﻿Sinocyclocheiluslongicornus (Cypriniformes, Cyprinidae), a new species of microphthalmic hypogean fish from Guizhou, Southwest China

﻿Abstract Sinocyclocheiluslongicornussp. nov. is described from the Pearl River basin in Hongguo Town, Panzhou City, Guizhou Province, Southwest China. Based on the presence of the long horn-like structure on the back of the head, Sinocyclocheiluslongicornussp. nov. is assigned to the Sinocyclocheilusangularis species group. Sinocyclocheiluslongicornussp. nov. is distinguished from its congeners by a combination of morphological characters: (1) presence of a single, relatively long horn-like structure on the back of the head; (2) pigmentation absent; (3) reduced eyes; (4) dorsal-fin rays, ii, 7; (5) pectoral-fin rays, i, 13; (6) anal-fin rays, iii, 5; (7) pelvic-fin rays, i, 7; (8) lateral line pores 38–49; (9) gill rakers well developed, nine on first gill arch; and (10) tip of adpressed pelvic fin not reaching anus.

. List of 76 currently recognized species of the genus Sinocycheilus endemic to China and references. Recognized species modified from Jiang et al. (2019).

Introduction
The golden-line fish genus Sinocyclocheilus Fang, 1936, is endemic to China, and is mainly distributed in the karst areas of Southwest China, including Guangxi, Guizhou, Yunnan, and Hubei provinces Jiang et al. 2019). The narrow distribution, morphological similarities, and morphological adaptations to cave environments, such as the degeneration or loss of eyes and body scales, have made classification of the genus difficult and often controversial (Chu and Cui 1985;Shan and Yue 1994;Wang et al. 1995;Wang and Chen 1998;Wang et al. 1999; Wang and Chen 2000;Xiao et al. 2005;Mao et al. 2021Mao et al. , 2022Wen et al. 2022). A phylogenetic study based on the mitochondrial cytochrome b gene (Cyt b) showed that all members of Sinocyclocheilus clustered as a monophyletic group, divided into four species groups, namely the S. jii, S. angularis, S. cyphotergous, and S. tingi groups . However, phylogenetic studies based on restriction site-associated DNA sequencing and mitochondrial genome reconstruction suggest that the S. angularis and S. cyphotergous species groups are not monophyletic (Xiang 2014;Liu 2018;Mao et al. 2021Mao et al. , 2022Wen et al. 2022). Sinocyclocheilus comprises 76 valid species, of which 71 species are grouped into five species groups (Table 1). 56 S. luopingensis Li & Tao, 2002 S. cyphotergous group Yunnan Nanpanjiang River Li et al. 2002a57 S. macrolepis Wang & Chen, 1989 S. cyphotergous group Guizhou; Guangxi Liujiang River Wang and Chen 1989 58 S. macrophthalmus Zhang & Zhao, 2001 S. cyphotergous group Guangxi Hongshuihe River Zhao 2001 59 S. macroscalus Li, 1992 S. tingi group Yunnan Nanpanjiang River Li 1992 60 S. multipunctatus (Pellegrin, 1931) S  Zhu, Zhu & Lan, 2011 S. jii group Guangxi Hejiang River Zhu et al. 201171 S. jii Zhang & Dai, 1992 S. jii group Guangxi Guijiang River Zhang and Dai 1992 72 S. gracilis  No assignment Guangxi Guijiang River Li and Li 2014 73 S. pingshanensis Li, Li, Lan & Wu, 2018 No assignment Guangxi Liujiang River Wu et al. 2018 74 S. wenshanensis Li,Yang, Li & Chen, 2018 No assignment Yunnan Panlonghe River Yang et al. 201875 S. wui Li & An, 2013 No assignment Yunnan Mingyihe River Li and An 2013 76 S. luolouensis Lan, 2013 No assignment Guangxi Hongshuihe River Lan et al. 2013  Species of Sinocyclocheilus have variably developed eyes and horn-like structures on the back of the head. Eye morphology includes normal, microphthalmic, and anophthalmic conditions (Mao et al. 2021). Normal-eyed and microphthalmic species are distributed from eastern Guangxi through southern Guizhou to eastern Yunnan, and eyeless species are mainly distributed in the Hongshuihe river basin in northern Guangxi and the Nanpanjiang river basin in eastern Yunnan (Mao et al. 2021). It may be absent, short, long, or single and forked. The horn-like structure is present mainly in species of the S. angularis and S. microphthalmus species groups Mao et al. 2021;Wen et al. 2022). These horned species are distributed in the Nanpanjiang, Beipanjiang, and Hongshuihe river basins of the upper Pearl River (Fig. 1).
We collected specimens of a horned, scaleless, and unpigmented species of Sinocyclocheilus in a completely dark cave in southwestern Guizhou Province in China. Molecular phylogenetic analyses and morphological comparisons showed that these specimens represented an undescribed species of Sinocyclocheilus. Here, we provide the formal description of that species as Sinocyclocheilus longicornus sp. nov.

Specimen sampling
During a cavefish diversity survey in southern China in May 2021, 32 specimens of the genus Sinocyclocheilus were collected in southwestern Guizhou Province. Among these, 15 specimens represented an undescribed species, subject of this, paper from Hongguo Town in Panzhou City; seven were S. angularis from Baotian Town in Panzhou; two were S. bicornutus from Xiashan Town in Xingren City; and eight were S. zhenfengensis from Zhexiang Town in Zhenfeng County (Fig. 1). Gill muscle tissues used for molecular analysis were preserved in 95% alcohol at −20 °C. All specimens were fixed in 10% buffered formalin and later transferred to 75% ethanol for long term preservation. All specimens were deposited in Guizhou Normal University, Guiyang City, Guizhou Province, China.

DNA Extraction, PCR amplification, and sequencing
Genomic DNA was extracted from muscle tissues using a DNA extraction kit from Tiangen Biotech Co., Ltd. (Beijing, China). Because the most used molecular markers in Sinocyclocheilus are fragments of the mitochondrial cytochrome b (Cyt b) and NADH dehydrogenase subunit 4 (ND4) genes, we selected these fragments for amplification and sequencing. Primers used for Cyt b were L14737 (5'-CCAC-CGTTGTTAATTCAACTAC-3') and H15915 (5'-CTCCGATCTCCGGATTA-CAAGAC-3'), following Xiao et al. (2005). Primers used for ND4 were L11264 (5'-ACGGGACTGAGCGATTAC-3') and H12346 (5'-TCATCATATTGGGT-TAG-3'), following Xiao et al. (2005). PCR amplifications were performed in a 25-μl reaction volume with the following cycling conditions: an initial denaturing step at 95 °C for 3 min; 35 cycles of denaturing at 94 °C for 50 s, annealing at 52 °C (for Cyt b and ND4) for 1 min and extension at 72 °C for 1 min, and a final extension step of 72 °C for 10 min. The PCR products were sequenced on an ABI Prism 3730 automated DNA sequencer at Chengdu TSING KE Biological Technology Co. Ltd. (Chengdu, China). All sequences were deposited in GenBank (Table 2).
All sequences were assembled and aligned using the MUSCLE (Edgar 2004) module in MEGA 7.0 (Kumar et al. 2016) with default settings. Alignment results were checked by eye. Phylogenetic trees were constructed with both maximum likelihood (ML) and Bayesian inference (BI) methods. The ML was conducted in IQ-TREE 2.0.4 (Nguyen et al. 2015) with 2000 ultrafast bootstrap (UBP) replicates (Hoang et al. 2018) and was performed until a correlation coefficient of at least 0.99 was reached. The BI was performed in MrBayes 3.2.1 (Ronquist et al. 2012), and the best-fit model was obtained based on the Bayesian information criterion computed with Partition-Finder 2.1.1 (Lanfear et al. 2017). In this analysis, the first, second and third codons of both Cyt b and ND4 genes were defined. The analysis suggested the best partition scheme for each codon position of Cyt b and ND4 genes. GTR+I+G, HKY+I+G, and TRN+I+G were selected for first, second, and third codons, respectively for both Cyt b and ND4 genes. Two independent runs were conducted in BI analysis, each of which was performed for 2 × 10 7 generations and sampled every 1000 generations. The first 25% of the samples were discarded as burn-in, resulting in a potential scale reduction factor of < 0.01. Nodes in the trees were considered well supported when Bayesian posterior probabilities (BPP) were ≥ 0.95 and the ML ultrafast bootstrap value (UBP) was ≥ 95%. Uncorrected pdistances (1000 replicates) based on Cyt b and ND4 genes were calculated in MEGA 7.0 (Kumar et al. 2016

Morphological comparisons
Morphometric data were collected from 44 well-preserved specimens of Sinicyclocheilus (Suppl. material 1). A total of 31 measurements were recorded to the nearest 0.1 mm with digital calipers following the protocol of Zhao et al. (2006) and . The following measurements were taken: TL total length (from the tip of snout to the end of the caudal-fin); SL standard length (from the tip of the upper jaw to the position of the last half-centrum); BD body depth (from the insertion of the dorsal fin vertically to the ventral midline); PL predorsal length (from the tip of the upper jaw to the origin of the dorsal-fin); DFL dorsal-fin depth (from the origin of the dorsal-fin to the tip of the longest ray); DBL dorsal-fin length (from the origin to the insertion of dorsal-fin base); PAL preanal length (from the tip of the upper jaw to the origin of the anal-fin); ABL anal-fin base length (from the origin to the insertion of anal-fin base); AFL anal-fin depth (from the origin of the anal-fin to the tip of the longest ray); PPTL prepectoral length (from the tip of the upper jaw to the base of anterior pectoral-fin ray); PTBL pectoral-fin base length (from the anterior to posterior end of pectoral-fin base); PTFL pectoral-fin length (from the base of the first pectoral-fin ray to the tip of the longest ray); PPVL prepelvic length (from the tip of the upper jaw to the base of the first pelvicfin ray); PVBL pelvic-fin base length (from the anterior to the posterior end of the pelvicfin base); PVFL pelvic-fin length (from the base of the first pelvic-fin ray to the tip of the longest ray); CPL caudal peduncle length (from the anal-fin insertion to the position of the last centrum); CPD caudal peduncle depth (depth at the narrowest part of the caudal peduncle); HL head length (from the tip of the upper jaw to the posteriormost point of the operculum); HD head depth at nape; HW head width (widest distance between the two gill covers); SNL snout length (from tip of snout to the anterior corner of the eye); ED eye diameter (diameter of the exposed portion of the eyeball); IOD interorbital distance (minimum distance between the eyes); IPND prenostril distance (the tip of the upper jaw to the anterior margin of the anterior nostril); POND distance between posterior nostrils (the shortest distance between posterior nostrils); UJL upper jaw length (from the tip of the upper jaw (the symphysis of the premaxilla) to the corner of the mouth); LJL lower jaw length (from the symphysis of the dentary to the corner of the mouth); MW mouth width (the distance between the two corners of the mouth); RBL rostral barbel length; MBL maxillary barbel length; FHL forehead horn length; PFPVL distance from the pectoral-fin insertion to the ventral-fin origin; and PVAFL distance from the insertion of the pelvic fin to the origin of the anal-fin.
We compared the morphological characters of the new species with literature data for 21 other species in the S. angularis and S. microphthalmus species groups (Table 3). We also examined the type and/or materials from the type-localities of S. angularis, S. bicornutus, S. hyalinus, S. rhinocerous, and S. zhenfengensis (Appendix 1). Principal component analyses (PCAs) of size-corrected measurements and simple bivariate scatterplots were used to explore and characterize the morphometric differences between the new species and S. rhinocerous and S. hyalinus. Mann-Whitney U tests were used to determine the significance of differences in morphometric characters between the new species and similar species, i.e., S. angularis, S. bicornutus, and S. rhinocerous. All statistical analyses were performed using SPSS 21.0 (SPSS, Inc., Chicago, IL, USA), and differences were considered statistically significant at P < 0.05. PCAs of morphological data were performed after logarithmic transformation and under conditions of no rotation. In addition, as reported by other researchers (Parsons and Jones 2000;Polaszek et al. 2010), canonical discriminant analysis (CDA, George and Paul 2010) was used to classify individuals into different groups, where a priori membership was determined based on specimens belonging to different species. All pre-processing of morphological data was performed in Microsoft Excel (Microsoft Corporation 2016).

Phylogenetic analyses and genetic divergence
ML and BI phylogenies were constructed based on two concatenated mitochondrial gene sequences, including 1140 bp Cyt b and 1380 bp ND4. The ML and the BI phylogenetic trees showed identical topology (Fig. 2). The monophyly of the genus Sinocyclocheilus was strongly supported by both phylogenetic analyses but the monophyly of the S. angularis and S. cyphotergous species groups was rejected (Fig. 2). In both analyses, the S. longicornus sp. nov. formed a highly supported clade (0.99 in BI and 96% in ML) with S. hyalinus and S. rhinocerous.
The smallest p-distances between S. longicornus sp. nov. and other species of Sinocyclocheilus were 6.0% in Cyt b (with S. rhinocerous) and 5.6% in ND4 (with S. bicornutus). These levels of divergence were similar to those between pairs of other recognized species. For example, the Cyt b p-distance was 2.4% between S. anatirostris and S. angularis, 3.4% between S. bicornutus and S. brevibarbatus, while the ND4 p-distance was 2.7% between S. anatirostris and S. angularis and 2.6% between S. bicornutus and S. anatirostris (Suppl. materials 2, 3).

Morphological analyses
Mann-Whitney U tests showed that the Sinocyclocheilus longicornus sp. nov. differed from S. angularis, S. bicornutus, and S. rhinocerous in several morphological characters (Table 4). This was specially most obvious comparing S. longicornus sp. nov. and S. rhinocerous, in wihich 87% of the morphometric characters were significantly different (p = 0.00-0.03) ( Table 3). Tips of the pelvic-fin rays without reaches to the anus  Table 2. Different colored rectangular and triangular boxes in addition to the nodes denote the different states of the presence of horn-like structures of species within the genus Sinocyclocheilus. Based on PCA of the morphological data, two principal component factors with eigenvalues greater than two were extracted. These accounted for a total of 89.86% of the total variation (Suppl. material 4). The first principal component (PC1) accounted for 83.37% of the variation and was positively correlated with all variables (eigenvalue = 27.22), thus reflecting the morphological differences between S. longicornus sp. nov. and similar species. The second principal component (PC2) accounted for 4.85% of the variation and was dominated by the length of the lower jaw (LJL), length of the upper jaw (UJL), and length of the head (HL) (eigenvalue = 0.44). On the two-dimensional plots of PC1 and PC2, S. longicornus sp. nov. can be clearly distinguished from S. angularis, S. rhinocerous, and S. hyalinus, and can be almost separated from S. angularis (Fig. 3A). A total of 29 characters were loaded on the PC 1 axis and were mainly influenced by body length, head, and fin ray characteristics (Suppl. material 4). CDA correctly classified 100% of the individuals in the initial grouping case for the four sample groups (N = 36). Canonical axes (CAN) 1-3 explained 59.8%, 30.6%, and 9.6% of the total variation, respectively ( Fig. 3B; Suppl. material 5). Therefore, based on PCA and CDA, the 15 specimens of S. longicornus sp. nov. regions in the space of morphological characters compared to four similar species.  Diagnosis. Sinocyclocheilus longicornus sp. nov. can be distinguished from all other congeners by the following combination of characters: (1) having a single, relatively long horn-like structure on the back of the head; (2) body scaleless, albinotic body without pigmentation; (3) reduced eyes; (4) dorsal-fin rays, ii, 7; (5) pectoral-fin rays, i, 13; (6) anal-fin rays, iii, 5; (7) pelvic-fin rays, i, 7; (8) lateral line pores 38-49; (9) gill rakers well developed, 9 on first gill arch; (10) tip of the pelvic-fin rays not reaching the anus when pelvic-fin rays extended backward.

Sinocyclocheilus longicornus
Description. Body moderately elongate and compressed. Dorsal profile convex from nape to dorsal-fin; greatest body depth at dorsal-fin insertion; ventral profile slightly concave, tapering gradually toward the caudal-fin; greatest body depth slightly anterior to dorsal-fin insertion.
Head short, compressed laterally, length longer than maximum head width, depth longer than maximum head width. large and long anterior horn-like structure present on back of head not forked at tip, at about 45° angle to horizontal and curved downward at tip. Reduced eyes present in upper half of head; eye diameter less than interorbital distance; interorbital distance larger than distance between posterior nostrils. Snout short, U-shaped, and projecting beyond lower jaw in dorsal view, less than half head length.
Mouth subterminal, with slightly projecting upper jaw. Two pairs of nostrils, anterior and posterior nostrils neighboring, nares at 1/3 between snout tip and anterior margin of eye; anterior nares possessing an anterior rim with a posterior fleshy flap forming a half-tube. Two pairs of barbels; rostral barbels long, insertion of rostral barbel in front of anterior nostril, not reaching anterior edge of operculum when rostral bent backward; maxillary barbel slightly shorter compared to rostral barbel, tip surpassing eye but not reaching anterior edge of operculum when bent backward. Gill opening large, opercular membranes connected at isthmus, gill rakers well developed, nine on first gill arch. Pharyngeal teeth in three rows with counts of 2, 3, 5-5, 3, 2; pharyngeal teeth strong and well developed, with curved and pointed tips.
Dorsal fin with two unbranched and seven branched rays; last unbranched dorsalfin ray hard at base, softening toward tip, with strong serrations along posterior edge; distal margin slightly concave, origin slightly anterior to, or superior to, pelvic-fin insertion and closer to caudal-fin base than to snout tip. Pectoral fin long with one unbranched and 13 branched rays; tip of depressed fin extending about midway between pectoral fin and pelvic-fin insertion; extending from posterior to pelvic-fin insertion and reaching to 35.44% of pelvic-fin length. Pelvic-fin long with one unbranched and seven branched rays, insertion slightly in front of dorsal-fin insertion, tip of the pelvic-fin rays not reaching the anus when pelvic-fin rays extended backward. Anus closer to anal-fin insertion than pelvic-fin insertion; anal fin with three unbranched and five branched rays; tip of anal-fin not reaching to caudal-fin base. Caudal fin with 17 branched rays and 14 unbranched rays, strongly forked; upper and lower lobes broadly pointed, unequal in length and shape.
Lateral line complete, slightly straight, curved upward at the anus position, originating from posterior margin of operculum and extending to end of caudal peduncle. Body scaleless, lateral line pores 38-49.
Coloration of holotype. In life, body overall white, slightly pink posterior to dorsal fin; barbels and gills red (Fig. 5); with white granular nuptial organs on dorsal surfaces of horn-like structure on back of head and snout (Fig. 5). In 10% formalin, body overall light yellow; posterior part of operculum and all fins partially transparent (Fig. 4).
Geographical distribution and habitat. Sinocyclocheilus longicornus sp. nov. is only known from the type locality, a vertical cave some distance from Hongguo Town, Panzhou city, Guizhou, China at an elevation of 2276 m. There was no light inside the cave. Individuals of S. longicornus sp. nov. were located in a small pool ~ 25 m from the cave entrance. The pool was ~ 1.8 m wide and 80 cm deep, with a water temperature of ~ 16 °C at collection time and a water pH of 7.4. The 15 specimens collected on 3 May 2021 were all adult males. Therefore, we believe that the breeding period started from mid-April. Within this cave, Sinocyclocheilus longicornus sp. nov. co-occurred with Triplophysa sp., and Sinocyclocheilus sp. Outside the cave, the arable land was farmed to produce maize, wheat, and potatoes.
Etymology. The specific epithet longicornus is an invariable noun in apposition, derived from the Latin words longus, meaning long, and cornu or cornus, meaning horn of the forehead, in reference to the presence of a long horn-like structure on the forehead of the species. We propose the English common name Long-Horned Goldenlined Fish and the Chinese common name Cháng Jiǎo Jīn Xiàn Bā (长角金线鲃).

Discussion
Morphological comparison and phylogenetic analysis support the generic assignment and and separate species status of Sinocyclocheilus longicornus sp. nov. The genetic differences between the new species and its close relatives, S. hyalinus and S. rhinocerous, were greater than the known genetic distances between other species (Suppl. materials 3, 4). Sinocyclocheilus longicornus sp. nov. the number of species of Sinocyclocheilus to 77, of which 13 species are recorded from Guizhou Province, China.
The genus Sinocyclocheilus is recognized as monophyletic, but there is no consensus on the classification of species groups Xiang 2014;Liu 2018;Mao et al. 2021Mao et al. , 2022Wen et al. 2022). Initially, Sinocyclocheilus was divided into four species groups, S. jii, S. angularis, S. cyphotergous, and S. tingi, based on mitochondrial Cyt b and morphological differences ). Phylogenetic trees reconstructed using mitochondrial ND4 and Cyt b, mitochondrial genome, and restriction site-associated DNA sequencing supported monophyly of the S. jii and S. tingi species groups and rejected monophyly of the S. angularis and S. cyphotergous species groups (Xiang 2014;Liu 2018;Mao et al. 2021Mao et al. , 2022Wen et al. 2022;this study). These studies proposed new classification schemes, such as two new clades (Clades E and F) from Mao et al. (2022) and a new species group (S. microphthalmus group) from Wen et al. (2022). Inconsistent topological differences may be related to molecular marker types, number of species and evolutionary models. For example, a phylogenetic tree reconstructed by Mao et al. (2021) for 49 species of Sinocyclocheilus using the GTR+I+G model for both mitochondrial ND4 and Cyt b rejected monophyly of the S. cyphotergous group. We reanalyzed their data for codon partitioning and found that the monophyly of both S. angularis and S. cyphotergous species groups was rejected. Different genes and different codons may have different evolutionary rates (Degnan and Rosenberg 2009), so the analysis may produce conflicting results when the same untested model is applied to different gene segments. Therefore, to resolve classification disagreements among species groups, the use of genomic data and a sufficient number of species is needed for future studies.
Variable or specialized morphological characters of Sinocycheilus are closely related to the orogeny producing dark cave environments (Yang et al. 2016;Mao et al. 2021Mao et al. , 2022Wen et al. 2022). For example, horn-like structures (single or forked, long or short) or bulges on the back of the head, and degeneration or loss of eyes . Sinocyclocheilus longicornus sp. nov. has a relatively long, unforked horn-like structure on the forehead, and small, degenerated eyes. It clustered with eight species of the S. angularis species group on the phylogenetic tree and could be divided into Clade I and Clade II. (Fig. 2). Long and short/indistinct horn-like structures are present in Clade I and Clade II, respectively (Fig. 2). Based on the present study and previous phylogenetic trees (Mao et al. 2021(Mao et al. , 2022Wen et al. 2022), we hypothesize that the evolution of the forehead horn may have occurred in at least two independent formations, one weakening event and one loss event (Fig. 2). As for the eye, no corresponding clade was found within the S. angularis species group, and variable eye phenotypes were also reported within S. bicornutus (in press), which may be related to the reduction of eye size during evolution or to the abundance and deprivation of food resources during growth and development, as well as related gene mutations (Ma et al. 2020;Mao et al. 2021).