Bitis harenna sp. nov. (Figs. 1 –7, 12; Tables 1–2; Appendices 1–2) Bitis parviocula Böhme, 1977: Largen & Rasmussen (1993: 383–385, in part); Spawls & Branch (1995: 120–121, in part); Largen & Spawls (2010: 615, in part); Wallach et al. (2014: 92, in part); Wittenberg et al. (2015, in part) undescribed species of Bitis: Böhme (1990: 12–13, fig. 4) Holotype. ZMUC R 68255 (Figs. 1–6), Dodola (from maps: 6.98 ° N, 39.18 ° E c. 2,400 m elevation), Oromia Region, Ethiopia; deposited by Sven Joergen Birket-Smith (1920–1983); preserved in ethanol; female. The exact date of collection is unknown, but was possibly 1966 or 1967, when Birket-Smith was in charge of the Addis Ababa Natural History Museum (M.J. Largen, pers. comm.). Diagnosis. A Bitis of the subgenus Macrocerastes (sensu Lenk et al. 1999) that differs from all other species except B. parviocula in lacking horns between the nostrils, in having less than four scales between the rostral and first supralabial, and in having a pale cross bar on the dorsal surface of the head behind the eyes (Figs. 1, 2). Differs from all other species of the subgenus Macrocerastes (including B. parviocula) in having a concavity on the lateral wall of the braincase for the origin of the M. retractor pterygoidei that is limited posteriorly by flange on the parietal, rather than extending onto the prootic (see Groombridge 1980; Fig. 3 / 4); in being mostly blackish dorsally with narrow pale (cream to yellowish) markings (Fig. 1); in lacking a regular, parallel-sided mid-dorsal stripe (and a series of dark, approximately semicircular markings immediately lateral to this stripe; Figs. 1, 2, 6); and in having a black middorsal marking on the head that extends anteriorly to between the nostrils (Figs. 1, 2). Among species of the subgenus Macrocerastes, the new species differs also from B. rhinoceros and B. gabonica in having a mostly dark rather than pale dorsal surface of the head. Unlike B. nasicornis, the new species lacks posteriorly notched scales. Unlike B. parviocula, the new species has facets on the frontals for articulation with the prefrontals that do not meet medially (Fig. 3), and it also lacks a subhorizontal ridge on the lateral surface of the prootic overhanging the foramen for the mandibular branch of the trigeminal nerve (Fig. 4). See Remarks section for additional possible differences between the new species and its most similar congener, B. parviocula. Description of holotype. Some morphometric and meristic data presented in Tables 1 and 2 and Appendix 2. Condition generally good, some scales rucked and/or abraded; specimen somewhat macerated in parts; 90 mm long approximately midventral incision extending back from approximately 150 mm behind snout tip, with approximately 10 mm 2 midventral section of body wall excised; outer layer of many scales lost. A large, relatively slender Bitis. Widest point slightly anterior to midbody; body tapers gently over anterior one-third of SVL, much more abruptly over posteriormost one-eighth of SVL. Tail short, very slender. Head posteriorly much wider than anteriormost end of body (‘neck’). Distance between snout tip and posterior end of retroarticular process 43.7 mm (longitudinal distance between level of these two points 42.5 mm). Eyes surrounded by 12,15 circumorbital scales; separated from supralabials by 2 to 3 scales; pupil vertically eliptical. Left and right circumorbital series separated by at least 10 interorbital scales. Nostrils broadly oval (anteroventrally-posterodorsally oriented). Supranasal shields overhang nasals substantially, separated middorsally minimally by 2 scales, separated from circumorbitals minimally by 2 scales. Nasals and supranasals separated from rostral by 3,3 scales (> 3 mm), from first supralabial by 2,3 scales (2.5 + mm). Rostral much wider (5 mm) than high (2.1 mm at midline). Most dorsal and lateral head scales between and behind eyes keeled except supralabials and adjacent row posterior to 9 th or 10 th supralabial; no keels on scales of lower jaw or chin. Posteriormost left and right supralabials separated by at least 31 scales. Supralabials 14,14. Infralabials 15,17, last of which especially small on each side; first pair contact at midline, second to fifth contact genials. Mental triangular, wider (5.1 mm) than long (3.3 mm). First pair of genials largest, contact at midline; second pair of genials much smaller and separated by one pair of smaller gulars; first pair of genials separated from first preventral by four gulars. First ventral separated from last infralabial by 8,7 scales. Inside of mouth without pigment (except for at least tips of tongue, dark). Teeth mostly hidden in gingivae. Left fang intact, approximately 11.6 mm long. Three preventrals (four discrete scales distributed over the length of the equivalent of three ventrals), 145 ventrals; ventrals mostly with orthoplicate posterior edges, notably convex in posteriormost ventral. Umbilical notch on posterior margin of 10 th to 13 th ventral scale anterior to anal scale. Anal shield single, strongly convex posterior margin; subcaudals 20,19 (excluding unpaired terminal scute), without keels; first subcaudal on left divided. Immediately behind posterior end of lower jaw (level with sixth ventral) 43 dorsal scale rows, reducing to 36 rows ten ventrals further posteriorly. Numbers of dorsal scale rows thereafter fluctuate, most stable between approximately 50 th and 110 th ventral (35–40 rows, see Appendix 2). Dorsal rows reduced to 30 by 120 th ventral; 28 rows at vent. Dorsal scale rows characterized by splits/duplications/divisions on upper part of flanks, fusions on lower part of flanks, beginning (anteriorly) by at least level of 15 th ventral. Pattern of divisions, fusions produces posteroventrally oriented ‘lines’ (posteriorly diverging dorsal scale rows) accentuated visually by longitudinal keels on scales (Figs. 5, 6). Scale row formula for 60 th to 80 th ventral as follows: Midvertebral scales variably single (vertebral) or double (paravertebral), for example single from level of 60 th to 72 nd ventral and from 74 th to 80 th, but double at level of 73 rd ventral (Fig. 5). Twentyseven (left) and 25 (right) vertebral scales aligned with the 60 th– 80 th ventral scales (Table 2). Approximately 17 dorsal scale rows on base of tail at level of anteriormost subcaudal. Keels on body scales moderate, most prominent on dorsum, less prominent on lower than upper flanks, absent on first dorsal scale row; where prominent, keels extend to distal tip of scales; posterior margins of scales entire (not notched). Difficult to determine presence or absence of apical pits; seemingly present in at least some dorsal body scales, paired, close to scale tips. All scales on tail keeled except for subcaudals. Colour in preservative essentially two tone (Fig. 1), with the dark varying from black to dark brown and the pale being ivory (off- or cream-white). Upper surface of head satin-black with two diverging ivory streaks united at the rostral, extending on to the nasal scute, contacting the upper edge of the eye, caudad to the throat and terminating at the retroarticular process. Approximately midway along length of head, the diverging ivory streaks are connected by a narrow (one scale wide), widely-open, V-shaped pale narrow band, broken at midline by 1 or 2 dark scales. Posteriorly a second pale, narrow band extends for five scales slightly obliquely posteriorly towards midline, then turns posteriorly and unites its antimere on the dorsum behind the head. The dark region on the dorsal surface of the head can be seen as a black (narrowly subtransversely bisected) arrowhead, 21 scales wide at its widest point, and reaching to within 3 scales of the rostral anteriorly. At its widest, the dark area bears a small middorsal spot across two scales. Laterally, a preorbital band, narrow but its width spread across two scales, extends from the dorsal streak, immediately anterior to the eye, to the posterior half of the 4 th supralabial. A narrow postorbital bar (less than one scale wide) extends from the posterior margin of the eye to the 10 th supralabial. Supralabials mostly with pale lower edges. Underside of head entirely pale except for dark anterior and posterior margins of anteriormost infralabials and few flecks posterolaterally, close to posteriormost infralabials. Upper and lateral surfaces of body mostly dark with pale markings; ventral surface mostly pale with dark markings (Figs. 1, 2, 5). Dorsal body pattern comprises a chain of 21 black sub-lozenge-shaped saddles with approximate maxima of eight scales long by 12 (rarely 14) scales wide. Lateral margins of dark saddles demarcated by narrow (generally one scale wide), irregular pale lines, often with small breaks. The flank pattern comprises broad dark patches that approximately alternate with the dorsal saddles, separated from each other by narrow pale, more or less vertical bars, 1 scale increasing ventrally to 2–3 wide. Within these dark flank markings are small, irregular pale marks. On the mid to upper flanks these pale markings are either isolated flecks or connected to form an inverted V; approximately aligned with these on the lower flanks are larger, wider deltoid pale marks issuing from the belly. Most scales in first dorsal row are pale, such that pale deltoid marks on lower flans are often continuous (ventrally) with vertical bars. The belly has scattered series of irregular spots roughly in three ranks, the innermost two brownish on each ventral plate, the outermost more black, singly or more frequently paired longitudinally. The tail has the same general colour pattern as the body (dark dorsally, pale ventrally), though the dorsal markings are less regular. Observation of live specimen. On 15 October, 2013, one of us (E.R.B.) observed a live specimen of B. harenna sp. nov. in the Bale Mountains National Park. The observations were made with colleagues Mark Chynoweth, James Kuria Ndung’u, Sisay Sayfu, Abdu Ibrahim and Khalifa Ali – the latter four of whom have worked in the Bale Mountains for at least three years without seeing this species before. The observation was made while driving along the road between the settlements of Rira and Delo Mena, as it passes through the Harenna Forest at approximately 2,140 m elevation (6.694525 ° N, 39.728576 ° E). This section of road is within a swath of forest that has openly spaced canopy trees (approximately 10 % cover) and a very dense understory of shrubs and vines. The area is approximately 1 km from an illegal settlement (“Aboyea”) within the Bale Mountains National Park. The snake was on a large roadside verge under bright but partly cloudy skies at 12.30 pm. As a result of recent road repair the verge was formed by a large amount of earth and stone (approximately 2 m wide on either side of the road) that had been moved mechanically. This section of road requires regular maintenance in the wet season thus creating fairly regular disturbance for a distance of c. 3 km. Upon sighting, the driver Sisay Sayfu stopped the vehicle within 3 m of the snake, which was apparently basking on the verge and did not move for a few minutes while being observed and photographed. The animal was matt black dorsally, more glossy along the lower flanks. The pale markings were distinctly yellowish. From memory the animal was approximately 1m in total length. It moved slowly off the side of the road and into undergrowth when the observers left the vehicle for closer inspection. The photographs (one shown in Fig. 7) show an animal clearly very similar to the holotype of B. harenna sp. nov. The colour pattern is the same with minor differences, including a more transverse (not posteromedially deflected) pale bar dorsally on the head behind the eyes. The tail is relatively longer than in the holotype, such that the live specimen is likely male. The animal has a general overall similarity (the ‘jizz’ of birdwatchers) to the puff adder Bitis arietans. Remarks. The holotype of Bitis harenna sp. nov. was identified previously as B. parviocula (Largen & Rasmussen 1993, Largen & Spawls 2010) and was used by Wittenberg et al. (2015) to score morphological characters for that species in their phylogenetic analyses. The skulls of ZMUC R 68255 (B. harenna sp. nov.) and R 68254 and ZFMK 63067 (B. parviocula), as analysed with HRXCT data, differ in several respects (Figs. 3, 4). Groombridge (1980: 140) reported a derived condition (present in the subgenus Macrocerastes species B. nasicornis and especially B. gabonica [including its then subjective junior synonym B. rhinoceros]) in which the concavity on the lateral wall of the braincase for the origin of the M. retractor pterygoidei is not limited posteriorly by a flange on the posterior edge of the parietal but instead extends onto the prootic. ZMUC R 68255 exhibits the plesiomorphic condition, with a limiting posterior parietal flange (Fig. 4 D) and R 68254 (Fig. 4 E) and ZFMK 63067 (Fig. 4 F) a moderate manifestation of the derived condition, somewhat more like that in B. nasicornis (Fig. 4 B) than in B. gabonica (Fig. 4 C). The condition in B. parviocula was not previously reported. In the B. parviocula specimens ZFMK 63067 (Fig. 4 F) and ZMUC R 68254 (B. parviocula: Fig. 4 E) the prootic forms a strong ridge overhanging the foramina for the mandibular branch of the trigeminal nerve, whereas this ridge is absent in R 68255 (B. harenna sp. nov.: Fig. 4 D). A strong ridge here is present also in the dried skulls of B. gabonica and B. arietans examined in this study, but much smaller or absent in B. nasicornis. The facets on the anterodorsal edges of the frontals for the medial process of the prefrontals are clearly separated at the midline in the holotype of the new species (ZMUC R 69255: Fig. 3 A) but continuous in the B. parviocula specimens ZMUC R 68254 (Fig. 3 D) and ZFMK 63067 (Fig. 3 G). The B. harenna sp. nov. condition for this character is also present in the skeletal material of B. arietans, B. gabonica, and B. nasicornis examined here. Böhme (1990) suggested that the fangs of B. harenna sp. nov. are relatively longer than those of B. parviocula, but we do not find that to be the case. Although there are very small differences in numbers of teeth between ZMUC R 68255 and 68254, making accurate and precise tooth counts from HRXCT data is not easy, and counts for ZFMK 63067 suggest there are not clear interspecific differences (Table 1). The holotype (ZFMK 16803) and the only other known museum specimens (ZFMK 60367, ZMUC R 68254) of Bitis parviocula are shown in Figs. 8–10. Although the skull of ZMUC R 68254 is partly smashed (Fig. 3), the proportions of the head of this B. parviocula specimen are nonetheless interpreted as being genuinely different to (more slender than) those of ZMUC R 68255. The distance between the snout tip and eye in the holotype of the new species (ZMUC R 68255) is 25.7 % of the head length but this value in the ZMUC B. parviocula specimen is 32.1 % - both specimens were measured on the left side of the head, the left mandible being largely intact in both. This proportion in the less damaged head of ZFMK 63067 is 28.2 %, and is approximately 28.4 in the smashed (Fig. 8) head of the holotype ZFMK 16803. Although a small individual of B. parviocula (from S.S.’s memory approximately 25 cm long) has the appearance of being relatively slim (Fig. 11 top left panel; Dobiey & Vogel 2007: 116; Nečas et al. 1993 a: fig. 1; Nečas 1997: 59), all photographs of recently wild-caught, larger and presumably adult individuals (examples shown in Fig. 11) show heavily bodied, stout animals approximately similar to B. nasicornis, B. rhinoceros and B. gabonica, whereas the holotype of B. harenna sp. nov. and conspecific animal observed live in the Harenna Forest appear somewhat more slender, contributing (along with an apparently more slender, posteriorly elongate head) to its more puff-adder-like form. Bitis parviocula is clearly quite variable in colour and colour pattern (Fig. 11; Dobiey & Vogel 2007) but no specimens approach the distinctive appearance of B. harenna. The scales of ZMUC R 68255 (B. harenna sp. nov.) are slightly different in texture to those of R 68254 (B. parviocula), being less horny, but this might be attributable to differences in preservation. The dorsal scalation pattern of ZMUC R 68254 (B. parviocula) is generally similar to the holotype of B. harenna sp. nov., though the fluctuations in the formula for the 60 th to 80 th ventral are even more pronounced: The midvertebral scales in R 68254 are also more variable: single from the level of 60 th to 73 rd ventral and from 74 th to 77 th, but double at level of 73 rd and of 78 th to 80 th ventral. Compared with the holotype of B. harenna sp. nov., R 68254 has more (30–31 versus 25–27: Table 2) vertebral scales aligned with the 60 th– 80 th ventrals, and the posteroventral ‘lines’ of dorsal scales on the lower flanks are more steeply inclined (Fig. 5). The range of variation in all characters noted in this paper for B. harenna sp. nov. and B. parviocula can be expected to expand (with unknown magnitude) as more specimens are examined, such that characterization of these two species will need revision in future. A map of the known distribution of the three species of Ethiopian Bitis is presented in Fig. 12). Böhme (1977) named B. parviocula for its seemingly small eyes, and this morphological feature was interpreted as an indication that the species was somewhat fossorial. The eyes of the type (and at that time only known) specimen of B. parviocula are small by artefact—this specimen was found dead and squashed on the road, and live specimens of the species do not have notably small eyes (Spawls 1994). Etymology. Named for the Harenna forest and escarpment, in the Bale Mountains National Park, where a live specimen of the new species was recently sighted. For nomenclatural purposes the specific epithet is considered a noun in apposition. Suggested common name. Bale Mountains adder (English). Conservation status. We obviously know very little about Bitis harenna sp. nov. and thus its conservation status is likely to be Data Deficient based on IUCN Red List criteria. However, any large snake, especially a probably venomous one, is likely to face persecution in Ethiopia. We are unable to support Nečas et al. ’s (1993 b) comments about B. parviocula being revered and protected by local folklore, and all three specimens of this species in museum collections have been damaged and were likely killed (or at least smashed post mortem) by local people. The region as a whole is under great pressure from agriculture and increasing urbanization, exacerbated by very high human population growth. The only known precise locali, Published as part of Gower, David J., Wade, Edward O. Z., Spawls, Stephen, Böhme, Wolfgang, Buechley, Evan R., Sykes, Daniel & Colston, Timothy J., 2016, A new large species of Bitis Gray, 1842 (Serpentes: Viperidae) from the Bale Mountains of Ethiopia, pp. 41-63 in Zootaxa 4093 (1) on pages 43-55, DOI: 10.11646/zootaxa.4093.1.3, http://zenodo.org/record/270951