New species of reed frog from the Congo basin with discussion of paraphyly in Cinnamon-belly reed frogs Author Schick, Susanne Author Kielgast, Jos Author Rödder, Dennis Author Muchai, Vincent Author Burger, Marius Author Lötters, Stefan text Zootaxa 2010 2501 23 36 journal article 10.5281/zenodo.195830 82cdbb30-e80e-48fd-8eb1-83bd91302fc8 1175-5326 195830 Hyperolius veithi sp. nov. ( Figs. 2 A, 4A) Holotype . ZFMK 89607 (field number A519), adult male, collected by J. Kielgast at a flooded area in the middle of primary forest away from rivers and streams in Salonga National Park ( 02.88 S , 20.41 E , ca. 415 m above sea level), Province of Bandundu, Equateur Kasaï Oriental and Occidental, Democratic Republic of Congo , 24 January 2008 . Paratypes . Fifty-eight adult specimens of both sexes ( ZFMK 89608 89645, ZMUC R771393 R771412) from the same locality; data as the holotype but collected 24 26 January 2008 . Diagnosis. A sexually monochromatic species of Hyperolius based on the following characters: pupil horizontal; tips of toes and fingers broadened; last phalanx of fingers out of alignment; fingers not arranged in opposing pairs; gular flap in males ( Schiøtz 1999 ; Channing & Howell 2006 ). It is defined by: (1) SVL males 25.7 ± 1.7 mm (22.7 28.4 mm , N = 32), females 33.9 ± 1.3 mm (31.7 36.5 mm , N = 27); (2) TIBL/SVL in males 0.54 ± 0.02 (0.52 0.61, N = 32) and in females 0.51 ± 0.02 (0.48 0.55, N = 27), HW/SVL in males 0.33 ± 0.02 (0.3 0.37, N = 32) and in females 0.33 ± 0.01 (0.29 0.35, N = 27); (3) dorsal surface smooth to coarse with few scattered pointed warts, most prominent and tubercular below eye; (4) snout shape dorsally and laterally rounded, nares visible from above; (5) E N/EYE in males 0.78 ± 0.09 (0.53 0.95, N = 32) and in females 0.85 ± 0.09 (0.71 0.99, N = 27), canthus rostralis convex from tip of snout to nostril and straight or slightly concave from nostril to eye; (6) tympanum covered by thick skin but with annulus tympanicus still visible; (7) FOOT/TIBL in males 0.67 ± 0.04 (0.57 0.75, N = 32) and in females 0.75 ± 0.05 (0.67 0.86, N = 27); (8) foot webbing formula 1(1), 2i (1) 2e(½), 3i (1 1½) 3e(½ 1), 4i (1 2) 4e(1 1½), 5(1 ½); (9) no sexual dichromatism; both sexes dorsally tan, olive brown to dark brown, usually lighter towards the sides and dorsally with more or less intense, diffuse dark markings; there is always a bright yellow dorsolateral line from tip of snout (lines from both sides of the body meet here) to sacral region which is commonly bordered by dark brown; it is broader posterior to eye and often contains tiny dark brown spots which may continue posterior into groin and onto hind limbs; always a cream to yellow spot on the heel; ventral sides are white or cream; a red marking on the inner femur is lacking; the gular flap of males is yellow; the iris is bronze to dark brown; (10) for DNA barcodes access GenBank (http://www.ncbi.nlm.nih.gov): GU443988 (male holotype , ca. 530 bp), GU443975 (female paratype , ZFMK 89629, ca. 480 bp). By being brownish and having a light dorsolateral lines, H. veithi males and females ( Fig. 2 A) are similar in colour and pattern to numerous Hyperolius species (see Schiøtz 1999 ), including PhJ H. cinnamomeoventris ( Fig. 2 C D) and its suggested synonyms H. fimbriolatus Buchholz & Peters, 1876 and H. ituriensis Laurent, 1943 ( Lötters et al. 2001 , Frost 2009 ). Hyperolius schoutedeni from the Congo basin deserves special attention, as this reed frog, like H. veithi , is also sexually monochromatic. The new species can be distinguished from any other Hyperolius species which are brownish and have a light dorsolateral lines (including H. cinnamomeoventris and H. schoutedeni ) as follows: (i) dorsolateral lines relatively short and broad ending in the sacral region (versus in the groin in other species); (ii) dorsolateral lines in life are bright yellow (versus whitish to tan in other species); (iii) dorsolateral lines often contain dark spots (absent in most other species); (iv) presence of a light spot on the heel, cream to yellow in life (absent in other species); (v) absence of a in life red marking on the inner femur (present at least in H. cinnamomeoventris and its suggested synonyms and occasionally in H. schoutedeni ; Schiøtz 2006 ). In addition to (i) to (v), in H. schoutedeni , a mid-dorsal line is common (absent in H. veithi ) and the new species is larger than H. schoutedeni and develops less foot webbing. Five individuals (three males, two females from N'Sele and Garamba, DRC ; Schiøtz 2006 ) and the female holotype of H. schoutedeni ( Fig. 4 B) have maximum SVL in males 22.9 (minimum SVL in H. veithi 22.7 mm ) and in females 25.2 mm (minimum SVL in H. veithi 31.7 mm ); holotype foot webbing formula is 1(1), 2i (1) 2e(½), 3i (1½) 3e(1), 4i (2) 4e(1½), 5(1) (for H. veithi see above) (note that measurements and foot webbing data for H. schoutedeni differ slightly from those given by Schiøtz 2006 ). Several additional Hyperolius species have been described from DRC ( Schiøtz 1999 ; Frost 2009 ) but remain little known as they are based on one to a few specimens only and poor original descriptions. Of these, H. atrigularis Laurent, 1941 , H. kibarae Laurent, 1957 and H. polli are similar to H. veithi in colour and pattern in preservative, but none has continuous bright dorsolateral lines (based on holotype examinations and Laurent 1941 , 1957 ). In terms of uncorrected p-distances of the 16S mitochondrial rRNA, H. veithi is little differentiated genetically from some, but not all populations assigned to H. cinnamomeoventris ( Table 2 ). We reject conspecifity, however, as H. veithi syntopically occurs with H. cinnamomeoventris (uncorrected p-distance 2.8 4.4 %, but see comment to Table 2 ) and amplexus was observed 27 times between H. veithi individuals but never between H. veithi and H. cinnamomeoventris . Additional arguments include presence of monochromatism in the new species versus dichromatism in H. cinnamomeoventris ( Fig. 2 A versus Fig. 2 B D) and H. veithi having white eggs ( Fig. 2 A) versus the half-pigmented eggs in H. cinnamomeoventris ( Lötters et al. 2004 ) . The genetically related H. molleri and H. thomensis ( Table 2 ) are dorsally brownish to greenish, lack dorsolateral lines (present in the new species) and possess a more coarse skin texture ( Fig. 2 A versus Fig. 2 E F). Description. The type series contains 32 males and 27 females . In external morphology, mature sexes differ in size ( Table 3 ) and the presence of gular flap and vocal sac in males only. Body slender with sacrum width about one fifth of SVL; head short and broad with distance from tip of snout to posterior corner of eye about one sixth of SVL; HW about one third of SVL; snout tip in dorsal and lateral views rounded; nostril dorsolateral, protruding and visible from above; choanae rounded; maxillary teeth present, vomerine teeth lacking; tongue about as long as wide, free for half of its length and anteriorly bifurcated; canthus rostralis convex from nostril to tip of snout, straight and longer from nostril to anterior corner of eye; EYE> E N; loreal area barely concave; tympanic membrane covered by thick skin but tympanic annulus visible, TYMP <EYE. Male gular flap relatively small compared to other Hyperolius species (see Schiøtz 1999 ). Tibia long, about half of SVL, tibiotarsal articulation extending to loreal region when hind limb adpressed forward along body; FOOT <TIBL; relative length of toes: 1 <2 <3 <5 <4; metatarsal tubercles ill-defined, rounded, inner somewhat larger in size and more prominent than outer, rest of sole smooth with distinct subarticular tubercles present at joints of phalanges of all toes; tips of toes broadened; foot webbing formula 1(1), 2i (1) 2e(½), 3i (1 1½) 3e(½ 1), 4i (1 2) 4e(1 1½), 5(1 ½). Relative length of fingers: 1 <2 <4 <3; palmar tubercle ill-defined, rounded, thenar tubercle less ill-defined than palmar tubercle, ovoid; rest of palm wrinkled with well visible subarticular tubercles at joints of phalanges of all fingers; tips of fingers broadened, tips of fingers 3 and 4 more than twice the width of finger base; hand webbing absent. Skin of dorsal surfaces smooth to coarse with few scattered pointed warts, most prominent and tubercular below eye, skin of ventral surfaces wrinkled. For measurements and proportions see Table 3 . In life ( Fig. 2 A), both sexes are dorsally tan, olive brown to dark brown, usually lighter towards the sides and dorsally with more or less intense, diffuse dark markings. A yellow dorsolateral line always runs from the Hyperolius cinnamomeoventris and related reed frogs studied in this paper ( Fig. 3 ; Table 1). For meaning of #, ## and E see FIGURE 2. Life aspects of: (A) Hyperolius veithi sp. nov. in amplexus (unidentified paratypes; photo J. Kielgast), note that there is a yellow spot on the heel visible in the male frog; (B) H. cinnamomeoventris male in PhJ from the type locality (AC 3008; photo A. Channing); (C) H. cinnamomeoventris female from the Kakamega Forest, Kenya (not collected; photo S. Lötters); (D) H. cinnamomeoventris male in PhJ from Semliki, Uganda (SL 555; photo A. Channing); (E) H. molleri from São Tomé (photo D. Lin, CAS) and (F) H. thomensis from São Tomé in amplexus (photo D. Lin, CAS). TABLE 2. Uncorrected p-distances of the 16S mitochondrial rRNA gene (ca. 530 bp sequences) of populations of Table 1. Distinct difference of the two H. veithi sp. nov. samples against the other samples is because the sequence of GU443975 is ca. 480 bp only (versus ca. 530 bp in GU443988 ). Sample, locality (GenBank accession number) 1 2 3 4 5 6 7 8 9 10 11 12 13 1 H. cinnamomeoventris E, Kenya : Kakamega - Forest ( AY323925 ) 2 H. cinnamomeoventris E, Uganda : Bundibugyo 1.6 - ( GU443994 ) 3 H. cinnamomeoventris E, Uganda : Semliki 2.2 1.2 - ( GU443993 ) 4 H. cinnamomeoventris E, Uganda : Kampala 0.2 1.5 2.2 - ( GU443996 ) 5 H. cinnamomeoventris E, Uganda : Bwindi 1.6 1.1 1.6 1.5 - ( AY603985 ) 6 H. cinnamomeoventris E, Uganda : Kibale 0.6 1.8 2.6 0.4 1.9 - ( GU443995 ) 7 H. cinnamomeoventris E, DRC : Salonga 3.4 3.0 3.4 3.4 4.0 3.8 - National Park ( GU443998 ) 8 H., cinnamomeoventris E, CAR : Dzanga- 4.4 4.0 4.2 4.3 3.6 4.7 2.1 - Sangha Reserve ( GU443997 ) 9 H. cinnamomeoventris #, Gabon : Lamabréné 8.4 8.5 8.1 8.3 8.6 8.7 7.5 7.6 - ( FJ594077 ) 10 H. cinnamomeoventris ##, Angola : Kalandula 3.8 3.8 3.4 3.8 4.2 4.2 4.2 3.8 5.6 - ( GU443990 ) 11 H. molleri , São Tomé ( AY603990 ) 6.0 5.7 5.6 6.1 5.1 6.5 6.2 6.2 6.6 4.7 - 12 H. thomensis , São Tomé ( AY603991 ) 7.0 6.4 6.2 6.8 5.9 5.8 6.7 6.8 7.2 4.9 1.4 - 13 H. veithi sp. nov. , DRC : Salonga National 9.4 5.2 8.5 5.4 4.6 5.8 4.4 8.7 8.6 3.8 6.2 6.4 - Park ( GU443988 ) 14 H. veithi sp. nov. , DRC : Salonga National 4.2 3.5 2.9 3.9 3.3 4.4 2.8 3.5 7.3 2.2 4.9 4.8 0.0 Park ( GU443975 ) TABLE 3. Measurements (mm) and proportions of Hyperolius veithi sp. nov. The mean is followed by the standard deviation and the range in parentheses. For abbreviations see text.

SVL	Females (N = 27) 33.9 ± 1.3 (31.71–36.5)	Males (N = 32) 25.72 ± 1.72 (22.69–28.35)
TIBL	17.4 ± 0.73 (15.55–18.49)	13.97 ± 0.98 (12.15–16.06)
HW	11.16 ± 0.42 (10.23–12.12)	8.56 ± 0.60 (7.62–10.06)
E–N	3.43 ± 0.27 (2.98–4.05)	2.64 ± 0.22 (1.99–2.96)
EYE	3.91 ± 0.39 (3.33–4.45)	3.4 ± 0.31 (2.64–4)
TYMP	Invisible	Invisible
FOOT	13.0 ± 0.75 (11.42–15.02)	9.31 ± 0.9 (7.08–10.92)
TIBL/SVL	0.51 ± 0.02 (0.48–0.55)	0.54 ± 0.02 (0.52–0.61)
HW/SVL	0.33 ± 0.01 (0.29–0.35)	0.33 ± 0.02 (0.31–0.37)
E–N/EYE	0.84 ± 0.09 (0.71–0.99)	0.78 ±0.09 (0.53–0.95)
FOOT/TIBL	0.75 ± 0.05 (0.67–0.86)	0.67 ± 0.04 (0.57–0.75)

tip of snout (lines from both sides of the body meet here) to the sacral region; this line is commonly bordered by dark brown colour and is broader posterior to eye where it often contains tiny dark brown spots. Spots may continue posterior to the dorsolateral lines into groin and onto hind limbs. The heel always has a cream to yellow spot. Ventral sides are white or cream; a red marking on inner femur is lacking; the gular flap of males is yellow. The iris is bronze to dark brown. In preservative, colours fade as follows: tan and brown to greyish, yellow to cream; dark brown, cream and white remain. In characters with intraspecific variation, male holotype conditions are as follows: gular flap and vocal sac present; foot webbing formula 1(1), 2i (1) 2e(½), 3i (1) 3e(½), 4i /e(1), 5(½); SVL 26.3 mm , TIBL 13.9 mm , HW 9.2 mm , E N 2.4 mm , EYE 3.7 mm , TYMP 1.4 mm , FOOT 10.6 mm . For individual aspects of colour in preservative and pattern see Figure 4 A; individual information on life colour was not recorded. For molecular genetic studies, some muscle tissue had been taken from the ventral side of the left femur. Distribution and natural history. The new species is only known from Salonga National Park in the central Congo basin, DRC ( Fig. 1 ). The habitat is lowland rainforest (ca. 400 m above sea level) with seasonal rainfall (short dry period in February, long dry period in June August); annual precipitation is ca. 1,865 mm and monthly mean temperature ca. 25.0 °C (data obtained from WorldClim; Hijmans et al. 2005 ). Due to limited sampling efforts in the Congo basin and an expected lack of distribution barriers to Hyperolius veithi , we expect this species to show a geographically wider range than currently known. Twenty-seven amplectant pairs and five single males were found at the end of the rainy season (January) at night on vegetation, ca. 1 m above water in a flooded but apparently permanent pond deep within primary forest. Frogs were not heard calling and some pairs were observed laying small clutches of whitish eggs on vegetation above water ( Fig. 2 A). It is expected that tadpoles (unknown) develop in water. Syntopic Hyperolius species included H. cinnamomeoventris , H. cf. platyceps and another undescribed species, not closely related to H. cinnamomeoventris and H. veithi . Conservation status. Due to the limited knowledge on both the geographic range encompassed by this species and population trends, we consider it Data Deficient when applying IUCN Red List categories and criteria ( IUCN 2008 ). Due to the apparent lack of physical borders to dispersal and the observation that numerous other amphibian species in the Congo basin show relatively large distributions, we expected that Hyperolius veithi occupies a larger geographic range than currently known in the central Congo basin (see above). If so, it may perhaps later well rank under Least Concern. Also, we do not consider chytridiomycosis, an emerging infectious disease causing amphibian decline in other regions of the planet, a potential threat to H. veithi . The central Congo basin shows limited suitability for the emergence of this fungal pathogen ( Rödder et al. 2009 ) and elsewhere in tropical Africa, Hyperolius species well survive despite high prevalence and individual parasite load ( Kielgast et al. 2010 ). Etymology. The specific name is a patronym for Michael Veith, acknowledging his support of amphibian research in tropical Africa. Paraphyly of Cinnamon-belly reed frogs. As shown in Figure 3 , Hyperolius cinnamomeoventris is paraphyletic when H. molleri , H. thomensis and H. veithi are considered distinct species. In addition to H. veithi (see above), we concur with Drewes & Wilkinson (2004) that specific distinctness of the sexually monochromatic H. molleri and H. thomensis is warranted. Although both these reed frogs may show similar colour and pattern ( Fig. 2 E F) to PhF H. cinnamomeoventris ( Fig. 2 C) (which is more applicable to H. molleri ), they possess a more coarse skin texture, lack (in life) a red marking on the inner femur (present in H. cinnamomeoventris ) and have parts of the toe and fingers tips bright orange (not different to dorsal or ventral colours in H. cinnamomeoventris ). Hyperolius thomensis is distinguished further by being larger in body size and having a black, white and orange marbled venter in life (cream in H. cinnamomeoventris ), while that of H. molleri is white or red in life and thus can be similar to that of H. cinnamomeoventris . When comparing uncorrected p-distances of H. molleri and H. thomensis with other reed frogs discussed in this paper, divergence is ≥ 4.7 % and ≥ 4.9 %, respectively ( Table 2 ). An operable threshold for the consideration of putative species (to be confirmed by an integrative approach) in the 16S mitochondrial rRNA gene for anuran amphibians is at about 3 % ( Fouquet et al. 2007 , Vieites et al. 2009 ) but can even be lower between well accepted species (e.g. Zimkus & Schick 2010 ). Applying this threshold to reed frogs studied here, specific distinctness of both H. molleri and H. thomensis against H. cinnamomeoventris is supported. Likewise, divergence of H. veithi and H. cinnamomeoventris is ≥ 4.4 % (or ≥ 2.2 % in the shorter H. veithi sequence, but see comment to Table 2 ). In addition, specific distinctness is still supported by morphology (colour and pattern in life and monochromatism versus dichromatism) and by having white versus half-pigmented eggs (see above). FIGURE 3. Bayesian phylogram of Hyperolius species including Cinnamon-belly reed frogs inferred from nucleotide sequence data from 16S mitochondrial rRNA. Bayesian posterior probabilities> 0.95 each are marked by an asterisk on branch. We here apply species names as given in Table 1; in the Cinnamon-belly reed frog clade, numbers in parentheses give sample size of the same haplotype, which corresponds with localities. Note that Hyperolius cinnamomeoventris is paraphyletic. Topotypic material of H. olivaceus (#) and H. cinnamomeoventris (##) are indicated. Hyperolius sp. ‘Salonga’ is H. veithi sp. nov. Both the Hierarchical Likelihood Ratio Tests and Akaike Information Criterion implemented in MrModeltest selected a GTR+I+G model with a gamma distribution of 0.5912 and a proportion of invariable sites of 0.2291 (estimated base frequencies: A: 0.3294, C: 0.2260, G: 0.1813, T: 0.2633; rate matrix: A-C: 3.0708, A-G: 7.4382, A-T: 5.3616, C-G: 1.5864, C-T: 21.7283, G-T: 1.0000). FIGURE 4. Dorsal and ventral views of preserved (A) male holotype of Hyperolius veithi sp. nov. (photos by F. Feß), SVL 26.3 mm, (B) female holotype of Hyperolius schoutedeni (photos by MRAC, through the courtesy of D. Meirte), SVL 26.4 mm. As clades currently assigned to H. cinnamomeoventris are generally well supported, in a purely cladistic view, different names should be applied. Considering the available names in the synonymy of H. cinnamomeoventris ( Frost 2009 ) , the Gabon clade in Figure 3 could be referred to as H. olivaceus Buchholz & Peters, 1876 with H. fimbriolatus treated as a synonym ( Lötters et al. 2001 ), since this genetic sample was obtained at Lambaréné which is the type locality of both these names (Table 1). Likewise, the Angola clade ( Fig. 3 ) is based on samples from Kalandula which is the type locality of H. cinnamomeoventris and its synonym H. tristis Bocage, 1866 (Table 1). Therefore, this clade could be recognized as H. cinnamomeoventris sensu stricto . All samples from the eastern range of Cinnamon-belly reed frogs constitute a monophyletic group, here called Eastern Clade ( Figs. 1 , 3 ). Of the available names in the synonymy of H. cinnamomeoventris , two originate from the eastern portion of the Cinnamon-belly reed frog distribution, H. ituriensis Laurent, 1943 and H. wittei Laurent, 1947 ( Frost 2009 ) . Consequently, one of them could potentially be applicable to the Eastern Clade, apart from the fact that the type locality of H. wittei lays out of the suggested geographic range of Cinnamon-belly reed frogs, at all ( Fig. 1 ). In spite of this a priori conclusive scenario, we momentarily refrain from taxonomic action for two reasons. First, delimitating and phenotypic diagnosing putative species remains a problem. Not only is the material available limited for comparisons (see Appendix) and call and other natural history data are sparse or lacking ( Lötters et al. 2004 ) so that an integrative approach combining data from different fields to is hampered. It is also true that molecular data demonstrate considerable variation within putative species and likewise overlap between clades ( Table 2 ). Second, with regard to tree-based taxonomy, it has to be considered that only a fragment of one mitochondrial gene and relatively few samples have been involved in our analysis. Although, the 16S mitochondrial rRNA gene is useful for DNA barcoding and tree-based species recognition in anuran amphibians ( Vences et al. 2005 , Fouquet et al. 2007 , Vieites et al. 2009 ), it has been shown that when analysing mitochondrial DNA sequences, paraphyly does not always uncover distinct taxa. There are various explanations for paraphyly or polyphyly within species when constructing mitochondrial gene trees, in particular introgression, hybrid speciation and (although less of a concern for mitochondrial than for nuclear loci) incomplete lineage sorting ( Funk & Omland 2003 ). In these cases some alleles in a species may appear more close to those of another (related) species than to conspecifics. As a result, intraspecific variation is larger than variation at the species level and may lead to a mistaken consideration of variants for species when following the monophyly criterion ( Funk & Omland 2003 ). Intensive sampling and analyzing phylogeographic data including related species are helpful for an appropriate interpretation of paraphyly. There is on-going debate regarding the reliability of mitochondrial gene trees for species delimitation. For example, McKay & Zink (2010) recently found that paraphyly in birds most commonly results from incorrect taxonomy and concluded that mitochondrial gene trees are rarely misleading. With regard to H. cinnamomeoventris , we conclude that additional sampling and studies are necessary for allowing warranted taxonomic action.