Abstract

Subterranean biodiversity continues to be poorly known as a result of uncertainties, challenges and hazards associated with sampling in microhabitats such as aquifers and caves. Focusing on the narrow, lateritic aquifers and associated groundwater habitats in the Western Ghats freshwater ecoregion (southern peninsular India), we investigate the genetic diversity of an enigmatic group of eel-like fishes (family Synbranchidae). A maximum-likelihood phylogenetic analysis based on mitochondrial cytochrome oxidase subunit 1 (cox1) gene sequences recovered these synbranchid eels into two distinct clades comprising genera Ophichthys (fossorial eels ‘with eyes’) and Rakthamichthys (aquifer-dwelling ‘blind’ eels). Additionally, three species-delimitation approaches (based on the mitochondrial cox1 gene), revealed the presence of 11 Evolutionarily Distinct Lineages (EDLs) within Rakthamichthys separated by an inter-lineage divergence between 5.8–20.3%, and an intra-lineage divergence between 0–4.5%. Rakthamichthys in southern peninsular India exhibited a distribution pattern comprising both restricted-range and wide-ranging lineages. Fossorial eels of the genus Ophichthys, on the other hand, are widely distributed in southern peninsular India, with clear geographical boundaries separating the two known species. The genetic network of Rakthamichthys and Ophichthys revealed multiple haplotypes within various EDLs, with a large number of mutations separating the haplotypes within, and between species and/or lineages. Though represented by high levels of genetic divergence revealing the potential existence of at least 11 EDLs, their remarkable morphological uniformity combined with a complex distribution pattern makes it difficult to assign known species names to various Rakthamichthys lineages. Most subterranean habitats in southern peninsular India are under severe anthropogenic threats. Therefore, resolving the taxonomy of, and developing conservation actions for groundwater-dependent species is a priority, for which we suggest future steps.

Groundwater, endemics, laterite, subterranean, Western Ghats

Subterranean and groundwater habitats are known to harbour some of the planet’s most unusual fish assemblages (Niemiller et al. 2019), but very little fundamental information has been generated on their taxonomy, life history and ecology. For example, of around 322 species of cave and groundwater fishes, and 55 species of interstitial fishes (those living substantially, or obligately, in river bed sediments or marine sands; Proudlove 2025), more than 50% are known only from their type material. A major reason for this knowledge gap – the Racovitzan impediment (Ficetola et al. 2019), is because the habitats occupied by these species, e.g., caves, aquifers, wells, karsts and interstitial waters (Niemiller et al. 2019; Proudlove 2025) are often unexplored and unmapped, or are otherwise too small, challenging, and often dangerous to be accessed (Ficetola et al. 2019).

Challenges in understanding and protecting the unknown, necessitate an accelerated approach to scientific explorations and research, as most groundwater and subterranean species have not yet been described, and many could go extinct before they are officially discovered and named (Mammola et al. 2019). The lack of information on subterranean diversity and distribution (Zagmajster et al. 2018) is also complicated by the high prevalence of endemic, and mostly cryptic species (Niemiller et al. 2013; Devitt et al. 2019), significant levels of intra-specific morphological variation, as well as differences in diversification patterns across taxa and lineages (Juan et al. 2010). Added to this are the uncertainties and challenges associated with groundwater fish sampling, including the lack of standardised survey techniques (Gibson et al. 2019).

The family Synbranchidae comprises a group of elongate, eel-shaped, amphibious, air-breathing acanthomorph fishes commonly known as swamp eels or mud eels, with a native range extending from Central and South America to Western Africa, South and South East Asia, and the Indo-Australian archipelago (Nelson et al. 2016). As a result of their distribution on all southern continents except Antarctica, synbranchid eels have been considered by Harrington et al. (2024) as the most remarkable example of dispersal among living lineages of air-breathing fishes. Synbranchid eels primarily occur in freshwaters, where many species exhibit a fossorial life, others inhabit caves, and some reside in groundwater aquifers (Moore et al. 2018; Britz et al. 2021). Having lost all fins (except caudal fin in Macrotrema; Rosen and Greenwood 1976) and scales (except species of Ophichthys – see Britz et al. 2020), adult synbranchid eels are unique among bony fishes. Seven genera and 28 species are currently considered valid within Synbranchidae (Fricke et al. 2025) of which two genera, Ophichthys and Rakthamichthys are endemic to South and Southeast Asia (Britz et al. 2020, 2021). The genus Ophichthys includes at least five, primarily fossorial species, distributed throughout the Indian subcontinent and Myanmar (Dahanukar 2010; Britz et al. 2020), while Rakthamichthys includes five, greatly elongated, blind, and mostly aquifer-dwelling species, endemic to the southern peninsular India, and north-eastern India (Britz et al. 2021).

The Western Ghats freshwater ecoregion in southern peninsular India is considered a global hotspot of groundwater and subterranean fish diversity and endemism (Raghavan et al. 2021). A fascinating element of the region’s groundwater fish assemblage are the fossorial and the aquifer-dwelling synbranchid eels, comprising six species all of which are endemic – Ophichthys fossorius, O. indicus, Rakthamichthys digressus, R. indicus, R. mumba and R. roseni (Britz et al. 2020). All six species are poorly known and incompletely documented, most of them known only from their original description, and as a result, assessed as ‘Data-Deficient’ on the IUCN Red List of Threatened Species. These synbranchid eels are mostly encountered unintentionally or accidentally, either from inaccessible groundwater systems beneath paddy fields, narrow aquifers connected to homesteads, or dug-out wells (Anoop et al. 2019; Raghavan et al. 2021).

Based on samples obtained through multiple approaches including citizen-science facilitated surveys, opportunistic collections, and a series of targeted field collections in known subterranean fish habitats in the Western Ghats freshwater ecoregion, between the years 2013 and 2024, we present the results of the first systematic study on the diversity, distribution and genetic structure of these fossorial, and aquifer-dwelling synbranchids. Genetic analysis using the mitochondrial cytochrome oxidase 1 (cox1) demonstrates the monophyly of the blind, aquifer-dwelling Rakthamichthys, and the fossorial Ophichthys, while multiple species-delimitation methods reveal a remarkable case of ‘cryptic’ speciation, and unusual patterns of distribution in Rakthamichthys.

The unique assemblage of several fossorial and aquifer-dwelling synbranchid species in the Western Ghats ecoregion, and particularly the southern Indian state of Kerala, makes this region critically important for understanding the evolution and diversification of these enigmatic taxa. Currently, only four ‘named’ species of Rakthamichthys (R. digressus, R. indicus, R. mumba and R. roseni) are known from the Western Ghats ecoregion, which are ‘indistinguishable’ in terms of their vertebral counts, based on our results. Rakthamichthys digressus described from Kuthiravattom (approx. 11.25°N) was reported to have 166–170 vertebrae (Gopi 2002), R. indicus described from Kottayam (approx. 9.6°N) 159 vertebrae (Eapen 1963), and R. roseni described from near Thrissur (approx. 10°N) 147 vertebrae (Bailey and Gans 1998). These counts based on only few individuals overlap with those of specimens of the widely distributed lineage – EDL J/Rakthamichthys sp. J (134–173). Even if we exclude the single specimen of EDL J with only 135 vertebra, the remaining range 152–173, still overlaps with at least the three EDLs D, H, F. Given that the putative type localities of R. digressus, R. indicus and R. roseni overlap with different EDLs (see Table 1), it is impossible to assign any of these three names to any of our EDLs. Such a large variation of 21 vertebrae in ‘EDL J’ is uncommon in bony fishes, but ranges of up to 15 vertebrae are known from anguilliform eels with similarly high numbers of total vertebrae (Böhlke 1989).

The fourth species, R. mumba was described recently based on the perceived differences in a combination of characters with other species of Rakthamichthys, one of which was the vertebral count (Praveenraj et al. 2021). However, our observations on the variation in counts (Figs 1 and 5) of the different lineages now suggests that vertebral counts alone cannot be used to delineate R. mumba, as previously thought (see Praveenraj et al. 2021). Nevertheless, we recognize R. mumba in our phylogenetic tree, because its cox1 sequence information is available from its type material.

Together with the previous observations on the difficulties in morphologically distinguishing species of Rakthamichthys (see Britz et al. 2021), our study reinforces the issue of exceptional morphological homogeneity in this group. Such remarkable morphological stasis has recently been observed in another group of aquifer-dwelling fishes from the Western Ghats ecoregion – the blind catfishes of the genus Horaglanis (Raghavan et al. 2023), which incidentally co-occur with Rakthamichthys in many localities.

The lack of standard morphological characters that define species boundaries in ‘cryptic’ groundwater fauna can be attributed to the extreme conditions in the subsurface environments creating limited adaptive morphospace (Lefébure et al. 2006), or the lack of substantial environmental changes over time which may trap organisms in a local optimum morphospace (Sansalone et al. 2022). The groundwaters of the Western Ghats ecoregion, especially the narrow lateritic aquifer systems are likely to have provided a long-term stable environment for its faunal assemblages, thereby helping them retain their external morphology despite significant levels of genetic diversification – a reason that has been attributed for the extreme morphological uniformity in Horaglanis (Raghavan et al. 2023), and a likely scenario in the case of Rakthamichthys as well.

Though represented by high levels of genetic divergence revealing the potential existence of at least 11 EDLs, their remarkable morphological uniformity combined with a complex distribution pattern makes it difficult to assign names to the various lineages of Rakthamichthys. For example, two distinct EDLs (I and J) occur at distances of 12–25 km away from the type locality of R. digressus (approx. 11.25°N), and at least four distinct EDLs (C, F, G and J) occur in and around (between 45–70 km) the type locality of R. roseni. Interestingly, specimens that comprise the widely-distributed ‘EDL J’ co-occur in and around the likely distribution range of both R. digressus and R. roseni, and exhibit a range of vertebral counts (134–173) which overlaps those observed in the type series of these two species (Bailey and Gans 1998; Gopi 2002). Rakthamichthys indicus described from near the town of Kottayam (approx. 9.6°N; Eapen 1963) is probably the only species (apart from R. mumba) that can be identified based on its distinct distribution range (EDL D), and comparable vertebral count (154 in our specimen vs. 159 in the holotype). But for want of a specimen from the type locality that exactly matches the vertebral count observed in the holotype, we refrain from identifying ‘EDL D’ as R. indicus.

It is generally considered that groundwater species cannot disperse over long distances, and therefore most have a restricted range (Zakšek et al. 2009). Exceptions include some stygobitic fishes such as Prietella phreatophila – a subterranean catfish whose northern and southernmost populations in Mexico are separated by ca. 750 km (Hendrickson et al. 2001), and Ophisternon candidum the blind, subterranean cave eel from north-western Australia, the different populations of which are separated by >400 km (Moore et al. 2018). Such large distribution ranges for groundwater fauna are attributed to an extensive, continuous, hypogean habitat through which the species can disperse – known as the ‘interstitial highway’ hypothesis (Ward and Palmer 1994). The rather large distribution range (a north-south distance of ca. 450 km) for ‘EDL J/Rakthamichthys sp. J’ can be supported by two different hypotheses, one of which is the ‘interstitial highway’ hypothesis (discussed above). The second, alternative, hypothesis is that the original distribution range of the last common ancestor of southern Indian Rakthamichthys was large, and included the total area of all present lineages. This large range was then fragmented through vicariance events into the smaller, less inclusive, areas of today’s lineages. Only the Rakthamichthys ‘EDL J’ lineage would have retained the original larger distributional range. Under the ‘interstitial highway’ hypothesis we would expect that the last common ancestor of the Western Ghats species of Rakthamichthys occupied a restricted area from which ‘EDL J’, would have expanded its range to its the large current distribution. Our current analysis is in favour of the ‘vicariance’ hypothesis with a large original distribution, still retained in ‘EDL J’ and a vicariant fragmentation of distribution for the other EDLs. However, phylogenetic and biogeographical hypothesis analyses using multiple molecular markers, and greater taxonomic sampling over wider geographical area are, required to further test these different scenarios.

The ability of groundwater species to actively move both within, and between adjacent habitats, including long-distance dispersal remains unclear (Jordan et al. 2020), with dispersal abilities known to be influenced by the aquifer geology as well as life history and ecology of the species (Trontelj et al. 2009). Genetic evidence indicates that hydrological barriers among aquifers can be labile facilitating transient movement of fauna between distant aquifers (Malard et al. 2023). It is also known that groundwater and subterranean species are frequently washed out of their habitats during earthquakes, floods and other extreme climatic events, and can survive for several days in surface-water systems (Trontelj et al. 2009), before recolonizing potential groundwater habitats. Rakthamichthys could likely possess biological traits to support an ‘amphibious’ mode of life by which they are able to migrate between the subterranean and surface-water habitats, including shallow wetlands and paddy fields (see discussion in Raghavan et al. 2022 on the subterranean Aenigmachanna gollum). In addition, the surface rivers and their hyporheic zone may also provide corridors between aquifers (Malard et al. 2023), especially during the heavy monsoonal rains in the Western Ghats ecoregion, with an individual of Rakthamichthys collected from under a rock at the edge of a stream.

An interesting result relating to the large distribution of ‘EDL J/Rakthamichthys sp. J’ is that its range includes groundwater and subterranean systems separated by an otherwise significant biogeographic barrier – the Palghat Gap (at 11°N). While this gap has been known to be a major barrier for a number of taxonomic groups, shaping their intraspecific genetic diversity as well as interspecific diversification (Anoop et al. 2018; Sidharthan et al. 2020; Biswas and Karanth 2021), little is known about how this geological barrier influences the distribution of groundwater and subterranean fauna. In addition to Rakthamichthys and Ophichthys, two other groups of groundwater and fossorial fishes of the Western Ghats region are distributed across the Palghat Gap – the eel loaches of the genus Pangio, and the fossorial dragon snakehead Aenigmachanna gollum. While Pangio bhujia occurs in lateritic aquifers located both north (11.3°N) and south (10.5°N) of the Palghat Gap, the monotypic Aenigmachanna gollum has also been recorded from either side of the gap in localities between 11.5°N to 9.3°N (Raghavan et al. 2022; Sundar et al. 2022). The purple frog Nasikabatrachus sahyadrensis is another example of a fossorial/subterranean species with populations on either side of the Palghat Gap (IUCN SSC Amphibian Specialist Group 2022). Combined, these (preliminary) pieces of evidence show that the Palghat Gap may only have little influence on the genetic diversity of fossorial and subterranean fishes and amphibians.

We observed large genetic divergence both ‘within’ and ‘between’ species/lineages of Rakthamichthys similar to the divergence in Horaglanis – a blind, aquifer-dwelling catfish of southern India (Raghavan et al. 2023). We observed maximum intra-specific genetic divergence of 4.9% in Rakthamichthys ‘EDL J’, with most interspecific genetic distances estimated to be above 4.5%. However, ‘EDL F’ and ‘EDL G’, which were delimited as distinct units in mPTP and bPTP, had minimum interspecific genetic divergence of only 2.9%, the reason why it was not delimited in ASAP, an approach that utilizes barcode gaps. These observed differences in genetic divergence also suggest that a ‘single genetic threshold’ is unsuitable not only for different taxa, but even lineages of the same taxa. Similar to the large intraspecific divergences in Rakthamichthys, we also observed similar divergence patterns in the fossorial swamp eel genus, Ophichthys. The maximum intraspecific genetic distance in O. indicus was 5.6% and in O. fossorius 3.3%. The two sister taxa, however, showed a minimum divergence of 10.0%. The large intraspecific distances observed in both Rakthamichthys and Ophichthys, suggest that a generally-held belief of using a fixed threshold for delimiting fish species is impractical and should be avoided (Srivathsan and Meier 2012; Collins and Cruickshank 2013).

An inherent limitation of our study is the dependence on the mitochondrial cox1 gene, for delimiting putative species/lineages; despite our best efforts, additional genes including the mitochondrial cytochrome b (cyt b), and nuclear recombination activating gene 1 (Rag1), could not be amplified. The fact that there could potentially be a discordance between mitochondrial and nuclear gene-based phylogenies, with various divergent mitochondrial lineages recovered in a single clade based on nuclear genes, is well-recognized (e.g., Thielsch et al. 2017). Deep mitochondrial divergences have been considered a special case of mito-nuclear discordance (Toews and Brelsford 2012), where the actual divergence between populations or species is substantially shallower than otherwise indicated by only mitochondrial genes (Zhang et al. 2019). Even a future scenario, in which genome-wide studies may indicate that deeply-divergent EDLs based on mitochondrial DNA cluster together as a single, widely-distributed taxonomic unit, will not alter our current findings significantly, as the various deeply-divergent EDLs of Rakthamichthys based on mitochondrial genes, cannot be distinguished from one another, either using external morphological characters or vertebral counts. The challenge that EDLs or species cannot be distinguished morphologically therefore still applies.

This paper is dedicated to the late Francy Kakkassery (St. Thomas College, Thrissur, India) for his interest in these enigmatic eels, and for inspiring and motivating many of the authors of this paper to document and undertake research on these poorly-known taxa. We are greatly indebted to the members of local communities throughout the distribution range of Rakthamichthys for their whole-hearted support and help. Many of our friends and colleagues provided valuable specimens or facilitated the collection of specimens – the late Francy Kakkassery, V.K. Anoop, Liju Thomas, V.V. Binoy, Sandeep Das, C.P. Shaji, Abhilash Ravimohanan Nair, Preetha Karnaver, Ryan Babu, Krishnaprasad P.H, Shreehari KM, Somanath R Poojary, Prathvik Poojary, Rushab, Pradvin Poojary, Ranjit Poojary, Prajwal Poojary, Keerthish Poojary, Smrithy Raj, Kiran Dev, and Pradeep Kumkar. We are grateful to K. Ranjeet and Suresh Kumar (Kerala University of Fisheries and Ocean Studies), Sanjay Molur and Latha Ravikumar (Zoo Outreach Organization), Eleanor Adamson (Fish Mongers Company), and Mike Baltzer and Michael Edmonstone (SHOAL Conservation) for their help and support with logistics, funding and project administration. Several snake rescuers and villagers in Kerala provided valuable information on the fossorial and aquifer-dwelling eels. ACP is particularly indebted to Mohan Joseph I.F.S, and the National Biodiversity Authority (NBA), Government of India for permits. Critical comments and suggestions from Kevin Conway, Hiranya Sudasinghe and an anonymous reviewer greatly helped improve an earlier version of this manuscript. Funding for this project came from the Directorate of Environment and Climate Change (DoECC), Government of Kerala, India; the Simon Birch Memorial Funds, Fishmongers Company, London, United Kingdom; Sächsisches Staatsministerium für Wissenschaft, Kultur und Tourismus (SMWK) through the TG-70 funding stream, Germany; Mohammed Bin Zayed Species Conservation Fund, UAE; and SHOAL Conservation, UK.