endemic plant Eriocaulon carsonii was grown under a range of pH conditions indicated that the species is equally able to tolerate alkaline (pH 7.0-7.7) and neutral (pH 6.4-7.0) conditions but is intolerant of acid conditions (pH of 4.1–4.4). Experiments on endemic snails indicate conditions become lethal when salinity is more than 15,000 μS/cm and temperatures <5°C and >40°C, conditions corresponding to the variation within the aquatic environments of the springs but exceeded in the shoreline conditions where evaporation concentrates salt (Ponder et al., 1989; Rossini, et al., 2017b). There is no evidence from these experiments that the snails are impeded in ‘soft’ waters with low salinity and neutral pH. Breeding populations of the endemic fish Scaturiginichthys vermeilipinnis can be maintained in water from Brisbane main water supply (P. Johnson pers. comm.). Disparities between physical and microbial environments of natural and artificial habitats may account for the limited recruitment of populations of the devils hole pupfish in the artificial habitat (Sackett et al., 2018). The extent that the geochemical environment of springs is a critical determinant of the habitat of endemic species requires further study.

The species list generated here includes taxa that are not yet formally described and excludes species at the sub-species level. A comprehensive taxonomy of the species in spring wetlands supported by molecular data will help to resolve the true levels of endemism in spring habitats and may identify cryptic species that are difficult to distinguish using morphology alone but may be only distant relatives due to the long periods of isolation in separate spring complexes. Such potential for cryptic speciation already has been recognised among amphipods that occupy individual artesian springs in South Australia (Guzik et al., 2012) and the American Southwest (Adams et al., 2018). Populations of a grass species, Cenchrus purpurascens, in isolated spring wetlands are morphologically indistinguishable from other populations across Asia and Australia (where it has been considered an exotic weed) but have been shown to have diverged in the order of 10⁵–10⁶ ya (Toon et al., 2018). In other cases, as for the devils hole pupfish, further study will reveal when species have morphologically diverged.

The concept of species endemism only has meaning in reference to a defined area or habitat (Shipley and McGuire, 2022). Endemism can be associated with specialised habitats at local scales, for example, cloud-forests, distinctive geological formations, coral reefs or wetlands. In this context and in lieu of comparative studies, the spring ecosystems at Ash Meadows, Byarri and Cuatro Ciénegas have some of the highest documented concentrations of endemic species in the world and highlights the critical importance of desert springs as specialised habitats. There is evidence from genetic analyses of these organisms that some pre-date Tertiary aridification and subsequent isolation in the springs, while others result from much more recent dispersal and rapid divergence in these unusual but constant habitats. The biota of spring habitats consists of both paleo-endemic taxa that evolved before 10⁶ ya, as well as neo-endemic taxa that are likely much younger. There are many springs that do not have high concentrations of endemics and the environment, distribution and history of these springs will undoubtedly help explain the situations where the endemics occur. Our understanding of the biology of these three locations on two continents may be at least partly a product of the extent to which they have been biologically surveyed. Desert spring complexes around the world await investigation.