Journal of East African Natural History 99(2): 103–128 (2010) AFRICAN SHREWS ENDEMIC TO THE ALBERTINE RIFT: TWO NEW SPECIES OF MYOSOREX (MAMMALIA: SORICIDAE) FROM BURUNDI AND THE DEMOCRATIC REPUBLIC OF CONGO Julian C. Kerbis Peterhans College of Professional Studies, Roosevelt University, 430 S. Michigan Avenue, Chicago, IL 60605, USA & Department of Zoology, Field Museum of Natural History, Chicago, IL 60605, USA [email protected] Rainer Hutterer Zoologisches Forschungsmuseum Alexander Koenig Adenauerallee 160, 53113 Bonn, Germany [email protected] Jacques Mwanga, Benjamin Ndara Département de Biologie, Centre de Recherche en Sciences Naturelles (CRSN)/Lwiro B.P. 147, Cyangugu, Rwanda [email protected], [email protected] Leif Davenport United States Peace Corps 2040 Antananarivo Place, Dulles, VA 20189, USA [email protected] Innocent Balagizi Karhagomba Organisation pour la Conservation Environnementale au Kivu (D.R. Congo) B.P. 388, Cyangugu, Rwanda [email protected] Jay Udelhoven The Nature Conservancy, Global Marine Team 1917 First Avenue, Seattle, WA 98101, USA [email protected] ABSTRACT The genus Myosorex has a classic relict distribution within sub-Saharan Africa. Montane populations in eastern and western equatorial Africa are separated by ca. 2900 km. Until this study, the closest known populations in southern Africa were separated by nearly 2000 km from the closest populations in the Albertine 104 J. Kerbis Peterhans, R. Hutterer, J. Mwanga, B. Ndara, L. Davenport, I. Karhagomba & J. Udelhoven Rift Valley. Here we document previously unknown populations of Myosorex, representing two new endemic taxa from montane forests adjacent to the Albertine Rift. In conjunction with additional data from Malawi, we fill in major gaps in our knowledge of the biodiversity and distribution of this genus in the areas of the Albertine and Malawi Rift Valleys. We demonstrate that this gap is an artefact of survey effort and collecting serendipity. The two new species described herein, as well as other species of Myosorex from north of the Zambezi River, exhibit limited distributions and are confined to montane habitats, typically above 1000 m. Our new species of Myosorex from Kahuzi- Biega NP (DRC) is the second known species of Myosorex from that park where it is syntopic with Myosorex babaulti. This is the first time that two species of Myosorex co-occur in any forest north of the Zambezi River. This suggests either sympatric speciation or a secondary re-invasion during times of climatic amelioration and forest expansion. The two species described here (Myosorex jejei and Myosorex bururiensis) are associated with two phenetically-defined species groups: the former with a more narrow hexagon- shaped skull, long-tail, and short claws (‘narrow-headed group’) and the latter with a more broad hexagon-shaped skull, short tail and long claws (‘broad- headed group’). Keywords: African shrews; Myosorex; Albertine Rift; endemism; evolution; biogeography; Bururi Forest; Kahuzi-Biega National Park; conservation. INTRODUCTION The documentation and description of African shrews (Soricidae) has lagged behind the discovery and description of most other mammalian species because of the difficulty of their capture and their generally cryptic appearance. Between 1989 and 2009, only 10 species of shrews, but 45 species of rodents and 47 species of primates, were described as new for Africa (including Madagascar; Hoffmann et al., 2009). A review of the distribution of collecting localities in sub Saharan Africa (Schlitter & Delany, 1985) indicated that there have been substantial gaps in the documentation of African mammals. This may lead to incorrect assumptions on species distributions. In this paper we fill in gaps in small mammal collecting localities along the Albertine Rift (figure 1). Combined with our recent report (Kerbis Peterhans et al., 2008) our results suggest that disjunct distributions within the genus Myosorex (Gray, 1838) are an artefact of collecting effort, collecting techniques, collecting serendipity, and of course, disjunct montane distribution. Our results further confirm that north of the Zambezi, the genus Myosorex is confined to relictual montane forest habitats, suggesting that these forest- litter dependent species have limited dispersal abilities. Fortunately, morphological variation within the genus is notable, and, provided comparative materials are at hand, adequate characters are available for review and analysis. The Albertine Rift ranges from the northern end of Lake Albert in the north to the southern tip of Lake Tanganyika (Plumptre et al., 2007). This area is also known as Two new Myosorex from Burundi and Democratic Republic of Congo 105 Figure 1. Map of the Albertine Rift region, with localities of various species of Myosorex as indicated. the ‘Great Lakes’ region of Africa and it is a topographically distinctive feature of the continent. The Albertine Rift is part of a large tectonically active system that separates 106 J. Kerbis Peterhans, R. Hutterer, J. Mwanga, B. Ndara, L. Davenport, I. Karhagomba & J. Udelhoven the east-central part of the continent from the remainder. To the north, this system has already split off the Arabian Peninsula, creating the Red Sea. Tectonic activity has lead to uplift on both sides of the lake-bearing rift depression. Climatic vicissitudes during the Pleistocene and elevational gradients have isolated the uplifted montane forests. Some of these highlands are assumed to have had intermittent connections in the past; current isolation has led to speciation within some of these montane archipelagos (reviewed in Kingdon, 1989). As a consequence, these highlands house Africa’s richest endemic montane faunas of birds (Prigogine, 1985; Collar & Stuart, 1988), butterflies (Carcasson, 1964) and small mammals (Kityo et al., 2003; unpublished data). In this paper we address specimens collected in disjunct forests on opposite sides of the Albertine Rift Valley: Kahuzi-Biega National Park, a World Heritage site in the eastern Democratic Republic of Congo and the tiny and virtually unknown Bururi Forest (16 km2) in southern Burundi. MATERIAL AND METHODS The species reported on here result from two inventory programs conducted by different teams and organizations between 1990 and 2007. Work in Bururi Forest (Burundi) was coordinated by the U.S. Peace Corps/USAID Biodiversity Program in conjunction with the Institut National pour l’Environment et la Conservation de la Nature (INECN). Small mammal and bird surveys were conducted at three locations within the Bururi Forest Reserve, Burundi: Nyagatarugwa Valley (1785 m), 6–11 August 1992 by LD; Ruhinga Hill (2170 m), 9–15 March 1993 by JLU and LD; Mumushwizi Valley (1880 m), 17–25 August 1993 by JLU and LD. The survey of Musisi Swamp and environs was carried out by JCKP and the DR Congo’s Centre de Recherche en Sciences Naturelles team based at Lwiro: JM, BN, Kayeye, Prince Kaleme, Baliwa, Masunga, Katamba and Jeje (Norbert) Balulu during 18–23 July 2005. The survey of an an elevational gradient of Mount Kahuzi (DR Congo) was carried out by JCKP and CRSN (JM, BN, Linjanja) between 18–29 July 2007. The survey began in Mugaba Swamp, continued to the adjacent Cinya Swamp along Mugaba Gate, and proceeded up the slope, crossing the three summits of the mountain. Security issues prevented the field team from camping on the mountain, and resulted in lesser coverage of the higher elevations. Surveys of Kahuzi-Biega National Park, Democratic Republic of Congo (DRC) were coordinated with the scientific staff at Lwiro, Centre de Recherche de Sciences Naturelles (CRSN); the ICCN (L’Institut Congolais pour la Conservation de la Nature) and the Field Museum of Natural History (Chicago). In both instances, terrestrial small mammals (mice and shrews) were collected using a variety of snap traps (Victor Rat and Museum Special), live traps (Shermans) and pitfall lines (PF, described in Voss & Emmons, 1996). The traps were baited with peanut butter and oatmeal and a variety of other foods, including fish and worms. Measurements generally follow Dippenaar (1977) and are illustrated in Stanley et al. (2005a). External dimensions are represented by: total length (Tot), tail vertebrae (Tail), hind foot including claw (HF), fore claw (Fclaw), hind claw (Hclaw), length of Two new Myosorex from Burundi and Democratic Republic of Congo 107 ear from notch (Ea), and weight (Wt). Length of Head and Body (HB) was obtained by subtracting Tail from Tot. Cranial and dental measurements and their abbreviations include: condylo-incisive length (CI), upper tooth row length (UTR), maxillary breadth (MB), greatest width of braincase (GW), height of cranial capsule (HCC), least interorbital width (IO), mandible from tip of incisor to articular condyle (MD), and lower tooth row length (LTR). We also add breadth across the post-glenoid processes (PGL) and the maximum height of the coronoid process of the mandibular ramus (COR). Fore and hind claw measurements were taken along the dorsal aspect of digit III. Drawings of tooth and skull features were made with the aid of a camera lucida attached to a stereomicroscope. The nomenclature of cranial bones and teeth mainly follows Meester (1963), with the exception that upper unicuspid teeth following the first upper incisor are simply called U1 to U4 for convenience. Hutterer (2005b) discussed the homology and naming of these teeth in detail which probably represent incisors, a canine, and a premolar. Myosorex species with a more narrow-skull (GW/CI<0.5), long-tail, and short claws are defined herein as the ‘narrow-headed group’. The other group, with a more broad skull (GW/CI>0.5), short tail and long claws is defined herein as the ‘broad-headed group’. With the exception of the type and only specimen of Myosorex schalleri (Heim de Balsac, 1966), only adult specimens were included in the analyses. Adulthood was determined on the basis of the fusion of the basioccipital-basisphenoid suture. Specimens examined, measured, or figured are listed in the appendix 1. Acronyms for museum collections used in this report are: AMNH, American Museum of Natural History, New York; CMNH, Carnegie Museum of Natural History, Pittsburg; FMNH, Field Museum of Natural History, Chicago; JCK, collector number of Julian C Kerbis Peterhans, housed at FMNH; MNHN, Muséum National d’Histoire Naturelle, Paris; RMCA, Royal Museum for Central Africa, Tervuren; TBR, Tanzanian Belgian Rodent Program, University of Antwerp (specimens now transferred to RMCA); SMNS, Staatliches Museum für Naturkunde, Stuttgart; ZFMK, Zoologisches Forschungsmuseum Alexander Koenig, Bonn; and ZMUC, Zoological Museum and University, Copenhagen. MLWM refers to field numbers for uncatalogued Malawi mammals, Field Museum of Natural History. RESULTS The genus Myosorex is part of the subfamily Myosoricinae (Hutterer, 2005a), along with Congosorex (Heim de Balsac & Lamotte, 1956) and Surdisorex (Thomas, 1906). All species in this subfamily show the myosoricine crown pattern of the lower p4, in which the enamel surface of this tooth forms a reversed letter V (see figure 2), and all display a distinct interparietal bone (figure 3, as discussed by Heim de Balsac & Lamotte, 1956). The three genera were regarded as separate entities since Heim de Balsac (1967), and were redefined by Hutterer et al. (2001). The three species of Surdisorex (reviewed by Kerbis Peterhans et al., 2009) are characterized by the absence of ear pinnae, long front claws, minute eye openings, three upper unicuspids, and the absence of lower p3. Surdisorex is 108 J. Kerbis Peterhans, R. Hutterer, J. Mwanga, B. Ndara, L. Davenport, I. Karhagomba & J. Udelhoven also characterized by the shape of the interparietal bone (figure 3), which is represented by a flattened triangle in all three species. Congosorex (reviewed by Stanley et al., 2005a) has reduced ear pinnae, short front claws, minute eye openings, three upper unicuspids, and a variably occurring lower p3. The interparietal bone is wing-shaped (figure 3d) but closer to Myosorex. The latter genus has normal-sized ear pinnae, short to long front claws, Figure 2. Comparison of the fourth lower premolar (p4) in species of the narrow-skulled group (top row) and the broad-skulled group (bottom row) of Myosorex. Top row from left to right: M. schalleri (MNHN 1981-1107, holotype), M. varius (ZFMK 82.135), M. cafer (ZFMK 2003.357), M. gnoskei (FMNH 191568, holotype), M. eisentrauti (ZFMK 69.373), M. jejei nov. sp. (FMNH 189098, holotype), M. longicaudatus (ZFMK 81.1293); bottom row from left to right: M. zinki (SMNS 4505, holotype), M. cf. geata (TBP 6268), M. kihaulei (ZMUC M 2134), M. okuensis (ZFMK 69.376, holotype), M. bururiensis nov. sp. (FMNH 155923, holotype), M. rumpii (ZFMK 69.375, holotype), M. babaulti (FMNH 160175). Scale bar is 1mm. a larger eye opening, four upper unicuspids, and always a lower p3. The interparietal bone is rather variable in Myosorex, as rising species numbers seem to increase the morphological variability as well. Although all forms of Myosorex are diagnosed by their hexagon-shaped skulls (e.g. Heim de Balsac & Meester, 1971) our reviews of Myosorex, including that of Kerbis Peterhans et al. (2008) and the two taxa described here, enable us to now place artificial groupings into two morphological types: a more narrow-skulled group (but still with a hexagon-shaped skull) represented by M. schalleri, M. gnoskei (Kerbis Peterhans, et al. 2008), and seven other species (table 1), and a more broad-skulled group (with a markedly hexagon-shaped skull), represented by M. blarina (Thomas, 1906), M. babaulti (Heim de Balsac & Lamotte, 1956) and a further six species. These latter members of the genus have a more pronounced hexagonal shape to the braincase as reflected in a wide braincase (GW) relative to the condylo-incisive length (CI). The two new species described below Two new Myosorex from Burundi and Democratic Republic of Congo 109 Figure 3. Comparison of interparietal bones of species of Surdisorex (a-c), Congosorex (d-f), and Myosorex (g-t): a, Surdisorex norae (FMNH 190262); b, S. polulus (FMNH 43846); c, S. schlitteri (FMNH 195069, holotype); d, Congosorex phillipsorum (FMNH 177682); e, C. polli (MRAC 23.234, holotype); f, C. verheyeni (ZFMK 99.932, holotype); g, Myosorex varius (ZFMK 82.135); h, M. cafer (ZFMK 2009.336); i, M. longicaudatus (ZFMK 81.1293); j, M. eisentrauti (ZFMK 69.374); k, M. gnoskei (FMNH 191568, holotype); l, M. jejei nov. sp. (FMNH 189276); m, M. babaulti (ZFMK 68.545); n, M. bururiensis sp.nov. (FMNH 155923, holotype); o, M. rumpii (ZFMK 69.375, holotype); p, M. okuensis (ZFMK 69.388); q, M. cf. geata (TBP 2305); r, M. kihaulei (FMNH 10984); s, M. tenuis (CMNH 69654); t, M. zinki (FMNH 174120). Inset skull shows the position of the interparietal bone. Scale bar is 1mm. 110 J. Kerbis Peterhans, R. Hutterer, J. Mwanga, B. Ndara, L. Davenport, I. Karhagomba & J. Udelhoven are representative of each group. The divisionbetween these groups, however, is arbitrary and is here defined as a ratio of greatest width of braincase to condylo-incisor length (GW/CI) of 0.5 or greater for the broad-skulled group (table 1). This subdivision is also partly reflected by other cranio-dental features, such as the shape (long and narrow versus short and broad) of the lower p4 (figure 2). Based on holotypes, data on tail length (Tail) relative to head and body length (HB) were compiled by Hutterer et al. (2001). Tail length ranges from 41.7%–83.8% of HB within the genus Myosorex. This range also includes values for recently described taxa of Myosorex: M. kihaulei (Stanley & Hutterer, 2000: 52.4%, 8 M; 54.3% 2 F), M. gnoskei (60%, n=13, table 2) as well as one described below as M. jeje (66.5%, n=11, table 2). However, the second species described herein (M. bururiensis, 29.5%) is the only species of Myosorex to possess a tail to body ratio (ca. 30%, n=1) that falls within the range of the other highly fossorial endemic African soricid genera (Congosorex and Surdisorex). These include Congosorex verheyeni (Hutterer, et al. 2002) at 30.2%, Congosorex polli (Heim de Balsac & Lamote, 1956) at 40% (based on holotypes, Stanley, et al. 2005a), Surdisorex norae (Thomas, 1906) at 27.6% (n=19) and Surdisorex polulus (Hollister, 1916), 30.1% (n=11, Kerbis Peterhans et al., 2009). The recent description of the relatively long-tailed Congosorex phillipsorum (Stanley et al. 2005a, holotype=57.9 %) has further eliminated ‘relative tail length’ as a character useful in distinguishing Myosorex from Congosorex. Table 1 Phenetic separation of the species of Myosorex species into more narrow-headed and more broad-headed groups, based on length-width relations of the skull. Note that the new records of M. gnoskei have formalin-erosion damage to the first upper incisor; therefore, only the measurements of the type specimen are presented for the affected measurements. Mean values provided; sample size as indicated in column N. more narrow- N CI GW GW/CI MB MB/GW MB/CI Source headed taxa schalleri* 1 18.9 8.3 .439 5.3 .638 .280 MNHN 1981-1107 varius 13– 22.7 10.8 .476 7.0 .648 .308 Meester & Dippenaar, (Pt St. Johns) 22 1978 cafer 22– 23.3 11.1 .476 7.0 .631 .300 Meester & Dippenaar, 42 1978 gnoskei 5 20.47* 10.36 .484* 6.11 .589 .297* This study, table 3 tenuis 2–3 21.57 10.43 .484 6.15 .590 .285 Roberts, 1951 eisentrauti 5 23.4 11.34 .485 6.34 .559 .271 Heim de Balsac, 1968 sclateri 2–3 25.17 12.27 .487 7.75 .632 .308 Roberts, 1951 longicaudatus 25 22.7 11.25 .496 6.65 .591 .293 Meester & Dippenaar, 1978 jejei nov. sp. 6 20.1 10.02 .498 5.89 .588 .293 This study, table 3 more broad- headed taxa zinki 11 22.8 11.5 .504 6.5 .566 .285 Stanley et al., 2005b geata 6 20.89 10.54 .505 6.44 .611 .308 Stanley & Hutterer, 2000 kihaulei 10 20.26 10.40 .513 6.38 .613 .315 Stanley & Hutterer, 2000 okuensis 10 22.45 11.93 .531 6.72 .563 .299 ZFMK specimens bururiensis* 1 22.5 11.96 .532 7.55 .631 .336 This study, table 3 nov. sp. blarina 4–7 21.99 11.73 .534 7.24 .617 .329 FMNH specimens rumpii 1 23.3 12.6 .541 7.4 .587 .318 ZFMK specimen babaulti 11 21.84 12.01 .550 7.34 .611 .336 FMNH specimens * holotype only Two new Myosorex from Burundi and Democratic Republic of Congo 111 SPECIES DESCRIPTIONS Bururi forest shrew Myosorex bururiensis Kerbis Peterhans & Hutterer, nov. sp. Holotype FMNH 155923, old male with teeth in heavy wear. Alcoholic carcass with skull removed (figure 4). Collected 22 August 1993 in a pitfall trap (PF 2, bucket) by Jay L. Udelhoven; original number JLU 193. Figure 4. Clockwise from top: lateral, ventral and dorsal aspects of the cranium of Myosorex bururiensis nov. sp. (FMNH 155923, holotype). Scale bar is 1cm. 112 J. Kerbis Peterhans, R. Hutterer, J. Mwanga, B. Ndara, L. Davenport, I. Karhagomba & J. Udelhoven s e easuremente single typ Hclaw - 1.9 1.9 1.6 1.6 1.8 1.7 2.0 1.7 1.8 1.9 1.9 1.8 1.6–2.0 2.5 2.3 2.5 2.8 mh cords and ared with t Fclaw - 2.4 2.0 2.1 1.7 2.0 2.0 2.3 2.2 2.1 2.2 2.1 2.1 1.7–2.3 3.3 3.2 3.0 3.15* remp % dditional when co Tail/HB 83.8% 52% 60% 48-75% 60% 60% 58% 56% 78% 92% 69% 71% 54% 59% 75% 67% 56%-9252% 53% 44% 29.5% a er. Note the M. gnoskei, Wt - 8 12 9–13.5 7.3 7.6 7.8 6.3 7 6 7 7.4 7.7 8.5 7 7.2 6.0-8.5 14 12 13 14 papfor 0 0 sidered in this e new metrics HF Ea 12 - 13.5 8.5 13.7 8.4 12–14 7–111 6 13 8 13 9 12 8 13 7 12 10 14 7 11 10 12.5 8 13 9 11 7 12.3 8.1 11–14 6–114 6 14 6 15 8 15 5 nh ot cn Myosorex erences i Tail 44 39 42 35-47 43 42 43 42 46 49 45 45 42 45 44 44.2 42-49 43 40 35 26 pecies of otable diff HB 52.5 75 70 60–86 72 70 74 75 59 53 65 63 78 76 59 67.6 53–78 82 75 80 88 sN me wi. 28 21 of soMala ot 14 12 00–115 12 17 17 05 02 10 08 20 21 03 11.8 02–125 15 15 14 g) u, T- 11111111111111111111 nts (mm, Wt in e Nyika Plateaold. ecies Spschalleri gnoskei gnoskei gnoskei jejei iejei jejei jejei jejei jejei jejei jejei jejei jejei jejei jejei jejei babaulti babaulti babaulti bururiensis ehb Tasuremable 2. External meof d on tM. gnoskei collecteshted in pecimen, are highlig Voucher MNHN 1981.1107 MNH 191568 (type) FMNH, mean, n=13 FMNH range, n=13 FMNH 189098 (type) FMNH 189276 FCK 5307 JCK 5308 JCK 5335 JCK 5347 JCK 5351 JCK 5353 JCK 5393 JCK 5415 JCK 5425 JNH, mean, n=11 FMMNH, range, n=11 FCK 5346 JCK 5350 JCK 5354 JMNH 155923 (type) F