Garra cavernicola, Freyhof, 2025

Freyhof, Jörg, 2025, Garra cavernicola, a new species from the Hajar Mountains in Oman (Teleostei: Cyprinidae), with a short discussion on monophyly and the description of evolutionarily young species, Zootaxa 5601 (2), pp. 335-345 : 336-343

Garra cavernicola , new species

( Fig. 1–4 )

Holotype. NMW 100394 (443), 39 mm SL; Oman: Al-Hoota cave , main lake , 23,0844, 57.3571.

Paratypes. NMW 100394 , 3, 41–43 mm SL; NMW 100406 , 2, 32–37 mm SL; same data as holotype. NMW 100321 , 3, 39–42 mm SL; NMW 100408 , 2, 47–58 mm SL; Oman: Al-Hoota cave , first lake , 23,0811, 57.3506 .

Diagnosis. Garra cavernicola is distinguished from other Garra occurring in the Hajar Mountains in Oman and the United Arab Emirates ( G. barreimiae , G. gallagheri , G. longipinnis , G. shamal , G. sharq ) by lacking an external eye (vs. present) and having a whitish or pink-coloured body without any pattern (vs. brown or grey, with a dark-brown or dark-grey mottled pattern). It is further distinguished from G. longipinnis by having smaller scales, and the absence of tubercles on the posterior portion of the scales ( Esmaeili et al. 2025).

Description. Morphometric data in Table 1 . Body stout, moderately compressed laterally, more compressed along caudal peduncle. Dorsal head profile rising gently, straight or convex, interrupted by shallow hump from dorsal body profile in some individuals. Profile of back rising to about halfway between nape and dorsal-fin origin, gently convex until dorsal-fin origin. Ventral profile more or less straight to anal-fin origin. Head short and conical, with slightly convex interorbital distance. Snout blunt. No tubercles on head. Transverse lobe absent or very shallow, poorly demarcated posteriorly by transverse groove. Proboscis absent or very shallow. Eye externally absent, eye socket filled with fatty tissue. Barbels in two pairs; rostral barbel located antero-laterally, reaching to vertical of nares, or to base of maxillary barbel; maxillary barbel at corner of mouth, reaching vertical of eye socket. Rostral cap well-developed, with few grooves, papillate on ventral surface. Upper lip present as thin band without papillae. Upper jaw completely covered by rostral cap. Gular disc elliptical, longer than wide or as wide as long; narrower than head width through base of maxillary barbel; groove between antero-median fold and central callous-pad narrow and deep, papillae on latero-posterior flap small; anterior marginal surface of central callous pad with small papillae.

Caudal fin with 9+8 branched rays; dorsal fin with 3 simple and 7½ branched rays, last simple ray slightly shorter than head length; distal margin straight or slightly convex; origin closer to caudal-fin base than to snout tip; inserted anterior to vertical through pelvic-fin origin; first branched ray longest, tip of last branched ray reaching vertical in front of anus. Pectoral fin with 1–2 simple and 13–15 branched rays, shorter than head length. Pelvic fin with 1 simple and 7–8 branched rays, not reaching to, or slightly beyond anus, not reaching anal-fin base, origin closer to anal-fin origin than to pectoral-fin origin, inserted below third or fourth branched dorsal-fin ray. Anal fin short, with 3 simple and 5½ branched rays; first branched ray longest; distal margin concave; origin closer to caudal-fin base than to pelvic-fin origin. Caudal fin forked; tip of lobes pointed; upper lobe longer than lower.

Lateral line complete, or often incomplete on posterior part of flank, with 30–34 scales on body and 1–2 scales on caudal fin. Transverse scale rows above lateral line 4½; between lateral line and pelvic-fin origin 3½ and between lateral line and anal-fin origin 3½. Circumpeduncular scale rows 12, rarely 13 or 14. Predorsal scales deeply embedded, irregularly arranged, smaller flank scales. Chest without scales. One short axillary scale at pelvic-fin base, and 7–8, usually 7, scales between posteriormost pelvic-fin base and anus.

Colouration. In life, body pink or whitish, fins whitish or usually transparent at distal and median parts; yellowish white in preservative.

Distribution. Garra cavernicola is endemic to Oman, and is known only from the Al Hoota Cave and nearby Hoti Pit ( Fig. 5 ).

Etymology. The species name combines the Latin “caverna”, an underground chamber, and “cola”, meaning ‘inhabitant of’. A noun in apposition, indeclinable.

Discussion

This manuscript was first developed in 2023. Subsequently, Turner (2024) published a conceptually similar, yet more extensive paper on the topic, using G. cavernicola (as G. longipinnis ) as an example. The ideas advanced by Turner (2024) strongly align with those discussed previously by Kottelat (1997), and further supported in the current paper. Despite this, the arguments made below are presented in their original form, as written before Turner (2024) was published, to illustrate how convergent these independent lines of thought are.

Several fish species are believed to have long evolutionary histories in subterranean habitats as evidenced by their strong phylogenetic distinctness ( Freyhof et al. 2016, Hashemzadeh Segherloo et al. 2016, Armbruster et al. 2016, Nguyen et al. 2021, Mao et al. 2021, Freyhof et al. 2022a, Raghavan et al. 2023). However, adaptation to subterranean environments can also occur relatively quickly as suggested by instances in which ancestors of some currently troglomorphic populations have invaded subterranean waters in geologically recent times. Close relationships between hypogean and epigean fishes imply that some cave fish populations may have originated following the Last Glacial Maximum (<10.000 years), as proposed for Garra ( Kruckenhauser et al. 2011) , Astyanax ( Keene et al. 2015) and Barbatula ( Behrmann-Godel et al. 2017) .

Among cave-adapted fishes, the genus Astyanax (specifically the blind, cave tetras) offers the most striking and best-studied examples. Historically, all subterranean populations of Astyanax were lumped into a separate genus Anoptichthys comprising a single species, A. jordani ; while the epigean populations were assigned to the genus Astyanax . Romero & Paulson (2001) placed Anoptichthys in Astyanax and synonymised A. jordani with the epigean A. mexicanus . Strecker et al. (2004, 2012) suggested a more complicated evolutionary history, arguing that multiple independent invasions of cave systems by epigean Astyanax have taken place leading to several distinct, paraphyletic, subterranean populations. They found no significant gene flow between the cave populations and the surface fish, even when these co-occur syntopically. Elliott (2018) confirmed the findings of Strecker et al. (2004), and reported 31 caves known to be inhabited by the troglomorphic Astyanax , some related to epigean A. mexicanus , others to epigean A. aeneus .

Interestingly, there were no attempts to revise the epigean-subterranean Astyanax complex from a taxonomic perspective, despite Strecker et al. (2012) demonstrating reproductive isolation in sympatry between different populations of this complex. There seems to be an implementation gap, and more work needs to be undertaken. Kullander (1999: 340) stated, “If the diagnostic characters are stable, and if both the epigean and the cave populations are self-perpetuating, then they are species distinct from each other, and all other species.”. It is clear from Strecker et al. (2004) that the subterranean populations are diagnosable and reproductively isolated. Still, their recognition as distinct species would make the ancestral epigean species ( A. mexicanus ) paraphyletic, and thus violate the criterion of monophyly under the ‘Phylogenetic Species Concept’ (PSC) (see Wheeler & Meier 2000 for different versions of the PSC). This might be why this issue has not yet been resolved, and the reason why most recent authors avoid using scientific names, and instead use the term “ Astyanax mexicanus complex” ( Keene et al. 2015).

A similar, though simpler situation has emerged in the case of Omani Garra longipinnis . Kruckenhauser et al. (2011) demonstrated that the subterranean population of G. longipinnis is not sister to all other populations, but rather nested within them, and showing close relation to the surface populations in the same region. Kirchner et al. (2017) investigated the subterranean Garra in Al Hoota Cave and nearby Hoti Pit, showing that while the cave population is isolated from the surface populations, individuals may occasionally be washed out into the surface habitats. These washed-out cave fish hybridise with epigean G. longipinnis , resulting in individuals with intermediate phenotypes, yet still being different from purely epigean fish. This clearly rejects the hypothesis by Banister (1984), that the cave population is just a phenotypic variation, and by this not based on own genetic mutation. However, Kirchner et al. (2017) observed the gene flow to be effectively unidirectional; the subterranean habitat remains off-limits to surface fish, preserving the genetic integrity of the cave population. Similar to the case of Astyanax mentioned above, recognising the subterranean population as distinct would make G. longipinnis —the ancestral epigean species, paraphyletic.

Kirchner et al. (2020) lumped the two units that show distinct evolutionary trajectories into a single species as they do not differ based on the molecular markers applied. By this, they avoid the paraphyly of the parental species following PSC principles, where monophyly of all species is an essential criterion ( Donoghue 1985, Cracraft 1989). Indeed, survival of the ancestral species and peripheral speciation is a long-debated issue (see Wiley, 1978, Kullander 1999) that has lost much of its attraction today. In many contemporary taxonomic practices, species are increasingly recognised only from molecular phylogenetic trees, and the monophyly of species has become a “dogma”, partly because phylogenetic trees offer a visually compelling demonstration whenever a species violates monophyly. Consequently, the ESC has often been sidelined, despite repeated cautionary advice (e.g., Mayden 1999). In doing so, ichthyology ignores the ESC as the primary species concept and, thus ignores the full diversity of species resulting from a diverse array of natural speciation processes. Instead, we stick to a single secondary species concept (the PSC), excluding considerable parts of species diversity.As the adage goes “a picture is worth a thousand words”, it is understandable that we no longer read long and complicated texts (such as this one), instead only look at phylogenetic trees. Therefore, a tree violating the monophyly of a species is immediately obvious, and evolutionary species are rejected as “they do not form distinct clusters in the tree, but are nested within another species”. As phylogenetic distances shown in trees are largely a visualisation of the numbers of generations passed since the isolation of populations, all species concepts have been reduced to phylogenetic time. Mayden (1999) advocated that none of the secondary species concepts (such as the PSC) should be used to describe species, highlighting the ESC as the only appropriate primary concept being able to span the “discovery net” for biodiversity. However, additional discussions, e.g., De Queiroz (2007), have presented versions of the PSC inclusive of diagnostic features taken as among the forms of evidence for use in species descriptions, and that molecular phylogenetic distance might be indicative for different evolutionary trajectories of populations.

By taking the PSC as the primary concept, peripheral speciation, in combination with the survival of ancestral species, became something like a difficult proposition, as 1) species must be monophyletic, and 2) the ancestral species “must go extinct when the species separates into two”. This proposal comes from phylogenetic trees based on cladistic analysis with only limited connection to the diversity of the speciation processes found in nature. From an evolutionary standpoint, there is no conceptual barrier preventing the ancestral species from surviving, after one of its peripheral populations splits off to become a new species. The theoretical necessity of postulating the extinction of ancestral species in phylogeny reconstruction, as advocated by Hennig (1966), is theoretically sound. However, Wiley (1978) already explained that an implication of the ESC would seem to be that if the ancestral lineage can lose one or more constituent populations without losing its historical identity or tendencies, it could survive such a split. De Queiroz (2007:883) opined that the survival of the ancestral species “is not an operational criterion for deciding when a lineage is sufficiently divergent to be considered an (own) species”. By this, it is clear that the parental species might survive peripheral speciation event(s), resulting in a lack of autapomorphies in the ancestral species.

In the case of G. cavernicola, Kirchner et al. (2017) detected only a fraction of the genetic diversity present in G. longipinnis , suggesting that the founding population of the cave colonisers was small. While it is currently unknown how much genetic variation the cave lineage “took with it,” the key point is that G. cavernicola has its own evolutionary trajectory. Indeed, Mayden (1997), Kottelat (1997), and Kullander (1999) all note that whenever a peripheral population successfully becomes isolated and begins adapting to new conditions, the original species may enter a temporary period of paraphyly. Over time, gene flow across populations of the ancestral species can restore monophyly, explaining why we see relatively few examples of paraphyletic ancestral species in nature (this might be a circular conclusion as we define species by monophyly). Indeed, it is logical that peripheral speciation should happen frequently, especially in freshwater fish, which face several geographic barriers, and this might be the most common mode of allopatric speciation.

Wiley (1978) has already discussed whether „ecotypes“ such as those in subterranean fishes may represent variants of the ancestral species. Yet, they represent independent lineages, with own adaptive mutations, even if recently diverged, and thus meet the ‘Evolutionary Species Concept’. The ESC demand distinct evolutionary trajectories ( Kottelat 1997, Mayden 1997 ¸ de Queiroz 2007), and the evolutionary trajectory of a subterranean fish population is very clearly different from that of an epigean population. Subterranean environments are very different from surface waters, and as subterranean fishes genetically adapt to these environments, they follow their own evolutionary trajectory. As soon as these new adaptations, such as the reduction of the eye and pigmentation, become observable, they are a diagnosable unit fulfilling the criteria to be distinct species under the ESC. This is why species concepts based on distinct evolutionary trajectories easily allow young but morphologically, physiologically, ecologically, or ethologically divergent species (such as cave fishes) to be recognised as distinct. At the same time, it is not easy to recognise allopatric phylogenetic lineages without demonstrating morphological, physiological, ecological, or ethological differences as distinct species (see Freyhof et al. 2022b for discussion).

Mayden & Wood (1995), Mayden (1997, 1999), and Kottelat (1997) consistently advocated that ESC, as the most flexible of the different species concepts, should serve as the primary conceptual framework in biodiversity studies. Following Mayden (1999), we have no conflict between the Evolutionary, Phylogenetic and other species concepts. The PSC represent a useful secondary species concept, subordinate to the broader ESC. As with other secondary species concepts (e.g., ‘Biological Species Concept’), the ‘Phylogenetic Species Concept’ (PSC) allows to recognise (only) a part of the species diversity, while only the ‘Evolutionary Species Concept’ (ESC) allows to span the whole “discovery net” for biodiversity. While the PSC may underestimate biodiversity by excluding lineages that are geologically young or lack complete monophyly, the ESC readily captures newly formed lineages that exhibit distinct trajectories, as in the case of G. cavernicola . By strictly advocating for the monophyly of all species, the PSC would indeed allow recognising G. cavernicola as distinct species. However, this can leave G. longipinnis paraphyletic, and a common, though often overstated, claim is that taxonomists would then need to split G. longipinnis into multiple “mini-species” until all lineages form strictly monophyletic clusters. In reality, such extreme splitting is rarely necessary. A single or minimal taxonomic revision often suffices to resolve paraphyly, and the PSC does not inherently demand proliferating multiple superficially diagnosed species. By prioritising monophyly above everything else, some PSC-based approaches risk sidelining genuine evolutionary and ecological considerations. This becomes particularly problematic if one “waits” for the ancestral species to regain monophyly before recognising newly evolved forms, ignoring true species-level divergence. By doing so, we would give priority to the monophyly of species and its overall evolutionary and ecological realities, effectively ignoring many valid species recognised under the ‘Biological Species Concept’ (as in the case of many reproductively isolated, sympatric salmoniform species endemic to postglacial lakes). In daily practice, this dismissal of non-monophyletic, but evolutionarily distinct lineages often occur as most taxonomists have a molecular phylogenetic background, and have either ignored, overlooked, or given up on other species concepts. They are “foot-voting” to prioritise the PSC over everything else, blindly following what is seen in molecular phylogenetic trees, without fully considering morphological, ecological, or reproductive isolation criteria.

Tree topologies of mitochondrial (control region [CR], cytochrome c oxidase subunit 1 [COI], cytochrome b [Cytb]) DNA sequences and nuclear data presented by Kruckenhauser et al. (2011), Hamidan et al. (2014), and Kirchner et al. (2020) recovered G. cavernicola within the G. longipinnis ( G. barreimiae in older papers) clade, lacking any distinct haplotypes of these genes consistent with a recent divergence. This situation very likely results from a relatively recent onset of the independent evolutionary history of G. cavernicola . Kruckenhauser et al. (2011) and Kirchner et al. (2017), analysing microsatellites, suggested that G. cavernicola (= cave population of G. barreimiae ) is phylogenetically very young and potentially became isolated from adjacent G. longipinnis only in the last interglacial period (125 to 120 ka) or during the early Holocene (9.7 to 6.2 ka). Consequently, G. longipinnis did not have enough time to re-establish monophyly after this split. However, the lack of immediate monophyly does not negate the existence of a distinct evolutionary trajectory in the cave lineage. Hence, G. cavernicola is herein described as a separate species under the ‘Evolutionary Species Concept’.

Material examined. Garra longipinnis . BMNH 1968.10.11.1, holotype, 52 mm SL, BMNH 1968.10.11.2-8, 7, paratypes, 37–50 mm SL; Oman: Saiq, Jebel Akhdar.—FSJF 4078, 16, 43–58 mm SL; Oman: pool in front of Ghubrat Tanuf Cave, 23.071°N, 57.368°E.—FSJF 4080, 22, 42–59 mm SL; Oman: spring pool in old town of Saiq, 23.073°N, 57.663°E.

Figures

FIGURE 1. Garra cavernicola, NMW 100394 (443), holotype, 39 mm SL, (223), paratype, 41 mm SL, (445) paratype, 43 mm SL; Al-Hoota cave, Oman.

FIGURE 2. Garra cavernicola, NMW 100394 (443), holotype, 39 mm SL, (223), paratype, 41 mm SL, (445) paratype, 43 mm SL; Al-Hoota cave, Oman.

FIGURE 3. Garra cavernicola, NMW 100394 (223), paratype, 41 mm SL; Al-Hoota cave, Oman.

FIGURE 4. Garra cavernicola, not preserved, ~55 mm SL; Al-Hoota cave, Oman.

FIGURE 5. Records of Garra longipinnis and G. cavernicola, based on data published by Kirchner et al. (2020).