Do Chinese cavefish show intraspecific variability in morphological traits?

Abstract Cavefishes represent one of the most bizarre and intriguing life forms inhabiting groundwater environments. One‐third of the known cavefishes worldwide is endemic to China, and almost half of those belongs to a single genus, Sinocyclocheilus (Cypriniformes: Cyprinidae). Analyzing the morphometrics of three Sinocyclocheilus species, we aimed to assess whether variability among conspecific populations exists. We predict that populations inhabiting different subterranean habitats (shallow vs. deep) show divergences in specific morphological traits to better cope with the local ecological conditions. Our results showed that the populations showing bigger eyes and reduced humpback were those occurring close to the cave entrance (habitats with light and high food availability), while specimens with smaller eyes and increased humpback were collected from deeper groundwater areas (habitats laying in darkness with food scarcity). This explorative study paves the way for further researches aiming to collect novel data on Chinese cavefishes and highlights the usefulness of these species in evolutionary studies.

Indeed, cave-adapted species not only tend to waste less energies (Hervant, 2012), but also try to increase the amount of energy stored in their body (Fišer, 2019;Ma et al., 2019). For this purpose, some species of cavefish can develop a specific morphological adaptation: the humpback (Zhao & Zhang, 2009). This adipose tissue is located on the fish back right behind its head (see Figure 1 in Lunghi, Zhao, Sun, & Zhao, 2019), and it servesas energy storage (Vandel, 1965;Zhao & Zhang, 2009).
China holds more than one-third of the known cavefish species worldwide, and most of them (>46%) belongs to the genus Sinocyclocheilus (Cypriniformes: Cyprinidae), which is, overall, the largest genus of cavefish and is endemic to China (Zhao & Zhang, 2009). In this study, we assessed whether intraspecific morphological variability in three Sinocyclocheilus cavefishes exists.
Considering the high adaptability characterizing Chinese cavefishes (Fenolio, Zhao, Niemiller, & Stout, 2013;Ma et al., 2019), we hypothesized that conspecific populations inhabiting different habitats may diverge in some morphological traits as a result of a different ecological pressures (Parzefall, 2001). Specifically, we predict that populations inhabiting the deepest groundwater habitats (i.e., area laying in darkness with food scarcity) show smaller eyes and larger humpbacks compared to those living close to the cave entrance (i.e., more illuminated areas with higher food availability). Testing this hypothesis will provide a better understanding of the evolutionary processes behind species adaptation to subterranean environments. Indeed, it is not clear yet to which extent the similarity in morphological traits occurring between cave species is due to a common phylogenetic origin rather than to similar ecological pressures (Culver & Pipan, 2015;Howarth, 2019). Therefore, a potential divergence among conspecific populations may strength the hypothesis supporting a preponderant effect of the local ecological pressures.

| Analyzed data
We analyzed the data published by Lunghi, Zhao, et al. (2019). We focused our study only on three species (Sinocyclocheilus brevibarbatus, S. jii, and S. microphthalmus; Figure 1 (Lunghi, Zhao, et al., 2019; but see also Appendix S1), for S. jii 27 because the eyeball diameter equals that of the eye (Zhao & Zhang, 2009). Specimens' preservation was not always optimal, and sometimes, the fish body was damaged; the highest rate of damage occurred in fins (>82%).

| Statistical analyses
We used the principal component analysis (PCA) to explore the morphological data from the three Sinocyclocheilus species. The PCA helps in reducing the group of correlated variables (the cavefish morphometrics) into a set of linearly independent variables. The obtained variables (the principal components) are ranked according to the amount of variance they explain; the first usually explains the largest amount of variance. Data related to fish fins were excluded from the analysis as they were often damaged; this allows to analyze the highest number of specimens. PCA analysis was run for each species singularly and included 21 variables for S. brevibarbatus and S. microphthalmus, while only 20 for S. jii (see Appendix S1). All data were log-transformed to improve normality and reduce skewness. For each of the first two components (those explaining the highest amount of variance), the significance of the most important correlated variables (loading value ≥ 0.55) was assessed using the analysis of variance (ANOVA) or the multivariate analysis of variance (MANOVA).
The potential inconsistency of the food resources in subterranean environments may affect our analysis, as the humpback area of cavefishes strongly depends by the temporal availability of resources (Vandel, 1965). Indeed, food resources in groundwater environments are irregularly enriched through the seasonal supply of organic matters brought by the incoming water from the surface Analyses were performed using the software PAST and R (R Development Core Team, 2018).

| Sinocyclocheilus brevibarbatus
The PCA analysis identified two groups of specimens with little overlap ( Figure 3a); the first two components together explained 91.57% of the variance (Table 1). The most important variables for the principal components were the humpback area (PC1) and the eyeball diameter (PC2) ( Table 2). Both variables were significantly different between the two populations (eyeball diameter, df = 1, F = 19.12, p < .001; humpback area, df = 1, F = 8.35, p = .007); the Bb1 population included specimens with bigger eyes and smaller humpback area.

| Sinocyclocheilus jii
The PCA analysis grouped all populations into a single group (Figure 3b); the first two components together explained 93.16% of the variance (Table 1), and for both, the only important variable was the humpback area (Table 2). Although all specimens were grouped into a single group, specimens from Ji1 showed a significantly smaller humpback area compared to those from the other two populations (df = 2, F = 18.52, p < .001).

| Sinocyclocheilus microphthalmus
The PCA analysis identified two distinct groups, one including populations Mi2 and Mi5 and one with Mi6, Mi8, and Mi10 ( Figure 3c); the first two components together explained 94.91% of the variance (Table 1). The most important variables for the principal components were the humpback area (PC1 and PC2) and the eyeball diameter (PC2) ( Table 2). The two variables were significantly different between the two groups (eyeball diameter, df = 4, F = 8.36, p < .001; humpback area, df = 4, F = 63.39, p < .001); the group including Mi2 and Mi5 had specimens with bigger eye diameter and smaller humpback area.  The loading value of considered morphological traits is shown for each species separately and for the two reduced subsets of S. microphthalmus as well. The following codes are the same showed in the dataset of Lunghi, Zhao, et al. (2019): Eye (eye diameter); Eye_ball (eyeball diameter); Snout (distance between the mouth tip and the beginning of the eye); Mouth width (length between the two mouth angles); Mouth length (length of the lower jaw); AD (linear distance between the snout tip and the top end of the head); B_height (head height measured at the nostril); C_height (head height measured at the eye); D_height (head height measured at the upper end); DI (linear distance between the top end of the head and the beginning of the dorsal fin); AE (maximum head length, measured from the snout tip until the farthest end of the head); FG (length of the forward pectoral fin base); IK (length of the dorsal fin base); I_depth (body depth measured at the beginning of the dorsal fin base); JW (length of the backward pectoral fin base); K_depth (body depth measured at the end of the dorsal fin base); NO (length of the anal fin base); O_depth (body depth measured at the end of the anal fin base); QR (caudal fin height at its base); AS (standard length); DID (humpback area). For each studied species we show the list of variables along their loading values for the first two PCA components.
NA means that the relative morphometric is not present for the species.
a Factors for which we tested the divergence between populations (loading values ≥ 0.55).

| D ISCUSS I ON
With this study, we provided the evidence that intraspecific morphological variability occurs in Sinocyclocheilus cavefishes. In particular, in two out of the three studied species we observed a significant divergence for two morphological traits, namely the eyeball diameter and the humpback area (Table 2). According to our results, the first principal component likely represents the specimens' size, as all the considered morphological traits showed a positive correlation with this axis (  Figure 3a,c) likely inhabited areas not far from the connection with surface, where incoming light is still present, and thus, the use of sight is important (Culver & Pipan, 2014;Lunghi et al., 2015;Uiblein, 1992). Indeed, Mi2 specimens were collected from the cave entrance, while those belonging to populations Mi6, Mi8, and Mi10 were collected inside the cave more than 1 km far from the connection with surface, in areas laying in complete darkness.
Unfortunately, no information on the collection site is available for Bb1, Bb2, and Mi5, and thus, future ecological surveys are needed to confirm this pattern. The small humpback area observed in Mi2 and Mi5 likely reflects a reduced need to store fat, probably as a consequence of a constant food availability (Culver & Pipan, 2014;Vandel, 1965); on the other hand, the increased humpback area observed in the conspecific Mi6, Mi8, and Mi10 populations likely indicate the need to store energy to better cope with irregular abundance of food resources (Culver & Pipan, 2019 Although the similarity in morphometrics shared by these specimens (no distinct groups were identified by the PCA analysis; Figure 3b), the humpback area of the population Ji1 was significantly smaller  (Howarth, 2019).

ACK N OWLED G M ENTS
This study is supported by grants from National Natural Science

CO N FLI C T O F I NTE R E S T S
We declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available from the following publication: Lunghi, Zhao, et al. (2019).