A new species of the Rana japonica group (Anura, Ranidae, Rana) from China, with a taxonomic proposal for the R. johnsi group

Abstract Rana jiulingensissp. nov., a new species from Hunan and Jiangxi, southeastern China, is described. The new species is assigned to the R. japonica group. The clade comprising R. jiulingensissp. nov. and R. dabieshanensis from Anhui is the sister taxon of R. omeimontis from Sichuan. Rana jiulingensissp. nov. can be distinguished by the significant divergences in the 16S and COI genes, and the combination of following morphological characters: body size medium, SVL 48.3–57.8 mm in adult males and 48.2–57.5 mm in adult females; dorsolateral fold straight; digits without circummarginal grooves; dorsal skin smooth; tibio-tarsal articulation reaching forward beyond the tip of snout; heels overlapping; webbing formula of toes: I 1⅓ – 2 II 1⅓ – 2⅓ III 1½ – 2⅔ IV 3 – 1⅓ V; absence of vocal sacs in males; and presence of creamy white nuptial pad with tiny hoar spines on the finger I and reddish tubercles on loreal and temporal regions in breeding males. Furthermore, based on our results and the previous literature, R. zhengi is synonymized with R. sangzhiensis, and a new species group, the Rana johnsi group, is proposed for the R. johnsi and R. sangzhiensis. Currently, the Rana contains 41 recognized species, and the phylogenetic placements of several species remain unresolved.


Introduction
As the type genus of the family Ranidae Batsch, 1796, the concept of the true-frog genus Rana Linnaeus, 1758 has been discussed for a long time (Frost 2020). In a recent phylogenetic analysis (Yuan et al. 2016), Rana sensu lato was considered to be composed of nine clades, namely the subgenera Rana, Amerana Dubois, 1992, Liuhurana Fei, Ye, Jiang, Dubois & Ohler, 2010, Aquarana Dubois, 1992, Lithobates Fitzinger, 1843, Zweifelia Dubois, 1992, Pantherana Dubois, 1992, Pseudorana Fei, Ye & Huang, 1990, and an unnamed monotypic clade containing R. sylvatica (LeConte, 1825). However, this classification is still controversial, especially for the recognitions of the genera Lithobates and Pseudorana (Frost 2020). Nevertheless, the subgenus Rana, which is currently well recognized, contains 41 known species distributed from Europe to southeastern Asia. Among them, 23 species occur in China (AmphibiaWeb 2019). Recent researches on this subgenus have discovered new species from China and revised several taxonomic errors, indicating that the diversity and taxonomy of the subgenus Rana are still insufficiently understood (Yan et al. 2011;Zhou et al. 2015Zhou et al. , 2017Yuan et al. 2016;Yang et al. 2017;Zhao et al. 2017).
Based on morphological comparisons and geographical conditions, Fei et al. (2009) proposed three species groups for the Chinese species of the subgenus Rana: R. longicrus group, R. chensinensis group, and R. amurensis group. Subsequent phylogenetic analyses have revised several memberships of these groups (Yan et al. 2011;Zhou et al. 2015Zhou et al. , 2017Yuan et al. 2016;Wang et al. 2017;Zhao et al. 2017), and the nomenclature of the R. longicrus group was replaced by the R. japonica group . Currently, 16 Chinese species are recognized as members of the three species groups. The R. japonica group contains nine species: R. (R.) chaochiaoensis Liu, 1946 Shen, 1986;and R. (R.) zhengi Zhao, 1999. During herpetofaunal surveys in the Luoxiao Range, which is situated between the Jiangxi and Hunan provinces (Fig. 1), a series of Rana specimens was collected which can be assigned to the R. japonica group based on morphological characteristics. However, detailed examination of these specimens showed significant differences from all known congeners. Additional molecular analysis well supported the morphological identifications, demonstrating that these specimens formed an unnamed lineage within the R. japonica group. Therefore, we describe this series of specimens as a new species. Additionally, as revealed from our results and the previous literature, we suggest that R. zhengi should be synonymized with R. sangzhiensis, and we also propose a new species group, the Rana johnsi group, for the species R. johnsi and R. sangzhiensis.

Sampling and morphological analyses
Eight unnamed specimens were collected from multiple localities of Jiangxi and Hunan provinces. All specimens were fixed in 10% buffered formalin, later transferred to 70% ethanol, and deposited in the Museum of Biology, Sun Yat-sen University (SYS) and Chengdu Institute of Biology, Chinese Academy of Sciences (CIB), PR China. External measurements were made for the unnamed specimens with digital calipers (Neiko 01407A Stainless Steel 6-Inch Digital Caliper, USA) to the nearest 0.1 mm. These measurements are as follows:  Table 1. SVL snout-vent length (from tip of snout to posterior margin of vent); HL head length (from tip of snout to the articulation of the jaw); HW head width (head width at the commissure of the jaws); SL snout length (from tip of snout to the anterior corner of the eye); IN internasal distance (distance between nares); IO interorbital distance (minimum distance between upper eyelids); ED eye diameter (from the anterior corner of the eye to posterior corner of the eye); TD tympanum diameter (horizontal diameter of tympanum); TED tympanum-eye distance (from anterior edge of tympanum to posterior corner of the eye); HND hand length (from the proximal border of the outer palmar tubercle to the tip of digit III); RAD radio-ulna length (from the flexed elbow to the proximal border of the outer palmar tubercle); FTL foot length (from distal end of shank to the tip of digit IV); TIB tibial length (from the outer surface of the flexed knee to the heel).
The morphological description follows the consistent definition by Fei et al. (2009). Sex and age were determined by examining the gonads. Webbing formula was based on Savage (1975). Comparison characters of known congeners were obtained from the literature (Stejneger 1898;Liu 1946;Liu et al. 1993;Ye et al. 1993Ye et al. , 1995Lu et al. 2007;Shen et al. 2007;Li et al. 2008;Fei et al. 2009Fei et al. , 2012Yan et al. 2011;Wang et al. 2017;Zhao et al. 2017) and 80 examined museum specimens listed in the Appendix 1.

DNA Extraction, PCR amplification, and sequencing
A total of 56 muscular samples of Rana were used, encompassing nine samples of the undescribed specimens, and 47 samples from 12 recognized species. All samples were attained from euthanasia specimens and then preserved in 95% ethanol and stored at -40 °C. Genomic DNA were extracted from muscle tissue samples, using DNA extraction kit from Tiangen Biotech (Beijing) Co., Ltd. Two mitochondrion genes, namely partial 16S ribosomal RNA gene (16S) and partial cytochrome c oxidase 1 gene (COI), were amplified. Primers used for 16S were L3975 (5'-CGCCTGTTTAC-CAAAAACAT-3') and H4551 (5'-CCGGTCTGAACTCAGATCACGT-3') following Simon et al. (1994), and L2A (5'-CCAAACGAGCCTAGTGATAGCTGGTT-3') and H10 (5'-TGATTACGCTACCTTTGCACGGT-3') following Chen et al. (2013), for COI were dgLCO (5'-GGTCAACAAATCATAAAGAYATYGG-3') and dgHCO (5'-AAACTTCAGGGTGACCAAARAAYCA-3') following Meyer et al. (2005). PCR amplifications were processed with the cycling conditions that initial denaturing step at 95 °C for 4 min, 35 cycles of denaturing at 94 °C for 40 s, annealing at 53 °C (for 16S) / 48 °C (for COI) for 40 s and extending at 72 °C for 60 s, and a final extending step at 72 °C for 10 min. PCR products were purified with spin columns and then parameters (Thompson et al. 1997). For GenBank sequences that lack information for part of the segments, we filled the blank sites with "N". The aligned data was trimmed for allowing no gap positions and default parameters in Gblocks version 0.91b (Castresana 2000). All newly obtained sequences were deposited in GenBank (Table 1).
PartitionFinder2 was used to test the best partitioning scheme and jModelTest v2.1.2 was used to test the best fitting nucleotide substitution models, resulting in the best fit models for the partitions of COI and 16S as GTR + I + G. Sequenced data were analyzed using Bayesian inference (BI) in MrBayes 3.2.4 (Ronquist et al. 2012), and maximum likelihood (ML) in RaxmlGUI 1.3 (Silvestro and Michalak 2012). Two independent runs were conducted in a BI analysis, each of which was performed for 10,000,000 generations and sampled every 1000 generations with the first 25% samples discarded as burn-in, resulting in a potential scale reduction factor (PSRF) of <0.005. In ML analysis, the bootstrap consensus tree inferred from 1000 replicates was used to represent the evolutionary history of the taxa analyzed. Pairwise distances were respectively calculated b in MEGA 6 using the uncorrected p-distance model.

Morphological comparison
The unnamed specimens from Jiangxi and Hunan are assigned to the Rana japonica group based on the following combined characteristics: digits without circummarginal grooves, and dorsolateral fold distinct, extending straight from the posterior margin of the upper eyelid to above the groin. Therefore, we compare the new species with the species of the R. japonica group.
The new species differs from R. omeimontis as follows: body size smaller, SVL = 48.2-57.5 mm in adult females (vs 61.7-70.3 mm in females); head length significantly larger than head width, HW/HL = 0.82 in males and 0.85 in females (vs head length slightly larger than head width, HW/HL = 0.94 in males and 0.92 in females); and supernumerary tubercles present below the bases of each finger (vs absent).
The new species further differs from R. hanluica as follows: supratympanic fold absent (vs present); toe webbing formula I 1⅓ -2 II 1⅓ -2⅓ III 1 ½ -2⅔ IV 3 -1⅓ V (vs I 1⅓ -1⅔ II 1 -2 III 1⅓ -2½ IV 2⅓ -1 V); reddish tubercles present on loreal and temporal regions in breeding males (vs absent, but white horny spines present around loreal and temporal regions, upper eyelids, and snout in breeding males). The new species differs from R. longicrus in having: internarial distances larger than interorbital distances (vs smaller) and toe webbing formula I 1⅓ -2 II 1⅓ -2⅓ III 1 ½ -2⅔ IV 3 -1⅓ V (vs I 1⅔ -2⅓ II 1½ -2⅔ III 1⅔ -3½ IV 3⅓ -1½ V); from R. zhenhaiensis: supratympanic fold absent (vs present), dorsolateral fold extending straight from posterior margin of upper eyelid to above groin (vs dorsolateral fold slightly curved above tympanum), two outer metacarpal tubercles distinctly separated (vs merged at base), tibio-tarsal articulation reaching forward beyond tip of snout (vs around nostril), and nuptial pad creamy white in breeding males (vs gray or gray-brownish); from R. culaiensis: dorsolateral fold extending straight from posterior margin of upper eyelid to above groin (vs dorsolateral fold slightly curved above tympanum), and tibio-tarsal articulation reaching forward beyond tip of snout (vs at nostril); from R. jiemuxiensis: dorsolateral fold extending straight from posterior margin of upper eyelid to above groin (vs dorsolateral fold slightly curved above tympanum), head length significantly larger than head width (vs slightly larger), internarial distances larger than interorbital distances (vs smaller), and two outer metacarpal tubercles distinctly separated (vs merged at base); from R. chaochiaoensis: supratympanic fold absent (vs present), internarial distances larger than interorbital distances (vs smaller), and toe webbing formula I 1⅓ -2 II 1⅓ -2⅓ III 1 ½ -2⅔ IV 3 -1⅓ V (vs I 1 -1⅔ II 1⅓ -2 III 1½ -2½ IV 2⅔ -1 V); from R. japonica: outer metacarpal tubercles present (vs absent), tibio-tarsal articulation reaching forward beyond tip of snout (vs reaching or beyond tip of snout in males, reaching at center of eye or beyond nostril in females), nuptial pad creamy white and divided into three parts (vs nuptial pads grayish brown or yellowish brown and divided into two parts).
From Rana chevronta, which lacks molecular data, the new species can be distinguished by its larger body size, SVL = 48.3-57.8 mm in adult males (vs 39.7-44.0 mm), head length significantly larger than head width (vs almost equal), relative finger lengths I < II < IV < III (vs II < IV < I < III), and nuptial pad creamy white and divided into three parts in breeding males (vs purplish gray and undivided).

Phylogenetic analyses
The ML and BI analyses resulted in essentially identical topologies and are integrated in Figure 2, in which the major nodes are sufficiently supported with the Bayesian posterior probabilities (BPP) >0.95 and the bootstrap supports (BS) for maximum likelihood analysis >85. The pairwise distances based on COI and 16S genes among all samples are given in the Supplementary material, Tables S1 and S2, respectively.
The Rana samples representing the new species are grouped in a distinct and robust monophyletic lineage with high support (BPP = 1.00 and BS = 100) and low divergence (mean 0.3%, ranging 0.0-0.6% in COI, and mean 0.1%, ranging 0.0-0.5% in 16S); they form a separate evolutionary lineage within the R. japonica group. This lineage from Jiangxi and Hunan is close to R. dabieshanensis from Anhui and R. omeimontis from Sichuan. The smallest genetic distance between this lineage and a previously recognized species is 3.4-4.0% in COI (with R. omeimontis) and 1.6-2.0% in 16S (with R. dabieshanensis), which are significant when compared to all other recognized species (e.g. 2.8-3.6% in COI between R. longicrus and R. culaiensis; 1.2-1.3% in 16S between R. dabieshanensis and R. omeimontis).
Forearms 0.19 of SVL and hand 0.26 of SVL; fingers slender, without web but with narrow fringe; tip of fingers rounded, not expanded, without circummarginal grooves; relative finger lengths I < II < IV < III; subarticular tubercles significantly prominent, rounded; distinct, small, rounded supernumerary tubercles below the bases of each finger; inner metacarpal tubercle indistinct, ovoid, partly covered by nuptial pad; two outer metacarpal tubercles distinctly separated, slightly larger, long elliptic; nuptial pad with tiny spines on the finger I, divided into three parts, the basal one around the inner metacarpal tubercle and partly covering it, the largest one from the edge of the basal one to the subarticular tubercle of finger I, the smallest one extending from the edge of the biggest one to the tip of finger I.
Tibia 0.63 of SVL and foot 0.88 of SVL; heels overlapping when hindlimbs flexed at right angles to axis of body; tibio-tarsal articulation reaching forward beyond the tip of snout when hindlimb stretched along the side of the body; relative toe lengths I < II < III < V < IV; toes webbing formula: I 1⅓ -2 II 1⅓ -2⅓ III 1 ½ -2⅔ IV 3 -1⅓ V; absence of lateral fringes on the lateral edges of toes I and V; subarticular tubercles oval and distinct; inner metatarsal tubercle large, ovoid, outer metatarsal tubercle small.
Dorsal skin smooth with sparse tiny granules; several small tubercles on flank; supratympanic fold absent; dorsolateral fold distinct and thin, extending straight from posterior margin of upper eyelid to above groin; several tiny granules on the skin of loreal and temporal regions; ventral surface smooth, large flattened tubercles densely arranged on the rear of thigh and around vent.
Coloration of holotype. In life, dorsal surface yellowish brown with few black spots; black speckles forming a linear stripe between eyelids; dorsolateral fold intermittently edged with black on two sides; loreal region yellowish; temporal region yellowish, slightly tinged with grey; tiny granules on loreal and temporal regions reddish; dorsal forelimbs and hindlimbs reddish with indistinct greenish grey transverse bars. Throat yellowish; chest and belly creamy white; ventral surface of forelimbs and hindlimbs flesh color; nuptial pad creamy white; tubercles around vent yellowish.
In preservative, dorsal surface turns grey with black spots and light grey patches; limbs taupe with brown transverse bars. Ventral surface white, with greyish mottling on throat and belly; ventral surface of limbs beige; hands and toe webs dark grey.
Variations. Measurements of type series specimens are given in Table 2. Coloration of dorsal skin varies from brown to yellowish brown (Fig. 4). Black edges on dorsolateral fold indistinct in all paratypes. SYS a006495 and 6496 with V-shaped mark. The number of transverse bars ranges from two to five on forearms, three or four on thigh, and three to six on tibia.
Distribution and ecology. Currently, Rana jiulingensis sp. nov. is known from Guanshan Nature Reserve in the Jiuling Mountains and Mount Wugong in the Wugong Mountains of northwestern Jiangxi, and Mount Mufu and Mount Dawei in the Mufu Mountains of northeastern Hunan. This suggests that its geographic distribu- tion is the central and northern parts of the Luoxiao Range (Fig. 1). All individuals were found on the surface of paths or on the bush leaves beside paths in subtropical evergreen broadleaved forests. Males SYS a005511 and 5519, which were collected in mid-September, bear a well-developed nuptial pad, while males SYS a006494 and 6495, collected in early August, are without a nuptial pad. This suggests that the breeding season of this species might begin in September.

Discussion
All recognized species of the subgenus Rana from China (except for R. chevronta) are included in our work for morphological and molecular analyses. Four monophyletic clades are supported by high values (BPP = 1.00 and BS > 85, respectively) in the phylogenetic tree. Three of them correspond to the morphologically recognized R. japonica group, R. chensinensis group, and R. amurensis group. The fourth, unnamed monophyletic clade includes R. johnsi, R. sangzhiensis, and R. zhengi. Within this unnamed clade, R. sangzhiensis and R. zhengi cluster together with significant support (BPP = 1.00 and BS = 100) and little divergence (0.0-0.4% in COI and 0.0-0.4% in 16S), which is consistent with the original morphological identification by Zheng et al. (1997). Therefore R. zhengi is considered a synonym of R. sangzhiensis. Furthermore, all species of this clade were morphologically previously assigned to Pseudorana (Fei et al. 2009). Thus, based on the phylogenetic relationships and morphological similarities, this monophyletic clade is proposed as a new species group, the Rana johnsi group. For the remaining species, their exact placements remain unresolved due to the insignificant support. Further study of these species is needed, and new species groups might be proposed for these outcast species.
Within the Rana japonica group, the genetic divergences among three species, R. longicrus, R. zhenhaiensis, and R. culaiensis, are relatively closer than other species. Additionally, the validations of these species have been supported by the morphological examinations (Li et al. 2008;Fei et al. 2009). Anuran frogs are suggested with conservative phenotypes (Cherry et al. 1978). Cryptic species, which are morphologically identical but genetically differentiated, are also common in most species complexes (e.g. Yan et al. 2011;Kuraishi et al. 2013;Xiong et al. 2015;Lyu et al. 2019Lyu et al. , 2020. With remarkable morphological diversity, but relatively smaller genetic differentiation, R. longicrus, R. zhenhaiensis, and R. culaiensis show a special situation. This suggests that an integrative taxonomic approach is especially important in delimitation of anuran species, and that reliance solely on morphological or molecular evidence would be misleading. The discovery of Rana jiulingensis sp. nov. increases the diversity of the genus Rana in the Luoxiao Range to five species (Fig. 1). This situation indicates that the Luoxiao Range has the greatest diversity of Rana species in southern China and may be key to speciation of the genus Rana.