A new brevicipitid species (Brevicipitidae: Callulina) from the fragmented forests of the Taita Hills, Kenya Author Loader, Simon P. Author Measey, John Author De, Rafael O. Author Malonza, Patrick K. text Zootaxa 2009 2123 55 68 journal article 10.5281/zenodo.188158 c2416f9c-8e07-4c45-91f1-26965f7a07f5 1175-5326 188158 Callulina dawida sp. nov. ( Figures 2–5 ) Holotype . NMK A/4267 (Field No. JM01239), adult male, collected by J. Mwandoe & T. Mwakio, on 16 December 2004 , in Kiangungu forest patch near Iyale forest, 37M 426787, UTM 9623928. Paratypes . NMK A 3913 (MW03197) collected by G. J. Measey on 7 May 2002 , in Ngangao forest), A 3534/3 (NK091), A 3534/2 (NK090), and A 3534/1 (NK089) collected by R. Odhiambo on 10 February 1998 , in Ngangao forest, A 1499/1 collected by H. J. Beentje and J. Lorubun on 10 May 1985 in Ngangao forest, A 3703 ( CHA 2) collected by P. K. Malonza, V. Muchai and R. Mwakondi on December 1999 , in Chawia forest, A 4266 (JM01235) collected by B. A. Bwong and G. J. Measey on 16 December 2004 , in Fururu forest, A 4268/2 (JM01632) and A 4268/1 (JM01628) collected by G. J. Measey and J. Mwandoe 28 December 2004 in Ngangao forest, A 3535 (NK592) collected by R. Odhiambo on 22 March 1998 , in Chawia forest. All the following specimens were collected by P. M. Odongo: BMNH 2005.1574 (formerly NMK A 3647/ 4- FNK998) and BMNH 2005.1575 (formerly NMK A 3647/3 – FNK1206) collected between 4-8 June 1998 in Kenya , Taita-Taveta district, Taita Hills, Ndiwenyi forest, A 3613/2 (FNK906) collected on 29 May 1998 , in Fururu forest, A 3647/5 (FNK956) and A3647/6 (FNK 1205) collected on 4 June 1998 in Ndiwenyi forest, A 3613/1 (FNK817) collected on 21 May 1998 , in Fururu forest, A3647/1 (FNK106) and A 3647/2 (FNK 1207) collected on 8 June 1998 in Ndiwenyi forest, A 3619 (FNK826) collected on 25 May 1998 , in Ndiwenyi forest, and A 3617(FNK957) collected on 1 June 1998 , in Fururu forest. FIGURE 2. Dorsal, ventral and lateral views of the holotype of Callulina dawida (NMK A/4267). Diagnosis . The new species is assigned to the genus Callulina based on the following characteristics: Truncated or expanded terminal phalanges (simple in Spelaeophyrne, Probreviceps , Breviceps , and Balebreviceps ); single posterior denticulated row in the palate of Callulina (two denticulated rows in Probrevicpes , and a glandular mass in Breviceps ); choanae present (absent in Balebreviceps ); moderately sized wedge shaped lobes on the mentomeckelian elements, posteroventrally directed (in Probreviceps , Balebreviceps , and Breviceps variously reduced and enlarged, Largen and Drewes, 1989 ); cultriform process of the parasphenoid with broad based but narrow alary processes, tapering laterally (cultriform process of the parasphenoids in known brevicipitids show wide diversity, Largen and Drewes, 1989 ); nasals almost meet at midline (broadly separated in Breviceps and Balebreviceps ); clavicle well-developed and straight though slightly curved anteriorly at the point of contact of the coracoid and scapulae (clavicle straight in Breviceps , Probreviceps , and Spelaeophryne ) (see Figure 3a ); omosternum large (rudimentary or small in Breviceps , Probreviceps , and moderate in Balebreviceps , Figure 3a ); tympanum present and usually well-differentiated (absent in Balebreviceps and Probreviceps uluguruensis ); double condylar articulation between the urostyle and the sacral vertebra (fused in Balebreviceps , Breviceps , and Probreviceps , Figure 3 b). FIGURE 3. a) Urostyle elements of Callulina dawida (A3535), Abbreviations, uro= Urostyle, sa= Sacrum, att= ilialurostyle muscle attachment, il= ilium. b) Pectoral elements of Callulina dawida (A3535). Abbreviations, omo= Omosternum, clav= Clavicle, cor= Coracoid, procor= Procorocoid, ster= Sternum. Callulina dawida is a slightly smaller, robust shaped Callulina that is morphologically distinct from the two other species in this genus ( C. kreffti and C. kisiwamsitu ). The new species is distinguished by the following characters: truncate finger tips, rounded at edges, and only slightly expanded (if at all) beyond the width of the subarticular tubercle (finger tips expanded beyond the width of subarticular tubercle in C . kisiwamsitu and C. kreffti ; see Figure 4 ); terminal phalanges Y-shaped but not expanded beyond the width of the base of phalange. Callulina kisiwamsitu and C. kreffti have terminal phalanges expanded distally beyond the width of the base of the phalange, C. kreffti has T-shaped terminal phalanges and some individuals exhibit an intermediate ‘T’ and ‘Y’ shape (e.g. CAS162505). Remarks on diagnosis. Tubercles on the hands and feet were treated as diagnostic for species in Callulina ; de Sá et al . (2004) stated that in C. kisiwamsitu “there is no contact between inner and outer metatarsal”, however an examination of additional specimens indicates that this character is more variable than previously considered. Because inner, mid and outer tubercles seem to be quite variable in their size and position, and definition they cannot consistently discriminate between these species. Given that tubercles can also be desiccated, they can be difficult to discern precisely. We suggest that this character needs to be further examined, among and within populations of Callulina to determine its systematic utility in the genus. The new species shows the inner and outer tubercles separated on the hand by a mid palmar tubercle ( Figure 4 ) that is also present in other Callulina species. FIGURE 4. Ventral close up of hands (1mm scale) (a) Callulina dawida (NMK A/4267) (b) C. kreffti (2000.196) (c) C. kisiwamsitu (BM 2002.45). FIGURE 5. Callulina dawida in life. Description of the Holotype . Body stout, head wider than longer. Snout truncate in lateral view ( Figure 2 ), snout tip extending anteriorly beyond the jaws. Snout tip flattened not pointed at apex, with slightly rounded edges. Lower jaw has small warts, concentrated on chin and lateral sides near to tympanum, less dense on upper jaw. Tongue rounded. Canthus rostralis rounded, loreal region sloping slightly concavely. Nostrils rounded, directed laterally, nearer to the tip of snout than the eye. Interorbital space flat, larger than width of eyelid. Pupil horizontal. Tympanum distinct and ovoid, transversely smaller than longitudinally. Tympanum defined by smooth, light coloured skin with warts around edge of the disc. Tympanic ridge is prominent, marked by a darker brown colouration, starts at posterior corner of eye to the arm insertion. Warts on dorsal surface of head and eyelids small, rounded. Dorsal and ventral surfaces of body granular, with rounded, shallow warts; lateral warts slightly larger. Forelimbs slender, covered with small warts that are larger and more densely concentrated at the arm joint than on ventral surfaces. Hind limbs stout and also covered with small warts, which are larger and concentrated at the leg joint. Digits of hands moderately long, relatively slender. Tip of digits truncate, slightly rounded at the edges, but not expanded beyond the width of the first subarticular tubercle. First, second, and fourth fingers of similar length, first being the shortest, third finger longest. Toes are truncate with slightly rounded edges, first smallest in length, followed by second, third, and fifth, with fourth toe being the longest. An evident fold of skin defines the dorsal junction between the penultimate and ultimate phalanges in hands and feet, marked as a narrow white coloured band. No webbing present on either hands or feet. Subarticular tubercles are distinct and rounded. Palmar tubercles are distinct and well-separated, inner and outer metatarsal tubercles distinct, not contacting each other and separated by a medial tubercle, inner one larger. Vent is ventro-posteriorly positioned. Measurements . SUL = 29.6; TL = 10.4; ED = 3.0; TD = 1.6; ETD = 1.4; ND = 1.9; NED =2.5; JW = 8.9; LF3 = 3.8; LT4 = 5.2, TSL = 6.8; HL = 8.9; NLD =1.4; WDF3 = 0.9; WDTF3 = 0.9; IOD = 4.7. Variation. The tympanum is usually distinct in Callulina dawida . However, in a small number of specimens (NMK A3534/3, A3534/1, A 3534/2, A 4268/1), the tympanum poorly demarcated. In this case, the tympanum was distinguished either by smoother skin or by dissection of the tympanic region to verify the presence of the tympanum. Variation in the size, shape and presence/absence of the tympanum is a character often utilized in brevicipitine systematics (e.g. Parker, 1931; de Sá et al. 2004 ; Loader et al. 2006 ), though caution is needed because the granular skin of brevicipitids can obscure its presence. The width of the distal phalange in Callulina dawida is less expanded than other Callulina species currently recognized, with the truncate end being almost always equal to the width at the first subarticular tubercle. However, in some specimens the finger-tips are slightly more expanded than the width of the tubercle (BMNH 2005.1574, A4268, and NMK A3647/1), but never reaching the degree of expansion found in either C. kreffti or C kisiwamsitu . The size and condition of hand and foot tubercles vary slightly between specimens, with some being slightly desiccated, such as specimen NMK A3534/3 where tubercles are less clear. Sexual dimorphism. There is some size morphological difference between the sexes, mean SVL males = 27.1 (n = 9: 21.2—31.3) whereas mean SVL females = 29.1) (n = 9: 24.1—38.3). Three specimens were immature and not sexed or analysed. Females are larger than males, though not significantly (t-test, p = 0.5962). The position and size of the tympanum between male and females differ. Males have a significantly larger tympanum, e.g. tympanum-snout urostyle (t-test => 0.001); and the position of the eye to the tympanum (t-test => 0.001) is also greater in males. Besides these differences, we did not identify any other significant differences in body proportions. Males often appear lighter or more brightly coloured than females. Colouration. In life, the dorsal colour is quite variable, ranging from shades of light yellow through orange, brown to dark brown. In darker individuals, the flanks, hind- and fore quarters are usually lighter, sometimes with the tubercles having white tips ( Figure 3 ). The ventral surface of body is pale and occasionally spotted. In preservative, the dorsal surface is dark tan brown with lighter patches towards the lateral sides of mid-body and head. A distinct, broad, and posteriorly directed light brown interorbital band emanates from both eyelids, meeting at centre of cranium. Almost continuous with the interorbital patterning, a light brown dorsolateral stripe continues from the eyelid along the sides to the hind limb insertion. The brown dorsolateral stripe darkens slightly posteriorly. Limbs are lighter in colour to dark brown dorsum, with their ventral side a light brown beige colour. The underside of the chin has distinctive brown markings. Dark brown colouration encroaches on the middle of throat region, resembling two triangles with the apex of each point meeting at the centre of the throat. The patterning varied between specimens, perhaps influenced by preservation, with some specimens showing a strong interorbital band and dorso-lateral stripe, and others with less clear markings. Advertisement Call. We made a detailed temporal analysis of 1.5 minutes of calling from a single male recorded on 15 September 2008 in Wundanyi forest. This period comprised 10 calls, each call being made up of a mean of 7 "chirps" (range 5 to 9; std dev 1.15) with a mean duration of 0.06 seconds (range 0.05 to 0.07; std dev 0.007), and a mean interval of 0.17 seconds between chirps (range 0.14 to 0.28; std dev 0.03), three such chirps are shown in Fig. 6 . The interval between calls had a mean of 7.21 seconds (range 0.75 to 15.11, std dev 4.82). Peak dominant frequency of Callulina dawida is at 1.6 KHz, with a notable harmonic at 3.2 KHz. The call and harmonic fall below 3.5 KHz ( Figure 6 ), whereas C. kisiwamsitu is always below 2 KHz, and peak dominant frequency in C. kreffti is always above 2 KHz but below 3 KHz (de Sá et al , 2004 ). The call can best be described as a fast repeated "brrr brrr brrr...". Etymology. The specific name dawida is a noun in apposition. It is derived from Dawida , the language spoken by the inhabitants of Taita Hills where this species is found. The local Ki-dawida name king’ombe refers to the similarity in the movement of the animal to that of a cow (Ng’ombe = Ki-swahili for Cow). The common name Taita Warty Frog is given to reflect the restricted distribution of this species to Dawida and Mbololo blocks of Taita Hills, Kenya . Morphometrics Analysis. We conducted a morphometric analysis on 101 Callulina specimens: Callulina kreffti (57 individuals), C. kisiwamsitu (23 individuals), and C. dawida (21 individuals). Using an ANOVA analysis, we tested three sets (each species) for all measures including ratios using snout-urostyle length, to see if there were any statistically significant differences. Statistical significance was shown in the following characters: snout-urostyle length ratios, distal phalange width (p = ≥ 0.001), infraorbital distance (p = ≥ 0.001), and jaw width (p = ≥ 0.001) ( Table 1 ). The degree of digital expansion is identified as a key diagnostic feature for discriminating Callulina species. Furthermore, the robust shape of the head is shown to be statistically different in the new species from other species, as indicated by the significant larger jaw width ( Table 1 ). The infraorbital distance is also shown to be significantly different between species. The size of the tympanum is largest in the new species, although not statistically significant. All other morphological characters examined were not shown to be statistically significant, reflecting the generally similar shape and morphology of all species of Callulina . TABLE 1. ANOVA analysis on select morphological characters in Callulina . See text for abbreviations. Group 1 — C. kreffti , Group 2— C. kiswamsitu , and Group 3— C. dawida . SUL/JAW WIDTH Group N Mean Min Max Variance St. dev 1 57 0.264 0.228 0.359 0.000490 0.022144 2 23 0.252 0.223 0.278 0.000228 0.015089 3 21 0.322 0.274 0.392 0.000928 0.030458 ANOVA Current effect (group): F (2, 40)= 76.223, p= 0.00000

SUL/DISTAL PHLANGE WIDTH
Group	N Mean Min	Max	Variance	St. dev
1	57 0.0454 0.0340	0.0580	0.000036	0.006038
2	23 0.0345 0.0314	0.0415	0.000005	0.002340
3	21 0.0338 0.0283	0.0388	0.000009	0.002924
ANOVA	Current effect (group): F (2, 40)= 76.923, p= 0.00000

SUL/INTERORBITAL
Group	N Mean Min	Max	Variance	St. dev
1	57 0.1988 0.1604	0.2463	0.000306	0.017496
2	23 0.1819 0.1643	0.2037	0.000088	0.009377
3	21 0.1780 0.1587	0.2157	0.000207	0.014374
ANOVA	Current effect (group): F (2, 38)= 25.508, p= 0.00000

Phylogenetic Analyses. An alignment of sequences was gathered to test two specific hypotheses, 1) that the new species forms a clade with Callulina species and not other brevicipitids, and 2) Callulina dawida was sister group to a clade including Callulina kreffti and Callulina kisiwamisitu . To test these hypotheses an alignment of eleven DNA-sequences were assembled. This resulted in a matrix of 1129 unambiguously aligned characters, of which 725 were constant, 110 variable, and 294 parsimony informative. Hoplophryne and Phrynomantis , two microhylids, were used as outgroups. An exhaustive search option using parsimony yielded one best tree (863 steps; Fig. 7a ). Maximum Likelihood analyses were also conducted (using a heuristic search option using 10 random addition sequence replicates and TBR swapping method under a GTR + G model as suggested by Modeltest 3.04, Posada and Crandall, 1998). The tree resembles the parsimony tree ( Fig. 7 b), only differing in the position of Spelaeophryne – shown to have an ambiguous phylogenetic position in previous analyses (Loader et al. 2004). Support for clades was measured with bootstrap proportions (Felsenstein, 1985, 10,000 pseudoreplicates). In all analyses, Callulina species form a well-supported clade – as demonstrated by high bootstrap values. Within this grouping, the geographically close Usambara species C. kreffti and C. kiswamsitu form a clade – albeit weakly supported – with C. dawida a sister group to this clade. Overall the analyses demonstrate the genetic distinctiveness of the three Callulina species. FIGURE 6. Sonogram of Callulina dawida . FIGURE 7. a) Maximum likelihood tree showing branch lengths. (- ln likelihood = 5576.01891). Nucleotide frequencies A= 0.3306 C= 0.2645 G= 0.1413 T= 0.2636, Nst=6, Rmat= 1.0000 2.2193 1.0000 1.0000 and 4.7793, Rates=gamma, Shape= 0.7896, Pinvar= 0.3855. Numbers below branches refer to bootstrap support. b) Maximum Parsimony exhaustive tree showing branch lengths. Numbers below branches refer to bootstrap support. Natural History. During the day this species can be found buried in deep and wet leaf litter or associated with decomposing logs. At night, individuals are encountered moving through the leaf litter, where males have been seen to call. A minority of specimens were found low down on branches in trees. Distribution and Conservation Status The new species is presently only known from remnant forest patches in the Taita Hills, Kenya . Specimens were recorded from Mbololo and Dawida blocks (Mbololo, Ngangao, Chawia, Fururu, Vuria, Ndiwenyi, Mwachora and Boma-Wundanyi forests; Figure 2 ). Based on surveys conducted in the Taita Hills, we calculated the spatial distribution of Callulina dawida . The estimated extent of occurrence of C. dawida is equal to 168.2 km 2 and the estimated area of occupancy is about 4.3 km 2. These are respectively the area included in the polygon (minimum convex hull polygon) obtained by linking the localities where presence of the species was recorded and the area of eight forest fragments in the Taita Hills where C. dawida is known to occur. Based on these spatial estimates we applied the criteria set out by the IUCN for assessing the conservation status of species. The area of occupancy, consisting of eight severely fragmented locations, was estimated to be less than 10 km 2. In addition, some of the forest fragments in which this species occurs are under severe pressure from an increasing local population who utilise forest products (both cutting sticks and collecting dead wood). Therefore, according to the IUCN criteria, the species conservation status is critically endangered (CR B1a, b (ii, iii)). Positive steps are being made to conserve the indigenous forest of the Taita Hills. Several eucalyptus and pine plantations in the area have been earmarked for conversion back to indigenous forest, and it is hoped that these measures will increase the suitable habitats available for many of the endemic species of the Taita Hills.