JU Bed) Pane AA tunen Bi lt milan Fe ited nh. ee A ce Sie u Bi ec ET nn ut aha EEE an ag eg nn en met a ee o~ A De (USEUM A. KOENIG BONN | Es od. ey. ST ? 4 BONNER ZOOLOGISCHE MONOGRAPHIEN Die Serie wird vom Zoologischen Forschungsinstitut und Museum Alexander Koenig herausgegeben und bringt Originalarbeiten, die fiir eine Unterbringung in den ,,Bonner zoologischen Beiträgen“ zu lang sind und eine Veröffentlichung als Monographie rechtfertigen. Anfragen beziiglich der Vorlage von Manuskripten sind an die Schriftleitung zu richten; Bestellungen und Tauschangebote bitte an die Bibliothek des Instituts. This series of monographs, published by the Alexander Koenig Research Institute and Museum of Zoology, has been established for original contributions too long for inclu- sion in „Bonner zoologische Beiträge“. Correspondence concerning manuscripts for publication should be addressed to the editor. Purchase orders and requests for exchange please address to the library of the institute. L’Institut de Recherches Zoologiques et Muséum Alexander Koenig a établi cette série de monographies pour pouvoir publier des travaux zoologiques trop longs pour étre inclus dans les „Bonner zoologische Beiträge“. Toute correspondance concernante des manuscrits pour cette série doit étre adressée a l'éditeur. Commandes et demandes pour échanges adresser a la bibliotheque de l'insti- CUE Saved: BONNER ZOOLOGISCHE MONOGRAPHIEN, Nr. 46, 2000 Preis 100,- DM Schriftleitung/Editor: G. Rheinwald Zoologisches Forschungsinstitut und Museum Alexander Koenig Adenauerallee 150-164, D-53113 Bonn, Germany Druck: JFeCARTHAUS, Bonn ISBN 3-925382-50-X ISSN 0302-671 X ISOLATED VERTEBRATE COMMUNITIES IN THE TROPICS Proceedings of the 4"" International Symposium of Zoologisches Forschungsinstitut und Museum A. Koenig, Bonn, May 13 - 17, 1999 BONNER ZOOLOGISCHE MONOGRAPHIEN, Nr. 46 2000 Herausgeber: ZOOLOGISCHES FORSCHUNGSINSTITUT UND MUSEUM A. KOENIG BONN Die Deutsche Bibliothek - CIP-Eiheitsaufnahme Isolated vertebrate communities in the tropics : Bonn, May 13-17, 1999 / Hrsg.: Zoologisches Forschungsinstitut und Museum A. Koenig, Bonn. - Bonn : | Zoologisches Forschungsinst. und Museum A. Koenig, 2000 (Bonner zoologische Monographien ; Nr. 46) (Proceedings of the ... international symposium of Zoologisches Forschungsinstitut und Museum A. Koenig ; 4) | ISBN 3-925382-50-X Das Zustandekommen und die Durchführung des Symposium wurde gefördert durch Zuwendungen der Deutschen Forschungsgemeinschaft Deutschen Lufthansa AG Deutschen Telekom und der Stadt Bonn Contents INTRODUCTION Isolated vertebrate communities in the tropics . RID EEELT OTOL Se ie a Bee SNORT RN RN NIE DR EN RE 7 NEOTROPICAL REGION Analyzing species composition in fragments Paes POLAR SOURCE War ee nen ee ee 9 Chronology and mode of speciation in the Andean avifauna Bea Moreno OG) Bieldsar.. 2... ks LE Re al es ee Se a 25 Biogeography, geographical variation and taxonomy of the Neotropical hummingbird genus Polyerata Heine, 1863 (Aves: Trochilidae) Ast WON W Gish Rae LGRA 5 ae Or hl ORES REE a tr a 47 The endemic genus Orestias Valenciennes, 1839 (Teleostei: Cyprinodontidae) of the Chilean Altiplano emvOciG ULsSienknecnt &A: Bussen. 2.2....2.2....22.2. 2.222. 38 AFROTROPICAL REGION West Africa Diversity of asnake community in a Guinean rain forest (Reptilia, Serpentes) Wi DONE ee ee ee 69 The distribution of Chromaphyosemion Radda, 1971 (Teleostei: Cyprinodonti- formes) on the coastal plains of West and Central Africa IL DRRENDERT ee ee A 79 The ALSCO Cameroon herpetological expedition 1998: the sampling of a mountain rainforest LEW elleremann,W Bohme& BP. Herrmann.........2.......2....... I) New record of Grauer’s Blind Snake Rhinotyphlops graueri (Sternfeld, 1912) (Reptilia: Serpentes: Typhlopidae) in western Uganda Mmmesiapeia.P.'Necas & D. Modry .............. con. 105 Developmental plasticity and behavioural adaptations of two West African anurans living in an unpredictable environment (Amphibia, Anura) Mes ADDEN a N A EEE 109 Central and East Africa Flora and vegetation of the afromontane region in Central and East Africa 2: FISCHOTE Sees A a N era gD a eee 124 Systematics and distribution patterns of Afrotropical Canaries (Serinus-species group, Aves, Passeriformes, Carduelidae) R. van den. Elzen nn... en POE ee ean 183 Patchy versus continuous distribution patterns in the African rain forest: the problem of the Anomaluridae (Mammalia: Rodentia) A.C. Schunke & R: Hutterer 2... Vee ee en ee 145 Non-insect arthropods (Isopoda, Arachnida and Myriapoda) on the high mountains of tropical Africa P. Beron..... en nn a tase Be se ae Oe oe ee 158 Two sympatric Lygodactylus-species in coastal areas of Eastern Africa (Reptilia, Gekkonidae) Bo ROU 2... 2 ne een ol a 2 189 Comparative ecology and morphology of snipes (Aves, Scolopacidae) in Africa C.M. Gichuki & NN, Gichuki.. „2... 2a. ee ss. 199 Review of karyological studies and the problems of systematics of Ethiopian Arvicanthis Lesson, 1842 (Rodentia: Muridae) M.I. Baskevich & LA, Lavrenchenko ...2....2..2......0 eee 209 Isolation of bird populations in the North Nandi Forest, western Kenya H.'Schifter.2.-. ae ne 217 The mammals of the isolated Harenna forest (southern Ethiopia): structure and history of the fauna L.A..Lavrenchenko: ...... 2. 2.2. 2... 0.00 2 22 Ecology of Otomys barbouri Lawrence & Loveridge, 1953 (Mammalia, Rodentia): an endemic of the afro-alpine zone of Mt. Elgon/East Africa V. Clausnitzer i Oe Oe a ns ee rr YDS The allozymic phylogeny: evidence for coherent adaptive patterns of speciation in ethiopian endemic rodents from an isolated montane massif L.A. Lavrenchenko, A.N. Milishnikov & A.A. Warshaysky 2.22 se 245 Primates in Eritrea — distribution and habitat D. Zinner,;E. Torkler & F. Pelaez 00 ee eee 255 Islands east of Africa Effects of fragmentation and assessing minimum viable populations of lemurs in Madagascar J.U. Ganzhorn, St.M. Goodman, J.-B.Ramanamanjato, J. Ralison, D. Rakotondravony & B. Rakotosamimanana re 265 Priority areas for forest bird and primate conservation in Madagascar — do they match up? 1, SLC IWIERLES can soe igre tite ne ee eis cin ZI Conservation planning in the Mantady-Zahamena corridor, Madagascar: Rapid Assessment Program (RAP) 1s NNIEIEN Sve sia eae GB RR” a OS SG cana ers UR ARE a ee PRSEURGE 285 Fragmentation effects on reptile and amphibian diversity in the littoral forest of south-eastern Madagascar EN EL CL ILE BILCULU | LUO sr ote ot Sy en ee oaay ae faite SE Male ws er sce ne 297], Using fossils and phylogenies to understand evolution of reptile communities on islands EY, AUTO Sr Be a Be N a ara 309 Bird communities on the Comoro islands I LEBE Sa 58 re lr sh Rat ete i eg Zine ee ee ster ASOT Re EN 325 The reptile fauna of the Sogotra archipelago DOT a N ee N ie ke Ne le wes 32) i ORIENTAL REGION Preliminary chromosomal results of Niviventer Marshall, 1976 (Mammalia, Rodentia, Muridae) from the Dalat plateau in southern Vietnam HOES ICN UG IL aa G Vee UEZTLCDS OW. 3 ea i a eee 35) || Mammals of coastal islands of Vietnam: zoogeographical and ecological aspects Sie 5 UDI ZOE SOW? BROT LOE oon te aera cE CaO tte”) area a aa ee a Sal PALEARTIC REGION On the herpetofauna of the Sultanate of Oman, with comments on the relationship between the Afrotropical and Saharo-Sindian fauna ws UV RUIEOS) Ce TESTE ITH DOT tise Chee re Resta nee 367 The evolution of Microtia Freudenthal, 1976 (Mammalia, Rodentia), an endemic genus from the neogene of the Gargano island, southern Italy 2. LAG CRRA te Fo aS soe ON CAAT ON Ce sR RE en eee 381 mb hRACTS Arboreal squamates in tropical forests of Vietnam: species richness, natural history and conservation MTCC SINE GOTLON eared. a EN Men soe he eae Na ey De eee Ot 390 Does frugivore diversity influence seed dispersal and seedling establishment of fruiting plants? PONCE Mee ONMUDSGOACSC newer ee se ee es 390 The Ichthyological Section of the ZFMK: Topics of research in biodiversity K. Busse, J. Freyhof, A. Nolte, R. Sonnenberg & I. Steinmann .......... SM Influence of tree species and habitat structure on the canopy beetle fauna in Semliki Forest, Uganda M "Cousin & T.-Wagner '. ... u.a. ne aR apes Sie ele 392 Geckos milk honeydew produced by Cicada ın Madagascar M. Follinge & W. Böhme '.. 2.2... ....0 ee ses 392 Morphological age-dimorphism and its eco-ethological consequences illustrated for the White-tailed Hawk (Buteo albicaudatus) A.::Gamauf en ne ee Sa BC ea ae 393 Centers of endemic terrestrial vertebrates in Colombia: a methodological approach N.A. Guzman: Ballesteros 3.) 2. 2.00... 2 20 2 394 Mammalian biogeography of the Japanese Archipelago and rodent communities size structure V.:Millien-Parra (eee. 8 Oo i a eee 394 The real biodiversity is inside D. Modry, J.R. Slapeta, P. Necas & B. Koudela =. 4) 2 35 395 The amphibia of an isolated archipelago: The Eastern African Arc forests, Tanzania A. schidtz... „2 2... ne a ee 396 The Alcolapia species-flock (Teleostei: Cichlidae) of Lakes Natron and Magadi, Kenya and Tanzania: another piece in the puzzle of cichlid evolution? R. Sonnenberg, L. Seegers &R. Yamamoto ... ..... 2.0 0 396 Comparative adaptive behaviour of Derby Eland in the zoo of the Wildlife College, Garoua and the Benoue National Park Zoo C. Tangwing-Njeke on .:........ ee 397 Phylogeny of the anuran genus Mantella from Madagascar M.. Vences, F. Glaw, A. Hille & W. Böhme... 2.2 3 eee 398 Distribution patterns and specificity of Chrysomelids (Coleoptera) in central African forests 1. Wagner... 2... Os ee Re sen Lessons on evolutionary processes from an unwanted experiment in a tropical fish community F. Witte & C.D.N. Barel (i foe a 00 > INTRODUCTION Isolated Vertebrate Communities in the Tropics Goetz Rheinwald Since 1984, at intervals of five years, the Alexander Koenig Research Institute and Museum of Zoology hosts symposia on zoological problems in the tropics with a focus on Africa. The background has been described in the Proceedings volume of the first symposium (Schuchmann 1985). The first two were organized by sections of the vertebrate department and were entitled “African vertebrates: systematics, phylogeny and evolutionary ecology” (1984) and “International symposium on vertebrate biogeography and systematics in the tropics” (1989). The entomological department was responsible for the third: “International symposium on biodiversity and systematics” (1994). All Proceedings volumes are still available (see felchences): This fourth symposium in 1999 was again under the auspices of the vertebrate department and was entitled: “Isolated Vertebrate Communities in the Tropics”. It was mainly organized by Dr. Renate van den Elzen. It was the intention of the scientific staff of the department to focus on the evolutionary aspects that have led to the extreme species diversity and ecological variety so typical of the tropical biota. Since isolation 1s thought to be the main or even the sole precondition for the evolution of diversity, the symposium should focus on this aspect. The second idea was that evolution never takes place in individuals or isolated populations, but always in communities. So we arrived at the title of the symposium and of this volume. We invited several speakers whose recent publications showed that they are actively working in our field of interest. More speakers were attracted by the provisional agenda, until finally we had from 13 to 17 May 1999 a symposium with 26 lectures and 20 posters. We thank all for their valuable contributions. The topic ‘vertebrate communities’ was not absolutely adhered to in all contributions. Several papers did not fit into ‘communities’, but into ‘tropical vertebrates’. We had one presentation dealing with non-insect arthropods. Since this paper presents valuable aspects of how isolation works it is also included in this volume. We were pleased to find a speaker who supplied an introduction to phytogeography, undoubtedly the basis for all evolutionary aspects in animals. There are papers that predominantly deal with theoretical aspects, others that provide insight into biodiversity and evolutionary traits, and some that are concerned with the conservation of these isolated communities. The contributions are arranged according to their main geographic reference. So we start in South America, and pass over West, Central and East Africa to the islands between Africa and India; we have two contributions on the Oriental region and close with two on the south of the Palearctis. Those presentations of the symposium not reported in this volume are added, with their abstracts, in alphabetical order of the authors at the end of this book. It is my conviction that proceedings only make sense when they are published 8 immediately after the symposium. Therefore it was my special concern to bring together all the manuscripts in a very short time. I have to thank most contributors for their understanding on this point. Even the last ones were kind enough to ensure that the publication date was not affected. All papers passed one or two reviewers. I am most indebted to N. Arnold — G. Arratia — A. Barker — K. Busse — R. van den Elzen — H. Enghoff - J. Fjeldsa — J. Ganzhorn — J. Haffer — R. Hutterer — I. Ineich — U. Joger — M. Louette — A. Ohler - B. Patterson — G. Peters — K.-L. Schuchmann — H. Winkler, who had consideration for my desire for quick publication; nevertheless they did their job with great accuracy and responsibility, for which I thank them all sincerely. My special thanks belong to Brian Hillcoat, who very carefully went through all those papers that came from non-native English speakers. He contributed much more than only an improvement of the English. In her usual manner, S. Rick helped to solve the many large and small problems that arise with such a project. We have to thank the great number of nameless volunteers under the guidance of A. Schunke; without their expert assistance such an international event would be impossible. Deutsche Forschungsmeinschaft, Bonn gave financial support in two ways: firstly DFG supplied the largest part of the budgets for the organization of the symposium (GZ 4851/34/99). In addition DFG supported the travel facilities for our East European colleagues (GZ 436 114/88/99). We thank DFG very much for this help. Deutsche Lufthansa AG sponsored the flight of one invited speaker from abroad. Deutsche Telekom contributed gifts in kind for the handling of the symposium. Stadt Bonn gave support to persons from less developed countries. References Peters, G. & R. Hutterer (1990) (eds.): Vertebrates ın thes topics. Proceedings of the International Symposium on Vertebrate Biogeography and Systematics in the Tropics. - Alexander Koenig Zoological Research Institute and Zoological Museum, Bonn. 424 pp. Schuchmann, K.-L. (1985) (ed.): Proceedings of the International Symposium on African Vertebrates: Systematics, Phylogeny and Evolutionary Ecology. - Zoologisches Forschungsinstitut und Museum Alexander Koenig. 585pp. Ulrich, H. (1997) (ed.): Tropical Biodiversity and Systematics. - Proceedings of the International Symposium an Biodiversity and Systematics in Tropical Ecosystems. - Zoologisches Forschungsinstitut und Museum Alexander Koenig. SS pp: 9 Analyzing species composition in fragments Bia Dre Prat te ns om oc Wo Atmar Abstract: Fragmented systems tend to exhibit distinctive patterns of species richness and species composition. Because local extinctions on fragments are no longer balanced by immigration from surrounding areas, their species-area relationships characteristically exhibit high slopes. Small fragments often support many fewer species than an equivalent area of contiguous habitat. In addition, fragments often support “nested subsets” of species, where the species comprising smaller local assemblages constitute a proper or included subset of the species in richer ones. Nestedness appears most prevalent and most strongly developed in fragmented systems, where species composition has been sculpted by local extinction. However, this structure also characterizes other kinds of ecological systems. Extinction, colonization, disturbance, habitat distribution, hierarchical niche relationships, and passive sampling may all shape local assemblages into nested patterns. Although nested structure per se can tell us little about the processes that produced it, the ordering of species and sites in a nested matrix can tell us a great deal about possible causation. Characteristics of species or sites that correlate strongly with these vectors become plausible contributors to the nested structure shown by the system as a whole. Key words: Biogeography, extinction, habitat fragments, nested subsets, species composition Introduction Fragmentation of natural habitats is a progressive by-product of escalating resource consumption by humans. Continuous tracts of natural landscapes are eroded by human activities into increasingly dissected patches in an anthropogenic matrix. Fragmentation poses an especially severe threat to tropical organisms, which often have relatively small geographic ranges (Rapoport 1982, Terborgh & Winter 1983) and occur at low density because of their generally patchy distributions (Foster 1980). Fragments differ from isolates in only one fundamental respect: their history. However, biogeographic history has pervasive effects on the types of species that are present on islands (Ricklefs & Schluter 1993, Patterson 1999; table 1). Although colonization, speciation, and extinction must simultaneously affect all islands continuously (at least as rates), their cumulative impact on a fauna or flora varies enormously according to its prior history. Patterns of species richness on fragments commonly differ from those shown on isolates or in non-isolated areas of contiguous habitat. Lawlor (1986) showed that the species-area slope, z (from regressions of log species on log area), averaged twice as high in archipelagos fragmented by rising sea-levels as in oceanic ones in which islands have always been isolated. In fact, the distribution of slopes for the two classes of islands did not overlap. Smaller fragments are subject to a surfeit of Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 10 Table |: Hallmarks of two distinctive classes of islands Isolates Fragments Relicts Endemics') characteristic species favored species Invaders, r-selected Generalists, K-selected species richness Low species-area slopes Steep species-area slopes species composition Distributional checkerboards Nested subsets species flux Colonization-limited Extinction-driven ') Of course, endemics are also produced by vicariance, which involves fragmentation of once- continuous ranges. However, in such cases endemism arises with evolutionary divergence after the vicars have become isolated. The former identity of vicars (that is, as surviving daughter lineages of a widespread common ancestor that predated divergence) is underscored by the term relict. In a similar fashion, relicts may exist within currently contiguous habitats. local extinctions as species number falls to a level commensurate with their newly circumscribed areas. Conversely, the biotas of isolates — regardless of their size — are often limited by rates of colonization (Lawlor 1986). Historical derivation also influences the species composition of archipelagos. Many insular systems exhibit a pattern of species composition termed nested subsets, in which the species comprising a small fauna or flora represent a proper or included subset of those on larger, richer islands, rather than a random draw of those found in the entire species pool (Patterson 1990, Ganzhorn 1998; Wright et al. 1998). Nestedness is especially well-developed in systems thought to have undergone “faunal relaxation” (Patterson & Atmar 1986, Wright & Reeves 1992, Wright et al. 1998). During relaxation, the common species pool that occupied each island prior to fragmentation is ravaged by local extinctions until the species inhabiting the fragment are again in balance with their new surroundings (e.g., Diamond 1976). Because the extinction risk of species is often set by attributes such as body size, trophic position and habitat specialization (Brown 1971, Diamond 1984), in turn influencing their rarity, the same species tend to disappear from each of the islands, and in approximately the same order (Patterson & Atmar 1986, Blake 1991, Bolger et al. 1991, Atmar & Patterson 1993). Nestedness is acommon response to faunal relaxation, as dramatically shown by frogs on Amazonian forest fragments in the “Minimum Critical Size of Eco- systems” project (fig.l in Wright et al. 1998; cf. Zimmerman & Bierregaard 1986). Yet most ecological systems are nested, and the factors influencing this ecological structure are as diverse as nature itself. Extinction, colonization, disturbance, habitat distribution, hierarchical niche relationships, and passive sampling may all shape local assemblages into a nested pattern (Patterson 1990, Patterson & Brown 1991, Simberloff & Martin 1991, Cutler 1994, Andrén 1994 a,b). Causes of nestedness in these specific cases are usually identified on the basis of other information available in each study. However, Lomolino (1996) proposed that causality could be assessed directly by organizing rows in the presence-absence 1] matrix according to each plausible determinant, and measuring the nestedness of each resulting matrix. Thus, matrices organized by area (for extinction) and by isolation (for colonization rate) could be directly compared, and the strength of these often-opposing forces could be directly assessed. However, separate analyses are required to evaluate each potential determinant (Lomolino 1996). The Nestedness Calculator The “Temperature Calculator” of Atmar and Patterson (1993, 1995) seems ideally suited to explore various features of nestedness, including causation. Other metrics (Patterson & Atmar 1986, Cutler 1991, Simberloff & Martin 1991, Wright & Reeves 1992, Lomolino 1996) measure nestedness in relation to the distribution of each species. These partial scores are then summed over species to compile community-wide patterns (1.e., counting “up” in a site x species matrix, then summing across columns). Each ordering of rows (=sites) produces a potentially different distribution of “holes” and “outliers” and a different nestedness score. In contrast, “temperature”, T, assesses degrees of nestedness simultaneously across species and sites (i.e., counting diagonally across the matrix). Just as the species compositions of fragments can be nested within one another, so too can the incidence distributions of species. Neither deserves logical precedence in ecological analyses. Moreover, there is only one arrangement of “presences” in a distribution matrix that maximizes nestedness across both these dimensions. This essential structure can then be compared to numerous possible correlates without matrix reorganization. Atmar & Patterson (1995) developed a Windows-based Visual Basic” program for implementing analyses of T. Their “Temperature Calculator” is freely available over the Internet. Subsequent sections detail steps needed to calculate the biogeographic temperature of any set of distributions and to assess their biological determinants: (1) packing matrix elements so as to maximize nestedness; (2) measuring matrix order; (3) assessing the statistical significance of nestedness; (4) identifying and interpreting biogeographic discordance; (5) assessing causation; and (6) calculating state-occupancies. Packing the matrix Presence-absence matrices contain two levels of information. In addition to specifying which species occur at which sites, these matrices also reflect the relative hospitality of sites to the species under study, as well as the prevalence of environmental conditions needed to support each species. This secondary information becomes apparent only after a matrix has been packed into a state of maximal nestedness. The “hospitality” of islands or sites declines from top to bottom of the matrix. Likewise, the prevalence and width of species = niches are ordered from the left to right. Actually, these rankings of rows and columns emerge from reordering the rows and columns of the matrix to minimize the unexpectedness of occurrences. Although few matrices prove to be perfectly nested, all matrices can be packed into a state of maximal nestedness. Matrices are packed to a condition of maximal nestedness by reordering entire rows and columns until unexpectedness is minimized. Changing the order of rows and columns does not alter which species occur at which sites, but it does change the overall appearance of the matrix. Nested matrices can be readily distinguished without tests because their presences are tightly clustered into the upper-left corner of the matrix. The topmost site in a packed matrix is judged the most hospitable. Similarly, the leftmost species is the one whose niche requirements are most commonly and consistently met. In practice, it may be the most resistant to extinction, most prone to colonization, or most probably encounter resources essential to species persistence. The concept of redundancy can also be appreciated with reference to a packed matrix. When a series of sampled sites each supports all of the species in the pool, the information they provide on the ordering of species and of sites is entirely redundant. For purposes of measuring temperature and depicting matrices, such sites are combined into a single all-inclusive site in the topmost row of the matrix. For purposes of assessing the rank of individual sites (see below), all such sites are tied at first. Calculating matrix “temperature” The concepts of heat, information, noise, order and disorder are all closely related. The calculated metric measures the “biogeographic heat” of the matrix based on the distribution of unexpected presences and absences. In a perfectly nested matrix, the set of species on any island will be a proper or included subset of the species on all islands that precede it in the matrix. The hypothetical line that separates the occupied area of the matrix (i.e., the upper-left corner of the matrix) from the unoccupied portion is termed the boundary line. The boundary line for a perfectly ordered matrix is not arbitrary nor does it depend on specific distribution patterns. Instead, the line is specified only by the size, shape, and “fill” of the matrix (i.e., how many presences it contains). Species absences above and to the left of the line are defined as unexpected, as are species presences below and to the right of it. Because the boundary line defines the condition of maximal nestedness of species and sites, every unexpected presence beyond the line is accompanied by a corresponding absence within it, and vice versa. When stochasticity is low (i.e., when T is low), unexpected presences and absences cluster near the line. As system randomness increases, the unexpected presences and absences move further away from it. The “temperature” of the matrix is a measure of that penetration. Following Atmar and Patterson (1993), the formula for local unexpectedness of cell ij is: Dalal Da) where d ij measures the distance of the cell from the boundary line along the skew diagonal, and D jj is the length of the matrix parallel to the skew-diagonal. Similarly, total unexpectedness is: U=1/(mn)2& u, 13 summed over mrows and n columns, and system temperature is calculated using the constant K, where K = 100/U as max ? Woes ouh Assessing probabilities A nested distribution pattern represents a highly specific type of distribution. Monte Carlo techniques can be used to estimate the probability that the nested structure in any distribution pattern could be produced at random. To assess that probability, any number of matrices can be drawn at random to produce a baseline expectation. These matrices are generated wholly at random (1.e., 100°). However, because random events often clump together, and the matrix-packing algorithm compiles these clustered events to maximize nestedness, the randomized matrices will be cooler after they are packed. The extent to which the characteristic temperature of these matrices deviates from 100° depends on the size, shape and degree of fill of the matrix. Colder characteristic temperatures will be produced in smaller matrices, those that are mostly filled or mostly empty, and in matrices that are highly rectangular (i.e., many more rows than columns, or vice versa). In each case, it is possible for the packing algorithm to cluster clumps of “presences” into the upper-left corner of the distribution matrix. 126 : | = d=0.0217us. 3, E W = 0.98746 ns. > | we fi ©: 07 = = 3 So i a ee ow ‘ Oey au“ 300° system temperature Fig.1: Testing the significance of observed nestedness. The top graphs depict tests of nor- mality for 1000 simulation scores generated for mammals in the southern Rocky Mountain mammals (Patterson & Atmar 1986). Results of both tests for normality, the Shapiro- Wilks’ W and the Kolmogorov-Smirnov d, showed non-significant differences (P < 0.05), validating the probability inference made in the lower graph, in which the likelihood that observed structure was part of the simulated scores is ~10”. 14 It then becomes a simple problem to ask “What is the probability of an observed temperature of, say 4.15°, given the trial distribution of simulated scores?” The trial distribution may be taken either as a sample of scores or as a universe. Hugueny & Guegan (1997) criticized the use of normal-probability distributions when assessing the distribution of test statistics that were generated using Monte Carlo techniques. Their argument revolved around the distribution-free nature of Monte Carlo methods, and the often-skewed distributions of test parameters. Fig. l contains a test for normality of 1000 temperature scores for the Rocky Mountains mammal data set of Patterson and Atmar (1986), generated using the Nestedness Calculator. Both the Shapiro-Wilks = W score and the Kolmogorov-Smirnov d test for normality are non-significant (P > 0.05), showing that T scores are approximately distributed as a normal variable. The result is that after measuring T in 1000 simulated archipelagos, we have far greater assurance than one in 1000 that the observed temperature of 4.15° falls outside the range of simulated scores. In fact, using a normal approximation, the probability that observed T is included within the range of simulated scores is ~10°”. Detecting idiosyncratic species and sites Two forms of noise contribute to the temperature of a matrix: (1) the random variation of environmental, demographic, and genetic stochasticity; and (2) the “coherent” noise of specific biogeographic events or of ecologically distinctive species. Random noise creates a gray band of mixed presences and absences along the entire length of the boundary line. In contrast, coherent noise creates idiosyncratic “spikes” that correspond to species or islands that contribute much more noise than the remainder. Where temperatures are not uniformly distributed across islands and species, contrasting biogeographic histories or current ecologies are implied for those islands and species, respectively, than those that characterize the system as a whole. Distributions of 82 species of bats along the Eastern Versant of the Andes in Peru (Patterson et al. 1996) illustrate the potential of the Calculator for identifying idiosyncrasy (fig.2). Bat species richness falls precipitously with elevation, so that highland faunas generally represent attenuated versions of those below, forming a nested pattern. However, the individualized responses of highland bat species are strikingly discordant. Each of Manu’s endemic bat species is identified by temperature “spikes” in the lower pane of fig.2. Auxiliary information is required to determine why given species or islands are idiosyncratic. Idiosyncrasies often indicate unmarked heterogeneity in the original data set. Idiosyncrasies among sites may result from habitat heterogeneity (especially for groups of habitat specialists) or signify sites that experienced con- trasting biogeographic histories. Idiosyncratic species might variously recolonize some islands from which they have been locally extirpated, reach their range limits in the midst of the archipelago, or be victims of competitive exclusion or products of local speciation. In this sense, idiosyncrasy may often indicate that the “rules” of the system as a whole have been violated. AT TT IAL Ht INTUTE EET SL yl an Il N mens ai | | | nie ll | Carollia sp. nov. i: Vampyressa melissa Anoura sp. nov. | Sturnira erythromos Fig.2: Interpreting idiosyncrasies and assessing the basis for nested distributions. The upper graph shows the nested distributions of 82 bat species (columns) at 19 sampling points (rows) along an elevational gradient in the Andes in southeastern Pert (Patterson et al. 1996). In this region, species richness of bats declines rapidly with elevation, and species do not generally replace one another along the gradient; higher-elevation communities are attenuated versions of those below. The temperature of this system is 10.2° and the probability that this could be due to chance is -10°°. The lower graph shows the contrasting distributions of a handful of highland endemics, which appear as idiosyncratic “spikes” in the plot of species idiosyncrasies. The Spearman rank-correlation coefficient between rows (in the packed matrix) and elevation equals R = 0.936 (P < 0.00001); nested structure may be plausibly attributed to elevation (or its correlates). Assessing causation The existence of strongly nested distributions suggests an universal ordering mechanism for species or sites in the system under study. However, the rows and columns of the matrix submitted to analysis have been rearranged in the packing step to maximize nestedness. Where are the matrix alterations recorded, and how can matrix order be compared to independently determined features? Matrix reorganization vectors record the rearrangements of rows and columns between matrices as submitted and those that emerged from the matrix-packing step. Descriptive vectors for either species or sites (e.g., body size estimates of species or area estimates for sites) may then be compared to the order of each in the packed matrix. Because the ordering of species by the Nestedness Calculator is only relative, rank-correlation analysis of such variables is preferred (cf. Siegel 1950): Causal analyses are applied to two well-studied biogeographic systems. In the first (table 2), nestedness of montane mammals in the southern Rocky Mountains can be correlated with various physical features of the mountains themselves. The order of rows (mountain ranges) in the packed matrix correlates with their physical characteristics (see table) at tabulated levels. In the second, distributions of Peruvian bats over 19 sampling stations along the Eastern Versant of the Andes 16 Table 2: Rank-correlations of assemblages of montane mammals on mountaintops in the southern Rocky Mountains (Patterson & Atmar 1986). Row order determined by matrix packing and then correlated with various extrinsic factors. IP a Elevation -.327860 n.s. Latitude -.808813 <.00001 Longitude .406460 <.05 Coniferous forest area -.510545 <.01 Mesic forest area -.678533 <.0001 Distance to source .491228 <.05 Notes: Elevation and coordinates determined for the highest peak. Areal measure taken from Patterson (1984), and distance measure from Lomolino et al. (1989). are also highly nested. Correlations of row order in the packed matrix with elevation reveal strong correlation (R,= 0.9364, P < 0.0001), suggesting that elevation or a co-variate determines nested structure. As in other non-experimental fields, causation of nestedness must be inferred from correlations and appropriate disclaimers are necessary. Calculating state occupancy The preceding steps for measuring nestedness apply to any matrix, regardless of its historical derivation, its current dynamics, or the uniformity of its species and sites. If more stringent assumptions of the system are warranted (such as its being at colonization-extinction equilibrium and with homogeneous resource distributions; cf. Atmar & Patterson 1993), several powerful inferences are possible. Nested distributions imply that the cell most likely to be occupied in any matrix is the one in the upper left-hand corner; that is, the most ubiquitous species will virtually always be present on the most hospitable island. Similarly, the cell least likely to be occupied will be the bottom, right-most cell, where the most marginal species would be found on the least hospitable island. All other matrix cells vary between these extremes in a manner that is specified by their distance from the boundary line. One may therefore calculate the probability of each matrix cell being occupied. At zero degrees (the condition of perfect order), all cells within the boundary line will be filled. However, as matrix temperature increases, unexpected absences or presences begin to appear in those cells closest to the boundary line, where unexpectedness values are lowest, spreading out across the matrix as a whole. Two graphs prepared by the Nestedness Calculator can be used to indicate the probability of a cell’s occupancy. The first, a “band diagram”, shows unexpected- ness values for expected occurrences where the species is present, as well as un- expected presences and absences. The second, a cumulative probability diagram, shows state occupancy as a function of unexpectedness for the current matrix. It 1s from this second diagram that the stability of the various populations can be estimated for each cell in the matrix (fig.3). Calculating state occupancy presumes that the biogeographic noise is random rather than coherent. Thus, unexpected presences resulting from idiosyncratic causes will be deemed more unstable than they actually are. Similarly, the unexpectedness of idiosyncratic absences will be overestimated. excellent candidates for reintroductions == 5 a = HE; = = == = =F ses, = A Be: Be, = eA poor candidates | : sosition : for salvage efforts ; Fig. 3: Using a packed and highly nested matrix to make management decisions; data are boreal mammals of the southern Rocky Mountains (Patterson & Atmar 1986). “Holes” in the packed matrix signify populations that ought to be present, based on species co-occurrence data; introductions of such species are presumably favorable. On the other hand, “outliers” indicate populations not expected under the nested subset pattern, and seemingly a poor risk for strenuous management efforts. In fact, the lower arrow identifies an officially endangered population — Tamiasciurus hudsonius — on the Pinaleno Mountains, and one placed under notice of review by the U.S. Fish and Wildlife Service — Microtus longicaudus. Both lie outside the boundary line of nested relationships and hypothetically are next at risk of local extinction. Discussion Nestedness appears to be a near-universal property of the assemblages studied by ecologists and biogeographers. Its virtual ubiquity in the natural systems that scientists select for analysis is striking because, like other ecological patterns, nestedness is scale dependent (Patterson 1990). Conditions for the development of nested structure include: 1) the species and sites being compared must have a common biogeographic history; 2) both must be exposed to similar contemporary ecological conditions; and 3) some kind of hierarchical niche relationships among species or sites (Patterson & Brown 1991). 18 The first two conditions ensure that sites have been open to colonization by a common species pool. The third ensures that there are differences in the numbers of species among sites, and/or differences in the incidence of species among sites. Differences among species in extinction risk, vagility, or habitat affinities affect their incidence, whereas differences in area, isolation, elevation, or habitat composition affect the species richness of sites. Each of these variables may contribute to nested structure and deserves causal investigation. Nestedness is a property of assemblages, not one of individual species, so it is reasonable to ask whether nestedness reveals anything that cannot be determined from the ordered distributions of constituent species (cf. Simberloff & Martin 1991). In fact, nestedness analyses can simultaneously track both overall and specific features of distributions, as illustrated in fig.2. The predominant faunal pattern shows Amazonian bat faunas becoming attenuated along the elevational gradient, with highland faunas representing nested subsets of those at lower elevations. However, the responses of a handful of highland bat species exhibit strikingly discordant patterns, as shown in the plot of idiosyncrasies. Some, such as Sturnira erythromos and Anoura geoffroyi are also perfectly ordered by elevation, only inversely with respect to the remaining bat species. Nestedness requires a-hierarchical ordering of sites, species or both. As widely noted (e.g., Williams 1964), habitats are often distributed in a patchy manner, and fragments may differ substantially in habitat composition, in ways affecting both species richness and composition. Sites may also be ordered with respect to isolation, area, or other variables (table 2). Orderings of species are commonly expressed in terms of niche preferences. Most biologists are familiar with the “Hutchinsonian” niche concept, based on the probability of an organism’s survival/reproduction within a n-dimensional hyperspace. Over all resources, this probability distribution equals the fundamental niche, but because other species preempt resources, the observed or realized niche is expressed as the occupancy of species in shared resource space. A different concept was articulated by James et al. (1984) and seems especially appropriate to geographic analyses based on presence-absence data. The “Grinnellian” niche reflects the range of environmental values that are necessary and sufficient for a species to carry out its life history. Within a geographic region, species occupy a geographic region congruent with the distribution of its niche. The density and incidence of species within their geographic ranges reflect the prevalence of these conditions (James et al. 1984). Seen in these terms, the rankings of species across an archipelago provide an objective ordering of species with respect to Grinnellian niche width. Differences in incidence may influence species interactions, as detailed by Hanski (1982) in discussing core and satellite species. The nested ordering of species differs from core-satellite species mainly in being graded, not classified. A fundamental asymmetry of ecological relationships is implied by nested subsets: the “core” species in nested systems (e.g., A) are predictable elements of the environment for more marginal species (e.g., D), but the reverse is not true. Species’ responses to competition and other coevolutionary pressures should 19 therefore be asymmetric, with direction being set by the prevalence or incidence of the species involved. Moreover, as noted by Simberloff and Martin (1991), nested biotas are not apt to be dominated by competitive exclusion, which tends to produce distributional checkerboards, the opposite of nestedness. Guidelines for conservation The distinctive patterns of species richness and composition in fragments have important implications for conservation. First, the elevated slopes of species richness changes with area means that insular impoverishment with areal reduction is substantially greater on fragments than on islands (Lawlor 1986). The result is that generalized “island” slopes are inappropriate for predicting the species richness values of fragments (Patterson 1991; Doak & Mills 1994). By using such “typical” slopes, previous forecasts of tropical species loss with deforestation (e.g., Simberloff 1986) seriously underestimate the eventual costs of habitat conversion. Nested subsets also have relevance to biological conservation, most obviously in resolving the SLOSS conundrum. In a perfectly nested archipelago, small reserves each support the same set of species (Patterson & Atmar 1986). However, few systems are perfectly nested, and nestedness is seldom perfectly correlated with island area. In addition, systems of small reserves, when selected on the basis of faunal or floral dissimilarity, will often preserve more species than the single largest island (Simberloff & Gotelli 1984). The utility of nestedness in reserve planning appears to be limited, at least when the compositions of all areas are known — “smart” amalgamation routines invariably perform better than those based solely on area considerations, whether from small-to-large or large-to-small (Cook 1995, Lomolino 1996). But nested subsets have other implications as well. In any strongly nested archipelago, the boundary line marks the distributional limits of the fauna or flora, beyond which population lifetimes are expected to be nil, and species are either absent or expected to be absent. This leads to the expectation depicted in fig.4, that population lifetimes should vary systematically over a nested matrix, reaching their zenith at the origin. In general, we can expect greatest survivorship for the most widespread species on the most hospitable fragment, and declining probabilities for more narrowly distributed species and less hospitable fragments until we reach the boundary line. For a biological system that is undergoing faunal relaxation, the matrix cells distributed along the boundary line denote populations that are most at risk. For a highly dynamic biological system comprised of metapopulations, those boundary populations will wink on and off with greatest frequency, as local populations momentarily recover from local extinctions only to disappear again. In general, rarity should be inversely correlated with population lifetimes shown in fig.4 (Diamond 1984). The precise position of the boundary line will then depend on the intensity or scale of sampling. Exhaustive sampling (that will uncover the rarest, most narrowly distributed species population in a fragment) or long-scale sampling (that uncovers and records populations that are present only episodically in a given site; cf. Andrewartha & Birch 1954) will shift the boundary line to the lower right of the matrix. This does not, however, render the nested pattern artifactual because nestedness depends on the pattern of co-occurrence, not population lifetimes jie Hees a Hey Heo a POOH Pon RH PS Hy Ss fe HHS ESCHE HP CHAT SHHHIT ie HHH HPSS Mesa oe ERSTEN tt ieee ett Hee hs Tate He Puh HE HERR KHHHREUHRFR ie Sn HH: populations at greatest risk of local extinction Fig.4: Hypothetical relationship between population lifetimes and the position of popula- tions in a highly nested matrix. By definition, lifetimes fall to zero where the species is absent, but should generally increase — perhaps more-or-less regularly — as one approaches the most widespread species on the most hospitable fragment. All along the boundary line, populations of different species are thought to be most at risk. the number of occurrences. Livingstone & Grayson (1994) showed that sampling artifacts for Great Basin mammals actually obscured the underlying nested structure, and that additional records of occurrence from more intensive sampling improved the fit of these mountaintop mammals to the nested pattern. Nestedness is fundamentally a type of hierarchical organization. Hierarchies are familiar to biologists, forming the basis of the Linnean system of nomenclature, as well as the foundation for most attempts to classify the world’s biogeographic regions. Despite their simplicity, hierarchies provide a powerful means of storing and retrieving abundant information. As one example, few people can have personal knowledge of Pseudoryx nghetinhensis, an animal discovered in central Vietnam in 1992. However, knowing the hierarchical position of this species (Mammalia, Artiodactyla, Bovidae, Bovinae, Boselaphini or Tragelaphini; Schaller & Rabinowitz 1995) permits a host of inferences regarding its morphology, genetics, and evolution. In a similar manner, knowledge of the hierarchical 21 structure of nested biotas permits numerous inferences — for example, if species M is present, then so too are species A, B, C, ..., K, and L. Although Simberloff & Martin (1991) questioned whether community-wide statistics had any transcendental value (compared to those calculated on a per-species basis), surely this predictivity is one important reason. The relative rankings of species and fragments and the ancillary analyses they permit provides another. Many questions in ecology and evolutionary biology require knowledge of species-abundance relations, but nestedness analyses are based solely on presence- absence data. Although this limits the potential applications of nestedness, this feature is seen as a distinct advantage because of “data economy”. Even the most rudimentary ecological sampling yields a set of species occurring together at a place and time. Such so-occurrence data are often the best we can manage for many poorly sampled tropical areas. Moreover, given spatial and temporal variation, the presence or absence of species should be more stable than their relative abundance rankings. Development of theory that is compatible with such data should be a high priority for conservationists, again especially in the tropics. Understanding the potential biases of such analyses can only refine our ability to fashion useful predictions from them. Acknowledgments We are grateful to Renate van den Elzen for making travel arrangements, Rainer Hutterer and Gustav Peters for their hospitality during the meeting, and to colleagues David Wright, Alan Cutler, and Greg Mikkelson for extended discussions of the meaning and significance of nested subsets. We thank Jorg Ganzhorn and Jon Fjeldsa for helpful comments on the presentation and manuscript. The Nestedness Calculator was developed with the support of the National Science Foundation (BSR-9106981) and our home institutions. References Andrén, A. (1994a): Effects of habitat fragmentation on birds and mammals in landscapes with different proportions of suitable habitat: a review. — Oikos 70: 355-366. Andren, H. (1994b): Can one use nested subset pattern to reject the random sample hypothesis? Examples from boreal bird communities. — Oikos 70: 489- 494. Andrewartha, H.G. & L.C. Birch (1954): The distribution and abundance of animals. — University Chicago Press, Chicago. Atmar, W., & B.D. Patterson (1993): The measure of order and disorder in the distribution of species in fragmented habitat. — Oecologia 96: 373-382. Paaitaiemey.. oc B.D Patterson (1995): "Ihe nestedness temperature calculator: a visual basic program, including 294 presence-absence matrices. — AICS Res., Inc., University Park, New Mexico, and The Field Mus., Chicago (download at http://aics-research.com/nestedness/tempcalc.html). Blake, J.G. (1991): Nested subsets and the distribution of birds on isolated 27 woodlots. — Conserv. Biol. 5: 58-66. Bolger, D.T., A.C. Alberts &M.E. Soule (99). Occurence pattemsior bird species in habitat fragments: sampling extinction, and nested species subsets. — Amer. Nat. 137: 155-166. Brown, J.H. (1971): Mammals on mountaintops: Nonequilibrium insular biogeography. — Am. Nat. 105: 467-478. Cook, R.R. (995): The relationship between” nested ssubsets) habiran subdivision, and species diversity. — Oecologia 101: 204-210. Cutler, A.H. (1991): Nested faunas and extinction in fragmented habitats. — Conserv. Biol. 5: 496-505. Cutler, A.H. (1994): Nested biotas and biological conservation: metrics, mechanisms, and meaning of nestedness. — Landscape Urb. Plan. 28:73-82. Diamond, J.M. (1976): Relaxation and differential extinction on land-bridge islands: applications to natural preserves. — Proc. 16th Int. Ornithol. Congr., Canberra, Australia, 616-628. Diamond, J.M. (1984): “Normal” extinctions of isolated populations. — pp. 191-246 in: Extinctions. (ed. Nitecki, M.H.). — Univ. Chicago Press, Chicago. Doak, D.F. & L.S. Mills (1994): A useful role for theory in conservation. — Ecology 75: 615-626. Foster, R.B. (1980): Heterogeneity and disturbance in tropical vegetation. — pp. 75-92 in: Conservation biology: an evolutionary-ecological approach (eds. Soule, M.E. & B A. Wilcox). — Sinauer Assoc., Sunderland, Mass. Ganzhorn, J.U. (1998): Nested patterns of species composition and their implications for lemur biogeography in Madagascar. — Folia Primat. 69: 332-341. Grayson, D.K. & S.D. Livingston (1993): Missing mammals on Great Basin Mountains: Holocene extinctions and inadequate knowledge. — Conserv. Biol: 1.527532. Hanski, I. (1982): Dynamics of regional distribution: the core and satellite species hypothesis. — Oikos 38: 210-221. Hugueny, B. & J.F. Guegan (1997): Community nestedness and the proper way to assess Statistical significance by Monte-Carlo tests: some comments on Worthen and Rohde’s (1996) paper. — Oikos 80: 572-574. James, F.€., R.E. Johnston; N.O. Wamer, 6 J Niemeyer Boecklen (1984): The Grinnellian niche of the wood thrush. — Amer. Nat. 124: 17-30. King, W.B. (1978-1979): Red data book, 2: Aves. 2nd ed. — IUCN. Morges, Switzerland. Lawlor, T.A. (1986): Comparative biogeography of mammals on islands. — pp. 99-125 in: Island biogeography of mammals (eds. L.R. Heaney & B.D. Patterson). — Biol. J. Linnean Soc. 28(1&2) and Academic Press, London. Lomolino,M.V.,J.H. Brown &R. Davis (1989): Island biogeography of montane forest mammals in the American Southwest. — Ecology 70: 180. Patterson, B.D. (1984): Mammalian extinction and biogeography in the Southern Rocky Mountains. — pp. 247-293 in: Extinctions. (ed. Nitecki, M. H.) — Univ. Chicago Press, Chicago. Patterson, B.D. (1990): On the temporal development of nested subsets patterns of species composition. — Oikos 59: 330-342. Pane ns on. BeDr (1991): The integral’ role of biogeographic theory in the conservation of tropical forest diversity. — pp. 124-149 in: Latin American mammals: History, Biodiversity, Conservation. (eds. Mares, M.A. & D.J. Schmidly). — Univ. Oklahoma Press, Norman, OK. Patterson, B.D. (1999): Contingency and determinism in mammalian biogeo- graphy: the role of history. — J. Mamm. 80: 345-360. Patterson, B.D. & W. Atmar (1986): Nested subsets and the structure of mammalian faunas and archipelagos. — pp. 65-82 in: Island biogeography of mammals. (eds. Heaney, L.R. & B.D. Patterson). — Biol. J. Linn. Soc. 28(1&2) and Academic Press, London. Batuverson, 5.De & TH. Brown (1991): Regionally nested patterns of species composition in granivorous rodent assemblages. — J. Biogeogr. 18: 395-402. Basserson, BD Ve Pacheco & S. Solari (1996): Distributions of bats along an elevational gradient in the Andes of south-eastern Peru. — J. Zool. 240: 637-658. Quinn, J.F. & S.P. Harrison (1988): Effects of habitat fragmentation and isolation on species richness:evidence from biogeographic patterns. — Oecologia 75: 132-140. Rapoport, E.H. (1982): Areography: geographical strategies of species. — Fund. Bariloche Ser., Pergamon Press, Oxford. Retcieletsen ce DE Schluter (1993) (eds.): Species diversity in ecological communities: historical and geographical perspectives. — Univ. Chicago Press, Chicago. Schaller, G.B. & A. Rabinowitz (1995): The saola or spindlehorn bovid Pseudoryx nghetinhensis in Laos. — Oryx 29: 107-113. Siegel, S. (1956): Nonparametric statistics for the behavioral sciences. — McGraw-Hill Book Co., New York. Simberloff, D.S. (1986): Are we on the verge of a mass extinction in tropical rain forests? — pp. 165-180 in: Dynamics of extinction. (ed. Elliott, D.K.) — Wiley-Interscience, New York. Sımberloit, D.S. &N. Gotelli (1984): Effects of insularization on plant species richness in the prairie-forest ecotone. — Biol. Conserv. 29: 27-46. Simberloff, D.S. & J.-L. Martin (1991): Nestedness of insular avifaunas: simple summary statistics masking complex species patterns. — Ornis Fenn. 68: 178-192. Terborgh, J.W. & B. Winter (1983): A method for siting parks and reserves with special reference to Colombia and Ecuador. — Biol. Conserv. 27: 45-58. Williams, C.B. (1964): Patterns in the balance of nature. — Academic Press, London. Wright, D.H. & J.H. Reeves (1992): On the meaning and measurement of nestedness of species assemblages. — Oecologia 92: 416-428. 24 Wright, D.H., B.D. Patterson, EG. Mikkels one AH Cuvier Same Atmar (1998): A comparative analysis of nested subset patterns of species composition. — Oecologia 113: 1-20. Zimmerman, Bla) & RO] Bierre caand 2 198, Relevance oraihe equilibrium theory of island biogeography and species-area relations to conservation with a case from Amazonia. — J. Biogeogr. 13: 133-143. Authors’ addresses: Bruce D. Patterson, Department of Zoology, The Field Museum, Chicago IL 60605- 2496, USA Wirt Atmar, AICS Research Inc, P O Box 4691, Las Cruces NM 88003-4691 (WA), USA YS Chronology and mode of speciation in the Andean avifauna J. Garcia-Moreno & J. Fjeldsa Abstract: This paper reviews speciation studies of Andean forest birds, using mtDNA sequence data, and explores possible driving mechanisms. Typical biogeographic patterns in the Andean forest zones comprise sharp (parapatric) replacements along the Andes and elevational replacements where species sometimes occupy different ecological zones along several thousand kilometers of Andean slopes. The DNA data (if we accept the existence of a molecular clock) reveal much variation in the timing of speciation events: the diversification has proceeded during the entire uplift of the northern Andes, with speciation in the Pleistocene glacial periods only in some lineages. Some parapatric replacements may have been maintained since before the onset of major glaciations. This suggests that vicariant forms are valid species which remain ecologically incompatible for long periods, possibly because they are subject to uniform selection pressures. Establishment of sympatry (through dispersal) may be long delayed, partly because sharp geographical replacements are maintained by physical barriers (such as a valley deeply intersecting the cloudforest zone). We suggest that most Andean species originate as relict populations which persist because of local orographic protection against ecoclimatic disturbance. The process of relictuation, which may be related to high levels of local community drift in large parts of the region, affects only some species. It therefore does not follow a consistent chronology across lineages. In order to assess whether relict taxa are recruited into the regional species pool we now need detailed studies using broader sample schemes and statistical documentation of historical community structures based on coalescence theory. Key words: speciation, Andes, avifauna, mtDNA Introduction By comparing regional patterns of species richness of tropical birds representing recently radiated groups with those representing deep lineages, Fjeldsa (1994) and Roy et al. (1997) demonstrated that in the recent geological past the cradle of new Species was in the montane regions. The biodiversity of the South American lowland rainforest is to a large extent a legacy from the Tertiary “super-rainforest”’ (Hooghiemstra & van der Hammen 1998, Flenley, in press). In such lowland areas, the biogeographic signals concerning initial speciation events are buried by subsequent redistribution, and the variation in endemism reflects that of species richness and current ecology (Fjeldsa et al. in press). In the montane areas, in contrast, distinctive patterns of endemism are more likely to reflect recent historical isolations (Fjeldsa 1995). In this paper we will evaluate questions about the causes and mechanisms of rapid diversification in South American montane areas. Speciation in montane areas is often explained by the many isolating barriers and the tremendous elevational gradients, where selective pressures on birds vary over Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 26 short distances (Vuilleumier 1969, 1984a, Simpson-Vuilleumier 1971, Fjeldsa 1992, O’Neill 1992). Attention is also given to the temporary isolation caused by Pleistocene glaciation, although this is often limited to the most recent (and best documented) glacial cycles (e.g., Vuilleumier & Simberloff 1980, Traylor 1985, Fjeldsa 1992, Heindl & Schuchmann 1998). These interpretations need to be re- evaluated in the light of new kinds of palaeoclimatological data and new insights into the mechanisms which drive global climatic change (see Andersen & Borns 1994 for a review). The global climate is subject to continuous orbital forcing, resulting in frequent but slight oscillations throughout the upper Tertiary, a marked cooling and longer (40 000-year) cycles in the early Pleistocene, and an abrupt change only 800 000 years ago to a period characterized by long (100 000 year) large-amplitude cycles (Hooghiemstra et al. 1993). Thus, if biological diversification were driven by periodic isolation of fragments of once-continuous biomes (Haffer 1997), and if new species were efficiently redistributed in the intervening periods, we would predict an exponential increase in speciation events in the upper Pleistocene. However, speciation could also be driven by high local biotic turnover (community drift, Hubbell 1999), which is caused by ecoclimatic disturbances on any time scale. We assume that this 1s possible because some species will persist only as relict populations ın places which are inherently stable and where the biotic turnover is slight. Using a long series of satellite images, Fjeldsä et al. (1999) were able to document that peaks of endemism in the Andes correspond to places where the interannual variability in surface conditions is slight, reflecting local orographic moderation of the climate. As traditional ideas about processes and kinds of isolating mechanisms are currently being questioned, the understanding of processes underlying the diversification of tropical biota remains a hot topic. Rather than entertaining plausible but unfalsifiable scenarios we need clear conceptual models which allow robust testing. Lynch (1988) asked for documentation of consistent hierarchies of phylogenetic branching patterns across many taxonomic groups. However, if the diversification process was dominated by divergence of relict populations or was randomized over time, as suggested above, then only some species would be affected, and not necessarily at the same time. The predictions would then be that speciation patterns are non-synchronous in time (across lineages), that speciation was allopatric (by isolation of relict populations) and that endemism is locally aggregated. If speciation involves persistence of relict populations where the climate is predictable and the level of faunal turnover is slight, then neutral genetic drift would dominate, resulting in random variation in phenotypes between isolates. In this paper we will review a number of case studies of Andean birds where molecular data were used. We will use these data to evaluate the ideas outlined above. Materials and methods We studied several widespread and species-rich groups of Andean birds, focusing on groups which show linear arrangements of species along the upper eastern Andean slope. This focus is based on the following rationale: Because of the 2] linearity of the band of humid montane forest (Graves 1988), we presume that sister-group relationships are mainly one-dimensional in space (and therefore easy to analyze). Populations which replace eachother along the Andes, in a specific ecological zone, would also be subject to rather uniform selection pressures. It is already well documented that such groups often exhibit a ‘checkerboard’ variation of phenotypes, as different characters vary independently between populations, sometimes in a way which produces decided ‘leapfrog’ patterns (e.g., Rensch 1929, Remsen 1984a, Remsen & Graves 1995 a,b, Graves 1980, 1982, Maijer & Fjeldsa 1997). Most groups also include species which replace each other sharply in different elevational zones, or on humid slopes and adjacent rainshadow slopes. This means that there will also be cases of consistent directional selection, which might cause extremely rapid phenotypic change (Carson 1990, Walter & Paterson 1994, Smith et al. 1997). DNA was extracted from preserved blood and tissue samples. Fragments of the mitochondrial cytochrome b and/or ND2 genes were amplified and sequenced following standard protocols. The two genes exhibit similar substitution rates (Johnson & Sorensen 1999). We constructed phylogenies for those genera for which we had a good taxonomic sampling: Arlapetes brushfinches (Garcia-Moreno & Fjeldsä 1999), Metallura hummingbirds (Garcia-Moreno et al. 1999a), Cranioleuca spinetails (Garcia-Moreno et al. 1999b), various tanagers (Hemispingus, Iridosornis, Chlorospingus, Diglossa; Garcia-Moreno unpubl.), Lepidocolaptes woodcreepers (Garcia-Moreno & Silva 1997), and Ochthoeca chat- tyrants (Garcia-Moreno et al. 1998). Opportunistic sampling was supplemented by specific efforts to fill in collecting gaps and to make comparisons across a potential dispersal barrier. We also used published molecular data for other Andean groups. While ca. 100 species were covered altogether, the sampling was insufficient for describing population structures within species. To estimate evolutionary time we used genetic distances (uncorrected), as well as Kimura’s two parameters, and assumed three different rates of nucleotide substitutions: 2% per million years (Klicka & Zink 1997, and references therein), 10% 3rd position substitutions per million years, and 0.5% 3rd position transversions per million years (Irwin et al. 1991). The two latter measures complement each other: The total 3rd position rate is ideal for comparing closely related taxa, but is poor when comparing distantly related ones, because of saturation; the 3rd position transversion rate tracks the relative age of branches over a long evolutionary time frame (Burns 1997) but provides few data for comparing closely related taxa where there are few (or no) transversions. We compared divergence data for taxa occurring in allopatry, parapatry, and sympatry. To minimize effects due to the possibility of different substitution rates in different groups we restricted these comparisons to those within genera and groups of related genera. We assume that the substitution patterns within such groups reflect the chronology of events rather than variation in substitution rates. Even in comparisons between groups, there is little reason to fear dramatic differences in substitution rates, since all groups (except the hummingbirds) are passerine birds of fairly constant size and generation time characteristics. A large portion of them even belong in one close-knit group of ‘nine-primaried oscines’. 28 ‘Hotspot’ of mean endemism Species richness ‘hotspot’ Fig.1:‘Hotspots’ of species richness and mean endemism for highland birds in the tropical Andes region. ‘Hotspots’ are defined here as the 200 top-scoring cells: the first 50 cells have the largest symbols, etc. The mapping was done using the WorldMap software (Williams 1994) in the form of conservative range maps (in a 15'x15' grid) for all bird species which are well established above 2500 m (at least locally) (see Fjeldsä et al. 1999). The mean endemism is calculated as the mean inverse range-size for all species. These maps were made from a datafile constructed using a phylogenetic species concept (see Fjeldsa in press for details); however, with the 15' resolution, the species richness is virtually unaffected by the species concept used. 29 Results Biogeographic patterns The regional patterns of species richness and endemism for high Andean birds are illustrated in fig.1. The species richness ‘hotspot’ (fig.1A) follows the band of humid montane forest fringing the upper Amazon Basin. Despite some local varia- tion, there is no general latitudinal trend within this range (Fjeldsa et al. 1999). Congeneric species often replace each other in different elevational zones (long parallel distributions in fig.2) or they replace each other sharply along the Andes, sometimes across physical barriers, such as the North Peru Low, but sometimes in places with no conspicuous barrier (fig.2). The endemism is locally aggregated (fig.1B). Some aggregates extend right across a habitat gap, as range-restricted birds may inhabit cloudforest fragments on either side of a deep (dry) valley, many of which interrupt the eastern Andean cloud-forest band (Fjeldsa 1995). Case studies (starting with groups which radiated mainly before the Pleistocene) Ochthoeca chat-tyrants. These are common birds of Andean shrubby vegetation, most species being widespread along the length of the Andes, with broadly overlapping ranges (fig.2) but some habitat segregation (Garcia-Moreno et al. 1998).On the basis of the level of genetic divergence, and a few sharp replacements along the Andes, some of the traditionally accepted species may represent pairs of sister species (cinnamomeiventris/thoracica and frontalis/spodionota). All species which are distributed along long sections of the Andes show some local variation, e.g., in the expression of wingbars. Apart from this subspecific differentiation, the radiation predates the upper Pleistocene (figs.2 and 3, Garcia-Moreno et al. 1998). Because of the extensive sympatry, the biogeographic signals are often lost. However, an adaptive shift is apparent from bushy highland habitat in the deep branches to closed montane forest, and bamboo areas in the lower montane zone in the top branch. Scytalopus tapaculos (Arctander & Fjeldsa 1994, and some unpublished data). This is a very complex group of secretive, sooty-grey birds which scurry like mice through dense understory of humid montane forest. A large number of species (which are most safely identified by songs and calls) replace eachother sharply in different elevational zones, or in different mountain ranges. As most species are poorly known or only recently recognized (Krabbe & Schulenberg 1997) the sampling was incomplete (and the group is therefore not shown in figs.2 and 3). However, a 10-15% sequence divergence in most comparisons suggests that the population subdivision is ancient. In all comparisons of species which replace each other elevationally on a slope, it was found that each was most closely related to a taxon inhabiting another mountain range (four-taxon tests). This strongly supports the view that the speciation was allopatric and that sympatry (with sharp elevational segregation) is secondary. Tanagers, Thraupinae. This large group showed a strong radiation during the Miocene (Sibley & Ahlquist 1990, Burns 1997). We studied variation in the genus 30 <—N ECUADOR NPL PERU Cuzco BOLIVIA Sr; Ochthoeca STICK „Jeiskii 5, 7- i ulchella ] diadema 5.7 BS p = 81-117 frontali „ [ spo ionota trots 8.2 EP __spodionsia ag 67; — cinnamomeirentris 5.6 thoracica ] 9.6-15.7 rufipectoralis ] fumicolor Be ken z 2 berlepschi Chlorospingus TS semifuscus sss ws a EEE NEE I er pig OP Nee i parvirostris (oS SST a — — canigularis =? = Hemispingus 11.9 Eng Sr fests | En Eee 8.4-10.3 [LD__aphys —] 12 superciliaris ] verticalis 4 3. 7 xanthopht. ] 11.7-17.1 [__ rufosupercil. 1 so ?[L__] parodii BRETTEN RL Le peer BE UN HE en EIERN ER ET Re OES EIERN =] frontalis 1-11.12.—— il, See or men He 2 a] , —46.2= Iridosornis rufivertex lal reinhardti ] 4.5 > 81 -9,8— = jelskii SER ] ? analis ] :porphyro. Atlapetes tricolor 5 __latinuchus 14——_atinuchus | 71? ee . 5.6 [ schistac. | See ‘ing: 13. 8[ schist. 3, 9 ienllee melanolaemus 1 ieee 7b canigenis 6.8 rufinucha 5 S| terborghi be fulviceps ss: Metallura rianthina ] 2.6——44 6.5 williami ] N atri. <; 7 odom. !, theresiae PB u En = 5.1 aeneocauda 1.2 Tae ral = baroni Diglossa laffresnayi 27 er 4 mystacalis ] humeralis | brunneiventris 0.7 — aba 1 Cranioleuca a mare 41,6, „ae 3 = albicapilla ] E 3,7 am BESTER EEE +-37> [0.8 pyrrhophia curtata ] Fig. 2. Spatial replacements of taxa along the cis-Andean slope of Ecuador, Peru and Bolivia (with taxa inhabiting adjacent intermontane and trans-Andean areas shaded). Figures mark the sequence divergence between closely related groups which inhabit adjacent slopes or different altitudinal zones on the same slopes. 3] Ampelion and Doliomis ‘et Seer ee ee ee SS Be U ee a) ie ee es ee ee eS eee eS eS SE See (Se ee ee ee) a a SES nn u Ko Ochthoeca ne en ee m ne] EEE EN — ES SS ee) x xXx — Se Chlorospingus ee eee > . — * — | Hemispingus * r - not covered Fe * = Iridosornis not covered = een seen BD & N RER, = Allapetes ze Se en Seel not covered ESE Sur * 1 z— 0 | Metallura Pree ccs ee ee ee E | = = — [= a eat *x ——— (Je a Diglossa not covere SES Lt Cranioleuca : 5 Lower Upper Miocene Pliocene Pleistocene Pleistocene 8 7 6 5 4 3 2 1 Millions of years before present Fig. 3. Estimated ages for phylogenetic branches in ten groups of Andean birds. Some of these groups began to diversify before the upper Miocene, but because of the incomplete sampling we will not estimate their time of origin. The bars mark the maximum and minimum estimates of the timing, according to three methods specified in the text, of a specific phylogenetic node. The margins of uncertainty are large for early events,but much smaller for the most recent events. An * indicates that a species was lacking in our sampling, and this information is placed according to our judgement of where the species should be in the phylogeny. Chlorospingus, which belongs in a rather deep branch, and three more recently radıating groups (Hemispingus, mountain tanagers (mainly /ridosornis), and flowerpiercers Diglossa). Hemispingus shows similar distribution patterns to Ochthoeca, with many widespread but elevationally segregated species, but also some replacements (fig.2). Three main clades form a trichotomy that we date to the upper Miocene: (1) a clade with long branches and short internodes comprising eyebrowed birds (tree topology trifasciatus [atropileus {auricularis, calophrys}]); (2) mostly grey birds (superciliaris [xanthophthalmus, verticalis]); 3) mostly ochraceous birds (rufosuperciliaris [piurae {frontalis, melanotis}]). The differentiation may precede the Pleistocene, even among sister species like verticalis and xanthophthalmus (fig.3). Two M. melanotis representing two subspecies have identical DNA sequences but show an early (Pliocene) divergence from the Tumbesian ‘subspecies’ piurae (which may be a valid species). For Chlorospingus our sampling was incomplete, but among the Andean species we found canigularis to be basal, flavigularis and parvirostris being sister species, 32 and the widespread and polytypic opthalmicus being closest to semifuscus of the wet northern Pacific slope. Interestingly, the Bolivian population of opthalmicus, which superficially resembles birds from Colombia and Venezuela rather than adjacent Peruvian birds, may be a distinctive species (diverging since the mid- Pleistocene). The radiation predates the upper Pleistocene (fig.3). Iridosornis, a group of colourful mountain tanagers, shows vicariance as well as elevational segregation (fig.2). The tree topology is still ambiguous. Most speciation events were in the Pliocene, but the two parapatric sister species rufivertex and reinhardtii may have diverged in the upper mid-Pleistocene. We studied the Diglossa carbonaria group and samples of D. laffresnayii and mystacalis, using D. sittoides as an outgroup (see Hackett 1995). The mystacalis- lafresnayii complex may be paraphyletic, diverging in the Pliocene. The D. carbonaria complex is widely distributed in bushy habitats throughout the tropical Andes region, with sharp geographical displacements (fig.2) of birds with strikingly different pigment saturations. There is very little genetic variation in this branch (figs.2 and 3) from Colombia to Bolivia despite the sharp replacements, and there is limited hybridization only in one of the contact zones (Graves 1982). We also have data for a few Basileuterus wood warblers (unpublished), which may be very close to the tanagers (Sibley & Ahlquist 1990, Burns 1997). This group is incompletely sampled but our results suggest speciation events of a similar age as In tanagers (tab.1). Metallura hummingbirds. Garcia-Moreno et al. (1999a) evaluated the idea of Graves (1988) that the differentiation is strongest at the treeline since the narrow physical configuration of this habitat implies high risks of local elimination and isolation of relict populations. We compared the relationship between two assumed (and verified) sister taxa: M. tyrianthina, ubiquitous in the humid montane zone of the tropical Andes region, and the M. aeneocauda group, which 1s narrowly bound to humid treeline shrubbery. Diversification proceeded through the Pliocene and Pleistocene, with one very recent speciation event (M. eupogon/theresiae, fig.3). We confirmed that the clade of treeline forms was deeply differentiated, starting with an early isolation of the southernmost population (aeneocauda) and range fragmentation progressing northwards. An aberrant species in mist vegetation above the Peruvian coastal desert (M. phoebe, large and melanistic) is closest to its geographically nearest neighbours in southern Ecuador and may owe its aberrant morphology to strong directional selection in its special environment. Otherwise, the treeline clade is characterized by checkerboard variation in phenotypic characters (Graves 1980), while in M. tyrianthina there is a more consistent latitudinal trend with the most derived phenotypes in the north (Heindl & Schuchmann 1998). Atlapetes brush-finches (Garcia-Moreno & Fjeldsa 1999). Phenotypically distinct forms replace each other sharply, at different altitudes, on adjacent humid and rain- shadowed slopes, along the Andean slope, or in some cases in a mosaic (fig.2, Remsen & Graves 1995a,b). In the traditional classification (Paynter 1970), grey and yellow-and-green species are placed in different groups, with many similarly 33 coloured populations ranked as subspecies. The DNA data did not support this subdivision (Garcia-Moreno & Fjeldsa 1999a). With the reservation that northern and west-slope populations were not included in our study, three main branches were demonstrated, diverging at the Plio-Pleistocene transition, each comprising geographically adjacent but morphologically different forms, and with most speciation events in the mid-Pleistocene (figs.2 and 3). Plumage colours did not faithfully reflect the evolutionary trajectories in this genus, as adjacent and related populations may be strikingly differently pigmented. The data also show that many forms, currently ranked as subspecies, are genetically more divergent than some Atlapetes species which are sympatric (although segregated in adjacent elevational zones). Since the replacements are abrupt, sometimes in places with no habitat discontinuities, and with scarcely any signs of hybridization, we will consider these forms as valid and competing species. Table 1: Sequence divergence (uncorrected) between related species which are sympatric (syntopic or segregated in adjacent elevational zones along the Andean slopes) and parapatric (replacing each other in adjacent territories along the Andes, with or without a separating physical barrier) (Mann-Whitney U-test). genus mean distance proba- significance (range) bility level Scytalopus sympatric 0.161 (0.084-0.168) parapatric 0.098 (0.063-0.119) Ochthoeca sympatric 0.118 (0.044-0.157) parapatric 0.079 (0.056-0.143)' Hemispingus sympatric 0.143 (0.071-0.222) parapatric 0.103 (0.087-0.119) Chlorospingus | sympatric 0.122 (0.087-0.165) parapatric 0.066 (0.025-0.130) Iridosornis sympatric 0.080 (0.053-0.098) parapatric 0.020 Diglossa sympatric 0.107 (0.027-0.134) parapatric 0.033 (0.007-0.082) Metallura sympatric 0.072 (0.064-0.080) parapatric 0.037 (0.002-0.086) Atlapetes sympatric 0.128 (0.042-0.143)° parapatric 0.040 (0.014-0.066) Cranioleuca sympatric 0.044 (0.042-0.045) parapatric 0.019 (0.005-0.041) Anairetes sympatric 0.044 (0.054-0.079) parapatric 0.013 Basileuterus sympatric 0.083 (0.070-0.105) parapatric 0.033: DD SINE DOW NW RRARNDND — = a OO - OO ') Comparisons include Tumbezia salvini 7 . . . ~) Comparisons include Buarremon brunneinucha 34 Cranioleuca (Garcia-Moreno et al. 1999b) belongs in the complex ‘spinetail’ group, which radiated mainly along the eastern Brazilian-Andean track. We focused on what seemed to be a natural group of species which replace each other along this track, mainly in woodland and thornscrub habitats, and with a distinctive checkerboard variation of phenotypes (Maijer & Fjeldsa 1997). The DNA data suggested a recent and rapid radiation (mainly in the upper Pleistocene, fig.3) with somewhat deeper splits between the main clades, both of which include species which inhabit other ecological zones and which diverge phenotypically from the focal group. This suggests that in addition to a random sorting of phenotypic traits in the focal group there were several adaptive shifts into humid forest or even lowland forest (swamp-forest and species-poor islands of whitewater rivers). The radiation is not fully resolved and within the most closely knit species groups geographically unstructured haplotypes were demonstrated, suggesting retained ancestral polymorphism and random lineage sorting (Avise 1994). Other groups Ampelionid cotingas (Arctander & Fjeldsä manuscript; fig.3) show deep branches indicating a mid-Miocene differention of plantcutters (Phytotoma, in dry Prosopis woodlands) and the humid forest lineages Ampelion (A. rubrocristatus, which is ubiquitous in the montane forests of the tropical Andes region, and A. rufaxilla, which is more local in high-rainfall areas), Doliornis (very local in elfin forests along the eastern Andean ridge) and Zaratornis (endemic to Polylepis and cloud forests above the central Peruvian coastal desert and in the adjacent Cordilleras Blanca area). Doliornis shows a mid-Pleistocene vicariance (D. remseni in the northern central Andes of Colombia and locally in Ecuador; D. sclateri in central Peru). Flycatchers of the genus Leptopogon are represented by four species, amaurocephalus being widespread in the tropical lowlands, superciliaris in the submontane zone of the Andes, and taczanowskii and rufipectus in the lower montane forest north and south of the North Peru Low. MtDNA data (Bates & Zink 1994) show the topology (amaurocephalus [superciliaris {taczanowskii/ rufipectus}]), with the most recent split at 4 Mya (fig.2). Lepidocolaptes woodcreepers (Garcia-Moreno & Silva 1997) are widespread in Neotropical forests and savanna woodlands. The mtDNA data suggest that the split between the species of forest and savanna woodland took place as recently as 2 Mya, and the separation of L. lacrymiger, which is distributed in montane forest along the entire tropical Andes region, is a similar recent event. Adelomyia melanogenys. The distribution of haplotypes among ten individuals from different sites in Ecuador, Peru and Bolivia (Garcia-Moreno unpubl.) showed a single (poorly resolved) cluster of haplotypes, except for the single Bolivian taxon (the distinctive subspecies inornata) which may have diverged in the Pliocene. The latter could represent a distinctive species, although Zimmer (1951) claims evidence of hybridization in Cuzco. Tit-tyrants (Anairetes, Uromyias), widespread birds of bushy Andean habitats, diversified mainly around the Plio-Pleistocene transition with one vicariance event in the upper Pleistocene (fig.3, Roy et al. 1999). Andean pipits (Anthus), which are 39 widespread grassland birds with considerable range overlaps between species, speciated in the lower to mid-Pleistocene (Volker 1999). Asthenes spinetails, inhabiting the same ecological zone, show as slight genetic differences as Cranioleuca (Arctander 1991 and unpublished). American Carduelis species, which are widespread in bushy habitats, with large range overlaps, radiated within the last 3 Mio years (Arnaiz-Villena et al. 1998). A molecular study of conebills (Conirostrum), whıch are also widespread in bushy and wooded habitats, suggests that this group radiated in the Andes in the early Pleistocene, and that it is paraphyletic with respect to Oreomanes fraseri, a highly specialized bird of the high-elevation Polylepis woodlands (Nielsen & Arctander submitted). The divergence is estimated at 2.9 Mya, which matches well with the abundance of Polylepis pollen since 3.1 Mya. Chronogram Fig.3 shows the temporal distribution of phylogenetic node since the upper Miocene. It should be noted that each node is represented by a bar which shows the range of different time estimates, which is considerable for the Mio- and Pliocene events but much smaller for those in the upper Pleistocene. Assuming that all groups have similar substitution rates, that calibrations estimated for mammals (Irwin et al. 1991) apply, and that our species sampling is representative, we find that the diversification of Andean birds was well spaced throughout the time period, except maybe for more intensive speciation in the upper Miocene, Pliocene and mid-Pleistocene. It should be noted that the tanager genera in fig.3 show very different chronograms, although they are closely related (Burns 1997) and inhabit similar ecological zones. A few cases in Diglossa and Cranioleuca suggest that speciation may sometimes take place within a time frame of one or two glacial periods (see Graves 1982 and Maijer & Fjeldsa 1997 regarding species rank). However, this happened only in few lineages, and never as a Series of events which might correspond to successions of glacial periods. Other lineages showed no speciation in the upper Pleistocene, and this even applies to species inhabiting ecoregions which were extensively glaciated (e.g., Anairetes alpinus, Ochthoeca fumicolor and oenanthoides, Oreomanes fraseri). Speciation events within the period of marked Pleistocene glaciations are: Metallura williami/atrigularis/baroni and M. theresiae/eupogon, Cranioleuca antisiensis/baroni/curtata and erythrops, Cranioleuca marcapatae/albiceps and C. obsoleta/pyrrhophia and henricae, Diglossa humeralis/brunneiventris/carbonaria. In Atlapetes, some speciation events may correspond to the early part of the period. Parapatric and sympatric species We compared the divergence data for taxa occurring in allopatry, parapatry, and sympatry. Sympatry refers to co-existence on the same slope, although sometimes with elevational segregation. The allopatric species represent a ‘mixed bag’, ranging from closely related species isolated in different mountains to remotely related species which inhabit different parts of the Andes and sometimes different life zones (fig.2). The number of comparisons are insufficient for testing for 36 differences between sister species which are local/patchy and not strictly parapatric, those which are widespread and common up to a habitat discontinuity which separates them, and those which are in actual parapatric contact. We therefore present a statistical comparison of para- and sympatric species only (tab.1). For every taxonomic group, the sympatric species were all strongly divergent compared with most parapatric species. In all cases, except for Anairetes, Basileuterus and Iridosornis (where we had only a single parapatric comparison), we found a significant difference in the genetic divergence exhibited by sympatric and parapatric species (tab.1), even though some species that replace each other in adjacent mountains (i.e., parapatric in our terminology) are genetically distinct. Also taking into account the results by Arctander & Fjeldsa (1994) (and Patton & Smith 1992 for Andean mice), we reject the view that speciation can proceed through differential selection across an ecological gradient (Endler 1982). Speciation starts with isolation on different slopes, and sympatry (with elevational and other kinds of segregation) becomes possible only after a considerable time lag. The relative role of the North Peru Low The North Peru Low (NPL, see fig.1) is a geologically complex zone with a relatively low and tectonically splintered western cordillera, and an erosion gap through the eastern Andean ridge probably predating the uplift of this cordillera (Räsänen et al. 1987). This gap has been regarded as a main barrier causing speciation of highland birds (Vuilleumier 1969, Graves 1985, O’ Neill 1992, Stotz et al. 1996). The species replacements across the gap represent a wide range of time scales, from 0.4 to 11.9% sequence divergence (5.7 = 4.06 %, N=12) Ihıszeanshe compared with the variation across the NPL within species, which varied from 0.3 to 6.1 (in Adelomyia melanogenys) (2.8 + 1.81 %, N=9; or 0.3, 0.7, 2.4, 4.1 and 6.1 % for five pairwise comparisons of samples collected within 150 km of the gap). The species replacements, and the intraspecific comparisons, demonstrate isolation and dispersal at many time scales. The sister species separated by the NPL included in our study inhabit mainly upper montane forest. Parker et al. (1985) suggested that major rivers such as the Maranon may not affect the dispersal of mid-elevation forest birds so much. However, a considerable divergence was found in the mid-elevation genera Adelomyia and Leptopogon. Several northern taxa are established (without conspicuous morphological differentiation) immediately south of the NPL, on the eastern ridge, displacing a southern (sister?) taxon slightly to the south, where there is no conspicuous barrier (Fjeldsa & Krabbe 1990). Very few cases exist of a displacement slightly north of the NPL. For several groups we found that the principal split was not at the NPL but in the southern part of the tropical Andes region (in the Metallura aeneocauda group with 5.1-8.5% sequence divergence, Atlapetes with 4.9-8.0%, Cranioleuca with 2.7- 4.0%, Hemispingus (eye-browed group) with 8.4-10.3%) (Stotz et al. 1996). This split separates populations in a southern area which includes the Apurimac-Cuzco and La Paz centres of endemism (fig.1). Within currently recognized species, deep 3 southern splits were documented in Adelomyia melanogenys (7.7-10.1%) and Chlorospingus ophthalmicus (5.7%). A dozen montane forest birds, which are widely distributed near the equator, may have an isolated population south of Cuzco or in the Bolivian yungas (Fjeldsa & Krabbe 1990). The endemics of the dry forests and local cloudforests of the Tumbes region (southwestern Ecuador and northwestern Peru) also vary in age, which may be related to the predictable mist front (and therefore persistence of mist vegetation) in this region, in addition to the isolating effect of the Andes (Best & Kessler 1995, Fjeldsä 1995, Garcia-Moreno et al. 1998). Discussion In the past, many authors related the speciation patterns in the Andes to topographic barriers. We accept that the uplift of the northern Andes must have provided numerous opportunities for speciation by orographic isolation. However, it is important to note that barriers which separate sister species today were not necessarily the initial cause of speciation (Cracraft & Prum 1988, Louette 1992). Only a detailed study of historical population structures can reveal whether populations inhabiting different slopes diverged because of the barrier between them, or whether the replacements are secondary contact zones. Speciation events have also been related to specific Pleistocene glaciations (Traylor 1985, Fjeldsa 1992, Heindl & Schuchmann 1998). This may offer plausible scenarios. However, direct evidence of the timing is lacking and it is difficult to provide robust tests of effects of climatic cycles with no beginning or end phase. We tried to evaluate the traditional interpretations using the molecular clock concept, assuming that it is better to have a crude clock than no clock at all (Bromham et al. 1999). We will stress that reliable calibrations of evolutionary rates are difficult to make, especially as the fossil record is very fragmentary. Various statistical tests of the relative uniformity of base substitutions along different lineages have been performed (see Mindell & Thacker 1996, Avise & Walker 1998, Bleiweiss 1998 a, Waddell et al. 1999). Bleiweiss (1998 a,b) found that hummingbirds show a high rate of mtDNA divergence, but with a markedly reduced rate in highland hummers. The different rate tests suggest fairly uniform accumulation of base substitution within restricted taxonomic groups (such as nine- primaried oscines, comprising tanagers and Atlapetes in our study; Burns 1997). Many authors come to a similar 2% estimate using very different groups of birds (note 11 in Klicka & Zink 1997). Because of the uncertainty we used several different estimates. This does not help with the uncertainty caused by the relatively short sequences we used, although because of the good taxon sampling we are confident about the phylogenetic results (see e.g., Graybeal 1998). Our time estimates should be taken as a rough measure and should not be overinterpreted. We could not support a synchrony of speciation events across taxonomic groups (and particularly not among the closely related ‘nine-primaried oscines’, where each genus shows its own pattern in fig.3). In our data set there are no indications of successive speciations during the sequence of eight large-scale climatic cycles in 38 the upper Pleistocene (Hooghiemstra et al. 1993), as is sometimes implied in the literature. It is important to note that several groups showed no speciation in the upper Pleistocene, while others showed several speciation events near the mid- to upper Pleistocene boundary. Some of these (e.g., Atlapetes) could represent the simultaneous break-up of several lineages, in which case attempts to resolve them would be futile (Stepien & Kocher 1997). We have to take into account that the density of nodes in the most recent past is accentuated by our sampling regime, and also potentially by the extinction of deeper lineages. A high level of phylogenetic resolution suggests slow rates of historical isolation compared with the coalescence time, except in the Diglossa carbonaria group and Cranioleuca. We therefore conclude that the diversification of the Andcan avifauna proceeded throughout the period of uplift and gradual formation of new habitats in the northern Andes since the Miocene (Wijninga 1995, Hooghiemstra & van der Hammen 1998). Similarly, Bleiweiss (1998 c) related the rapıd radiation of hummingbirds since the lower Miocene to the creation of new habitats by major geological and climatic upheavals and multiple dispersals. Even though the usefulness of the molecular clock is debatable, we prefer to accept this kind of evidence rather than perpetuating a model which is based on belief more than data. Our interpretation of those groups which show a marked vicariance pattern of mid-Pleistocene age is that this pattern could often persist despite the dramatic ecoclimatic chances in the upper Pleistocene. The idea of speciation pulses driven by glacial-interglacial cycles appears to be a failed paradigm (Klicka & Zink 1997) (although this does not exclude population structuring or some speciation events related to glacial isolation). The typical mode of speciation must be allopatric or parapatric, by isolation in different (but usually adjacent) montane areas (as also suggested by Patton & Smith 1992 for Andean rodents, and by Arctander & Fjeldsa 1994). In the groups we studied, species with small distributions are found mainly in recently differentiated clades (fig.2). We infer from this (and from tab.1) that sympatry comes later, after long isolation, adaptive change and dispersal. A rapid dispersal can in some cases be supported by low levels of morphological and genetic differentiation along the Andes (with examples showing even zero divergence over 1000 km). In many lineages, the biogeographic signals lie buried in multiple cycles of redistribution, and detailed studies of historical population structures would be needed to retrieve them. We could also use our current data under the assumption that rates of base substitution will be modest in populations which remain in situ compared with those which colonize new areas, with founder effects and shifting selection pressures. Using outgroup comparisons, we found that the Ochthoeca radiation involved southern origins, followed by northwards dispersal for O. fumicolor, leucophrys/oenanthoides, pulchella and spodionota. However, no such polarity could be established in Metallura, and for other taxa with long linear distributions, where our sampling was inadequate. The random phenotypic variation in lineages with species replacements along the Andes (Graves 1982, 1985, Remsen 1984 a,b, Remsen & Graves 1995 a,b, Maijer & Fjeldsä 1997, Garcia-Moreno & Fjeldsä 1999) suggest that speciation starts with 39 divergence of tiny isolates. This is also indicated by the random sorting of haplotypes in Cranioleuca populations inhabiting relictual forest patches (Garcia- Moreno et al. 1999b). Morphological characters would fail to retrieve the branching patterns that were substantiated by the molecular markers. Subtle differences between small isolated populations in relict woodlands within the highlands may also reflect random processes in small populations but may not be conducive to speciation (Vuilleumier 1984 b). Our data provide strong indications that geographic replacements can sometimes be maintained over long time spans (e.g., Atlapetes). The reason could be a lack of directional selection where populations remain in the same area. In this situation we expect populations to diverge only by neutral drift. It can also be inferred that parapatric taxa are valid species: they represent cohesive and genealogically concordant lineages of individuals that share a common fertilization system through time and space, represent independent evolutionary trajectories, and demonstrate essential but not necessarily complete reproductive isolation from other such units (Johnson et al. 1999). Naturally, replacements between such incompatible species are most easily maintained where there is a physical barrier, although a few replacements do exist on slopes with no obvious barrier. Overall, our results are consistent with our initial predictions based on the model of divergence of relict populations (p.26). Within the Andes, certain areas may be inherently stable because of local high pressure areas, topographic conditions which cause stable (laminar) mist formation on certain slopes, or specific projecting mountain ridges moderating the impact of cold south polar winds. The most conspicuous of such areas are just north of the Tunari highlands (Cochabamba, Bolivia), where the projecting Cordilleras Mosetenes and Cocapata form a marked shelter against the impact of cold southern winds, causing a benign and stable climate in the Cotacajes drainage and in the southern Yungas of La Paz. A similar area exists in Cuzco and Apurimac in Peru, where a fan of mountain ridges provides a similar shelter. The south polar winds today cause short but frequent spells of cold winter weather in the lowlands and Andean slopes in the Southern part of the tropical zone. Servant et al. (1993) suggest that the vegetational changes in tropical South America during the Pleistocene were driven mainly by the increased impact of these winds; however, they are potentially disturbing also on other time scales. Species respond individually when tracking the environmental oscillations within their geographical ranges (Walter & Paterson 1993, Colinvaux et al. 1996). The high levels of environmental disturbance (even within fairly continuous habitat) imply high levels of faunal turnover sorting out some of the rarer species from local communities (Hubbell 1999). The best prospects for persistence of relict populations are therefore in places with slight local faunal turnover. According to this interpretation, speciation may proceed without barriers which would isolate entire communities. Instead diversification 1s a random process where only some species undergo relictuation and independent evolution of each isolate. Speciation relates in this case to the intrinsic properties of specific places rather than to the barriers between them. 40 Hooghiemstra & van der Hammen (1998) suggest that because of the concave arrangement of the cis-Andean slope, this region would have been continuously forested during the Pleistocene. According to the mechanism outlined above, some species ranges could nevertheless be fragmented at any time, and the isolation of populations would last a long time because of the high general levels of climatic change and faunal turnover. It is conceivable that many relict populations do go extinct. However, repeated bottleneck effects may also remove deleterious alleles which could give inbreeding problems (Templeton & Reed 1984, part III in Loeschcke et al. 1994). Possibly, some small isolates are exposed to novel selective regimes in unique environments to adapt to new conditions (as indicated by size variations in Cranioleuca |Maijer & Fjeldsa 1997] and by rapid phenotypic changes in Oreomanes and Metallura phoebe) and this opens new avenues for speciation (Walter & Paterson 1994). Perspectives We now need to ask ourselves whether relict populations are isolates which persist for some time but eventually vanish - or whether they represent a potential source of recruitment to the regional species pool? We have already suggested that dispersal takes place. Over the geological period considered, even a low rate of successful expansions could substantially contribute to the regional species pool. However, dispersal and species pump models are extremely difficult to evaluate. We need to study (1) community structure (as high local turnover means that range-restricted species will be rare, while low turnover allows them to become locally abundant; Hubbell 1999) and (2) molecular data to document the historical community structure using large samples from many localities, rapidly evolving sequences (e.g., control region, nuclear microsatellites) and statistics based on coalescence theory (Hudson 1990, Hartl & Clark 1997). Many field studies (Fjeldsa & Kessler 1996, Andersen et al. 1999) indicate that viable populations of Andean birds can persist in tiny fragments of natural habitat in severely degraded highland landscapes. This gives some hope for conservation, but the aggregated pattern of endemism (fig.1b) nevertheless means that in order to minimize extinction we need to define clear priorities. The key problem for conservation of Andean highland birds 1s that peak concentrations of endemics are often immediately adjacent to densely populated areas (Fjeldsa & Rahbek 1999, Fjeldsä et al. 1999). Possibly, the ecoclimatic conditions moulding the pattern of endemism also provided predictable conditions for agriculture in the highlands. This means that the main challenge today for preventing extinction in this region will be to maintain patches of natural habitats, and biodiversity, within areas that are populated. The key problem will be to support other landuse methods, raising the production in some areas and stopping the intensive use of fire on all high ridges to maintain extensive grazing (Fjeldsa & Kessler 1996). 41 Acknowledgements The molecular work was supported by the Danish Natural History Research Council (grant no. 11-0390). We thank P. Arctander for all his support and assistance during this phase, and R.B. Payne for valuable comments on the manuscript. References Andersen, B.G. & H.W. Borns (1994): The ice age world. — Scand. Univ. Press, Oslo. Dndersen, PN... I. Hjarsen &N.M. Williams (1999): Monitoring and management of high Andean biodiversity — a study from Cochabamba, Bolivia.- DIVA Tech. Rep. 6. National Environmental Research Institute, Kalg, Denmark. Arctander, P. (1991): Avian systematics by sequence analysis of mtDNA. — Acta XX Congr. Int. Orn.: 619-628. Arctander, P. & J. Fjeldsä (1994): Andean tapaculos of the genus Scytalopus (Aves, Rhinocryptidae): a study of speciation using DNA sequences. — pp. 205- 225 in: Conservation Genetics (eds Loeschcke, V., J. Tomiuk & S.K. Jain).- Birkhauser, Basel. Arctander, P. & J. Fjeldsä (manuscript): The relationships of plantcutters, (Phytotoma, Aves), re-analyzed using DNA sequence data. — Steenstrupia. ana Nena, A. M. Alvarez-Tejado, V. Ruiz-del-Valle, C. Grctasce-la-Lorre, P. Varela, M.J. Recio, S. Ferre & J. Martinez-Laso (1998): Phylogeny and rapid northern and southern hemisphere speciation of goldfinches during the Miocene and Pliocene epochs. — Cell. Nol. Life Sci. 54: 1031-1041. Avise, J.C. (1994): Molecular markers, natural history and evolution. — Chapman & Hall, New York. Avise, J.C. & D. Walker (1998): Pleistocene phylogeographic effects on avian populations and the speciation process. — Proc. R. Soc. Lond. B. 265: 457- 463. Bes c& REM. Zink (1994): Evolution imto the Andes: molecular evidence for species relationships in the genus Leptopogon. — Auk 111: 507-515. Best, B.J.& M. Kessler (1995): Biodiversity and conservation of Tumbesian Ecuador and Peru. — BirdLife, Cambridge, U.K. Bleiweiss, R. (1998 a): Relative-rate tests and biological causes of molecular evolution in hummingbirds. — Mol. Biol. Evol. 15: 481-91. Bleiweiss, R. (1998 b): Slow rate of molecular evolution in high-elevation hummingbirds. — Proc. Natl. Acad. Sci. USA 95: 612-616. Bleiweiss, R. (1998 c): Origins of hummingbird faunas. — Biol. J. Linn. Soc. 65: 77-97. Bromhame 172 MI. Phillips & D. Penny (1999): Growing up with dinosaurs: molecular data and the mammalian radiation. — Trends Ecol. Evol. 14: 113-118. Burns, K.J. (1997): Molecular systematics of tanagers (Thraupinae): evolution and biogeography of a diverse radiation of Neotropical birds. — Mol. Phyl]. Evol. 42 72: 334-348. Carson, H.L. (1990): Increased genetic variance after a population bottleneck. rend Ecol; Evol. 32 228230: Colinvaux. P. AYP EV de Olıiveitra, J E. Moreno MIET NIN EEE M.B. Busch (1996): A long pollen record from lowland Amazonia: forest and cooling in glacial times. — Science 274: 85-88. Cracraft, J. & R.O. Prum (1988): Patterns and processes of diversification: speciation and historical congruence in some Neotropical birds. — Evolution 42: 603-620. Endler, J.A. (1982): Pleistocene forest refuges: Fact or fancy?. — pp. 641-657 in: Biological diversification in the tropics (ed. Prance, G.T.). — Columbia Univ. Press, New York. Fyeldsa, J. (1992): Biogeographic patterns and evolution of the avifauna of relict high-altitude woodlands of the Andes. — Steenstrupia 18: 9-62. Fjeldsa, J. (1994): Geographical patterns for relict and young species of birds in Africa and South America and implications for conservation priorities. — Biodiver. Conserv. 3: 207-226. Fjeldsä, J. (1995): Geographical patterns of neoendemic and older relict species of Andean birds: the significance of ecological stability areas. — pp. 89-102 in: Biodiversity and conservation of Neotropical montane forests (eds Churchill, S.P., H. Balslev, E. Forero & J.L. Luteyn). — The New York Botanical Garden. Fjeldsa, J. (in press): Choosing priority areas for global conservation: the relevance of systematics. — Envir. Conserv. Fjeldsä, J. & N. Krabbe (1990): Birds of the High Andes. — Apollo Books, Svendborg, and the Zoological Museum, Copenhagen. Fjeldsä, J. & M. Kessler (1996): Conserving the biological diversity of Polylepis woodlands of the highland of Peru and Bolivia. A contribution to sustainable natural resource management in the Andes. — Nordeco and Zoological Museum, Copenhagen. Fjeldsa, J., E. Lambin & B. Mertens (1999); Comelationm@bcicen endemism and local ecoclimatic stability documented by comparing Andean bird distributions and remotely sensed land surface data. — Ecography 22: 63-78. Fjeldsä,J. & C. Rahbek (1999) Continent-wide diversification processes and conservation priorities. — pp. 139-160 in: Conservation in a changing world (eds Mace, G.M., A. Balmford & J.R. Ginsberg). - Cambridge U.P., Cambridge, U.K. Fyeldsa, J., M.K. Bayes, M.W. Bruford &MS Roy, (in press) Biogeography and diversification of African forest faunas: implications for conservation. — In: Rainforests — past and future (ed. Moritz, C.).- Chicago U.P., Chicago. Flenley, J. (in press): Palyno-diversity and the super rainforest) — Mm: Rainforests — past and future (ed. Moritz, C.). — Chicago U.P., Chicago. Garcia-Moreno, J., P. Arctander & J. Fjeldsa (1998): Pre-Pleistocene differentiation among chat-tyrants. — Condor 100: 629-640. Garcia-Moreno, J., P. Arctander & J. Ejeldsa (SI Stone diversification at the treeline among Metallura hummingbirds. — Auk 6: 702-711. 43 Garcia-Moreno, J., P. Arctander & J. Fjeldsä (1999b): A case of rapid diversification in the Neotropics: Phylogenetic relationships of Cranioleuca spinetails (Aves, Furnariidae). — Molec. Phyl. Evol. 12:273-281. Garcia-Moreno,J. & J. Fjeldsä (1999): A re-evaluation of species limits in the genus Atlapetes based on mtDNA sequence data. — Ibis 141: 199-207. Garcia-Moreno,J.& J.M.C. Silva (1997): An interplay between forest and non-forest South American avifaunas suggested by a phylogeny of Lepidocopaltes woodcreepers (Dendrocolaptinae). — Studies Neotr. Fauna Envir. 32: 164-173. Graves, G.R. (1980): A new species of metaltail hummingbird from northern Peru. — Wilson Bull. 92: 1-7. Graves, G.R. (1982): Speciation in the carbonated flower-piercer (Diglossa carbonaria) complex of the Andes. — Condor 84: 1-14. Graves, G.R. (1985): Elevational correlates in speciation and intraspecific geographical variation in plumage in Andean forest birds. — Auk 102: 556-579. Graves, G.R. (1988): Linearity of geographical range and its possible effect on the population structure of Andean birds. — Auk 105: 47-52. Grayiocale A) (1998): Is it better to add taxa or characters to a difficult phylogenetic problem? — Syst. Biol. 47: 9-17. Eatekeer S.J. (1995): Molecular” systematics and zoogeography of flowerpiercers in the Diglossa baritula complex. — Auk 112. 156-170. Haffer, J. (1997): Alternative models of vertebrate speciation in Amazonia: an overview. — Biodiver. Conserv. 6: 451-476. hele ce AG. Clark (1997): Principles of population genetics. — Sinauer Ass. 304 pp. Heindl, M. & K.L. Schuchmann (1998): Biogeography, geographical variation and taxonomy of the Andean hummingbird genus Metallura GOULD, 1847. —J.Orn. 139: 425-473. Mmomoeonlemsira, He. Jy. Mielice, A. Berger & N.J. Shackleton (1993): Frequency spectra and paleoclimatic variability of the high-resolution 30-1345 ka FUNZA I pollen record (Eastern Cordillera, Columbia). — Quatern. ser Rey. 12: 141-156. Foo niemsinar THe & Lf: van der Hammen (1998): Neogene and Quaternary development of the neotropical rain forest: the forest refugia hypothesis, and a literature overview. — Earth-Science Reviews 44: 147-183. Hubbell, S.P. (1999): A unified theory of biogeography and biodiversity. — Princeton Univ. Press. Hudson, R.R. (1990): Gene genealogies and the coalescence. — Oxford Surveys Ecol. Biol. 7: 1-44. Lane vie ED Kocher & A.©. Wilson (1991): Evolution of the cytochrome b gene of mammals. — J. Mol. Evol. 32: 128-144. Toms one NK. EV Remsen &’C7 Cicero (in press) Resolution of the debate over species concepts in ornithology: a new comprehensive biologic species concept. — 11 pp in: Proc. 22nd Int. Orn. Congr., Durban. — (SD-ROM, eds. Adams, N.J. & R.H. Slotow). — BirdLife South Africa, Johannesburg. 44 Johnson, K.P. & M.D. Sorensen (1999): Phylogeny and biogeography of babbling ducks (genus: Anas): a comparison of molecular and morphological evidence. — Auk 116: 792-805. Klicka, I. é& RM. Zink) (1997) ihe importanee oT Tecemtmicemacecmim speciation: a failed paradigm. — Science 277: 1666-1669. Krabbe, N. & T.S. Schulenberg (1997): Species limits and natural history of Scytalopus tapaculos (Rhinocryptidae), with descriptions of the Ecuadorian taxa, including three new species. — pp. 74-88 in: Neotropical Ornithology Honoring Ted Parker (ed. Remsen, J.V.). — The American Ornithologists’ Union, Washington D.C. Loescheke, V., J. Tomiuk & S.K. Jain, ‘eds’ (994). Consemanion Genetics. — Birkhauser, Basel. Louette, M. (1992): Barriers, contact zones and subspeciation in central equatorial Africa. — Bull. Brit. Orn. Club Centenary Suppl. 112(A): 209-216. Lynch, J.D. (1988): Refugia. — pp. 311-342 in: Analytical Biogeography (eds Myers, A.A. & P.S. Giller). - Chapman & Hall, London. Maijer, S. & J. Fjeldsä (1997): A new Cranioleuca spinetail from Bolivia, with description of a “leapfrog pattern” of geographical variation in the genus. — Ibis 139: 606-616. Mindell, D.P. & C.E. Thacker (1996): Rates of moleculanveyolmiione phylogenetic issues and applications. — Ann. Rev. Ecol. Syst. 27: 279-303. Nielsen, K.K. & P. Arctander (submitted): Multiple mitochondrial pseudogenes with apparent recombination in a passerine genus. — Mol. Phyl. Evol. O’Neill, J.P. (1992): A general overview of the montane avifauna of Peru. — Mem.Mus. Hist. Nat. UNMSM (Lima) 21: 47-55. Parker, T.A., T.S. Schulenbere et al. (19859, The aviiaunagonmeuac Huancabamba region, northern Peru. — pp. 160-197 in: Neotropical Ornithology (eds Buckley, T.A., S.F. Mercedes, E.S. Morton, R.S. Ridgely & F.G. Buckley).- The American Ornithologists’ Union, Washington, D.C. Patton, J.L. & M.F. Smith (1992): mtDNA phylogeny and Andean mice: a test of diversification across ecological gradients. — Evolution 46: 174-183. Paynter, R.A., Jr. (1970): Subfamily Emberizinae. — pp. 3-214 in: Checklist of Birds of the World, Vol. 8 (ed. Paynter, R.A.). Museum of Comparative Zoology, Cambridge Mass. Räsänen, M.E.,J.S.Salo & R.J. Kalliola (1987): Fluvial perturbance in the western Amazon basin: regulation by long-term sub-Andean tectonics. — Science 283: 1398-1401. Remsen, J.V. (1984a): High incidence of ‘leapfrog’ pattern of geographic variation in Andean birds: implications for the speciation process. — Science 224: 171-173. Remsen, J.V. (1984b): Geographic variation, zoogeography, and possible rapid evolution in some Cranioleuca spinetails (Furnariidae) of the Andes. — Wilson BML, Yor 31525235 Remsen, J.V. & W.S. Graves (1995a): Distribution patternswsand zoogeography of Atlapetes brush-finches (Emberizidae) of the Andes. — Auk 112: 45 210-224. Remsen, J.V. & W.S. Graves (1995b): Distribution patterns of Buarremon brush-finches (Emberizidae) and interspecific competition in Andean birds. — Auk 112: 225-236. Rensch, B. (1929): Das Prinzip geographischer Rassenkreise und das Problem der Artbildung. — Borntraeger, Berlin. Move vViEs see Lhorres- Muna dco: Hertel (1999) Molecular systematics of the South American tit tyrants (genus Anairetes) including A. fernandezianus. — Mol. Phyl. Evol. 11: 67-76. One Views. J5€-M? Silva, P. Aretander, J. Garcia-Moreno & J. Fjeldsa (1997): The role of montane regions in the speciation of South American and African birds. — pp. 325-343 in: Avian Molecular Evolution and Systematics (ed. Mindell, D.P). - Academic Press, London. ScmuanuMv. J. Maley, B. Turg, M-L. Absy, P. Brenac, M. Fournier & M-P. Lendry (1993): Tropical forest changes during the Late Quaternary in African and South American lowlands. — Global and Planetary Change 7: 25-40. sibley, EG. & J.C. Ahlquist (1990): Phylogeny and classification of birds. — Yale Univ. Press, New Haven. Simpson-Vuilleumter, B. (1971): Pleistocene changes in the fauna and flora of South America. — Science 173: 771-780. Smee hak Wayne, D.J. Girman & M.W. Bruford (1997): A role for ecotones in generating rainforest biodiversity. — Science 276: 1855-1857. Stepien,C.A.& T.D. Kocher (1997): Molecules and morphology in studies of fish evolution. — pp. 1-11 in: Molecular systematics of fishes (eds. Kocher, T.D. & C.A. Stepien). — Acad. Press, London. Saez IW. Fitzpatrick, T.A. Parker IIT & D.K. Moskovits (1996): Neotropical Birds. Ecology and Conservation. — Univ. Chicago Press, Chicago. ieomplcton, AR. & B. Read (984): Factors eliminating inbreeding depression in a captive herd of Speke’s gazelle (Gazella spekei). — Zoo Biol. 3: I OR Traylor, M.A. (1985): Species limits in the Ochthoeca diadema species-group (Tyrannidae). — Ornithol. Monogr. 36: 430-442. Volker, G. (1999): Molecular evolutionary relationships in the avian genus Anthus (pipits: Motacillidae). — Mol. Phyl. Ecol. 11: 84-94. Vunhicume@ren, El (1969): Pleistocene speciation in bitds-living in the high Andes. — Science 223: 1179-1180. Vuilleumier, F. (1984a): Zoogeography of Andean birds: two major barriers, and speciation and taxonomy of the Diglossa carbonaria superspecies. — Nat. Geoer Soc. Res. Reports 16: 713-731. Vuilleumier, F.(1984b): Patchy distribution and systematics of Oreomanes fraseri (Aves, Coeribidae) of the Andean Polylepis woodlands. — Am. Mus. Novit. No. 2777. Mimneommilconr, ke cc DD Simbierlotr, De (980); Ecology vs. history as 46 determinants of patchy and insular distributions in high Andean birds. — pp. 235- 379 in Evolutionary Biology, Vol. 12 (eds. Hecht, M.K., W.C. Steere & B. Wallace). — Plenum, New York. Waddell, P.]., Y. Gao, M. Hasesawa &2D7Rr7 Ninde lee GSO: Assessing the Cretaceous superordinal divergence times within birds and placental mammals using whole mitochondrial protein sequences and an extended statistical framework. — Syst. Biol. 48: 119-137. Walter,.G.H. & H.E.H. Paterson .(1992)2 Ihe impiliicanoncere palaeontological evidence for theories of ecological communities and species richness. — Aust. J. Ecol. 19: 241-250. Wijninga, V.M. (1995): A first approximation of montane forest development during the late Tertiary in Colombia. — pp. 23-34 in: Biodiversity and conservation of Neotropical montane forests (eds. Churchill, S.P., H. Balslev, E. Forero & J.L. Luteyn). — The New York Botanical Garden: 23-34. Williams, P.H. (1994): WORLDMAP. Priority areas for biodiversity. Ver. 3. Privately distributed computer software and manual. — The Natural History Mus., London. Zimmer, J.T. (1951): Studies of Peruvian birds No. 60. The genera Heliodoxa, Phlogophilus, Urosticte, Polyplancta, Adelomyia, Coeligena, Ensifera, Oreotrochilus, and Topaza. — Amer. Mus. Novit. 1513: 1-45. Authors’ addresses: Jaime Garcia-Moreno, Museum of Zoology - University of Michigan, 1109 Geddes Ave., Ann Arbor, MI48109-1079, USA. Jon Fjeldsa, Zoological Museum - University of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen, Denmark. 47 Biogeography, geographical variation, and taxonomy of the Neotropical hummingbird genus Polyerata Heine, 1863 (Aves: Trochilidae) André-A. Weller Abstract: Recent studies on distribution and morphological characters of the Neotropical hummingbird genus Amazilia indicate a polyphyletic origin of this taxon, thus contradicting the presently assumed monophyly of Amazilia. One of the proposed subgenera, Polyerata (Heine, 1863), is here regarded as distinct at the generic level. These findings are based on recent biogeography and geographical variation studies. It is hypothesized that the six species of the current Polyerata group are derived from a common ancestor that originated in northern South America. Key words: Amazilia, Polyerata, Trochilidae, biogeography, taxonomy, Neotropics Introduction The genus Amazilia, Trochilidae, is considered by recent reviewers the largest taxon of the hummingbird subfamily Trochilinae, comprising 29 (Peters 1945) to 33 (Wolters 1982) species. Its current distribution ranges from the southern Nearctic region (southwestern USA) to the southern Neotropics (southward to Bolivia, Argentina, and Uruguay). In previous studies, classification was based on general similarity and distributional aspects. Existing morphological patterns within Amazilia have often been neglected, even by Simon (1921) and Peters (1945), either by creating new genera (Simon) or including other species groups (e.g., Polyerata, Saucerottia) as subgenera in the taxon Amazilia (Peters). A recent study (Weller 1998) indicates that several members of the current subgenus Polyerata can be distinguished from Amazilia by plumage patterns and zoogeographic aspects. As a taxonomic consequence | propose that Polyerata represents a valid genus. Material and methods Altogether 1116 specimens of all taxa of Polyerata were compared with those of Amazılia and studied for geographical variation. For the evaluation of colour characters I examined specimens under natural light conditions and by means of a magnifying lamp (x 10). Morphometric data were obtained using a digital caliper (measured to the nearest 0.5 mm). Range maps were drawn based on distributional data derived from skin specimens and literature (Sanchéz Osés 1995). Collecting sites were located using ornithological gazetteers (Paynter 1982, 1985, 1989, Paynter & Traylor 1977, 1981, 1983, 1991, Stephens & Traylor 1985) and diverse travel maps. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 48 Results Taxonomy and distribution of Polyerata The genus Polyerata was initially separated from Damophila (Gould 1861) by Heine (1863), who designated P. amabilis as type species. Later taxonomists, such as Elliot (1878) and Hartert (1900), often placed Polyerata species in the genus Agyrtria. Simon (1921) erroneously included several species in the genera Damophila, Arenella (nom. nov.), and Chionomesa (nom. nov.). More recently, Peters (1945) merged Agyrtria and Polyerata under the latter name as a subgenus of the older taxon Amazilia (Lesson 1843). This classification has been maintained in various subsequent revisions (e.g., Zimmer 1950, Wolters 1982, Sibley & Monroe 1990). My studies of the currently recognized genus Amazilia (Weller 1998) suggest that six species should be separated as members of the genus Polyerata: P. fimbriata, P. lactea, P. amabilis, P. rosenbergi, P. boucardi, and P. luciae. The last three species are monotypic, the others polytypic. Polyerata ranges from interior Central America to southeastern South America. 1. Polyerata fimbriata (Gmelin, 1788) This most widespread, polytypic member of the genus is found exclusively in South America, from eastern Colombia to southeastern Brazil (plate 1). Seven subspecies are recognized (from N to S): P. f. elegantissima (Todd, 1942): N and NW Venezuela, adjacent parts of E Colombia; P. f. fimbriata (Gmelin, 1788): NE Venezuela to Guyana and N Brazil (to Amapä); P. f. apicalis (Gould, 1861): E Colombia at base of Andes and adjacent lowlands; P. f. fluviatilis (Gould, 1861): SE Colombia to NE Ecuador; P. f. laeta (Hartert, 1900): NE Peru; P. f. nigricauda (Elliot, 1878): C and E Brazil from Bahia to Sao Paulo, E Bolivia; P. f. tephrocephala (Vieillot, 1818): SE Brazil from Espirito Santo to Rio Grande do Sul. 2. Polyerata lactea (Lesson, 1829) Besides P. fimbriata, this species is the only one that occurs exclusively east of the Andes in South America. Unlike P. fimbriata it shows a highly scattered mosaic-like distribution, mainly south of the Amazon basin (plate |). Three subspecies are recognized (from N to S)): P. I. zimmeri (Gilliard, 1941): tepuis of SE Venezuela; P. |. bartletti (Gould, 1866): tropical zone from E Ecuador to N Bolivia; P. |. lactea (Lesson, 1829): C and SE Brazil. Plate 1: Distribution of Polyerata species in South America based on data obtained from skin specimens. Upper left: P. rosenbergi; upper right: P. fimbriata; lower birds: P. lactea; female (left) and male (right) Plate 2 (next page): Distribution of Polyerata species in Central and northwestern South America based on data obtained from skin specimens. Upper birds: P. /uciae; mid birds: P. boucardi; lower birds P. amabilis; female (left) and male (right) = 100071 a ~ 7 2 a 10 Lo Tos 49 3. Polyerata rosenbergi (Boucard, 1895) This monotypic species 1s confined to the tropical lowland rain forests from W Colombia to NW Ecuador (plate 1). 4. Polyerata amabilis (Gould, 1851) P. amabilis is found both in Central and South America (plate 2). Several reviewers have regarded the race decora as a distinct allospecies (e.g., Wetmore 1968, Stiles et al. 1989, Sibley & Monroe 1990). The race costaricensis (Todd 1942) was described from the northwestern range of the species but is not recognized here as colour variations in the dorsal plumage occur in the morphologically heterogeneous nominate form. Two subspecies are recognized (from N to S): P. a. amabilis (Gould, 1851): NE Nicaragua to NE Colombia and W Ecuador; P. a. decora (Salvin, 1891): SW Costa Rica to W Panama (Chiriqui). 5. Polyerata boucardi (Mulsant, 1877) P. boucardi is a monotypic endemic of Costa Rica, restricted to mangrove areas along the Pacific Coast from Gulf of Nicoya to Rio Grande de Terraba (plate 2). 6. Polyerata luciae (Lawrence, 1867) This monotypic taxon occurs only in thorn forests of N and C Honduras (plate 2) where it has become a vulnerable species. Morphology and geographical variation Variation in plumage characters All members of the genus share the golden to bronze-green dorsal coloration and tail, a glittering gorget or throat patch and a grayish to whitish abdomen. In part, these colour patterns differ strongly from those of other species groups within the Amazilia complex. For example, species of the Amazilia tzacatl group (including, e.g., A. amazilia, A. rutila, A. yucatanensis) have a rufous tail and a rufous to brownish- grayish abdomen. Furthermore, birds of the Agyrtria group (e.g., A. leucogaster, A. franciae) show whitish underparts and more or less contrasting, often glittering caps. Sexual dimorphism in Polyerata species is slight; females usually have more whitish fringes or discs on the throat. P. fimbriata shows the strongest geographical variation within the group, with a gorget that is glittering golden (northern, eastern subpopulations) to bluish green (western subpopulations fluviatilis, laeta). Unlike P. fimbriata, P. lactea has a violet- blue gorget. Both taxa have in common the pure white center of belly and abdomen (reduced in P. /. bartletti). P. amabilis and P. rosenbergi are violet-blue only on the throat, a character reduced in females. The abdominal region is grayer. As apomorphic characters, P. amabilis has a golden to bronze-green glittering crown and a bronzish chin. A different throat coloration occurs in P. boucardi (golden green) and P. luciae (turquoise-green). The abdomen is more grayish white and there are light green (boucardi) to grayish green rectrices (luciae). 50 Morphometric variation Polyerata spp. are medium-sized trochilids (9-11 cm, 4-7 g) with straight or slightly decurved, medium-length bills (20-25 mm). The South American taxa P. fimbriata and P. lactea are morphometrically very similar. Within P. fimbriata, the bill length increases from eastern to western subpopulations in the Amazon basin (P. f. fimbriata, P. f. nigricauda: 19-21 mm; P. f. apicalis, P. f. fluviatilis: 23-24 mm; averages for males). P. f. tephrocephala is the largest form, with a bill length of 23 mm and a wing length of 59 mm (males). In P. lactea, the wing is longer in P. I. bartletti and P. 1. zimmeri (54-57 mm, males) than in the nominate race (52-53.5 mm). P. lactea and P. amabilis are similar in measurements. Although P. amabilis and P. rosenbergi occur in sympatry, they differ in bill length (P. a. amabilis: 18.5-19.5 mm; P. rosenbergi: 19-21.5 mm, males). However, southern subpopulations of the latter have shorter bills. In P. amabilis, P. a. decora has a significantly longer bill (22.5-24 mm, both sexes) than the nominate race. The Centra] American taxa P. boucardi and P. luciae differ from P. amabilis chiefly in length of outer rectrices (males of P. boucardi/P. luciae: 34-34.5 mm; P. a. amabilis/P. a. decora: 29-30.5 mm) which are longer than the inner ones by 6.0 mm in P. boucardi/P. luciae vs. 1-3.5 mm in P. amabilis. Moreover, P. luciae has a longer bill (22 mm, males) than P. boucardi (20 mm, males). Discussion When considering biogeographical aspects of the Neotropical fauna and flora, it has often been suggested that refuges played an important role in speciation processes (e.g., for birds see Haffer 1967, 1969, 1979; for butterflies, Brown & Mielke 1972; for plant families, Prance 1973). Alternatively, several reviewers found that range patterns of some bird species are correlated with river systems, implying that rivers are potential barriers to gene flow (riverine theory; e.g., Hellmayr 1910, Sick 1967, Salo et al. 1986). Based on distributional data and morphological patterns in Polyerata, I suggest elsewhere (Weller 1998) that the evolutionary center of the genus 1s in South America. Morphometric and colour characters indicate that plesiomorphic characters are found in P. fimbriata and P. lactea. Presumably, both taxa are derived from a common ancestor by ecological segregation (tropical vs. subtropical habitats) prior to the Pleistocene. Probably P. fimbriata evolved in rain forest areas north of the Amazon (Guyana shield), and P. lactea in semideciduous subtropical forest of southeastern Brazil or the Andes. Subsequently, the radiation of P. fimbriata was bidirectional as subpopulations invaded the western Amazon basin and Venezuela, and others reached central (P. f. nigricauda) and eastern Brazil (P. f. tephrocephala). It seems possible that either the development of the Amazon river system or Pleistocene flooding of the lower Amazon basin separated the northern and southern subpopulations (compare Willis 1969); however, they have come in secondary contact along the river more recently (zone of intergradation of P. f. fimbriata / f. nigricauda). If P. lactea originated in southeastern Brazil, westward emigration could have resulted in invasion of the submontane zone of the Andes (Peru, Bolivia: P. I. 5] bartletti) and the Pantepui region of southeastern Venezuela (P. /. zimmeri). As P. 1. bartletti is a very divergent form in colour, and P. I. zimmeri similar to P. I. lactea, there could have been “leapfrog” evolution (Remsen 1984). Considering the origin of the taxa west of the Andes and in Central America, I hypothesize that this group was derived from the eastern Andean P. lactea population. Based on present-day ecological requirements, a proto-amabilis-rosenbergi associated with montane rain forest could have crossed the Andes either in S Colombia or S Ecuador. Subsequently, the precursors of both taxa were isolated in different refuges (P. rosenbergi: Choc6 Center; P. amabilis: Panama Center; see Müller 1973). For P. amabilis decora, it is postulated that it developed in a Pleistocene submontane forest refuge of Cordillera de Talamanca. On the basis of morphological similarities, the P. boucardi-luciae group possibly evolved from P. amabilis during the mid-Pleistocene. Strong habitat differences (P. boucardi: mangroves; P. luciae: thorn forest) in the current taxa indicate ecological segregation of the boucardi-luciae precursor populations that, however, once may have been connected due to greater habitat tolerances. The present, rather relictual populations of both species may be caused by man-made influences (e.g., habitat fragmentation) as well as interspecific competition (amabilis-boucardi). As a consequence, P. boucardi and P. luciae are two of the rarest and most endangered trochilids of Central America (Collar et al. 1992). Acknowledgements I thank the curators and staff of the following research institutes and museums for supporting my studies: American Museum of Natural History, New York (AMNH); The Academy of Natural Sciences, Philadelphia (ANSP); The Natural History Museum, Tring (formerly British Museum [Natural History], BMNH); California Academy of Sciences, San Francisco (CAS); The Carnegie Museum of Natural History, Pittsburgh; Delaware Museum of Natural History, Wilmington (DMNH); Field Museum of Natural History, Chicago (FMNH); Natural History Museum of L. A. County, Los Angeles (LACMNH); Museum of Natural Science, Louisiana State University, Baton Rouge (LSUMNS); Museu de Biologia Mello-Leitao, Santa Teresa (MBML); Museum of Comparative Zoology, Cambridge, Mass. (MCZ); Moore Laboratory of Zoology, Occidental College, Los Angeles (MLZ); Museu Nacional, Rio de Janeiro (MN); National Museum of Natural History, Smithsonian Institution, Washington, D.C. (NMNH); Forschungsinstitut Senckenberg, Frankfurt/Main (SM); Museum of Zoology, University of Michigan (UMMZ); Museu de Zoologia, Universidade de Sao Paulo, Sao Paulo (USP); Western Foundation of Vertebrate Zoology, Camarillo (WFVZ); and Alexander Koenig Research Institute and Museum of Zoology, Bonn (ZFMK). References Boucard, A. (1895): Genera of humming birds. — Humming Bird 5: 1-402. Brown, K.S., Jr. & O.F.F. Mielke (1972): The Heliconians of Brazil (Lepidoptera: Nymphalidae). Pt. II. Introduction and general comments, with a supplementary revision of the tribe. — Zoologica, N. Y. Zoological Society. a2 Collar, N.J., E.B. Gonzaga N. Krabbe? Ay Miadirono Nieto eier Naranjo, L.A. Parker, IT SD. Weg ID threatened bindsrermautne Americas: The ICBP/IUCN Red Data Book, third edition, part 2. — International Council for Bird Preservation, Cambridge, U.K. Elliot, D.G. (1878): A classification and synopsis of the Trochilidae. — Smithson. Contrib. Know. 317: 1-277. Gilliard, E.T. (1941): The birds of Mt. Auyan-Tepui, Venezuela. — Bull. Amer. Mus. Nat. Hist. 77: 439-508. Gmelin, J.F. (1788): Syst. Naturae, ed. 13, 1: 448. Gould, J. (1851): On some new species of Trochilidae. — Proc. Zool. Soc. London 19: 115-116. Gould, J. (1861): An introduction to the Trochilidae, or family of humming-birds. — London. Gould, J. (1866): pp 175-201 in: Catalogue of birds collected by Mr. E. Bartlett on the River Ucayali, Eastern Peru, with notes and descriptions of new species (eds. Sclater, P.L. & O. Salvin). — Proc. Zool. Soc. London 34. Haffer, J. (1967): Speciation in Colombian forest birds west of the Andes. — Amer. Mus. Novit. 2294: 1-57. Haffer, J. (1969): Speciation in Amazonian forest birds. — Science 165: 131-137. Haffer, J. (1979): Quaternary biogeography of tropical lowland South America. — pp. 107-140 in: The South American Herpetofauna: its origin, evolution, and dispersal (ed. Duellman, W.) — The University of Kansas, Lawrence. Hartert, E. (1900): Trochilidae. Das Tierreich. 9. Lieferung. — R. Friedlander & Sohn, Berlin. Heine, F. (1863): Troechilidiea =a. Omithol 12 723277 Hellmayr, C.E. (1910): The birds of the Rio Madeira. — Novit. Zool. 17: 257-428. Lawrence, G.N. (1867): Descriptions of six new species of birds of the families Hirundinidae, Formicariidae, Tyrannidae, and Trochilidae. — Ann. Lyc. New York 8: 400-405. Lesson, R.P. (1829): Histoire naturelle des oiseaux-mouches. — Arthus Bertrand, Paris. Lesson, R.P. (1843): Echo du Monde Sav., col. 757. Müller, P. (1973): The dispersal centres of terrestrial vertebrates in the Neotropical realm. A study in the evolution of the Neotropical biota and its native landscape. — Biogeographica 2: 1-244. Mulsant, E. (1877): Descriptions des especes nouvelles des Trochilidés. — Ann. Soc. Linn. Lyon: 6. Paynter, R.A., Jr. (1982): Ornithological Gazetteer of Venezuela. — Mus. Comp. Zool., Cambridge, Mass. | Paynter, R.A., Jr. (1985): Ornithological Gazetteer of Argentina. - Mus. Comp. Zool., Cambridge, Mass. 53 Paynter, R.A., Jr. (1989): Ornithological Gazetteer of Paraguay. 2nd edition. — Mus. Comp. Zool., Cambridge, Mass. Paynter, R-A., Jr. & M.A. Traylor, Jr. (1977): Ornithological Gazetteer of Ecuador. — Mus. Comp. Zool., Cambridge, Mass. Paymeen, RAS, Jn. & M-A. Traylor, Jr. (1981): Ornithological Gazetteer of Colombia. — Mus. Comp. Zool., Cambridge, Mass. Boyer, Ren. Jn. & M.A. Traylor, Jr. (1983): Ormithological Gazetteer of Peru. — Mus. Comp. Zool., Cambridge, Mass. Besen ho A. In. & MA, Wraylor, Jr. (1991): Omithological Gazetteer of Brazil. — Mus. Comp. Zool., Cambridge, Mass. Peters, J.L. (1945): Check-list of birds of the world. Vol. 5. - Cambridge, Mass. Prance, G.T.(1973): Phytogeographic support for the theory of Pleistocene forest refuges in the Amazon basin, based on evidence from distribution patterns in Caryocaraceae, Chrysobalanaceae, Dichapetalaceae and Lecythidaceae. — Acta Amazonica 3: 5-28. Remsen, J.V. (1984): High incidence of “leapfrog” patterns of geographic variation in Andean birds: implication for the speciation process. — Science 224: 171-173. BOP eke aliiola, IT Hakinnenm. Y° Mäkinen, P. Niemelä, M. Pukakka, & P.D. Coley (1986): River dynamics and the diversity of Amazon lowland forest. — Nature 322: 254-258. Sanchéz Osés, C. (1995): Distribuciön geogräfica de la familia Trochilidae (Aves: Apodiformes) en Venezuela con algunas implicaciones biogeogräficas. — Unpubl. Master’s thesis, Universidad Central de Venezuela, Caracas. Salvin, O. (1891): Descriptions of new species of Upupe and Trochili in the collection of the British Museum. — Ann. Nat: Hist. 7: 374-378. Sibley, C.G. & B.L. Monroe (1990): Distribution and taxonomy of birds of the world. — Yale University Press, New Haven. Sick, H. (1967): Rios e enchentes na Amazonia como obstaculo para a avifauna. — pp. 495-520 in: Atas do simpösio sobre a biota Amazönica, 5 (Zool.) (ed. Lent, H.). — Conselho Nacional de Pesquisas, Rio de Janeiro. Simon, E. (1921): Histoire naturelle des Trochilidés (synopsis and catalogue). — Encyclopédie Roret, L. Mulo, Paris. Seeless i G., AF. Skutch & D. Gardner (1989): A guide to the birds of Costa Rica. — Christopher Helm. Bromley, London. Biephens, E. & M.A. Traylor, Jr.(1985): Ornithological Gazetteer of the Guianas. — Mus. Comp. Zool., Cambridge, Mass. Todd, W.E.C. (1942): List of hummingbirds in the collection of the Carnegie Museum. — Ann. Carnegie Mus. 29: 271-370. Vieillot, J.L.P. (1818): Nouvelle dictionnaire d histoire naturelle, 13. — Paris. Weller, A.-A. (1998): Biogeographie, geographische Variation und Taxonomie der Gattung Amazilia (Aves: Trochilidae). — Unpubl. doctoral thesis, Rhein. Friedrich- Wilhelms-Universität Bonn. 54 Wetmore, A. (1968): The birds of the Republic of Panama. Part 2. — Columbidae (Pigeons) to Picidae (Woodpeckers). — Smithsonian Institution Press, Washington, D.C. Willis, E.O. (1969): On the behavior of five species of Rhegmatorhina, ant- following birds of the Amazon basin. — Wilson Bull. 81: 363-395. Wolters, H.E. (1982): Die Vogelarten der Erde. — Parey, Hamburg. Zimmer, J.T. (1950): Studies of Peruvian birds. No. 59. The genera Polytmus, Leucippus, and Amazilia. — Amer. Mus. Novit. 1475: 1-27. Author’s address: Dr. André-A. Weller, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 150-164, 53113 Bonn, Germany 58 Taxonomic considerations on the genus Orestias Valenciennes, 1839 (Teleostei: Cyprinodontidae) of the Chilean Altiplano Wolfgang Villwock, Ulrike Sienknecht & Arne Lüssen Abstract:The present distribution of the teleost genus Orestias Valenciennes, 1839, endemic to the Altiplano, can be explained by the extent of the lake and river systems during the Plio- and Pleistocene. The last main revisions of the genus defined 20 and 43 species respectively. Investigations by Villwock and co-workers have shown that it is rather unlikely that such a number of species might have evolved under the given historic, geographic and ecological conditions. Additionally, some of the characters used for the separation of the species are highly varıable. Recently, crossbreeding experiments with four Orestias-species were started to clarify their intrageneric relationships. All F, are fertile and as far as allready produced - the further offspring too. According to all aspects and results of the different investigations the four “species” of the Chilean Altiplano belong to one species, Orestias agassizii Cuvier & Valenciennes, 1846. Key words: Orestias, Orestias-agassizii assemblage, crossbreeding experiments, Altiplano, South America Introduction The cyprinodontids of the genus Orestias Valenciennes, 1839 are the only endemic and autochthonous teleosts inhabiting the remnants of ancient water bodies in the Altiplano of Pert, Bolivia and Chile, with its center in recent Lake Titicaca (fig.1). Tchernavin (1944) ranked the genus Orestias as an own subfamily (family Cyprinodontidae) while Costa (1997) reduced it to a tribe Orestiini within the subfamily Cyprinodontinae. We prefer the differentiation suggested by Tchernavin because (1) he published the last, in many aspects most convincing revision of the Orestias species-group, and (2) because the question of the real interrelationship with other cyprinodontid units seems to be unsolved and open to further discussion — despite Costa’s very thorough morphological and morphometric investigations. According to our view, Costa’s phyletic and biogeographic conclusion of three different, equally ranked tribes also allows the postulation of three different subfamilies (see fig.2) — a point of view not that far from Tchernavin’s and our own. This view, however, is contrary to the opinion of Parenti (1984) or Parker & Kornfield (1995), who believe that a sister-group relationship exists between Orestias and Anatolian representatives of Old World Aphaniini. We will not discuss here in detail why we are sceptical of Parenti’s hypothesis. However, her postulation of a total of 43 different species of Orestias (24 of them solely within the borders of recent Lake Titicaca!) is not feasible, neither on the basis of many of the characters she used (c.f. comments by Müller 1993, Villwock & Sienknecht 1993) nor of events in the historic development from ancient to recent Altiplano Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 L. Chungara oO ei ©. = © = » a. ® o a 7 a = a ; Salar de Uyuni ess Lagos / Lakes (recent) Salares / Salt lakes (recent) Hig EM Eh et t IN .. ig.1: Map of the ancient water Lago Ballivian (Pliocene Wes ; ; : body and ist recent remnant in the Altiplano (draft by BSG ae ca. ani é Sienknecht, PC-drawing by Stiewe) Lago Minchin (Pleistocene) water bodies (see Moon 1939, Newell 1949, Ahlfeld & Branisa 1960, Zeil 1979, 1984, Wörner et al. 1988, Kött et al. 1995) (fig.1). Whichever is true, the orestiine cyprinodontids are as a whole one of the “isolated vertebrate communities in the topics (seele 2E) We will concentrate on some aspects resulting from own investigations on a geographically restricted group of Orestias in the Chilean part of the Altiplano which has been said to belong to four different species: O. parinacotensis, O. laucaensis (Arratia 1982), O. ascotanensis (Parenti 1984) and O. chungarensis (Vila & Pinto 1986) (fig.3; fig. 4: points 1, 2, 3 and 7). In 1991, Villwock and Sienknecht collected fishes of three additional populations: from the rivers Isluga (fig.4: 4) and Collacagua (fig.4: 5) and from springs at the eastern border of the Salar de Huasco (fig.4: 6). These new points are located between - but not actually connected with - the three northern “species” and “O. ascotanensis” in the south of the Chilean Altiplano. These populations (and probably more may exist) were only named according to their locations, and not described as new species, because of reasonable suspicion that they might be the Fig.2: Assumed phylogeny of Cyprinodontinae (after Costa 1997): A: Old World Aphaniini - B: New World Cyprinidontini — C: New World Orestiini nearest relatives of an already known species: Orestias agassizii' Cuvier & Valenciennes, 1846, including the four above mentioned “species”. This taxonomic em Fig.3: Typical representatives of Chilean Orestias-populations; above: female from Rio Lauca; below: male from Rio Isluga. (Drawings by Ruud Wildekamp, Gemert, The Netherlands) ') According to Eschmeyer (1998) the correct name should be "agassizii", whereas the most commonly used name "agassii" (introduced by Tchernavin 1944) is a typesetting error 58 concept derived from the common geological history (see above) and the supposed young separation of their present habitats in the Chilean Altiplano, i.e. own suggestions of an approximates age of separation of less than 15 000 years B.P. (Villwock et al. 1987; based on '*C-aging reported by Geyh, Niedersächsisches Landesamt für Bodenforschung, March 1986: 11 730+100 ys). This was meanwhile confirmed by geological investigations, e.g. Francis & Self (1987): 15 000 to 17 000 ys, Harmon et al. (1987): < 15 000 ys, Wörner et al. (1988): 13 500 ys and others. Additional results concerning the age of Chilean Altiplano lakes and their fauna were published by Villwock & Sienknecht (1993, 1995, 1996) and by Schwalb et al. (1999). The results of the comparison of the four species descriptions given by Arratia, Parenti and Vila & Pinto (see above), with observations from our own investigations (including the thesis data by Thomann 1990), are presented in the following paragraphs. Rio Desaguaderg BOLIVIA Fig.4: The distribution of Chilean Orestias (Draft Sienknecht, PC-drawing Stiewe) A - Lago Titicaca; B - Lago Aca A Poopo; C - Salar de Surire; D - Iquique 2 Salar de Coipasa; E - Salar de Uyuni; F - Salar d’Ascotän 1 - Lago Chungarä (LC); 2 - Laguna de Parinacota (LP); 3 - Rio Lauca (RL); 4 - Rio Isluga (RI); 5 - Rio Collacagua (RC); 6 - Salar de Huasco (SH); 7 - Salar d’Ascotän (SA) 59 Morphometric and meristic characters Head-length / standard-length Thomann (1990) reports on some basic data such as the main morphometrics (body- indices) and meristic characters (number of fin rays in the dorsal and anal fins) which he found in a total of 858 specimens, among them 419 of the population of Lago Chungara (fig.2:1), 122 of Parinacota (fig.2:2), 172 of Lauca (fig.2:3), 25 of Huasco (fig.2: 6) and 110 of Salar d’Ascotän (fig.2:7). One of the most commonly used metric characters for discriminating different species is that of the coefficient between head-length and standard-length (see fig.5: B/A). Fig.5: Distances to be measured (after Lauzanne 1981) These indices do not show any significant differences between these five compared localities (Thomann 1990: 63, table 14). Other metric data, such as body height (fig.5: F) or body width (fig.5: G) are not very reliable because they depend on nutritional condition, stage of maturity (differing between female and male) and many other physiological conditions of the studied specimens. The main morphometric similarities of the four “species” and the population of Salar de Huasco were calculated from random samples of 25 specimens of each population. The results led to the dendrogram shown below, reflecting the narrow relationship between the Chilean Orestias tested (fig.6): 0.12 0.10 0.08 0.06 0.04 CHUNGARA (LC) — PARINACOTA ( P) — _ LAUCA (RL) i : : ee ays ASCOTAN (SA) Fig.6: Morphometric similarities eee 22 YUASCO-(SH) (Thomann 1990) Moreover, the average values of Thomann correspond to those given by Lauzanne 1982: 42) and Loubens & Sarmiento (1985) concerning O. agassizii from Lago Pequenio (the smaller part of Lake Titicaca) and the Lago Grande (main Lake Titicaca). This correspondence shows a close relationship in these important characters between Orestias agassizii and the five cited Chilean populations, an assumption that was first made by Villwock & Thomann (1987) and again by Villwock & Sienknecht (1993), based on their own numerous measurements. 60 Finrays in the dorsal and anal fins A similar result is obtained from the comparison of meristic characters, e.g. the variability of finrays in the dorsal and anal fins, which do not differ significantly neither within the four northern Chilean species® (Gnelusive the Huascor population: Thomann 1990) nor between the latter and O. agassizii from different parts of its range (Lauzanne 1982 and own collections: Villwock & Sienknecht 1993). The average values of the dorsal fin (D) range from 13.3 (Lauca) to 15.2 (Chungarä); the average values for the anal fin (A) range from 13.5 (d’Ascotän) to 15.0 (Chungara and Huasco). These values are again within the normal distribution of O. agassizii sensu Lauzanne (1982). Scales and scalation The number of scales and the degree of scalation are no useful characters for species discrimination since scale reductions occur in all orestiine cyprinodonts. Moreover, scales show a high variability in counts (and size) within the same popu- a. em Fig.7: Different stages of scale development (alizarine-S- c. staining) a: TL 18-19mm; b: TL 20- 2lmm; c: TL 25-26mm (Lüssen 1998) 61 lation, as it is generally known for comparable characters that undergo reductions (Villwock 1963). Therefore already Tchernavin (1944) stated correctly that “the number of scales in Orestias varies widely and a difference of two or three scales obtained in examining a few specimens cannot be considered specific” (p. 148). In cases of irregularities of the scalation, another aspect should also be considered: Lüssen (1998) demonstrated that the number of scales in the longitudinal line, and especially those in the transversal line (corresponding with differences of the extent of naked parts of the body), depend directly on growth (and age) (figs.7a-c). Accordingly, Tchernavin’s statement should be completed by the reservation that in cases of reduced scalation only specimens of about the same size should be used for counts of scales. If specimens of the same size are not available this character should not be used for species discrimination at all. Aspects of growth, manifestation of characters and sexual maturation As just pointed out, some of the morphometric and meristic characters depend directly on the ontogenetic stage. However, this is not only true with respect to reduction characters (as scales and scalation) but also to sexual maturity. Normally, sexual maturity 1s or may be used as synonym for being “full grown” or “adult”. This statement may work in general, but does not apply in the genus Orestias: sexual maturity will be reached in the Chilean group at an age of about 9 months, their full-grown stage however at not below two years (controlled in the aquarium). Also, within this period and phenotypical variation the scalation will continue to become individually completed. In other words, sexual maturity of the specimens and their special scalation characters cannot be used for species discrimination. This appears to be one of Parenti’s (1984) chief errors when she created at least some of her new species for Lake Titicaca (see also Müller 1993). Feeding habits as a tool for species distinction Vila & Pinto (1986) and Pinto & Vila (1987) were asking in the context of their species description for special feeding habits; however they could not state significant differences in food consumption. Those differences described by Thomann (1990) obviously depended on seasonal differences in food availability. There is no doubt that the Chilean Orestias populations consist of omnivorous specimens — as does O. agassizii (Villwock (1966), Lauzanne (1982), Loubens & Sarmiento (1985). Genetic aspects in species discrimination Number of chromosomes The first counts were reported by Lueken (1962), who found in samples of O. luteus and O. agassizii the haploid number of n=24 (2n=48). Arratia (1982) found the same result with respect to O. parinacotensis. The other new species described by Arratia, O. laucaensis, has 2n=51 chromosomes in females and 2n=52 in males — the discriminating feature between the two species. In general, it is not very rare in fishes that males and females differ by one chromosome; however, normally the males have the lower counts: e.g., goodeid males 2n=41, females 2n=42 (Uyeno & Miller 1972, Uyeno et al. 1973). According to general chromosomal investigations on fishes (Makino 1951) and many 62 comparative studies since then (e.g., Nogusa 1960, Roberts 1967, Gyldenholm & Scheel 1971, Foerster & Anders 1977, Rab et al. 1987, Vitturi et al. 1996), including some contributions with special concern to oviparous cyprinodonts (cf. Karbe 1961, Gold et al. 1980, Vitturi et al. 1995) the by far most common karyotype in fishes is 2n = 48 chromosomes. Preliminary enzyme investigations In addition to the facts discussed above, Thomann (1990) initiated the first electrophoretic investigations on the Chilean Orestias from Lago Chungarä, Parinacota and Rio Lauca (see fig.4), using tissue homogenates, separated in vertical starch gels. Twelve enzymes were assayed (24 loci): on the basis of these results (see Thomann 1990: table 29, p.99) the author calculated the genetic similarities according to Nei (1972), illustrated in the following scheme and dendrogram: | Lago Chungara/LC Parinacota / P Rio Lauca / RL EC 24) = 6.000 R 0.011 0.000 RL 0.006 0.013 0.000 0.025 0.02 0.015 0.01 0.005 0.000 CHUNGARA _—— 1m LAUCA SS ele NO Fig.8: Evaluated genetic distances according to Nei (1972) (above) and dendrogram of genetic similarities (below) (Thomann 1990) According to this study, the genetic distance between the three tested Chilean Orestias, O. chungarensis, O. parinacotensis and O. laucaensis is even beyond the level of local populations and therefore should belong to the same species. Crossbreeding for valid species detection (sensu Mayr 1967) Because of lack of confirmation in some genetic aspects (see above) crossbreeding experiments between all of the available Chilean populations (including the four dealt with here “species”; see fig.4: 1-7) were started some years ago. According to Mayr’s concept of valid species (“biospecies concept”), hybrid fertility indicates that the tested forms will belong to the same species while hybrid sterility shows the borders between true or valid species. This statement concerning hybrid fertility is - as well known - regularly applicated to hybrids which occur under natural conditions. However, cases as Orestias or as many other Old und New World oviparous cyprinodonts, which interspecifically do neither show ecological nor (courtship) behavioural differences under natural conditions, do not present considerable reasons for excluding their fertile “species” hybrids from Mayr’s biospecies concept — only because they are products of controlled aquarium crossbreeding. This opinion is basically confirmed by the 63 comparison of identic results in histological gonad structures of F,-hybrids - natural vs. aquarium hatched ones - of two Mediterranian Aphanius species (cf. Villwock 1985, 1987). The above mentioned crossbreeding in Chilean Orestias resulted without any exception in fertile F,-hybrids, which led to numerous F, and already F, offspring. This is one additional argument that the different Chilean Orestias populations belong to the same species, which is according to our suggestions O. agassizil. Conclusions The results of the various investigations that have been referred to above lead to the following conclusions: (1) Evidently all Chilean Orestias populations belong to one single species. (2) This species is presumed to be identical with Orestias agassizii, the most widely spread species within the distribution patterns of the genus Orestias and the one with the broadest ecological tolerance. This assumption is based on the high degree of similarity, or highly corresponding characters as illustrated in morphometric and meristic evaluations and is confirmed by results of genetic investigations. A similar interpretation has already been published by Mann (1954), and even Parenti (1984) ranks the Chilean “species” in an Orestias agassizii complex. (3) The assumption that the species belong solely to one single unit which is identical with ©. agassizii is confirmed by aging the given geological facts concerning the development of the present water bodies inhabited by the Chilean Orestias populations concerned, including the distribution patterns of Orestias agassizii. Geographic distance and time since isolation of this tropical fish community agree with this concept. Discussion remarks at the Symposium Question: Have you any information on longtime population density under natural conditions? Answer: No, no precise ones, however there is no indication that the population size will differ very much from year to year. The Chilean populations remain seemingly on a numerical high level (about thousands / population). Question: Are there obvious differences between observations in nature vs. aquaria controlled strains concerning ontogenetic stages? Answer: No, as far as we know. References Ahlfeld, F. & L. Branisa (1960): Geologia de Bolivia. — Instituto Boliviano eeueciuoleo: ya Paz: Arratia, F.G. (1982): Peces del Altiplano de Chile. — pp. 93-133 in: El hombre y os ecosistemas de Montana. (ed. UNESCO). — MAB 6. El ambiente natural y Las poblaciones humanas de los Andes del Norte Grande de Chile (Arica, Lat. 18°28'S). Vol. I. La vegetation y los vertebrados inferiores de los pisos altitudinales entre Arica y el Lago Chungara. Santiago de Chile. Costa, W.J.E.M. (1997): Phylogeny and Classification of the Cyprinodontidae revisted (Teleostei: Cyprinodontiformes): Are Andean and Anatolian killifishes sister taxa? — J.Comp.Biol. 2(1): 1-16. 64 Eschmeyer, W.N. (1998): Catalog of fishes. — http.//www.fishbase.org/ Eschmeyer. Foerster, W. & F. Anders (1977): Zytogenetischer Vergleich der Karyotypen verschiedener Rassen und Arten lebendgebährender Zahnkarpfen der Gattung Xiphophorus. — Zool. Anz. 198: 167-177. Francis, P. & S. Self (1987): Verheerende Flankenabbrüche an Vulkanen. — Spektrum Wiss. 8: 86-93. Gold, J.R., W.J. Karel & M.R. Strand (1980): Chromosome formulae of North American fishes. — Progr. Fish-Cult. 42: 10-23. Gyldenholm, A.O. & J.J. Scheel (1971): Chromosome number of fishes. — J. Fish Biol. 3: 479-486. Harmon, R.S., J. Davidson, G. Wörner, ©. Wopez [FE sc obama Moreno, N.J. McMillan, D. Turner & €. Nye (19700037 chronology, geochemistry, and isotope character of the Nevados des Payachata volcanic group, N-Chile. — Act. X. Congr. Geol. Argentino, Tucumän, Argentina 4: 344-345. Karbe, L. (1961): Cytologische Untersuchung der Sterilitätserscheinungen bei anatolischen Zahnkarpfen, ein Beitrag zum Speziationsproblem. — Mitt. Hamburg. Zool. Mus. Inst. 59: 73-104. Kött, A., R. Gaupp & G. Wörner (1995): Miocene to Recent history of the western Altiplano in northern Chile revealed by lacustrine sediments of the Lauca basin. — Geol. Rundschau 84: 770-780. Lauzanne, L. (1981): Description de trois Orestias nouveaux du Lac Titicaca, O. ispi n.sp., O. forgeti n.sp. et O. tchernavini n.sp. (Pisces, Cyprinodontidae). — Cybium, 3 DEUS, Dy 38 791 Lauzanne, L. (1982): Les Orestias (Pisces, Cyprinodontidae) du Petit lac Titicaca. — Rev. Hydrobiol. trop., Bondy 15(1): 39-70. Loubens, G. & J. Sarmiento (1985): Observations sur les poissons de la partie Bolivienne du lac Titicaca II. Orestias agassii Valenciennes, 1846 (Pisces, Cyprinodontidae). — Rev. Hydrobiol. trop., Bondy 18(2): 159-171. Lueken, W. (1962): Chromosomenzahl bei Orestias. — Mitt. Hamburg. zool. Mus. Inst. 60: 195-198. Lüssen, A. (1998):Vergleichende Untersuchung zur frühen Ontorenmte neuweltlicher Zahnkarpfen der Gattung Orestias VAL., 1839 (Teleostei, Cyprinodontidae) unter besonderer Berücksichtigung der Schuppenanlagen. — Diplom- Arbeit, Fachbereich Biologie, Universität Hamburg: 51 pp. Makino, S. (1951): An atlas of the chromosome numbers in animals. 2. ed. — The Iowa State College Press. Ames, Iowa, 27. Mann, G. (1954): Vida de los Peces en aguas chilenas. — Universidad de Chile: 195 umd 199 Mayr, E. (1967): Artbegriff and Evolution, Parey, Hamburg (lst ed.: 1963: “Animal species and Evolution”, Harvard University Press, Cambridge, Mass.). Moon, H.P. (1939): III. The geology and physiography of the Altiplano of Peru and Bolivia. — Trans. Linn. Soc. London 1: 27-43. 65 Miller, R. (1993): Kritische Anmerkungen zur Revision der Gattung Orestias (Pisces: Cyprinodontidae) durch Parenti aus dem Jahr 1984. — Zool. Jb., Abt. Anatom. Ontogen. Tiere. G. Fischer, Jena. Nei, M. (1972): Genetic distance between populations. — Am. Nat. 106: 283-290. Newell, N. (1949): Geology of lake Titicaca region, Peru and Bolivia. — Geol. Soc. Am., Memoir, 36: 10-104. Nogusa, S. (1960): A comparative study of the chromosomes in fishes with particular considerations on taxonomy and evolution. - Mem. Hyogo Univ. Agric. 3(1), Biol. Ser. 62 pp (app.). Parenti, L.R. (1984): A taxonomic revision of the Andean killifish genus Orestias (Cyprinodontiformes, Cyprinodontidae). — Bull. Am. Mus. Nat. Hist. 178: 110-214. Parker, A. & I. Kornfield (1995): Molecular perspective on evolution and zoogeography of Cyprinodontid killifishes (Teleostei; Atherinomorpha).- Copeia 12827: Pinto,C.& I. Vila (1987): Relaciones tröficas y caracteres morfofuncionales de Orestias laucaensis Arratia, 1982 (Pisces, Cyprinodontidae). — Ann. Mus. Hist. Nat. Valparaiso 18: 77-84. Paneer. Rotvun & B. Mayr (1987): Karyotype study of eight species of European percid fishes (Pisces, Percidae). — Caryologia 40 (4): 307-318. Roberts, F.L. (1967): Chromosome cytology of the Osteichthyes. — Progr. Fish- Cule9: 75-83. Schwalb, A., S.J. Burns & K. Kelts (1999): Holocene environments from stable isotope stratigraphy of ostracods and authigenic carbonate in Chilean Altiplano lakes. — Palaeogeogr., Palaeoclimatol., Palaeoecol. 148: 153-168. Tchernavin, V.V. (1944): A revision of the subfamily Orestiinae. — Proc. Zool. Soe., London 114, 9: 140-233. Thomann,R. (1990): Systematische und ökologische Untersuchungen an Orestias — Populationen (Valenciennes, 1839) (Pisces: Cyprinodontidae) im chilenischen Hochland., Dissertationsschrift, Fachbereich Biologie der Universität Hamburg., 135 S. Biicmonmhe co ROR. Miller (972): Second discovery of multiple sex chromosomes among fishes. — Experientia 28: 223-225. Bene ts ROR. Miller & J.M. Fitzsimonis (1973): Karyology of the cyprinodontoid fishes of the Mexican family Goodeidae. — Copeia 2: 497-510. Vila, I. & M. Pinto (1986): A new species of killifish (Pisces, Cyprinodontidae) from the Chilean altiplano. — Rev. Hydrobiol. trop., Bondy 19(3-4): 233-239. Villwock, W. (1963): Genetische Analyse des Merkmals “Beschuppung” bei anatolischen Zahnkarpfen (Pisces, Cyprinodontidae) im Auflöserversuch. — Zool. Anz 1710.4-2): 23-45. Villwock, W. (1966): Zur Biologie der Orestiinae (Pisces: Cyprinodontidae), unter besonderer Berücksichtigung von Darminhaltsuntersuchungen. Ein Beitrag zum Problem der Speziation in der Gattung Orestias. — Abh. Verh. nat.wiss. Ver. Hamburg, (NF) 10: 153-166. 66 Villwock, W. (1985): Uber Naturbastarde zwischen zwei validen Arten der Gattung Aphanius Nardo, 1827 (Pisces: Cyprinodontidae) aus der Bardawil-Lagune, Nordsinai/Agypten. — Mitt. hamb. zool. Mus. Inst. 82: 311-317. Villwock, W. (1987): Further contributions on natural hybrids between two valid species of Aphanius, Aphanius dispar dispar (Rüppell) and Aphanius fasciatus (Valenciennes) (Pisces, Cyprinodontidae) from the Bardawil-Lagoon, North Sinai and Al-Qanatir, west of the Suez Canal, Egypt. — pp. 238-244 in: Proc. Symp. Fauna and Zoogeogr. Middle East (eds. Krupp, F., W. Scheider & R. Kinzelbach). — Beih. Tiibinger Atlas Vord. Orient, Reihe A (Naturwiss.) 28. Villwock, W. & U. Sienknecht (1993): Die Zahnkarpfen der Gattung Orestias (Valenciennes, 1839) (Teleostei: Cyprinodontidae) aus dem Altiplano von Bolivien, Chile und Peru. Ein Beitrag zur Entstehung ihrer Formenvielfalt und intragenerischen Verwandtschaftsbeziehungen. — Mitt. hamburg. zool. Mus. Inst. 90: 321-362. Villwock, W. & U. Sienknecht (1995): Intraspezifische Variabilität im Genus Orestias Valenciennes, [839 (Teleostei: Cyprinodontidae) und zum Problem der Artidentität. — Mitt. hamburg. zool. Mus. Inst. 92: 381-398. Villwock, W. & U. Sienknecht (1996): Contribuciön al Conocimiento e Historia de los Peces Chilenos. Los Cyprinodöntidos del Género Orestias Val., 1839 (Teleostei: Cyprinodontidae) del Altiplano Chileno. - Medio Ambiente, Ambiente Acuaticos 13 (1): 119-126. Villwock, W. & R. Thomann (1987): Mitteilung über einen neuen Fundort von Orestias (Valenciennes, 1839) (Pisces: Cyprinodontidae) aus dem nördlichen Altiplano von Chile. Ein Beitrag zur Kenntnis der Verbreitung der Gattung Orestias. — Mitt. hamburg. zool. Mus. Inst. 84: 167-174. Villwock, W., L. Kies, F. Thiedig & R. Thomann (1987)2 Geolosische ökologische Untersuchungen am Lago Chungara/ Nord Chile: Zielsetzungen und erste Ergebnisse (Investigaciones geolögicas y ecolögicas en el Lago Chungara/ norte de Chile: Objetivos y primeros resultados). — IDESIA (Chile) 9: 21-34. Vitturi, R., E. Catalano, M S. Colomba, E. Montaemıo 7 Pellerito (1995): Karyotype analysis of Aphanıus fasciatus (Pisces, Cyprinodontiformes): Ag-NORs and C-band polymorphism in four populations from Sicily. — Biol. Zentr.bl. 114: 392-400. Vitturi, R., A. Libertini, A. Mazzola, M.S. Colomba & Ges 332 (1996): Characterization of mitotic chromosomes of four species of the genus Diplodus: Karyotypes and chromosomal nuclear organizer region phenotypes. — J. Fish Biol. 49: 1128-1137. Wörner, G., R.S. Harmon: J. Davidson. S. Mo0orbAcha Da Turner, N. McMillan, €. Nye, L. Lopez-Escobar.& H2 Mionemnte (1988): The Nevados de Payachata volcanic region (18°S/69°W, N. Chile). I. Geological, geochemical and isotopic observations.- Bull. Volcanol. 50: 287-303. Zeil, W. (1979): The Andes. A geological review. — In: Beiträge zur regionalen Geologie der Erde (eds. Bender F., V. Jacobshagen et al.). — Gebrüder Borntraeger, Berlin, Stuttgart. 67 Zeil, W. (1984): Die Salare in Nordchile — Geologie, Struktur und Geochemie.— Geotekton. Forsch. 67(I-Il): 1-146. Authors’ address: Zoologisches Institut und Zoologisches Museum, Universitat Hamburg, Martin-Luther- King-Platz 3, 20146 Hamburg, Germany 3 ne ee coe i 69 Diversity of a snake community in a Guinean rain forest (Reptilia, Serpentes) Wolfgang Böhme Abstract: During a one month excursion to the Ziama Forest (Forét de Ziama), SE Guinea, atotal of 38 species of snakes was collected, 5 of them being new for the fauna of Guinea. The only two earlier herpetological or ophidiological expeditions in this area (see Chabanaud 1920/21, Condamin 1959) recorded 24 snake species. Of these, 4 were not present in my collection, so that a total of 42 species is known from this small part of the West African rain forest, which approaches the published record (45: see Trape 1985) of any African snake assemblage. The species accumulation curve even suggests a much higher species richness in this forest. Whereas the anuran fauna in the same area was also comparatively rich (29 species recorded in one month), the lizard assemblage proved to be less rich (11 species only). An evaluation of the food niches of the 38 syntopic snake species encountered demonstrates a correspondingly low number of lizard-eaters (which are predominant in savannas), but many forms specializing in small mammals, birds, frogs and fish, some even in other snakes (Mehelya) and arthropods (Aparallactus). Key words: Reptilia, Serpentes, species richness, abundance, niche segregation, SE Guinea, West Africa Introduction On behalf of the PROGERFOR (Forest Gestation Resource Project, Ministry of Agriculture and Animal Resources at Conakry, Republic of Guinea), I spent the entire month of October 1993 in this West African country. PROGERFOR has a strong nature conservation component and tries to withhold the major part of two large and relatively intact forests from exploitation by forestry management. These forests are (1) the Ziama Forest (Forét de Ziama) of approx. 1300 km?, and (2) the Diecke Forest (Forét de Diécké) of about 700 km?, both situated in the extreme SE of the country where it meets, in the triangle between Liberia and Ivory Coast, the northern margin of the West African (= Upper Guinean) rain forest block (fig.1). My efforts to survey the herpetofauna of the two forests to be protected, within the framework of other zoological and botanical inventories of both forests, were concentrated on the Ziama Forest, and I was based in the PROGERFOR compound at Sérédou, hosted by the head of this project, Prof. Dr. Wilfried Bützler. In contrast to the Diecke Forest, which occupies a rather flat area as a lowland forest, the Ziama Forest covers a hilly area with steep slopes, fast-running torrents and waterfalls reaching up to 1400 m a.s.l., but also has more level areas with slow-running forest rivers and creeks. Primary forest is still extant on the slopes, and also in the level areas, and where it has Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 70 \ GUINEE FORESTIE mas succneno nn MACENTA a x RZ 4) leoreet de ; EN Ziama > a Nzebela Fig.l: Map of SE Guinea with the Ziama and Diecke forests. Stars indicate collection sites. been (illegally) destroyed (including deforestation by fire) it 1s reforested with local timber such as Terminalia. From 1300-1400 m a.s.l. the vegetation already contains montane floral elements, e.g. the widely distributed African tree fern Cyathea manniana. For a detailed description of habitats, see Böhme (1993, 1994 and in prep.). Material and methods Of the four weeks I spent in Guinea, I was actually present for 20 days in the study area. During this period, a total of 81 amphibian and reptile species was collected: | caecilian, 29 anuran, 11 lizard and 40 snake species. Of these, 38 snake species were recorded only in the Ziama Forest, a total of 154 specimens, which are deposited in the ZFMK collection, Bonn. Due to the limited time available, we tried to be as effective as possible and included the local people in our efforts. Moreover, we left buckets with preservative fluid in some villages which we controlled repeatedly. Some specimens included in the analysis were collected by Wilfried Biitzler before and also after my visit, and a few other specimens, left for deposition in the teaching collection of the “Centre de Recherches” at Sérédou, are not included in the present analysis. In Bonn, the snake material was checked with regard to the stomach contents. Aspects of faunistic, morphological, systematic, and zoogeographical importance will be published elsewhere, together with information on the remaining herpetofauna obtained. Table I: The snake fauna of SE Guinea as documented by various authors Forét de Ziama, between Macenta and N’Zebela Reference Family Subfamily Species Condamin this paper) (1920/21) Typhlopidae Typhlops angeli Typhlops punctatus liberiensis Boidae Calabaria reinhardti Python sebae Colubridae Colubrinae Bothrophthalmus lineatus Chamaelycus fasciatus Gastropyxis smaragdina Hapsidophrys lineatus Lamprophis lineatus Lamprophis olivaceus Lamprophis virgatus Gonionotophis klingi Lycophidion irroratum Lycophidion semicinctum Hormonotus modestus Mehelya guirali Mehelya poensis Mehelya stenophthalmus Philothamnus carinatus Philothamnus heterodermus Philothamnus irregularis Philothamnus semivariegatus Meizodon coronatus Meizodon regularis Thrasops occidentalis Natricinae Afronatrix anoscopus Grayia smithii Natriciteres variegatus Boiginae Crotaphopeltis hotamboeia Dipsadoboa brevirostris Dipsadoboa elongata Dipsadoboa unicolor Dispholidus typus Polemon acanthias Polemon bocourti Thelotornis kirtlandii Psammophis phillipsi Toxicodryas blandingii Toxicodryas pulverulenta Dasypeltinae Dasypeltis fasciata Dasypeltis scabra Forét de Diecke, between N’Zérékoré and Diecke ++ t+ 444 Mt. Nimba region ++ +++++++++ + 71 2 Forét de Ziama, Forét de Diécké, between Macenta between N’ Zérékoré and N’Zebela and Diécké Reference Family Subfamily Species Condamin (this paper) (this paper) Atractaspididae Aparallactus lineatus Aparallactus modestus Aparallactus niger Atractaspis irregularis Elapidae Dendroaspis viridis Naja melanoleuca Naja nigricollis Pseudohaje nigra Viperidae Causinae Causus lichtensteini Causus maculatus Viperinae Atheris chlorechis Bitis gabonica Bitis nasicornis Results and discussion Species richness: The herpetofauna of Guinea is extremely poorly known (see BOhme 1993, 1994 and in prep. for a detailed historical overview). For the study area dealt with here, this poor knowledge is illustrated by the following facts: 1. I was the first person to search for amphibians and reptiles in this area (since 1959 in the entire country!) since 1958, when Mr. R. Pujol finished a 5-month expedition at Sérédou (Condamin 1959). Of the 24 snake species collected by R. Pujol, 4 were not represented in my collection, resulting in a total number of 42 species for the Ziama Forest snake assemblage! 2. No less than 5 species collected during my visit have not been recorded for Guinea before. They are: Calabaria reinhardti, Meizodon regularis, Thrasops occidentalis, Mehelya guirali, and Dasypeltis fasciata (for details see Böhme, 1993, 1994 and in prep.). 3. Several of the remaining species collected by us were known in Guinea only from Mt. Nimba, to which the French herpetological activities in Guinea were restricted. They made Mt. Nimba famous as a center of endemism, mainly because it was the only site where data were available. The inclusion of the Ziama and Diécké Forests into herpetological surveying demonstrated that nearly all Mt. Nimba snakes also occur elsewhere. The information summarized above is documented in more detail in table 1. The Aparallactus lineatus Aparallactus modestus Bitis gabonica rhinoceros Bitis nasicornis Calabaria reinharatii Dasypeltis fasciata Dipsadoboa brevirostris Grayia smithii Hapsidophrys lineatus Mehelya stenophthalmus Naja nigricollis Thelotornis kirtlandii Thrasops occidentalis Atheris chloroechis Bothrophthalmus lineatus Dipsadoboa unicolor Lamprophis lineatus Mehelya guirali Mehelya poensis Philothamnus carinatus Lamprophis olivaceus Naja melanoleuca Philothamnus heterodermus Psammophis phillipsii Typhlops punctatus Aparallactus niger Atractaspis irregularis Causus maculatus Dendroaspis viridis Miodon acanthias Lamprophis virgatus Meizodon regularis Gastropyxis smaragdina Toxicodryas blandingii Toxicodryas pulverulentus Natric@teres variegatus Crotaphopeltis hotamboeia Afronatrix anoscopus 5 10 15 20 25 30 35 40 45 Number of Individuals Fig.2: Species richness and abundance in the Ziama Forest snake community. Abundance: Fig.2 combines the species richness with the abundance data in order to give an idea of the diversity of this snake community. By far the most dominant species (during October 1993) was Afronatrix anoscopus. Of 154 specimens in total, 42 belonged to this aquatic species, which is 27.3 % of the entire community. Together predominantly Micro- habitat Fre ja ja Fig.3: Species and niche occu- pation of the Ziama Forest snake community. Some species may appear twice as they can be diurnal and nocturnal as well as semiter-restrial, -aquatic, and - arboreal predominantly 74 Table 2: Ecological classification of the Ziama Forest snake community; n - nocturnal; d - diurnal; aq - aquatic; ar - arboreal; f - fossorial; t - terrestrial. For the numbers of food categories see text. Snake species Activity & Microhabitat Food categories Aparallactus lineatus n ft HW © Aparallactus modestus n f/t 7b/ 6a Bitis gabonica rhinoceros Met la, c/ 2a/ 4a Bitis nasicornis not la/ 4a/ 5/ -2a Calabaria reinhardtii net la/ 7/ 8a Dasypeltis fasciata a 2b Dipsadoboa guineensis n ar 4a/ ? Grayia smithii d aq _ d/ 4a Hapsidophrys lineatus d ar 3a/ 4a Mehelya stenophthalmus net 3a,b,c/ 4a Naja nigricollis dt la/ 2a,b/ 3a,b,c/ 4a/ 6a/ -7/ -8 Thelotornis kirtlandii dar 3a,c/ 2a,b/ 4a/ 6a Thrasops occidentalis dar 2a/ 3a/ 4a Atheris chlorechis n ar 3a/ la Bothrophthalmus lineatus Det la Dipsadoboa unicolor n ar Aa Lamprophis lineatus eat la/ 2a,b/ 3a,b,c/ 4a Mehelya guirali net 3c/ 3a/ 4a Mehelya poensis nt 3a,b,c/ 4a Philothamnus carinatus d t/ar 4a Lamprophis olivaceus ne la/ 3a Naja melanoleuca n/d t la/ 4a/ 3a,b,c/ 5/ 2a,b Philothamnus heterodermus dar 4a/ 3a Psammophis phillipsii det la/ 3a/ 4a/ 2a Typhlops punctatus nut 6a,b/ 8a Aparallactus niger n ft 7/6 Atractaspis irregularis me fi la/ 3c Causus maculatus d t/aq 4a Dendroaspis viridis d t/ar la/ 2a Miodon acanthias n ft 3b,c/ 4b Lamprophis virgatus not la/ 3a/ 4a Meizodon regularis dt la Gastropyxis smaragdina dt 3a/ da Toxicodryas blandingii n ar 2a,b/ 3a/ 1b/ -4a Toxicodryas pulverulentus n ar la/ 2a/ 3a Natriciteres variegatus d aq dal 5/ 7c Crotaphopeltis hotamboeia net 4a/ 3a/ la Afronatrix anoscopus d aq 5/ 4a with the next five commonest species - II Crotaphopeltis hotamboeia, 9 Natriciteres variegata, 6 Toxicodryas pulverulentus, 6 Toxicodryas blandingii, and 6 Gastropyxis smaragdina - already more than 50 % (viz. 51,8 %) are represented. The remaining 32 species, represented by five to one specimens each, form roughly the second half of the community. Spatial and temporal niche segregation: In table 2, an ecological classification is given for the spatial (terrestrial, fossorial, aquatic, arboreal) and temporal (diurnal, noc- turnal) niche occupation of the snake assemblage. Fig.3 summarizes the result: The dominant species are either aquatic or terrestrial, with the exception of the terrestrial Crotaphopeltis hotamboeia. The sum resulting from the figure exceeds the 38 species studied as some of them can show up in more than one category, as they may be diurnal 13 and nocturnal, or terrestrial and aquatic, terrestrial and arboreal etc. (see table 2). Fossorial snakes are of course not differentiated in diurnal vs. nocturnal forms. The lack of nocturnal, aquatic snakes is remarkable, which could be expected as an antipredator strategy of ectothermic reptiles which, however, might be unable to compensate nocturnal temperature loss in an aquatic environment. Trophic niche: The trophic niche is correlated with the habitat. Birds are mostly preyed upon by arboreal forms, whereas the aquatic forms prey predominantly on fish and frogs. The species of the Ziama snake community were classified as to their main prey types. These were divided in the following categories: 1. Mammals 5. Fishes a - small mammals: shrews, rodents b - bats 6. Insects c - larger mammals a - imaginal stages b - larval stages 2. Birds a - adults and nestlings 7. Other arthropods b - eggs a - millipedes b - centipedes 3. Reptiles c - spiders a - lizards with well developed limbs b - serpentiform lizards 8. Other invertebrates c - snakes a - worms b - molluscs 4. Amphibians a - anurans b - caecilians The assignment of the snakes to these prey categories is also summarized in table 2. This is based on my own data, supplemented with information from Pitman (1974), Leston & Hughes (1968), Stucki-Stirn (1979) and Broadley (1983). Euryphagous species can easily be differentiated from stenophagous ones. A striking phenomenon is the poor lizard assemblage of the Ziama Forest with only 11 species encountered. Whether this is a collecting artefact or the real species richness (or rather “poorness”!) might also be reflected by the number of lizard-eating (prey type 3 b) snake species. Out of 38 species, they are only 17 in number, most of them not narrowly specialized in this kind of prey. This suggests that lizards do not play a comparable role as snake prey in this forest community as they do in savanna reptile communities. The great trophic diversity of this species-rich snake community is, apart from rodent-, bird-, lizard-, frog-, fish-, and even arthropod-eaters, best demonstrated by the extreme specialists such as Mehelya, feeding to a great extent on other snakes, and Dasypeltis, feeding only on bird’ eggs: a unique specialization lacking in the Amazonian and Malayan tropics. Comparison with other snake communities: Species richness data for African snake communities have been compiled by Trape (1985). Four assemblages from Ivory Coast, Ghana, Gaboon, and CAR each ranged between 35 and 39 species. He himself recorded 45 species (represented by 351 specimens) at the forest site Dinamika (People’s Democratic Republic of Congo = Congo-Brazzaville), which is still the published continental African record. The Ziama assemblage approaches this record. 76 A recent inventory of the snakes of the Comoe National Park (Ivory Coast) (Rodel et al. 1995) revealed 37 species collected over a period of 4 years. Rödel et al. (in press) were able to add 7 more species, resulting in a total of 44 species in this area. Their suspicion that up to 25 further snake species remain to be discovered in Comoe NP leads to a new quality of African snake species richness. There is, however, already evidence that such numbers are actually reached by some snake communities in Cameroon (C. Wild, pers. comm.)! There are, however, some serious problems in comparing these data. The most important are (1) varying sizes, reliefs and habitat structures of the respective areas; and (2) varying collecting effort (so-called “man-days” sensu Murphy et al. 1994). This is the more true for snake diversity data from other places in the tropics. Neotropical forest snake communities from comparable areas vary between 36 (Arima Valley, Trinidad) and 60 species (Chagres, Panama), whereas Malayan richness data are available for a Bornean (36) and a Thai (47) forest site (Dunn 1949, Inger & Colwell 1977, Murphy et al. 1994). In the Bornean study area (Danum Valley: Murphy et al. 1994), 161 snakes - representing 36 species - were collected in 166 days. This means 0.9 snakes per day. The Guinean study area (Forét de Ziama) revealed similar numbers of specimens and species, viz. 154 snakes representing 38 species a considerably shorter time, but also with a differing number of collectors. In any case, considering the fact that in both cases formerly uncollected species were found even on the very last day of the field-trip (see fig.4), the total of 42 snake species recorded for the Ziama forest snake community accumulated species 2 4 6 8 10 12 14 16 18 20 22 days Fig. 4. Species accumulation curve of the Ziama Forest snake community Wi in SE Guinea is very likely to be much exceeded when more time is available for future investigations. Acknowledgements First of all, I wish to thank my host at Sérédou, Guinea, Prof. Dr. Wilfried Biitzler, Göttingen. He provided the entire organisation and logistics of my excursion and moreover took a most active part in collecting specimens. My thanks are also due to H.U. Gross and M. Midre of the Deutsche Forst Consult (DFC) for employing me as a short-time expert. Furthermore, I am indebted to the respective authorities in Conakry, the personnel of PROGERFOR, the Direction Nationale des Foréts et Chasse (DNFC), particularly to its director M. Sagnah Satenin, for providing all necessary permits; and to the Deutsche Kreditanstalt ftir Wiederaufbau (KfW), Frankfurt am Main, for creating the financial basis for the whole project including my visit. I am grateful to Dr. Barry Hughes, London, and to Dr. J.B. Rasmussen, Copenhagen, for sharing their ophidiological competence with me. Last but not least, I wish to express my gratitude to Birgit Rach, Ursula Bott and Wolfgang Bischoff, all from ZFMK Bonn, for their valuable assistance with collecting data and preparing the manuscript. References Angel, F., J. Guibé, M. Lamotte & R. Roy (1954): La Réserve Naturelle Inté- grale du Mont Nimba. Serpents. — Mem. Inst. fr. Afr. Noire, Dakar, 40 (2): 381-402. Böhme, W. (1993): Mission d’ études herpétologiques dans les foréts de Ziama et Diécké, Guinée forestiere. — Unpubl. project report, Neu-Isenburg (DFC), 49 pp. Böhme, W. (1994): Frösche und Skinke aus dem Regenwaldgebiet Südost-Guineas, Westafrika. I and II. — Herpetofauna, Weinstadt, 16 (92): 11-19, and 16 (93): 6-16. Broadley, D.G. (1983): Fitzsimons’ snakes of southern Africa. — Delta Books, Johannesburg. Chabanaud, P. (1920): Contribution a l’eEtude de la faune herpétologique de |’ Afrique occidentale. Note préliminaire sur les résultats d’une mission scientifique en Guinée francaise (1919-1920). — Bull. Com. Etud. Hist. Sci. Afr. Occ. Fr., Paris, 1920: 489-497. Chabanaud, P. (1921): Contribution a l’etude de la faune herpétologique de l’ Afrique Occidentale. Deuxiéme note. — Bull. Com. Etud. Hist. Sci. Afr. Occ. Fr., Paris, 1921: 445-472. Condamin, M. (1959): Serpents récoltés a Sérédou (Guinée) par R. Pujol. — Bull. Inst. Fr. Afr. Noire, Dakar, 21 (A 4): 1351-1366. Dunn, E.R. (1949): Relative abundance of some Panamanian snakes. — Ecology, 39 0023957. Inger, R.F. & R.K. Colwell (1977): Organisation of contiguous communities of amphibians and reptiles in Thailand. — Ecol. Monogr., 47: 229-253. Leston, D. & B. Hughes (1968): The snakes of Tafo, a forest cocoa-farm locality in Ghana. — Bull. Inst. Fond. Afr. Noire, Dakar, 30 (A2): 737-770. Murphy, J.C.,H.K. Voris & D. Karns (1994): A field guide and key to the snakes of the Danum Valley, a Bornean tropical forest ecosystem. — Bull. Chicago mlenmperoly Soc., 29 (/): 133-151. Pitman, C.R.S. (1974): A guide to the snakes of Uganda. Revised edition. — Wheldon & Wesley, Codicote, 290 pp. 78 Rodel, M.-O., K. Grabow, C. Bockheler & D. Mahsberg (1995): Die Schlangen des Comoé-Nationalparks, Elfenbeinktiste (Reptilia: Squamata: Serpentes). — Stuttg. Beitr. Naturk., (A) 528: 1-18. Rödel, M.-O., K. Kouadio & D. Mahsberg (1999): Die Schlangenfauna des Comoé-Nationalparks, Elfenbeinküste: Ergänzungen und Ausblick. — Salamandra, Rheinbach, 35(3): 165-180. Roman, B. (1976): Serpents mortels de l’Ouest Africain. — Etud. Sci., Paris 6: 1-56. Stucki-Stirn, M.C. (1979): Snake report 721. A comparative study of the herpetological fauna of the former West Cameroon/Africa. — Herpeto-Verlag, Teuffenthal, 650 pp. Trape, J.F. (1985): Les serpents de la région de Dimonika (Mayombe, République Populaire du Congo). — Rev. Zool. Afr., Bruxelles 99: 135-140 Wallach, V. (1994): Aparallactus lineatus (Peters) and Aparallactus niger Boulenger: two valid species from West Africa. — J. Herpetol. 28 (1): 95-99. Author’s address: Prof. Dr. Wolfgang Böhme, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 150-164, 53113 Bonn, Germany 2) The distribution of Chromaphyosemion Radda, 1971 (Teleostei: Cyprinodontiformes) on the coastal plains of West and Central Africa Rainer Sonnenberg Abstract: The species of the monophyletic genus Chromaphyosemion Radda, 1971 live in small streams and rivulets (“marigots’) in the coastal plains from Togo to Gabon. In their distribution area, some glacial forest refuges are postulated, mainly in Cameroon, Equatorial Guinea and Gabon. It is assumed that evolution and biogeography of this genus is connected with the evolution of the Central African rainforest during the Quaternary. In this paper Chromaphyosemion is grouped into 16 forms, defined by their colour patterns. The distribution data show that the maximum diversity of this genus is found in the proposed glacial rainforest refuges of Cameroon, suggesting a structurally complex history of these old rainforest blocks. Key words: Aplocheilidae, Chromaphyosemion biogeography, rainforest refuges Introduction Responsible for the biodiversity of African tropical ecosystems, according to some authors, is the influence of climatic changes during the Quaternary and the resulting fragmentation of rainforest (Maley 1991, Mayr & O’Hara 1986). This probably induced major extinctions and, as an effect of the fragmentation, an isolation of populations, genetic bottlenecks and later a recolonization, which affected rainforest-bound species especially (Huber 1998b). Of special interest are position and number of rainforest refuges. Some authors provide a map of various currently assumed refuges in Western and Central Africa (Hamilton & Taylor 1991, Lévéque 1997, Maley 1991, Sayer et al. 1992). In this paper I want to test if the distribution of a mainly rainforest-bound group of small aplocheilid fish with limited dispersal abilities correlates to these assumed refuges in their distribution area. The recent distribution and evolution of aplocheilid species and species groups seems to be closely connected with the historical changes of rainforest distribution and climates through the Quaternary (Huber 1998b). This means that a detailed study of the phylobiogeography of these fish should give further insight into the history of the tropical rainforest in coastal lowlands. Especially the different forms of Chromaphyosemion, with their relatively small ranges, could contribute additional data to the studies of historical rainforest fragmentation. The aplocheilid genus Aphyosemion Myers, 1924 is subdivided into several subgenera (Myers 1924, Huber 1977, Huber & Seegers 1977, Kottelat 1976, Radda 1971b, 1977). One of them, Chromaphyosemion, is clearly monophyletic (Amiet 1987, 1991, Seegers 1981) and here considered as a genus. Chromaphyosemion are Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 80 small fish which reach about 5-7 cm total length and live in small rivers and streams in the coastal lowlands from Togo to northern Gabon (Eberl 1996, Radda 197 1a,b, 1975, Radda é Huber 1976, Radda cu Piunz lah ite 98 7) Then isomorphic with marked sexual dimorphism. The different forms are commonly distinguished by male coloration and sometimes by female colour pattern as well. Scheel (1966, 1968, 1974, 1990) did much comprehensive work on this group, based on crossbreeding experiments and karyotyping of different populations. His results reveal a high degree of reproductive isolation and variable chromosome numbers between populations. At the beginning of his studies he synonymized all described species with Aphyosemion bivittatum (Lönnberg, 1895) which then was considered as a ‘morphospecies’ because no morphological character could be found to distinguish subunits, and colour patterns are not visible in preserved material (Scheel 1966, 1968: 166 ff.). Scheel himself was aware of the fact, that the BIV-group - as he named them after A. Divittatum - consists of different ‘bio- species’ (sensu Mayr). In 1974 and 1990 Scheel revalidated some of the ‘species’, which in his opinion also consist of groups of different ‘bio-species’. Amiet (1987), Daget et al. (1986), Scheel (1968, 1990) and Seegers (1986) tried to clarify the taxonomic confusion in this group. Fish hobbyists as well tried to redefine the known species by their coloration (see Eber! 1996 for a comprehensive review, e.g. of Poliak and Legros). According the actual state of knowledge it is most practicable to treat the forms which can be separated by their colour pattern as species; this does not anticipate any other decision in the future. The aim of this study is a review of the distribution data of the different Chromaphyosemion species and the comparison of the results with the data on rainforest refuges in that area. Material and Methods Preserved specimens from the studied live populations and slides of their live coloration are stored in the collection of the ‘Zoologisches Forschungsinstitut und Museum Alexander Koenig’ (ZFMK). Additional distribution data and pictures are taken from Amiet (1987, 1991), Eberl (1996), Huber (1996), Langton (1996), Radda (1971a,b, 1975), Radda & Huber (1976), Radda & Piirzl (1977, 1981, 1982, 1987), Scheel (1966, 1968, 1974, 1990), Seegers (1997) and Teugels et al. (1992). Populations are grouped according to shared characters in body coloration and fin Plate 3: A selection of forms (probably species) of Chromaphyosemion Radda. Figures 1-6 show forms, which are all named C. splendopleure by most authors, 7 C. loennbergii and 8 the population marked on the map with ?; Amiet refers this also to C. loennbergii. 1 (upper row, left) C. splendopleure, Tiko; 2 (upper row, right) C. cf. splendopleure, Muyuka, Police Station; 3 (second row, left; a.s.o.) C. sp. aff. splendopleure ‘Dizangue’, Mangoule; 4 C. sp. aff. splendopleure ‘Kopongo’, Kopongo CSK 95/27; 5 C. sp. 07, near Nkolbonda ; 6 C. sp. “Likado’, Likado River; 7 C. loennbergii, Song Bibai; 8 C. sp., Chutes d’Ekom Photographs by Eigelshofen (1,4,7, and 8) and Pohlmann (2,3,5, and 6) vee HGAE AEAUAG UN 5 Br A t i ) t £ i \ te x rs iS = ws a lon) Sl colour patterns of adult males. The populations of one phenotype can be separated from other related phenotypes by one or more diagnostic characters or a combi- nation of characters. 1 C. bivittatum 2 C. cf. splendopleure 3 C. splendopleure ae 4C. volcanum | 5C. poliaki 5C.sp. aff. splendopleure ‘Dizangue’ _7C. sp. aff. splendopieure ‘Kopongo’ 8 C. riggenbachi 9 C. Icennbergii Bere 10 C. sp. 96 11 C. sp. 07 12 C. sp. 04 == 13 C. lugens 14 C.sp. ‘Likado’ ? refer to C. loennbergii | I 100 km O Fig.1: Map of the coastal plains of Cameroon with the schematic ranges of the Chromaphyosemion species in Cameroon (modified after Amiet 1987). The numbers on the map correspond to the following species. If known, the number is placed at or near the terratypica. Within the species accounts species names are used according to Daget et al. (1986) and Huber (1996) with addition of the species, described after the publication of these checklists. Here only those synonyms are given which occur in the literature used; additional synonyms are given in Daget et al. (1986), Seegers (1986) and Wildekamp (1993).Only a short description of the coloration is given, especially in those cases where I differ from previous authors; otherwise the first description is cited where a useful colour description is given. A more detailed description of body coloration and fin colour pattern of different species is given by Amiet (1987, 82 1991), who supplies a very detailed description for Aphyosemion s.l. along with a determination key, and Eberl (1996). For all species, the known distribution area and a reference to published pictures and / or preserved material is given. Poliak, Legros and Eberl (reviewed in Eberl 1996) separated populations of ‘Aphyosemion splendopleure’ into three ‘phenotyes’, which are here redefined and in some aspects changed. The distribution data of the species from Cameroon are presented in a map (fig.1). The pattern of species density and distribution is compared with published maps of postulated rainforest refuges in the area in question (Hamilton & Taylor LOO Malley 1991, Sayeret al 192% Results Genus Chromaphyosemion Chromaphyosemion Radda, 1971: 118 (type species: Fundulopanehax multicolor Meinken, 1930 by original designation). This name is considered to be a synonym of Chromaphyosemion bitaeniatum (Ahl, 1924) (see Seegers 1986). Chromaphyosemion was described by Radda (1971b) as a subgenus of Aphyosemion. Diagnosis: Chromapyhosemion is distinct from all other Aphyosemion s.l. by two dark lateral bands in both sexes (vs. one or none). The male coloration can change very fast, depending on mood and hierarchical status of specimens (vs. less conspicuous changes in Aphyosemion). No blue or green metallic base colours on the sides of the body (vs. present in most other Aphyosemion). In males, two or three dorsolateral rows of scales with a metallic sheen (C. riggenbachi with none to two rows) (vs. no metallic sheen on these scales or a metallic sheen on the whole body). The form of unpaired fins in Chromaphyosemion is unique and differs from all species of Aphyosemion (Amiet 1987, Radda 1971b). Caudal fin rounded with pronounced dorsal and ventral streamers in males, sometimes trilobated in large specimens. Dorsal fin large and triangular. Anal fin trapezoid. In males, white, yellow or orange streamers are developed at unpaired fins, mostly with a different hue to that of the fins. Amiet (1987), Huber (1996) and Scheel (1974) give the following additional combination of characters to define Chromaphyosemion: dorsal fin inserts above or slightly behind anal fin origin. Number of dorsal rays 9-14, number of anal rays I 1- 16, D/A- position 1/1-4 fin rays (vs. a higher number of anal and dorsal rays if the D/A-position is in the same range). Amiet (1987), Murphy & Collier (1999) and Seegers (1981) showed that Chromaphyosemion is a monophyletic group with no apparent intermediate forms to Aphyosemion or related genera and therefore it 1s considered as a genus. The relationship to other species-groups remains unclear. A recent study on molecular systematics of Aphyosemion and Fundulopanchax fails to provide a well supported hypothesis — except the monophyly of most of the studied species groups (Murphy & Collier 1999). Distribution: Chromaphyosemion species live in the coastal plains from Togo to northern Gabon. In the following species accounts the approximate distribution from West Africa to Central Africa is provided. 83 The species of Chromaphyosemion Chromaphyosemion bitaeniatum (Ahl, 1924) Terra typica: Niger, Nigeria Synonyms: Fundulus bitaeniatus Ahl, 1924 - Aphyosemion bitaeniatum (Eberl 1996, Huber 1996, 1998a, Seegers 1986, 1997) - Aphyosemion bivittatum (Radda 1971b, Radda & Pürzl 1987, Scheel 1968, Schröder 1967) - Aphyosemion multicolor (Radda 1975, Radda & Pürzl 1981, Seegers 1981) Examined material: ZFMK 21986 - 21987, Afanyangan (TMBB 90/13), Togo; ZFMK 21988 - 21989, Agbetiko (RT 97), Togo; ZFMK 21990 - 21991, Ibeju Creek, Nigeria; ZFMK 21992 - 21993, Yemoji River, Nigeria Pictures: Eberl 1996 [p. 34, Lagos]; Radda & Piirzl 1987, [p.69, southern Benin]; Scheel 1968 [p. 116, Meko, Nigeria; p. 117, Umudike, Nigeria; p. 120 upper fig., Lagos, Nigeria]; Seegers 1997 [p. 59, fig. AO3240-4 - A03244-4; p. 60, A03747-4 - A03749-4; p. 61, A03752-4] Distribution: This species is found in the coastal plains from Togo to the Niger delta in Nigeria, which is about half of the distribution area of the genus. Along the Niger this species occurs further inland. From the eastern border of the distribution area one population is known from Umudike, which is between the river systems of the Niger and the Cross. Remarks: Unpaired fins mainly orange with a blue or blue-green groundcolour. The amount of blue-green and orange in the anal fin can vary between and within different populations. A very conspicuous dark wound-like marking behind the operculum (Scheel 1974). Scheel (1974) and Schröder (1967) showed that between different populations various degrees of reproductive isolation are found. Despite these findings there are only relatively small variations between the coloration of different populations. Chromaphyosemion bivittatum (Lönnberg, 1895) Terra typica: rivulet..., near the waterfall of river N’dian [Cameroon] Synonyms: Aphyosemion bivittatum (Amiet 1987, Eberl 1996, Huber 1996, 1998a,b, Radda 1971a,b, 1975, Radda & Pürzl 1981, 1987, Radda & Wildekamp Peo? Senec! 1966, 1968; 1974, 1990, Seegers 1981, 1986, 1997) Examined material: ZFMK 21994 - 21995, Kwa Riverfalls Plantation, Nigeria; ZFMK 21996 - 21997, Funge, Cameroon Pictures: Amiet 1987 [plate 34, fig. 5-6]; Eberl 1996 [p. 22, Mundemba; p. 23, Funge and Biafra]; Radda & Piirzl 1987, [p.71, Funge]; Seegers 1997 [p. 61, A03245-4, A03246-4 and A03970-4] Distribution: This species is known from the area between the Cross River (Nigeria) and Funge (Cameroon). Remarks: Chromaphyosemion bivittatum is characterized by one, seldom two, dark spots on the base of the caudal fin in males (vs. absent in the closely related species C. lugens and C. sp. 06 (Amiet 1987, 1991, Eberl 1996)). Chromaphyosemion cf. splendopleure (Brining, 1929) Synonyms: Aphyosemion bivittatum (Radda 1975, Scheel 1966, 1968) - A. bitaeniatum (Eberl 1996, in part) - A. aff. multicolor (Radda & Pürzl 1982, in part) 84 - A. splendopleure (Amiet 1987, Huber 1996, 1998a,b, Radda 1971a,b, Radda & Purzl 19877 Scheel 1974 19905 Seesensl OS lO Sone pa splendopleure ‘Meme’-group (Eberl 1996, in part) - A. sp. aff. splendopleure (Radda & Wildekamp 1977) Examined material: ZFMK 21998 - 21999, Lykoko, Cameroon; ZFMK 22000 - 22001, Muyuka Police Station (C 89/15), Cameroon Pictures: Amiet 1987 [plate 37, fig. 16 and 19]; Eberl 1996 [p. 45, Muyuka Police Station]; Radda & Pürzl 1982, [p. 19, Oron, Nigeria], Radda & Pürzl 1987 [p. 70, Ekondo Titi, Cameroon]; Scegers 1931 pP. 1562 118 olf See OA: 122, A03530-4, A03945-4, A03947-4 only the right-hand picture; p. 123, A03948- 4] Distribution: These populations are found in the lower Cross in Nigeria (Radda 1975), and south of Funge to the Mungo River (Cameroon). Strains are known from Oron, Ekondo Titi, Mbonge, Lykoko, Owe and Muyuka. Remarks: The colour of the unpaired fins is blue-green with yellow, the metallic sheen in the dorsal scales is yellow to gold and body colour can range from blue to pink, the throat and part of the belly are yellow. This separates C. cf. splendopleure from the nominate C. splendopleure. Eberl (1996) listed both forms under his ‘Meme ’-phenotype. It is not known if the populations of the Cross River area are in connection with those of the Meme area or if they are divided by the distribution area of C. bivittatum. C. cf. splendopleure is closely related to C. splendopleure. It is possible that a cline between both species may be found in the area around Kumba, but in the area around Muyuka and Tiko they are easy to distinguish. Chromaphyosemion splendopleure (Briining, 1929) and C. volcanum (Radda & Wildekamp, 1977) Terra typica: Tiko in Kamerun [Tiko in Cameroon] (C. splendopleure) “...ın einem kleinen Bächlein, welches durch den südwestlichen Stadtteil Kumbas fließt” [small rivulet, which flows through the southwestern part of Kumba = Kake River] (C. volcanum) Synonyms: A. splendopleure (Amiet 1987, and Eberl 1996, in part) - A. bivittatum ( Scheel 1968, in part) - A. volcanum (Amiet 1987, Eberl 1996, Huber 1996, Radda 1997, Radda & Pürzl 1987, Radda & Wildekamp 1977, Scheel 1990) Examined material: ZFMK 22002 - 22003, Moliwe, Cameroon; ZFMK 22004 - 22005, Bamukong, Cameroon; ZFMK 22006 - 22007, Tiko, Cameroon; ZFMK 22008 - 22009, Bombe (CXC 23), Cameroon Pictures: Amiet 1987 [plate 37, fie. 17-18]; Eberl 1996 p DIA Zyoleennaa: Seegers 1997 [p. 122, A03946-4, A03949-4 and A03950-4] Distribution: These species are known from Moliwe, Bamukong, Tiko, Yoke, Bombe and Kumba, which is an area from the foothills of Mount Cameroon to the region south of Lake Barombi Mbo. Most of the rivulets, in which this form is found, are tributaries of the River Mungo. Remarks: These species have in general a red to orange body colour with a copper-red to reddish metallic sheen in the dorsal scales. Mainly orange-coloured unpaired fins with yellow to orange streamers. The throat is orange. 85 This group comprises two described species which are phenetically closer to each other than to the populations of C. cf. splendopleure. These species seem to be closely related compared with the following species, which are also included in C. splendopleure by various authors. C. splendopleure and C. volcanum are here grouped together due to their apparent similarity and are treated as one group of closely related populations. There is an ongoing discussion in the aquaristic literature about the validity of the taxon C. volcanum (e.g., Eberl 1996, Radda 1997) but neither reproductive isolation nor its reverse has been so far tested; herein both taxa are considered as valid until further data are provided. Chromaphyosemion sp. aff. splendopleure (Brüning, 1929), “Dizangué’ Synonyms: Aphyosemion splendopleure (Amiet 1987, Huber 1996, 1998a,b, BaddarezPürzl71987, Scheel 1974, 1990, Seegers 1981, 1986, 1997, in part) - A. splendopleure ‘Dizangué’ ( Eberl 1996) - A. bivittatum (Scheel 1968, in part) Examined material: ZFMK 22010 - 22011, Mangoule, Cameroon Reemumnes: Amiet 1987 plate 36, fie. 13-14]; Eberl 1996 [p. I, Dizangue]; Seegers 1997 [p. 122, A03943-4 and A03944-4] Distribution: This species is found along the coast between Douala and Kribi. The western limit is the Atlantic Ocean and in the east it reaches the distribution areas of the species living further inland, C. riggenbachi and C. loennbergii. Remarks: C. sp. aff. splendopleure (‘Dizangué’) differs from the nominate C. splendopleure in body coloration and fin colour pattern. The lower lateral band is characteristically very conspicuous. The dorsal part of the body is brown, below beige to green-yellow. The caudal fin has a mainly flame pattern but not on a blue groundcolour (like C. Joennbergii). Together with the following species these fish show a marginal blue to blue-green band in the anal fin (vs. absence in the other two forms which were formerly included in C. splendopleure). The anal fin is in most populations translucent yellow-green (Amiet 1987) (vs. orange, orange-red or yellow in all other populations previously included in C. splendopleure). A wound- like marking as in C. bitaeniatum is present (vs. absence in the previous forms). Chromaphyosemion sp. aff. splendopleure (Brüning, 1929), “Kopongo’ Synonyms: A. splendopleure (Amiet 1987, Huber 1996, 1998a,b, Radda & Pürz] 1987, Scheel 1974, 1990, Seegers 1981, 1986, 1997, in part) - A. splendopleure ‘Kopongo’ (Eberl 1996) - A. bivittatum (Scheel 1968, in part) E 10 cm diameter) and leaf litter were removed simultaneously from all four sides, starting at the perimeter and working inward. 97 Table |: Description of the ten sampling quadrats with the represented number of amphibian and reptile species each. altitude short description sampling no. of no. of m a.s.l. date amphib. reptile species species forest between camp and plantation 25-11-98 ecocline: forest to plantation 26-11-98 sunpatch, montane forest 27-11-98 ecocline: fern meadow to montane forest 28-11-98 Eyimba, secondary forest 30-11-98 Eyimba ecocline: sec. forest to plantation 1-12-98 forest on creek side 2-12-98 forest on creek side 3-12-98 hilltop, tree island in plantation 4-12-98 forest 5-12-98 Results and Discussion Hand capture Although not quantitative, this is a very successful method. Generally hand capture is often the most effective method of accumulating the most species in the minimum amount of time (Scott 1994). This technique resulted in a total of 40 amphibian species in five families and 41 reptile species in 13 families. Glue traps Despite precipitation the glue traps maintained their stickiness throughout the duration of the study. However they proved to be largely ineffective, catching a total of only four lizards (Agama and Mabuya), all found within the sunny plantation area. Insects were often represented as well as feathers of several bird species and one live bat. Drift fence-pitfall traps Drift fences with pitfall traps have been used successfully to determine species richness and to detect the presence of rare species (Corn 1994). Here this method proved marginally effective, producing one snake (Natriciteres), two lizards (Panaspis) and six frogs (Arthroleptis, Phrynobatrachus). A limitation of this method is that such traps tend to capture some species more readily than others (Corn 1994). Parris et al. (1999) found drift fence-pitfall traps much less effective than stream-searching during their amphibian survey in Queensland, Australia. 98 Table 2: Number of specimens of amphibians and reptiles found in the 10 sampling quadrats. Uncertain species determination (cf.) is indicated by °, juveniles are indicated by * Amphibia Bufonidae Nectophryne afra Nectophryne batesi Wolterstorffina parvipalmata Arthroleptidae Arthroleptis adelphus Arthroleptis cf. adolfifriderici Arthroleptis bivittatus Arthroleptis poecilinotus Arthroleptis variabilis Arthroleptis sp. Astylosternus diadematus Astylosternus montanus Cardioglossa melanogaster Leptodactylodon bamilekianus Leptodactylodon mertensi Ranidae Dimorphognathus africanus Petropedates parkeri Phrynobatrachus cricogaster Phrynobatrachus werneri Phrynobatrachus sp. Phrynodon sandersoni Hyperoliidae Hyperolius ocellatus Leptopelis brevirostris Leptopelis calcaratus Chamaeleonidae Chamaeleo montium Scincidae Panaspis amieti Panaspis reichenowii Panaspis vigintiserierum Colubridae Chamaelycus fasciatus Lycophidion laterale 99 3 Bufonidae 3 sp. O Arthroleptidae 11 sp. El Ranidae 6 sp. MHyperoliidae 3 sp. Ei Chamaeleonidae 1 sp. LO Scincidae 3 sp. @ Colubridae 2 sp. Fig.2: Herpetofauna diversity of all quadrats arranged in families They found only two species using the traps versus 13 species by stream-searching. Both trapped species were ground-dwellers whereas a high percentage of the species recorded while stream-searching were tree frogs. Additionally, 20 shrews and rodents were also captured in the Mt Nlonako study as well as some terrestrial crabs. Quadrat sampling This quantitative method proved to be effective especially for the sampling of leaf litter- and ground-dwelling species (tables | and 2). Quadrat sampling yielded 23 amphibian species in four families, whereas hand captures, for comparison, produced 40 species in five families. Fifty-eight percent of the hand captured amphibian species were found in the quadrats. Only 15% of the hand captured reptile species (41 species from 13 families) were represented in the quadrats (six species in three families). cee 14% a Bufonidae 21 ind. | 2 Arthroleptidae 72 ind. ERanidae 26 ind. MHyperoliidae 20 ind. Zi Chamaeleonidae 1 ind. [1Scincidae 11 ind. @ Colubridae 2 ind. Fig.3: Relative abundance in all quadrats according to family. 100 EM Colubridae LC Scincidae 21 Chamaeleonidae 35 MHyperoliidae 0 30 YP ©) ae El Ranidae E (1 Arthroleptidae ° 20 ; 1 E] Bufonidae u 15 10 i 0 1 2 3 4 5 6 7 8 9210 quadrat Fig.4: Diversity of amphibian and reptile species per family and quadrat. Reptiles represented 20% of all herpetological species (fig.2) found in the quadrats and nine percent of all herpetological specimens (fig.3). Quadrat four (1600 m elevation) contained reptiles but no amphibians (table 2, fig.4 and 5). This can be related to the fact that there was no running or open water at this altitude, thus restricting amphibians dependent upon such water for reproduction. Quadrat 3 (1600 m elevation), however, contained four specimens of Arthroleptis cf. adolfifriderci. This species does not require water bodies for reproduction because it lays its eggs in moist substrate (Duellman & Trueb 1986). A similar case is reported from America, where Eleutherodactylus, which also undergoes a development independent of free water, may even ecologically replace small lızards (Scott 1976a). Accordingly, Inger (1980) reports that in South-east Asia and Central America floor-dwelling species of frogs and lizards in tropical forests are ecologically very sımilar. Quadrats 3 and 4 are characterized by low species diversity and low specimen abundance (fig.4 and 5). Arthroleptidae and Ranidae accounted for 60% of all species (fig.2) and 64% of all reported specimens (fig.3). Arthroleptid specimens represented almost half (47%) of all specimens found and were present in eight out of ten plots. Interestingly, hyperohids were found only in quadrats where bufonids were present as well (figs.4 and 5). The reason may be that all reported toads were at least partially arboreal species and thus somewhat similar to treefrogs in their habitat requirements. lO] @ Colubridae 1 Scincidae ZI Chamaeleonidae MHyperoliidae 2 10 8 % El Ranidae o 6 DArthroleptidae 2 Bufonidae quadrat Fig.5: Abundance of individuals per family and quadrat. Quadrat 7 had the highest species diversity and specimen abundance (figs.4 and 5). This can be explained by its position at a creek and its possibly higher microhabitat diversity. However, such an explanation may be wrong since quadrat 8, an apparently similar plot, produced only average diversity and less than average abundance values. In general, quadrats varied greatly with regard to species diversity and specimen abundance. Thus conclusions drawn from statistical analysis of results obtained by the quadrat method will show significant trends only when larger numbers of quadrats are investigated (at least 50 quadrats; Jaeger & Inger 1994). The lack of animals in quadrat 9 can only be explained by variation within the samples. Scott (1976b) found a total of eight species of three amphibian families (Bufonidae, Arthroleptidae, Ranidae) in 15 rainforest plots in the Cameroon lowland (up to 30 m elevation). Quadrat size was 58 m?. Specimens of Arthroleptidae greatly outnumbered Bufonidae and Ranidae. In the Mt Nlonako study a total of 24.1 specimens/100 m? were found; amphibians with 21.9 and reptiles with 2.2 specimens/100 m?. Scott (1976b) noted 9.4 individuals/100 m? in his Cameroon lowland rainforest study area. He reports 1.5 specimens/100 m? for a Bornean site, and in Central America 14.6 (Panama) and 17.1 (Costa Rica) individuals/100 m?. This clearly means that the Mt Nlonako abundances are even higher than the Central American herpetofauna abundances studied so far. 102 Acknowledgements We are greatly indebted to Mr. Horst Nobis of ALSCO Germany for providing funding for this project. Chris Wild of WWF (Mt Kupe Forest Project) provided invaluable assistance, support and expertise. Many thanks to our co-worker Oliver Euskirchen and our field assistants from the Mt Kupe Forest Project, Nyasoso, and from Nguenge village. References Corn, P.S. (1994): Straight-Line Drift Fences and Pitfall Traps. — pp. 109-117 in: Measuring and Monitoring Biological Diversity. Standard Methods for Amphibians (eds. Heyer, W.R., et al.). — Smithsonian Institution Press. Duellman, W.E. & L. Trueb (1986): Biology of Amphibians. —- McGraw-Hill Book Company, New York. Eisentraut, M. (1973): Die Wirbeltierfauna von Fermando Boo und Westkamerun. — Bonn. zool. Monogr. 3: 1-428. Euskirchen, O. (1998): Untersuchungen zur Okologie und Ethologie der Herpetofauna einer montanen Regenwaldregion in Kamerun. — Unpublished Master’s Thesis, Universitat Bonn, 176 pp. Inger, R.F. (1980): Densities of floor-dwelling frogs and lizards in lowland forests of Southeast Asia and Central America. — The American Naturalist 115: 7.61- 7:70. Jaeger, R.G. & R.F. Inger (1994): Quadrat sampling, — pp I721067m: Measuring and Monitoring Biological Diversity. Standard Methods for Amphibians (eds. Heyer, W.R., et al.). - Smithsonian Institution Press. Müller, M.J. (1996): Handbuch ausgewählter Klimastationen der Erde. — Forschungsstelle Bodenerosion, Universität Trier, Heft 5: 400 pp. Parris, K.M., Norton, T.W. &R.B. Cunningham (1999)2 4 companson of techniques for sampling amphibians in the forests of south-east Queensland, Australia. — Herpetologica 55: 271-283. Schmitz, A. (1998): Systematik und Zoogeographie der Herpetofauna einer montanen Regenwaldregion in Kamerun. — Unpublished Master’s Thesis, Universitat Bonn, 261 pp. Scott, N.J. (1976a): The abundance and diversity of the herpetofauna of tropical forest litter. — Biotropica 8: 41-58. Scott, N.J., Jr. (1976b): The herpetofauna of forest litter plots from Cameroon, Africa. — pp. 145-150 in: Herpetological communities (ed. Scott, N.J., Jr.). —U.S. Department of the Interior, Fish and Wildlife Service, Wildlife Research Report 13 Scott, N.J., Jr. (1994): Complete species inventories! — PP 18.8.2. m Measuring and Monitoring Biological Diversity. Standard Methods for Amphibians (eds. Heyer, W.R., et al.). -— Smithsonian Institution Press. Authors’ addresses: Dr. Hans-Werner und Patricia Herrmann, Kölner Aquarium am Zoo, Zoologischer Garten Köln, Riehler Straße 173, 50735 Köln, Germany Prof. Dr. Wolfgang Böhme, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany eG vy ues ann) N Re nh, N a. al gear Win Pa 105 New record of Grauer’s Blind Snake Rhinotyphlops graueri (Sternfeld, 1912) (Reptilia: Serpentes: Typhlopidae) in western Uganda JanR. Slapeta, Petr Neéas & David Modry Abstract: During a herpetological excursion to Western Uganda in August 1996, one specimen of Rhinotyphlops graueri (Sternfeld, 1912) was collected. The specimen was found dead in a shallow periodic water pool in the remaining rain forest on the pathway in the Semuliki National Park (0°50°N 30°5 E). The reported specimen is 271 mm long and 4.2 mm wide (at midbody) and has 24 scale rows at midbody. The head, with a rounded slightly prominent snout, is indistinct from the neck. The specimen is deposited in Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn (ZFMK 63.138). The presented finding is situated about 500 km north of the previous northernmost record and represents the first report of Rhinotyphlops graueri (Sternfeld, 1912) from Uganda. There is no evidence of a geomorphological or ecological border between the previously published localities and this new record, which lies just on the eastern margin of the Congo basin. Key words: Rhinotyphlops graueri, new record, Uganda Introduction Rhinotyphlops graueri was originally described as Typhlops graueri by Sremmcidn(! 912) The type locality is originally mentioned as “Ufer des Tanganyika, Urwald hinter den Randbergen am Nordwestufer des Tanganyika”. Laurent (1956) synonymized Typhlops avakubae Schmidt, 1923 with T. graueri Sternfeld, 1912, but this was stated as an error by Roux-Esteve (1974) who synonymized T. avakubae Schmidt, 1923 with Rhinotyphlops caecus (Dumeril, 1856). Finally, Roux-Esteve (1974) placed Typhlops graueri ınto the genus Rhinotyphlops and considered T. gracilis polli Laurent, 1960 (originally described as T. leptosoma polli Laurent, 1956) to be its junior synonym. R. graueri (Sternfeld, 1912) belongs to the “Group VI” of the african typhlopids (sensu Roux-Esteve 1974), together with a further eight species: R. caecus, R. pallidus, R. lumbriciformis, R. gracilis, R. rufescens, R. sudanensis, R. kibarae and R. wittei, all from tropical equatorial Africa from Fernando Poo (Bioko) to Zanzibar. The recently described Rhinotyphlops debilis from the Central African Republic also matches the mentioned “Group VI” (Joger 1990). During a herpetological excursion to Western Uganda, undertaken by the authors in August 1996, one specimen of Rhinotyphlops graueri was found dead in a shallow periodic water pool in tropical rain forest after a heavy night rain on a pathway crossing the Semuliki National Park. The specimen is now deposited at the Zoologisches Forschungsinstitut und Museum Alexander Koenig Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 106 Fig.1: Line drawing of a head ot Rhinotyphlops graueri (Sternfeld, 1912), ZFMK 63.138): A: ventral view, B: dorsal view, C: lateral view. in Bonn (ZFMK' 63.138). The collected specimen 1s 271 mm long and 4.2 mm wide and has 24 scale rows at midbody. The head is indistinct from the neck, with a round, slightly prominent snout (fig.1). The entire body is of a uniform pinkish colour. It is in all characters (sensu Sternfeld 1912, Roux-Estéve 1974, Broadley & Howell 1991, Meirte 1992) in accordance with Rhinotyphlops graueri, though was discovered out of the known distribution area of this species. According to Roux-Esteve (1974), Rhinotyphlops graueri is distributed in central tropical Africa below the equator around Lake Tanganyika in the Democratic Republic of Congo (former Zaire), Burundi and Tanzania. The northernmost record (RGMC 21495 —496), was collected near Rumonge in Burundi, identified by Roux-Esteve (1974) out of 14 examined Rhinotyphlops specimens. Further known records come from Burundi (two specimens from one locality), Tanzania (six specimens from two close localities) and the Democratic Republic of Congo (six specimens from four localities) — for details see appendix. The localities in Burundi and Tanzania are located within the remaining rainforest close to the Lake Tanganyika shore. The ') The following abbreviations for institutions are used in the text: ZFMK = Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn, Germany RGMC = Musée Royal de I’ Afrique Centrale, Belgium IRSN = Insitut Royal des Sciences Naturelles de Belgique MCZ = Museum of Comparative Zoology, Harvard University, USA 107 Fig.2: The preserved specimen of Rhinotyphlops graueri from | Uganda (ZFMK 63.138) - scale © bar=1cm localities in the Democratic Republic of Congo lay in the rainforest of the Congo Basin, 300 km west of the Lake Tanganyika shore. Our specimen was found about 500 km north of the previous northernmost record of Rhinotyphlops graueri. The Semuliki National Park is a relatively small area along the river Semuliki within the Ituri forest, belonging to the Congo basin. The fauna and flora of this area consists predominantly of species typical for the Congo basin, with some East African elements. There is no obvious geographical or ecological border between the previously known localities and our new record. Loveridge (1937) and Laurent (1954) published lists of forest snakes from the area around the “Great Lakes”. Both works were commented on by Hughes (1983) in his attempt to summarize the African snake fauna. According to all above-mentioned papers Rhinotyphlops graueri belongs to the species which are restricted to forests. Blind snakes are generally secretive snakes adapted to a fossorial life, usually found on the surface after rains and/or floods. This is the main reason why there are rather limited numbers of specimens in museum collections. This report extends the distribution range of Grauer’s Blind Snake, Rhinotyphlops graueri, ca. 500 km northward and represents the first record of this species north of the equator. Additionally, this finding adds a further species to the checklist of Ugandan herpetofauna. It can be expected that the distribution of this species extends along the “Great Lakes” (Tanganyika, Kivu and Edward) in tropical equatorial Africa, being confined to lowland rainforests. Acknowledgements We are deeply indebted to Dept./IUCN Kibale & Semuliki Development Project, Fort Portal, and Semuliki National Park for the opportunity of field work in western Uganda, and to all staff of Semuliki National Park for their understanding, technical assistance and facilitating our stay and research. We thank also Wolfgang Bohme for useful comments and assistance in the examination of the material in the Herpetological Collection of the Zoologisches Forschungsinstitut und Museum A. Koenig, Bonn. 108 References Broadley, D.G.& KM. Howell (1991)) A-CheekEist of the Repullesson Tanzania, with Synoptic Keys. — Syntarsus 1: 1-70. Hughes, B. (1983): African snake fauna. — Bonn. zool. Beitr. 34: 311-356. Joger, U. (1990): The herpetofauna of the Central African Republic, with description of a new species of Rhinotyphlops (Serpentes: Typhlopidae). — pp. 85-102 in: Vertebrates in the tropics (eds. Peters G. & R. Hutterer). —- Alexander Koenig Zoological Research Institute and Zoological Museum, Bonn. Laurent, R.F. (1954): Apercu de la biogéographie des batraciens et des reptiles de la région des grands lacs. — Bull. Soc. zool. France 79 (4): 290-310. Laurent, R.F. (1956): Contribution a l’herpetologie de la région des Grands Lacs de I’ Afrique centrale. I. Généralités. II. Chéloniens. III. Ophidiens. — Ann. Mus. r. Congo Belg., sér. in-8°, Sci. Zool., 48: 1-390. Loveridge, A. (1937): Scientific results of an expedition to rain forest regions in eastern Africa. IX. Zoogeography and itinery. — Bull. Mus. comp. Zool. Harv. 79: 479-541. Meirte, D. (1992): Cles de determination des serpents d’ Afrique. — Ann. Mus. Roy. Afr Cent. Ser. 8vo, Sci Zool, 267: 1 16]. Roux-Esteve, R. (1974): Revision systematique des Typhlopidae d’Afrique (Reptilia-Serpentes). — Mém. Mus. Nat. d’Histoire Natur. (N.S.), LXXXVII: 1-313. Sternfeld, R. (1912): Reptilia. pp. 197-279 in: Wissenschaftliche Ergebnisse der Deutschen Zentral-Africa Expedition 1907-1908.1V. — Leipzig. Appendix: Records of Rhinotyphlops graueri (Sternfeld, 1912) Democratic Republic of Congo: RGMC 1.985: (holotype of T. leptosoma poli): Katanga, Tanganyika district, ter. Albertville, Niemba; — IRSN 2.057: riv. Lubunduy, ter. Albertville; -RGMC 18.002-18005: (paratypes of T. leptosoma poli): Nyunzu, ter. Albertville; - RGMC 15.353: (paratype of T. leptosoma poli and T. kibarae ): Kabalo, ter. Kabalo. Burundi: RGMC 21.495-496: Rumonge, ter. Bururi. Tanzania: MCZ 48.051-53: Ujiji; - MCZ 48.054-56: Bagilo-Ujiji. Uganda: ZFMK 63.138: Semuliki National Park. Authors’ addresses: Jan R. Slapeta, Department of Parasitology, University of Veterinary and Pharmaceutical Sciences, Palackého 1-3, Brno, Czech Republic, CZ-612 42, e-mail: slapetaj @ vfu.cz. Petr Necas, Sportovni 11 Brno, Czech Republic, CZ-602 00 David Modry, Department of Parasitology, University of Veterinary and Pharmaceutical Sciences, Palackého 1-3, Brno, Czech Republic, CZ-612 42, e-mail: modry @ics.muni.cz or Institute of Parasitology of Czech Academy of Sciences of the Czech Republic, Ceské Budéjovice, Czech Republic 109 Developmental plasticity and behavioural adaptations of two West African anurans living in an unpredictable environment (Amphibia, Anura) Marko Spieler Abstract: Inthe Comoé National Park in Ivory Coast (West Africa) alternative reproductive strategies of two anuran species were investigated. Laboratory experiments showed that larvae of Bufo maculatus responded adaptively to lowered water levels by accelerating their development and metamorphosing smaller and faster than tadpoles grown in constantly high water volumes, whereas larvae experiencing increased water levels did not significantly change growth or developmental rates. First results in Hoplobatrachus occipitalis indicate that larvae of this species showed a fixed minimum developmental rate. Only the size at metamorphosis varies according to different water level conditions. In accordance with the high predation risk by different aquatic predators, two different types of adaptations were found. Tadpoles of B. maculatus aggregated in shallow water regions if attacked by cichlid fishes, the most important predators in the river spawning site of this toad species. These shallow areas are not accessible to the larger-sized predators. Experiments to determine the mechanism of aggregation behaviour showed that an alarm pheromone from injured B. maculatus tadpoles in combination with a mechanosensitive cue from rapid movements of aquatic predators elicits aggregation behaviour in conspecific tadpoles. The most important predators in the little rock pools, the main spawning site of H. occipitalis,are carnivorous conspecifics. The adults minimize the risk of cannibalism of their tadpoles by preferring of pools as spawning sites that have no or only few other conspecific tadpoles. Experiments showed that the density as well as the size of conspecific larvae in a pond is determined by the adults via chemical cues. Key words: Hoplobatrachus occipitalis, Bufo maculatus, phenotypic plasticity, reproductive strategy, aggregation behaviour, environmental stress Introduction Thirty-four anuran species are known from the Comoé National Park in Ivory Coast (West Africa) (Rödel 1996). In view of such a high number of species in a small area, different reproductive strategies to reduce competition can be expected (Mayhew 1968). The choice of different spawning sites is one of the potential possibilities (Resetarits and Wilbur 1989, Resetarits 1996). Therefore different adaptations to the specific conditions of the breeding sites utilized are to be expected. While most of the 34 anuran species breed in ephemeral savannah ponds, three frog species bred in two alternative spawning sites that were completely different to that of the savannah (Spieler 1997a). Bufo maculatus and B. regularis use shallow depressions on the flooded river plains with quickly changing water levels as their main breeding habitat. There the tadpoles have to cope with two main problems: 1) high desiccation risk once the larval habitat becomes isolated from the river, Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 110 2) heavy and unpredictable predation risk by cichlid fish that invade the breeding habitats when the river water level increases, leaving again when the water level decreases. Occasionally, these spawning sites allowed very successful breeding, but usually complete reproductive failure occurred (Linsenmair 1998). Hoplobatrachus occipitalis breeds at the very beginning of the rainy season in little pools on rocky river banks that are already completely filled after small amounts of the first local rainfall. If there is no additional precipitation during the next few days these little rock pools dry out completely. In contrast, the bigger savannah ponds did not exist at this early time. Only after several heavy rains the savannah soil is saturated with water, the ponds are filled and remain for longer periods. Then they are used as spawing sites by H. occipitalis, as well as by other anuran species in the area. Besides the high risk of desiccation in the rock pools, tadpoles have to cope with the high and long lasting predation risk from conspecifics which drastically increases in times of food shortage. In favourable years breeding in rock pools showed an unusually high reproductive success of more than 30 % survival rate, whereas in unfavourable years over 90% of the breeding pools did not produce a single metamorphosed froglet. In contrast, in the savannah pools the differences in survival rate between years are less pronounced, but usually not more than 2% of the larvae survived (Linsenmair 1998). The following two questions were at the center of our attention in this study. Do the larvae of B. maculatus and H. occipitalis respond adaptively to the decreased water levels that indicate the start of desiccation of their larval environment? In other words, do larvae in this situation accelerate their developmental rate at the cost of growth by metamorphosing earlier but at a smaller size than siblings that grow under conditions of constant or increasing water level? Do these larvae respond differently to the different predators, cichlid fish in the case of B. maculatus and carnivorous conspecific tadpoles in the case of H. occipitalis? Materials and Methods Study area The study was conducted in the Guinean savannah of Comoé National Park (8°5'- 9°6'N, 3°1'-4°4'W) in the northeastern part of the Ivory Coast, the largest National Park (11,500 km’) in West Africa. The study area is situated in the immediate surroundings of a research camp of the University of Würzburg. The tree-scrub- savannah in this region is characterized by distinct rainy and dry seasons, with a mean annual precipitation between 750 and 1100 mm during the last 7 years (our own measurements) and on average 1100 mm between 1977 and 1987 (Poilecot 1991). The core dry period lasting from December to February is usually without any precipitation. Most bodies of water in the study area are ephemeral ponds that always dry up in the dry season and usually also several times in the rainy period. River beds of the large rivers are the only places where pools of water persist throughout the dry season. The date of the first rain, as well as the amount and spatial distribution of the following rains at the beginning of rainy season, are extremely variable and unpredictable from year to year. 111 Studied animals Bufo maculatus Hallowell, 1854, a medium-sized toad (female: 41-60 mm, n=5; male: 40-54 mm, n=4) is widely distributed over Africa south of the Sahara. The clutches of B. maculatus had 4000-8000 eggs per string (n=5). The tadpoles did not disperse far from their hatching site but remained together at the same locality for several days. Larvae feed on algae and detritus. We chose to observe aggregation behaviour in B. maculatus instead of B. regularis Reuss, 1833 living sympatrically with B. maculatus on the banks of the Comoé river because this species was more common in the study area than B. regularis. Hoplobatrachus occipitalis' (Giinther, 1858) is broadly distributed over West Africa (Lamotte 1967). Sexually mature individuals vary greatly in body size fiomales- oles =ll.7 mm, range: 59-111 mm, n=42; males: 76.8+7.9 mm, range: 65-99 mm, n=43). Females attach their eggs, one at a time, to the bottom of shallow areas of pools. The number of mature eggs in the gonads of five medium sized females was 1590+310. The tadpoles are carnivorous and cannibalistic, preying on insect larvae and tadpoles of conspecifics and other anurans (Spieler & Linsenmair 1997). They have well developed horny jaws with a characteristic spine. Experimental design to investigate developmental plasticity In laboratory experiments both individuals and groups of tadpoles were raised under constantly low (LL-treatment group), constantly high (HH), decreased (HL) and increased (LH) water level conditions. The size of aquaria used was 19x9x6 cm. It was not known which water levels represented low or high water conditions for tadpoles. Therefore in one set of experiments, 320 ml of water were used for low water conditions (2 cm water) and 640 ml were used in aquaria with high water conditions (4 cm water). In another set of experiments, 160 ml of water (1 cm water) represented low water conditions and 320 ml (2 cm water) high water conditions. In experiments with individual tadpoles, five aquaria per treatment group were used in the case of H. occipitalis and 10 in the case of B. maculatus. Experiments with groups of 5 or 10 tadpoles in each aquarium, coming from the same clutch, were made only with B. maculatus (10 aquaria/treatment group). The water level was changed in treatment groups HL and LH when tadpoles had reached developmental stage 38 after Gosner (1960) in B. maculatus and 34 in H. occipitalis. Metamorphosis was reached with developmental stage 46. The water temperature in all experiments was 28+2°C (daily range: 26-30°C). Every fourth day acomplete water exchange was made. The larvae were fed ad lib, B. maculatus with Tetra Phyll (dry plants) and cleaned lettuce pieces of 5x5 cm, H. occipitalis with mosquito larvae. Means are given + standard deviations. For statistical analysis a STATISTICA/ Mac program by StatSoft 1991 was used. The significance level was set at p<0.05 unless stated otherwise. ' We have decided to classify this species as part of the genus Hoplobatrachus (formerly: Dicroglossus) listed by Duellman (1993) in accordance with a revision by Dubois (1992). This decision does not presuppose any intention to support this revision given the present state of knowledge. Results Developmental plasticity Larvae of B. maculatus responded adaptively to lowered water levels by accelerating their development and metamorphosing smaller and faster than tadpoles grown in constantly high water volumes. Siblings experiencing increased water levels did not significantly change growth or developmental rates in comparison with tadpoles grown in constantly low water volumes (fig.1). 80 70 D oO Ol (jo) larval period (day) body weight (mg) 40 30 LL 320 LH20 H Leo H Hsso Embers H Em H IM Fig.1: (A) Mean body mass of different treatment groups of B. maculatus at metamorphosis. (B) Mean developmental time from hatching to metamorphosis of the same tadpoles. Different letters above bars indicate significant differences (Tukey-type multiple comparisons). 40 body weight (mg) ow O1 larval period (day) & oO D O1 20 LL 160 LHıso HL o2o HA x20 alee Erlen ul HH, Fig. 2 (A) Mean body mass of the first metamorphosing froglet in each aquarium in the different treatment groups of B. maculatus. (B) Mean developmental time from hatching to metamorphosis of the same tadpoles. Different letters above bars indicate significant differences (Tukey-type multiple comparisons). 113 Similar results were found in experiments in which the water level was changed earlier, at Gosner-developmental stage 34, as well as in experiments in which 160 ml and 320 ml water levels were used (Spieler 1997b). Experiments with groups of B. maculatus tadpoles showed that in most cases the first metamorphosing froglet in each aquarium in the HL treatment group was smaller but earlier than the first metamorphosing froglet in each aquarium of the other three treatment groups (fig.2). Larvae of A. occipitalis experiencing decreased water levels metamorphosed at a smaller size; those experiencing increased water levels metamorphosed at a larger size than siblings grown in corresponding constant water levels. Unexpectedly, larvae of all treatment groups showed a similar developmental rate (fig.3). 60 50 = oS = 3 ° 2 30 = 2 > — xe) a) 8 = 20 LL 350 LAs 0 H Loo H Hy le 320 L es H 25 H His Fig. 3 (A) Mean body mass at, and (B) mean timing of metamorphosis of the different treatment groups of A. occipitalis tadpoles. Because of low sample size in some treatment groups no statistical comparisons can be made. Reaction of tadpoles to different predators The main spawning sites of B. maculatus are shallow depressions on the margins of rivers, where tadpoles of this toad species are either evenly distributed or aggregated. Two types of aggregations were observed. One moving in swarms on the bottom and one stationary, with the latter concentrated on small sections at the edges of puddles in extremely shallow regions of 0.1-0.5 cm of water (fig.4). The stationary aggregations contained up to several thousand individuals on an area of 10-20 cm in diameter. Observations and exchange-experiments showed that tadpoles formed stationary aggregations in shallow water regions only in the presence of predators (Spieler and Linsenmair 1999). These areas are not accessible to the larger predators. The most frequent predators found were young cichlid fish and at lower abundance the carnivorous tadpoles of A. occipitalis. Fig.4: Aggregated (left) and non-aggregated (right) tadpoles of B. maculatus in the Comoé river Semi-natural experiments showed that a species-specific or genus-specific alarm pheromone released by injured B. maculatus tadpoles in combination with a mechanosensitive cue from the quick movements of aquatic predators elicits aggregation behaviour in conspecific tadpoles (Spieler & Linsenmair 1999). 40 30 =~ O) = St eed c oO) 2 0 ©® = > ge) 2 10 0 day 0 day O day 5 day 5 without with without with predators predators predators predators Fig.5: Relationship between mean body mass (+/-SD) of aggregated medium-sized B. maculatus tadpoles (aggregation caused by predators) and non-aggregated sibling tadpoles in artificial ponds without predators. After 6 days the mean size of aggregated tadpoles was significantly lower in comparison with non-aggregated larvae (Mann-Whitney U-Test). 115 Since feeding opportunities should be reduced within a stationary aggregation, spending any extended time aggregated will lead to either a longer developmental time and/or will negatively affect the size at metamorphosis. Groups of tadpoles in natural pools, one aggregated and the other evenly distributed, with an initially identical body mass, showed significant differences in mass gain already after 2 days (Spieler & Linsenmair 1999). Similar results were found in experiments with tadpoles in artificial ponds (fig.5). Cichlids usually leave the puddles when falling water levels reach a critical limit. The aggregated tadpoles then dissolve their aggregations after some hours and begin feeding to an increased extent. The most effective predators of H. occipitalis tadpoles in the rock pools are the carnivorous conspecific larvae (Spieler 1996). Hatchlings have only low survival rate in pools with older conspecifics. Therefore reproducing A. occipitalis lay their eggs preferably in breeding ponds containing no or only a few other conspecific tadpoles. Experiments showed that the density of conspecific larvae in a pond is determined by the adults by means of chemical cues (Spieler & Linsenmair 1997). In addition, adult frogs showed risk-spreading behaviour. Most radio-tracked frogs visited and also spawned in more than one rock pool during one laying bout (Spieler and Linsenmair 1998b). They seemed to spawn only a small fraction of their mature eggs in one night. Only some females, laying during the first spawning bout of the year in the largest rock pool that contained no other conspecific larvae, spawned all their approximately 1600 mature eggs in this pool. When rock pools were flooded by the increasing water level of the river, most AH. occipitalis left the river bed and migrated to savannah pools. Several individuals travelled distances of up to 6 km and each frog visited thereby up to 10 savannah ponds to spawn there (Spieler 19970): Discussion Sympatrically-living anurans in the study area possess very different life-history strategies (Spieler 1997a, Linsenmair 1998). All these anurans are confronted with one major problem, the unpredictable variability of the rainfall pattern in time and space. For species that produce only few but large eggs, one of the best tactics to confront such unpredictable conditions is to spread the risk by sparingly spending eggs and always retaining enough for the next breeding opportunity, as well as to select only suitable pools as spawning sites by including in their choice all available abiotic and biotic parameters (e.g. size and seepage of pools, density of predators in the pool) that characterize an individual pool (Spieler & Linsenmair 1998a). This strategy seems to have evolved in H. occipitalis. Another tactic is to Spawn numerous and small eggs and to leave it up to the tadpoles to react to the variable conditions of their environment. This seems to be realized in B. maculatus. At the larval stage, the best tactic seems to have a plastic aquatic developmental rate corresponding to water level variations. This strategy seems to have evolved in tadpoles of B. maculatus. Hoplobatrachus occipitalis is the only species in the frog community of the Comoé National Park that breeds regularly after the earliest rains (Spieler & Linsenmair 1997). This is possible because the species uses small rock pools at the river bank as spawning sites that have been completely filled after the first small 116 rainfalls. But these pools have severe disadvantages for reproducing frogs. Because of the relatively small size of these rock pools compared with the bigger savannah ponds the risk of desiccation is very high. On rocky areas at the river bank of approximately 0.01 km’ more than 100 rock pools could be present. The quality of these pools varies to their water depth and seepage. According to the prediction, adult H. occipitalis minimize the risk for their tadpoles to dry out by distributing their eggs over several pools and prefer those pools that contain water for a longer period (water-holding capacity: WHC, Spieler & Linsenmair 1997). In addition, adult frogs minimize the risk of their tadpoles to becoming victims of cannibalism by preferring pools as spawning sites that are still unoccupied by conspecific tadpoles (Spieler & Linsenmair 1997). The important role of cannibalism in the survival rate of tadpoles was emphasized by Petranka & Daphne (1995). But few indications are present that predators influence the choice of spawning sites in frogs (Beebee 1985, Resetarits & Wilbur 1989, IIdas & Ancona 1994). In contrast, Bufo maculatus uses spawning sites on the margins of rivers. The quality of these sites depends on the level of groundwater, the water level of rivers and the amount of local rainfall. If the water level increases, predatory fish can invade the spawning sites, which they have to leave when the water level falls. In view of the quickly and unpredictably changing water level, the quality of these spawning sites ıs not detectable for frogs at the time of spawning. That is possibly the reason why we found no strong spawning-site selection in this species. Here, the tadpoles have to cope with the high risk of predation and desiccation and have evolved different strategies against these risks. As a reaction to the high predation pressure, tadpoles form in the presence of predators tight and long-lasting aggregations, at the cost of a marked decrease in growth (Spieler & Linsenmair 1999). Similar behaviours are described in a few other anuran species (Bragg 1965, Hews 1988, Caldwell 1989, Waldman 1991, Rodel & Linsenmair 1997). Developmental plasticity An further controverse point is, whether developmental rate and growth of tadpoles are strongly coupled and could vary only together in relation with biotic and abiotic parameters or whether developmental rate and growth of tadpoles are more independent and could allow a trade off between this two. It should be highly adaptive in an unpredictable environment to accelerate the developmental rate at the cost of growth and to metamorphose faster but with smaller size in the case of decreasing water level in comparison with siblings that grown under constant or increasing water level conditions after new rain falls (Wilbur & Collins 1973). This hypothesis was supported in some studies under natural conditions (Savage 1961, Brockelmann 1969, Collins 1979, Newman 1988, 1989, Crump 1989). First results in A. occipitalis indicate that larvae of this species showed a fixed very fast development (about 25 days from egg laying to metamorphosis under natural conditions) at different water level conditions that indicate a drying out of pools or improving conditions after new rain falls. Only the size at metamorphosis varies depending on the different water level conditions. But perhaps this is a result of small sample size in our study. Preliminary observations suggested that final developmental stages, from the appearance of the front limbs to the complete 17 resorption of the tail, are very flexible, facilitating a prolonged stay in water in the case of favourable aquatic environment. Larvae of B. maculatus react to a large extent as predicted by the hypothesis of Wilbur & Collins (1973). Single tadpoles that had grown in decreasing water level conditions accelerated their development, metamorphosing smaller and faster than tadpoles that had grown in constantly high or increasing water volumes, indicating improved conditions after rainfall. Likewise, Bufo bufo and Bombina variegata that were investigated under similar conditions showed similar results (Böll et al. 1997). The best evidence for a tradeoff between development and growth in anurans was found in Scaphiopus hammondii (Denver et al. 1998). The control of the adaptive phenotypic plasticity is presumably carried out by hormones stimulated by different environmental cues (Denver 1997a, 1997b, 1998). If groups of tadpoles are kept together in a single aquarium a crowding effect results, described in many studies (Brockelmann 1969, Gromko et al. 1973, Heusser & Blankenhorn 1973, Wilbur 1977a, 1977b, Collins & Cheek 1983, Newman 1987, Semlitsch 1987, Wong & Beebee 1994). If conspecific tadpoles are kept together, the larvae grow slower and metamorphose at a smaller size. Accordingly, in experiments with groups of B. maculatus tadpoles, the developmental time was longer and frogs at metamorphosis were smaller in comparison to experiments with single tadpoles. In treatment group HL increasing crowding effects in comparison to the HH group are to be expected. In agreement with this prediction, the frogs at metamorphosis were very small. Surprisingly, developmental time of the first metamorphosing froglet in each aquarium was shorter than in the HH-treatment group; not longer as we expected because of crowding (Spieler 1997b). Therefore, first metamorphosing froglets of each aquarium react exactly as predicted by the hypothesis of Wilbur & Collins (1973). A possible explanation is this: the crowding effect enlarged the variation in size and developmental stages of larvae within each group (Licht 1967, Gromko et al. 1973, Wilbur 1977a, Sokol 1984). At the time of changing water level in the HL group, the biggest tadpoles in each aquarium could accelerate their developmental rate at the cost of the smaller siblings and at the cost of individual growth. One further open question is whether all anuran species show potential developmental plasticity or whether this possibility is only restricted to anurans living in — at least sometimes — unpredictable environments. Acknowledgements I thank W. Böhme, U. Grafe and A. Ohler for valuable comments on an earlier draft of the manuscript, as well as K.E. Linsenmair for help during the fieldwork and laboratory work. I also thank my students participating in our field course on tropical ecology and helpful graduates for assistance with fieldwork. This study was supported by Deutsche Forschungsgemeinschaft (SFB 251, Project No. B3) and by the Volkswagen-Stiftung (AZ 1/64 102). We thank the Ministre des Eaux et Foréts and the Ministre de la Récherche Scientifique, République de Cote d Ivoire for granting the research permit for doing field work in the Park National de la Comoé. 118 References Beebee, T.J.C. (1985): Discriminate analysis of amphibian habitat determinant in South-East England. — Amphibia-Reptilia, Leiden 6: 35-43. Boll, S.,M.Spieler, U. Grafe & K.E. Linsenmair (1997): Adaptations of temperate and tropical anuran larvae to the risk of desiccation and predation. — pp. 24-25 in: Third World Congress of Herpetology (eds. R. Zbynék & H. Scott). - Durabo Celakovice, Prag. Bragg, A.N. (1965): Gnomes of the night: the spadefoot toads. — University of Pennsylvania press. Brockelmann, W.Y. (1969): An analysis of density effects and predation in Bufo americanus tadpoles. — Ecology 50: 632-644. Caldwell, J.P. (1989): Structure and behavior of Hyla geographica tadpole schools, with comments on classification of group behavior in tadpoles. — Copeia 1989: 938-950. Collins, J.P. (1979): Intrapopulation variation in the body size at meta- morphosis and timing of metamorphosis in the bullfrog, Rana catesbeiana. — Ecology 60: 738-749. Collins, J.P. & J.E. Cheek (1983): Effect of food and dans on development of typical and cannibalistic salamander larvae in Ambystoma tigrinum nebulosum. — Amer. Zool. 23: 77-84. Crump, M.L. (1989): Effect of habitat drying on developmental time and size at metamorphosis in Ayla pseudopuma. — Copeia 3: 794-797. Denver, R.J. (1997a): Environmental stress as a developmenealr we Corticotropin-releasing hormone is a proximate mediator of adaptive phenotypic plasticity in amphibian metamorphosis.— Hormon. Behav. 31: 169-179. Denver, R.J. (1997b): Proximate mechanisms of phenotypic plasticity in amphibian metamorphosis. — Am. Zool. 37: 172-184. Denver, R.J. (1998): Hormonal correlates of environmentally induced metamorphosis in the Western spadefoot toad, Scaphiopus hammondii. — Gen. Comp. Endo. 110: 326-336. Denver, R.J., N. Mirhadi & M. Phillips (1998): Adaptive plasticiiy, mi amphibian metamorphosis: Response of Scaphiopus hammondii tadpoles to habitat desiccation. — Ecology 79: 1859-1872. Dubois, A. (1992): Notes sur la classification des Ranidae (Amphibiens Anoures). — Bull. mens. Soc. linn. Lyon 61: 305-352. Duellman, W.E. (1993): Amphibian species of the world: additions and corrections. — Univ. Kansas Mus. Nat. Hist., Special Publications 21: 1-372 Gosner, K.L. (1960): A simple table for staging anuran embryos and larvae with notes on identification. — Herpetologica, Lawrence, Kan. 16: 183-190. Gromko, M.H., F.S. Mason & S.J. Smith-Gill (1973): Analysis of the crowding effect in Rana pipiens tadpoles. — J. Exp. Zool. 186: 63-72. Heusser, H. & H.J. Blankenhorn (1973): Crowding — Experimente mit Kaulquappen aus homo- und heterotypischen Kreuzungen der Phänotypen 119 esculenta, lessonae und ridibunda (Rana esculenta- Komplex, Anura, Amphibia). — Rev. Suisse Zool. 80: 543-569. Hews, D.K. (1988): Alarm response in larval western toads, Bufo boreas: release of larval chemicals by a natural predator and its effect on predator capture efficiency. — Anim. Behav. 36: 125-133. Ildas, A.S. & N. Ancona (1994): Analysis of amphibian habitat preferences in a farmland area (Po plain Italy). - Amphibia-Reptilia, Leiden 15: 307-316. Lamotte, M. (1967): Les batraciens de la region de Gpakobo (Cöte d’ Ivoire). — Bull. Inst. Fondam. Afr. Noire 19: 219-294. Licht, L.E. (1967): Growth inhibition in crowded tadpoles: intraspecific and interspecific Effects. — Ecology 48: 736-745. Linsenmair, K.E. (1998): Risk spreading and risk reducing tactics of West African anurans in an unpredictable and stressful environment. — pp. 221-241 in: Population and community dynamics in the tropics (ed. Newbery, D.M.). — Annual symposia of the British Ecological Society 1996. Mayhew, W.W. (1968): Biology of desert amphibians and reptiles. — pp. 196- 356 in: Desert biology. (ed. G.W. Brown). — Academic Press, New York. Newman, R.A. (1987): Effects of density and predation on Scaphiopus couchii tadpoles in desert ponds. — Oecologia 71: 301-307. Newman, R.A. (1988): Adaptive plasticity in development of Scaphiopus couchii tadpoles in desert ponds. — Evolution 42: 774-783. Newman, R.A. (1989): Developmental plasticity of Scaphiopus couchii tadpoles in an unpredictable environment. — Ecology 70: 1775-1787. Petranka,J.W.& A.G.T. Daphne (1995): Explosive breeding reduces egg and tadpole cannibalism in the wood frog, Rana sylvatica. — Anim. Behav. 50: WON 739. Poilecot, P. (1991): Un écosystéme de savane soudanienne: Le Parc National de la Comoé (Cote dIvoire). — Paris, Organisation des Nations Unies, pour l’education, la science et la culture, pp 346. Resetarits,W.J.& H.M. Wilbur (1989): Choice of oviposition site by Ayla chrysoscelis: Role of predators and competitors. — Ecology, Washington 70: 220- LUCE Resetarits, W.J.J. (1996): Oviposition site choice and life history evolution. — Amer. Zool. 36: 205-215. Rodel, M.O. (1996): Amphibien der westafrikanischen Savanne. — Chimaira, Frankfurt. Rodel, M.O. & K.E. Linsenmair (1997): Predator induced swarms in the tadpoles of an African savannah frog, Phrynomantis microps. — Ethology 103: 902-914. Savage, R.M. (1961): The ecology and life history of the common frog. — Sir Isaac Putnam and Sons Ltd., London. Semlitsch, R.D. (1987): Paedomorphosis in Ambystoma talpoideum: Effects of density, food, and pond drying. — Ecology 68: 994-1002. 120 Sokol, A. (1984): Plasticity in the fine timing of metamorphosis in tadpoles of the hylid frog, Litoria ewingi. — Copeia 1984: 868-873. Spieler, M. (1996): Strategien der Streßvermeidung bei westafrikanischen Anuren. — Verh. dtsch. Zool. Ges., Stuttgart, 89: 260. Spieler, M. (1997a): Anpassungen ausgewählter Froschlurche an Trockenstreß und Räuberdruck in einer westafrikanischen Savanne. — Salamandra 33: 133-151. Spieler, M. (1997b): Anpassungen westafrikanischer Anuren an Austrock- nungsrisiko und Räuberdruck in einem saisonalen Lebensraum. — Wissenschaft und Technik, Berlin, pp. 181. Spieler, M. (1997c): Radio-telemetrische Untersuchungen zur Laichplatzwahl eines westafrikanischen Raniden. — Mertensiella, Bonn 7: 203-220. Spieler, M. & K.E. Linsenmair (1997): Choice of optimal oviposition site by Hoplobatrachus occipitalis (Anura: Ranidae) in an unpredictable and patchy environment. — Oecologia 109: 184-199. Spieler, M. & K.E. Linsenmair (1998a): Biotic and abionestactons influencing the spawning site selection of savanna anurans. — Zoology, Supplement I, 101: 15. Spieler,M.& K.E. Linsenmair (1998b): Migration patterns and diurnal use of shelter in a ranid frog of a West African savannah: a telemetric study. — Amphibia-Reptilia 19: 43-64. Spieler,M. & K.E. Linsenmair (1999): Aggregation behaviour of tadpoles of Bufo maculatus as an effective antipredator mechanism. — Ethology 105: 665- 686. Waldman, B. (1991): Kin recognition in amphibians. — pp. 162-219 in: Kin recognition (ed. Hepper, P.G.). - Cambridge University Press. Wilbur, H.M. (1977a): Density-dependent aspects of growth and metamorphosis in Bufo americanus. — Ecology 58: 196-200. Wilbur, H.M. (1977b): Interactions of food level and population density in Rana sylvatica. — Ecology 58: 206-209. Wilbur, H.M. & J.P. Collins (1973): Ecological aspects of amphibian metamorphosis. — Science 182: 1305-1314. Wong,A.& T. Beebee (1994): Identification of a unicellular, non-pigmented alga that mediates growth inhibition in anuran tadpoles: a new species of the genus Prototheca (Chlorophyceae: Chlorococcales). — Hydrobiologica 277: 85- 96. Author’s address: Dr. Marko Spieler, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 150-164, 53113 Bonn, Germany Flora and vegetation of the Afromontane region in Central and East Africa Eberhard Fischer Abstract: A survey of the Afromontane and Afroalpine vegetation is provided. The vegetation belts of Mt. Kahuzi, the Virunga volcanoes and Mt. Kenya are briefly described. Phytogeography of the East African mountains is discussed and the distribution patterns of Dendrosenecio, the giant Lobelias and Impatiens within the Afromontane region are described. Some hypotheses on the origin of the Afroalpine flora are briefly discussed. Key words: Afromontane region, Mt. Kahuzi, Virunga volcanoes, Mt. Kenya, Dendrosenecio, Lobelia, Impatiens Introduction Africa contains several isolated mountain ranges (fig.1). Their flora consists of more than 80% endemics (Hedberg 1961a). They have been called the afromontane and afroalpine archipelago-like centre of endemism (White 1983). In Central and East Africa, these mountains are situated along the Rift valley, the most important being Mt Kahuzi (3305 m a.s.l.) in eastern Congo, the eastern crest of the Central Rift valley with the Nyungwe Forest (up to 3000 m) in Rwanda, the Virunga volcanoes (4507 m) in Rwanda, Congo and Uganda, the Ruwenzori (5109 m) in Congo and Uganda, Mt Elgon (4321 m) in Uganda and Kenya, the Aberdare Mountains (3994 m) and Mt Kenya (5194 m) in Kenya, and Kilimanjaro (5895 m) and Mt Meru (4565 m) in Tanzania. There exists a vast literature on afromontane and afroalpine flora and vegetation (ee nauman 1955, Hedberg 1951, 1957, 1961a, 1965, 1969, 1986, White 1978, 1983). Several studies on single mountains have also been published (e.g. Mt Kahuzi: Fischer 1996, Hendrickx 1946, Scaetta 1934; Nyunge Forest: Fischer & Hinkel 1990, 1992, 1993, 1994, Habyaremye 1993; Virunga volcanoes: Snowden 1933; Ruwenzori: Demaret 1958, Hauman 1933, Schmitt & Beck 1992; Mt Kenya: Beck et al. 1981, Rehder et al. 1988; Aberdare Mountains: Schmitt 1991; Kilimanjaro: Klötzli 1958), but there is still a lack of vegetation analysis. The following paper intends to give an overview of the afromontane and afroalpine vegetation and their distribution patterns in Central and East Africa. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 RAT rN La 8 — Se eee ATTA iM IH (crane PH Ee ae 4 ay ail VAN ned ae ER ae 20 Fig.1: Main phytochoria of Africa (after White 1983): 1. Guineo-Congolian regional centre of endemism; 2. Guineo-Congolia-Sudania-Transition zone; 3. Lake Victoria regional mosaic; 4. Guineo-Congolia-Zambesia Transition zone; 5. Sudania regional centre of endemism; 6. Zambezian regional centre of endemism; 7. Zanzibar-Inhambane regional mosaic; 8. Tongoland-Pondoland regional mosaic; 9 Afromontane archipelago-like regional centre of endemism (black); 10. Somalia-Masai region; 11. Kalahari-Highveld Transition zone; 12. Karroo-Namib region; 13. Cape region; 14. Sahel Transition zone; 15. Madagascar - Altitudinal zonation A detailed description of altitudinal zonation in East Africa was provided by Hedberg (1951). Generally, these high mountains can be divided into the western group (Mt Kahuzi, Nyungwe, Virunga volcanoes and Ruwenzori) with a much more humid climate, and the eastern mountains (Mt Elgon, Aberdares, Mt Kenya, Kilimandjaro, Mt Meru) with a considerably drier climate. Only Mt Kahuzi and Ruwenzori have a direct contact to the lowland rainforests of the Congo basin, while the other mountains are much more isolated and surrounded by savanna vegetation. The vegetation belts of the Kahuzi massif (fig.2), not mentioned by Hedberg, are briefly described here (see also Fischer 1996). From 850 m to 1300 m a.s.l. a mountain rainforest is present. There is a Transition Forest zone from 1300 to 1700 m, followed by a montane forest from 1700 to 2300 m (2400 m). From 2400 to 2600 m the bamboo zone occurs, intermixed with Podocarpus-forest on hill summits, swamps and in various montane forest communities. A Hagenia- Hypericum zone, which is well developed on the Virunga-Volcanoes, is lacking. The Ericaceous belt begins at 2600 m (Mt Biega) and reaches to 3200 m on Mt Kahuzi. The summit of Mt Kahuzi bears a Dendrosenecio johnstoni-Erica- subparamo and a Helichrysum-Lobelia stuhlmannii-paramo. In spite of the isolate position, Mt Kahuzi has several afroalpine species in common with the Virunga 5000 A A En (1a Ba Ns [I lle RITA G, 2 Di WIP \ RECA _———S SS (ees ee ES en en a b C Fig.2: Altitudinal zonation of Ruwenzori (a), Virunga volcanoes (b) and Kahuzi (c) (after Fischer 1996): 1 Paramo, 2 Podocarpus forest, 3 Ericaceous belt, 4 Hagenia-Hypericum zone, 5 Bamboo zone, 6 Montane forest belt. 124 volcanoes and the Ruwenzori. Only Swertia macrosepala, Erica johnstoni, Bartsia macrophylla, Lobelia stuhlmannii, Helichrysum guilelmi and Disa stairsii need be mentioned. The giant groundsel Dendrosenecio johnstoni has proved to be a different subspecies (ssp. kahuzicus). The zonation of the vegetation at the Virunga volcanoes (fig.2) is comparable. Below 2700 m, the original montane forest has been destroyed by clearcutting for agricultural purposes. From 2700 to 3000 m, a secondary Dombeya-montane forest with scattered Hagenia has developed, followed by a Hagenia-Hypericum belt from 3000 to 3300 m, where large moss cushions are found. On the saddle of Karisimbi at 3400 m a moorland with Dendrosenecio johnstoni and Erica Johnstoni occurs. Around Lake Muderi, a Sphagnum peat bog with Carex runssorensis has developed. Above 3400 m there is a Dendrosenecio johnstoni- Hypericum revolutum-subparamo. Due to acidic soil conditions, an Erica- Philippia forest, comparable in its floristic composition with that on Mt Kahuzi, has developed above Lake Muderi and on the steep slopes of the Susa-valley. The beginning of the paramo is at 3600 m. It can be divided into two types: the Dendrosenecio johnstoni-Lobelia stuhlmannii-paramo from 3600 to 3900 m, and the Dendrosenecio johnstoni-Lobelia wollastoni-paramo from 3900 to 4200 m. Above 4200 m no more giant groundsel are found and nearly pure meadows of Alchemilla johnstoni are well developed. The summit at 4507 mis covered by an alpine desert, where bryophytes and lichens dominate. As an example of a drier eastern mountain, the vegetation belts of Mt Kenya are briefly described (see also Hedberg 1951, Rehder et al. 1988, Bussmann & Beck 1996). The montane forest belt has been divided into a lower montane forest zone from 2000 to 2700 m, the bamboo zone with dominating Sinarundinaria alpina from 2700 to 3200 m and the Hagenia-Hypericum zone from 3200 to 3400 m. This forest belt has been studied phytosociologically by Bussmann & Beck (1996). From 3400 to 3600 m, an Ericaceous belt has developed, which is considered by Rehder et al. 1988 to be part of the lower alpine zone. This Ericaceous bush is characterized by a dense cover of tussock, mainly Carex monostachya, which alternates with heath formations (Erica arborea, Philippia keniensis, Stoebe kilimandscharica). Thus it differs fundamentally from the dense Ericaceous bush on the humid western mountains. Above 3600 m, a paramo is developed, where several Dendrosenecio and Lobelia species (D. keniodendron, D. keniensis, Lobelia telekii, L. deckenii subsp. keniensis) are found. This alpine vegetation is also much less dense than that on the Virunga volcanoes or Ruwenzori. Dendrosenecio keniodendron is mainly distributed in the upper paramo up to 4400 m. The nival zone is characterized by bare soil due to the abundant solifluction and herbaceous vegetation. The rocks are densely covered with lichens most abundant among them Alectoria sarmentosa and Usnea spp. Phytogeographical aspects About 80% of the Afromontane and Afroalpine flora are endemic to the massifs of Central and East Africa (Hedberg 1961a). Vicarious taxa occur of different status. Some species are distributed on all or most of these mountains, like Subularia monticola, Ranunculus oreophytus, Swertia macrosepala or Disa stairsii. Hedbergia abyssinica is also found in the high mountains of Ethiopia, Cameroon and Malawi. In some cases, species are confined to group of mountains. A good example is the genus Bartsia, where B. macrophyyla occurs on Mt Kahuzi, the Virunga volcanoes and the Ruwenzori, while B. longiflora and B. decurva are found on Mt Elgon, the Aberdares, Mt Kenya and Kilimanjaro. Fig.3: Distribution of Dendrosenecio johnstoni (after Hedberg 1986 and Mabberley 1986). The most interesting and also most complicated genus of Afroalpine plants is Dendrosenecio (fig.3). Much literature exists on distribution and biology (e.g. Mildbraed 1922, Fries & Fries 1922a, Humbert 1934, Beck et al. 1982), but the taxonomy is still controversial. While Hedberg (1957, 1969) distinguished 17 species, Mabberley (1974, 1986) lumped them into four species, one of them (D. Johnstoni) with eight subspecies. Recently, Knox (1994) argues for a narrower species concept, but more studies on ecology, morphology and molecular biology are required. Until a modern revision, the slightly modified concept of Mabberley (1986) is accepted. These taxa do not occur below 3000 m and are likely to have originated from one or two diaspores brought by long-distance dispersal. The most widespread species is Dendrosenecio johnstoni (fig.3). D. johnstoni subsp. kahuzicus is endemic on Mt Kahuzi, subsp. adnivalis occurs on the Virunga 126 volcanoes and Ruwenzori, while the var. friesiorum is endemic to Ruwenzori. Dendrosenecio johnstoni subsp. cheranganiensis and subsp. dalei are both endemics on the Cherangani Hills in Kenya, the subsp. elgonensis and subsp. barbatipes are confined to Mt Elgon, and subsp. battiscombei is restricted to Mt Kenya and the Aberdare Mountains. On Kilimanjaro subsp. cottonii and subsp. Johnstoni are found, the latter being present also on Mt Meru. Mt Kenya harbours two endemic species, Dendrosenecio keniodendron and D. keniensis (= Senecio brassica). One species, D. brassiciformis, is confined to the Aberdare Mountains. A similar picture is provided by the giant lobelias (Fries & Fries 1922b, Hedberg 1957, Thulin 1984). Lobelia deckenii, with its vicarious subspecies, is a good example of a rapid genetic drift, which originated in differences in splitting of corolla and pubescense of bracts, sepals, petals and anthers (Hedberg 1969). L. deckenii subsp. deckenii is endemic on Kilimanjaro, subsp. elgonensis on Mt Elgon, subsp. keniensis on Mt Kenya, subsp. sattimae on the Aberdare Mountains, and subsp. burtii on Mt Meru and the Ngorongoro crater. Lobelia deckenii subsp. bequaertii is the westernmost taxon of this group and confined to the Ruwenzori. Other Lobelia species are found on several mountain systems, but no differentiation into subspecies can be observed. Lobelia wollastonii is endemic to the Virunga volcanoes and the Ruwenzori, while its vicarious species L. telekii is restricted to Mt Elgon, Mt Kenya and the Aberdare Mountains. The closely related L. stuhlmannii is found on Mt Kahuzi, the Virunga volcanoes and Ruwenzori. Other vicarious species pairs are Lobelia bambuseti on Mt Kenya and Aberdare Mountains and L. petiolata on Mt Kahuzi and the Virunga volcanoes, as well as L. aberdarica on Mt Elgon, Mt Kenya, and the Aberdare Mountains and L. mildbraedii on Mt Kahuzi, the Virunga volcanoes, the Congo-Nile watershed and Ruwenzori. Finally, the distribution of several /mpatiens species (Grey-Wilson 1980) can be regarded as a model system. The /mpatiens filicornu aggregate 1s represented on Mt Kahuzi, the Congo-Nile watershed, the Virunga volcanoes and Ruwenzori with Impatiens bequaerti, I. mildbraedii subsp. mildbraedii, I. erecticornis and I. purpureo-violacea. The closest relative, Impatiens mildbraedii subsp. telekii is restricted to Mt Kenya and the Aberdare Mountains. The /mpatiens stuhlmannii aggregate, with /. apiculata, I. masisiensis and I. warburgiana, 1s endemic to the western group (Mt Kahuzi, the Congo-Nile watershed, the Virunga volcanoes, Ruwenzori) and only /. stuhlmannii is also found on Mt Elgon. Of the /mpatiens tinctoria aggregate, only /. tinctoria subsp. finctoria is found on Ruwenzori, while I. tinctoria subsp. elegantissima is endemic to Mt Elgon, Mt Kenya and the Aberdares. Impatiens fischeri is restricted to Mt Kenya and the Aberdare Mountains. The western humid mountains are apparently a centre of endemism for Impatiens. The Impatiens kilimanjari aggregate 1s represented here by /mpatiens runssorensis (Ruwenzori), /. bururiensis, I. gesneroidea, I. quadrisepala (Congo- Nile watershed) and /. superglabra (Mt Kahuzi), while /. kilimanjari is endemic to Kilimanjaro. /mpatiens keilii and I. paucidenta from the Impatiens gomphophylla aggregate are also restricted to the Congo-Nile watershed in the Central Rift Valley. On Mt Elgon, the following taxa of this group are endemic: /mpatiens tweediae, I. miniata and I. digitata subsp. phlyctidoceras. Impatiens digitata subsp. digitata is confined to Kilimanjaro. The Impatiens rubromaculata aggregate is 127 found only on the eastern mountains. /mpatiens hoehnelii and I. meruensis subsp. cruciata are endemic on Mt Kenya and Aberdares, while /. meruensis subsp. meruensis, I. rubromaculata subsp. rubromaculata and I. nana are found only on Kilimanjaro. These distribution patterns may be explained by long-distance dispersal and sub-sequent radiation of a taxon. However, we also have to consider the refugial hypothesis of Hamilton (1974, 1982). There is strong evidence (Bonne- le mi Bea N ee LAU Sans AIRES . Nabe rer ai BAT ER? : JERE el ELLER & ECE reat BERBRN SUR nn. aaa wi I a Beit, HY Fig.4: Distribution of Erica arborea (after Fischer 1996). Closed distribution area hatched, subfossil records (open circles). 128 fille et al. 1990, Grunderbeek et al. wee t dur So ela a en Be ee the climate in Africa and an ulti a lo etation belts and a for ir in a sts ns nly in ee artic Ae SEE RTS te a er S Sr ei Be FERN RN 23 Se / meee 12, 202 lee: la a REIN no | ie EP hee eetka COEUR ES Tg ee CREAT HY me cUCCP ET Fig.5: Distribution of Canarina eminii (circles) and C. < s (stars) (after Hedberg 1961b). 129 refugial areas (see also Fischer 1996) and the distribution pattern of some Panomontane species periectly reflects these refugia. The orisin of the Afromontane flora by migrating taxa during the Tertiary period is supported by an interesting phytogeographical link between Macaronesia and East Africa (Lösch & Fischer 1994). Here, in the Ericaceous shrub vegetation, numerous vicarious taxa could be demonstrated, among them Erica arborea (fig.4) which is present in Macaronesia, the Mediterranean region and the East African mountains as well. Steping-stones can be found in the Saharan region by subfossil records and even a recent locality. Another example for vicarious taxa is provided by the genus Canarina (Hedberg 1961b) (fig.5), where C. canariensis 1s endemic to the Canary islands and C. eminii and C. abyssinica are East African species. Also the genus Dendrosenecio and the giant Lobelias may have originated during the Tertiary. As pointed out above, a long distance dispersal between isolated mountains and subsequent radiation might have been important for the evolution of the afromontane flora. References moe aMoowNehder. BP. Pomeratz & M. Senser (1981): Ecological analysis of the boundary between afroalpine vegetation types “Dendrosenecio woodlands” and “Senecio brassica-Lobelia keniensis community” on Mt Kenya. — J. East African Nat. Hist. Soc. and Nat. Mus. 172: 11 p. Beck, E., M. Senser, R. Scheibe, H.M. Steeiwer & 2. IPOMmerAty7 (1982): Frost avoidance and freezing tolerance in Afroalpine “giant rosette” plants. — Plant Cell & Environm. 5: 215-222. Bonnefille,R.,J.C. Roeland & J. Guiot (1990): Temperature and rainfall estimates for the past 40,000 years in equatorial Africa. — Nature 346: 347-349. Bussmann, R.W. & E. Beck (1996): The forests of Mount Kenya — A phytosociological synopsis. — Phytocoenologia 25: 467-560. Demaret, F. (1958): Apercu sur la flore et la végétation des foréts a Hagenia abyssinica (Bruce) Gmel. du Ruwenzori occidental. — Bull. Jard. Bot. Brux. 28: SS 337: Fischer, E. (1996): Die Vegetation des Parc National Kahuzi-Biéga/Zaire. — Erdwiss. Forschung 35: 1-240. Becher col. Hinkel (1990); Untersuchungen zur Einnischung der Herpetofauna in Nebelwäldern Ruandas (Zentralafrika). — Mainzer Geowiss. Studien 34: 485-502. Fischer, E. & H. Hinkel (1992): Natur Ruandas, La nature au Rwanda. Einführung in die Flora und Fauna Ruandas, Apercu sur la flore et la faune rwandaise. Materialien zur Partnerschaft Rheinland-Pfalz/Ruanda, 1992/1. — Rhein-Main-Verlagsgruppe, Mainz. Fischer, E. & H. Hinkel (1993): Zur Flora und Fauna der Walder Ruandas/ Zentralafrika. I. Der Cyamudongo-Wald. — Mainzer Naturwiss. Archiv 31: 239- 213: 130 Fischer, E. & H. Hinkel (1994): Zur Flora und Fauna der Walder Ruandas/ Zentralafrika. II. Der Gishwati-Wald. — Mainzer Naturwiss. Archiv 32: 39-85. Fries, R.E. & T.C.E. Fries (1922a): Uber die Riesen-Senecionen der afrikanischen Hochgebirge. — Svensk. Bot. Tidskr. 16: 321-340. Fries,R.E. & T.C.E. Fries (1922b): Die Riesen-Lobelien Afrikas. — Svensk. Bot. Tidskr. 16: 383-416. Grey-Wilson, C. (1980): Impatiens of Africa. - A.A. Balkema, Rotterdam. Grunderbeek, M.-C. van, E. Roche & H. Doutre Kepone (OS premier Age du Fer au Rwanda et au Burundi, archéologie et environnement. — I.N.R.S. Butare — Rwanda, Publication 23: 1-57. Habyaremye,F.X. (1993): Analyse Phytosociologique des Foréts Primaires de la Créte Zaire-Nil au Rwanda. - Belg. Journ. Bot. 126: 100-134. Hamilton, A.C. (1974): Distribution patterns of forest trees in Uganda and their historical significance. — Vegetatio 29: 21-35. Hamilton, A.C. (1982): Environmental history of East Africa. — Academic Press. Hauman, L. (1933): Esquisse de la vegetation des hautes altitudes sur le Ruwenzori. — Bull. Acad. roy. Belg. (Classe des Sciences), Je série 19: 602-616, 702-717, 900-917. Hauman, L. (1955): La “Région afroalpine” en phytogéographie centro- africaine. -— Webbia 11: 467-469. Hedberg, O. (1951): Vegetation belts of the East African mountains. — Svensk Bot. Tidskr. 45: 140-202. Hedberg, O. (1957): Afroalpine vascular plants. A taxonomic revision. Symbolae Bot. Upsalienses 15, 1: 1-411. Hedberg, ©. (196la): The phytogeographical position of the afroalpine flora. — Recent Advances Bot. 1: 914-919. Hedberg, O. (1961b): Monograph of the genus Canarina L. (Campanulaceae). — Svensk Bot. Tidskr. 55: 17-62. Hedberg, ©. (1965): Afroalpine flora elements. — Webbia 19: 519-529. Hedberg, O. (1969): Evolution and speciation in a tropical high mountain flora. — Biol. J. Linn. Soc. 1: 135-149. Hedberg, O. (1986): Origins of the afroalpine flora. — pp. 444-468 in: High altitude tropical biogeography (eds. Vuilleumier, F. & M. Monasterio.) — Oxford University Press. Hendrickx, F.L. (1946): Esquisse de la végétation du Kahusi. — Rev. Agron. col., Bukavu, 3e trim. 1946, 25-32. Humbert, H. (1934): Sur un Senecio arborescent nouveau des hautes montagnes du Congo belge et sur les liens phylogénétiques des especes alliées. — Bull. Soc. Bot. 81: 830-848. Klötzlı, F. (1958): Zur Pflanzensoziologie des Südhanges der alpinen Stufe des Kilimandscharo. — Ber. Geobot. Inst. Rübel 1957: 33-59. 131 Knox, E. (1994): Species concepts in the giant Lobelias and the giant Senecios. — Proc. XIIlth Plenary Meeting AETFAT, Malawi 1: 467-471. Lösch, R. & E. Fischer (1994): Vikariierende Heidebuschwälder und ihre Kontaktgesellschaften in Makaronesien und Zentralafrika. — Phytocoenologia 24: 695-720. Mabberley, D.J. (1974): Evolution of the giant groundsels. — Kew Bull. 28 (1): 61-96. Mabberley, D.J. (1986): Adaptive syndromes of the afroalpine species of Dendrosenecio. — pp. 81-102 in: High altitude tropical biogeography (eds. Vuilleumier, F. & M. Monasterio.) — Oxford University Press. Mildbraed, J. (1922): Zur Kenntnis der Senecio-Bäume der afrikanischen Hochgebirge. — Feddes Repert. 18: 227-232. meom@cmnien Fa Beck & JO. Kokwaro (1988): The afroalpine plant communities of Mt Kenya (Kenya). — Phytocoenologia 16: 433-463. Scaetta, H. (1934): Le climat écologique de la dorsale Congo-Nil (Afrique centrale équato-riale). — Inst. Roy. Col. Belge, Sect. Sc. Nat. Méd., Mém. 4°, 3. B ruxelles. Schmitt, K. (1991): The Vegetation of the Aberdare National Park, Kenya. — Hochgebirgsforschung 8: 1-259. Innsbruck. Schmitt, K. & E. Beck (1992) On the afroalpine vegetation of the Ruwenzori Mountains, Uganda. — Phytocoenologia 21: 313-332. Snowden, J.D. (1933): A study in altitudinal zonation in South Kigezi and on Mounts Muhavura and Mgahinga. — Uganda. J. Ecol. 21: 7-27. Thulin, M. (1984): Lobeliaceae. — pp. 1-59 in: Flora of Tropical East Africa (ed. Polhill, R.M.). — A.A. Balkema, Rotterdam. White, F. (1978): The Afromontane Region. — pp. 463-513 in: Biogeography and ecology of Southern Africa (ed. Werger, M.J.A.). — Junk, The Hague. White, F. (1983): The vegetation of Africa. A descriptive memoir to accompany the UNESCO/AETFAT/UNSO vegetation map of Africa. — Paris. Author’s address: Prof. Dr. Eberhard Fischer, Institut für Biologie, Universität Koblenz-Landau, Rheinau 1, D-56075 Koblenz, Germany Systematics and distribution patterns of Afrotropical Canaries (Serinus species group, Aves, Passeriformes, Carduelidae) Renate van den Elzen Abstract:A brief comparison of results obtained from a morphological and a molecular approach to canary phylogeny is given. Based on the molecular hypothesis, distribution patterns of species under study in both analyses are reevaluated. In molecular trees, nodes of high bootstrap values in stable clusters support monophyletic species groups also found in the morphological dataset. The respective species show mostly allopatric distribution, similar gross habitat demands and are of younger phylogenetic age than species clustered at deep nodes and/or with low bootstrap values. Distribution patterns of these species are either disjunct or sympatric, and gross habitat demands differ. Key words:systematics, distribution patterns, Carduelidae, Afrotropis Introduction Carduelidae is a family of songbirds spread almost worldwide. It comprises about 140 species, of which 32 (Paynter 1968) to 45 taxa (Sibley & Monroe 1990) are placed within the genus Serinus Koch, 1816. Distribution of these species is focussed, with 38 members, in the Afrotropis. Two species of the Afrotropical stock also inhabit the Arabian Peninsula, five occur in the Palaearctic and two in the Oriental Regions. In two recent papers, rapid radiation (Arnaiz-Villena et al. 1999) and systematics and distribution patterns (van den Elzen & Khoury 1999) of this species group were analyzed, using two different methods, a molecular (Arnaiz-Villena et al. 1999) and a combined morphological-ethological approach (van den Elzen & Khoury 1999). As the two studies correspond in some of their results and complement each other, both analyses are summarized here, a short comparison is given and, based on the molecular hypothesis of canary phylogeny, distribution patterns of species under study in both analyses are reevaluated Phylogenetic analyses Results of the morphological approach In the systematical analysis, 5 ethological and 6 morphological characters found as synapomorphies in 23 extant canary species were polarized from two outgroups Fringilla coelebs and Chloris chloris and P. striolata and C. thibetana (see table |, details in van den Elzen & Khoury 1999). A first-hand cladogram of all 25 species obtained 5 species groups (see also van den Elzen 1985), later given generic rank,mainly following the nomenclature of Wolters (1979): Serinus (sensu stricto, 5 species: alario, citrinella, canicollis, syriacus, pusillus within this study), Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 134 Table 1: Behaviour and plumage patterns of Serinus species used for cladistic analysis (nomenclature follows Wolters 1979). Fringilla coelebs /juv % 0 1 0 0 0 1 Chloris chloris 0 0 0 0 0 0 | | 0 2 2 N | = —) S oj~o;]o Alario alario Serinus citrinella l i l 1 canicollis | Serinus Serinus Syriacus Serinus pusillus Serinus Crithagra leucoptera Crithagra striolata Crithagra gularis tristriata Crithagra Poliospiza Dendrospiza citrinelloides Dendrospiza hyposticta Dendrospiza scotops Dendrospiza Serinops flaviventris | Crithagra sulphurata Crithagra donaldsoni Crithagra albogularis Ochrospiza leucopygia — — Ochrospiza reichenowi — Ochrospiza atrogularis Ochrospiza citrinipectus Ochrospiza mozambica Ochrospiza dorsostriata Ochrospiza xanthopygia Ochrospiza Characters 01-11 Ol attraction call of juveniles: 0= identical to begging calls, short, no frequency modulations; 1= frequency modulated, of long duration 02 Head-up posture (aggressive display): 0= wings spread; |= head-up posture 03 courtship display: O= lateral displaying, pivoting; 1= fluffed display, 2= head-up posture, 3= heraldic-eagle display 04 nestbuilding: 0= by ? only, l= also by & 05 transport of nesting materials: 0= by ? only, l= also by & 06 rump coloration: O= as colour of back; 1= contrasting 07 ear patch: 0= weak; 1= prominent, contrasting; 2= non-existent 08 moustachial stripe: O= weak; 1= dark and contrasting; 2= non-existent 09 head pattern: 0= uniform, no supercilium; 1= yellow supercilium; 2= s. black and yellow; 3= s. white 10 juvenile plumage: 0= as adult pl.; 1= different to ad. pl. 11 gape coloration of nestlings: O= red; |= blue patches on red 0 = plesiomorphic condition, 1(2,3 in transformation series) = apomorphic conditions, ? = no information 139 Ochrospiza (7 species: leucopygia, reichenowi, atrogularis, citrinipectus, mozambica, dorsostriata, xanthopygia), Crithagra (4 species: flaviventris, sulphurata, donaldsoni, albogularis), Poliospiza (4 species: leucoptera, striolata, gularis, tristriata) and Dendrospiza (3 species: citrinelloides, hyposticta, scotops). As relationships between genera were still unsolved, the dataset was analyzed with the computer programme HENNIG 86. Six trees were found but are not taken into consideration, as they have only poor solutions and low tree stability with a consistency index of 51. Tree stability in computer-based phylogenetic analyses needs a support of 5 characters per node (Mooers et al. 1995), not given in the dataset of van den Elzen & Khoury (1999) as species outnumbered characters by 25 to 11. Lacking additional clear synapomorphies, the number of taxa was reduced by summarizing characters of species groups supported in the first analysis. Thus only 7 taxa represented 5 canary species groups (genera Serinus, Ochrospiza, Crithagra, Poliospiza and Dendrospiza) plus the two outgroups (table |, generic names and details in: van den Elzen & Khoury 1999). Despite the fact that 11 characters had to support 6 nodes, the HENNIG 86 analyses provided one single tree of rather good quality (21 steps, consistency index 80, retention index 80). It places (fig.1) the Serinus species next to the outgroups Fringilla and Chloris at the base of the tree. Thus Serinus is the (phylogenetically older) sister taxon of a monophyletic entity comprising four taxa in two sister groups: Ochrospiza plus Crithagra on the one hand, Poliospiza with Dendrospiza on the other. Fringilla Chloris Serinus Crithagra Ochrospiza Poliospiza Dendrospiza Fig.l: HENNIG 86-tree from behavioural and morphological characters of species groups mediated under generic names given in table 1. 136 Results of the molecular analysis Arnaiz-Villena et al. (1999) studied 18 (17, if S. canicollis flavivertex is treated as a subspecies) “Serinus”’ species using mitochondrial DNA of the cytochrome b gene. Their sample comprises, in the nomenclature put forward in van den Elzen & Khoury (1999), the following species: Serinus alario, S. canaria (wild birds and domesticated form), S. serinus, S. canicollis (in two populations S. c. canicollis and S. c. flavivertex); Ochrospiza atrogularis, O. leucopygia, O. citrinipectus, O. dorsostriata, O. mozambica; Crithagra flaviventris, C. sulphurata, C. albogularis; Poliospiza gularis, P. striolata; Chionomitris thibetana. As outgroups, DNA sequences of Fringilla coelebs and Passer luteus were compared, as well as two non-passerine galliform species: Lophura nycthemera and Gallus gallus. The authors present three dendrograms obtained by three different mathematical procedures (algorithms): unweighted maximum parsimony heuristic search (PAUP), neighbour-joining bootstrap and UPGMA. In their parsimony and neighbour-joining trees Arnaiz-Villena et al. (1999) show bootstrap values above 50, found in 15 of 20 nodes in the parsimony analysis and 14 of 20 nodes in the neighbour-joining tree. High bootstrap (80 and above) supports nodes and thus phylogenetic relationships of the following species and species-groups in both dendrograms: Fringilla as an outgroup to the Carduelidae under study (fig.2: node A); canaria and serinus (fig.2:cluster D, node 4), alario and canicollis (fig.2: cluster D, node 3), flaviventris and sulphurata (fig.2: cluster C, node 2), atrogularis and leucopygia (fig.2: cluster C, node 1), as sister taxa. Neighbour-joining defended a clade alario-canicollis-pusillus (fig.2: D, 3a) and another consisting of atrogularis-leucopygia-citrinipectus-dorsostriata (Fig.2: C, lb). Not shown in fig.2 are the relationships of domesticated and wild Serinus canaria (bootstrap 100) and of Serinus canicollis flavivertex to nominate canicollis (bootstrap 98). Within all three dendrograms three species-groups were placed concordantly. — One node (fig.2, D) postulates phylogenetic relationships of species within two subclusters: canaria-serinus and alario-canicollis-pusillus, and thus supports the morphological-ethological findings and species limits of the restricted genus Serinus proposed by van den Elzen & Khoury (1999). — A second node (fig.2, C, b) places atrogularis-leucopygia-citrinipectus- dorsostriata into one cluster that includes not only mozambica but also citrinelloides, from a different monophyletic unit, the Dendrospiza-species group, as depicted in two dendrograms, parsimony and neighbour-joining, but not the UPGMA (see above). A monophylogenetic relationship of only 5 of these 6 taxa (as members of the genus Ochrospiza) was suggested by van den Elzen & Khoury (1999), but the two genera Ochrospiza and Dendrospiza proved not to be closely related (fig.1).Thus the hypothesis that Ochrospiza species are a monophyletic entity is only partly supported. — A third invariable node (fig.2: C, c) groups four species: flaviventris, sulphurata, albogularis, gularis. It also does not support a closer relationship of Crithagra species within the limits of the morphological analysis as it places gularis into a distinct species group, genus Poliospiza. The three tree topologies of the DNA 137 analysis also do not sustain a broad genus Crithagra (sensu Wolters 1979, lumping Poliospiza and Crithagra species), as they place P. striolata either next to C. thibetana (fig.2, B, a), and their common node (fig.2, B) in a sister group position to the Serinus s.str. node (parsimony tree) or to the node of the Afrotropical clade (neighbour-joining tree, fig.2), or both taxa separately in an outgroup position to the first node of the other 15 species of the whole sample (UPGMA tree). Of the five lineages proposed by van den Elzen & Khoury (1999), none is unquestionably verified by the mtDNA study. Serinus and Ochrospiza are partly supported, whereas Poliospiza appears to be paraphyletic and the species limits of Crithagra may have to be reevaluated. The choice of only one single representative of the Dendrospiza species group, D. citrinelloides, may have disturbed the tree topology as indicated by the low bootstrap values of 53 and below. Phylogenetic relationships between the three monophyletic entities described above could not be proved by the molecular analysis. Serinus s.str. appears in the three trees figured by Arnaiz-Villena et al. (1999) as the sister taxon to either the Afrotropical clade of van den Elzen & Khoury (1999) (genera Ochrospiza, Crithagra, Poliospiza, Dendrospiza; neighbour-joining tree), the P. striolata-C. thibetana node (parsimony tree), or the flaviventris-sulphurata-albogularis-gularis node (UPGMA tree). In the latter, the canaria-serinus node is joined by mozambica and citrinelloides as sister taxa, with the cluster alario-canicollis-pusillus in a stable sister-group position. The parsimony tree depicts a common node of Serinus s.str. (including P. striolata and C. thibetana with bootstrap values below 50) with Ochrospiza (including mozambica), and with “Crithagra” as their sister taxon. The neighbour- joining tree sketches a Serinus s.str. cluster opposing a cluster formed by three subgroups, one including the complete sample of Ochrospiza species plus mozambica, another the “Crithagra” subgroup with the already mentioned restriction of P. striolata, which forms, combined with C. thibetana, the third subset. In the UPGMA dendrogram, Serinus species are disrupted by O. mozambica and D. citrinelloides. Their sister taxon is the mixed Crithagra-Poliospiza gularis group, and both clusters have Ochrospiza species (solely O. mozambica) as a sister taxon, opposed by P. striolata and C. thibetana. The UPGMA tree in Arnaiz-Villena et al. (1999) also gives an approximate time calculation calibrated on the known divergence time of Lophura nycthemera and Gallus gallus, 19 Mya. Based on this assumption, C. thibetana appears as the phylogenetically most distant and thus oldest species in the sample, diverging about 9 Mya, Ochrospiza species are relatively older than the other Afrotropical species under discussion, while the true canaries Serinus s.str. are the youngest clade. Analyses of distribution patterns According to the vicariance model of Nelson & Platnick (1981), phylogeny can be inferred from distribution patterns. Speciation is assumed to proceed in geographical isolation, a partial requirement also of the Biological Species Concept (Mayr 1942, 1992). Vicariants are allopatric in their distribution and have similar ecological demands. Therefore species ranges of sister taxa, or of a proposed 138 monophyletic species group, should be allopatric in early stages of differentiation. It can also be expected that phylogenetically younger species have roughly similar ecological demands and are to be found in more similar habitats than phylo- genetically older species but in different niches, as postulated in the extended species concept (Mayr 1992). Distribution patterns were presented by van den Elzen & Khoury (1999) on counts of species Occurrence in 37 x 5~ erids. Ihe main) Souncesmwmene mune distribution maps in Hall & Moreau (1970), supplemented by individual maps in Ash & Miskell (1998), Clemetn et al. (1993), Gatter (1997), Harrison et al. (1997), Lewis & Pomeroy (1989) and Irwin (1964). Centres of diversity of Afrotropical “Serinus” s.lat: (38 species in Sibley & Monroe 1990) are montane habitats of the Albertine Rift Mountains where 12 species occur. Three minor centres with 8 species each are in the Angolan, east Zimbabwean and Eastern Arc Mountains. West Arica is poorest, with a species maximum of 3 and an average of 1.7 within 5° grids, montane areas in East Africa, with an average of 7.4 species per square, are richest in canary species abundance. Counted by altitudinal distribution, lowland species outnumber montane taxa: 21 versus 15. Two species (S. canicollis and Dendrospiza scotops) are occurring in both habitats (details in van den Elzen & Khoury 1999). The four species groups which were separated as the genera Crithagra, Ochrospiza, Poliospiza and Dendrospiza from Serinus s.lat. are genuine Afrotropical inhabitants. Most species of the genus Serinus s.str. breed in restricted ranges around the Mediterranean basin, Serinus pusillus occurs from Turkey into the Himalayas, whereas Serinus canicollis and Serinus (Alario) alario inhabit the Afrotropical region. The two species used as outgroups in the HENNIG 86 analysis, Fringilla coelebs and Chloris chloris, are distributed in the Palaearctic region. Fringilla reaches the African continent and has breeding populations in the Canary Islands and parts of North Africa. Chloris chloris has its nearest relatives in the Himalayas. Within the Afrotropical endemic genera, species ranges were found to be allopatric to parapatric in most cases or, if sympatric, species are separated by diverging ecological requirements (details in van den Elzen & Khoury 1999). Centres of species diversity for Poliospiza and Dendrospiza are montane in the northeastern Rift Valley, for Crithagra montane to lowland areas in southern Africa, and for Ochrospiza lowlands in northeastern Africa. A topographical divergence of lineages, that is adaptation of species of any of the proposed genera to either montane or lowland habitats, could not be confirmed. As each has representatives in both environments, genera can only be characterized by their distribution centres. As indicated by the ranges of different Serinus canicollis- populations — reaching from high altitude montane areas in East Africa to man- made environments in southern Africa — landscape requirements are seemingly not key factors for habitat choice of the species. As large-scale investigations have shown, distribution patterns are known to correspond best to climatic factors (O’Brien 1998). Van den Elzen & Khoury (1999) explained canary distribution ranges and patterns by climate and climatic changes in the past leading to habitat fluctuations and shifts. The geographic posi- 139 eAUI+¢ ase onouasojAyd sa eR RG Kc KC a en 2 = n a G © . P>UISIUOI EXE] suoneJ2I exe] jexeyu Sessel anne ae ee ee a ere t= | ees ee ‘QpOU JoIoJJIp WOAF pospnl ase sNouasojAyd (+9) ‘A-V JO 90 UONL[NI[eS sum ay) ur pajuasaid jou spou ="d'u ‘(666] AINOYY 2 uazjq Up UBA JO SHUM) SNUIS Juo Jo Satdods JUDIAJJIP =SNuIs ‘V1IDUIS JUIIIJJIP 0) =P1OUIS ‘Sar[IUUe] =ATIUIL] :Ju9I2JJIp 0) SUOTAG P212ISN[ exe) =Paul29uU09 exe ‘dno1s Ja\sts Jad exe] Jo JoquinU =SUoNeJ2! exe) !opou Aq pataysnyjo exe} Jo Jaquinu =exeju ‘(666]) AINOYY 2 UaZg Uap ueA Aq PoULITJUOS OsTe sapou=soUapTIOUTOS Ma :uonnginsıp SJıyeduAs = duis ‘satsads Jo uonngrnsıp o1nedeied ='desed ‘apou ayy Aq paasısnj9 saroeds Jo uonngınsıp omedoyye =dorye :05 ueyJ Jomo] densjoog ‘umoys Jou ='s'N "uonersıdızyur May) pur (GT “SY :6661) Te 19 BUSTIIA-zZleuTY Jo 901 Zururol-ınoqygıpu ay) WON SOPON :T OIQEL 140 F. coelebs W PalWcmed C. thibetana M PalE P. striolata M AfrE O. mozambica L AfrWES a] B b D. citrinelloides M AfrE Ic O. dorsostriata L AfrE 80 O. citrinipectus L AfrS € sel: O. leucopygia L AfrWNE O. atrogularis L AfrSE C. albogularis L AfrS 100] A 63 © : P. gularis L AfrWES 50 C. sulphurata L AfrSE DD C. flaviventris L AfrS 89 S. pusillus M PalEemed S. canicollis LM AfrES 221 26 S. alario L AfrS S. serinus L PalWcmed sl Ä S. canaria L PalWmed Fig.2: Neighbour-joining bootstrap tree from Arnaiz-Villena et al. (1999, fig. 1b), slightly modified. Neighbour-joining (1000 replications) based on 924 bases of mtcyt b genes from 17 canary species and Fringilla coelebs. Bootstrap values above 50 are shown. For details see text. tion of the areas of canary diversity coincides with endemic bird areas already outlined for Africa by Bibby et al. (1992) and Stattersfield et al. (1998) and with areas of primary montane and temperate floristic elements (Miehe & Miehe 1994, Wickens 1976). Ecological requirements in terms of food plant species were not discussed but may count as a limiting factor as most of the canary species are pure seedeaters and thus dependent on food plants, but the requirements of only a few species are as yet documented, including generalists (Serinus syriacus, Khoury 1998a,b) as well as specialists (Poliospiza leucoptera, Milewsky 1976, 1978). Provided with a new hypothesis of “Serinus” phylogeny, the neighbour-joining tree of Arnaiz-Villena et al. (1999) was chosen to test topographical and geographical distributions of canary lineages under the vicariance model again. The neighbour-joining tree was favoured because of its convincingly high bootstrap values. Moreover it depicts the phylogeny of genera suggested by the analysis of behavioural and morphological characters carried out by van den Elzen & Khoury (1999). It is presented in fig.2 with slight differences from the original: the two non-passerine species are omitted and nodes for nominate Serinus c. canicollis - S. canicollis flavivertex and of Serinus canaria for wild and domestic forms suppressed. Under the premises of the vicariance model, nearest neighbours showing up in the dendrogram should verify the following characteristics: allopatric ranges, similarity in their topographic distribution and broad habitat requirements if they are young nodes. Deep nodes in contrast are allowed to display either sympatry, broad habitat differentiation, or lowland-montane divergence, or even show 14] differentiation in all premises within but not between subclusters. Results of an analysis of the molecular tree are summarized in table 2, and details are also shown in fig.2. For the first-species pair supported by high bootstrap ancestral nodes, Serinus canaria-S. serinus, (fig.2: 4) ranges are allopatric with a small parapatric zone and their ecological requirements are roughly similar, with S. serinus having a larger range in more diverse habitat types and at lower elevations. In the S. alario- canicollis-pusillus clade (fig.2: node 3a) distribution patterns are puzzling. The allopatry argument is proven, two montane species combined with a lowland form, two Afrotropical canaries combined with a Palaearctic species. Serinus canicollis and alario overlap in distribution in the southernmost parts of their range, where they even associate in mixed-species flocks. If the distribution patterns of the entire Serinus clade (fig.2: node D) are investigated, no range overlap between the species can be found. The second well-established pair by bootstrap is Crithagra flaviventris- sulphurata (fig.2: node 2). Species ranges are allo- to parapatric, overlap occurs in the southernmost part of range, but breeding habitats differ distinctly in the sympatric situation, C. sulphurata preferring in general moister habitats at higher altitudes than C. flaviventris. The assemblage of C. sulphurata-flaviventris plus P. gularis (fig.2: node 2a) together with C. albogularis (fig.2: node c) sketches a situation not as easily interpreted. P. gularis and C. albogularis both share the flaviventris habitat demands, and the ranges of the congeners albogularis and flaviventris overlap. In the third high bootstrap node (O. leucopygia-O.atrogularis, fig.2: 1) species with allopatric distribution are again combined: both are savanna inhabitants, O. leucopygia living in the drier environment. The taxa adjoined to this species pair in node Ib and la by also high bootstrap, O. citrinipectus and dorsostriata, are allopatric and have roughly equal ecological requirements. If mozambica and D. citrinelloides, their nodes (fig.2: lc, b) are of low significance, are joined, the latter, as a montane element, is allopatric to all other taxa and prefers woodland. O. mozambica — as the most widespread species — occurs in sympatry with atrogularis, leucopygia and citrinipectus (not with dorsostriata!), but is found in moister environments. As this species is allopatric to dorsostriata, Hall & Moreau (1970) placed both into one superspecies, the atrogularis-group comprising the species atrogularis (including reichenowi and xanthopygia), leucopygia, citrinipectus, as well as flavigula into a second one. The last node (fig.2: a) supports faintly (since it is only of low significance, its bootstrap values are below 50 and are not given by Arnaiz-Villena et al. 1999) the relationship of C. thibetana and P. striolata. Both species are disjunct, occurring in high altitudes of different biogeographical regions. Summarizing, it may be concluded that best results (i.e. high bootstrap values, tree stability, concordance with the vicariance model and systematics deduced from the morphological approach) are in the younger nodes, with the exception of node A. The suggested phylogenetic relationships of species in clusters with nodes of low bootstrap values, with inconsistent arrangements according to the algorithms used, are in most cases also contradicted by inconsistent distribution patterns and 142 broad habitat requirements of the species concerned. Deep nodes (B, C and partly a) are seemingly more difficult to prove than younger ones (from 6 Mya on). Following the HENNIG 86 tree-topology of genera (van den Elzen & Khoury 1999), Serinus s.str. had to be the phylogenetically older clade. The ancestor of this taxon entered Africa south of the Sahara, where radiation of Ochrospiza, Poliospiza, Dendrospiza and Crithagra was favoured by mainly Pleistocene climatic changes, whereas Serinus s.str.-species (e.g. Serinus canicollis) entered the Afrotropics in a second immigration wave (see also van den Elzen 1985). From their calculations of the phylogenetic age of different canary “lineages”, Arnaiz-Villena et al. (1999) concluded that the “Serinus”-lineage s.lat. appeared in the Miocene and that Pleistocene glaciations may have driven subspeciation only, the majority of genera in the nomenclature of van den Elzen & Khoury (1999) being already present in their ancestral forms 6-7 Mya. These authors discuss an Asian origin for “Serinus”, an old hypothesis favoured by Mayr (1946) and Marten & Johnson (1986) for other Carduelidae. Research on this topic will have to continue. References Arnaiz-Villena, A., M. Alvarez-Tejedo, V. Ruiz-dei_ Da C.Garcia-de-la-Torre, P. Varela: M.J. Recto. S; Kenner Martinez-Laso (1999): Rapid radiation of Canaries (Genus Serinus). — Mol. Biol" Evol. 1022 17. Ash, J.S. & J.E. Miskell (1998): Birds of Somalia. — Pica Press, Sussex. Bibby, €., N.J. Collar, M.J. Crosby, M.E. Heath: (© Irmpordene T.H. Johnson, A.J. Long, A.J.’ Stattersiteld & SI Ekıneood (1992): Putting biodiversity on the map: priority areas for global conservation. — Cambridge, U. K. (International Council for Bird Preservation). Clement, P., A. Harris & J. Davis (1993): Finches and) Spamows= an identification guide. — Christopher Helm & Black, London Elzen, R. van den (1985): Systematics and evolution of African canaries and seedeaters (Aves: Carduelidae). — pp. 435- 451 in: Proc. Intern. Symp. African Vertebr.(ed. Schuchmann, K.-L.).- Bonn. Elzen, R. van den & F. Khoury (1999): Systemauk, phylogenetische Analyse und Biogeographie der Großgattung Serinus Koch, 1816 (Aves, Carduelidae). — Cour. Forsch. Senck. 135: 55-65. Gatter, W. (1997): Birds of Liberia. — Pica Press, Sussex. Hall, B.P. & R.E. Moreau (1970): An atlas of speciation in African passerine birds. — British Museum of Natural History, London. Harrison, J.A., D.G. Allan, L.G. Underhill, MH erzemansı ee Tree, V. Parker & C.J. Brown (1997): The Atlas of Southern African Birds. — 2vols. BirdLife South Africa, Johannesburg. Irwin, M.P.S. (1964): An african canary superspecies. — Occas. Pap. Nat. Mus. S. Rhodesia 4 (27B): 16-25. Khoury, F. (1998a): Habitatwahl und Nahrungsökologie des Zederngirlitzes Serinus syriacus in Jordanien. — Dissertation, Bonn. 143 Khoury, F. (1998b): Habitat selection by Syrian Serin Serinus syriacus in south- west Jordan. — Sandgrouse 20(2): 87-93. Lewis, A. & D. Pomeroy (1989): A Bird Atlas of Kenya. — A.A. Balkema, Rotterdam, Brookfield. Marten,J.& N.K. Johnson (1986): Genetic relationships of North American Cardueline finches. — Condor 88: 409-420. Mayr, E. (1942): Systematics and the Origin of Species. — Columbia University Press, New York. Myr, E. (1946): History of North American bird fauna. — Wilson. Bull. 58: 4-41. Mayr, E. (1992): A local flora and the biological species concept. — Amer. J. Bot. 295292238. Miehe, S. & G. Miehe (1994): Ericaceous forests and heathlands in the Bale Mountains of South Ethiopia, ecology and man’s impact. — Traude Warnke Verlag, Hamburg. Milewsky, A.V. (1976): Feeding ecology and habitat of the Protea Seedeater. — MSc thesis, University of Cape Town, Cape Town. Milewsky, A.V. (1978): Diet of Serinus species in the southwestern Cape, with species reference to the Protea Seedeater. — Ostrich 49: 174-178. amen On. RD.M. Page, A. Purvis & P.H. Harvey (1995): Phylogenetic noise leads to unbalanced cladistic tree reconstruction. — Syst. Biol. 44: 332-342. Nelson, G. & G. Platnick (1981): Systematics and biogeography, cladistics and vicariance. — Columbia University Press, New York. O’Brien, E.M. (1998): Water-energy dynamics, climate, and prediction of woody plant species richness: an interim general model. — J. Biogeogr. 25: 379- 398. Paynter, R.A. (1968): Checklist of Birds of the World 14 (ed. J. L. Peters) . — Museum of Comparative Zoology, Cambridge U.S.A. Sibley,C.G. & B.L. Monroe (1990): Distribution and taxonomy of the birds of the world. — Yale University Press, New Haven U.S.A. Based N. 1, MJ: Crosby, A.J. Long &D.E. Wege (1998): Endemic Bird Areas of the world. — BirdLife International, Cambridge Wickens, G.E. (1976): The Flora of Jebel Marra (Sudan Republic) and its geographical affinities. - Kew Bull. Add. Ser. V: 1-368. Wieltens Heb (979): Die Vogelarten der Erde.- Paul Parey, Hamburg, Berlin. Author’s address: Dr. Renate van den Elzen, Alexander Koenig Research Institute and Museum of Zoology, Adenauerallee 160, 53113 Bonn, Germany; e-mail: r.elzen.zfmk @uni- bonn.de Bed ne x 7 “2 REN ane HAN st tor Patchy versus continuous distribution patterns in the African rain forest: the problem of the Anomaluridae (Mammalia: Rodentia) Mopase= Sichiunikers Rainen Hutterer Abstract: We investigated the distribution patterns of Anomalures (scaly-tailed squirrels), a family of gliding rodents confined to African old-growth forests. The evolution of these spezialized mammals might be connected to the evolution of these forests. Current knowledge, however, does not allow conclusive confirmation of this assumption. Our analysis of both literature and specimen-based locality data has revealed two conflicting views of the distribution patterns of these animals. In most current textbooks the existence of broad continuous ranges throughout the African forest belt is postulated. By mapping actual locality data we found that patchy discontinuous distributions are equally supported. The contradictions are discussed for the respective countries in detail. Key words: Anomaluridae, Rodentia, biogeography, African rain forest Introduction Anomalurids (or scaly-tailed squirrels) are highly specialized gliding rodents confined to African old-growth forests. Their existence depends on factors such as specific tree height, distance between trees and sufficient suitable tree holes. The evolution of these gliding mammals should thus be closely tied to the evolution of the African rain forest. Current knowledge, however, does not allow conclusive confirmation of this assumption. Material and methods For the analysis of the postulated distribution of Anomaluridae maps were taken from current textbooks (see below), enlarged to the same scale and the respective areas combined on a single map. The outlines in figs.1-3 give the maximum distribution assumed. Documented specimen-based localities were taken from various authors (see below), either redrawn from maps or from localities with given coordinates. Localities less than ca. 100 km apart were regarded as representing a continuous area and thus combined. Distribution patterns Our analysis of both literature and specimen-based locality data has revealed two conflicting views of the distribution patterns of these animals. In most current textbooks (McLaughlin 1984, Rahm 1988, Kingdon 1997) the existence of broad continuous distributions throughout the African forest belt is postulated (figs.1-3). Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 ANS | | | | Nu / ) Y A EN zz % Ben ING: SEO». | AZ on f f Ae ERS x i > — ER é bw pam 4 a \ ER Fig.l: Presumed continuous distribution of the Anomalurinae as postulated by various authors (hatched area) contrasted with the documented localities (solid black areas); see text for references. Scale = 1000 km. Fig.2: Presumed continuous distribution of /diurus as postulated by various authors (hatched area) contrasted with the documented localities (solid black areas); see text for references. Scale = 1000 km. 147 By mapping actual locality data from the literature (Sanderson 1940, Rosevear 1953, Rahm 1960, Eisentraut 1963, Rahm & Christiaensen 1963, Verheyen 1963, Kuhn 1965, 1966, Rahm 1966, Rahm & Christiaensen 1966, Verheyen 1968, Jones 1971, Roche 1972, Kingdon 1974, Delany 1975, Ansell 1978, Happold 1987, Ansell & Dowsett 1988, Pérez del Val et al. 1995, Robbins & Van der Straeten 1996) and from voucher specimens we found that patchy discontinuous distributions are equally supported (figs.1-3). Our preliminary results thus suggest that the patterns of distribution in anomalurids are more complex. The discrepancy between assumed and real data differs among countries, therefore some areas will be discussed in detail in the following sections. < m \ | ) Ga 35 / \A Bar S J \ ra Ne a real i 3 A \ er “eu ny = / aX x Y if J a Sy y\ N] ——— ( OD JS ye —> ri ( T \ Ss ( R Be > Na h [8 _—| JENS f ( N re ER 3 Q S { f N A \ 5 1} PR ) r | \ 7 I A ‘ sree ı | wt) / < Ly ar 3 | | = De J ~ Soe ( SEN \ , } \ \ / \ ) \ SS ene x 2 Yi lS —— \ 2 / SL a] / Se: Game ER We N I a, l N N) y) Y lace ) F fr A, ) f aw | 4 ( he { nd a IS V Se Wa > Fe N uns SL Z i f / 9 Fig.3: Presumed continuous distribution of Zenkerella as postulated by various authors (hatched area) contrasted with the documented localities (solid black areas); see text for references. Scale = 1000 km. 1. Senegal to Togo In most current textbooks a broad belt of continuous distribution over West Africa is assumed, with a northwestern extension of Anomalurinae into The Gambia and Senegal, and an extension of /diurus to Liberia. Contrary to the postulated distribution, documented localities from the western range of the Anomaluridae are extremely scattered. In most cases only a single animal was recorded, which often became the type specimen of a new taxon. Except for small clusters along the Liberia-Ivory Coast border and in Benin the specimen-based localities are usually isolated, with several hundred km distance between them. From The Gambia and Guinea-Bissau no findings are recorded. Certainly the small number of records is partly due to the lack of expeditions to this area. But apart from this a broad continuous distribution is impossible because of the very scattered occurrence of rain forests. Particularly the postulated 148 distribution for /diurus seems exaggerated, as it is only based on two isolated findings in Sierra Leone and Ghana. 2. Benin Contrary to the postulated distribution of /diurus and the Anomalurinae in this area no findings were recorded in the literature investigated. Analogous to the more western countries this may partly be due to the lack of collections, but in Benin the Savanna extends down to the coastal region and separates the rain forest areas (Dahomey Gap: Grubb 1978, Robbins 1978), so in this country the occurrence of animals dependent on rain forest is unlikely. 3. Nigeria This is the westernmost country with several observations of Anomalurinae, often less than one hundred km apart. Without a thorough investigation of additional material it is impossible to decide whether the gaps between the recordings are due to a lack of information or to the occurrence of an insurmountable obstacle. Despite numerous findings of Anomalurinae no /diurus was recorded, which seems to indicate that the genus does not occur in this area, contrary to the postulated distribution. 4. Cameroon and Equatorial-Guinea From these two countries findings of all groups of Anomaluridae are recorded. The distribution can be divided into several areas. Records of Anomalurinae and Idiurus originate from the Bambouto Mountains and the Cameroon Mountain in western Cameroon. The nearest findings to these are separated by a gap of some two hundred km and lie in southwestern Cameroon and western Equatorial Guinea and include Anomalurinae, /diurus and Zenkerella. The island of Bioko is inhabited by at least two members of the Anomalurinae and Zenkerella, but no Idiurus is documented. From the center and east of southern Cameroon only a few scattered findings, mainly of Zenkerella, are recorded. 5. Gaboon, Congo and Central African Republic In these countries the records are very rare and scattered, with findings of Anomalurinae near the northern coastal area of Gaboon and three widely separated localities for Zenkerella, two in Congo and one in the Central African Republic. Obviously the southern extent of the distribution of /diurus and Zenkerella in Gaboon has been overestimated. 6. Democratic Republic of Congo and Cabinda The records from this area are the most evenly distributed, at least for the Anomalurinae. From Cabinda and the extreme west of the Democratic Republic of Congo, between Popular Republic of Congo and Angola, several localities have been recorded, separated from the rest by a gap of some three hundred km. 149 The distribution of Anomalurinae in the main part of the Democratic Republic of Congo is difficult to interpret, because on the one hand the localities are very evenly distributed and on the other hand they show an average inter-locality distance of 100-150 km. As in the case of Nigeria, a better interpretation of the distribution pattern in the central Democratic Republic of Congo would require more information about these localities. A different situation is found in the northeast of the country, where documented localities of Anomalurinae are usually less than 100 km apart. In this area a continuous distribution is very likely. The /diurus records are more scattered, but an area that supports a continuous distribution of Anomalurinae may also provide suitable habitats for /diurus. However a connection between the /diurus population in the Congo Basin and that in West Africa, as postulated in the literature, is extremely unlikely. Although the region between the two areas was thoroughly investigated, no specimens were recorded; therefore the /diurus population of the Congo Basin must be regarded as geographically separate from the population in Cameroon. 7. Angola From Angola only two isolated findings are reported in the literature studied. Because it is mainly gallery forest that 1s available for anomalurids in this country, localities that do not belong to the same river system must be regarded as separated from one another. 8. Zambia In western Zambia the situation is similar to Nigeria and central Congo: numerous records, some in clusters and obviously belonging to a continuous distribution, gaps of about 150 km between clusters, and also some isolated findings. No records are known from the southern part of Zambia, and only two isolated findings from the eastern part. 9. Rwanda and Burundi From each of these countries a single finding is recorded. The locality in Rwanda is probably connected with the Anomalurinae distribution in Uganda, since the distance to the nearest locality there is just about 100 km. 10. Uganda and Kenya The distribution in Uganda seems to be divided into two parts: the recorded localities near the border to the Democratic Republic of Congo very likely belong to the Anomalurinae area in northeastern Congo. Clearly separated from this area are the localities north of Lake Victoria in Uganda and Kenya. 11. Tanzania and Malawi The Anomalurinae in Tanzania are separated from the populations of Kenya in the north and Zambia in the south respectively by gaps of more than 500 km, so a connection is very unlikely. Only one locality is known from Malawi. 150 Systematics At present the 39 described taxa of the Anomaluridae are grouped into seven species (table 1). Our preliminary data, however, suggest that the taxon diversity has been underestimated. Some taxon names, such as the genus Anomalurops Matschie, 1914, will have to be resurrected. Table 1: Concepts of the number of genera, species and subspecies of Anomaluridae from 1897 to 1997 Author no of genera no of species no of subspecies Trouessart 1897 Allen 1939 Ellerman 1940 Malbrant & Maclatchy 1949 Grassé & Dekeyser 1955 Rosevear 1969 Misonne 1971 Rahm 1988 Dieterlen 1993 | Starck 1995 Kingdon 1997 WHWWWWAKRN HRY Conclusions The level of discrepancy between assumed and documented distributions differs between countries, but generally the presumed area appears to be an overestimate. Particularly the northern distribution of the Anomalurinae and the southern range of Idiurus and Zenkerella are not supported by recorded findings. In addition, well- known rain forest clearings such as the Dahomey Gap in Benin are often ignored. Sometimes all recorded localities of a taxon are presumed to belong to one continuous distribution area, even in cases where no findings were documented for some 1500 km, as in the genus /diurus. The analysis of distribution patterns in this group is also complicated by the lack of information due to the low level of zoological exploration of some areas, particularly in West Africa. For Nigeria, Cameroon and the Congo Basin countries, data from already existing museum specimens may solve some of the problems. Such data may support either separated clusters with clearly defined gaps or continuous distributions. Another source of confusion regarding the distributions of the Anomaluridae is the controversial taxonomy of the group. It is sometimes impossible to tell which species (or even genus) an author refers to in a publication dealing with locality records. Our further research will thus focus on the taxonomy and phylogeny of this interesting group of mammals in order to answer some of the questions encountered, particulary concerning their phylogeography. 151 References Allen, G.M. (1939): A checklist of African Mammals. — Bull. Mus. Comp. Zool. Harvard College 83: 1-763. Ansell, W.F.H. (1978): The Mammals of Zambia. — The National Parks & Wildlife Service, Chilanga, Zambia. Moscow W oh a. ce RJ. Dowsett (1988): Mammals of Malawi. = The Mrendrne Press, Zennor, U.K. Delany, M.J. (1975): The Rodents of Uganda. — Trustees of the British Museum (Natural History), London. Dieterlen, F. (1993): Family Anomaluridae. —pp.757-758 in: Mammal Species of the World. A taxonomic and geographic reference. 2nd edition (eds. Don E. Wilson & DeeAnn M. Reeder). Smithsonian Institution Press, Washington, London. Eisentraut, M. (1963): Die Wirbeltiere des Kamerungebirges unter besonderer Berücksichtigung des Faunenwechsels in den verschiedenen Höhenstufen. — Paul Parey, Hamburg & Berlin. Ellerman, J.R. (1940): The families and genera of living rodents. Volume I: Rodents other than Muridae. — British Museum (Natural History), London. Besserer cc bie Dekeyser (1953). Ordre des Rongeurs. — pp. 1321-1525 in: Traité de Zoologie. Anatomie, Systematique, Biologie. (ed. Pierre P.Grassé). 1972 Paris. Grubb, P. (1978): Patterns of speciation in African mammals. — Bull. Carneg. Mus. nat. Hist. 6: 152-167. Happold, D.C.D. (1987): The mammals of Nigeria. — Clarendon Press, Oxford. Jones, C. (1971): Notes on the anomalurids of Rio Muni and adjacent areas. — J. Mammal. 52: 568-572. Kingdon, J. (1974): East African mammals. An Atlas of Evolution in Africa. Volume II Part B (Hares and Rodents). — Academic Press, London, New York. Kingdon, J. (1997): The Kingdon Field Guide to African Mammals. — Academic Press, London. Kuhn, H.-J. (1965): A provisional check-list of the mammals of Liberia. — Senckenbergiana biol. 46: 321-340. Kuhn, H.-J. (1966): Anomalurus pelii auzembergeri in Liberia. — J. Mammal. 47: 334-338. Malbrant, R. & A. Maclatchy (1949): Faune de l’Equateur Africain Francais. Tome 2. Mammiferes. Encyclopédie biologique 36. Lechevalier, Paris. McLaughlin, C.A. (1984): Protogomorph, Sciuromorph, Castorimorph, Myomorph (Geomyid, Anomalurid, Pedetoid, and Ctenodactyloid) Rodents. — pp. 267-288 in: Orders and families of recent mammals of the world (eds. S. Anderson & J. Knox Jones Jr.). John Wiley & Sons, New York: 267-288. Misonne, X. (1971): Rodentia. — pp. 1-39 in: The Mammals of Africa. An identification Manual (eds. J. Meester & H.W. Setzer). Smithsonian Institution Press, City of Washington. 1592 Perez del Val I. 3. Juste Castrovie, on C995). Aumevile waco Zenkerella insignis Matschie, 1898 (Rodentia, Anomaluridae). First records in Bioko island (Equatorial Guinea). - Mammalia 59: 441-443. Rahm, U. (1960): L’ Anomalurus jacksoni de Winton: Répartition, biologie et observations en captivité. — Bull. Soc. Roy. Zool. Anvers 18: 3-13. Rahm, U. (1966): Les mammiferes de la forét équatoriale de l’Est du Congo. — Ann. Mus. Roy. Afr. Centr. 149: 39-121. Rahm, U. (1988): Dornschwanzhornchenverwandte. — pp. 116-125 in: Grzimeks Enzyklopädie Säugetiere (ed. B. Grzimek). Kindler Verlag, München, Band 3. Rahm, U. & A. Christiaensen (1963): Les mammiferes de la region occidentale du lac Kivu. — Ann. Mus. Roy. Afr. Centr. 118: 1-83. Rahm, U. & A. Christiaensen (1966): Les mammiferes de l’ile Idjwi (Lac Kivu, Congo). — Ann. Mus. Roy. Afr. Centr. 149: 1-35. Robbins, C.B. (1978): The Dahomey Gap — a reevaluation of its significance as a faunal barrier to West African high forest mammals. — Bull. Carneg. Mus. nat. Hist. 6: 168-174. Robbins, C.B. & E. Van der Straeten (1996): Small mammals of Togo and Benin. — Mammalia 60: 231-242. Roche, J. (1972): Capture de Zenkerella insignis (Rongeurs, Anomaluridés) en République Centrafricaine. - Mammalia 36: 305-306. Rosevear, D.R. (1953): Checklist and Atlas of Nigerian Mammals with a foreword on Vegetation. — F. Howard Doulton & Co., London. Rosevear, D.R. (1969): The rodents of West Africa. — British Museum (Natural History), London. Sanderson, I.T. (1940): The mammals of the North Cameroons forest area. — Trans. Zool. Soc. Lond. 24: 623-725. Starck, D. (1995): Lehrbuch der speziellen Zoologie Band 27 Fol Säugetiere. — Gustav Fischer Verlag Jena. Trouessart, E.-L. (1897): Catalogus Mammalium tam viventium quam fossilium. Fasciculus I. - Berolini, R. Friedlander & Sohn. Verheyen, W.N. (1963): Contribution a la systématique du genre /diurus (Rodentia-Anomaluridae). — Rev. Zool. Bot. Afr. 68: 157-197. Verheyen, W.N. (1968): The Anomalurinae of the Congo (Rodentia: Anomaluridae). — Rev. Zool. Bot. Afr. 77: 392-411. Authors’ address: A.C. Schunke, Dr. R. Hutterer, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, D-53113 Bonn, Germany Non-insect Arthropoda (Isopoda, Arachnida and Myriapoda) on the high mountains of tropical Africa Petar Beron Abstract: For the first time an overview is given of the distribution of non-insect arthropods (Isopoda, Arachnida and Myriapoda, or IAM) of the montane zone of tropical Africa: Mt Cameroon, Mt Kenya, Kilimanjaro, Ruwenzori, Mt Elgon, Mt Meru and some Ethiopian mountains including Semien. Although “montane” is usually defined as the area above the forest, here we arbitrarily indicate two lines above which the fauna is analyzed: 2200 m (allowing comparison with the Eurasian oreal) and 3500 m, above which throughout the world lower atmospheric pressure, higher UV radiation and the same ecological factors are present. We present the montane fauna of Isopoda, Arachnida and Myriapoda of the Afrotropical oreal and compare it with the fauna of those zones in the northern regions of the Old World, e.g. Europe, Himalaya and Central Asia. We list in the appendixall species of IAM found in Africa above 3500 m and thus give a complete list of all hypsobionts so far described. In Africa, we found IAM up to 4600 m (Isopoda), 2600 m (Schizomida), 3500 m (Scorpionida), 4100 m (Pseudoscorpionida), 4600 m (Opilionida), 4930 m (Araneida), 4590 m (Acari), 4500 m (Symphyla), 4500 m (Chilopoda), and 4200 m (Diplopoda). The level of endemism is low and, at least concerning the IAM, the Afrotropical montane fauna does not form a separate zoogeographical unit. Comparisons between the IAM of the Afrotropical oreal and the oreals of Europe, Central Asia and the Himalaya present a very diversified pattern. This phenomenon is discussed under evolutionary aspects. Key words: Afrotropical oreal, Isopoda, Arachnida, Diplopoda, Chilopoda, Symphyla, biodiversity Introduction While there are several important studies on the ecology of the insect life at high altitude (especially by Mani 1968, centred on the Himalayan fauna), no attempt is known concerning such important groups as the non-insect arthropods (essentially Isopoda: Oniscidea, Arachnida and Myriapoda). Most of the recent contributions to these groups are based on the rich collections of J. Martens and other zoologists from the Himalaya. But the high mountains of Africa have also been investigated by many expeditions and we can now put together the available information and try to come to some conclusions. For several decades the author has explored high mountains in all continents and so has had the chance to get a personal impression of the high mountain environment, as well as to collect important series of Isopoda, Arachnida and Myriapoda, most of which are still under study. This research also included tropical West, Central and East Africa (the mountains of Cameroon, the Ruwenzori, Kilimanjaro and Mt Elgon). Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 154 Here under “high mountains of East and Central Africa” we understand the mountains north of the Zambezi, east of 30°E and south of Egypt. In this area the highest mountains in Africa are situated: the volcanos Kilimanjaro (5895 m), Mt Kenya (5199 m), Mt Meru (4567 m), Karisimbi (4507 m), Mt Elgon (4322 m), and the massif of Ruwenzori (5109 m). Some other high mountains in this area are: in Tanzania: Uluguru (2653 m), Ngorongoro (Lulmalasin 3648 m, Oldeani 3188 m), Usumbura (2570 m), Pare (2463 m), Kipengere (Mtorwi 2961 m), and Rukwa (Tapepo 2694 m); in Kenya: Aberdare (Satima 3963 m, Kinangop 3960 m); and in Malawi: Sapitva (3000 m). Aberdare is only 100 km distant from Nairobi and is therefore well explored. Mt Elgon has also been visited by collectors of ground fauna. Larger collections of [AM exist only from Kilimanjaro, Mt Kenya, Mt Meru, Mt Elgon, Ruwenzori, and Aberdare (see fig.1). The mountains of Ethiopia are much less explored, but are very interesting from a zoological point of view. The highest tops are situated in the Semien massif (Ras Dashan 4623 m), followed by Mandebo (4310 m in the Bale National Park), Guna (4231 m), Guge (4203 m), Kakka (4193 m), Abune Josef (4193 m), and Mangestu (Tala 4100 m). The Semien mountains were glaciated up to 2600 m. Fako (height 4090 m) is the highest summit of the volcano range on the border between West and Central Africa. The zone with forest or higher scrub vegetation in the mountains ascends according to temperature and precipitation. At an altitude of 2200 m.a.s.l. on every mountain in tropical Africa we find forest or a man-made landscape. But everywhere at 3500 m a.s.l. we find the conditions typical of the montane environment: lower atmospheric pressure, higher UV radiation and a vegetation adapted to these conditions. We present here the Isopoda, Arachnida and Myriapoda' fauna reaching or surpassing 2200 m to have the basis for a comparison with similar montane regions of the Old World. And we present the IAM fauna for 3500 m, the zone where the fauna consists of true hypsobionts. In the following paragraphs we will give some general information on the distribution of IAM in the Old World (Africa plus Eurasia), with special reference to the montane regions of Europe, the Himalaya and Central Asia. Data on IAM in Africa are given here only if this is essential for the general overview. But every zoologist who is not specialized in IAM will have problems understanding the system of these non-insect athropods; this 1s especially true because the endings of classes, orders or suborders are often not self-explanatory (s. table 1). Isopoda. In contrast to the high diversity of montane woodlice in Africa, the number of families, genera and species in the oreals of the rest of the Old World is rather poor. In the oreals of Europe and Asia there are no endemic families. The number of genera is much smaller and none is endemic. Out of 23 species known in the Old World occurring higher than 3500 m, 12 live in East and Central Africa. ') Myriapoda is used here as a superclass, including the classes Symphyla, Pauropoda, Chilopoda and Diplopoda; we know that Myriapoda possibly is not a real taxon, but a polyphylum. , Hoggar N 3003 > S Emi Kussi 3415 = Mara 3088 Ethiopian Highlands cy Je, Suge A 4 Fako 4208 409 ‚Mlande 3000 4 Brandberg 2600 Aa Ntlenyana 3482 Fig.1: Map of Central and South Africa, showing the main elevations. Arachnida. All 8 orders of Arachnida are known from the high mountains of the Old World and most from high altitudes in Africa (except Solpugida). They are also present in the oreal of the Americas. The orders not represented in the montane fauna of Eurasia and Africa (Palpigradi, Amblypygi, Uropygi, Ricinulei, Opilioacarida), are also absent from the high mountains of the western hemisphere, although these orders do occur there at lower altitudes. 156 Table 1: The higher taxa of Isopoda, Arachnida and Myriapoda dealt with in this paper (the groups found in tropical Africa above 2200 m are in bold): Class Crustacea Superclass Myriapoda Order Isopoda Class Symphyla Suborder Oniscidea Class Pauropoda Class Chilopoda Order Lithobiomorpha Order Geophilomorpha Order Scolopendromorpha Order Scutigeromorpha Class Arachnida Order Schizomida Order Scorpionida Order Solpugida Order Araneida Order Pseudoscorpionida Class Diplopoda Order Opilionida Suborder Laniatores Suborder Palpatores Order Acariformes Suborder Acaridida Order Polyxenida Order Glomerida Order Craspedosomatida Order Julida Order Chordeumatida Suborder Prostigmata Order Stemmiulida Suborder Oribatida Order Spirostreptida Order Parasitiformes Order Polydesmida Suborder Gamasida Suborder Ixodida Scorpionida. Five families in the Old World are known to contain species living over 2200 m. Some genera of Buthidae (Uroplectes, Lychas) are known to live up to 3500 m on Mt Meru and at 3000 m on Kilimanjaro. In South America, scorpions live much higher (Orobothriurus crassimanus Maury, family Bothriuridae, up to 5560 m in Peru). Polis (1990) writes: “Such high-elevation species are all small. They feed on a diverse array of arthropods that are also found at these heights (Mani 1968), and their small size may be due to the short period during which they are able to forage... ‘Cold hardiness’ allows at least some species to survive freezing temperatures. Surprisingly, high-altitude scorpions live under rocks, in scrapes, and in relatively short burrows (...), rather than in deep burrows with terminal chambers below the frost line”. Solpugida. At least five of the Old World families contain species living higher than 2200 m. In South America, other families of this order (Ammotrechidae) can reach even 5000 m (Dasycleobis crinitus Mello-Leitao in Argentina). The highest record for the Old World is of Galeodes setulosus Birula from Tajikistan (4000 m). Schizomida. These small arachnids are warm-loving and clearly avoid the high mountain environment. The highest record (2600 m) is from Tanzania (Reddell & Cokendolpher 1995). Pseudoscorpionida. This order shows a high diversity all over the montane regions of the Old World. However table 2 demonstrates that the different families and genera have their typical distribution; none is evenly spread over the four regions. 1577 Table 2: Number of species of different genera and families of Pseudoscorpionida living at or above 2200 m in Europe, Central Asia, Himalaya and tropical Africa Centr. Asia Himalaya Centr. Asia Himalaya Chthoniidae Cheiridiidae Centrochthonius Apocheiridium Chthonius S Cheiridium Lagynochthonius Cryptocheiridium Tyrannochthonius Saaoe SOS © ore bt NOH w Atemnidae Lechytiidae Atemnus Lechytia Cyclatemnus Micratemnus Tridenchthoniidae Paratemnoides Compsaditha Titanatemnus Ditha Pycnodithella Cheliferidae Verrucadithella Dactylochelifer Gobichelifer Geogarypidae Chelifer Afrogarypus Hansenius Geogarypus Hysterochelifer Lophochernes Olpiidae Microchelifer Calocheiridius Garypinus Chernetidae Olpium Alochernes Caffrowithius Hyidae 0 Ceriochernes Stenohya Lasiochernes syn. Laevigatocreagris Dendrochernes Lamprochernes Neobisiidae Megachernes Bisetocreagris Nudochernes Microbisium Orochernes Neobisium Pselaphochernes Nepalobisium Roncus Withiidae Ectromachernes Syarinidae Stenowithius Ideoblothrus Trichotowithius Withius SoS eo ©o © ooooO rm = SisSızSszSsıDED Wee fo we NH QoS Oe 2] SS) SS) &S) NOCGCrrFr OC & So oS oo SIISESISESESZESESESE | 3 SEES oo Cc Ole dS USS) US) Sy SNS SiS Soo o's SS Wo NOR Ke S] Opilionida. The short-footed Cyphophthalmi do not live above 2000 m. The numerous species and genera belonging to Laniatores are confined mostly to tropical countries, including high mountains. In the mountains of tropical Africa, Laniatores are clearly predominant. In the high mountains of Europe, Laniatores do not live higher than 2000 m. In the Old World 227 species of harvestmen (84 Laniatores and 143 Palpatores) reach or go higher than 2200 m. Here 11 families are known to exist, but much less 158 are true high mountain-dwellers and only three families of Laniatores contain true members of high mountain species: Phalangodidae, Biantidae and Assamiidae. The latter two families live also above 4000 m (Biantidae up to 4250 m in Nepal, Assamiidae up to 4600 m on Kilimanjaro). Between 2200 and 4000 m the number of genera and species almost halves every 500 m. Above 4500 m only one species was found: Hypoxestus accentuatus Sorensen (Assamiidae) on Kilimanjaro. Table 3: The nine families of Old World Araneida living above 3500 m (listed according to altitude): Lycosidae: 70 species up to 6100 m Dictynidae: 9 species up to 4930 m Salticidae: 85 species up to 5947 m Thomisidae: 48 species up to 4880 m Linyphiidae: 447 species up to 5545 m Theridiidae: 24 species up to 4600 m Hahniidae: 14 species up to 5181 m Araneidae: 23 species up to 4500 m Gnaphosidae: 74 species up to 4980 m Araneida. High-altitude spiders belong almost exclusively to the suborder Araneo- morphae. The only three species (all African) of the suborder Mygalomorphae are not known from localities higher than 2500 m. Above 3500 m only representatives of Araneomorphae are met. The nine families, reaching or surpassing 4500 m, outline the habitus of the spider fauna in the highest regions of the Old World. They contain 794 of the total of 904 species of spiders living above 2200 m in the Old World (table 3). Table 4: Number and percentage of species of the family Linyphiidae that reach or surpass different altitudes of the total of Araneida species Araneida Linyphiidae altitude no. of species no. of genera no. of species (percentage) 2000 m a.s.l. 447 (49,6%) 3000 m .a.s.l. 37 179 (51%) 4000 m a.s.l. 18 af (0%) 5000 m a.s.l. 3 36370) From these data we can see that the bulk of species living in all high mountains of the Old World belong to the family Linyphiidae (s.lat.). Clearly none of the genera can be considered to be dominant in the high mountains of the Old World, except Lepthyphantes s.lat., which contains 10 % of all high altitude spiders and was recorded up to 5545 m. Pardosa and Xysticus also have many species widespread in higher altitudes. Acariformes. All three suborders occur in high mountains. The present maximum recorded altitude is 6100 m (in Nepal). — Acaridida are not well studied. Prostigmata are represented at altitudes above 2200 m in the Old World by not less than 256 species in 95 genera and 31 families (a tiny fraction of the 199 prostigmatic mites known in the world with more than 14 000 species in 140 families and 1100 genera). The most species-rich are the families of Trombiculidae (53), Trombidiidae (41), Erythraeidae (38), Bdellidae (18) and Rhagidiidae (18). The genus Balaustium (Erythraeidae) contains the most species (19). The families of the Prostigmata are cosmopolitan. Seventeen of them reach 3000 m, 11 (with 23 genera and 35 species) 3500 m, seven 4000 m, four 4500 m, and one 5000 m. None of all identified prostigmatic mites was recorded up to 6100 m. The highest records are of Adamystidae (5100 m in the Hindu Kush), Anystidae (4950 m in Nepal), Rhagidiidae (4800 m on Kilimanjaro), Pygmephoridae (4500 m in New Guinea), Erythraeidae (4260 m in New Guinea), Trombidiidae (4200 m on Ruwenzori), Trombiculidae (4155 m in Pakistan). With more intense research the upper limit of Bdellidae and other families most probably will rise. The bulk of the material so far collected remains unidentified due to the lack of specialists. Oribatida. Among the montane Acariformes, the soil-inhabiting species of the suborder Oribatida are most numerous. With their 66 families found at or above 2200 m, the Oribatids are the most diversified suborder of all high-altitude Arthropods, even more varied than the spiders. In the whole world more than 7000 oribatid mites have been described, belonging to 177 families. Their importance in natural ecosystems is enormous (in some places more than 70 % of all soil mites). Together with the theoretical interest they represent, this has led to an intensive research of the group by many specialists. The families Oppiidae (36), Ceratozetidae (33), and Phthiracaridae (18) contain most species. With increasing altitude the number of families and species decreases (table 5). Table 5: Decrease of Oribatida (families and species in the Old World) with increasing altitude > 2200 m: 66 families, 328 species > 4500 m: 16 families, 19 species > 3000 m: 41 families, 116 species > 5000 m: 13 families, 15 species > 3500 m: 31 families, 70 species > 5500 m: 9 families, 11 species > 4000 m: 19 families, 28 species Oribatida have been recorded in the Himalaya as high as 6100 m (Janetschek 1990). This author found in Nepal at up to 5800 m representatives of the families Brachychthoniidae, Hermanniidae, Damaeidae, Ceratoppiidae, Tectocepheidae, Suctobelbidae, Oribatulidae, Ceratozetidae and at up to 5500 m Oribatellidae, which are not yet identified to species level. Certainly at 5000 m tens of species live which are at the base of the scarce nival biota. Parasitiformes. All three suborders in this order are known from high mountains of the Old World. The Gamasida form the bulk of representatives of the order Parasitiformes, but they are as yet poorly studied, except the fauna of the Alps. Several species (mostly parasites and commensals on mammals) have been recorded from only a few places (Hindu Kush up to 4550 m, Kilimanjaro up to 4285 m). Members of at least eight families live higher than 3000 m: Parasitidae, Macrochelidae, Halolaelapidae, Zerconidae, Rhodacaridae, Hypoaspididae, Dermanyssidae, Laelapidae. Two of the three families of the suborder Ixodida are known from the montane environment (parasites on birds and mammals). Thanks to intensive research of the Himalayan fauna between 1960 and 1980 mainly by H. 160 Hoogstraal and his colleagues, Argasidae have been found up to 4575 m in Nepal and Ixodidae as high as 5488 m. Ixodes berlesei (on the Snow Partridge Lerwa lerwa in Nepal at 5488 m) is the highest representative of the Parasitiformes ever found. Symphyla. With about 160 species this is a small class. Both of its families (Scolopendrellidae and Scutigerellidae) live in all high mountain regions of the world and are represented in this environment with a total of 25 species. The genus Hanseniella prevails in the tropical mountains and is practically the only genus of Symphyla known to live in the mountains above 3000 m. Chilopoda. The centipedes contain about 2800 known species of which at least 80 live in the Old World above 2200 m. Identified species of Geophilomorpha in the Old World are known up to 3600 m (Tygarrup nepalensis Shinohara, Mecistocephalidae, Nepal), but Janetschek (1990) has collected Geophilidae in Nepal even at 4400 m. Our observations confirm the occurrence of Geophilomorpha at high altitudes. The genus Schendylurus has been recorded in Peru slightly higher (4500 m). Scolopendromorpha reach 4000 m on Mt Elgon (Scolopendra afra Mein.), 4500 m in Nepal (Cryptops doriae Pocock) and 4150 m in the Andes. The representatives of Lithobiomorpha (Lithobiidae) reach the highest altitude of all Myriapoda (up to 5700 m in the Himalaya). They are the most numerous within the four classes of Myriapoda and one of the most conspicuous arthropods in the high mountains of Eurasia. Only one other family of Lithobiomorpha has been recorded from considerable altitudes — two species of the genus Lamyctes (Henicopidae) reach 4000 m, one occurs above 4000 m in the mountains of tropical Africa. Scutigeromorpha reach 4250 m in Nepal (our unidentified collections); they are not recorded from high altitudes in Africa. Diplopoda. Out of the total of 16 orders in the class Diplopoda, 12 include species that are found above 2200 m in the Old World. As far as we know, in the orders Glomeridesmida, Platydesmida, Spirobolida and Siphoniulida no species are found at that altitude. Out of more than 12 000 species of Diplopoda described in the world, at least 237 have been recorded in the Old World from altitudes higher than 2200 m, including 20 from the mountains of tropical Africa. Material and method We summarized the available information from the literature on the species of Isopoda, Arachnida and Myriapoda in Africa living higher than 2200 m. This dataset was completed by data obtained by our own collecting in the mountains of tropical Africa: —in December 1977 from Buea to the top of Fako, the summit (4090 m) of Mount Cameroon, and on the island of Bioko (formerly Fernando Poo) on the slopes of Pic de Malabo (3008 m) — in 1983 on Kilimanjaro in Tanzania —in 1993 on Ruwenzori in Uganda and Mt Elgon in Kenya. 161 We also collected on many European mountains, in the Karakorum, Himalaya and other mountains in Asia. Here, too, the available information in the literature was gathered. When comparing of the oreals of Africa, Europe, Himalaya and Central Asia, special attention was paid to the distribution of IAM in different altitudinal belts. Parts of the collected material is still under study and therefore does not appear in the lists and in the different comparisons. Within the accounts there is some help for the reader: classes and orders, when they are named for the first time, are in bold and suborders are underlined. Isopoda, Arachnida and Myriapoda in the mountains of tropical Central and East Africa When we here describe the [AM from the montane Afrotropis so far identified, some aspects should be kept in mind: a great number of species, genera, families, or orders are known from lowland Africa. In most cases we do not point out that the taxon may occur elsewhere in Africa but that it is absent from the tropical mountains. The conclusions about endemism would change considerably if we were to include lowland taxa. Isopoda The essential information on the woodlice of the mountains of East Africa comes from the collections of the Swedish Expedition of Y. Sjöstedt in 1905-1906 (identified by Budde-Lund), from the material collected by R. Jeannel and the other participants to the Omo Expedition (identified by L. Paulian de Felice), from the collection of H. Franz (published by K. Schmölzer), and from the results of the Zoological Mission in East Africa of IRSAK (P. Basilewsky and N. Leleup), the material having studied by F. Ferrara and S. Taiti. Data on the terrestrial Isopoda of the high mountains of Ethiopia can be found in the paper by Barnard (1940), who identified the material collected by J. Omer-Cooper in 1926-27, as well as in the interesting note by A. Vandel in the general paper by Scott (1958). The species of the well-explored East and Central African mountains (Ruwenzori, Mt Elgon, Mt Kenya, Kilimanjaro, Aberdare, Uluguru, etc.) are well known, whereas we have little information about Ethiopia and Malaw, and the fauna of Mt Cameroon above 3000 m is still unexplored. Considerable material, presently being identified in Florence, was also collected by us on the mountains of Kilimanjaro, Mt Elgon and Ruwenzori. At the moment we know that 66 species of Isopoda in Afrotropical mountains reach or surpass 2200 m: Mt Elgon 9 species, Kilimanjaro 12, Mt Meru 11, Ruwenzori 3, Nyiragongo 2, Aberdare 6, Kivu 1, Uluguru 7, Mt Kenya 4, Ethiopia 4, Malawi 6, Mt Fako 1. At least 42 species live above 2500 m, 23 above 3000 m, 11 above 3500 m and 5 at or above 4000 m. In the East and Central African mountains above 2200 m there are at least 49 species belonging to 27 genera and five families. These are: Philosciidae (Afrophiloscia, Uluguroscia, Arcangeloscia, Pleopodoscia, Buddelundiscus), Porcellionidae (Thermocellio, Uramba, “Porcellio”), Oniscidae (Alloniscus), Eubelidae (Eubelum, Gelsana, Gerutha, Rufuta, Periscyphops, Stegosauroniscus, Benechinus, Kenyoniscus, Periscyphis, Mesarmadillo, Hiallum, Hiallelgon, 162 Arthiopopactes, Oropactes, Microcercus) and Armadillidae (Pseudodiploexochus, Ctenorillo, “Synarmadillo”). Twelve genera in four families (Eubelidae 9, Philosciidae |, Porcellionidae |, and Armadillidae I) reach or surpass 3500 m. The tropical family Eubelidae contains the highest number of genera and species living in high mountains. Eighteen genera contain species that live above 2200 m. The highest living are Aethiopopactes Verh (Ae. chenzemae) (4600 m on Kilimanjaro) and Benechinus Budde-Lund (B. armatus) (4600 m on Mt Meru), but also in the genera Eubelum, Hiallelgon, Hiallum, Mesarmadillo, Periscyphis, and Microcercus there are species recorded from altitudes at or over 3500 m. Genera of Isopoda endemic to the Afrotropical mountains are Benechinus, Gerutha, Hiallelgon, Hiallum, Kenyoniscus, Stegosauroniscus, (Eubelidae), and Ctenorillo (Armadillidae). High altitude endemics are Benechinus (only on Meru and Kilimandjaro, 2200-4600 m), Stegosauroniscus (Meru, 2200-2600 m, Kilimandjaro, 2200 m) and others. All 66 Isopoda species from Afrotropical mountains above 2200 m are endemics. Most species obviously occur only on one massif. Species that were collected from two mountain massifs are: Afrophiloscia uncinata, Stegosauroniscus horridus, Benechinus armatus (Kilimanjaro and Meru, which are close together but nevertheless separated by savanna) and Oropactes maculatus (Uluguru and mountains in Malawi). Arachnida Scorpionida. Some genera of the family Buthidae (Uroplectes, Lychas) are known to live at up to 3500 m on Mt Meru and at 3000 m on Kilimanjaro. Solpugida. Although this order is well represented in Africa, none of its many species is known to live in this continent higher than 2200 m (lack of research?). Schizomida. These small Arachnids are warm-loving and clearly avoid the high mountain environment. They have been found at up to 2600 m in Tanzania (Reddell & Cokendolpher 1995). Since at this altitude we find the usual vegetation cover we think that Schizomida are not a part of the hypsobionts of Africa. Pseudoscorpionida. The studies of Tullgren, Beier, Mahnert, Redikorzev and our own collecting in Kenya, Tanzania and Uganda have shown that 56 species (in 25 genera and 13 families) live at or above 2200 m in East and Central African mountains; 22 reach 3000 m, six 3500 m and one 1s known to live above 4000 m. Of the 56 species above 2200 m, almost all are endemic to the mountains of the Afrotropis (see footnote p. 166), but there are almost no endemic genera: Trichotowithius (Withiidae) is known only from two high-mountain species (above 2100 m), one in Ethiopia another on Mt Elgon. Most of the remaining genera are widespread in Africa, some of them even outside the continent. Besides the two species of Trichotowithius, we can consider as hypsobionts four species of the genus Nudochernes (N. nidicola 2470-3000 m, N. montanus 3500 m, N. crassus 3000-3700 m, and N. granulosus 2600 m) inhabiting the nests of Tachyoryctes (Rodentia, Rhizomyidae), as well as Tyrannochthonius brevimanus (2280-3300 m), Lechytia maxima (2350-2650 m), Cryptocheiridium elgonense (2650-3200 m), and Titanatemnus palmquisti (2000-4100 m), all from the well researched Mt Elgon, can be considered as endemics. T. palmquisti ıs found also on 163 Kilimanjaro and Mt Meru, raising again the question of how the “island” fauna of the Afrotropical mountains (spiders, Chilopods and other groups) might have developed. Opilionida. Knowledge of the rich montane Afrotropical fauna (at least 66 species living above 2200 m) comes from the studies of Sörensen (1910), Goodnight & Goodnight (1959), Roewer (1913, 1941, 1956), Lawrence (1962), and Kauri (1985). The short footed Cyphophthalmi do not live above 2000 m. The numerous species and genera of the Laniatores are confined mostly to tropical countries, including high in the mountains where they are strikingly predominant. Of the Palpatores, predominant in the Palaearctic Region, in the Afrotropical oreal we find only the largest family, Phalangiidae, which contains the bulk of the high altitude opilions, i.e. more than half of all genera and species known within the order ocurring above 2200 m. Among the 66 species of opilionids (46 from Laniatores and 20 from Palpatores, 11 and 6 genera respectively; both relations close to 2:1) recorded above 2200 m (on Kilimanjaro, Mt Kenya, Mt Meru, Mt Elgon, Aberdare, Uluguru, Oldeani, Ruwenzori, Hanang, and Semien), 34 species were found above 3000 m, 26 above 3500 m, 12 above 4000 m, and only two (Hypoxestus accentuatus, Assamiidae, and Rhampsinitus bettoni, Phalangiidae) above 4500 m. For an understanding of the sources of the Opilionida fauna of the high mountains of East and Central Africa table 6 may be useful. Table 6: Distribution of the two suborders of Opilionida in Africa (after Lawrence 1962) Laniatores Palpatores Phalangodidae Assamiidae Triaenonychidae Phalangiidae East Africa numerous dominant absent numerous Central Africa numerous dominant absent absent (Congo Area ) Madagascar numerous absent dominant absent Southern Africa rather numerous absent dominant numerous In the Afrotropical oreal at least 20 genera have been recorded, most of them from the Assamiidae (11 or 55 %), and six from the Phalangiidae (30%). The endemic genera of Afromontane opilions living between 2200 and 3500 m (usually for several mountains) are : Monobiantes (2200 m), Proconomma (2400 m), Abdereca (3100 m), Bambereca (2900 m), Erecula (2300 m), Eusidama (2400 m), Metarhabdopygus (2800 m), Sesostrellus (2900 m), Spinixestus (2400 m), and Hindreus (3300 m). Endemics at or above 3500 m can be found among the members of the genera Metabiantes, Monobiantes, Proconomma, Aberdereca, Bambereca, Ereca, Erecula, Eusidama, Hypoxestus, Metarhabdopygus, Metaereca, Randilea, Sesostrellus, Simienatus, Spinixestus, Cristina, Dacnopilio, Guruia, Hindreus, Odontobunus, Rhampsinitus. 164 Araneida. Much has been done in the study of Afrotropical spiders, but some of the most important publications were produced by older authors (E. Strand, A. Tullgren, R. de Lessert, H. Fage, E. Simon, L. Berland) and need revision. Newer contributions come from Benoit (1978), Berland (1914, 1920), Bosmans (1977, 1978, 1979, 1981a,b), Bosmans & Jocque (1983), Denis (1950, 1962), Fage (1940), Lessert (1915-1926), Russel-Smith & Jocqué (1986), Jocqué (1981), Scharff (1992), and Wesolowska (1986). When we analyze the high-altitude spiders of tropical Africa, we can see that only representatives of the suborder Mygalomorphae (Migidae, Dipluridae) live above 2200 m. Some genera are endemic to Africa (Mallinella, Aberdaria). The large number of African spiders in the suborder Araneomorphae fit into the general pattern of the Old World (see Introduction). Some genera living above 2200 m are: Lepthyphantes, Microlinyphia, Erigone, Asthenargus, Oedothorax, Meioneta, Trichopterna, Walckenaeria (Linyphiidae) and Heliophanus (Salticidae). Altitudes above 3500 m are reached by species belonging to the genera Pelecopsis (4930 m) and Heliophanus, Salticidae (4650 m), Araeoncus (4650 m), Callitrichia (4930 m), Microcyba (4320 m), Oreocyba (4300 m), Erigone (4200 m), Lepthy-phantes (4000 m), Toschia (3920 m), Walckenaeria (3820 m), Tybaertiella (3750 m), Asthenargus (3550 m), and Mallinella, Zodariidae (3500 m). In Africa no spiders have been found above 5000 m (on the three mountains higher than this). The family Linyphiidae, the most numerous with 97 species and 30 of the 36 genera, 1s particularly well studied (Scharff 1992). This family, with Callitrichia and Pelecopsis, reaches the highest altitudes in Africa. Sixteen genera living above 3000 m belong to the Linyphiidae. Some genera of Linyphiidae have developed many species: Ctenus (13), Microcyba (12), Pelecopsis (12), Callitrichia (11), or Lepthyphantes (9). The available data on the Afrotropical spiders are incomplete and unreliable; nevertheless we believe that - as the case is in other montane areas (see later) - in the Afrotropis at an altitude of 2200 m about 50% of all species of spiders belong to Linyphiidae. Acariformes. Acaridida are not well studied. Some are known from altitudes at or near to 3500 m (family Glycyphagidae with Glycyphagus domesticus and family Anoetidae with Histiostoma telatum) (André 1945, Mahunka 1982a). In the mountains of tropical Africa Prostigmata have been studied mainly by Evans and André. Our extensive collections from various mountains are still being studied. The data from the Afrotropical mountains mainly include the families Anystidae, Bdellidae, Erythraeidae, Trombidiidae and Microtrombidiidae. Rhagidiidae (also found in the Himalaya) were collected by us on Kilimanjaro at 3500 m, but are still under study. From the mountains of tropical Africa many genera have been recorded which are either unknown in Europe or not found in the higher zones of European mountains: Compsothrombium, Dinothrombium, Dromeoothrombium, Eutrichothrombium (Trombidiidae), Camerotrombidium, Enemothrombium (Microtrombidiidae). On the other hand families, genera and 165 even species (Anystis baccarum L. on Kilimanjaro) are shared with European oreal. The highest record for prostigmatid mites in tropical Africa is at 4200 m (Trombidiidae on Mt Elgon). Oribatida have been studied by Evans, Balogh, Mahunka, Stary and others. In the Afrotropical oreal at least 19 families have been recorded (see appendix). The oribatids, reaching or surpassing 3500 m in the Afrotropics, belong to the genera Liochthonius (3890 m), Tectocepheus (3890 m), Microtegeus (4285 m), Dampfiella (3890 m), Amerioppia (3820 m), Quadroppia (3820 m), Oppia (4285 m), Scutovertex (4438 m), Incabates ? (3820 m), Nannerlia (3890 m), Zygoribatula (3810 m), Scheloribates (4590 m), Africoribates (4590 m), Ghilarovizetes (3900 m), and Oribates (3810 m). Scheloribates laevigatus C.L. Koch was recorded (with questionmark) from Kilimanjaro up to 4590 m. This is also the maximum altitude at which oribatids (and Acari in general) in Africa have been recorded. Parasitiformes. The suborder Gamasida is not well known, especially in the Afrotropical oreal. Of Ixodida, Neumann (1910) and other authors have recorded the genera /xodes and Rhipicephalus in the high mountains of tropical Africa (up to 3500 m on Mt Meru). Symphyla. This group was studied by Ribaut, Silvestri and Scheller. In equatorial Africa seven species are known to live above 2200 m. The genus Hanseniella is dominant in the tropical mountains and is practically the only genus of Symphyla known to live above 3000 m. According to Silvestri (1907) and Scheller (1954), Hanseniella lives on Ruwenzori up to 4500 m. Symphylella vulgaris is found in Africa (above 2200 m) and in Europe. Chilopoda. Centipedes were studied mainly by Attems and Ribaut. Scolopendro- morpha reach 3500 m (Cryptops on Mt Meru and Mt Elgon) and 4000 m (Scolopendra afra Mein. on Mt Elgon). Lithobiomorpha are very scarce on the mountains of Central and East Africa. Instead of Lithobiidae we find Henicopidae. The genus Lamyctes goes much higher (up to 4200 m on Ruwenzori) than the European Lithobiomorpha. Geophilomorpha: In East Africa one species of Schendylurus (Schendyluridae) occurs up to 2210 m. This is a family which East Africa does not share with the other mountains of the Old World, but with South America. Species of Alloschizotaenia, Geophilidae, live up to 2800 m, those of Mecistocephalus, Mecistocephalidae, reach at least 3900 m (on Nyiragongo). Diplopoda. This class was studied especially by Attems (1909, 1939), Brölemann (1920), and Mauries & Heymer (1996). At least four orders live in tropical Africa above 2200 m: Polyxenida, Stemmiulida, Spirostreptida and Polydesmida. The tiny Polyxenida are represented in tropical Africa by one family (Polyxenidae) up to 2740 m (Mt Elgon). The family Stemmiulidae (Stemmiulida) reach 3500 m on Mt Meru, the Odontopygidae (Spirostreptida) reach 3000 m on Kilimanjaro. Polydesmida are represented in the Afromontane environment by the families Fuhrmannodesmidae (Elgonicola jeanneli Attems up to 3500 m on Mt Elgon, and Sphaeroparia petarberoni Mauriés et Heymer up to 4200 m on Ruwenzori) and Oxydesmidae (up to 3000 m on Kilimanjaro). 166 Comparison between the Isopoda, Arachnida and Myriapoda of the Palaearctic and the Palaeotropical mountain systems IAM in the Atlas and the mountains of tropical Africa None of the many Afromontane species, genera and even families of IAM is known to live in the oreal of the Atlas (Alderweireldt & Jocqué 1992, Denis 1961, 1967, Lépiney 1939, Verhoeff 1937, 1938 a,b). The montane IAM of the Atlas are of clearly Palaearctic type, which agree with the findings of other groups of animals and plants. IAM in the European oreal and in the mountains of tropical Africa Isopoda. If we compare the high mountain Isopods of Europe and of tropical Africa, we can see that between the 24 species in Europe and the 66 species in tropical Africa living higher than 2200 m there are no species in common. Of the 10 genera in Europe and the 35 in tropical Africa there are no genera in common. At the family level there is some overlap at this altitude: just two of the seven families living in Europe and the five in Africa are shared by the two continents (Porcellionidae and Oniscidae). In Europe the highest altitude 1s reached by representatives of the genus Porcellio (3300 m in Sierra Nevada). The family Porcellionidae is not very numerous in the montane Afrotropis. The highest occurring species belong to Uramba at 2500 m and Thermocellio at 2200 m — in the zone of tropical montane forests. The numerous Eubelidae (up to 4600 m), Philosciidae (up to 3700 m) and Armadillidae (up to 3100 m), so typical of the Afrotropis, are absent from the high mountains of Europe; Eubelidae do not live in Europe at all. Table 7: Comparison of typical families of Isopoda between European and Afrotropical oreals European oreal Afrotropical oreal Porcellionidae Porcellionidae Oniscidae Oniscidae Trichoniscidae Philosciidae Mesoniscidae Eubelidae Buddelundiellidae Armadillidae Trachelipodidae Armadillidiidae Pseudoscorpionida. Only 12 species of Pseudoscorpionida, belonging to 3 genera (Chthonius, Neobisium” and Allochernes) are known from the European oreal °) One exception is the European species Neobisium muscorum (= N. carcinoides), recorded by Mahnert (1981) on the basis of tritonymphs from Aberdare in Kenya (3203 m), the first representative of Neobisium in Africa south of the Sahara. In the extensive collections of Mahnert from African mountains, as well as in the collections of the older specialists (publications of M. Beier), we could not find Neobisium, so the authenticity and the autochthonous character of this species is doubtful. This is why we do not consider this record in our zoogeographical analysis. 167 (without the Caucasus). These genera are absent from the mountains of equatorial Africa. In the mountains of equatorial Africa all three European families are present, but also eight families more, three of which (Lechytiidae, Tridenchthoniidae and Ideoroncidae) do not live in Europe at all. Except the doubtful Neobisium, none of the remaining Afrotropical genera occurs in the European oreal. Only two of the genera (Withius and Apocheiridium) live on the European continent at all. Table 8: Comparison of families and genera of Pseudoscorpionida in the oreals of Europe and the Afrotropis Europe Afrotropis Chthoniidae Chthoniidae Chthonius - up to 3030 m Tyrannochthonius - up to 2900 m Neobisiidae Neobisiidae Microbisium - up to 3300 m Neobisium - up to 3600 m (Neobisium - up to 3203 m) Chernetidae Chernetidae Allochernes - up to 2600 m Caffrowithius - up to 3300 m Nudochernes - up to 3700 m Lechytiidae Lechytia - up to 2900 m Tridenchthoniidae Compsaditha - up to 2300 m Verrucadithella - up to 3200 m Geogarypidae Afrogarypus - up to 2900 m Ideoroncidae Negroroncus - up to 2250 m Cheiridiidae Apocheiridium - up to 2300 m Cryptocheiridium - up to 3200 m Atemnidae Cyclatemnus - up to 3000 m Micratemnus - up to 2200 m Paratemnoides - up to 3050 m Titanatemnus - up to 4100 m Cheliferidae Hansenius - up to 2250 m Microchelifer - up to 2700 m Withiidae Ectromachernes - up to 3000 m Stenowithius - up to 2180 m Trichotowithius - up to 3000 m Withius - up to 3500 Opilionida. All high-mountain harvestmen in Europe belong to five families of the Palpatores (28 species above 2200 m). In the mountains of equatorial Africa above 2200 m, 66 species of Opilionida (46 Laniatores and 20 Palpatores) have been recorded. In Europe only two species live above 3000 m (both in the genus Mitopus), in equatorial Africa 34. 168 Araneida. Out of the 18 genera of spiders, known in the Afrotropical oreal above 3000 m, more than half are also represented in similar habitats in Europe (Lepthyphantes, Microlinyphia, Erigone, Asthenargus, Araeoncus, Ceratinopsis, Pelecopsis, Meioneta, Walckenaeria, Heliophanus). From eight genera known above 4000 m, five (Lepthyphantes, Erigone, Araeoncus, Pelecopsis, Heliophanus) are also inhabitants of the higher parts of European mountains. Acariformes. The mites of the suborder Acaridida are not adequately studied in both areas. The genus Histiostoma (Anoetidae) is present both in European and Afrotropical mountains. In Europe and especially in the Alps Prostigmata are better known (over 2200 m with at least 20 families, 35 genera and more than 100 species, excluding the strict parasites such as Myobiidae, Listrophoroidea, etc.). This group is less well known in the Afrotropical oreal. Families, genera and even species (Anystis baccarum L.) are shared with the European oreal. The highest record for prostigmatid mites in Europe is at 3774 m (Bdellidae in the Alps), in tropical Africa at 4200 m (Trombidiidae on Mt Elgon). In Europe, many genera recorded on the mountains of tropical Africa are either unknown or are found only in the lower regions. We have already considerable information on the representatives of Oribatida in the oreals of Europe and tropical Africa. Most of the families in the Afrotropical oreal live also in the European mountains. Many genera are also shared (Liochthonius, Brachychthonius, Heminothrus, Nothrus, Nanhermannia, Scheloribates, Ceratozetes, Galumna, Tectocepheus, Suctobelba, Oppia, Quadropia, Rhysotritia). Scheloribates laevigatus C.L. Koch occurs in the Alps up to 2700 m. Parasitiformes. The suborder Gamasida is not well known in both regions. The other suborder, Ixodida, is represented in both with different species of the genus Ixodes. Symphyla. Both families of Symphyla live in both regions. In Europe, seven species are known above 2200 m, in equatorial Africa also seven. The two regions even share one species - Symphylella vulgaris. All other Afromontane Symphyla belong to the genus Hanseniella, represented also in Europe but only by a small number of lowland species. In the Afrotropical oreal the genera Geophilella and Scutigerella have not been found, but they are known from some European mountains. Chilopoda. In Europe, Geophilomorpha reach 2900 m (Geophilus) and Scolopendromorpha 2700 m (Scolopendra, both in the Sierra Nevada). Scutigeromorpha do not live in the higher parts of the European mountains. Lithobiomorpha (family Lithobiidae) is represented by the genera Eupolybothrus (up to 2500 m) and Lithobius (up to 2914 m), in total by at least 20 species. In the Afrotropical oreal Geophilomorpha reach at least 3900 m (Mecisto- cephalus on Nyiragongo), and Scolopendromorpha 3500 m (Cryptops, Lamnonyx on Mt Meru and Mt Elgon). Instead of Lithobiidae (Lithobiomorpha) we find 169 Henicopidae in the Afrotropical oreal. The genus Lamyctes occurs much higher than the European Lithobiomorpha (4000 m). Diplopoda. Of thel3 orders of Diplopoda living in high-altitude environments of the Old World, six live in the European and four in the Afrotropical oreal (table 9). Table 9: Comparison of the occurrence of orders (and families) of Diplopoda in the oreals of Europe and Africa European oreal African oreal Polyxenida (Polyxenidae) Polyxenida (Polyxenidae) Glomerida — Craspedosomatida — Julida — Chordeumatida — - Stemmiulida - Spirostreptida Polydesmida Polydesmida (Polydesmidae) (Fuhrmannodesmidae, Paradoxosomatidae) We can see that at the level of orders there are considerable differences between the two areas. Only two orders - the Polyxenida and Polydesmida - and one family - Polyxenidae - are shared. IAM in the Himalaya and mountains of tropical Africa The Himalaya is much higher and is situated farther north than the mountains of tropical Africa. Nevertheless, some IAM reach hisher altitudes in the Himalaya (over 5000 and even 6000 m) than in the African oreal. The Himalaya has a dual zoogeographical character: it belongs partly to the Palaeotropical realm, of which the mountains of tropical Africa are also a part, but their northern slopes form the southern limit of the Holarctic. Table 10: Comparison of the families of Isopoda from the Himalaya and Afrotropical mountains Himalaya above 2200 m Afrotropics above 2200 m Philosciidae Philosciidae Armadillidae Armadillidae Porcellionidae Porcellionidae Oniscidae Oniscidae Trachelipodidae Eubelidae 170 Isopoda. Only nine species of terrestrial Isopoda, belonging to five families, have been recorded so far from the Himalaya from altitudes over 2200 m. In the Afrotropical oreal their number is much higher: 66 species in 5 families. Of these five families four are shared by the two regions. The family Eubelidae in Africa is replaced by the Trachelipodidae in the Himalaya (see table 10). None of the five genera of woodlice in the high Himalaya and the 35 genera in the mountains of tropical Africa occurs in both regions. Pseudoscorpionida. In the Himalaya above 2200 m, 30 species of pseudoscorpions have been found, 17 species above 3000 m, 7 above 3500 m, and 2 above 4000 m. The species living higher than 2200 m belong to 20 genera and 12 families. In the mountains of tropical Africa above 2200 m, 56 species in 25 genera and 13 families are known (without the doubtful finding of Neobisium - see footnote p.166). There is a considerable similarity in the number of taxa of all ranks. In the Himalaya, the family Hyidae (the highest) is additional, in the Afrotropical mountains the Ideo- roncidae. Both mountain systems share 10 families and 5 genera above 2200 m: Tyrannochthonius, Lechytia, Apocheiridium, Calocheiridius and Withius. There are no species in common. In the Himalaya, pseudoscorpions have been recorded up to 5000 m (Stenohya), and in the Afrotropical oreal up to 4100 m (Titanatemnus). Table 11: Decrease in number of species of Pseudoscorpionida with altitude in the Himalaya and the mountains of the Afrotropics Himalaya Afrotropics > 2200 m 30 56 > 3000 m 17 2D > 3500 m 7 6 > 4000 m 2 | Opilionida. As already said, the opilionid fauna of the Himalaya reflects the dual zoogeographic character of this mountain system. In the higher parts of the Himalaya both suborders Laniatores and Palpatores (the last suborder being predominant) are present. In the Himalayan opilionids above 2200 m (including four species found up to 2150 m) the ratio of the genera of Laniatores:Palpatores ıs 9:21, and of species 29:58 (in both cases approximately 1:2). For comparison: in the mountains of Central and equatorial East Africa above 2200 m, the ratio of genera is 11:5, and of species 46:20. The ratio is just the opposite. In the Afrotropical oreal, 26 opilionid species (out of 66) have been found above 3500 m, 12 above 4000 m and only two (Hypoxestus accentuatus, Assamiidae, and Rhampsinitus bettoni, Phalangiidae) above 4500 m. In the Himalaya above 2200 m, at least 87 opilionid species are known, 45 above 3000 m, 15 above 3500 m, seven above 4000 m and two above 4500 m (Himalphalangium palpale, Phalangiidae, and Sabacon sp., Sabaconidae, which occurs even above 5000 m). In the mountains of tropical Africa above 2200 m, three families of Opilionida (Assamiidae, Biantidae, Phalangiidae) are found, and seven in the Himalaya. All high-altitude wa African families are also present in the high Himalaya, but four additional families occur (Oncopodidae, Sabaconidae, Phalangodidae, Sclerosomatidae). In the Afrotropical oreal at least 20 genera have been recorded, | 1 (55%) of Assamiidae and six (30%) of Phalangiidae, more typical of the Holarctic realm. In the high Himalaya, Martens, Suzuki and other specialists have found representatives of at least 30 genera above 2200 m, 6 (20%) of Assamiidae and 18 (60%) of Phalangiidae. There are no genera in common between the two mountain systems. Araneida. There are some differences between the Himalayan and the Afrotropical oreals in families and even in suborders. In the Afrotropical oreal (up to 2500 m) representatives of Mygalomorphae occur that are absent in the Himalaya above 2200 m. Here, only species of the suborder Araneomorphae are known (Anapidae, Tetrablemmidae, Sicariidae, Salticidae, etc). In both areas the family Linyphiidae dominates. Both systems also share some genera belonging to this family (Lepthyphantes, Erigone, Asthenargus, Walckenaeria, Oedothorax) which also occur in the European oreal. Some genera have developed many species in the high Himalaya: Lepthyphantes (19), Yaginumaella (14), Xysticus (9), and Pardosa (8). But there are no species in common between the Himalaya and Central Africa. We found that four families of spiders in the Old World form the highest associations, inhabiting the regions above 5000 m: Lycosidae, Salticidae, Linyphiidae and Gnaphosidae. In the Himalaya above 3500 m, we find 18 genera, the highest at 6100 m. In Africa no spiders have been found above 5000 m (in the three mountains overpassing this altitude). Acariformes. No information exists on Himalayan Trombidiidae, but the other recorded Prostigmata in both areas are completely different. Rhagidiidae have been collected by us on Kilimanjaro at 3500 m, but are still under study and it is not possible to compare them with the five species of five genera found in the Himalaya at 3900 m (Zacharda & Daniel 1987). Oribatida. Almost all families and genera, recorded from the high Himalaya (above 2200 m) are also known from the higher mountains of tropical Africa. In the Himalaya a few species have been recorded from this altitude, mostly by Aoki, Piffl, Sheals, Travé, Mahunka and other authors. Janetschek (1990) mentions many species - identified only up to the genus, found during his expedition up to 5800 m - that are some of the highest in the world. The genera, however, are widespread and are present also in the European mountains (Liochthonius, Hermannia, Belba, Oribatula, Scheloribates, Trichoribates, Oribatella, Ceratoppia, Tectocepheus, and Suctobelba). Parasitiformes. The suborder Gamasida is not well known in both regions. The suborder Ixodida reaches 5488 m in the Himalaya (/xodes berlesei Birula, the highest locality for Ixodida and Parasitiformes in the world). At least six species of the genus /xodes in the Himalaya live above 3600 m. In the Himalaya Haemaphysalis aponommoides Warburton also occurs up to 4880 m, as does the endemic genus Anomalohimalaya. In tropical Africa /xodes (with different species) and Rhipicephalus reach considerable altitudes (up to 3500 m). 172 Symphyla. The two families in the class Symphyla and the genus Hanseniella are shared by the Himalaya and the mountains of equatorial Africa. In the high Himalaya only 1-2 species of Symphyla have been recorded (up to 4900 m), as opposed to seven species in the higher parts of equatorial Africa (up to 4500 m). Chilopoda. Scutigeromorpha. Not yet identified scutigeromorphs have been collected by us in Nepal up to 4250 m. There is no information concerning this order in the high- altitude tropical Africa. Lithobiomorpha. The centipedes of this order are widespread in the Himalaya and known up to 5545 m (Lithobius hirsutipes khumbensis): In these mountains above 2200 m at least ten species have been found, all belonging to Lithobiidae (Lithobius with two species). The family Lithobiidae and genus Lithobius are known from the High Atlas, but are absent from the Africotropical mountains. There we find (up to 4000 m on Mt Kenya) two species of the genus Lamyctes (family Henicopidae), which is unknown in the high Himalaya. The difference between the two oreals in the Lithobiomorpha is at the family level. Geophilomorpha. As in the Lithobiomorpha, the difference here also lies at the family level: inthe Himalaya at least two species of Tygarrup (Mecistocephalidae) occur, in East Africa one species of Schendylurus (Schendyluridae). Scolopendromorpha. Several Scolopendromorpha (Scolopendridae) live ın the high Himalaya above 2200 m (Cryptops up to 4500 m, Otostigmus up to 3000 m, Ethmostigmus up to 2700 m and Rhysida up to 3400 m). That means that only Crytops is shared by the oreals of the Himalaya and Central Africa. Diplopoda. In the last two decades the Diplopoda fauna in the Himalaya has become one of the best known in the world, mainly because of the intensive field work of J. Martens and the research of Golovatch, Mauries, Enghoff, Shear and Table 12: Comparison of the orders and families of Diplopoda between the Himalayan and Afritropical oreal Himalaya Africa Order Spirostreptida Harpagophoridae Odontopygidae Order Polydesmida Fuhrmannodesmidae Fuhrmannodesmidae Paradoxosomatidae Paradoxosomatidae Polydesmidae Oxydesmidae Opisotretidae Order Polyxenida Polyxenidae Polyxenidae 173 other specialists. The two regions share three orders: Polyxenida, Spirostreptida and Polydesmida, but only three families (Polyxenidae, Fuhrmannodesmidae, and Paradoxosomatidae) and not a single genus (table 12). At least five genera in the family Fuhrmannodesmidae live in the Himalaya above 2200 m, while on Ruwenzori in the Afrotropical oreal the genus Sphaeroparia reaches 4200 m (S. petarberoni Mauries et Heymer). IAM in the mountains of tropical Africa and Central Asia The mountains of Central Asia (Afghanistan, Pakistan, Hindu Kush, Karakorum, Pamir, Tien Shan, Kunlun, and Tibet - from the border between Iran and Afghanistan to 120° E) are situated more to the north and are higher than the mountains of tropical Africa (which do not reach 6000 m). The two systems have a different vertical zonation, origin, climate and vegetation. They also belong to different zoogeographical regions and even realms, and their faunas have been isolated from each other at least since the Quaternary. Isopoda. From the mountains of Central Asia only ten species of Isopoda have been recorded above 2200 m, belonging to three families. The two areas in question share two families: Porcellionidae and Armadillidae. In the mountains of tropical Africa the family Eubelidae dominates, but does not occur in Central Asia. The Central Asian mountains share the family Trachelipodidae with Europe and the Himalaya. None of the four genera in the Central Asian oreal and the 35 genera in Afrotropical mountains is common to both. The arid mountains of Central Asia are characterized by genera like Protracheoniscus and Desertoniscus (as high as 4725 m), the Afrotropical oreal by the more hygrophilous representatives of Eubelidae, Philosciidae, etc. (up to 4600 m). Pseudoscorpionida. At least 12 species in eight families have been recorded from Central Asia above 2200 m, including 11 above 3000 m and four above 3500 m. From the oreal of tropical Africa over 2200 m, 28 species of Pseudoscorpionida have been recorded, belonging to 21 genera and 11 families (up to 4100 m). The highest Central Asian records within the Neobisiidae are of Bisetocreagris kaznakovi (Red.) (Tibet 4810 m), and within the Cheliferidae Dactylochelifer brachialis Beier (Karakorum 4200 m), Gobichelifer chelanops (Red.) (Karakorum 3650 m) and “Chelifer” baltistanus di Cap. (Karakorum 3950 m). All families of pseudoscorpions from montane Central Asia live also in the high mountains of tropical Africa; of the African families, the Lechytiidae, Tridenchthoniidae and Withiidae are not represented in Central Asia above 2200 m. There are no genera in common between the two areas. Opilionida. Few species of Opilionida are known to live in the vast arid mountains of Central Asia. This fauna is much poorer than the Himalayan or the Afrotropical. 174 They all belong to the suborder Palpatores and mainly to the family Phalangiidae s.lat. (one species belongs to Sclerosomatidae). The highest species is Homolophus nordenskioeldi, living at up to 5600 m (a world record for Opilionida). Representatives of the genera Diabunus, Egaenus, Opilio(?) and Phalangium(?) live above 3500 m. These data are rather fragmentary and after research in some remote corners of Tibet with Indomalayan influence the picture is likely to change. For the moment there are no Opilionid genera in common between the Afrotropical and the Central Asian oreals. Araneida. In Central Asia above 2200 m more than 180 species of spider have been found, belonging to 18 families, but only four families contain at least 148 of these species: Linyphiidae (87), Gnaphosidae (22), Salticidae (19) and Lycosidae (20). The 88 species of the Linyphiidae belong to 36 genera (in the Afrotropical oreal there are 97 species in 30 genera). The figures are quite close. We have already pointed to the amazing coincidence in the percentage of high-altitude Linyphiidae versus the other spiders ın the high Alps (51,2%) and in the mountains of Central Asia (51,8%). Nine genera (probably more) are common to the two areas above 2200 m: Lepthyphantes, Microlinyphia, Erigone, Asthenargus, Oedothorax, Meioneta, Trichopterna, Walckenaeria (Linyphiidae) and Heliophanus (Salticidae). We mentioned already that in tropical Africa more than half of the 18 genera of spiders living above 3000 m are also represented in the oreal of Europe. Of the genera living above 3500 m (38 in Central Asia, 14 in the Afrotropical oreal) four are shared, of those above 4000 m (8 ın Africa, 26 in Central Asia) three are shared, and above 4500 m only Erigone and Heliophanus live in both areas. The four shared genera above 3500 m are: 3 Heliophanus (Salticidae): 4600 m (Central Asia), 4650 m (tropical Africa) Erigone (Linyphiidae): 4950 m (Central Asia), 4200 m (tropical Africa) Lepthyphantes (Linyphiidae): 4250 m (Central Asia), 4000 m (tropical Africa) Pardosa (Lycosidae): 5170 m (Central Asia), 3700 m (tropical Africa). Acariformes. The suborder Oribatida deserves special attention, as there is reliable information about these important mites in both regions. There are a number of common families: of the 20 families in the Afrotropical oreal and the 35 in Central Asia, 12 are shared, but many others may also be in common since they occur in one of the regions but have so far been recorded only at lower altitudes. There are also some genera in common (Liochthonius, Heminothrus, Nanhermannia, Tectocepheus, Amerioppia, Quadroppia, Oppia, Scutovertex, Zygoribatula, Scheloribates) but no species. In the mountains of Central Asia an altitude of 3500 m is reached by the genera Hypochthonius, Liochthonius, Heminothrus, Nothrus, Scutovertex, Zygoribatula, Gerloubia, Scheloribates, Ceratozetella, Diapterobates, Punctoribates, Eupelops, Unduloribates, Suctobelbella, and Novosuctobelba, that of 5000 m by Tectocepheus, Oribotritia, Oribatella, and Sphaerobates. 175 In this (clearly incomplete) list of 19 Central Asian montane genera we find five living in both oreals: Liochthonius, Tectocepheus, Scutovertex, Zygoribatula, and Scheloribates. The limit of 3500 m for many of the Central Asian genera is an artificial one, because this was the upper limit of the research carried out by Maria Hammer in the Hindu Kush. In the Himalaya, oribatids have been recorded up to 6100 m, which may be true for Central Asia too. For the other Acari, /xodes is the shared genus (up to 3600 min Tien Shan, up to 3500 m on Mt Meru). Symphyla. So far no specimens of this class have been recorded from Central Asia. Chilopoda. Lithobiomorpha. As in Europe and the Himalaya, in Central Asia representatives of the genus Lithobius (Lithobiidae up to 4300 m) dominate, and they do not occur in the Afrotropical oreal. In Tajikistan we found another genus of the family Lithobiidae: Hessebius, as high as 4500 m. In tropical Africa the litho-biomorphs are represented by the family Henicopidae (Lamyctes up to 3500 m). In the Scolopendromorpha one genus is shared by the two oreals - Scolopendra (in Central Asia up to 2700 m, in tropical Africa up to 2710 m). This is the only genus of Scolopendromorpha collected in Central Asia so far. The order Geophilomorpha in Central Asia is represented by two families (Geophilidae up to 2820 m in Afghanistan and Mecistocephalidae up to 3500 m in Tibet). They live also in the Afrotropical oreal, but the genera are different. Scutigeromorpha. Verhoeff has described one species of the genus Thereuopoda from Central Asia (3100 m); they are unknown in the oreal of tropical Africa. Diplopoda. Central Asia, and especially its high and arid parts, are among the poorest regions for Diplopoda in the world. According to Golovatch, in the deserts of Central Asia huge regions exist where this forest-dwellers are absent. The few genera we know of are close to the Himalayan fauna and have nothing in common with the rich diplopod fauna living in the Central African mountains. Discussion The comparison between the non-insect arthropod faunas of the montane Afrotropics and the oreals of Eurasia reveals many agreements but more differences. If we exclude the case of Neobisium carcinoides (see footnote p. 166), there are no shared species. Table 13 presents what we see in the genera. Obviously the mutuality depends largely on the mobility of the animals. In groups with low mobility (Isopoda, Opilionida, Diplopoda) there are no common genera. Some others (Araneida, Acariformes, Parasitiformes) are more widespread. In the case of Araneida the higher mobility may be caused by balooning, in the parasitiform mites the mobility of the hosts is obviously the reason for their extension. 176 But we know many examples which do not fit in this scheme. According to Scharff (1992), 14 linyphiid spiders (Araneida) are endemic to the Afroalpine region. He quotes: “The high degree of endemism among forest linyphiid species on nearby mountains in eastern Tanzania does not support a theory of common or Table 13: Number of genera in different taxa in the montane Afrotropis (above 3500 m) and those shared with in the corresponding altitudes of Eurasia Class or order no. in the Afrotropis shared genera in Eurasia Isopoda 10 0 Scorpionida 0 Pseudoscorpionida 4 2 Opilionida 9 0 Araneida 18 13 Acariformes 2! 18 Parasitiformes 4 4 Symphyla l i Chilopoda 5 2 Diplopoda 8 0 occasional contact between the faunas as a result of balooning”. A mite of the family Bdellidae (Acariformes, Prostigmata) is not more mobile than a spider, but we can see that Bdella, Erythraeus and other Prostigmata can be found in all or most continents. A Pardosa spider (Araneida, Lycosidae) is not more mobile than any of the Opilionida, but the harvestmen on African mountains are largely endemic to the continent whereas those lycosids are widespread. Most genera of Oribatida (Acariformes) are also widespread although they have low mobility. These patterns obviously depend on other factors, such as age of the group, ecological plasticity, type of speciation, etc. The distribution of some families, e.g. Eubelidae (Isopoda), Assamiidae and Biantidae (Opilionida), is palaeotropical or pantropical. Their genera, however, are entirely Afrotropical and have evolved in Africa. Very few of them are confined to the montane regions. Fage (1940) noted that the separate high mountain massifs in tropical Africa (Kilimanjaro, Mt Kenya, Mt Elgon, Ruwenzori and others), rising like islands amidst the dry and hot savanna, share many species of spiders. It is obvious that their establishment must have taken place under quite different climatic conditions, when the montane environment was more widespread. An active exchange at present is impossible. According to Scharff (1992) the idea of Fage (1940) is not verified by the actual data: “It is a general trend that the different mountains harbour different linyphiid species”. In the opilionids, however, we see that many of them live on two or more mountain massifs, with the reservation that until now only some of the many high mountains have been sufficiently investigated. 77 References A selection of papers on the non-insect arthropods in tropical Africa above 3500 m and papers used in the text: Alderweireldt,M. & R. Jocqué (1992): A contribution to the knowledge of the wolf spider fauna of Morocco: new species and new records (Araneae: Lycosidae). — Phegea 20: 15-119. Alluaud, Ch. (1908): Les coléopteres de la fauna alpine du Kilimandjaro avec notes sur la faune du Mont Meru. — Ann. Soc. ent. France 77: 21-32. Pane. Ch. ER. Jeanne! (1912): Voyage de Ch: Alluaud et R: Jeannel en Afrique Orientale (1911-1912). Liste des Stations, 24 p. — Limoges et Paris. André, M. (1936): Descriptions de plusieurs Microthrombidium de |’ Afrique Orientale (Mission de l1’?Omo, 1932-1933). — Bull. Mus. nat. Hist. nat. Paris, 2e s. 8(5): 406-422. André, M. (1938): Mission scientifique de I’Omo.- Mém. Mus. nat. Hist. nat. Paris, Zoologie, 37(4): 212-254. André, M. (1945): Mission scientifique de !’Omo, VI. 57. Faune des terriers des rats-taupes. XI. Acariens. - Mém. Mus. nat. Hist. nat. NS 19(1): 199-210. André, M. (1957): Contribution a 1’ Etude des Thrombidions du Congo Belge. — Rev. Zool. Bot. Afr. 56(3-4): 301-344. André, M. (1965): LXXXVI — Acari Thrombidiidae. Mission zoologique de VPI.R.S.A.C. en Afrique orientale. (P. Basilewsky et N. Leleup, 1957).- Ann. Mus. Roy. Afr. Centr., 8, Zool. 138: 65-75. Attems, C.G. (1909): Myriopoda. — In: Wissenschaftliche Ergebnisse der Kilimanjaro-Meru Expedition (ed. Sjöstedt, Y.). — Stockholm, 3: 1-64. Attems, C.G. (1937): Chilopoden und Symphylen des Belgischen Congo. — Rev Zool. Bot. Afr. 29(3): 317-332. Attems, C.G. (1939): Myriopoda. — In: Mission Scientifique de l’Omo, V. - Mem. Mus. Nat. Hist. nat., NS, 5(55): 303-318. Balogh, J. (1966): On some oribatid mites from Tshad and East Africa collected by Prof. H. Franz, Vienna. — Opusc. Zool. Budapest 6(1): 69-77. Barnard, K.H. (1940): XXXVI. Entomological Expedition to Abyssinia, 1926- 7: Woodlice collected by Mr. J. Omer-Cooper. — Ann. Mag. nat. Hist. (1) 6: 355- 366. Beier, M. (1944). Uber Pseudoscorpioniden aus Ostafrika. — Eos, Madrid 20: 212. Beier, M. (1951): On some Pseudoscorpioidea from Kilimanjaro. — Ann. Mag. Nat. Hist. 12(4): 606-609. Beier, M. (1955): Pseudoscorpionide, gesammelt während der schwedischen Expeditionen nach Ostafrika 1937-38 und 1948. — Ark. Zool., ser. 2, 25: 527-558. Beier, M. (1959): Pseudoscorpione aus dem Belgischen Congo gesammelt von Herrn N. Leleup. — Ann. Mus. Congo Belge, Sci. Zool. 72: 5-69. 178 Beier, M. (1962): Pseudoscorpionidea. In: Mission Zoologique de !’I.R.S.A.C. en Afrique orientale VIII. - Ann. Mus. Roy. Afrique. Centr. Tervuren, ser. 8, Sci. Zool. 107: 9-37. Benoit, P.L.G. (1978): Hahniidae du Mont Kenya (Araneae). Scientific Report of the Belgian Mt. Kenya Bio-Expedition, 1975, No 16. — Rev. Zool. afr. 92(3): 609-621. Berland, L. (1914): Arachnida III: Araneae (Ire partie). — Voyage de Ch, Alluaud et R. Jeannel en Afrique Orientale (1911-1912), Résultats scientifiques: 37-94. Berland, L. (1920): Arachnides IV: Araneae, Qe partic). — Vovagerdes er Alluaud et R. Jeannel en Afrique Orientale (1911-1912), Résultats scientifiques: 95-180. Bosmans, R. (1977): Scientific Report of the Belgian Mt. Kenya Bio-Expedition No 3. Spiders of the subfamily Erigoninae. — Rev. Zool. afr. 91(2): 449-472. Bosmans, R. (1978): Description of four new Lepthyphantes species from Africa, with a redescription of L. biseriatus Simon & Fage and L. tropicalis Tullgren. — Bull. Br. arachnol. Soc. 4(6): 258-274. Bosmans, R. (1979): Scientific report of the Belgian Mt.Kenya bio-expedition No 17. Spiders of the subfamily Linyphiinae from Mount Kenya (Araneae- Linyphiidae). — Rev. Zool. afr. 93(1): 53-100. Bosmans, R. (198la): Scientific Report of the Belgian Mt, (Cameroon Expedition 1981. I. Situation of the collecting sites on the altitudinal gradient. Description of Hahnia leopoldi n.sp. — Rev. Zool. afr. 96(3): 670-682. Bosmans, R. (1981b): Etudes sur les Hahniidae (Araneae) africains II. Les especes du genre Hahnia de la collection Simon. — Bull. Mus. natn. Hist. nat., Panis) de ser, sect. A. 122052 Bosmans, R. (1985): Les genres Troglohyphantes Joseph et Lepthyphantes Menge en Afrique du Nord (Araneae, Linyphiidae). Etudes sur les Linyphiidae nord-africaines, III. - Rev. Arachnol. 6(4): 135-178. Bosmans, R. & R.Jocqué (1983): Scientific report of the Belgian Mount Cameroon expedition 1981. No 9. Family Linyphiidae (Araneae). — Rev. Zool. alt. 973): 381617. Bosmans, R. & E. Thijs (1980): Spiders of the family Hahniidae from Mount Kenya. Scientific report of the Belgian Mt. Kenya bio-expedition no 20. — Revue Zool. afr. 94(3): 559-569. Brolemann, H.W. (1920): Myriapodes III Diplopoda. — In: Voyage de Ch. Alluaud et R.Jeannel en Afrique orientale (1911-1912). — Rés. Sci.: 49 — 298. Budde-Lund, G. (1910): Crustacea. Isopoda. — In: Wissenschattliche Ergebnisse der Kilimanjaro-Meru Expedition (ed. Sjöstedt, Y.) 3: 3-20. Cernik, A. & J. Sekyra (1969): Zemepis velehor. — Academia, Praha. Coe, M.J. (1967): The ecology of the Alpine zone of Mount Kenya Monograph. Biol. 17: VIII + 136 p. 79 Coe, M.J. (1989): Biogeographical affinities of the high mountains of tropical Africa. — In: Quaternary and environmental research on East African mountains (ed. Mahaney, W.C.) — Rotterdam: 257-278. De Lattin, G. (1967): Grundriss der Zoogeographie. — Gustav Fischer Verlag, Jena. Denis, J. (1950): Spiders from East and Central African mountains collected by Dr. George Salt. — Proc. zool. Soc., London 120: 497-502. Denis , J. (1961): Notes d’ aranéologie marocaine. IX. Les Araignées du Maroc. I. Orthognatha et Haplogynae. — Soc. Sci. Nat. Phys. Maroc 41: 141 - 167. Denis, J. (1962): Notes sur les Erigonides. XX. Erigonides d’ Afrique orientale avec quelques remarques sur les Erigonides éthiopiens. — Rev. Zool. Bot. Afr. 65(3-4): 169-203. Denis, J. (1967): Notes d’Aranéologie marocaine. X. Les Erigonides du Maroc. — Bull. Soc. Sci. Nat. Phys. Maroc 47: 137-164. Evans, G.O. (1953): On a collection of Acari from Kilimanjaro. — Ann. Mag. nat. Hist. 12(6): 258-281. Fage, L. (1940): Le peuplement en Arachnides des hauts massifs montagneux de l’ Afrique Orientale. — VI-e Congres intern. Entom., 1935, 2: 487-491. Ferrara, F. & S. Taiti (1984): New species and records of Philosciidae (Terrestrial & Isopods) from Eastern Africa. — Rev. Zool. afr. 98(2): 293-307. Franz, H. (1979): Ökologie der Hochgebirge. — Ulmer Verlag, Stuttgart, 495 p. Fries, R.E. & T.C.E. Fries (1948): Phytogeographical researches on Mount Kenya and Mt. Aberdare, British East Africa. — K. Svenska Vetensk. Akad. andl 25(5). Goodnpcsiiie. @ 17 & ML Goodnieht (1959): Report on a collection of Opilionids from East Africa. — Ark. Zool. 12(15): 197-222. Hauman, L. (1955): La “région afro-alpine” en phytogéographie centro- africaine. - Webbia 11: 467-469. Hedberg, O. (1951): Vegetation belts of the east African mountains. — Svensk. Bot. Tidskr. 45: 140-202. Hedberg, O.(1965): Afro-alpine flora elements. —- Webbia 19: 519-529. Hingston, R.W.G. (1925): Animal life at high altitudes. — Geogr. J. 65: 185- 198. Holm, A. (1962): The spider fauna of the East African mountains. — Zool. Bidr., Uppsala 35: 19-204. Janetschek, H. (1990): Als Zoologe am Dach der Welt. — Ber. Nat.-Med. Ver. Innsbruck, Suppl. 6: 119p. Jeannel, R. (1950): Hautes montagnes d’ Afrique. — Mus. Nat. Hist. Nat. Paris (Suppl.) 1: 1-253. Jocqué, R. (1981): Some Linyphiids from Kenya with the description of Locketidium n. gen. (Araneida, Linyphiidae). — Rev. Zool. afr. 95(3): 598-569. 180 Kauri, H. (1985): Opiliones from Central Africa. — Ann. Mus. R. de Il’ Afrique Centrale, Tervuren 245 — Sci. Zool.: VIII+168 p. Killick, D.J.B. (1978): The Afromontane. — biogeography and ecology of Southern Africa. (ed. Werger, M.J.A.). - Monogr. Biol. 31: 515-560. Lange, H. (1985): Kilimanjaro. The white roof of Africa. — Leipzig, 175 pp. Lawrence, R:E. (1962): Mission zoologique de IER eS A. ©. sen Adaline orientale ( P. Basilevsky et N. Leleup, 1957). LX XIV. - Opiliones. — Ann. Mus. Roy. Afr. Centr., 8, Zool. 110: 9-89. Lépiney, J. de (1939): Solifuges et Opilionides du Maroc determines par le Dr. C.-Fr. Roewer. — Bull. Soc. Sci. nat. Maroc 19: 116-117. Lessert, R. de. (1915-1926): Araignées du Kilimanjaro et du Mérou. — Rev. suisse 2001. 23.27.28 3133. Mahnert, V. (1981): Die Pseudoskorpione (Arachnida) Kenyas I. Neobisiidae und Ideoroncidae. — Rev. suisse Zool. 88(2): 535-559. Mahnert, V. (1982a): Die Pseudoskorpione (Arachnida) Kenyas, IV. Garypidae. — Ann. Hist., Nat. Mus. Nat. Hungarici 74: 307-329. Mahnert, V. (1982b): Die Pseudoskorpione (Arahnida) Kenyas V. Chernetidae. — Rev. suisse Zool. 89(3): 691-712. Mahnert, V. (1983): Die Pseudoskorpione Kenyas VI. Dithidae (Arachnida). — Rev. Zool: afr. 97(1): 141-157. Mahnert, V. (1985): Die Pseudoskorpione (Arachnida) Kenyas. VIII. Chthoniidae. — Rev. suisse Zool. 92(4): 823-843. Mahnert, V. (1988): Die Pseudoskorpione (Arachnida) Kenyas. Familien Withiidae und Cheliferidae. — Trop. Zool. 1: 39-89. Mahunka, S. (1982a): Äthiopische Tarsonemiden (Acari: Tarsonemida), I. — Acta Zool. Acad. Sci. Hung. 21 (3-4): 369-410. Mahunka, S. (1982b): Oribatids from the eastern part of the Ethiopian region. I. — Acta Zool. Acad. Sci. Hung. 28 (3-4): 293-336. Mahunka, S. (1983a): Oribatids from the eastern part of the Ethiopian region. II. — Acta Zool. Acad. Sci. Hung. 29 (1-3): 151-180. Mahunka, S. (1983b): Oribatids from the eastern part of the Ethiopian region (Acari). HI. - Acta Zool. Acad. Sci. Hung. 29 (4): 397-440. Mahunka, S. (1984a): Oribatids of the eastern part of the Ethiopian region (Acari). V. — Acta Zool. Hung. 30 (1-2): 87-136. Mahunka, S. (1984b): Oribatids of the eastern part of the Ethiopian region (Acari) VI. — Acta Zool. Hung. 30 (3-4): 393-444. Mani, M.S. (1968): Ecology and biogeography of high altitude insects. — Dr. W. Junk Publ., The Hague, Ser. Entomologica 4: 527 p. Martens, J. (1993): Bodenlebende Arthropoda im zentralen Himalaya: Bestandsaufnahme, Wege zur Vielfalt und ökologische Nischen. — In: Neue Forschungen im Himalaya (ed. Schweinfurth, U.). — Erdkundl. Wiss. 112: 231- ASO). 181 Mauries. Ep & A. Heymer (1996): Nouveaux micropolydesmides d’ Afrique centrale: essai de rassemblement pour une révision du genre Sphaeroparia (Diplopoda, Polydesmida, Fuhrmannodesmidae). — Bull. Mus. nat. Hist. nat. Paris, 4e ser. 18, Sect. A (1-2): 165-184. Monod, Th. (1957): Les grandes divisions chorologiques de |’ Afrique. - C.S.A., Publ. 24, Londres, 147 p. Moreau, R.E. (1963): Vicissitudes of the African biomes in the late Pleistocene. — Proc. Zool. Soc. London 141: 395-421. Neumann, L.G. (1910): Ixodidae. — In: Wissenschaftliche Ergebnisse der Kilimanjaro-Meru Expedition. (ed. Sjöstedt, Y.) 3: 17-30. Nobili, G. (1906): Spedizione al Ruwenzori di S. A. R. Luigi Amedeo di Savoia Duca degli Abruzzi, IX. Crostacei. — Boll. Mus. Zool. Anat. Torino 21(544): 1-2. Paulian de Félice, L. (1941): Oniscoides de la cote occidentale d’ Afrique III. Eubelidae. — Bull. Soc. Zool. France 66: 50-59. Paulian de Félice, L. (1945a): Isopodes. — In: Mission scientifique de 1’Omo, VI, 57. Faune des terriers des Rats-taupes. - Mém. mus. nat. Hist. nat. Paris, N. SOG) 313. Paulian de Félice, L. (1945b): Isopodes terrestres. — In: Mission scientifique del’ Omo, VI, 60 (Zool.). - Mem. mus. nat. Hist. nat. Paris, N.S. 19(2): 331-344. Polis, G.A.(1990): Ecology. — pp. 247-293 in: The biology of scorpions (ed. Polis, G.A.). — Stanford Univ. Press, Stanford. Bed Week. c 1 7C) Cokendolpher (1995): Catalogue, bibliography and generic revision of the order Schizomida (Arachnida). — Texas Mem. Mus., Speleol. Monogr. 4: 1-170. Ribaut, H. (1914a): Myriapodes. I. Chilopodes. — In: Voyage de Ch. Alluaud et R.Jeannel en Afrique Orientale (1911-1912), Résultats scientifique: 1-35. Ribaut, H. (1914b): Myriapodes. II. Symphyla. — In: Voyage de Ch. Alluaud et R.Jeannel en Afrique Orientale (1911-1912), Résultats scientifique: 39-48. Roewer, C.F. (1913): Arachnides I. Opiliones. — In: Voyage de Ch. Alluaud et R.Jeannel en Afrique Orientale (1911-1912), Résultats scientifique 1: 1-22. Roewer, C.F. (1941): Expedition to South-West Arabia 1937-8. Vol. I, Nos 1-8. Arachnida: Opiliones. — Brit. Mus. (Nat. Hist.): 51-52. Roewer, C.F. (1952): Opilioniden und Solifugen aus Ost-Afrika. — Veroff. Mus. Nat.kde. Bremen 2: 87-90. | Borna @ Er (1956): Journey to Northern Ethiopia (Simien), 1952-3: Arachnida, Opiliones. — J. Linn. Soc. Zool., London 43: 93-95. ur Bass Smich, A oc Rk. Jocque (1986) New species and records of Linyphiidae (Araneae) from Kenya. — Rev. Zool. afr. 99: 439-453. Salt, G. (1954): A contribution to the ecology of Upper Kilimanjaro. — J. Ecol. 42: 375-423. 182 Scharff, N. (1992): The linyphiid fauna of Eastern Africa (Araneae: Linyphiidae) — distribution patterns, diversity and endemism. — Biol. J. Linn. Soc. 45: 117-154. Scheller, U. (1954): Symphyla from the Swedish East Africa expeditions in 1937-1938 and 1948, with a summary of the known African species. — Ark. Zool. 7(13): 227-240. Schmölzer, K. (1974): Landisopoden aus Zentral- und Ostafrica (Isopoda, Oniscoidea). — Sitzungsber. Österr. Akad. Wiss., Mathem.-naturw. K1., 1, 182(1- 5): 147-200. Scott, H. (1958): Biogeographical research in High Simien (northern Ethiopia), 1952-533. - Proc. Linn. Soc.) London 0G): E33 Silvestri, F. (1907): Spedizione al Ruwenzori di S.A.R. Luigi Amedeo di Savoia, duca degli Abruzzi. XXVI. Nuova specie de simfili Scutigerella ruwenzorii. — Boll. Mus. Zool. Anat. comp. Univ. Torino 22. Sjöstedt, Y. (1910): Wissenschattliche Ergebnisse, der schwedischen zoologischen Expedition nach dem Kilimanjaro, dem Meru und den umgebeden Massaisteppen Deutsch-Ostafrikas 1905-1906. — Stockholm, 3 vols. SGrensen, W. (1910): Arachnoidea. Opiliones. In: Wissensehantirehne Ergebnisse der Kilimanjaro — Meru Expedition (ed. Sjöstedt, Y.). Stockholm, 3: 59-82. Starega, W. (1984): Revision der Phalangiidae (Opiliones), III. Die afrikanischen Gattungen der Phalangiinae, nebst Katalog aller afrikanischen Arten der Familie. — Ann. Zool., Warszawa, 38 (1): 1-79. Swan, L.W. (1968): Alpine and aeolian regions of the world. — In: Arctic and Alpine environments (ed. Wright, H.E. Jr. & W.H. Osborn).— Indiana Univ. Press, Bloomington: 29-54. Taiti, S. & F. Ferrara (1980): The family Philosciidae (Crustacea: Oniscidea) in Africa, south of the Sahara. — Mon. zool. ital., N.S., Suppl. 4: 53-98. Taiti, S. & F. Ferrara (1987): Contributions to the knowledge of the montain fauna of Malawi 6. Terrestrial isopods (Crustacea). — Rev. Zool. Afr. 101 (1): 69- 102 Trägärdh, I. (1910): Acari. — In: Wissenschaftliche Ergebnisse der Kilimandjaro-Meru Expedition (ed. Sjöstedt, Y.). Stockholm, 3: 31-57. Troll, C. (1959): Die tropischen Gebirge. Ihre dreidimensionale klimatische und pflanzengeographische Zonierung. — Bonner Geogr. Abh. 25: 93 p. Troll, C. (1961): Klima und Pflanzenkleid der Erde in dreidimensionaler Sicht. — Naturwiss. 48(9): 332-348. Troll, C. (1975): Vergleichende Geographie der Hochgebirge der Erde in landschaftsökologischer Sicht. — Geogr. Rundsch. 5: 185-198. Tullgren, A. (1910): Araneae. — In: Wissenschaftliche Exgebuissem@en Kilimandjaro-Meru Expedition (ed. Sjöstedt, Y.). Stockholm, 3: 85-172. Vachon, M. (1945): Mission scientifique de l!’Omo, VI. 57. Faune des terriers des rats-taupes. X. Chernétes. -Mém. Mus. Nat. Hist. Nat., N.S. 19(1): 188-197. 183 Van Steenis, C.G.G.J. (1935): On the origin of Malaysian mountain flora II. Altitudinal zones, general considerations and renewed statement of the problem. — Bull. Jard. Bot. Buitenzorg, ser. 3, 13(3). Verhoeff, K.W. (1937): Uber einige Porcellio-Arten aus Marokko, gesammelt von Herrn J. de Lépiney. — Zool. Anz. 119: 302 - 307. Merhoeft K.W. (1938a): Uber Land-Isopoden aus Marokko. Il. Teil. — Zool. Anz. 124: 61-67. Verhoeff, K.W. (1938b): Myriapoden aus Marokko. Il. Teil (Gesammelt von BlercnnledeBepiney). - Zool. Anz. 123: 295-302. Wesolowska, W. (1986): A revision of the genus Heliophanus C.L.Koch, 1833 (Aranei: Salticidae). - Ann. Zool., Warszawa, 40(1): 1-254. White, F. (1978): The Afromontane region. — pp. 463-513 in: Biogeography and ecology of southern Africa (ed. Werger, M.J.A.) — Mon. Biol. 31, Dr W. Junk Publ., The Hague. Zacharda,M. & M. Daniel (1987): The first record of the family Rhagidiidae (Acarina: Prostigmata) from the Himalayan region. — Vest. (s. Spole). zool. 51: 58-59. 184 Appendix Isopoda, Arachnida and Myriapoda known to live in the Afrotropical oreal above 3500 m (in brackets the main authors for those groups) Isopoda (Barnard 1940, Budde-Lund 1910, Paulian de Félice 1941, 1945a,b, Ferrara & Taiti 1984, Taiti & Ferrara 1980, Nobili 1906, Scott 1958, Schmölzer 1974) Fam. Eubelidae Aethiopopactes chenzemae Ferrara et Taiti - 4600 m, Kilimanjaro Benechinus armatus Budde-Lund - 4600 m, Mt Meru Mesarmadillo chappuisi Paulian de Félice - 4000 m, Mt Elgon M. arambourgi Paulian de Félice - 4000 m Mt Elgon Hiallelgon jeanneli Paulian de Félice - 4000 m on Mt Elgon “Periscyphis” montanus Schmölzer - 3800 m, Mt Kenya Hiallum richardsoni Paulian de Félice - 3500 m, Mt Elgon Periscyphis ruficauda Budde-Lund - 3500 m, Mt Kenya Eubelum tachyorictidis Paulian de Félice - 3500 m, Mt Elgon Microcercus sp. - 3455 m, Ethiopia _ Fam. Philosciidae Afrophiloscia uncinata Ferrara - 3700 m, Kilimanjaro Fam. Porcellionidae “Porcellio” spatulata Barnard - 3600 m, Ethiopia Fam. Armadillidae “Synarmadillo” marmoratus Budde-Lund - 3500 m, Mt Meru Arachnida Scorpionida (Tullgren 1910) Fam. Buthidae Uroplectes fischeri Tullgren - 3500 m, Mt Meru Pseudoscorpionida (Beier 1944, 1951, 1955, 1959, 1962, Vachon 1945, Mahnert 1982a,b, 1983, 1985, 1988) Fam. Olpiidae Calocheiridius crassifemoratus Beier - 3650 m, Mt Elgon Fam. Atemnidae Titanatemnus palmquisti (Tullgren) - 4100 m, Kilimanjaro Fam. Chernetidae Nudochernes crassus Beier - 3700 m, Mt Elgon N. montanus Beier - 3500 m, Mt Elgon N. robustus Beier - 3500 m, Mt Elgon Fam. Withiidae Withius somalicus Beier - 3500 m, Mt Elgon Opilionida (Sorensen 1910, Goodnight & Goodnight 1959, Roewer 1913, 1941, 1952, 1956, Lawrence 1962, Kauri 1985) Laniatores Fam. Assamiidae Metaereca abnormis Roewer - 4000 m, Ruwenzori Hypoxestus accentuatus Sör. - 4600 m, Kilimanjaro H. holmi Goodnight et Goodnight - 4200 m, East Africa Hypoxestus mesoleucus Sor. - 3500 m, Kilimanjaro H. patellaris Sör. - 4000 m, Ruwenzori Eu Ereca undulata Sor. - 4025 m, Kilimanjaro E. simulator Sor. - 4000 m, Kilimanjaro E. maculata Roewer - 3975 m, Kilimanjaro E. lata Sor. - 3500 m, Kilimanjaro E. affinis Sor. - 3500 m, Kilimanjaro E. modesta Sor. - 3500 m, Kilimanjaro Randilea scabricula Roewer - 3630 m, Mt Elgon Simienatus scotti Roewer - 3505 m, Semien, Ethiopia Fam. Biantidae Metabiantes punctatus Sör. - 4000 m, Klimanjaro M. trifasciatus Roewer - 3600 m, Mt Meru M. convexus Roewer - 3500 m, Ruwenzori Palpatores Fam. Phalangiidae Odontobunus africanus Roewer - 3770 m, Kilimanjaro O. armatus (Sör.) - 4000 m, Kilimanjaro Rhampsinitus bettoni Pocock - 4600 m, Kilimanjaro Rh. discolor Karsch - 3870 m, Ruwenzori Rh. salti Roewer - 3800 m, Kilimanjaro Rh. soerenseni Starega (= Rh. pictus Sör.) - 3500 m, Mt Meru Rh. (?) mesomelas Sör. - 4000 m, Kilimanjaro Dacnopilio scopulatus Lawrence - 3600 m, Mt Meru Guruia africana Karsch - 4000 m, Kilimanjaro G. frigescens Loman - 4000 m, East Africa Cristina pachylomera Simon - 3870 m, Ruwenzori; 3658 m, Semien Araneida (Benoit 1978, Berland 1914, 1920, Bosmans 1977, 1978, 1979, 1981a,b, Bosmans & Jocqué 1983, Denis 1950, 1962, Fage 1940, Holm 1962, Jocqué 1981, Lessert 1915-1926, Russel-Smith & Jocqué 1986, Scharff 1992, Tullgren 1910, Wesolowska 1986) Fam. Zodariidae Mallinella vittiventris Strand - 3500 m, Karisimbi Fam. Linyphiidae Araeoncus picturatus Holm - 4650 m, Kilimanjaro A. subniger Holm - 4400 m, Mt Kenya A. impolitus Holm - 3900 m, Aberdare Asthenargus expallidus Holm - 3550 m, Aberdare A. inermis Simon et Fage - 3480 m, Mt Kenya A. marginatus Holm - 3450 m, Ruwenzori Callitrichia ruwenzoriensis Holm - 4930 m, Ruwenzori C. kenyae Fage - 4530 m, Mt Kenya C. glabriceps Holm - 4200 m, Mt Elgon C. aliena Holm - 4150 m, Mt Elgon C. paludicola Holm - 3975 m, Kilimanjaro C. hamifer Holm - 3800 m, Mt Elgon C. monticola Tullgren - 3500 m, Kilimanjaro Ceratinopsis fako Bosmans et Jocqué - 4000 m, Fako, Cameroon Erigone aethiopica Tullgren - 4200 m, Mt Kenya; 4000 m (Cameroon) Lepthyphantes kilimanjaricus Tullgren - 4000 m, Kilimanjaro L. ruwenzori Jocqué - 3800 m, Ruwenzori Meioneta obscura Denis - 4723 m, Ruwenzori Microcyba erecta Holm - 4300 m, Ruwenzori M. hamata Holm - 4300 Mt Elgon M. annulata Holm - 4000 m, Mt Elgon Microcyba brevidentata Holm - 3900 m, Kilimanjaro M. projecta Holm - 3800 m, Ruwenzori M. hedbergi Holm - 3730 m, Muhavura M. falcata Holm - 3450 m, Ruwenzori 186 Neriene kibonotensis Tullgren - 3500 m, Kilimanjaro Oreocyba propinqua Holm - 4300 m, Mt Elgon O. elgonensis Fage - 4200 m, Mt Elgon Pelecopsis ruwenzoriensis Holm - 4930 m, Ruwenzori P. biceps Holm - 4300 m, Kilimanjaro P. alticola Berland - 4165 m, Mt Elgon; 3850 m, Mt Kenya; 3400 m, Aberdara P. tenuipalpis Holm - 4000 m, Ruwenzori P. infusca Holm - 4000 m, Ruwenzori P. senecicola Holm - 3930 m, Ruwenzori P. pasteuri Berland - 3800 m, Kilimanjaro P. reclinata Holm - 3760 m, Mt Elgon P. varians Holm - 3450 m, Mt Elgon P. flava Holm - 3450 m, Ruwenzori Toschia telekii Holm - 4300 m, Mt Kenya T. aberdarensis Holm - 3750 m, Aberdare Tybaertiella kruegeri Simon - 3750 m, Karisimbi Walckenaeria meruensis Tullgren - 3820 m, Kilimanjaro; 3280 m, Mt Meru W. aberdarensis Holm - 3550 m, Aberdare, Mt Kenya W. ruwenzorensis Holm - 3450 m, Ruwenzori Fam. Hahniidae Hahnia tabulicola Simon - 3900 m, Cameroon H. major Benoit - 3750 m, Mt Kenya H. sirimoni Benoit - 3750 m, Mt Kenya Fam. Lycosidae Pardosa alticola Alderweiteldt et Jocqué - 3700 m, Rwanda; 3650 m, Semien Fam. Salticidae Heliophanes crudeni Lessert - 4650 m, Kilimanjaro H. gladiator Wesolowska - 4450 m, Mt. Kenya H. imperator Wesolowska - 4200 m, Mt Kenya H. kenyaensis Wesolowska - 3650 m, Mt Elgon Acariformes (Andre 1936, 1938, 1945, 1957, 1965, Balogh 1966, Evans 1953, Mahunka 1982a, b, 1983a, b, 1984a, b, Trägärdh 1910) Acaridida Fam. Glycyphagidae Glycyphagus domesticus (De Geer) - 3500 m, Mt Elgon Fam. Anoetidae Histiostoma telatum Mahunka - 3450 m, Kilimanjaro Prostigmata Fam. Bdellidae Bdella piggotti Evans - 3810 m, Kilimanjaro Fam. Anystidae Anystis baccarum L. - 3810 m, Kilimanjaro Fam. Erythraeidae Charletonia areolata Trägärdh - 3810 m, Kilimanjaro Fam. Trombidiidae Trombidium bipectinatum Trägärdh - 3800 m, Mt Meru T. tinctorium L. - 3800 m, Mt Meru Trombidium meruense Trägärdh - 3800 m, Mt Meru T. simile Trägärdh - 3800 m, Mt Meru Dinothrombium trispilum Berlese - 4200 m, Ruwenzori D. tarsale Berlese - 4000 m, Mt Elgon Allothrombium pergrande Berlese - 4000 m, Kilimanjaro Fam. Microtrombidiidae Enemothrombium bipapillatum Berlese - 3500 m, Mt Elgon E. carduigerum Berlese - 3500 m, Mt Elgon E. jeanneli André - 3500 m, Mt Elgon Oribatida Fam. Brachychthoniidae Liochthonius tanzanicus Mahunka - 3890 m, Kilimanjaro Fam. Haplozetidae Protoribates shiraensis Evans - 4590 m, Kilimanjaro Fam. Euphthiracaridae Rhysotritia ardua (C.L. Koch) - 3800 m, Kilimanjaro Fam. Microtegeidae Microtegeus undulatus Berlese - 4285 m, Kilimanjaro Fam. Ceratozetidae Africoribates ornatus Evans - 4590 m, Kilimanjaro Ghilarovizetes africanus Mahunka - 3900 m, Kilimanjaro Fam. Tectocepheidae Tectocepheus spinosus Mahunka - 3890 m, Kilimanjaro Fam. Scutoverticidae Scutovertex africanus Evans - 4438 m, Kilimanjaro Fam. Dampfiellidae Dampfiella setosa Mahunka - 3890 m, Kilimanjaro Fam. Oppiidae Oppia nasalis Evans - 4285 m, Kilimanjaro O. africanus Evans - 4285 m, Kilimanjaro Amerioppia foveolata Mahunka - 3820 m, Kilimanjaro Fam. Quadroppiidae Quadroppia crenata Mahunka - 3820 m, Kilimanjaro Fam. Oribatulidae ? Incabates longisacculus Mahunka - 3820 m, Kilimanjaro Nannerlia elongatissima Mahunka - 3890 m, Kilimanjaro Zygoribatula setosa (Evans) - 3810 m, Kilimanjaro Fam. Scheloribatidae Scheloribates laevigatus C.L. Koch - ? 4590 m, Kilimanjaro Fam. Oribatidae Oribates geniculatus (L.) - 3810 m, Kilimanjaro Parasitiformes (Andre 1938, Neumann 1910) Gamasida Fam. Macrochelidae Macrocheles elgonensis Andre - 3500 m, Mt Elgon Fam. Hypoaspididae Hypoaspis praesternalis Willmann - Kilimanjaro Ixodida Fam. Ixodidae Ixodes ugandanus djaronensis Neumann - 3500 m, Mt Meru I. rasus Neumann - 3500 m, Mt Meru Rhipicephalus simus C.L.Koch - 3500 m, Mt Meru 187 188 Symphyla (Ribaut 1914b, Silvestri 1907, Scheller 1954) Fam. Scutigerellidae Hanseniella ruwenzorii Silvestri - 4500 m, Ruwenzori, 3650 m, Mt Kenya H. pillipes Attems - 4200 m, Ruwenzori, 4000 m, Mt Elgon H. afromontana Scheller - 4000 m, Ruwenzori H. producta Ribaut - 3650 m, Mt Kenya Chilopoda (Attems 1937, 1939, Ribaut 1914a) Geophilomorpha Fam. Mecistocephalidae Mecistocephalus insularis Lucas - 3900 m, Nyiragongo M. (sub “Lamnonyx’) punctifrons Newprt - 3500 m, Mt Elgon Lithobiomorpha Fam. Henicopidae Lamyctes africana Poc. - 4200 m, Ruwenzori L. fulvicornis Meinert - 4000 m, Mt Kenya Scolopendromorpha Fam. Scolopendridae Scolopendra afra Mein. - 4000 m, Mt Elgon Fam. Cryptopidae Cryptops numidicus tropicus Attems - 3500 m, Mt Meru C. incerta Attems - 3500 m, Mt Elgon C. bottegoi kenyae Rib. - 3500 m, Elgon Lamnonyx punctifrons Newport - 3500 m, Mt Elgon Diplopoda (Attems 1909, 1939, Brölemann 1920, Mauries & Heymer 1996) Stemmiulida Fam. Stemmiulidae Stemmiulus sjostedti Attems - 3500 m, Mt Meru Polydesmida Fam. Fuhrmannodesmidae Elgonicola jeanneli Attems - 3500 m, Mt Elgon E. jeanneli microchaeta Attems - 4000 m, Mt Elgon Sphaeroparia petarberoni Mauries et Heymer - 4200 m, Ruwenzori S. minuta Attems - 3500 m, Mt Meru Author’s address: Dr Petar Beron National Museum of Natural History 1, Tsar Osvoboditel Blvd Sofia 1000, Bulgaria 189 Two sympatric Lygodactylus species in coastal areas of Eastern Africa (Reptilia, Gekkonidae) Beate Roll Abstract: The genus Lygodactylus (Reptilia, Gekkonidae) comprises about sixty species of relatively small, diurnal geckos living mainly in Africa and Madagascar. A large group within the genus is the ‘Lygodactylus picturatus’ complex whose taxa are either described as subspecies or are treated as individual species depending on authors. Two taxa of this complex occur sympatrically without interbreeding on the coast of Kenya between Mombasa and the Tanzanian border. Whereas these taxa differ only slightly in size and other morphological features like scalation pattern, there are definite differences in the colour patterns and other biological features. L. picturatus is characterized by a distinct sexual dichromatism and strong physiological and ontogenetic colour change; in contrast, L. mombasicus has only slight sexual dichromatism, only weak physiological and no ontogenetic colour change. The somewhat complicated nomenclature of the species concerned is also discussed. Key words: East Africa, Reptilia, Gekkonidae, Lygodactylus, sympatric species Introduction Geckos are typically small, agile lizards, and they are found in tropical and subtropical regions of all continents. Most geckos are nocturnal. However, members of 14 out of approximately 90 genera are diurnal, and between the categories ‘nocturnal’ and ‘diurnal’ there are several transitions. Ophthalmological evidence indicates that diurnal geckos have re-acquired diurnal habits from nocturnal gekkonid ancestors (Walls 1942, Röll 1996). Diurnal geckos are restricted to New Zealand, Central and South America, and Africa, the Middle East and the Indian Ocean Islands. One strictly diurnal genus 1s Lygodactylus, which comprises numerous species of relatively small geckos living in Africa (from about 15° N to 30° S) and Madagascar. A large and difficult group within the genus is the ‘Lygodactylus picturatus’ complex, with many taxa in West, Central and East Africa. Depending on authors, these taxa are either described as subspecies of L. picturatus (e.g. Loveridge 1947) or are treated as individual species (e.g. Pasteur 1965). Many of these taxa are indeed conspecific, despite recognizable differences in size and colour patterns. In Kenya, animals from the coast (L. p. picturatus) are able to produce viable offspring with animals from, e.g. Mtito Andei (about 250 km northwest of Mombasa) and from Olorgesailie (in the Rift Valley, about 60 km southwest of Nairobi) (Röll 1994 and unpubl. observ.). Thus these populations have to be treated as subspecies. As animals from Olorgesailie come close to L. p. keniensis Parker, 1936 they are parts of a cline, which may well extend at least to ‘typical’ L. p. keniensis from the northern Rift Valley or even beyond. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 190 On the other hand, there are two Lygodactylus taxa on the Kenyan coast that are easily distinguished phenologically and that occur sympatrically, even sharing one tree. They do apparently not interbreed, at least there are no intermediates. This paper deals with the ecology of these two taxa and with some nomenclatorial problems around them. Materials and methods Field observations Geckos of the genus Lygodactylus were studied in February 1992 and March 1997 in Kenya, from the Rift Valley (Lake Bogoria) to the coast. The two taxa concerned were observed primarily in Tiwi Beaches on the coast between Mombasa and the Tanzanian border (4°13'S, 39°36'E). Animals were observed with binoculars for several days. They were common on trees (e.g. coconut palms, but more often other trees, notably those with several trunks) and on walls of buildings. Breeding Geckos were kept and bred under laboratory conditions for several years. They were housed in glass terraria containing a layer of sand and various plants, branches, and pieces of bark as hiding places. A 12-hour light/dark cycle was maintained. The temperature varied from 28 to 34 °C during the day and from 23 to 25°C during the night. The geckos were fed a variety of insects and pieces of banana or mashed fruit. Drinking water was enriched with mineral preparations and vitamins. Eggs were incubated either in the terraria or in a special incubating chamber with thermostat, and the young were reared individually in smaller cages. Morphological characters Specimens preserved in 70 % ethanol were investigated from different localities on the East African coast (Kenya: Kikambala, Tiwi Beaches, Diani Beach; Tanzanian coast). For comparison with results by Loveridge (1947), the same morphological characters, 1.e. scalation of upper and lower jaw, number of preanal pores, number of pairs of adhesive lamellae and, additionally, number of pairs of adhesive lamellae under the tail tip, were examined. These characters were also checked in living animals. Results For description of the two taxa of Lygodactylus treated here, the terms L. p. picturatus and L. ‘p.’ mombasicus are used sensu Loveridge (1947). Colour patterns L. p. picturatus is characterized by a distinct sexual dichromatism, which in this taxon attains its greatest development within the genus (Loveridge 1947). Head, neck and shoulders of the territorial male are bright mustard yellow with dark spots or broken streaks (plate 4: a,c, page 192). Back, limbs and tail are bluish-grey, sometimes with light eyespots. Pale lateral stripes extend from the shoulders to the 19] base of the tail. Breast and belly are white and yellowish orange respectively. The throat is uniformly black (plate 4: d). Males are capable of changing coloration very rapidly, e.g. after disturbance. Within seconds they can become very dark, almost black, without any trace of yellow. Only males have preanal pores. Head and shoulders of the ‘well-poised’ female are light yellow with darker spots or streaks (plate 4: b). Back, limbs and tail are light brown to beige-grey with pale lateral stripes extending from the shoulder to the beginning of the tail. Breast and belly are white and whitish yellow respectively. The throat is either uniformly white or marked with a more-or-less faint, mottled greyish chevron-shaped pattern (plate 4: e). Similar to the male, this coloration can change to brown or dull grey. In both sexes this physiological colour change is confined to the dorsal side of the animals. Hatchlings have a colour pattern similar to that of females but with the head brownish grey, not yellow (plate 4: f). They are able to change their general colour from brownish grey to dark brown. Male offspring develop the typical coloration at an age of about 6 to 8 months. In contrast, L. ‘p.’ mombasicus has no sexual dichromatism (apart from the colour pattern of the throat) (fig.la,b). Both males and females are characterized by a whitish or cream head and neck with clear-cut black markings: black lateral streaks from tip of the nose through the eyes to the shoulders, between the eyes a marking resembling a crown or the letter *W’, on the occiput two fused blotches, on the neck two fused spots which unite with the lateral streak, and above the forelimbs another pair of spots extending posteriorly (fig.lc). Breast and belly are white and yellow respectively. The sexes can be distinguished by the preanal pores occurring only in males and by the colouring of the throat. The throat of males is almost uniformly black (fig.ld), whereas that of females is white with a dark chevron-shaped marking (fig.le). Hatchlings have the same coloration as adults (fig.1f). Males, females and hatchlings are capable of changing from a brighter coloration to a duller one, but the typical markings on their heads are always clearly recognizable. In specimens of both taxa preserved in 70% ethanol, the yellow colour of the belly is destroyed probably because of leaching of pigments into the fixative, whereas the yellow of the head is retained. Morphological characters The scalation patterns of the head do not differ in the two taxa of Lygodactylus described here (data are summarized in table 1). This is in correspondence with results by Loveridge (1947). Additionally, both taxa possess 6 pairs of adhesive lamellae under the fourth toe. Undamaged tail tips of L. p. picturatus and L. ‘p.’ mombasicus are provided with 6 to 8 and 7 to 9 pairs of adhesive lamellae (scansors) respectively. The number of the preanal pores in males was found to be 6 to 8 in L. p. picturatus and 10 to 11 in L. ‘p.’ mombasicus. Adult animals of both taxa are of similar size. Generally, males are slightly larger than females. Eggs of both Lygodactylus taxa are indistiguishable. Fig.1: Lygodactylus ‘p.’ mombasicus Loveridge, 1935 a) Adult male: head, neck and shoulders with typical black markings, body beige-grey. b) Adult female: colouring as in the male. c) Dorsal view of the head of a female. d) Black throat of a male. e) Throat of a female with a darker chevron-shaped marking. f) Hatchling, two days old: same colour pattern as the adults. Scale bars I cm. All photographs are of live specimens. Plate 4: Lygodactylus p. picturatus (Peters, 1870) a) Adult male, b) adult female, c) dorsal view of the head of a male. d) Black throat of a male. e) Throat of a female with a faint greyish chevron-shaped marking. f) Hatchling, five days old: juvenile pattern resembling the colour pattern of the female. Scale bars 1 cm. All photographs are of live specimens. ee nn mn m He FL 5 x ae Au CAS = = a a re . 5 = A . “2 i 5 i 4 TT 7 : u 4 : 0 iy 7 | 1 i i ee i : x ; \ 2 f al 4 A x ‘ i idee i i f ot i x | i i hes ö z Ta i bit 5 Fe ES 1 f ¢ 5 va i 08 N \ o hu 56 s + Y rears . { Ak . e Ss \ 2 we 193 Table I: Morphological characters of Lygodactylus p. picturatus and L. ‘p.’ mombasicus Lygodactylus p. picturatus Lygodactylus ‘p.’ mombasicus data from Loveridge 1947 data from Loveridge 1947 own data own data internasal granules nasals surrounding nostril rostril enters nostril upper labials lower labials postmentals scansors under fourth toe scansors under tail tip male (preanal) pores snout to vent length “ (mm) snout to vent length ? (mm) snout to vent length of hatchlings diameter of eggs (mm) no data no data no data no data Field observations Tiwi Beaches is a ‘standard’ East African coastal area. L. p. picturatus inhabits trees on the beach and in gardens, as well as walls of buildings. These geckos live in small groups, usually consisting of one male and one or two females. On palm trees, usually only one pair of L. p. picturatus was found. Big trees with several trunks and branches may be inhabited by more than one group. L. ‘p.’ mombasicus, also strictly arboreal and living in groups with one male, was not observed on palm trees and buildings but only on bigger trees on the beach and in gardens. Here, groups of L. ‘p.’ mombasicus and L. p. picturatus can coexist on the same tree without apparent interference. The typical intraspecific threat display (curved, Table 2: Biological characters of Lygodactylus p. picturatus and L. ‘p.’ mombasicus. Lygodactylus p. picturatus Lygodactylus ‘p.’ mombasicus — cd’: head bright yellow with irregular — cd‘: head whitish with characteristic, dark spots or broken streaks; body bluish well-defined dark markings; body beige- grey; throat uniformly black grey; throat uniformly black — distinct sexual dichromatism: ? head — sexual dichromatism slight (in ? throat yellowish brown with dark spots or with dark chevron only) streaks; body beige greyish; throat white or weakly marbled — strong physiological colour change — physiological colour change weak — marked ontogenetic colour change, — no ontogenetic colour change especially in do’ sympatric occurrence of both taxa without interbreeding on Kenyan coast 194 arched, laterally flattened body) of Lygodactylus males was not observed between males of different taxa. Only animals with the typical colour pattern of either L. p. picturatus or L. p. mombasicus were found both at Tiwi Beaches and Diani Beach. There were no intermediates between the two distinct colour patterns. Discussion Two taxa of the genus Lygodactylus occur sympatrically at the coast between Mombasa and the Tanzanian border. They were treated ne Loveridge (1947) as L. picturatus picturatus and L. ‘p.’ mombasicus. Whereas these taxa differ only slightly in their sizes and other morphological features (like scalation pattern), there are definite differences in their colour patterns and other biological features (tables 1+2). L. p. picturatus 1s characterized by a distinct sexual dichromatism and strong physiological and ontogenetic colour change; in contrast, L. ‘p.’ mombasicus has only slight sexual dichromatism, only weak physiological and no ontogenetic colour change. Both Lygodactylus taxa are strictly arboreal and live in small groups, usually consisting of one male with one or a few females. Bigger trees with multiple trunks can be inhabited by several groups of one or both taxa. While there certainly are olfactory elements, the bulk of the social behaviour of Lygodactylus is visually guided. Males of both taxa occupy and patrol territories that are defended by visual threat displays and fights. Aggressive displays are commonly directed towards conspecific males and females. In contrast, there seems to be little or no interference between members of different taxa. It is remarkable that two Lygodactylus taxa with such similar ecological requirements can coexist in the same area, since interspecific competition should be expected. Although it is not uncommon to find both taxa on the same tree, more often each taxon inhabits its own tree. Interspecific competition may be reduced by partly partitioning the habitat, e.g. by inhabiting different vegetation types. L. p. picturatus often uses palm trees and buildings where L. ‘p.’ mombasicus was not found. The two taxa meet on multi-stemmed trees where the different trunks could be ‘divided’ between the taxa. However, it was also observed that both taxa shared a single-stemmed tree. In the South Turkana region of Kenya, Greer (1967) observed a very similar situation between two Lygodactylus species (L. picturatus keniensis Parker, 1936 and L. somalicus battersbyi Pasteur, 1962). Although being ecologically very similar, these species occur sympatrically in this area, where they inhabit Acacia trees. In general, one species occupies one tree, but sometimes both species occur on the same tree, usually a larger one. L. picturatus keniensis and L. somalicus battersbyi differ conspicuously in their colour patterns, as do L. p. picturatus and L. ‘p.’ mombasicus, and Greer also never observed the typical display behaviour interspecifically. In that case, however, there was a clearly dominant taxon: confrontations between the two species resulted in the considerably smaller L. somalicus battersbyi being chased away. 195 Another example is the sympatric occurrence of L. chobiensis and L. capensis in Zambian woodlands (Simbotwe 1983). These species, too, are ecologically very similar. They reduce competition by partitioning their habitat. The larger L. chobiensis was more often found on live trees, whereas L. capensis was more commonly found on dead ones. Additionally, L. chobiensis occupies on average higher perches than L. capensis. This could be confirmed in Popa Falls (Caprivi Strip, Namibia), although it was not unusual there to find both species on the same live tree (Röll 1999). In contrast to these cases local variants of L. picturatus e.g. the populations found in Mtito Andei and Olorgesailie are not separated ecologically. Apparently they are elements of a continuous cline. Taxonomical and nomenclatorial problems As L. p. picturatus and L. ‘p’. mombasicus occur sympatrically without interbreeding, and show distinct differences in male/female colouring and in physiological and ontogenetic colour change, they should consequently be treated as distinct species. However, the nomenclature of this group is somewhat complicated. Peters described L. picturatus (as Hemidactylus picturatus) in 1870. The use of the name L. picturatus for the ‘yellow-headed’ taxon was the common usage during the following hundred years (e.g. Boulenger 1885, Parker 1942, Pasteur 1960, Wermuth 1965, Howell 1981, Welch 1982). In 1935, Loveridge described the ‘black-and-white-headed’ taxon as subspecies L. p. mombasicus. In a revision of the genus Lygodactylus, Pasteur (1965) states that ‘to his knowledge’ the holotype of L. picturatus is atypical L. ‘p.’ mombasicus as defined by Loveridge (1947). Though this statement would upset the widely used nomenclature of the whole group, it is not substantiated further. It remains unclear whether Pasteur actually saw the types; anyway, the otherwise rather vague original description by Peters (1869, sic!) states that the male has a yellowish head, and does not mention any bold black markings. This question has to be left open for the time being. Consequently, Pasteur treated L. picturatus mombasicus as a synomym of L. picturatus and described the ‘yellow-headed’ dwarf gecko as anew species, L. luteopicturatus. However, he did apparently not check whether any of the numerous older names for this taxon were nomenclatorially available (see e.g. the synonymic list in Loveridge 1947). Since then, the name L. luteopicturatus has been used by only a few authors for the ‘yellow-headed’ dwarf gecko (e.g. Pakenham 1983, Broadley & Howell 1991), while the name L. picturatus was equally rarely used for the ‘black-and-white-headed’ dwarf gecko (e.g. Broadley & Howell 1991). In a recent checklist of the geckos of the world, all names for the ‘yellow-headed’ and the ‘black-and-white-headed’ dwarf geckos are listed simultaneously (Welch 1994). So, there are two general lines of argument. If one follows Loveridge’s (1947) definitions, the ‘yellow-headed’ taxon has to be named L. picturatus, while the ‘black-and-white-headed’ taxon is definitely L. mombasicus. In this case, Pasteur’s L. luteopicturatus (1965) is merely asynomym of L. picturatus. 196 If one were to accept Pasteur’s (1965) statement about the identity of the holotype of L. picturatus, the situation is more complicated. The ‘black-and-white- headed’ taxon would simply become L. picturatus. The ‘yellow-headed’ taxon, however, has been described several times since Peters’s original description of L. picturatus. It is rather probable that there are available names for this taxon, e.g. Tornier’s (1896) ‘variations’ (perhaps ‘modern’ subspecies; cf. Kraus 1973) griseus, septemlineatus or quinquelineatus, Boettger’s quinque-lineata (1913) or Loveridge’s manni (1928). If any of these 1s indeed available, /uteopicturatus again becomes a synomym. For these reasons, and to retain stability of a name in common usage, I suggest rejecting Pasteur’s questionable proposal until definitely substantiated. This will require examining the type material of the various taxa mentioned in the previous paragraph. This study is currently undertaken. Until then, the widely accepted name L. picturatus (Peters, 1870) for the ‘yellow-headed’ taxon should be retained, as e.g. faunistic appers can suffer substantially from an unclear nomenclatorial situation. Consequently, the -“"black-and-white-headed’ taxon would have to be treated as L. mombasicus Loveridge, 1935, as it has been shown. hereinziosbe specifically distinct from L. picturatus. Zusammenfassung Die Gattung Lygodactylus (Reptilia, Gekkonidae) umfaßt ca. 60 Arten relativ kleiner, tagaktıver Geckos, die hauptsächlich in Afrıka und Madagaskar leben. Eine große Gruppe innerhalb der Gattung ist der “‘Lygodactylus picturatus’-Komplex, dessen Taxa je nach Autor entweder als Subspecies oder als individuelle Arten beschrieben werden. Zwei Formen dieses Komplexes kommen sympatrisch und reproduktiv isoliert an der Küste Kenias zwischen Mombasa und der Grenze Tanzanias vor. Die beiden Taxa unterscheiden sich nur geringfügig in der Größe wie auch in anderen morphologischen Mermalen, z.B. in der Beschuppung. Dagegen gibt es deutliche Unterschiede in ihren Farbmustern und anderen biologischen Merkmalen. So ist L. picturatus charakterisiert durch auffälligen Geschlechtsdichromatismus und durch ausgeprägten physiologischen sowie ontogenetischen Farbwechsel. Im Gegensatz dazu weist L. mombasicus einen nur gering entwickelten Geschlechtsdichromatismus und einen schwach entwickelten physiologischen sowie keinen ontogenetischen Farbwechsel auf. Die Nomenklatur der betreffenden Arten wird diskutiert. References Boettger, ©. (1913): Reptilien und Amphibien von Madagaskar, den Inseln und dem Festland Ostafrikas. — pp.269-376, pl. XXIII-XXX in Reise in Ostafrika (ed. Voeltzkow, A.). Vol.3. — Stuttgart Boulenger, G.A. (1885): Catalogue of the lizards in the British Museum (Natural History). - London 2nd ed. Vol. 1: 1-497. Broadley, D.G. & K.M. Howell (1991): A check list of the repeileszor Tanzania, with synoptic keys. — Syntarsus 1: 1-70. 197 Greer, A.E. (1967): The ecology and behavior of two sympatric Lygodactylus geckos. — Breviora 268: 1-19. Howell, K.M. (1981): Notes on two diurnal geckos, Lygodactylus picturatus and Phelsuma dubia. — East Afr. Nat. Hist. Soc. Bull. 3: 43-45. Kraus, O. (1973): Internationale Regeln ftir die Zoologische Nomenklatur: Bericht über Änderungen, gültig ab 1. Januar 1973. — Senckenbergiana biol. 54: 219-225. Loveridge, A. (1928): Description of a new species of gecko from Tanganyika Territory. — Proc. U.S. Nat. Mus. 72 (Art.24): 1-2. Loveridge, A. (1935): New geckos of the genus Lygodactylus from Somaliland, Sudan, Kenya, and Tanganyika. — Proc. Biol. Soc. Washington 48: 195-200. mocnnases 9. (1947): Revision of the African lizards of the family Gekkonidae. — Bull. Mus. Comp. Zool. 98: 1-469. Pakenham, R.H.W. (1983): The reptiles and amphibians of Zanzibar and Pemba Islands (with a note on the freshwater fishes). — J. East Afr. Nat. Hist. Soc. Nat. Mus. 177: 1-40. Parker, H.W. (1942): The lizards of British Somaliland. — Bull. Mus. Comp. Zool. Harvard 91:1-101. Pasteur, G. (1960): Notes préliminaires sur les Lygodactyles (Gekkonides). I. Remarques sur les sous-especes de Lygodactylus picturatus. — Bull. I. F. A. N., Ser. A, 22: 1441-1452. Pasteur, G. (1962): Notes préliminaires sur les lygodactyles (Gekkonides). II. Diagnose de quelque Lygodactylus d Afrique. — Bull. I. F. A. N., Ser. A, 24: 606-614. Pasteur, G. (1965): Recherches sur l’Evolution des lygodactyles, lézards afro- malgaches actuels. — Trav. Inst. Scient. Chérif., Sér. Zool., 29: 1-132. Peters, W. (1869): Mitteilung über eine neue Nagergattung, Chiropodomys penicillatus, so wie tiber einige neue oder weniger bekannte Amphibien und Fische. — Monatsber. Akad. Wiss. Berlin: 448-461. Peters, W. (1870): Beitrag zur Kenntnis der herpetologischen Fauna von Südafrika. - Monatsber. Akad. Wiss. Berlin: 110-115. Röll, B. (1994): Lygodactylus picturatus (Peters). — Sauria 16 (3) Suppl.: 307-310. Röll, B. (1999): Lygodactylus chobiensis (FitzSimons). — Sauria 21(3) Suppl.: 457-460. Röll,B., R. Amons & W.W. de Jong (1996): Vitamin A, bound cellular retinal-binding protein as ultraviolet filter in the eye of the gecko Lygodactylus picturatus. — J. Biol. Chem. 271: 10437-10440. momotwe. MP. (1983): Comparative ecology of diurnal geckos (Lygodactylus) in Kafue flats, Zambia. — Afr. J. Ecol. 21: 143-153. Tornier, G. (1896): Die Kriechtiere Deutsch-Ost-Afrikas. — Berlin. Walls, G. (1942): The vertebrate eye and ist adaptive radiation. — Cranbrook Institute of Science, Bloomfield Hills. Welch, K.R.G. (1982): Herpetology of Africa: A checklist and bibliography of the orders Amphisbaenia, Sauria and Serpentes. — Robert E. Krieger Publishing Company, Malabar, Florida. 198 Welch, K.R.G. (1994): Lizards of the world: a checklist. 1. Geckos. — Longdunn Press Ltd, Bristol, England. Wermuth, H. (1965): Liste der rezenten Amphibien und Reptilien: Gekkonidae, Pygopodidae, Xantusiidae. — Walter de Gruyter & Co., Berlin. Author’s address: Dr. Beate ROI, Lehrstuhl für Allgemeine Zoologie und Neurobiologie, Fakultät für Biologie, Ruhr-Universität Bochum, 44780 Bochum, Germany 199 Comparative ecology and morphology of snipes (family Scolopacidae) in Africa ee VM Giehukr& Nathan N. Giehukı Abstract: Six of the 19 existing snipe species occur in Africa, four as Palearctic migrants and two as resident breeding birds. The resident species are the continental African Snipe Gallinago nigripennis and Madagascar Snipe Gallinago macrodactyla which is endemic to the island of Madagascar. The Palaearctic migrant snipes which winter in Africa (i. e. Jack Snipe Lymnocryptes minimus, Common Snipe Gallinago gallinago, Great Snipe Gallinago media and Pintail Snipe Gallinago stenura) show some eco-morphological similarities with the African Snipe but all exhibit distinct differences in their social behaviour and life-history patterns from the Madagascar Snipe. In this study we investigated the breeding biology of the African Snipe in the highlands of central Kenya and sought to explain the evolutionary mechanisms that would have produced species segregation, variation in social behaviour and life-history patterns of snipes in Africa. Key words: Kenya, Africa, Gallinago nigripennis, snipes, evolution Introduction The ecology and morphology of a species provide special windows for investigating the process of evolution. This is mainly because variations that are produced through evolutionary processes and adaptation to different environments can be inferred or observed from changes in body structure, social behaviour, ecology and life-history patterns (Futyuma 1979). This study seeks to determine the ecological and morphological adaptations that have permitted the survival and spread of six species of snipes to different environments and geographical regions, especially in Africa. Much of this paper deals with a synthesis of the principles of ecology, morphology and social biology in as much as they assist in the understanding of evolutionary trends in snipes. Our synthesis draws from existing information and from our field studies of the African Snipe in Kenya. Taxonomy and distribution Snipes belong to a heterogeneous and cosmopolitan order (Charadriiformes) which contains small waders that inhabit shorelines of lakes and seas, banks of rivers, marshes, cultivated fields and short grasslands. The various species that constitute this diverse group of birds are ecologically isolated through differences in food, habitat and geographical region. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 200 Hayman et al. (1986) recognized 18 species, while Sibley & Monroe (1990) recognized 19 species of snipes occurring in different parts of the world. The continent of Africa has two resident species and four Palaearctic migrant species of snipes (Table I). The Madagascar Snipe (Gallinago macrodactyla) is endemic to the island of Madagascar while the African Snipe (G. nigripennis) only occurs on the continent. Table 1: The movement status and distribution of the various species of snipes in Africa Distribution Resident/breeding Somalia and Kenya Pintail Snipe Gallinago stenura African Snipe Gallinago nigripennis Migrant/non-breeding Central, East and S. Africa Madagascar Snipe | Gallinago macrodactyla | Resident/breeding Eastern Madagascar Great Snipe Gallinago media Resident/breeding All Africa Common Snipe Gallinago gallinago Migrant/non-breeding | All Africa Jack Snipe Lymnocryptes minimus | Migrant/non-breeding | East and West Africa The African Snipe and the Madagascar Snipe exhibit distinct distributions (Urban etal. 1986). The Madagascar Snipe is confined to the island of Madagascar and is therefore geographically isolated from other snipes that occur on the continent. This species exhibits distinct morphological and ecological differences from the continental snipes, reflecting its long period of geographical and genetic isolation. The African Snipe is resident and breeds in eastern and southern Africa. The species distribution in the continent shows three distinct sub-populations or races. The sub-population Gallinago nigripennis aequatorialis (Rueppell) occurs in the region between the Ethiopian highlands and the Democratic Republic of Congo, including East Africa south to Malawi and northern Mozambique. The Gallinago nigripennis angolensis (Bocage) sub-population occurs in Angola, Botswana, Namibia, western Zimbabwe, Zambia and south Katanga in Congo. The nominate subspecies Gallinago n. nigripennis (Bonaparte) inhabits South Africa and southern Mozambique. The Palearctic snipes (table I) occur widely in the continent but mainly within the tropics (Urban et al. 1986). The presence of these snipes in Africa enhances competition with resident snipes for the available food resources but has no significant impact on their reproductive success (Moreau 1972). In our central Kenya study site (Lake Ol Bolossat, 0° 09'S, 36° 09'E) the African Snipe is resident, while the Common Snipe is a regular visitor. The Jack Snipe, Great Snipe and Pintail Snipe occurred there in the 1980s, but over the last 20 years have become extremely rare. Snipe Populations The populations of snipes in Africa are not well documented. However, annual African Waterbird censuses, which are co-ordinated by Wetlands International, during the period between 1991 and 1997 indicate low numbers of individuals throughout the continent (table 2). These counts are usually carried out in January and February when Palearctic snipes are in the continent and in June and July when ee 201 LEGEND u African Snipe | @ Common Snipe | | +e Gallinago Species A Great Snipe * Jack Snipe gp Madagascar Snipe * Pintail Snipe Fig.1: Distribution of various species of snipes in Africa (source: annual waterbird census 1991- 1997): they have returned to their breeding areas in Europe and northern Asia. The Common Snipe and the African Snipe are abundant in all parts of their range within the continent. The Jack Snipe and Pintail Snipe occur in extremely low numbers. The Great Snipe has also become extremely rare in East Africa during the last 20 years (Zimmerman et al. 1996). The records of snipes often show an aggregated number of birds called Gallinago sp., which implies that the various species of snipes are not easy to differentiate in the field during the counts. 202 Table 2: Total number of various species of snipes recorded in Africa during January and July waterfowl counts between 1991 and 1997. [ Snipe species | 1901 | 1902 | 1903 | 1994 | 1995 | 1996 | 1997 | Tora | | Common Snipe | 464) | 2383 ri 24 (3) a ON ) 25 (4) 3 9 (3) 3 Ee ere ce as | | African Snipe | - I sı@ [io | 24) | am | 2760) | 227) | 1086 | Be 4 22 4 | Pintail Snipe | 2.4) | eran i ea Don aaa 2 | Madagascar Snipe | | [say eee eee ee BE i 1) ) |) ) as Nee sere) Note: Numbers in brackets represent the number of African countries that submitted records to Wetlands International There are better pre-1990 records and counts of snipes in West and South Africa than in East and Central Africa. However, this situation has been changing, with many countries in East and Central Africa starting to count and study snipes. Snipe populations in various African countries are relatively small as indicated by counts made between 1991 and 1997 (see fig.1). The low numbers may be linked to increasing aridity, water pollution, loss of wetlands, and counters insufficient in numbers or experience in the continent. However, the number of countries participating in the annual waterfowl counts in Africa has been gradually increasing and counts are becoming more regular. Data collection is also improving as more researchers study snipes in different parts of the continent. Morphological characteristics Body size, including bill size, are important characters influencing the outcome of interspecific and intraspecific competition in birds. Competition between indivi- duals for common resources as well as changes in resource abundance and availability generate adaptations which enable a particular population or sub- population to successfully colonize new habitats and exploit new niches. Differences in body size of various snipe species permit niche separation and hence minimize the negative effects of competition for food. Examination of variations in body measurements of Palearctic and resident snipes in Africa provides insight into the strategies that have enabled those waders to co-exist while utilizing similar habitats. Body size measurements provided in table 3 indicate that the Jack Snipe is the smallest and Madagascar Snipe the largest of the six species examined in this study. In evolutionary terms, Jack Snipe is perhaps the most advanced while Madagascar Snipe may be the most primitive. The African Snipe, Common Snipe, Pintail Snipe and Great Snipe are of intermediate size. Interspecific competition for food resources is likely to be intense among individuals of almost the same body size, especially where they share similar foraging habitats. The four species of snipes forage together in Africa during the boreal winter. 203 Table 3: Major body measurements of the various snipe species occurring in Africa Males and females combined). Ranges of body size measurements (mm) Wing length Tarsus length Bill length 170-190 105-121 22-25 Snipes are fast-flying birds. Migrant birds fly over long distances while the resident African Snipe makes considerable seasonal movements within its range. In terms of wing length, the Jack Snipe is again the smallest while the Madagascar Snipe is the largest. The other snipes fall in an intermediate wing length category. In terms of tarsus length, the Jack Snipe is the smallest and the Madagascar Snipe the largest, with the other four forming a cluster between the two. Females tend to be slightly larger than males, especially in bill length. Bill length has a direct relationship with food acquisition and hence competition. In terms of bill length, the Jack Snipe has the smallest with Common Snipe and Great Snipe occupying intermediate sizes. Madagascar Snipe and African Snipes have considerably longer bills than the Palearctic migrant snipes. Breeding Habitat All six species of snipes that occur in Africa, including Madagascar, inhabit and breed in wet areas. The Palearctic species, particularly the Jack Snipe and Great Snipe, breed in the cold areas of northern Europe and Asia. The Common Snipe breeds in warmer environments in Central Europe and Central Asia. The two species that breed in Africa also nest in cold but high altitude areas. The African Snipe breeds in highland bogs at altitudes of 1700-4000 m a.s.l., while the Madagascar Snipe breeds in upland marshes but occupies altitudes ranging from 0 Table 4: Breeding habitats of the various species of snipes occurring in Africa Open boreal marshland | Moist wooded boreal tundra Pintail Snipe Flooded sedge and moist grassland Fresh or brackish water, marshland with rich tussock sed Wet moorland, highland bogs and swamps Sedge swamps, rice fields, small bogs and banks of streams es, damp farmland 204 to 2500m in eastern Madagascar. The Common Snipe breeds in open wetlands with relatively short grass or sedges (table 4). The Jack Snipe breeds in open boreal marshland, while the Great Snipe nests in moist wooded boreal tundra. The breeding habitat of the Great Snipe extends to dry land with scattered clumps of vegetation. The breeding habitat of the Common Snipe is generally similar to that of the African Snipe and Madagascar Snipe. All three species build nests on moist ground with a rich community of tussock grass or sedges. The natural vegetation in which the nest is built shields the nesting bird and the nest contents from strong wind. It also provides shade for the eggs and brood during hot periods of the day as well as concealment from predators. Except for local differences in the choice of nest sites all the six species of snipes breed in similar habitats and cold environments. Behaviour Snipes are cryptically coloured, secretive, partly nocturnal and inhabit marshy areas, all of which make their study extremely difficult. Furthermore, the sexes look alike and young birds are difficult to distinguish from adults. Information on breeding behaviour is therefore scanty for most species. However, available information reveals considerable similarities in reproductive behaviour of the different snipe species that breed in Europe and Africa. One of the most conspicuous behaviours of breeding snipes is the aerial courtship display, known as drumming. This behaviour comprises circular flights broken by vertical dives from heights of between 60m and 100m (Reddig 1978, Gichuki et al. 1998). Most drumming is carried out by males but the receptive female may also drum early in the breeding season (Perrins & Middleton 1987). Snipes that occur in Africa show some common features of reproductive behaviour (table 5). Except for the African Snipe, which has a monogamous mating system (Urban et al 1986, Gichuki et al. 1998), the other five species exhibit a polygynous mating system. In the case of the Great Snipe, for instance, small groups of males display together in clearly defined lekking sites which can easily be recognized even after the birds have left (Hayman et al. 1986). In the Pintail Snipe, males competitively perform aerial displays which are equivalent to the leks of the Great Snipe. Females are attracted to the display areas where successful males mate with more than one female. Table 5: Common features of reproductive strategies and behaviour of the six snipe species found in Africa 205 Jack Snipe males hold a territory of up to 20 ha and advertise it by performing high aerial displays. A Great Snipe male defends a large territory of up to 100 ha in which it mates with several females and defends them against competitors. Madagascar Snipe males also defend females. In contrast, Common Snipe and African Snipe males make flight displays solitarily in the breeding area and do not maintain exclusive territories. In fact, African Snipe males make flight displays in the breeding area without any observable aggressive encounters (pers. obs.). There is little variability in nest site among the various species of snipes which breed in Europe and Africa. In all species nests are placed on moist ground with short grass or sedges. The African Snipe places its nest on raised platforms or on top of a tussock of grass which is about 10-15 cm above ground level (Gichuki et al. 1998). The nests of snipes are likely to be flooded by water if they are placed on lakeshores or seasonally flooded depressions. Water level or depth is therefore a critical factor influencing nest placement and nest success in snipes (Gichuki et al 1998). There is also little variability in clutch size, with females of most snipe species laying 3—4 eggs. The pear-shaped eggs are proportionally large, the clutch accounting for 60% of the female’s body weight in African Snipe (pers obs.). Females of the migrant snipes lay one clutch per year but lost clutches are replaced fairly quickly (Glutz et al. 1977). The African Snipe and the Madagascar Snipe appear to breed twice in a year. This is possible because of the high annual environmental productivity in tropical Africa where these birds breed. Parental,care varies between the various snipe species, with the majority having female care of the offspring. The African Snipe is monogamous and both parents take care of the young. It is not yet clear whether the parents share the task of incubating eggs and building the nest. In the Jack Snipe, Pintail Snipe and Great Snipe, females select nest sites, build the nests and incubate eggs while males guard them. While receptive females are guarded by breeding males early in the season, those males have opportunities to mate with other receptive females later, especially after the onset of egg-laying and incubation. The occurrence of 2 nestings in a season may enable the male African Snipe to maximize reproductive output by staying with one female. Food and its acquisition The diet of both the resident and Palearctic snipes is generally similar, consisting mainly of annelids, molluscs, insects, crustaceans and vegetable matter (Urban et al. 1986). Consumption of seeds is occasional but varies with habitat and season. The species of snipes with relatively short bills, especially Jack Snipe and Common Snipe, tend to consume more seeds and vegetable material than the species with longer bills (Devort et al. 1997). The African Snipe and the Madagascar Snipe which have relatively long bills eat predominantly invertebrates, especially earthworms. Conservation status Snipes are highly opportunistic and live in a wide range of wetland habitats. Areas that are conserved for more conspicuous waterfowl species, such as ducks, geese 206 and flamingos, are in most cases not suitable for snipes which at times inhabit small permanent wetlands or inconspicuous seasonal wetlands. As a result of landscape modifications in Europe and northern Asia where Palearctic snipes breed, and also in Africa where they winter, the global populations of migrant snipes appear to be gradually declining due to loss of suitable habitats. The populations of the Great Snipe and Pintail Snipe, for instance, have decreased considerably both in Europe and Africa. Habitat changes, due to changes in climate, drainage of wetlands and pollution of water systems, also threaten the survival of the African Snipe and Madagascar Snipe throughout their distribution ranges in the continent. Sport hunting of snipes poses no serious threat to the existence of this group of birds in Africa as the practice is largely uncommon. Summary and conclusion Snipes form a diverse group of waders that inhabit wetlands in various parts of the world. They have a strong association with moist or soft substrates from which they acquire food. The birds depend on their large and strategically placed eyes as well as their long and sensitive bill to detect food. The bills are weak and cannot penetrate hard surfaces. The males, females and juveniles are generally similar and well camouflaged. The birds are capable of rapidly “freezing” or taking off rapidly when disturbed or threatened by a predator. The morphological adaptations, such as long bill, and long wings (relative to the size of the body) that permit rapid take-off, have contributed significantly to the survival and success of snipes in a wide range of wet or moist habitats in their northern breeding areas and wintering areas in Africa. The populations of the various snipes are not well documented but declines can be inferred from rapidly shrinking or deteriorating habitat. Annual counts of snipes carried out in Africa do not provide accurate data, or show any pattern of population changes. There 1s considerable similarity in the food, breeding habitat and breeding behaviour of Palearctic and tropical snipes. Nearly all consume a wide variety of invertebrates and breed in wet and cool or cold environments. While Palearctic snipes make annual long-distance movements between their European breeding areas and wintering areas in Africa, the African Snipe and Madagascar Snipe make seasonal and altitudinal movements within their ranges. The three sub-populations of African Snipe are separated by large areas of arid land, and as long as they remain isolated from each other by physical barriers they could evolve into different species. In this context, the Madagascar Snipe appears to be the remnant of a larger population that was widespread in Africa but long ago disappeared from the continent. There is also a lack of variability in clutch size and nest site. The Palearctic and tropical snipes have similar clutch sizes and have retained the primitive ground- nesting habit of their presumably larger ancestors. A polygynous mating system and female parental care appear to be widespread among snipes. The evolution of monogamy in the African Snipe and the marked territorial behaviour of the Great Snipe and Jack Snipe appear to have been necessitated by high costs of parental care in the breeding areas of those species. 207 Greater attention in terms of research and conservation should be accorded to this complex and interesting group of waders. Snipes offer ample scope for research and opportunity for promoting conservation of shorebirds and the wetlands that support them. Snipes also provide useful tools for investigating various principles of ecology, evolutionary biology and sociobiology in animals. Acknowledgements Financial support for this project was provided by the Kenya Museum Society, the Frank Chapman Memorial Fund of the American Museum of Natural History, New York and a generous anonymous person.We are most grateful to Professor Lester Short and Jennifer Horne who were always helpful in times of need during the field work. The Director of Research and Scientific Affairs in the National Museum of Kenya, Dr. Rashid Aman was most supportive of this project in many ways. Thanks to Charles Waihenya, James Wachira, Lucy Gakua and several other volunteers who assisted with the fieldwork and endured patiently many hours of darkness, dangerous swamps and heavy rains.We are also grateful to the organizers of Isolated Vertebrate Communities in the Tropics who gave financial support to enable Cecilia to attend and present this paper at the symposium. Dr. Renate van den Elzen and Anja C. Schunke guided Cecilia in Bonn, Germany and made her stay very comfortable.We are indeed grateful. Last and not least we would like to thank Rose Warigia who kindly typed the manuscript. Special appreciation to our "most understanding children Jedida, Joshua and Elizabeth for their patience and support at home and in the field. References Pevwom view. Chevallier, H. Lethier, G. Olivier & J. Veiga (1997): The Snipe: Elements of an Action Plan. — Editions confluences, CICB, OMPO, Paris. Futuyma, D.J. (1979): Evolutionary Biology. — Sinauer Associates Inc., Sunderland, USA. Gichuki. C. M., N.N. Gichuki & E.W. Kairu (1998): Population Biology and Breeding Ecology of the African Snipe Gallinago nigripennis in Central Kenya. — Proc. 5th Int. Woodcock and Snipe Conf., 3-5 May, 1998, Czempin, Poland. Glutz von Blotzheim, U., K.M. Bauer& E. Bezzel (1977): Handbuch der Vogel Mitteleuropas. Band 7 (Charadriiformes). - AULA Wiesbaden, Germany. Hayman, P., J. Marchant & P. Prater (1986): Shorebirds: an identification guide. — Beckenham; Croom Helm, UK. Moreau, E.R. (1972): The Palearctic African Bird Migration Systems. — Academic Press, London, New York. Perrins,C.M. & A.L.A. Middleton (eds). (1987). The Encyclopedia of Birds. — Equinox Ltd., Oxford. Reddig, E. (1978): Der Ausdrucksflug der Bekassine. — J. Ornithol. 119: 357- 381. 208 Sibley, C.G. & B.L. Monroe (1990): Distribution and Taxonomy of Birds of the World. — Yale University Press, New Haven, US A. Urban, E.K., C.H. Fry & S. Keith (1986): Birds of Africa, vol. 2. - Academic Press, London . Zimmerman, D.A,, DA. Turner & DI. Bearson (1996) Binds omienya and northern Tanzania. — Russel Friedman Books Halfway House, South Africa. Authors’ address: Cecilia M. Gichuki & Nathan N. Gichuki, National Museums of Kenya, P.O. Box 40658, Nairobi, Kenya. 209 Review of karyological studies and the problems of systematics of Ethiopian Arvicanthis Lesson, 1842 (Rodentia: Muridae) Mere baskewichisée ErAT Davrenchenko Abstract: A review of karyological research in rats of the genus Arvicanthis in Ethiopia together with new data on the karyotype and intrapopulational chromosomal variability (especially variations in the amount of heterochromatin in A. somalicus from Awash National Park) are given. The chromosomal characteristics of the species are evaluated in view of their divergence as a result of long historical isolation. The possible role of chromosomal peculiarities in forming and supporting the isolating mechanisms among semisympatric species is discussed. The hypothesis on the possible positive correlation between species age and chromosomal variability is tested in the genus Arvicanthis. Taxonomic aspects of the accumulated chromosomal data for Ethiopian Arvicanthis are discussed. Key words: chromosomes, Muridae, Arvicanthis, variability, evolution, systematics, Africa, Ethiopia, endemics Introduction The present distribution of Grass rats of the genus Arvicanthis in Africa in different habitats (savanna, grassland, moorland, croplands and some others) reaches from the Nile basin, the Horn of Africa and East Africa across the central and western sub- saharan regions to West Africa. The history of the genus Arvicanthis in Africa started about 5 Mio years ago when the colonization of the continent from Asia took place (Jacobs 1978), and since that time the evolution and distribution of the genus has been significantly determined by climatic changes, leading to the isolation of populations over millions of years (Kingdon 1974). The results of these historical processes are obvious today on different levels, including that of the chromosomes. We studied the chromosomal peculiarities in the genus Arvicanthis in Ethiopia, where four species are recognized: either A. abyssinicus, A. blicki, A. somalicus and A. dembeensis (Yalden et al. 1976, Corbet & Hill 1991) or the first three and A. niloticus (Musser & Carleton 1993). While chromosomal differentiation is well documented for most Ethiopian representatives of the genus (Matthey 1959, Orlov et al. 1989, 1992, Baskevich et al. 1995, Corti et al. 1996, Lavrenchenko et al. 1997; see table 1) nothing was known about the karyotype of an endemic population in the Horn of Africa, A. somalicus — a small form adapted to low altitudes and dry climatic conditions (Yalden et al. 1976). We studied the chromosomal sets of 5 specimens of Arvicanthis from Awash National Park (table 1) identified according to their upper toothrow as A. somalicus. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 210 Results Within the studied specimens an intraspecific chromosomal polymorphism was obvious. Two specimens had a karyotype of 62, being composed of gradually decreasing acrocentrics, except one of the smallest pair of metacentric autosomes, and large submetacentric X- and medium-sized Y-chromosomes (NF=66). Following C-band staining all autosomes were revealed as medium-sized blocks of pericentro-meric heterochromatin, Y-chromosome entirely heterochromatinized and X-chromosome strongly stained in its pericentromeric region, in the whole short arm and before the distal end of the long arm (fig.1). The remaining specimens had karyotypes with 62 chromosomes with rearrangements in the first (standard and increased homologues) (1 sp. - fig.2a) or in one of the medium-sized (acrocentric and subtelocentric homologues) (2 sp. - fig.3a) pairs of autosomes. (In the karyotype of one 62-chromosomal specimen with rearrangement in a medium- sized autosomal pair, a fusion of acrocentric autosomes seemed to occur and we preliminary reduced its diploid number; but more detailed investigation supported its 62-chromosome state). It became obvious that the mechanism of this chromosomal polymorhism is due to the variability in the amount of hetero- chromatin (fig. 2b, 3b). We suggest that the system of intraspecific chromosomal polymorphism caused by changes in the heterochromatin quantity has an adaptive significance. When we compare our chromosomal results with published ones we can state that the 62-chromosome, unrearranged karyotype of A. somalicus is similar to that a» A A aa ar ** am Fig.1: Standard C-banded karyotype of Avicanthis somalicus ° from Awash National Park „a a) i) dr am au no ee np AR An OA GO DR AA nA do AXA AA HA EO Ei AR ea a ne BUN E.. ) Fig.2: Karyotype of female A. somalicus with rearrangement in the Ist pair of autosomes: a: routine staining; b: C-band staining of the Ist pair of A. dembeensis, found earlier at three localities in Ethiopia (Orlov et al. 1989, 1992, Baskevich et al. 1995, Corti et al. 1996; table 1). Obviously the same karyotype will be found in the specimens of A. dembeensis from Awash National Park where the ranges of these two species overlap (Yalden et al. 1976). We believe that the chromosomal peculiarities do not play an important role as an isolating mechanism among these sympatric species. But if they play a role in isolation their contribution is restricted to heterochromatin pattern only. In addition to the 62-chromosome A. dembeensis, a 60-chromosome Arvicanthis (sp.l) and a 56-chromosome Arvicanthis (sp. 2) have been distinguished in Ethiopia (table 1), all representing the “morphotype” of A. dembeensis. The first one has been recorded in the south of the Ethiopian Rift Valley, the second coexists with the 62-chromosomes A. dembeensis in the Gambella area. Distinct karyological forms within the morphotype of A. dembeensis in different localities of Ethiopia seem to correspond to relict populations, isolated from the rest of the species by dry areas since the last wet period. This assumption agrees with the 2.2 Table 1: Karyotypes of Ethiopian Arvicanthis Species/locality habitat/altitude no. of 2 reference coordinates (m a.s.l.) speci- mens A. abyssinicus l/unknown 2/ Ambo SRUSENEES 4/Managhesha DANO INS Skeets) 18) A. blicki 1/Bale Mts, Sanetti plateau (OP SIINE Soo 18 2/Bale Mts, Sanetti plateau 6°51'N, 39°53'E A. dembeensis 1/ Gambella COSINE SAS 223 2/70 km N of Awash Nat. Park 9°28'N, 40°18'E 8°24'N, 39°O1'E Arvicanthis sp.1 Konso ISIN SPA Arvicanthis sp.2 Gambella ($253 NB 422.215 A. somalicus Awash Nat. Park 9°00'N, 40° 10'E Cropland/ 2000 ditto cropland/ 2700 ditto Combretum savanna/2200-2300 alpine moorland/ 4000 Afroalpine belt/4050 Combretum savanna/450 Acacia arid savanna/900 open savanna, cropland/1550 open savanna/ 1650 savanna/ 900 Combretum savanna/450 Acacia arid savanna/900 Matthey 1959 Orlov et al. 1989 Orlov et al. 1992 Corti et al. 1996 Corti et al. 1996 Corti et al. 1996 Corti et al. 1996 Lavrenchenko et al. 1997 Orlov et al. 1989, 1992 Orlov et al. 1992 Baskevich et al. 1995 Corti et al. 1996 Orlov et al. 1992 Orlov et al. 1989 this study known relationship between Arvicanthis and humidity (see Kingdon 1974) and their recent distribution in Ethiopia (Yalden et al. 1976). Since chromosomal material from the terra typica of A. dembeensis (Tana lake area) is lacking, it is not possible to draw taxonomical conclusions for the various forms of “dembeensis”. Additionally, species rank of A. dembeensis was confused in the last taxonomic 213 AN RO 00 AR AMD AAR hn na on A as b os Baa OA Ne Oa My a * f Fig.3: Karyotype of A. somalicus © with rearrangement in the medium sized autosomal pair: a: routine staining; b: C-banding. review (Musser & Carleton 1993) where it is considered as a junior synonym of A. niloticus. A. abyssinicus and A. blicki are endemic to Ethiopia (Yalden & Largen 1992, Musser & Carleton 1993), thus indicating the isolation of Ethiopia over a 214 considerable period. Both species are adapted to higher altitudes on the Ethiopian plateau and to alpine grasslands, to which they may have been restricted during the upper Quartenary intertropical dry phase after the last glaciation (Adamson et al. 1980). They share a common ancestor and speciated during the Pleistocene (Corti et al. 1996). The analysis of chromosomal data for A. abyssinicus from three known localities (2n=62, NF=68) and A. blicki (2n=48, NF=68) reveals that their karyotypes are distinct (table 1): their differences consist of 7 Robertsonian translocations and similarities by the NORs localization pattern (Corti et al. 1996). Indeed, the specific localization of NORs in the two first pairs of autosomes - which are characteristic for only these two species - indicates on the chromosomal level that A. abyssinicus and A. blicki share a common ancestor. The variability in the X- and Y-chromosome morphology caused by changes in the amount of heterochromatin in the heterochromosomes and in the morphology of the two pairs of autosomes (2n=62, NF=68) characterizes the karyotype of A. abyssinicus (table 1). It is known that A. dembeensis and A. abyssinicus diverged earlier than the Ethiopian endemics (Capula et al. 1996, cit. from Corti et al. 1996). The karyotype of A. abyssinicus differs from that of the 62-chromosome A. dembeensis by one pericentromeric inversion (Corti et al. 1996) and from that of the 60-chromosome Avicanthis sp. | (“morphotype” A. dembeensis) by one Robertsonian translocation, four pericentric inversions and four chromosomal rearrangements connected with heterochromatin variations (Orlov et al. 1992). When we compare the karyotypes of the Ethiopian species and estimate their chromosomal variability we see the conservative structure of the sex chromosomes in the four Ethiopian species (Orlov et al. 1989, 1992, Baskevich et al. 1995, Corti et al. 1996; our data). The last observation can support the monophyly of the Ethiopian group. It has been shown that A. blicki is the most recent species in the genus Arvicanthis. Its divergence from the genetically related A. abyssinicus is dated to the end of Pleistocene; the latter diverged earlier from A. dembeensis; the time of divergence between West and East African lineages of Arvicanthis has been estimated to the Upper Pliocene (Capula et al. 1996, cit. from Corti et al. 1996).Using these data we can try to test the hypothesis of a positive correlation between the age of a taxon and its chromosome variability (Gileva 1990). The chromosome variability in Ethiopian Arvicanthis as demonstrated in table | and the age of the genus supports this assumption. When we summarize the chromosomal results on Ethiopian Arvicanthis we can state that the number of known chromosomal forms exceeds the recognized species number (Yalden et al. 1976, Musser & Carleton 1993); therefore it is possible that future revisions of this genus may increase the number of species and thus the Ethiopian theriofauna. The work was supported by the Russian Fund of Fundamental Investigations (Grant No 98-04-49210) and by the Programm “Frontiers in Genetics”. References Enkımson. DAN. PF: Gasse, R.A. Street & M.A.J. Williams (1980): Late Quarternary history of the Nile. — Nature, London 288: 50-55. Pere vicn. Mil.) VN. Orlov, A. Bekele & A. Mebrate (1995): New data on the karyotypes of Ethiopian small mammals (Insectivora, Rodentia). — pp. 58-72 in: Theriological studies in Ethiopia (ed. Sokolov, V.E.). — Nauka, Moscow. Bere Be & J.B. Hill (1991): A world list of mammalian species. — Oxford University Press, Oxford. Benue vie MV. Civitelli, R. Castiglia, A. Bekele & E. Capanna (1996): Cytogenetics of the genus Arvicanthis (Rodentia, Muridae). 2. The chromosomes of three species from Ethiopia: A. abyssinicus, A. dembeensis and A. blicki. — Z. Säugetierk. 61: 339-351. Gileva, E. A. (1990): Chromosomal variability and evolution. — Nauka, Moscow, 141 pp. Jacobs, L. (1978): Fossil rodents (Rhizomyidae and Muridae) from Neogene Siwalik deposits, Pakistan. — Bull. Mus. North. Arizona Press 52: 1-103. Kingdon, J. (1974): East African Mammals: an atlas of evolution in Africa. Vol. II, part. B: hares and rodents. — Academic Press, London. mamoncmemenko, Lb. A., A.N. Milishnikov, V.M. Aniskin, A. A. Warshavsky & W. Gebrekidan (1997): The genetic diversity of small mammals of the Bale mountains, Ethiopia. — SINET: Ethiop. J. Sci. 20: 213-233. Matthey, R. (1959): Formules chromosomiques de Murinae et de Spalacidae. La question du polymorphisme chromosomique chez les Mammiferes. — Rev. Suisse Zool. 66: 175-209. MoarsscreGoG., M.D. Carleton (1993): Family Muridae. — pp. 501-755 in: Mammal Species of the World. A taxonomic and geographic reference (eds. Wilson, D.E. & D.M. Reeder) — Smithsonian Inst. Press, Washington, London. Orlov, V.N.,M.I. Baskevich & N.Sh. Bulatova (1992): Chromosomal sets of the rats of the genus Arvicanthis from Ethiopia. — Zool. Zh. 71: 103-112. Saov VN. N. Sh. Bulatova & A.N. Milishnikov (1989): Karyotypes of some mammalian species (Insectivora, Rodentia) in Ethiopia. — pp. 72-94 in: Ecological and Faunistic studies in South-Western Ethiopia. — Nauka, Moscow. Bergen Dm WM oJ. Largen & D. Kock (1976): Catalogue of mammals of Ethiopia. 2. Insectivora and Rodentia. — Monit. Zool. Ital. (NS), suppl. 8: 1-118. Barden D Wo & MJ. Largen (1992): The endemic mammals of Ethiopia. — Mammal Rev. 22: 115-150. Authors’ address: Dr. Marina Baskevich and Dr. Leonid Lavrenchenko, Severtsov Institute of Ecology and Evolution, Rassian Academy of Sciences, Leninsky prospect 33, Moscow 117071, Russia. E-mail: sevin@ glas.apc.org N Isolation of bird populations in the North Nandi Forest, western Kenya Herbert Schifter Abstract: The North Nandi Forest is inhabited by some clearly isolated populations of small passerines which have been given subspecific rank only recently (Cunningham-van Someren & Schifter 1981). These include Phyllastrephus cabanisi, Bleda syndactyla, Sheppardia aequatorialis pallidigularis, Alethe poliocephala, Cossypha cyanocampter, Neocossyphus poensis, Turdus olivaceus porini and Illadopsis abyssinica. Some of them have been generally acknowledged. Indeed some isolated populations which have not yet been named as separate subspecies may exist. Among them the population of Zosterops senegalensis, should be mentioned, which is certainly intermediate between the populations to the east and those in the lower-lying Kakamega Forest. Also Pogonocichla stellata of the North Nandi Forest differs markedly from the population of the Mount Elgon area to the west. Furthermore there exists a very marked population of Platysteira concreta, which was formerly known only from the Kakamega Forest, therefore supporting extended protection of both areas. By contrast, larger birds seem to move regularly from the North Nandi Forest and do not represent isolated populations, but conservation of the North Nandi Forest is also important for their protection. Key words: Nandi forest, Kenya, new subspecies, conservation The study area The North Nandi Forest is an isolated forest in western Kenya which today is reduced to a narrow strip at the western edge of the North Nandi Escarpment, stretching about 32 km from north to south but only roughly 9 km wide. It ranges from about 1950 to more than 2150 m and is separated from the lower-lying and more western Kakamega Forest by the North Nandi Escarpment. The forest has been extensively logged in the past and has been widely cleared, mostly on its western edge. Only a small part of the forest has been declared as a Nature Reserve, in 1968 and in 1978. Nevertheless the remaining parts of the forest still house a great variety of animals and plants, though the number of species is lower than in the better known Kakamega Forest probably due to the greater altitude. Birds in general Among the resident bird species some have been found to represent obviously isolated populations, and have been given the rank of subspecies (Cunningham-van Someren & Schifter 1981), whereas larger birds undertake movements to the other forests in the area, especially the Kakamega Forest. The isolated populations of smaller passerines which have been given subspecific rank are: Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 218 Olive Greenbul Phyllastrephus cabanisi nandensis Common Bristlebill Bleda syndactyla nandensis Equatorial Akalat Sheppardia aequatorialis pallidigularis Brown-chested Alethe Alethe poliocephala nandensis Blue-shouldered Robin-Chat Cossypha cyanocampter pallidiventris White-tailed Ant-Thrush Neocossyphus poensis nigridorsalis Northern Olive Thrush Turdus olivaceus porini African Hill Babbler /lladopsis abyssinica poliothroax. Some of them have been generally acknowledged. Indeed isolated populations which have not yet been named as separate subspecies may exist. Among them is the population of the Yellow White-eye Zosterops senegalensis, which is certainly intermediate between the populations to the east and those in the lower-lying Kakamega Forest. Also the White-starred Robins Pogonocichla stellata of the North Nandi Forest differ markedly from the population of the Mount Elgon area to the west. Furthermore there is a very marked population of the Yellow-bellied Wattle-eye Platysteira concreta, which was formerly known only from the Kakamega Forest, which therefore supports extended protection of both areas. Larger birds which seem to move regularly from the North Nandi Forest to the Kakamega Forest do not represent isolated populations but conservation of the North Nandi Forest is important for their protection. Selected bird species The Common Bristlebill (Bleda syndactyla) is an inhabitant of the forests of West and Central Africa, reaching in western Kenya its eastern limit of distribution. Formerly included in the widespread subspecies B. s. woosnami, the Kenyan population has been separated as B. s. nandensis by Cunningham-van Someren & Schifter (1981) mainly because of its larger size. It is arare species in North Nandi, reaching here its upper limit of distribution. Also the Nandi population of the Olive Greenbul (Phyllastrephus cabanisi) has been separated as P. c. nandensis by its more greenish back. Formerly included in Phyllastrephus fischeri, the highland populations have been merged recently into P.cabanisi (Dowsett 1972, Short et al.1990: 154). The Equatorial Akalat Sheppardia aequatorialis is one of the common birds of the forest undergrowth. The population of the North Nandi Forest is different from those of other highland forests in Kenya and therefore has been separated as Sheppardia aequatorialis pallidigularis by being more “golden-yellow” than brown, with a conspicuous pale throat. Specimens from the Kakamega are darker without the distinct pale throat. Nevertheless both populations have been merged with the nominate subspecies by Fry et al. (1992). Many subadult birds caught between October and December in the study area indicate breeding late in the year in the North Nandi Forest whereas no subadult birds were found in September 1988. The Brown-chested Alethe (Alethe poliocephala) is a mostly West and Central African species of the family Turdidae. The population of the North Nandi Forest has been separated as Alethe poliocephala nandensis, based on the more pronounced Brussels brown of the back and also by its lack of a marked breast ae ar re + x IMATON G @ bom Gilo / “TL ‘* See, + Say en eo ENA Sy ; YA.p.nandensis x ER ELAGO: + @SBwWAMBA es BENI top (TORO) „” NORTH® NANDI 4) & ® 2 @ kisale + = - E 9k KAKAMEGA® \ ONE 3 9) er N N 3 ee 2 a a ty tog Bly Se A ® ar Zip A.p.akeleyae \ Ve # fi nx + a * » + 2 ‘es me 7 TANSANIA e St ; ee | » Paves Met tye Fig.1: Distribution of the subspecies of Alethe poliocephala in eastern Africa band, from the more eastern subspecies A. p. akeleyae. Though a very common species of the forest undergrowth, this alethe seems to be very sedentary, as shown by a specimen ringed as an adult bird on 23 November 1979 which was retrapped at a distance of only 100 m from the ringing site nearly nine years later on 21 September 1988 (Gichuki & Schifter 1990). The White-tailed Ant-Thrush (Neocossyphus poensis) of the North Nandi Forest has been separated as N. p. nigridorsalis. Though a rare bird in the area and confined to wetter parts near streams and swamps the bird seems to be very sedentary. The population of the Kakamega Forest has also been recently separated as N. p. kakamegoes by Cunningham-van Someren & Schifter (1981). Similarly, the clearly isolated population of the Northern Olive Thrush (Turdus olivaceus) might be separable as T. o. porini, differing even from nearby T. o. elgonensis by darker breast and back. Generally the populations of the Northern Olive Thrush in Kenya are certainly different from 7. abyssinicus of the Ethiopian highlands. The Blue-shouldered Robin-Chat (Cossypha cyanocampter) also is a West and Central African species reaching its eastern limit of distribution in the North Nandi Forest. The resident birds of the forest have been named C. c. pallidiventris, though this separate population is not generally acknowledged and has been merged iecenuy asain with C. c. bartteloti by Fry et al. (1992). The Nandi Forest population of the White-starred Robin (Pogonocichla stellata) may also represent 220 a distinct subspecies but has usually been included in the wider-ranging P. s. intensa. It should be noted that the nearby Mount Elgon is inhabited by its own subspecies P. s. elgonensis. The North Nandi population of the widespread African Hill Babbler (/lladopsis abyssinica[?]) has been separated as Alcippe a. poliothorax by its paler ventral surface. The distribution of the species, occurring in isolated mountain islands, is highly discontinuous in East Africa with even more significantly distinct populations in areas such as Mt. Loima in western Turkana. Therefore Hill Babblers from Kenya (formerly included in Alcippe) should certainly not be merged with the nominate subspecies from Ethiopia. Other Timaliidae of the North Nandi Forest belong to Central African species and may not be separable from the more westerly populations, including the rare Gray-chested Illadopsis Kakamega poliothorax. The Yellow-billed [oder bellied? ]Wattle-eye (Platysteira concreta) is alsoa West and Central African species. It was discovered in western Kenya only in 1923 and this population was named P. c. silvae by Hartert & Van Someren. In the North Nandi Forest it was discovered only in 1978 by our expedition (see also Schifter & Cunningham-van Someren 1998) where it is also obviously a rare bird. The isolated populations of western Kenya are likely to represent a separate subspecies as listed by Zimmerman (1972), though not generally acknowledged probably due to the very limited material available for studies in museum collections. Occurrence of the Yellow-bellied[s.o.!] Wattle-eye in North Nandi Forest, as well as the endemic subspecies of the area, strongly indicate that the site requires improved protection. The Black-throated Wattle-eye (Platysteira mentalis) and Jameson’s Wattle-eye (Platysteira jamesoni) of the North Nandi Forest probably do not represent separable subspecies. Migratory species such as the African Paradise Monarch- Flycatcher (Terpsiphone viridis) are not represented by distinct subspecies; in this species two different subspecies (T. v. suahelica and T. v. ferreti) have been confirmed by mistnetting. Also tits and sunbirds found in the North Nandi Forest do not differ significantly from populations found in the Kakamega Forest or in the highlands of Central Kenya. Yellow White-eyes (Zosterops senegalensis) are very common in the North Nandi Forest. Specimens of the probably resident population differ only slightly from those from the type locality of Z. jacksoni (Mau) and are intermediate between them and the more western Z. yalensis (named by Van Someren from Kaimosi). There is also considerable individual variation. Therefore the birds of the North Nandi Forest should still be included in Zosterops senegalensis jacksoni as long as further studies do not support new taxonomic Views. References Cunningham-van Someren, GR. & H. Schifter G98) Newaracesron montane birds from Kenya and southern Sudan. — Bull. Brit. Orn. Cl. 101: 347- 354, 355-363. Dowsett, R.J. (1972): Is the bulbul Phyllastrephus placidus a good species ? — Bulle Brit. Onn e 227952238 221 Beye Or. Keith, s. & Urban, E.K. (1992): The birds of Africa. Vol. 1V.— Academic Press,London. eienukı, EM. & H. Schifter (1990): Long life-span and sedentariness of birds in North Nandi Forest, Kenya. — Scopus 14. 24-25. Schifter, H. & G.R. Cunningham-van Someren (1998): The Avifauna of the North Nandi Forest, Kenya. — Ann. Nat.hist. Mus. Wien 100B: 425-479. ene. J.-F Me Horne & C. Muring-Gichuky (1990): Annotated check-list of the birds of East Africa. — Proc. West. Foundation of Vertebrate Zoology, 4 (3): 61-246. Zimmerman, D.A. (1972): The Avifauna of the Kakamega Forest, Western Kenya, including a bird population study. — Bull. Amer. Mus. Nat. Hist. 149 (3). 255-340. Author’s address: Dr. Herbert Schifter, Naturhistorisches Museum Wien, Erste Zoologische Abteilung, Vogelsammlung, Burgring 7, Postfach 417, A-1014 Wien, Austria uit 223 The mammals of the isolated Harenna Forest (southern Ethiopia): structure and history of the fauna eA aw Teme ne niko Abstract: The results of the faunistic survey of mammals conducted in the Harenna Forest are given. Seven mammalian species were recorded for the first time for the forest, two of them (Stenonycteris lanosus and Dasymys incomtus) were new for the whole area to the east of the Ethiopian Rift Valley. Altogether fifty-three mammalian species (including three species endemic to the forest) are so far known from the Harenna Forest. Distribution of mammals along an elevational gradient was studied. On the whole, the relationship between species-richness and elevation is strikingly hump-shaped, peaking in the 1900 — 2500 m zone. Furthermore, elevational ranges do not tend to increase with elevation, so “Stevens’s rule” is not corroborated. This effect may be associated with the absence of specialised low-elevation species as a result of the isolated position of the forest. In general, the mammalian fauna of the Harenna Forest was formed rather recently, mainly through a limited immigration of forest species and recruitent of species from adjacent altitudinal zones with their subsequent adaptive speciation. Although the fauna of the Harenna Forest demonstrates a superficial similarity with other African forest faunas, sharing with them such specialised ecological forms as forest guenon and arboreal hyrax, it 1s to a large extent unrelated to these mammalian faunas phylogenetically. Key words: biogeography, forest mammals, elevational gradient, endemics, Ethiopia Introduction The southern slope of the Bale Mountains represents a continuous range of undisturbed natural vegetation, from 1500 to 4400 m a.s.1. Such an environment is found in very few, if any, montane terrains in Africa. Montane evergreen tropical rain forest, known locally as the Harenna Forest, encompasses the largest part of this area (from 1500 to 3250 m a.s.l.). Furthermore, this forest is separated from other forest areas by the Rift Valley to the west, the flat high Sanetti Plateau to the north, and arid lowlands on the other sides. This makes the Harenna Forest a valuable model for the study of formation and evolution of an isolated tropical forest fauna. Until recently, the mammalian fauna of this forest was unexplored. The first zoological expedition reached the Harenna Forest only in August 1986 and found there three species new for Ethiopia (Cercopithecus albogularis, Cephalophus ?harveyi and Mus cf. triton) and two species new to science (Crocidura harenna and Crocidura bottegoides) (Afework Bekele 1988, Yalden 1988a, b, Hutterer & Yalden 1990, Yalden et al. 1996). It is likely that more undiscovered mammals occur in this forest. Furthermore, the Harenna Forest is an important but poorly known part of the Bale Mountains National Park (BMNP), which is one of the few Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 224 protected key areas in Ethiopia. So, basic information on mammalian fauna of this forest is of importance for local conservation projects. This paper presents the results of the faunistic survey of mammals conducted along an elevational transect in the Harenna Forest between February 1995 and January 1996. A biogeographical analysis of these findings and the resulting patterns of altitudional zonation of the mammalian fauna are also given. Study area and methods As a result of the great altitudinal range both, structure and composition of the Harenna Forest vary significantly. The following major vegetation belts have been visually recognized: | l) Podocarpus belt (1550 — 1900 m a.s.1.) 2) Aningeria belt (1900 — 2350. m) 3) Schefflera — Hagenia belt (2350 — 2600 m) 4) Hagenia — Schefflera — Hypericum belt (2600 — 3000 m) 5) Hypericum — Erica — Rappanea belt (3000 — 3250 m) (Lisanework Nigatu & Mesfin Tadesse 1989). A trapping-line of 100 live-traps was placed in each of these vegetation belts to study the distribution of small rodents and shrews (for detailed description of the trapping sites, see Lavrenchenko et al. 1998). Bats were caught in mist nets, positioned in clearings in the forest near these sites. Vehicle monitoring along the road, crossing the Harenna Forest, was carried out to study the distribution of the larger species. Additionally notes, measurements and photographs were made of any tracks, burrows and droppings of medium-sized and large mammals. I also used published records for the forest (Afework Bekele 1988, Yalden 1988a, b, Hutterer & Yalden 1990, Yalden et al. 1996) and unpublished data from archival files of the Bale Mountains National Park. The elevational range of each species was calculated. For most species it was determined by interpolating between elevational extremes (with the exception of few species — Crocidura thalia, C. olivieri, Dendromus mystacalis, and Stenocephalemys griseicauda — inhabiting only ericaceous bush near the upper limit of the forest and/or riverine habitats inside the forest). The elevational transect was divided into 18 arbitrarily defined 100-m units. A species was considered to be inhabiting an elevational unit if its altitudinal range covered more than a third of this unit. Similarities among species assemblages in these units, and turnover rates between adjacent zones on the elevational gradient, were evaluated by Jaccard’s coefficient (J) using the NTSYS program (Rohlf 1988). Results and discussion In total, fifty-three mammalian species were recorded in the Harenna Forest (table 1). Seven species (Rousettus aegyptiacus, Stenonycteris lanosus, Miniopterus schreibersi, Heterohyrax brucei, Heliosciurus gambianus, Lophiomys imhausii, and Dasymys incomtus) were recorded for the first time for the forest, two of them (Stenonycteris lanosus and Dasymys incomtus) were new for the whole area east of the Ethiopian Rift Valley. Furthermore, our study has revealed that Lophuromys 5) NM Nn Table 1. — Species of mammals recorded in the Harenna Forest. Zoogeographic categories modified fromYalden et al. (1996): CAF = Central African forest, EAF = East African forest, EAM = East African montane, EAS = East African savanna, EE = endemic to Ethiopia, EH = endemic to the Harenna Forest, P = Palaearctic, PAS = Pan- African savanna, SA = Somali-arid, SS = Saharo-Sindian, WAS = West African savanna, U = unclassified. No. Species Zoogeographic Habitat in the Range of records (m) Crocidura thalia EE grassland 1 935-3 300 Crocidura olivieri PAS grassland 1 935-2 400 Crocidura bottegoides BE forest/grassland 2 400-3 280 Crocidura harenna BEI forest 2 400-2 630 Rousettus aegyptiacus U forest 1 970 Lissonicteris angolensis CAF forest 1 550-1 880 Stenonycteris lanosus EAM forest 1 970 Rhinolophus cf. deckenii U forest 1 970 Rhinolophus clivosus U woodland 1580, Rhinolophus landeri PAS woodland 1550 Rhinolophus hildebrandtii EAS forest 2 400 Pipistrellus nanus BINS forest 15950 Plecotus austriacus B forest 1 980-3 280 Miniopterus inflatus U forest 1 950-3 280 Miniopterus schreibersi U forest 2 400 Galago senegalensis WAS forest 1 530-1 950 Papio anubis WAS grassland/forest 1 500-2 900 Cercopithecus djamdjamensis EH forest 2 400-2 800 Colobus guereza CAF forest 1 550-3 000 Lycaon pictus PAS grassland 1 550-2 400 Canis aureus SS grassland/forest 1 680-2 400 Viverra civetta U forest/grassland 1 600-2 600 Genetta maculata U forest 1 900-2 400 Atilax paludinosus U forest 2 760 Herpestes sanguineus U forest 1 650-2 220 Ichneumia albicauda PAS forest/grassland 1 500-3 075 Crocuta crocuta PAS forest/grassland 1 950-2 400 Felis serval PAS grassland/forest 1 680-2 400 Panthera pardus U forest 1 680-2 400 Panthera leo U forest/grassland 1 750-2 400 Heterohyrax brucei EAS forest 2 460-2 760 Orycteropus afer PAS woodland 5550 Hylochoerus meinertzhageni CAF forest 1 600-2 400 Potamochoerus larvatus EAF forest/grassland 1 600-2 760 Phacochoerus africanus PAS grassland 2 350 Sylvicapra grimmia PAS forest/grassland 2 400-3 300 Cephalophus ?harveyi EAF forest 2 400 Tragelaphus buxtoni BE forest/grassland 2140 Tragelaphus scriptus U forest 1 500-3 200 No. Species Zoogeographic Habitat in the Range of records (m) ESSE BAe Miata als EN categor Parenma Honest). U 21.22 2 eee 40 | Heliosciurus gambianus PASS forest 1 950-2 150 41 |Lophiomys imhausii U forest 2 760 42 |Dendromus mystacalis PAS grassland 2 400-3 280 43 |Tachyoryctes splendens EAM grassland 2 400-2 790 44 |Mus mahomet EAM grassland 1 950-1 970 45 |Mus cf. triton EH forest 1 950-2 920 46 |Praomys albipes BE forest 1 500-3 075 47 |Stenocephalemys griseicauda BE grassland 2 400 48 |Lophuromys brevicaudus EE forest 2 400-3 300 49 |Lophuromys chrysopus BE forest 1 550-2 760 50 |Dasymys incomtus PAS swamp 1979 5l | Graphiurus murinus PAS forest 3 280 I52 | Hystrix cristata WAS | forest/grassland | 1 900-2 350 “flavopunctatus” from the Harenna Forest comprises two distinct species, L. brevicaudus and L. chrysopus, both endemic to Ethiopia (Lavrenchenko et al. 1998). The relationship between species-richness and elevation is strikingly hump-shaped, with a peak ın the 1900-2500 m zone (fig.1). It is in agreement with the model of bounded random geographical ranges with two hard boundaries (model 4 in Rahbek 1997) where the pattern of species richness is produced on the assumption of random elevational association between the size and placement of 35 7 30 - 25 4 Species richness 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 3100 3200 Elevational range (m) Fig.1: Elevational variation in the mammal species richness in the Harenna Forest. 3600 3200 2800 2400 2000 Elevational range (m) 1600 1200 353221651275, 6 20924 29°30 52,23 2734919 46°39 15 47-45 43°31 124.3 36 51 10 6 25 44 7 50 33 28 40 22 38 49 27 17 26 35 11 37 53 4 18 14 13 41 42 48 Species Fig.2: Elevational ranges (ordered by elevational midpoint) of mammal species of the Harenna Forest. See table 1 for numbers of species. the species’ elevational ranges. On the other hand, the hypothesis of a hump-shaped relationship between species-richness and primary productivity (model 3 in Rahbek 1997) cannot be rejected because in our case the peak was reached at an elevation lower than the median of the gradient. Furthermore, elevational ranges do not tend to increase with elevation (fig.2). There is no significant correlation between the elevational range and the elevational midpoint (r = 0.202), so Stevens’s (1992) extension of “Rapoport’s rule” on the elevational gradient is not corroborated. It may be associated with the absence of specialised low-elevation species due to the isolated position of the forest. It is noteworthy that three forest species endemic to the Harenna Forest (Crocidura harenna, Cercopithecus djamdjamensis and Mus cf. triton) occur at medium elevations only, whereas more widespread Ethiopian woodland endemics (Tragelaphus scriptus meneliki, Praomys albipes, and Lophuromys chrysopus) show relatively broader altitudinal ranges in the forest. Jaccard’s similarity profile for the Harenna Forest elevational gradient is given in fig.3. The two lowest dips in pairwise similarity (1900 m and 2400 m), indicating high turnover rate in mammalian species composition, correspond to the zone boundaries between the Podocarpus and Aningeria belts, and the Aningeria and Schefflera-Hagenia belts respectively. Therefore the biomes of these three vegetation belts possess to some extent discrete mammalian communities. A characteristic feature of the Harenna Forest mammalian fauna is the nearly complete absence of both West and Central African forest species, which, however, occur in montane forests of south-west Ethiopia (Crocidura niobe, Hypsignathus monstrosus, Epomophorus gambianus, Micropteropus pusillus, Hipposideros ruber, Hipposideros fuliginosus, Pipistrellus tenuipinnis, Mimetillus moloneyi, 228 Mops nanulus, Cercopithecus mitis boutourlinii, Cercopithecus neglectus, Oenomys hypoxanthus) and of species found in riverine forests of southern Somalia (Epomophorus wahlbergi, Nycteris parisii, Pipistrellus eisentrauti, Galago crassicaudatus, Cercopithecus mitis zammaranoi). The Central African forest species Lissonycteris angolensis and Hylochoerus meinertzhageni, and the East African Cephalophus harveyi, are the exceptions (the occurrence of the latter species 1s supported by sightings only — for details see Yalden et al. 1996). Most of the forest species endemic to Ethiopia (Crocidura macmillani, C. phaeura, C. zaphiri, Grammomys minnae, Desmomys harringtoni, and Praomys ruppi) are absent from the Harenna Forest. The exceptions are two widespread Ethiopian forest endemics, Praomys albipes and Lophuromys chrysopus, for which a recent origin from aboriginal species of the ericaceous belt in the Bale Massif (Lophuromys brevicaudus and Stenocephalemys griseicauda respectively) has been proposed (Lavrenchenko et al. 2000). The contemporary distribution of these two species may be accounted for by dispersion from the Bale Massif during some periods in the Pleistocene when a more humid climate could have allowed the few forest and marsh species to spread across the Rift Valley, which at present is the major zoogeographic barrier in Ethiopia. Our finding of a relict and isolated population of Dasymys incomtus in a small swamp on the Shawe river in the centre of the Harenna Forest supports this hypothesis, assuming at the same time that the latter species could have dispersed in the opposite direction Swehnelacn populations of Dasymys incomtus are also known from Western Senegal (Duplantier et al. 1997). The lack of widespread forest-dwelling, small terrestrial mammals for which the risk of local extinction and the colonising capacity is low, as well as the presence of some Central and East African forest ungulates (with the opposite characteristics), support the concept that the mammalian fauna of the Harenna Forest was formed rather by colonization de novo than by the extinction of representatives of a forest fauna once common in the whole of Central and East Africa. On the other hand, the Harenna Forest is remarkable for the number of endemic mammals (Crocidura harenna, Cercopithecus djamdjamensis and Mus cf. triton) that inhabit this relatively small geographic area. Crocidura harenna is a typical member of a putative phylogenetic cluster of 36-chromosome Crocidura species from the Bale Massif. It is worth mentioning that the karyotype of this morphologically distinct species 1s practically identical with that of the montane Ethiopian endemic Crocidura glassi (Lavrenchenko et al. 1997). The other Harenna endemic, Cercopithecus djamdjamensis, being a member of the superspecies C. aethiops, is uniquely adapted to closed-canopy montane forest (Carpaneto & Gippoliti 1994). Its resemblance to C. mitis sensu lato, possibly resulting from this adaptation, is so amazing that a single female, collected in the course of the Harenna Forest Expedition in 1986, was identified as C. albogularis (Afework Bekele 1988). However, our repeated observations of these monkeys confirm the conclusion of Carpaneto & Gippoliti (1994) that C. djamdjamensis belongs to the C. aethiops species complex, but not to the C. mitis-albogularis lineage (peculiarities of acoustics, locomotion and scrotum colour are the diagnostic features). 229 0,8 - oO oO —| oO aN Pairwise Jaccard's similarity (J) (S) N 0,0 + 2100 2200 2300 2400 2500 2600 2700 + 2800 2900 3000 3100 3200 3300 Fo Al 1600 1700 - 1800 1900 2000 - Elevational levels (m) Fig.3: Pairwise Jaccard’s similarities of mammal assemblages between adjacent elevational units of the Harenna Forest. The distinctive, olive-brown Mus from the Harenna was assigned to M. triton (Yalden 1988 a). However, our detailed cytogenetic study (Aniskin et al. 1998) does not support this identification. Moreover, Mus cf. triton from the Harenna Forest, together with the widespread Ethiopian M. mahomet, belongs to the cytotaxonomic group that does not include the true M. triton and is characterised by primitive acrocentric heterochromosomes (for review, see Capanna 1985). So, all three endemic Harenna species originated apparently from aboriginal non-forest stocks and their resemblance to some species from other forest territories might be a result of convergence. Further, the arboreal browser Heterohyrax brucei found in the Harenna Forest occupies the ecological niche of a Dendrohyrax species. A comparison with sizable material in the Koninklijk Museum voor Midden-Afrika (Tervuren) reveals that the Harenna Heterohyrax resembles Dendrohyrax arboreus in teeth pattern (but possesses the other main diagnostic cranial features of Heterohyrax), which may be a result of this ecological shift. It is noteworthy that Kingdon (1971) assumed a direct origin of Dendrohyrax validus from some ancestral Heterohyrax population. The Harenna Forest is the southernmost limit for the essentially Palaearctic species Plecotus austriacus. It is interesting that the Harenna Plecotus differs strikingly in its colour pattern from the neighbouring Arabian — North African subspecies christii and resembles conspecific specimens from more humid European areas. It thus seems clear that the mammalian fauna of the Harenna Forest was formed rather recently and mainly through two different processes: a limited immigration of forest species and the recruitment of species from adjacent altitudinal zones (savanna and moorland) with their subsequent adaptive speciation. Apparently, the 230 latter process prevailed being promoted both by an ecological vacuum in the forest communities and richness of the non-forest Ethiopian fauna. The degree of the adaptation to forest habitats varies among representatives of different groups, reaching its ultimate expression in the three species endemic to this forest (and Heterohyrax brucei) found in the rather narrow altitudinal range most distant from the forest margins. These adaptations result in convergence with forest species from the rest of Africa. Although the fauna of the Harenna Forest demonstrates superficial similarity with other African forest faunas, sharing with them such specialised ecological forms as Forest guenon and Arboreal hyrax it is to a large extent not related to these mammalian faunas. Acknowledgements I am indebted to the Ethiopian Wildlife Conservation Organization for permission to work in the BMNP. I thank the warden and staff of the BMNP for permission to use data from the archival files of the Park. I thank Mr. A.A. Warshavsky and Drs. V.M. Aniskin, A.N. Milishnikov and S.V. Kruskop for assistance in the field. The work was supported by a grant from the Russian Foundation of Fundamental Researches (No. 99-04-49169). Presentaion of this paper to the Symposium was made possible by travel support from the Deutsche Forschungsgemeinschaft. References Afework Bekele (1988): A census of large mammals in the Harenna Forest, Ethiopia. — Sinet: Ethiop. J. Sci. 11: 27-39. Aniskin, V.M., L.A. Lavrenchenko, A.A. Varshavy slide? Milishnikov (1998): Karyotypes and cytogenetic differentiation of two African mouse species of genus Mus (Rodentia, Muridae). — Rus. J. Genetics 34: 80-85. Capanna, E. (1985): Karyotype variability and chromosome transilience in rodents: the case of the genus Mus. — pp. 643-669 in: Evolutionary relationships among rodents (eds. Luckett, W.P. & J.L. Hartenberger). — Plenum Publishing Corporation. Carpaneto, G.M. & S. Gippoliti (1994): Primates of the Harenna Forest, Ethiopia. — Primate Conserv. 11: 12-15. Duplantier, J.M.,L. Granjon & K. Ba (1997): Repartition biogeographique des petits rongeurs au Senegal. — J. Afr. Zool. 111: 17-26. Hutterer, R.& D.W. Yalden (1990): Two new species of shrews from a relic forest in the Bale Mountains, Ethiopia. — pp. 63-72 in: Vertebrates in the Tropics (eds. Peters, G. & R. Hutterer). - Museum Alexander Koenig, Bonn. Kingdon, J. (1971): East African mammals. An atlas of evolution in Africa. Vol. 1. — Academic Press, London & New York. Lavrenchenko; L.A:, A:N. Milishnikov, V 2M. Annus eine Warshavsky & W. Gebrekidan (1997): The genetic diversity of small mammals of the Bale Mountains, Ethiopia. — Sinet: Ethiop. J. Sci. 20: 215-235. 231 marnememenwon.A.; W.N. Verheyen & J. Hulselmans (1998): Systematic and distributional notes on the Lophuromys flavopunctatus Thomas, 1888 species-complex in Ethiopia (Muridae - Rodentia). — Bull. Inst. Roy. Sci. Nat. Belg., Biologie 68: 199-214. Bavnenchenko, F.A.,A.N. Milishnikov & A.A. Warshavsky (2000): The allozymic phylogeny: evidence for coherent adaptive patterns of speciation in Ethiopian endemic rodents from an isolated montane massif. — pp. 245-254 in: Isolated vertebrate communities in the tropics (ed. Rheinwald, G.). — Bonn. zool. Monogr. 46. Lisanework Nigatu & Mesfin Tadesse (1989): An ecological study of the vegetation of the Harenna Forest, Bale, Ethiopia. — Sinet: Ethiop. J. Sci. 12: 63-93. Rahbek, C. (1997): The relationships among area, elevation, and regional species richness in neotropical birds. — Am. Nat. 149: 875-902. Rohlf, F.J. (1988): NTSYS-pce: Numerical taxonomy and multivariate analysis system. Ver. 1.40. — Exeter Publishers, New York. Stevens, G.C. (1992): The elevational gradient in altitudinal range: an extension of Rapoport’s latitudinal rule to altitude. - Am. Nat. 140: 893-911. Yalden, D. (1988a): Small mammals of the Bale Mountains, Ethiopia. — Afr. J. Ecol. 26: 281-294. Yalden, D. (1988b): Small mammals in the Harenna Forest, Bale Mountains National Park. — Sinet: Ethiop. J. Sci. 11: 41-53. Bergen? DW, MJ. Larsen, D. Kock & J.C. Hillman (1996): Catalogue of the mammals of Ethiopia and Eritrea. 7. Revised checklist, zoogeography and conservation. — Tropical Zool. 9: 73-164. Author’s address: Dr. L.A. Lavrenchenko, Severtzov’s Institute of Ecology and Evolution RAS, Leninsky pr., 33, Moscow, 117071, Russia 233 Ecology of Otomys barbouri Lawrence & Loveridge, 1953 (Mammalia, Rodentia): an endemic of the Afro-alpine zone of Mt Elgon, East Africa Viola Claus nit zen Abstract: The ecology of Otomys barbouri was studied on Mt Elgon, Uganda at an elevation of about 3700 m a.s.l. between October 1996 and November 1997 and additionally in March 1999. Otomys barbouri reached its highest population densities (22 ind./ha; equal to a biomass of about 2200 g/ha) in herbaceous and structured grasslands. The diet was entirely vegetarian, consisting mainly of grasses, different flowers (Senecio, Helichrysum, Kniphofia), and some herbs. Population dynamics and diet were studied in relation to climate and plant phenology; Otomys barbouri is non-seasonal. The microhabitat use and diets of Otomys barbouri and the sympatric O. typus are analysed and discussed in respect to fire succession in the Afro-alpine environment. On Mt Elgon O. barbouri dominates in structured and species-rich grasslands, while O. typus mainly inhabits the pure tussock grasslands; both species avoid Erica and Stoebe stands. However, such ericaceous forest would cover large areas of Mt Elgon without fire, representing an unfavourable environment. The most important habitats for Otomys, the grassland communities, are the results of various fire histories. The homogenous tussock grasslands, dominated by Festuca pilgeri, indicate a very high fire frequency. Key words: Afroalpine zone, fire ecology, rodents, population ecology, microhabitat use, Uganda Introduction “The mountains are Africa’s Galapagos Islands” (Kingdon 1989). The flora and fauna of these montane islands differ from the surrounding lowlands, since they harbour a number of endemics and species with highly disjunct distributions, as well as a number of unique life-forms. Some of these conspicuous elements can be found on virtually all of the East African mountains. Although montane forests formed a migration corridor in the last ice age, no evidence for continuous Afroalpine environments has been found so far (Hamilton 1982). The isolated mountains of East Africa have never attracted the same degree of scientific attention as their neighbouring savanna biomes. Few studies on the Afroalpine rodent communities have been carried out, e.g. in the Simien Mts., focusing on the population ecology of Arvicanthis blicki (Miller 1977). This report will concentrate on two montane species of the Groove-toothed rat (Otomys). Species belonging to the genus Otomys were widespread and common in grassland biomes in Africa during the Plio-Pleistocene (Denys & Jaeger 1986, Denys 1989). Fossils of the whole subfamily Otomyinae predominate in southern African Pleistocene deposits (Bronner et al. 1988). The present distribution of Otomys is Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 Fig.l: Microscopic view of cell fragments from a faecal sample of Otomys typus. The larger cell compartments, with long, nearly rectangular cells, are Festuca pilgeri. disjunct, mainly restricted to montane habitats in tropical Africa and to South African grasslands. These distribution patterns seem to be the results of climatic changes and competition with Murinae, which invaded Africa in the early Pliocene (Denys & Jaeger 1986, Denys 1989). Otomys barbouri 1s endemic to the Afroalpine zone of Mt Elgon. This species was described by Lawrence & Loveridge (1953), was later considered to be a subspecies of O. anchietae and only recentlywas restored to species status again (Dieterlen & van der Straeten 1992). O. typus has a disjunct distribution on various East African mountains, and is most often restricted to elevations above 2000 m. Study site Mt Elgon (4321 m a.s.l.) is an isolated massif of volcanic origin straddling the Kenya/Uganda border (0°54'-1°25'N and 34°14'—34°45'E), which was deglaciated about 12 000 years BP. The volcano rises from a plateau 1850—2000 m high in the east and 1050-1350 m in the west; the longest north-south distance is about 80 km, the longest east-west distance 50 km. The study sites were situated on the western slopes of Mt Elgon, between 3650 and 3800 m, near the Dirigana Valley. During the study period the precipitation at that elevation was 911 mm/y, April and September were the wettest months, the average temperature was 5.4° C. i) ws Nn Fig.2: Male Otomys typus with a transmitter round its neck (only the antenna is seen) for a radio- tracking study. Methods Three one-hectare grids were laid out for a mark-recapture study. Each of these grids consisted of 11 parallel lines, with 10 m spacing. The lines had trapping stations with two traps every 10 m, resulting in 121 trapping stations and 242 traps per grid. Trapping was done with Sherman live traps (Large and Extra Large) baited with a peanut-cassava flour mixture for four days each month from October 1996 to November 1997. Since both Otomys species were not attracted by any bait the traps were placed directly in their surface runway system. The difficulties of catching Otomys have been described before by various authors (e.g. Kingdon 1974, Shore & Garbett 1991). The animals were individually marked by toe-clipping, weighed to the nearest gram, measured, and the sexual condition and ectoparasites were recorded. The vegetation was mapped for all three grids, and the vertical structure measured using the point-intercept method; all plants were identified by Wesche (Wesche, in prep.). Faeces of individuals were collected from the traps and later analysed using a microscope. For identification of the cell fragments in the pellets a cell catalogue was prepared from sampled plant species. Some of the cell fragments could be identified even to species level (fig.1). 236 Individuals of Otomys typus and O. barbouri were radio-tracked for several days (fig.2), using a TRX-10002 receiver (Wildlife Materials, Canada) and MD-2C transmitters (Holohil Systems, Canada). Along a transect on the western slope of Mt Elgon sampling was done for two consecutive nights every two months (Oct. 1996—Oct. 1997) in different habitats between 2900 m and 4300 m. The traps were placed in three trap-lines (N-S direction) with 40 m spacing. Trap points were set at 20-m intervals, one small and one large break-back trap baited with peanut-cassava flour mixture were placed at each trap point. Additionally, five localities were trapped in the same pattern only once. In 4200 trap-nights, 384 rodents (13 species) were caught. The voucher specimens from this study are deposited in the Staatliches Museum für Naturkunde, Schloß Rosenstein, Stuttgart, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn and the Zoology Museum of the Makerere University, Kampala. Rodents inhabiting the Afroalpine zone A clear turnover of the vertical species composition on Mt Elgon occurs at about 3200 m a.s.l., which coincidences with the present upper limit of the montane forest (fig.3). 1, 2500 3000 3500 4000 4500 5000 Altitude (m) Fig.3: Vertical distribution of rodents on the western slopes of Mt Elgon; white bars indicate endemics. Endemic rodents are only found in the Afroalpine zone, where they inhabit ericaceous vegetation, moorlands and tussock grasslands ranging between 3300 and 4300 m. The moorlands and tussock-grasslands above 3700 m on Mt Elgon are permanently inhabited by 9 rodent species, 8 of them belonging to the Muridae (table DD). 237. Table 1: Rodents found on Mt Elgon above 3700 m a.s.l.: the commonest ones are in bold. Species Author Abundant in grassland habitats Muridae Cricetomys gambianus Waterhouse, 1840 Dendromus mesomelas (Brants, 1827) X Grammomys dolichurus (Smuts, 1832) Lophuromys flavopunctatus Thomas, 1888 X Otomys barbouri Lawrence & Loveridge, 1953 x Otomys typus Heuglin, 1877 X Rhabdomys pumilio (Sparrman, 1784) xX Tachyoryctes ruddi Thomas, 1909 x Myoxidae Graphiurus murinus (Desmarest, 1822) Population ecology Otomys typus and Otomys barbouri are the commonest rodents in the Afroalpine zone of Mt Elson. They have been recorded at densities of 32 (O. typus) and 23 (O. barbouri) individuals per hectare, giving a biomass of 3200 and 2200 g per ha respectively. A total number of 87 Otomys barbouri and 177 O. typus were marked individually on the three grids. The recapture rates of all marked specimens were 67% for O. barbouri and 72% for O. typus. In the second half of the study, of the total monthly capture at least 80% were recaptured. The home-range sizes for both species did not vary between males and females; the given estimates were calculated using the 100% convex polygon method (fig.4). n= 49 n= 87 72] 500 45] O. barbouri O. typus Fig.4: Home-range sizes tor Otomys barbouri and O. typus based on a one-year mark-recapture study at about 3700 m on Mt Elgon; 100% convex polygons. 238 The population size estimates for the different grids, calculated by the “minimum number alive” method, did not fluctuate much and not in parallel patterns between the plots over the year. There was no correlation between climatic conditions and population fluctuations. Both species showed no seasonality, reproduction occurred throughout the year. The weight gain of individual males underlines this statement (fig.5). 150 x 0 So eg ge 6 O ® | N \ Weight (¢ \ \ 50 4 : O74 oe 0 11 192 | 2 3 4 5 6 7 8 9 10 150 O Q ‘ —e O > De EE —eu—g 5 100 ‚Eur = men een — >: ‘ei Sg go ss SS SS eS SS — = pee a ee: — > z 8 9: 3 = O 0 O 0 a ZL wy, EX \ N 500 ; oe Fr 11 12 | 2 3 4 5 6 7 8 > 10 Month 1996/97 Fig.5: Weight gain or loss in individual males of Otomys barbouri (top) and O. typus (bottom); Mt Elgon, 3600-3800 m. Although one might think that the climatic conditions in the Afroalpine zone have a strong influence on the rodents, Otomys barbouri and O. typus did not show any seasonality. Reproduction, population fluctuations, or weight gain/loss did not 239 correlate with climatic variations. Both species reproduce the whole year round and no evidence for unfavourable seasons could be found. Diet compostion The faecal analyses showed no significant fluctuation in diet composition over the year (fig.6). The diet of Otomys typus consisted mainly of grasses, of which Festuca pilgeri (fig.1) made up 50%. In the pellets of Otomys barbouri all grasses together totalled only about 50%, the other cell fragments being from dicotyledons, mainly parts of Compositae inflorescenses. Flowers Compositae mel il 3 “ 5 9 0 Compositae ~ 60 _ S 50 11 12 1 2 3 4 5 6 7 8 9 10 Month 1996/97 Fig.6: Faecal analyses, based on monthly collected pellets on Mt Elgon for Otomys barbouri (top) and O. typus (bottom). 240 The genus Otomys is a highly adapted herbivore, relying exclusively on grasses and herbs (Perrin & Curtis 1980). The two species studied on Mt Elgon were strictly herbivorous: Otomys typus feeding mainly on grasses, O. barbouri on grasses and dicotyledons. Food, especially grasses, is not a limiting factor for the species on Mt Elgon. The phenology of the plants is not markedly seasonal (Wesche in prep.) and flowers, a preferred food of Otomys barbouri, are available throughout the year. ED Swamps, Herbs Rocks GH) Grasslands cap Bushes Fig.7: Daily movements of Hike milhanges a male of Otomys barbouri (4.-10.Mar.1999). Arrows indicate the morning, noon and evening fixes, starting on the morning of the 4th; Mt Elgon, 3600 m. Movements The radio-track study gave an impression of the daily ranges of the animals, although the maximum time one individual could be followed up was 9 days. All radio-tracked individuals (n=15) moved a total maximum distance of 35 m, most often moving to and fro between the same areas (fig.7). Both species occurred in the same areas, but used different microhabitats (fig.8), the range used being very restricted over the observed days. 2) Bush- & Forb-rich Grasslands Rocks Tussock-Grassland Bushes ee N N . SSonee? IN Fire influenced (SED) . Braue kein Fig.8: Ranges of a male Otomys barbouri and a see male O0 np sake br Oromys Iris -5.Mar. 1999); Mt Elgon 241 Although being very unsociable, which is the case in most Otomys species (e.g. Davis 1972, Dieterlen 1968, Vermeuen & Nel 1988), the animals do not establish territories, the home-ranges of both species and different sexes overlapping strongly. If animals meet in the field, they give a warning noise and normally each moves away. Fights were not observed in the study. Additionally, no evidence for fights, like cut ears or tails, were found. Microhabitat use Both Otomys species coexist sympatrically in large areas on Mt Elgon, but they show significant differences in their microhabitat use (fig.9). The local distribution of each species depends on vegetation composition and structure and plant species richness. The differences in distribution of Otomys barbouri on structured and species-rich grasslands and of O. typus on Festuca pilgeri dominated tussock grasslands was significant (y?-test: n= 53, p= 0,018). Fig.9 compares the abundance (trap grids) of the Elgon endemic Otomys barbouri with the more widely distributed Otomys typus. Black squares indicate the position Otomys barbouri Otomys Eınus crete GT (1 ha) See a Re N nee ee oe: ie = = = : B 5 u. : : u = ee Ba = 8 = ag = © go ag Eu u - a Ce | = [| Lu Bam GB tere | aoe a a B 1 |} | | a a | | | oso 5 ge Fs a SG a ag" 8 = = I u oe a | a a = a 8 = eq a a a EB gs a 8 = » 8 : a s 8 a See Pee oS. ee Ei Otomys barbouri Otomys typus ai ee en a = ee Secs ea : a = 8 8 3 | u. u a a B | a i as = @ 8 ou a = a » Bee eee a | 2 EB 8 s He 8 a 5 | i [| a | | a : i os © ge st & B a a u a | u a o 6 a ia : gg: 5 o 8 8 | | en eneenesraetenn ee leeeeanes B-B- BB? 1-7 2.8 Tussock Grassland Scrubs > 5m? ©) Bush- and Forb-rich Grassland (Festuca pilgeri) (Erica, Stoebe) e.g. Helichrysum, Koeleria, Carex) Fig.9: Comparison of two trap grids, upper line at 3650 m a.s.l., lower line at 3750 m. Left squares indicate the position of traps within the grids, black squares symbolize numbers of individuals of the given species caught in a trap. Data for a full year summarized. Drawings on the right indicate the corresponding vegetation structure. 242 of traps within the grid where animals were caught, sizes of the squares symbolize the numbers of individuals. The sketch map of the vegetation in the grids allows an assessment of the pattern. Otomys barbouri prefers patchy grasslands or those rich in shrubs, whereas pure tussocks are avoided. In contrast, O. typus avoids patchy vegetation and prefers tussock grasslands. Dense scrub is inhabited by neither of the species. The lower grid is a much more uniform grassland. In consequence the endemic species is rare, whereas Otomys typus 1s abundant. The comparatively small home-ranges (fig.4) make such microhabitat analysis possible. Fig.10 gives an example of the two preferred habitats of Otomys on Mt Elgon. Radio-tracked Oromys typus (n=4) moved only within the Festuca pilgeri grassland shown in the foreground. Individuals of O. barbouri (n=3) spent most of their time in the Helichrysum and Dendrosenecio bush seen in the background. Fig.10: Habitats of Otomys; O. typus is mainly found in the Festuca pilgeri grasslands shown in the foreground, O. barbouri in the Helichrysum and Dendrosenecio thicket behind; Mt Elgon, 3600 m. Fire as an important factor in landscape formation The present distribution of the different grassland formations in the Afroalpine zone of Mt Elgon, which I have discussed as the important habitats for Otomys, is a result of a long history of fire. The importance of fires in Afroalpine ecosystems is well known and was described in detail by Hedberg (1964). Mt Elgon is mentioned as having been particularly affected by man (Hedberg 1951, Beck et al. 1987). The impact of fire on the ericaceous and grassland communities on Mt Elgon were directly observed (Wesche et al. in press). Fires, which are all man- made, appear to be the major agent in replacing the natural ericaceous vegetation 243 with grassland communities. Without fires the upper ericaceous belt up to about 4000 m on Mt Elgon would naturally consist of dwarf Erica forest, with more open and shrubby communities only in bogs or on shallow rocky soils. Megaherbivores ‚like Elephant and Buffalo probably had a large impact on the vegetation, leaving open areas of different successions and sizes. Today relics of unburned ericaceous vegetation mainly survive in fire-safe rocky sites. Both Otomys species profit from the fire-induced change towards dominating grasslands. Otomys barbouri prefers vegetation mosaics such as those created by rare fires. With increasing fire frequency the vegetation is converted into pure tussock grasslands, which are avoided by O. barbouri, but inhabited in high densities by O. typus. An example for this situation is given in the lower drawing of fig.8. Without any fire, both species would be very limited in their distribution, with scattered occurrences in naturally open or patchy vegetation in a dense ericaceous forest. Acknowledgements It is a pleasure for me to thank B. Moguli and Z. Gibaba for their help and company in the field, Dr. F. Dieterlen for numerous discussions, and my friend and colleague K. Wesche. References Bessere Render, ED. Schulze & J:O. Kokwaro (1987): Alpine Plant Communities of Mt- Elgon.- An Altitudinal Transect Along the Koitoboss Route. — J. East Afr. Nat. Hist. Soc. and Nat. Mus.76: 1-12. Boomnmer bo, S. Gordon & J. Meester (1988): Otomys irroratus. — Mammalian Species 308: 1-6. Davis, R.M. (1972): Behaviour of the Vleı Rat, Otomys irroratus (Brants, 1827). — Zool. Afr. 7: 119-140. Denys, C. (1989): Phylogenetic affinities of the oldest East African Otomys (Rodentia, Mammalia) from Olduvai Bed I (Pleistocene, Tanzania). — N. Jb. Geol. Palaont. Mh. 44: 705-725. Denys, C. & J.J. Jaeger (1986): A biostratigraphic problem: The case of the East African plio-pleistocene rodent faunas. — Modern Geol.10: 215-233. Dieterlen, F. (1968): Zur Kenntnis der Gattung Otomys (Otomyinae; Muridae; Rodentia). Beitrage zur Systematik, Okologie und Biologie zentralafrikanischer Formen. — Z. Säugetierk. 33: 321-352. Bee Sa van der Straeten (1992): Species ofthe genus Otomys from Cameroon and Nigeria and their relationship to East African forms. — Bonn. zool. Beitr. 43: 383-392. Hamilton, A.C. (1982): Environmental History of East Africa — A Study of the Quaternary. — Academic Press, London. Hedberg, O. (1951): Vegetation belts of the east-african mountains. — Svensk Bot. Tidskr. 45: 141-196. Hedberg, O. (1964): Features of afroalpine plant ecology. — Almqvist & Wiksells Boktryckeri, Uppsala. 244 Kingdon, J. (1974): East African Mammals — Hares and Rodents. — Academic Press, London. Kingdon, J. (1989): Island Africa. — Princetown University Press, Princetown. Lawrence, B. & A. Loveridge (1953): Zoological results of a 5" Expedition to East Africa. I: Mammals from Nasaland and Tete; with a note on the genus Otomys. — Bull. Mus. Comp. Zool. Harvard 110. Müller, J.P. (1977): Populationsökologie von Arvicanthis abyssinicus in der Grassteppe des Semien Mountains National Park (Äthiopien). — Z. Säugetierk. 42: 145-172. Perrin, MR. & B.A. Curtis (1980). Comparative morpholocvarommne digestive system of 19 species of Southern African myomorph rodents in relation to diet and evolution. — South Afr. J. Zool. 15: 22-33. Shore, R.F. & S.D. Garbett (1991): Notes on the small mammals of the Shira Plateau, Mt. Kilimanjaro. - Mammalia 55: 601-607. Vermeuen, H.C. & J.A.J. Nel (19838): The bush Karoo ram Ozona unisculatus on the Cape West coast. — South Afr. J. Zool. 23: 103-111. Wesche, K. (in prep.): Ecology of the high-altitude environment on Mt. Elgon (Uganda/Kenya). Vegetation and climatic conditions. — Ph.D. thesis, Univ. of Marburg. Wesche, K., V. Clausnitzer, S. Miche & G. Miche Gm pressSHabtat conditions in afroalpine communities — examples from East Africa. — Proc. Ist Symp. “AFW Schimper-Stiftung”. Author’s address: Viola Clausnitzer, Zum Lahnberg 14, 35043 Marburg, Germany. e-mail: wesche @ mailer.uni-marburg.de Allozymic phylogeny: evidence for coherent adaptive patterns of speciation in Ethiopian endemic rodents from an isolated montane massif bea Eavrenchenko, A.N. Milishnikov & A.A. Warshavsky Abstract: Genetic relationships within the two presumably monophyletic species groups, Lophuromys flavopunctatus species complex and Praomys albipes - Stenocephalemys spp. assemblage, which demonstrate clear patterns of elevational replacement in the Bale Mountains, were studied using a starch-gel electrophoretic analysis of 17 loci. Data were analyzed by both phenetic and cladistic procedures, considering both allele and loci as characters. The revealed phylogenetic patterns in two independently evolving endemic assemblages showed a high level of concordance. The species of the Afroalpine zone (L. melanonyx and S. albocaudata) belong to ancestral lineages. The low-altitude populations of the heathland species (L. brevicaudus and S. griseicauda) group with the forest species (L. chrysopus and P. albipes respectively) from the same altitudes. These clusters appear as sister groups to populations of L. brevicaudus and S. griseicauda from higher altitudes. Our results indicate an early origin of the Afroalpine rodent fauna and recent descent of the Ethiopian forest rodent fauna. We speculate that L. chrysopus and P. albipes descended from still extant parental taxa (L. brevicaudus and S. griseicauda respectively) by “explosive” adaptive evolution. Key words: Lophuromys, Stenocephalemys, Praomys, Ethiopia, allozymes, phylogeny, evolution Introduction The diverse and unique Ethiopian small mammal fauna demonstrates clear patterns of an altitudinal zonation, including some examples of elevational replacement of congeners (Rupp 1980, Yalden 1988). Three endemic species belonging to the Lophuromys flavopunctatus species complex, L. chrysopus, L. brevicaudus and L. melanonyx, occur in the Bale Mountains, replacing each other in the different altitudinal belts (tropical forest - heathland - Afroalpine zone) with very little overlap (Lavrenchenko et al. 1998). This altitudinal distribution closely matches the distributional pattern of another presumably monophyletic group of closely related endemic murides, including Praomys albipes, Stenocephalemys griseicauda and §. albocaudata, which inhabit the same vegetation zones respectively in the Bale Mountains (Yalden 1988). The habitat segregation occurring among these closely related taxa suggests an adaptive pattern of speciation. The present study used allozyme data to reconstruct the phylogenies of these two presumably monophyletic assemblages. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 Material and methods Sampling. Field work was carried out in the course of the Joint Ethio-Russian Biological Expedition (JERBE) between January and April 1995. Specimens were collected at the following localities in the Bale Mountains National Park (fig.1): (1) Podocarpus belt of the Harenna Forest (1780 m a.s.l., 6°31'N, 39°44°B) - 277; chrysopus, 3 P. albipes; (2) Aningeria belt of the Harenna Forest (1935 m a.s.l., 6°38'N, 39°44'E) - 2 L. chrysopus, 3 P. albipes; (3) Schefflera-Hagenia belt of the Harenna Forest in the Katcha area (2400 m .a.s.l., 6°42'N, 39°44'B) - 8 L. chrysopus, 2 L. brevicaudus, |3 2. aloipes aes: griseicauda, oO 5) 46 1s Fig.1: Map of the Bale Mountains National Park, showing trapping sites. See text for abbreviations of localities. 247 (4) Mosaic grassland/forest habitats in the Dinsho area (3170 m a.s.l., 7°06'N, 39°47'E) -33 L. brevicaudus, 2 P. albipes, 5 S. griseicauda: (5) Ericaceous belt of southern slope of the Sanetti Plateau (3700 m a.s.l., 6°46'N, 39°46'E) - 35. albocaudata: (6) Ericaceous belt of northern slope of the Sanetti Plateau (3750 m a.s.l., 6°54'N, 39°55'E) - 2 L. brevicaudus; (7) Afroalpine belt with sparse and short vegetation in Konteh area of the Sanetti Plateau (4050 m a.s.l., 6°51'N, 39°53'E) - 21 L. melanonyx, 6 S. albocaudata; feericaccous belt in Chorchora area (3500 m a.s.l., 6°56'N, 39°56'E) - 19 L. brevicaudus, | S. griseicauda, 2 S. albocaudata; wes wamp Shore areas in Kotera (3500 m a.s.l., 7°O00'N, 39°41'E) -15 L. brevicaudus, 3 L. melanonyx, | S. griseicauda, 3 S. albocaudata. Electrophoretic analysis of proteins. Standard horizontal starch-gel electrophoresis and staining techniques, as described by Selander et al. (1971) and Harris & Hopkinson (1978), were performed to analyze genic variation in 17 enzymatic and non-enzymatic proteins. The enzymatic proteins (their respective abbreviations and Enzyme Commission Numbers are given in parentheses) were alcohol dehydrogenase (ADH; 1.1.1.1); diaphorase (DIA-1, DIA-2; 1.6.2.2); glucose-6-phosphate dehydrogenase (G-6PD; 1.1.1.49); glutamate-oxaloacetate transaminase (GOT; 2.6.1.1); glicerophosphate dehydrogenase (GPD; 1.1.1.8); isocitrate dehydrogenase (IDH; 1.1.1.42); lactate dehydrogenase (LDH-A, LDH-B; 1.1.1.27); malate dehydrogenase (MDH-1, MDH-2; 1.1.1.37); malic enzyme (ME-1; 1.1.1.40); sorbitol dehydrogenase (SDH; 1.1.1.14); superoxide dismutase (SOD-1, SOD-2; 1.15.1.1). The non-enzymatic proteins were haemoglobin (HBB) and albumin (ALB). Phylogenetic analysis. The conspecific samples of L. chrysopus and P. albipes from adjacent localities 1, 2 and 3, and S. albocaudata from localities 5 and 7 were pooled basing on the analyses of homogeneity. The genetic divergence among populations and species were evaluated by Nei’s (1978) and Rogers’s (1972) genetic distances. Phylogenetic trees were computed from genetic distances using the Fitch-Margoliash least squares method (FM trees), neighbor-joining method (NJ trees) (PHYLIP v.3.5c (Felsenstein 1993)) and distance-Wagner procedure (DW trees) as employed in the BIOSYS-1 package (Swofford and Selander 1989). Cladistic analysis was performed using both loci and alleles as characters. In the former case we treated the most common allele as the character state and utilized Fitch parsimony criteria; both Wagner and Dollo parsimony criteria were applied in the latter case (PHYLIP v.3.5c). Bootstrap analyses involving at least 500 replications were conducted to assess the reliability of the branching patterns of FM and parsimony trees. Mus cf. triton, M. mahomet and two cryptic Otomys species from the Bale Mountains (for details, see Lavrenchenko et al. 1997) were used as outgroups in the phylogenetic analyses of the Praomys albipes - Stenocephalemys spp. assemblage and the Lophuromys flavopunctatus species complex respectively. serennenesecsennncenecoonenesnesneete 352100008415.1006013018658012.00280190610610836:000100650600001001.655001.6001.066100003006:0866061086300650001.06600.0.00:00550016 50 ee 248 Results Marked intraspecific genetic differentiation in L. brevicaudus, S. griseicauda, and S. albocaudata was revealed. The values of Nei’s unbiased genetic distance between some samples of these species (table 1, 2) are above those normally encountered for conspecific populations (D < 0.105; Thorpe 1983). The genetic distances between L. brevicaudus from locality 6 and all other “brevicaudus” populations fall within the value usually recorded for interspecific genetic differentiation in rodents. The mean genetic distance between the sample of L. brevicaudus from locality 9 and three remaining conspecific populations was D = 0.110, which corresponds to the genetic differentiation of rodent allopatric species. Table 1: Values of Nei’s genetic identity (above the diagonal) and genetic distance (below the diagonal) between Lophuromys populations. Lb. =D. brevicaudus; Uc. — LE. chm sopus. Em. _ 1. melanony», Ors Or Ce On ao) Ot.58 = Otomys (2n=58). ETC ONTLLONTOTTNLINLTONITTTTTTNLTNTOTTTNCTTNTTNTNLTTTNTNTETTLNTLTETTTENTTLLINTELLETTTKONLOITTETONT nace NTITETENETTLETTTTITTTNTEHLITLTTTOTNITONTTTTONTTTNTTTLIONLTNIONTETTLITNTOLTTNTITTTNTLTOTTITLTTTOKTOTNTNTIOHTNTLINTTTLTETLTOTTTIDNTNTOTTATOTTNTTITTHNTTTTTTTTTT Population U 8 9 10 12.9162 DNA 0.791 0.743. 70,21672092532 2. 1820.26 0.682 0.661 0241720980 SA aie: 0.750 0.779 021470951 Ab ES DO) 0.778 0.8307 OO AG Sl bE 3 0.8217 0.7522. 002257209298 Once I Brave nl 1238) 0.593. 0. 5312 022550187] u Wes inal 1 “+ 0.948720 2909 O31 8. Em. 9 0.054) Fr 028270 9.7:08.56 1.219 1.2070 02 22.097790 I TOEBS 1.146. 1.168. 00309 02 5 BERTRETRURDLLERRURERREERERERERTKERRERENKURKEERVERLERUERURDERRELRTEETEEEREERELRREREEERT DER KORLERREEERRERLRELEEENG caconoseennte: Table 2: Values of Nei’s genetic identity (above the diagonal) and genetic distance (below the diagonal) between Stenocephalemys and Praomys albipes populations. P.a. =P. albipes: S.g.=S. griseicauda: S.a. - S. albocaudata: Mim. —Viamahnomer Mn — Mus cf. triton. 2 3 4 5 6 7 8 9 Vion 1] 3 ERRENENTETTTTREH HT ROG ISH T OT OST ARATE I CT OA TEN rR OR RR RE TD TG OR TOLER Pa se 22.0.9990 088910 940 0. 654 “0.733 0.665 0.708 0.3353, 0065 a ONES DP 424 0.001 ***** 0.881 0.940 0.642 0.723 0.653 0.696 05200 0 0COm Ose 3 8.8.4 0.118 0.126 ***** 0944 0.783 0.864 0750), 0785 0.6053 O00 BD A S248 0.062 0.061 0,057 ***** 0.706 0.788 0.718 0.7607 70588, 00607750348 38.8.9 0.425 0.443 0.244 0.348 ***** 0.970 0.718 0.790 0.689 0.060 0.118 OS. 8 0.311 0324 0.146 0.238 0.030 *****" 10.709 0.784 0.679, 096770321 7. Sanaa 0.408 0.426 0.288 0.331 0.331 0.343 ***4* 0.827 0.9207 74270320 88.2.9 0.345 03653 02422 0274. 0236 0244 0.1%0 = 0.12 2 928.318 0.630 0.655 0.489. 0.531 0373 0387 0477 0337 N u, ONS A OF MIE en 12.734 2.806 2.789) 28201 2827 2.797 21108 a 20 a 0.417 WE ma | 2.130 2.125 2.108 2.140 2.140 2110 2.23 7 2.1277) ETO 1 0 Goran SPR RNR EISEN RE SSS SLT NN EEE sosnneansensce 249 species. Intraspecific genetic distances between all three samples of S. albocaudata and those between two groups of S. griseicauda (localities 3,4 versus localities 8,9) fall within the range generally recorded for interspecific genetic differentiation in African murines. On the other hand, it is amazing that the genetic distance between P. albipes and S. griseicauda from locality 3 and 4 corresponds to a genetic differentiation between subspecies or allospecies. Furthermore, the value of Nei’s D between P. albipes and any other population of Stenocephalemys falls significantly below the level usually recorded for taxa belonging to different confamilial genera. The DW dendrogram based on Rogers’s genetic distances and the trees computed from Nei’s and Rogers’s genetic distances using the FM and NJ methods demonstrates identical relationships among the Lophuromys at the population level (fig.2). The L. melanonyx branches basally from the L. chrysopus - L. brevicaudus clade, bootstrap value being 70%. The L. brevicaudus may be paraphyletic since the bootstrap analysis provides a modest support (52%) for a clade including L. chrysopus and populations of L. brevicaudus from localities 3, 4, 6 and 8, but not the conspecific population from locality 9. After all small samples (n < 5) had been excluded from the analysis, the bootstrap support for the position of these two branches markedly increased (88% and 64% respectively). Here, L. brevicaudus from locality 4 is linked to L. chrysopus (bootstrap value 52%). 0.00 | p.4 Fig.2: Fitch-Margoliash tree computed from Nei’s genetic distances between Lophuro- L.c.1,2,3 mys populations. Numbers above a line are distances; below are the bootstrap values (those in parentheses refer to the analysis with all small samples (n < 5) exclu- ded). The tree is rooted at the composite outgroup. See table 1 for abbreviations of 0.09 Ot.58 populations. 0.01 [5.3 0.23 NJ, FM and DW trees reveal that the genetic relationships among populations in the Praomys albipes - Stenocephalemys assemblage seem to be concordant with that two most basal splits of the tree, whereas there are two cladistically distinct groups of populations within the latter species. The first of them forms a clade with a of Lophuromys but are more complex (fig.3). Both S. albocaudata and S. griseicauda appear paraphyletic; two populations of the first species represent the bootstrap support of 77%, while the second is a part of the monophyletic group comprising the two populations of Praomys albipes as well (the bootstrap value for this association is 86%). Within the latter branch, the clustering of S. griseicauda from locality 3 and Praomys albipes is supported by a very high number of bootstrap replicates (97%). So the branching patterns of the two Species assemblages coincide in three respects: Fig.3: Fitch-Margoliash tree computed from Nei’s genetic distances between Steno- cephalemys and Praomys albipes populations. Numbers above a line are distances; below, the bootstrap values. The tree is rootede arzchte composite oUtELOoUp Bor abbre-viations of populations see table (1) the species of the Afroalpine zone (L. melanonyx and S. albocaudata) occupy basal positions in the phylogenies; (2) the youngest lineages are represented by the forest species (L. chrysopus and P. albipes); (3) the heathland species (Z. brevicaudus and S. griseicauda) may be paraphyletic. Indeed, there are preliminary indications that our samples of S. griseicauda include more than one species (Lavrenchenko et al. 1997). A discussion on this point will be published elsewhere. By contrast, our cytogenetic (Aniskin et al. 1997) and morphological (Lavrenchenko et al.1998) analyses support the conspecificity of all our samples of L. brevicaudus. Therefore we carried out parsimony analyses at the Species level ewinh populations being pooled together. The results of the two analyses treating both loci and alleles as characters were fully compatible with those based on the distance data. L. melanonyx diverges basally whereas L. chrysopus and L. brevicaudus are sister groups. The high bootstrap values indicate significant support for this branching pattern (fig.4). Fig.4: The most parsimo- nious tree sesultimesinom cladistic analyses of Lophu- romys. species branch lengths are drawn propor- tionately to those resulting from the analysis by locus using Fitch parsimony. Synapomorphies for the analysis by locus (changes of the most common allele in’ a locus) are sndreawed along the branches. Circled numbers, aces bOoOrsimEap values for the analyses by locus and by allele, the latter given in parentheses). 251 Discussion The most surprising result of our phylogenetic analyses — which, we believe, is not an artifact of sampling — is the association of low-altitude populations of the heathland species (L. brevicaudus and S. griseicauda) with the forest species (L. chrysopus and P. albipes respectively) from the same altitudes, so that the clusters thus formed appear to be sister groups to populations of the heathland species from higher altitudes. There are three possible theoretical explanations of this phenomenon: (1) possible limited hybridization between sympatric populations of the related species; (2) a high level of homoplasy (parallelism) in allozymic variation caused by climatic factors obscuring true patterns of relationships; (3) the explosive nature of speciation of the forest species. The first possibility seems unlikely in view of the lack of individuals heterozygous at a number of diagnostic loci, and chromosomal characteristics (Lavrenchenko et al. 1997) of both species assemblages from narrow overlap zones. Although the second possibility is difficult to exclude completely, we note that the allozyme expression patterns demonstrate no clear pattern of correlation with geoclimatic variables both in forest and heathland species. As fas as the Stenocephalemys complex is concerned, the difference between the two groups of populations of S. griseicauda is too high (table 2) to be accounted for by a local adaptive shift of allele frequencies. Therefore, the hypothesis postulating “explosive” evolution of L. chrysopus and P. albipes, both having descended from still extant parental taxons (L. brevicaudus and S. griseicauda respectively) seems to be favourable. It is worth mentioning that these forest species are the most divergent forms in their species assemblages with respect to chromosomal characteristics (Lavrenchenko et al. 1997). In addition, P. albipes was so divergent morphologically from Stenocephalemys spp. that it was even attributed to another genus. These evidences suppose an accelerated rate of chromosomal and morphological evolution under speciation of P. albipes and of chromosomal and genetic evolution of L. chrysopus. The morphological differentiation may be at least partly explained by adaptations to a forest environment. It has been suggested (Nevo 1985) that chromosomal repatterning may facilitate adaptive evolution (e.g. by providing new superior supergenes for ecotypic adaptation). In general, evolution of the two forest taxa may be interpreted within the framework of the invasive model of chromosomal speciation (White 1982). The recent origin of the endemic L. chrysopus and P. albipes from aboriginal lineages suggests that the Ethiopian forest rodent fauna has no direct relation to those of Central Africa. The resemblance of these two Ethiopian taxa to “true” forest species of Lophuromys and Praomys from the rest of Africa may be aresult of convergent evolution. The concordant phylogenetic patterns within two independently evolving endemic assemblages provide evidence of similar speciation histories and can be related to significant climatic and environmental changes during the Pliocene-Pleistocene. We presume that differentiation of the immediate ancestors of the two groups studied (probably similar morphologically to L. brevicaudus and S. griseicauda) occurred during the late Pliocene (2.4 Mya), when heathland had become established on the Ethiopian plateaux (Yalden & 252 Largen 1992). The specialized L. melanonyx and S. albocaudata adapted to the high altitudes apparently represent Lower Pleistocene splits connected with the formation of Afroalpine habitat. The recent origin of L. chrysopus and P. albipes may be associated with some humid period of the Pleistocene, when the treeline in the mountains was rising. It is worth mentioning that in the Bale Massif, in Pleistocene times, a significant area was covered by glaciers; their terminal moraines are at 3100 -3200 m a.s.l. (Messerli & Winiger 1980). A spatial fragmentation having resulted from such conditions may have faciliated the speciation events. On the other hand, an isolation of local populations by glaciers during some periods of the Pleistocene may explain the very high level of interpopulational genetic differentiation in L. brevicaudus, S. griseicauda and S$. albocaudata and their possible paraphyly. We can conclude that sharp altitudinal zonation, extremely diverse geomorphology, drastic environmental changes in the past, and the isolated position of the Bale Massif has made this unique area a real center of diversification and endemism for Ethiopian rodents. To clarify in more detail the interrelationships among these species, further analysis taking into consideration, an extended set of allozyme loci or other semenicraaiaums warranted. Acknowledgements We are indebted to the Ethiopian Wildlife Conservation Organization for permission to work in the Bale Mountains National Park. The work was supported by the Russian Foundation of Fundamental Researches (Grants NN. 99-04-49169 and 98-04-48566) during the stage of laboratory research. Presentation of this paper to the Symposium was made possible by travel support from the Deutsche Forschungsgemeinschaft. References Aniskin, V.M., L.A. Lavrenchenko, A.A. Varshauskin ar NE Milishnikov (1997): Karyotypic differentiation of three harsh-furred mouse species of genus Lophuromys (Murinae, Rodentia) from the Bale Mountains National Park, Ethiopia. — Russ. J. Genetics 33: 818-824. Felsenstein, J. (1993): PHYLIP Phylogeny Inference Packasen vets er University of Washington. Harris, H. & D.A. Hopkinson (1978): Handbook of enzyme electrophoresis in human genetics. — North-Holland Publ. Comp., Amsterdam; Amer. Elsevier Publ Comp, New York. Lavrenchenko L.A., W.N. Verheyen & J. Eulselmansı Gio: Systematic and distributional notes on the Lophuromys flavopunctatus Thomas, 1888 species-complex in Ethiopia (Muridae — Rodentia). — Bull. Inst. Roy. Sci. Nat. Belg. (Biol.) 68: 199-214. Lavrenchenko, L.A., A. N. Milishnikoev, V.M2 Anisikeineee eee Warshavsky & W. Gebrekidan (1997): The genetic diversity of small mammals of the Bale Mountains, Ethiopia. — Sinet: Ethiop. J. Sci. 20: 213.233. 258 Messerli, B. & M. Winiger (1980): The Saharan and East African upland during the Quaternary. — pp.87-118 in: The Sahara and the Nile (eds. Williams, M.A.J. & H. Faure). — Balkema, Rotterdam. Nei, M. (1978): Estimation of average heterozygosity and genetic distance from a small number of individuals. — Genetics 89: 583-590. Nevo, E. (1985): Speciation in action and adaptation in subterranean mole rats: patterns and theory. — Boll. Zool. 52: 65-95. Rogers, J.S. (1972): Measures of genetic similarity and genetic distance. — Stud. Genet. 7. Univ. Texas Publ. 7213: 145-153. Rupp. H. (1980): Beiträge zur Systematik, Verbreitung und Ökologie Beopıscher Nagetiere: Ersebnisse mehrerer Forschungsreisen. — Säugetierkdl. Mitt. 28: 81-123. Padi Ko, Vr. Smith, S.Y. Yane, W.E. Johnson & J.B. Gentry (1971): Biochemical polymorphism and systematics in the genus Peromyscus. |. Variation in the old-field mouse (Peromyscus polinotus). — Stud. Genet. 6. Univ. Texas Publ.7103: 49-90. Savononrd. P-L. & R.B. Selander (1989): BIOSYS-1: a computer program for the analysis of allelic variation in population genetics and biochemical systematics. Release 1.7. — Illinois Natural History Survey, Champaign. Thorpe, J.P. (1983): Enzyme variation, genetic distance and evolutionary divergence in relation to levels of taxonomic separation. — pp.131-152 in: Protein Polymorphism: Adaptive and Taxonomic Significance (eds. Oxford, G.S. & D. Rollinson). — Academic Press, London. Bean ID (1982): Rectaneularity, speciation and chromosome architecture. — pp. 75-102 in: Mechanisms of Speciation (ed. Barigozzi, C.). — Alan R. Liss, New York. Yalden, D. (1988): Small mammals of the Bale Mountains, Ethiopia. — Afr. J. Ecol. 26: 281-294. Yalden, D.W. & M.J. Largen (1992): The endemic mammals of Ethiopia — Mamm-Rev. 22: 115-150. Authors’ address: Dr. L.A. Lavrenchenko, Severtzov’s Institute of Ecology and Evolution RAS, Leninsky pr., 33, Moscow, 117071, Russia 155) In in Primates in Eritrea - distribution and habitat DeeZinnen E. Korkler & R Peläez Abstract: Eritrea, the former northern part of Ethiopia, is one of the youngest countries in the world since it became independent from Ethiopia in 1993. Due to several decades of civil war, information on status and distribution of wildlife and birds in Eritrea is outdated and incomplete. A stock-taking program of biological resources and biodiversity is one of the aims of the Eritrean government. We assisted in the program by doing a survey on primate (Papio hamadryas, Papio anubis and Cercopithecus a. aethiops) distribution and abundance in eastern and central Eritrea. Information obtained during the study will be integrated in a national wildlife management and conservation plan in Eritrea. We interviewed local people using a questionnaire to gather information about baboon (Papio sp.) and Grivet (= Vervet) monkey (Cercopithecus a. aethiops) sleeping sites, and about predator and wildlife presence. The sleeping sites were then visited to the nearest possible point and their position was recorded using a GPS. We took notes on the surrounding vegetation and the nearest watercourse. We also categorized vegetation at around 100 selected sites, independently of the primate sites, as reference points for a subsequent classification of Landsat MSS satellite data. We included the geographical position of sleeping sites, other evidence of baboons and monkeys, and information on the vegetation in a subsequent GIS analysis and mapping. Habitats were described by altitude, precipitation, and vegetation cover and plant productivity. In our survey area, we found 78 sleeping cliffs of Hamadryas baboons, 8 sites with Olive baboons and 43 sites with Grivet monkeys. A clear geographical separation between the two Papio species became obvious, whereas Grivet monkeys are sympatric with both baboon species. In contrast to other wildlife, the current distribution of baboons and monkeys seems not to deviate from their historical distribution and the proposed national parks will include several of the primate sites. Key words: Eritrea, Papio hamadryas, Papio anubis, Cercopithecus a. aethiops, distribution, ecology Introduction The central highlands of Ethiopia and Eritrea were refuge areas for plant and animal species during glacial times and have had a very profound effect upon the evolution, composition and distribution of the flora and fauna of NE Africa. Here we can find a unique assemblage of montane species, including many endemics. In the surrounding drier lowlands, a mixture of species of different zoogeographic regions meet: species of the East African savannas, of the sub-Saharan savanna belt, of the North African deserts and the Somali arid zone. Even species belonging to the Palaearctic Region are present in the area (Yalden et al. 1996). The number of mammalian taxa recorded from Ethiopia has been rising steadily during the past 30 years, as aresult of both taxonomic revisions and new discoveries. Yalden et al. (1996) recognized 277 terrestrial and 11 marine species, but state that these numbers are still provisional. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 256 Most of the biogeographic work of recent decades in Ethiopia was confined to the politically more stable central parts of the country, so information on species status in the northern areas is outdated and incomplete. In 1991, after a 30-year-long struggle for independence, the northern part of Ethiopia, the former Italian colony Eritrea, became an independent state. Presently the Ministry of Agriculture and the Department of Environment of Eritrea are preparing a National Biodiversity Strategy and Action Plan, but most of the relevant information on species distribution and abundance is missing. Actually for most of the taxa information was never collected or derives from sources older than 50 years. To overcome this serious data deficiency, the Ministry of Agriculture started a series of wildlife surveys. We focused on primates and’ gathered information about their distribution and abundance in central and eastern Eritrea, particularly Hamadryas baboons. Additionally, we were interested in the gross ecological condition of primate habitats and their variation by ecological zone and species. Information obtained during our study will be integrated in a national wildlife management and conservation plan in Eritrea. Eritrea Eritrea is located in NE Africa on the coast of the Red Sea with an area of 120,000 km’. It holds an estimated human population of 3.5 million. Eritrea encompasses several ecological zones which differ in altitude and distance from the Red Sea, and thus in annual rainfall and plant productivity. From east to west we find the coastal lowlands that stretch up to 40 km inland, with their south-eastern extension, Dankelia, an inhospitable, volcanic area with low rainfall and very high temperatures. The lowlands are followed by a sudden rise within 40 km up to 3000 m, the eastern escarpment. Next come the central highlands and the Hamasien plateau, including the catchment areas of the Anseba and Mareb rivers, the main area of settlement and agriculture. The terrain in the western escarpment drops steeply down again from 2200 m to 1000 m into the western lowlands, which extend in the west to the Sudanese border, there reaching an altitude of 500 m. Primate species There are historical reports of the occurrence of six primate species in Eritrea: Galago sp., Colobus guereza, Cercopithecus a. aethiops, Theropithecus gelada, Papio anubis and Papio hamadryas. The presence of Colobus is very doubtful, because there is almost no adequate habitat in Eritrea, and Galago and Theropithecus were not confirmed (Yalden et al. 1977). Method Survey We conducted our survey during October-November 1997 and March—April 1998 and October 1998. The survey covered more than 25 000 km’ (16°19' to 14°17'N, 37°13' to 39°53'E) with an additional extension to the west up to the Sudanese border (36°16'E) to include the Olive baboon (Papio anubis) region. We 231 interviewed local administrators and people in villages using a questionnaire to gather information about baboon (Papio sp.) and Grivet (Cercopithecus a. aethiops) sleeping sites, predator and wildlife presence. The people are normally very well informed about baboons and grivets in their neighbourhood, because in most areas baboons and monkeys are recognized as a threat to crops and gardens. We visited the sleeping sites of the baboons and monkeys to the nearest possible point and recorded their position and altitude by GPS. Whenever possible we tried to estimate the number of primates using the sleeping site and we took notes on the surrounding vegetation for a subsequent food analysis, and the nearest watercourse. Additionally, we recorded all other sites where we found evidence of primates. Primate habitats in Eritrea In our ecological description of primate habitats in Eritrea, we used the following parameters. — Precipitation: data on annual precipitation (mm y’') derive from the National Map of Eritrea 1:1,000,000 and from meteorological stations of the Ministry of Agriculture. — Altitude: the altitude and position of sleeping sites and other points of interest was directly determined by GPS. Average altitude of home-ranges and maximum elevational difference in home-ranges were calculated from an elevation model of Eritrea (pixel size: 0.9 x 0.9 km) to the nearest 100 m (U.S. Geological survey 1991): N N & I G A P. hamadryas @ P anubis NZ Eco-Zones cl = coastl. lowl. ee = east. escp. ch = centr. highl. Ethiopia we = west. escp. wl = west. lawl. 0 100 200 300 km Fig.1: Distribution of Hamadryas and Olive baboons (Papio hamadryas, Papio anubis) in central Eritrea. 258 — Normalized Difference Vegetation Index (NDVI): the NDVI is a measure of the amount and vigor of vegetation at the surface. The magnitude of NDVI is related to the level of photosynthetic activity in the vegetation. In general, higher values of NDVI indicate greater vigor and amounts of vegetation. Thus areas with higher NDVI normally have a higher plant productivity. Pixel size after rectification was 7.6 x 7.9 km. We used 10 years of dekadal data, starting in 1982, to calculate the average NDVI for each pixel of the terrestrial area Eritrea (U.S. Geological Survey QOD). — Vegetation classification: a vegetation classification of the survey area was carried out. We used the White (1983) classification for African environments with additional categories which seem to be important for baboons. Apart from the sites where we encountered primates, we categorized vegetation at around 100 selected sites as reference points for a subsequent classification of Landsat MSS satellite data (Zinner & Torkler 1996). Our classification corresponds reliably with official data from the Ministry of Agriculture. GIS analysis For an ecological description of primate habitats we encircled each baboon site with a buffer of 3 km radius, thus producing a 28.3 km’ area as a hypothetical = C aethiops ING Eco-Zones cl= coastl. lowl. ee = east. escp. ch = centr. highl. we = west. eScp. wl= west. lowl. Ethiopia Fig.2: Distribution of Grivet monkeys (Cercopithecus a. aethiops) in central Eritrea 299 A Aaa se Soin) aie ; N N aM Rn A \ N — “ A Ney 3 A \ \ Se \ aN 9 A { A ‘ / a N = / am 8 Die, a / ® a be oo ee g a mae un et Sn a P.hamadryas = if NG er TS el Sumer vs ee N e P.anubis EN S @ N cS =» C. aethiops a =e \ N Ethiopia . ma @aa te % a N \ = ya ° 0 100 200 300 km Me tes 3 | 2 en \ i = ® ei ER un oe a @ SS RR A EN pit & & E | = 8 e m Fig.3: Historical distribution of Papio hamadryas, Papio anubis and Cercopithecus aethiops aethiops in northern Ethiopia and Eritrea (Yalden et al. 1977). home-range of the group. Similarly, grivet sites were provided with a 0.5 km radius buffer. These home-range sizes correspond with data from other baboon and grivet studies (Cheney 1987). We included the geographical position of primate sites and information on the vegetation in a subsequent GIS analysis and mapping. Primate habitats were then described by their ecological parameters. We compared the habitats of the three species and primate habitats in the five different eco-zones. Results We found 78 sleeping cliffs of Papio hamadryas at all altitudes of all ecological zones, from the coastal area to the western lowland (fig. 1). In 49 cases, we were able to estimate the number of individuals that congregate at the sleeping cliff. In the south-western part of the western lowlands, Olive baboons replace Hamadryas baboons. Here we encountered 8 small groups of Olive baboons (6—28 individuals). We detected Grivet monkeys at 43 different sites in all ecological zones except the coastal lowlands (fig.2). Hamadryas and Olive baboons have exclusive ranges, whereas Grivet monkeys are sympatric with both baboon species. Overall, the current distribution of Papio hamadryas and Papio anubis seems not to deviate from their historical distribution as reported by Yalden et al. (1977) (fig.3). The precipitation and elevation data of primate home-ranges in Eritrea are given in table 1. Hamadryas baboons live in areas with less than 200 mm up to 1000 mm precipitation. 260 Table 1: Ecological parameters of Papio hamadryas (P.ham.), Papio anubis (P.anu.) and Cercopithecus a. aethiops (C.aeth.) home-ranges in the five ecological zones of Eritrea. Precipitation (mm va) altitude of site (m), maximum elevational difference within home- ranges (m). Eco-zone species n precipitation altitude of record (m) max. elevational (mm y'') difference in HR (m) coastal lowland eastern escarpm. central highland western escarpm. western lowland coastal lowland eastern escarpm. central highland western escarpm. western lowland western lowland P.ham. P.ham. P.ham. P.ham. P.ham. C.aeth. C.aeth. C.aeth. C.aeth. C.aeth. P.anu. 100-600 100-1000 400-900 400-800 350-400 400-1000 400-600 400-800 700-800 400-800 33-393 350-2300 1491-2736 830-2268 656-750 850-2064 1511-2575 975-2049 788-1030 100-700 200-1400 200-1200 100-1000 300-400 0-500 0-600 0-400 0-100 100-300 601-853 083 DL] eco-zone [22 P. hamadryas 0,30 LI C.aethiops BES P_ anubis NDVI 0,00 -0,05 | -0,10 coast. lowl. east. escp. centr. highl. west. escp. west. lowl. Fig.4: Normalized Difference Vegetation Indices (NDVI) of Eritrean eco-zones and of Papio hamadryas, Papio anubis and Cercopithecus a. aethiops home-ranges (means and ranges, single mean test; Papio hamadryas: coastal lowland n=7, eastern escarpment n=24, central highlands n=30, western escarpment n=13, western lowland n=4; Papio anubis: western lowland n=8; Cercopithecus a. aethiops: eastern escarpment n=11, central highlands n=20, western escarpment n=9; western lowland n=3. NDVI classes: -0.162 to —0.029 = very poor vegetation, middle; -0.028 to 0.105 = very poor vegetation, high; 0.106 to 0.155 = poor vegetation, low; 0.156 to 0.205 = poor vegetation, high; 0.206 to 0.259 = middle vegetation, low; 0.260 to 0.313 = middle vegetation, high). 261 Interestingly, in the western lowlands we detected them in the drier areas, whereas Grivets and Olive baboons live in more humid areas. Hamadryas baboons were found from sea level up to almost 3000 m. At the western border of their distribution, they occupy altitudes above 650 m. Olive baboons live in areas below the 900 m contour line. Grivets occupy similar altitude as the baboons. In the mountainous areas, the maximum elevation difference within Hamadryas home- ranges can reach up to 1400 m. In some of their home-ranges these baboons have to surmount these elevation difference on a daily basis, when they go down to the water source and up again to their sleeping cliff. The mean Normalized Difference Vegetation Indices (NDVI) for primate home- ranges in comparison with NDVIs of eco-zones are given in Fig. 4. Eritrea has only poor NDVIs, with the poorest condition in the coastal lowlands. In most eco-zones, primate home-ranges have an NDVI significantly higher than the respective eco- zone mean. Hamadryas baboons seem to be the most tolerant species. Their mean NDVIs for the coastal and western lowlands are the lowest of all primate home- ranges. A comparison of species reveals that in the eastern escarpment, the central highlands and the western escarpment Hamadryas baboons and Grivet monkeys use areas with similar mean NDVIs. In the western lowlands, however, the discrepancy between Hamadryas baboons and Olive baboons and Grivet monkeys becomes 45 [__] area of survey 40 [| P hamadryas Ap BET. aethiops 35 30 25 20 I u 15 Mn 10 ob OM i ni Fc FS We Ws Bu Gr Bs Ar Op Ot Fig.5: Proportion of vegetation and ground-cover classes in Eritrea (area of survey) and in home-ranges of Papio hamadryas and Cercopithecus a. aethiops. Area surveyed: 34087 km’; P. hamadryas n=68, 1923 km’; C. aethiops n=35, 27 km’. Fe = closed/medium-closed forest; Fs = open forest; Wc = closed/medium-closed woodland; Ws = open woodland; Bu = bush-/shrubland; Gr = grassland/wooded grassland; Bs = barren soil; Ar = agriculture; Op = Opuntia spp.; Ot = other (settlements, rivers, lakes). 262 obvious. On average, Olive baboons and Grivet monkeys occur in areas with higher NDVIs than Hamadryas baboons (P. anubis: t=2.06, p=0.0659; C. aethiops: t=4.27, p=0.0079). No difference was found between mean NDVIs of Grivet monkey areas and Olive baboon areas. The result of the vegetation classification of the eastern parts of our study area are given in fig.5. On average, the vegetation in Hamadryas baboon home-ranges was 2 % closed or medium-closed forest, 5 % open forest, 8 % closed woodland, 12 % open woodland, 23 % bush and shrubland, 8 % grassland and wooded grassland, 14 % barren soil and rocks, 16 % agriculture and 9 % prickly pear stands (Opuntia ficus-indica). Others, such as settlements and inland waters, represented less than 0.1 %. We could not classify 3 % of the satellite images because of cloud and shadow. A similar picture emerged for the grivet home-ranges, with one exception. In their home-ranges, open woodland was significantly more frequent than in baboon home-ranges. If we compare the overall vegetation distribution in our area of survey with the vegetation in primate home-ranges, it becomes obvious that both species select more wooded areas and reject the very open areas like grassland and barren soil. The number of Hamadryas baboons encountered at sleeping cliffs varied from 25 to 800, with a mean of 190. From this figure, we estimate a population of at least 15 000 Hamadryas baboons in our survey area. Discussion During a five-month survey, we gathered information on distribution and abundance of primates in central and eastern Eritrea, particularly on Hamadryas baboons. We detected Hamadryas baboons in all five ecological zones and Olive baboons only in the south-western part of the western lowland. Grivet monkeys are sympatric with both baboon species. A geographical separation of the two baboon species, and sympatry of Cercopithecus aethiops with Papio anubis was reported from Ethiopia by Dunbar & Dunbar (1974). An exception to the allopatry of the baboon species is an area of overlap that exists in central Ethiopia (Nagel 1973). So far we have not discovered such a zone in Eritrea, but there is probably an area west of Barentu where the ranges of the two species overlap. Ecologically, Hamadryas baboons seem to be the most tolerant species, occupying the widest range of habitats, from the marginal dry and hot coastal lowlands to the sub-humid and moist slopes and mountain tops of the eastern escarpment and the central highlands. Hamadryas baboons in Saudi Arabia and Yemen show a similar altitudinal distribution and ecological plasticity (Biquand et al. 1992, Al-Safadi 1994). Olive baboons are confined to the moist and low-lying areas in the south- west, with relatively high plant productivity. The likely borderline between the two species is the 700 to 900 m contourline in the western lowlands. Grivet monkeys share the habitat with both baboon species. Since they are much more arboreal than the baboons they are more restricted in their range to wooded areas. In most parts of our survey area we found grivets in river valleys, where they foraged in the riverine trees, particularly in large fig trees. They do not only search for their food in different parts of the habitat, but also feed on other parts of plants than the more 263 terrestrial baboons and competition is therefore reduced (Dunbar & Dunbar 1974). We can expect much more competition between the baboons because there is a broad overlap in diet and habitat use (Kummer 1968, Barton et al. 1993). It remains puzzling which factors hinder one or the other species in expanding its range, although, at least in some areas, the conditions seem to be suitable for both. In most parts of Eritrea, monkeys and baboons are the only larger wildlife stiil remaining and they are recognized as a threat to crops and gardens. Hunting and killing of wild animals is illegal, nevertheless people chase primates out of their areas and sometimes kill them. With intensification of agricultural activities through irrigation projects, the conflict between primates and humans will become more intense. However, the Eritrean government is planning national parks that will also include primate habitats (EAE 1995). Acknowledgements We would like to thank our field assistant Dawit Berhane for his help in the field and with the customs of the Eritrean people, the Ministry of Agriculture and the University in Asmara for administrative assistance, S. Heiduck for critical comments on the manuscript and the Deutsche Forschungsgemeinschaft (Ga 342/7- 1), the Acciones Especiales del Ministerio de Educaciön y Cultura (APC98-0012), Spain and the scientific agreement between the Universidad Autönoma de Madrid and the Deutsches Primatenzentrum for their financial support. References Al-Safadi, M. (1994): The hamadryas baboon, Papio hamadryas (Linnaeus, 1758) in Yemen (Mammalia: Primates: Cercopithecidae). — Zool. Middle East 10: 5-16. Bonbon? KoA, A. Whiten, R.W. Byrne & M. English (1993): Chemical composition of baboon plant foods: implications for the interpretation of intra- and interspecific differences in diet. — Folia Primatol. 61: 1-20. piguand) SV. biquand-Guyot, A. Bouse & J.P. Gautier (1992): The distribution of Papio hamadryas in Saudi Arabia: Ecological correlates and human influence. - Int. J. Primatol. 13: 223-243. Cheney, D.L. (1987): Interactions and relationships between groups. — pp. 267- Join erimate Societies (eds. Smuts, B.B., D.L. Cheney, R.W. Seyfarth, R.W. Wrangham & T.T. Struhsaker). — The University of Chicago Press, Chicago. Pumbar Rol M. & E.P. Dunbar (1974): Ecological relations and niche separation between sympatric terrestrial primates in Ethiopia. — Folia Primatol. 21: 36-60. Ppebeetrtirean Agency for the Environment (1995): National Environmental Management Plan for Eritrea. - Government of Eritrea, Asmara, Eritrea. Kummer, H. (1968): Social organization of hamadryas baboons. A field study. — The University of Chicago Press, Chicago. Nagel, U. (1973): A comparison of anubis baboons, hamadryas baboons, and their hybrids at a species border in Ethiopia. — Folia Primatol. 19: 104-165. 264 White, F. (1983): The Vegetation of Africa. - UNESCO. Yalden, D.W.,M.J. Largen. & D. Kock (1977): Catalogue of the mammals of Ethiopia. 3. Primates. — Monitore Zool. Ital. (NS) Suppl. 9: 1-52. Yalden, D.W., MI. Pargen, Do Kock, & JjGs Hillman 1996). Cataloeme of the mammals of Ethiopia and Eritrea. 7. Revised checklist, zoogeography and conservation. — Tropical Zoology 9: 73-164. Zinner, D. & F. Torkler (1996): GIS and remote sensing techniques as tools for surveying primates. — Ecotropica 2: 41-47. Maps and geographical data: National Map of Eritrea (1995): 1:1,000,000. — Government of Eritrea, Asmara, Eritrea. U.S. Geological Survey 1997: Edition: 1, Sioux Falls, SD, USA, EROS DataCenter / International Program. Authors‘ addresses: D. Zinner: Deutsches Primatenzentrum, D-37077 Göttingen, Kellnerweg 4, Germany, dzinner@gwdg.de F. Torkler: FH Eberswalde, D-16225 Eberswalde, Germany, ftorkler @fh- eberswalde.de F. Peläez: Universidad Autönoma, E-28049 Madrid, Spain, fpelaez@uam.es 265 Effects of fragmentation and assessing minimum viable populations of lemurs in Madagascar J.U. Ganzhorn, S.M. Goodman, J.-B.Ramanamanjato, J. Ralison, D. Rakotondravony & B. Rakotosamimanana Abstract: Lemur species assemblages on Madagascar show highly nested structures in forest fragments of different size. This allows us to predict with a high degree of probability which species will most likely go extinct if the available habitat decreases in size. We use this predictable pattern to identify the minimum viable populations of different lemur species that were able to survive for 20-40 years in forest fragments. According to this analysis, populations of about 40 adult individuals are able to survive for 20-40 years. Populations of this size require between 20 and 800 ha of suitable forest. Fragments of this size no longer exist in the littoral forests of the southeastern portion of the island and are rare in the western dry deciduous forests. Key words: Madagascar, lemurs, fragmentation, nested subset, minimum viable populations Introduction Over the last few decades the forests of Madagascar have been reduced dramatically in extent (Green & Sussman 1990, Nelson & Horning 1993). The situation is precarious, in particular for the dry deciduous forests of western Madagascar (Smith 1997) and for the littoral forests along the east coast (Du Puy & Moat 1996). Apart from forest reduction in general, fragmentation of the remaining forests is severe and several species of lemur occur only in a few remaining forest fragments (e.g., Propithecus tattersalli, P. diadema perrieri, Eulemur macaco flavifrons, Lepilemur dorsalis; reviewed in Mittermeier et al. 1994). In view of these developments, there are now several plans for reforestation in order to rehabilitate lost habitat for lemurs and other forest-dependent species (Holloway 1999). Reforestation can be used to extend and create buffer zones around natural forests, and to link remaining forest fragments that could then act as reservoirs from where native species could colonize the new habitats. In the domain of the western deciduous forest, secondary forests grow very slowly and do not provide suitable habitat for lemurs even after 40 years of regeneration (Sorg 1996, Ganzhorn & Schmid 1998), though mixed tree plantations can be used by lemurs after afew years (Ganzhorn & Abraham 1991). Under more favorable conditions in the domain of the eastern humid forests, tree plantations and native trees seem to grow fast enough to provide acceptable habitat for many (though not all) lemur species after a few years (Ganzhorn 1987, Goodman & Rakotondravony 1998). Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 266 While programs for reforestation and rehabilitation of fallow land progress in other parts of the world (e.g., Lamb 1998), there is little information available for Madagascar. Many reforestation projects on the island include patches of natural forest in their restoration strategy. These patches of forest should serve as sources from where species can colonize the regrowing habitat. One problem that we address here for Madagascar is the question of how big the patch of native forest has to be to allow the indigenous plant and animal species to survive in these forest fragments until the restored vegetation can provide suitable habitat for them to expand into. Lemurs were chosen for this study because they are among the largest forest- dependent vertebrate species of Madagascar. We therefore consider them “umbrella species”. If remaining forest patches are large enough to allow their survival, these fragments should also allow survival of many other forest-dependent species. To resolve this question we review the consequences of forest fragmentation in various regions of Madagascar to find out which extent of forest 1s required to allow a given lemur species to survive for the next 20-40 years (1.e., the time needed for establishing tree plantations). Methods We conducted a literature search to obtain information on lemurs occurring at several forest sites that had been fragmented some 20-40 years ago (table 1). This information was condensed into presence/absence data in these forest fragments. To assess the effect of fragmentation, the presence/absence data of lemurs in these fragments were compared to the composition of lemur communities in a nearby Table I: Summary of data used to assess the effects of forest fragmentation on lemurs Domain Range of Large forest nO. O18, Retenence Study site and forest fragment block for frag- type size [ha] comparison ments Menabe West; dry 0.6-600 Kirindy/CFPF 10 Ganzhorn et al. deciduous 1999b Mahatsinjo East; 240 Ankilahila Goodman & Schütz evergreen 1999 Sahivo East; 100; 231 Betampona Britneral2999 Antanamalaza evergreen Tampolo East; 675 Ambatovaky Goodman & littoral, Rakotondravony evergreen 1998 Manafiafy, South; Anosyennes, Lewis Environmen- Mandena littoral Vohimenas tal Consultants forest 1992; own data * only diurnal species 267 V4 / Ambatovaky R rf N 4 N Kirindy / Vy, SY CFPF S fan? =»; AS KS’ YZ GJ I A Alle ATE RR Rn Ru Fig. 1. Location of sites considered to assess the effect of fragmen- tation on : sr | STAID ANoSVenne> Shaded De —L aes Vohimenas a forests are in the east, dry aly f) 200 400 deciduous forests are in the west and spiny forests are in the south km large block of forest that was assumed to contain the complete set of extant lemur species known from each of these regions. Data were also supplemented with results from ongoing studies in the littoral forests near Tolagnaro, in the extreme southeastern portion of the island (fig.1). Population size of the different lemur species in a given fragment at the time of isolation was estimated by multiplying fragment size with the average population density of the lemur species in relatively undisturbed forests (Ganzhorn et al. 1999a). We then plotted population size against fragment size to assess the transitions between fragments of differing size where a given lemur species occurs in one fragment but is absent in the next smaller fragment. We assume that the lemur species was also present in the smaller fragment at the time of isolation, but that a population of that size was too small to survive the subsequent 20-40 years of isolation. We then assume that a population of a size intermediate between the population size of the larger fragment (where the species still occurs) and the next smaller fragment (where the species is absent) marks the minimum size of a stable population that would survive for 20-40 years. For example: the theoretical population size of species A in a fragment of 400 ha is about 300 animals and the species is still found in a fragment of this extent. The theoretical population size of this species in a fragment of 200 ha is only 150 animals. Now if species A does not occur in the 200 ha fragment we assume that a population of 225 animals would not 268 be able to survive for about 20 years. This is the population size at which a species would be doomed to extinction if they were not able to exchange individuals with other areas. This minimum viable population and the average density of a given species was then used to calculate the minimum area of fragments for different lemur species that should allow them to survive for the next 20-40 years. Table 2: Lemur species in differently sized forests (references as listed in table 1) A: Littoral forest Site M Hg Ef Dm | Pd Er VW li thal = SPP- Body mass [kg] 0. . 0.6 - ne 06 0.9 Ambatovaki z + + + + - + + Eu + B: Dry deciduous forest Anosyennes Body mass [kg] 0.04- 0.06 Vohimenas Tampolo Manafiafy S9 Mandena M3 Manafiaty S7 Mandena M15/16 Mandena M13 Mandena M20 Mandena M4 Mandena M5 Mandena M6 An. ar „ur ae + + + + + + + + + + + + + + + + + + + + + + + +--+ + + + + + t+ + + + + + + + + + + Kirindi Menabe | + Menabe + Menabe Menabe + Menabe + + Menabe Menabe nay Menabe Menabe : Menabe | Species: Mspp. = Microcebus spp., Cm = Cheirogaleus medius; Cspp. = Cheirogaleus spp.; Mc = Mirza coquereli, Pf = Phaner furciter, Lspp. = Lepilemur spp., Ef = Eulemur fulvus, Er = Eulemur rubriventer, Vv = Varecia variegata, Hg = Hapalemur griseus, Al = Avahi laniger, Pv = Propithecus verreauxi, Pd = Propithecus diadema, li = Indri indri, Dm = Daubentonia madagascariensis 269 © absent 1000 ° present | ‘3 Fig.2: Theoretical population size of lemurs present in any one fragment at the time of isolation some 20-40 years ago. The diffe- rent symbols mark whether or not the species in question is still 0,1 1 10 100 1000 present in a given fragment Fragment size [ha] today. Population size Results and discussion Presence/absence of lemurs ın relation to forest type and size The literature search revealed five studies that could be used to assess the effects of recent (20-40 year old) forest fragmentation on lemurs. The sources of these data and several parameters associated with each site are summarized in table 1. Pomussspeeies assemblages derived from littoral and dry forest lemur communities show highly nested patterns of species composition in fragments of declining size in both vegetation formations of Madagascar. This pattern held regardless of differences in the lemur communities in the large forested blocks that 250 200 150 100 Population size at extinction 50 0 1 2 3 4 5 6 7 Body mass [kg] Fig. 3. Size of populations that were not viable for more than 20 — 40 years. On the right side the figure depicts the median and 25-75% quartiles of the non-viable mean population size for all species. 270 are considered to represent the original state (table 2). Both data sets deviate from random assemblages of species according to the analyses for nested data sets provided by Atmar & Patterson (1993: littoral forest: T = 0.47, p < 0.001; dry forest: T= 0.17, p< 0.001; based on 1000 runs). This pattern is repeated at all sites in Madagascar studied so far (Ganzhorn 1998). Minimum viable populations Fig.2 shows the presence/absence of lemurs in forest fragments of different size. Assuming that at the time the forest had been fragmented the lemur density in each fragment was equivalent to the average density of this species, we can estimate the number of individuals that were present in the fragment at the time of isolation. The probability of being present after 20—40 years of isolation correlates positively with increasing fragment size and decreasing body mass. On average, populations of 36 adult individuals (median: 35.5; 25-75% quartiles: 17-97; n= 13 species) do not survive for more than 20-40 years (fig.3). The size of populations that are viable for 20-40 years is unrelated to body mass, litter size or inter-birth intervals. Minimum viable populations that can survive for 20 — 40 years seem to be smaller for folivorous and omnivorous species than for frugivorous species, though the effect is not significant due to small sample size. 100 10 Fig.4: Minimum size of forest fragments to maintain viable 10 100 1000 10000 Di en of differently sized emur species in the dry and in the Body mass [g] humid forest. Area requirements for MVP [ha] 1 The size of forest fragments required to maintain viable populations augments with increasing body mass and is larger for humid than for dry forests (fig.4). Thus, the small lemurs, such as Microcebus spp., need fragments of about 20 ha to survive for more than 20 years. But the largest lemurs of the eastern rain forest, such as Indri indri and Propithecus diadema which weigh up to about 7 kg (Mittermeier et al. 1994), need fragments of at least 8-10 km’ to maintain viable populations for the period needed to establish new habitats for them - if indeed suitable habitats can actually be established. In the littoral forest sites of Mandena, Manafiafy and Tampolo there are no remaining fragments of that size. The largest known lemur species occurring in these forests is Eulemur fulvus, which weighs approximately 2.5 kg (Mittermeier et al. 1994). These littoral forests on sandy soils have low plant 271 productivity, which might contribute to their impoverished lemur communities as compared to adjacent forests on lateritic soils. This point is further complicated by the fact that the nearby sites with lateritic soils are still attached to the large eastern humid forest block. The present analysis will have to be refined in the future as it rests upon a number of assumptions that do not take into account natural variation. But even if it were wrong by an order of magnitude, the future of the lemurs in Madagascar has to be a matter of grave concern. Acknowledgements Our work in Madagascar has been authorized by the Direction des Eaux et Foréts, the Association National pour la Gestion des Aires Protégées, and the Commission Tripartite. We thank QIT Madagascar Minerals and their environmental and conservation team for their support of the surveys and research in the region of Tolagnaro. The present paper is part of the Accord de Collaboration between the Université d’ Antananarivo and Hamburg University. We thank Anja Ganzhorn and Klaus Rupp for their help in the preparation of the figures. The participation of J. B. Ramanamanjato in the symposium was supported by a generous grant from the city of Bonn. Last but not least we thank R. van den Elzen and G. Rheinwald for their untiring efforts to make the symposium a success. References Atmar, W. & B.D. Patterson (1993): The measure of order and disorder in the distribution of species in fragmented habitats. — Oecologia 96: 373-382. met. A. A. Axel & R. Young (1999): Brief surveys of two classified forests in Toamasina Province, eastern Madagascar. — Lemur News 4: 25-28. Duekuy. DJ. & J. Moat (1996): A refined classification of the primary vegetation of Madagascar based on the underlying geology: using GIS to map its distribution and to assess its conservation status. — pp. 205-218 in: Biogeography of Madagascar (ed. Lourengo, W. R.). - ORSTOM, Paris. Ganzhorn, J.U. (1987): A possible role of plantations for primate conservation in Madagascar. — Amer. J. Primatol. 12: 205-215. Ganzhorn, J.U. (1998): Nested patterns of species composition and their implications for lemur biogeography in Madagascar. — Folia Primatol. 69 (Suppl.): 332-341. Ganzhorn, J.U. & J.-P. Abraham (1991): Possible role of plantations for lemur conservation in Madagascar: food for folivorous species. — Folia Primatol. ao: 171-176. Ganzhorn, J.U. & J. Schmid (1998): Different population dynamics of Microcebus murinus in primary and secondary deciduous dry forests of Madagascar. — Int. J. Primatol. 19: 785-796. Ganznvern, J.U., P.C. Wright & J. Ratsimbazafy (1999a): Primate communities: Madagascar. — pp. 75-89 in: Primate communities (eds. Fleagle, J.F., C. Janson & K. Reed). — Cambridge University Press, Cambridge. caaznorn. J.U., J. Fietz, E. Rakotovao, D. Schwab &D. Zinner 22. (1999b): Lemurs and the regeneration of dry deciduous forest in Madagascar. — Cons. Biol. 13: 794-804.. Goodman, S.M. & D. Rakotondravony (1998): Les lémuriens. — pp. 213- 222 in: Inventaire biologique de la Forét littorale de Tampolo (Fenoarivo Atsinanana), Recherches pour le Développement, Série Sciences biologiques, Vol. 14, (eds. Ratsirarson J. & S.M. Goodman ). — Centre d’ Information et de Documentation Scientifique et Technique, Antananarivo. Goodman,S.M. & H. Schütz (1999): Observations of lemurs in the forest of Tsinjoarivo, Ambatolampy. — Lemur News 4: 14-16. Green, G.M. & R.W. Sussman (1990): Deforestation history of the eastern rainforest of Madagascar from satellite images. — Science 248: 212-215. Holloway, L. (1999): Lemurs can’t do it all by themselves anymore. — Lemur News 4: 29-30. Lamb, D. (1998): Large-scale restoration of degraded tropical forest lands: the potential role of timber plantations. — Restoration Ecology 3: 271-279. Lewis Environmental Consultants (1992): Madagascar Minerals Project. Environmental Impact Assessment Study. Part I: Natural Environment. — Final Report to QIT Fer et Titane, Inc. Montreal, Quebec, Canada. Mittermeier, R.A.,T. Tattersall, W.R. Konstan, D NEST y72055% R.B. Mast (1994): Lemurs of Madagascar. — Conservation International, Washington, D.C. Nelson, R. & N. Horning (1993): AVHRR-LAC estimates of forest area in Madagascar, 1990. — Int. J. Remote Sensing 14: 1463-1475. Smith, A.P. (1997): Deforestation, fragmentation, and reserve design in western Madagascar. — pp. 415-441 in: Tropical Forest Remnants, Ecology, Management and Conservation of fragmented Communities (eds. Lawrence, W. & O.W. Bierregaard). — University of Chicago Press, Chicago. Sorg, J.-P. (1996): L’etude de la vegetation, un outil aussewzeer de l’amenagement et de la gestion des ressources forestieres a Madagascar. — Akon’ny Ala 18: 26-36. Authors’ addresses: Jorg U. Ganzhorn, Zoologisches Institut, Martin-Luther-King-Platz 3, 20146 Hamburg, Germany; Steven M. Goodman, Field Museum of Natural History, Chicago IL 60605, USA and WWE, BP 738, Antananarivo (101), Madagascar; Jean-Baptiste Ramanamanjato, QIT Madagascar Minerals, BP 225, Tolagnaro (614), Madagascar; José Ralison and Daniel Rakotondravony, Département de Biologie Animale, BP 906, Université d’ Antananarivo, Antananarivo (101), Madagascar; Berthe Rakotosamimanana, Département d'Anthropologie, BP 906, Université d’ Antananarivo (101), Madagascar. Priority areas for forest bird and primate conservation in Madagascar- do they match up? Frank Hawkins Abstract: Efforts to conserve biodiversity must be directed towards the most important areas. Aside from questions related to the feasibility of conservation, a key issue is the overlap of key areas (for species diversity or presence of threatened species) for different taxa. A premise of BirdLife International’s Important Bird Areas project is that the distribution of threatened and restricted-range birds can be used as a surrogate for other, less well-known taxa. In Madagascar, a recent study has identified 84 Important Bird Areas (IBAs), based on the presence of threatened, restricted-range and biome-restricted species. They are concentrated in primary forest and wetlands. The forest IBAs are concentrated in the north-eastern part of the humid forest, the north-western deciduous forest and parts of the southern spiny forest. A set of sites for primate conservation based on the presence of threatened species are concentrated in small patches of relict forest, mostly in the north and east, and in well-known sections of the eastern forest in the centre and south-east. There are no major priorities for primate conservation in the west or south. Using species richness as an indicator of conservation priorities gives a similar result. Thus overall there is no correlation between the priority rating of sites for bird and primate conservation, but with a few exceptions, the overall list of important bird areas includes all the important primate areas. This pattern is due, at least in part, to a lack of knowledge of taxonomy of nocturnal primate species, and poorly-known distributions of difficult-to-detect primates. In addition, the published threat categories of some widespread primate taxa seem inappropriate, partly because of the emphasis placed on taxonomic uniqueness. However, even when these factors are allowed for, the distribution of primate and bird priority areas seems to differ because of fundamental differences in the patterns of regional endemism. Key words:Important bird areas, Madagascar, primate conservation The Madagascar context Madagascar is widely regarded as one of the world’s top biodiversity conservation priorities. Humans only arrived about 2000 years ago (MacPhee & Burney 1991), but are widely considered to have caused extensive deforestation (Green & Sussman 1990, Nelson & Horning 1993). It is certainly true that forests are being cleared today. This is resulting in the loss of an astounding array of endemic biological diversity. Around 12 000 species of plants, including 6 endemic families (Dupuy & Moat 1996), five families, 14 genera, 32 species and 50 taxa of primates, all endemic (Mittermeier et al. 1995); 106 species, 37 genera and 5 or 6 families of endemic bird (Morris & Hawkins 1998); about 300 reptile species, about 95% of which are endemic (Raxworthy & Nussbaum 1997). In addition, almost all this Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 274 diversity occurs in forest. About 80% of plant species are found only in forest, as are 33 of 37 endemic genera of bird. All extant primate taxa are exclusively forest dwelling (except one in a marsh). Forest types All forest types (which include eastern rainforest, western deciduous forest, and southern spiny forest) are declining in extent, probably fairly rapidly. The most threatened is probably western deciduous forest (Dupuy & Moat 1996). Eastern forest occurs along a north-south scarp, with most forest occurring where there are frequent dry-season nocturnal mists that prevent burning. Western and southern forests are rather patchy, and occur mostly in areas of low human population density. Conservation issues Madagascar is currently half way through the second phase of a fifteen-year three- phase National Environmental Action Plan (World Bank 1995). A workshop funded by the Global Environment Facility in 1995 produced an essential first step in defining conservation priorities for Madagascar (Ganzhorn et al. 1997). However, lack of accurate distribution detail for most taxa, and lack of ecological knowledge of how species’ distributions vary with habitat, has prevented the development of a model that could be used for predicting conservation value of an area based on likely species presence (Hagen 1999). Such a model is essential if resources to make detailed inventories in all poorly-known forest areas are lacking. However, within the constraints of the National Environment Action Plan, there is a current urgent need to allocate scarce resources for the maximum biodiversity conservation good; — Managers need a way of identifying sites that contain the maximum valuable biodiversity, for as many taxa as possible — Managers would like to be able to use indicators to do this — species, families, higher taxa, habitats — Managers need to know, for other taxa, where extra effort will be required outside these priority areas. In Madagascar, only birds and primates are well enough known (at least in the published domain) for distributional comparisons to be useful. Thus in this paper the distributions of birds and primates are used to answer the following questions: — What is the degree of overlap in sites containing the most threatened birds and primates? — For birds and primates, if we choose a complement of sites that contains all key species for one taxon, which species in the other taxon are missed? — Does the distribution of birds best represent the distribution of primates, or vice- versa? — What are the implications of these results for conservation planning in the real world? 275 Methods Two largely independent data sets were compared; birds and primates within the eastern rainforest biome, and birds and primates in the western and southern forest biome. Priority sites for birds and primates were identified following the presence of threatened species (of IUCN categories: Critically Endangered, Endangered and Vulnerable, following Collar et al. 1994 and Mittermeier et al. 1994) present at a site. Bird distributions in potential priority areas came from ‘Projet Zones d’Importance pour la Conservation des Oiseaux a Madagascar’, a European Commission-funded collaboration between the Malagasy National Association for the Management of Protected Areas (ANGAP), the Ministry of Water and Forests (MEP), and BirdLife International (ZICOMA 1999). During Project ZICOMA, 117 sites were evaluated according to the presence of threatened, restricted-range or biome-restricted bird species. A book containing the details of the 84 sites (55 in forest) that qualified was published in July 1999. Primate priorities were set using data from Ganzhorn et al. 1996, plus Projet ZICOMA recent sightings (ZICOMA 1999), S.M. Goodman’s recent inventories for WWE (Goodman 1997, 1998, 1999, in press), and Conservation International’s recent RAP studies (Schmid 1999, and this volume). This protocol was considered superior to one using simple species richness, as the non-threatened species not considered in this protocol are generally widespread, and present in most primary forests. Complementarity analysis (Howard et al. 1998) used the “the simple greedy” algorithm (Williams 1998) which involves choosing a complementary set of sites within each major biome that included all the species of a particular taxon present in the biome. Ankara N, Analamera Andavak ® oO birds ndava aA al Darai araina & primates Manongarivo a HE both © Ankarafantsika Mikea <> Rakotosamimanana, A.T.C. Feistner & Y. Rumpler (996), mhc state of lemur conservation in Madagascar. — Primate Conservation 17: 70-80 283 Sammer KU Bo wRakotosamamana.» Hannah, J. Housh,L. Bros Oliver, Siok ajaobelina., Cy Rodstrom & Gy Tilkin (1997): Priorities for biodiversity conservation in Madagascar. — Primate Report 48: 1-25. Goodman, S.M. (ed) (1997): A floral and faunal inventory of the eastern slopes of the Réserve Naturelle Integrale d’ Andringitra, Madagascar; with reference to elevational variation. — Fieldiana, Zoology, new series 85. Goodman, S.M. (ed.) (1998): A floral and faunal inventory of the Réserve Spéciale d’Anjanaharibe-Sud, Madagascar: with reference to altitudinal variation. — Fieldiana Zoology, new series 90: 1-241. Goodman, S.M. (ed.) (1999): A floral and faunal inventory of the Parc Nationale d’ Andohahela, Madagascar; with reference to elevational variation. — Fieldiana Zoology, new series. Goodman, S.M. (ed.) (in press): A floral and faunal inventory of the Parc Nationale de Marojejy, Madagascar; with reference to elevational variation. — Fieldiana Zoology, new series. Goodman, S’M., A.E.A. Hawkins & C.A. Domergue (1997): A new species of Vanga (Vangidae) from south-western Madagascar. — Bull. Brit. Ormthol Gl 117: 5-10. Goodman, S.M. & L.M.A. Rakotozafy (1997): Subfossil birds from coastal sites in western and southwestern Madagascar: a palaeoenvironmental reconstruction. — pp. 257-279 in: Natural Change and Human Impact in Madagascar (eds. Goodman, S.M. & B.D. Patterson). — Smithsonian Institution, Washington. Hagen, R. (1999): Biodiversity conservation priorities in the three regions of the Landscape Development Interventions Program. — Unpublished report to USAID Madagascar. BRosvandı Er @, Po Viskanic, T.R.B. Davenport, F.W.Kisenyi,M. Baalerer C.J. Dickinson, J.S. Lwanga, R.A: Matthews & A. Balmford (1998): Complementarity and the use of indicator groups for reserve selection in Uganda. — Nature 394:472-475. Hawkins, A.F.A. (in press a): The primates of Isalo National Park, Madagascar. — Lemur News. Hawkins, A.F.A. (in press b): Altitudinal and latitudinal distribution of east Malagasy forest bird communities. — J. Biogeogr.. MacPhee, R.D.E. & D.A. Burney (1991): Dating of modified femora of extinct dwarf Hippopotamus from southern Madagascar: implications for constraining human colonisation and vertebrate extinction events. — J. Archaeol. Sci. 18: 695-706. Morris, P.A.J& A.F.A. Hawkins (1998): A photographic fieldguide to the birds of Madagascar. — Pica Press, Robertsbridge, East Sussex. Nelson, R. & N. Horning (1993): AVHRR-LAC estimates of forest area in Madagascar, 1990. — Int. J. Remote Sensing 14:1463-1475 Winesenmeirer, R.A., 1. Tattersall, W.R. Konstant, D.M. Meyers & R.B. Mast (1994): Lemurs of Madagascar. — Conservation International, 284 Washington, D.C.. Raxworthy, C.J. & R.A. Nussbaum (1997): Biogeographic patterns of reptiles in eastern Madagascar. — pp. 124-141 in: Natural Change and Human Impact in Madagascar (eds. Goodman, S.M. & B.D. Patterson). — Smithsonian Institution, Washington. Schmid, J. & P.M. Kappeler (1994): Sympatric mouse lemurs (Microcebus spp.) in Western Madagascar. — Folia Primatol. 63: 162-170. Schmid, J. (ed) (1999): Rapport préliminare d’inventaire rapide, corridor Mantadia- Zahamena. — Conservation International, Madagascar. Thompson, P.M. &M.I. Evans (1992): The threatened birds of Ambatovaky Special Reserve, Madagascar. — Bird Conserv. Int. 2: 221-237. Williams, P.H. (1998): Key sites for conservation: area-selection methods for biodiversity. — pp. 211-249 in: Conservation in a Changing World: Integrating Processes into Priorities for Action (eds. Mace, G.M., A. Balmford & J.R. Ginsberg). - Cambridge University Press, Cambridge, UK. W orld Bank (1995): Madagascar Environment Program Staff Appraisal Report. — World Bank, New York, USA. ZICOMA (1999): Zones d’Importance pour la Conservation des Oiseaux a Madagascar. — Projet ZICOMA, Antananarivo, Madagascar, 266pp. Zimmermann, E.,S. Cepok,N. Rakotoarison, V. Ziege mannssz0r Radespiel (1998): Sympatric mouse lemurs in north-west Madagascar: a new rufous mouse lemur species (Microcebus revelobensis). — Folia primatolog. 69: 106-114. Author’s address: Dr. Frank Hawkins, BirdLife International, Projet ZICOMA, BP 1074, Antananarivo 101, Madagascar; contact: zicoma @simicro.mg ——— Conservation planning in the Mantady-Zahamena corridor, Madagascar: Rapid Assessment Program (RAP) Jutta Schmid Abstract: Conservation International (CI) conducted a Rapid Assessment Program (RAP) of the biodiversity of the humid forest within the Mantady-Zahamena corridor in Madagascar. The RAP methodology is designed to quickly provide information needed to catalyze conservation activity and improve regional and national biodiversity protection. This eastern mountain range corridor links the protected areas of the Parc National (PN) de Mantady and the Réserve Naturelle Integrale (RNI) de Zahamena. This region is of particular interest and importance since not only does it comprise of primary and secondary lowland rainforest as well as moist montane forest, but it also connects two protected areas in the north and south. Survey work was divided into two expeditions (expedition 1: November 7 — December 14, 1998; expedition 2: January 15 — 28, 1999). Four camps were placed in different classified forests, and one site was placed in PN de Mantady. At each site, presence and relative abundance of lemur species were estimated using the line transect method. A total of 11 lemur species (four diurnal, two crepiscular, and five nocturnal species) were found in the corridor, including Aye-Aye Daubentonia madagascariensis. Forest clearance and settlement in the entire corridor are the major threats and might explain relatively low densities and absence of certain lemur species like Propithecus diadema diadema (site 1) and Varecia variegata variegata (site | and 5). Exceptional biodiversity cannot be maintained in areas where genetic exchange is impossible due to geographic isolation and disconnection of forest blocks. The most urgent and basic need is to stop forest fragmentation and habitat destruction, and to place exceptionally high conservation priorities on remaining lowland forests. Key words: Madagascar, Rapid Assessment Program, lemurs, conservation Introduction Endemism in Madagascar’s biota is extremely high, with species-level endemism of well over 90% for most taxonomic groups (Jenkins 1987, Langrand & Wilme 1997). One of the major adaptive radiations in Madagascar is among the primates. Malagasy lemurs (comprising about 33 species and 52 subspecies) are entirely endemic and occupy a wide range of forests, including eastern humid forests, western dry forests, and southern spiny forests (Harcourt & Thornback 1990, Mittermeier et al. 1994, Tattersall 1982). However, it is estimated that as much as 80% of these forests has disappeared in the 1500-2000 years since the arrival of man, to open land up for agricultural purposes, for mining, for extraction of building materials and fuelwood, and for commerical logging (Nelson & Horning 1993; Smith et al. 1997). Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 286 Forest fragmentation and habitat loss affect species richness and diversity (Laurance 1991), as they can cause changes in species-specific population dynamics. Reestablishement of the original biodiversity of isolated forest patches depends on distances to other wooded areas, and presence of fencerow connections (Gascon & Lovejoy 1998, Lindenmayer et al. 1993). Isolation of forested habitat may restrict rates of migration, and initially produce increases in density relative to continuous forest sites (Gascon & Lovejoy 1998). So far, little is known about possible consequences of disrupted gene flow and inbreeding ın small populations for the long-term survival of these species (Tomiuk et al. 1998). Corridors between remaining blocks of forest are required to maintain gene flow, facilitate the movement of animals across inhospitalbe terrain and promote dispersal between disjunct patches of habitats (Lindenmayer et al. 1993, Saunders et al. 1991, Tomiuk et al. 1998). Given the ongoing fragmentation of the Malagasy forests, corridors will be important for the survival of lemur species in otherwise fragmented habitats. Successful conservation planning for the unique habitats of Madagascar must be based on information on the biological recources existing in the region. Baseline inventory data can be obtained through “rapid assessments”. Conservation International’s Rapid Assessment Program (RAP) is designed to quickly provide the scientific information needed to catalyze conservation activity and improve regional and national biodiversity protection. RAP assembles teams of scientists to generate first-cut assessments of the biological value of poorly known areas. The Madagascar RAP expedition was conducted in the non-protected corridor between the Réserve Naturelle Intégrale (RNI) de Zahamena (to the north) and the Parc National (PN) de Mantady (to the south) in the eastern humid forest of Madagascar (Nicoll & Langrand 1989). As part of this multi-taxa inventory expedition, data were collected on primate species richness and abundance at four different sites within the corridor, as well as at one site in the PN de Mantady. Information on the distribution and density of lemur species in the Mantady- Zahamena region and the degree of threat to these populations is important to manage this part of Madagascar’s rainforests in the context of an integrated conservation and development project. Methods Preparatory work and planning In August and September 1998, local Malagasy stakeholders were interviewed to determine the specific types of data they desire, and to identify areas within the corridor of highest priority for biodiversity conservation. After that, recent vegetational and topographical maps (Ministere de |’ Agriculture et du Dévelop- pement Rural 1997, FTM 1968) of the area targeted for the RAP assessment were consulted to determine the extent of forest cover and likely areas for exploration. On October 12" 1998, an overflight in a small plane was conducted to identify forest types and points for field transects. Finally, ground reconnaissances were carried out to determine study sites, the route to be used for the transects and to prepare transport and logistics. Gt SE-\62 ae ) BF SE-160 f — Exceptional Exceptionnel j Fig.l: Map of Madagascar with areas of high biological importance. The area of the WAP asindicare d= (rom Ganzhorn et al. 1997) Study sites The biological inventory was conducted at five sites in the humid forest of the Mantady-Zahamena corridor (fig.2). The study sites were determined with the use of an overflight video, geographical and vegetational maps, and consultations with local people and national stakeholders. Survey work was divided into two expeditions (expedition 1: site | to 4: November 7 to December 14, 1998; expedition 2: site 5: January 15 to 28, 1999). Transects were centered around these sites within a radius of approximately 1.5 kilometers. Sampling lasted 6-7 days at each site. Names and positions of the five sites surveyed during the RAP are presented in table 1. Table |: Names and positions of sites surveyed during the RAP Iofa (Site 1) 18°42.1'S, 48°28.0'E 07-13 Nov Didy (Site 2) Orato ROUSE 4 8540/15, 16-22 Nov Mantady (Site 3) PALS) Sa GS DS)(0) 18 S 25 Nov-01 Dec Andriantantely (Site 4) 18°41.7'S, 48°48.8'E 04-10 Dec Sandranantitra (Site 5) Si O2ZE9IS 49 20525 u 18-24 Jan *) FC = Classified forest; PN = National Park ' Lake Alaotra ~ Betampona 05. Fig.2: Location of the study sites. Hatched areas indicate protected areas. Line transects The line transect method was employed for censusing lemurs. At each location two or three trails of varying lengths (750 m to 2700 m) were used for lemur surveys. We utilized preexisting trails left by bush pigs, cows and humans, as well as newly cut and laid out trails. Lemurs were censused by walking slowly (approximately 0.7 kmh’) along trails marked at 25 m cumulative intervals with flagging tape. Diurnal samples took place at times of increased lemur activity, in the morning (6:00 — 12:30h) and in the afternoon (15:00 — 18:00 h). Walks were always separated by a time interval of at least six hours when trails were censused twice during daylight hours. Nocturnal censuses commenced 10 to 30 minutes after dusk for two to four hours each night. The dim light of a headlamp was used at night to pick up the eyeshine from the reflective tapetum of nocturnal lemurs. Once detected, a more powerful hand-held flashlight and binoculars (7x40) were used for species identification. 289 For all detections, the species, time of contact, position on the transect, elevation, perpendicular distance from the trail, height from the ground, and habitat type were noted. Whenever possible, the number of individuals, age/sex composition and general behavior were also recorded. For each group of lemurs we estimated the perpendicular distance from the trail for the first individual seen. We did not spend more than 10 minutes on any single sighting. Censusing along each transect was repeated 1—3 times for nocturnal transects and 2-9 times for diurnal transects. Census walks were not conducted when viewing distance was restricted to less than 15 m because of heavy rain. Due to inadequate sample sizes because of the few repetitions of each transect, and the relatively short distances covered at each site, lemur densities were not determined (Ganzhorn 1994; Schmid & Smolker 1998; Sterling & Ramaroson 1996; Whitesides et al. 1988). However, we did calculate the mean number of sightings of lemurs per km transect and the mean detection distance perpendicular to the trail at which lemurs were seen was given for each species and trail. Since we did not find significant differences across trails, a single average detection distance was calculated for each species. For diurnal censuses the mean number of groups, and for nocturnal censuses the mean number of individuals observed within the transects were given. Lemurs heard but not seen during census walks, or seen outside of census walks by ourselves or other researchers, were not included in our calculations of encounter rates for either diurnal or nocturnal surveys. General observations Apart from the systematic transect surveys general observations were made during the day. At each site we explored the forest to look for secondary signs of the presence of certain lemurs such as the characteristic feeding signs of Daubentonia madagascariensis (gnaw marks from excavation of dead wood or living branches, opened seeds of Canarium madagascariense), or sleeping sites of nocturnal species (e.g. nests of Daubentonia; tree holes of Cheirogaleus or Microcebus). Furthermore, we paid special attention to vocalizations (songs) of Indris (Indri indri) and Ruffed Lemurs (Varecia variegata). Local people were questioned to collect information about the presence of lemur species and about the hunting pressure on the primate fauna. Results Species diversity In total, four diurnal species (Indri indri, Propithecus diadema diadema, Varecia variegata variegata, and Eulemur rubriventer), two crepuscular species (Eulemur fulvus fulvus and Hapalemur griseus griseus), and five typically nocturnal species (Microcebus rufus, Cheirogaleus major, Avahi laniger, Lepilemur mustelinus, and Daubentonia madagascariensis) were found in five survey sites within the Mantady-Zahamena corridor (table 2). All taxa were recorded by direct observation except for Daubentonia, which was recorded by its indirect feeding signs (Duckworth 1993, Erickson 1995). None of the five survey sites contained all species. Species diversity was highest at sites 2 and 3, where ten lemur species were present. Eight species were found at sites | and 4, and only six at site 5. Indri indri, 290 Table 2: The primate species found in the humid forests of the Mantady-Zahamena corridor listed by camp site and altitude. Species were recorded during survey walks or during additional observations. Site | Site 2 Site 3 Site 4 Site 5 835 m 960 m 895 m 530 m 450 m Microcebus rufus 4 + + + Cheirogaleus major Avahi laniger Lepilemur mustelinus Daubentonia madagascariensis Indri indri Propithecus diadema diadema Varecia variegata variegata Eulemur rubriventer Eulemur fulvus fulvus Hapalemur griseus griseus Total number of species + species present; # species seen by other participants; fd feeding signs; * vocalizations only Table 3: Mean number of sightings per km transect and mean detection distances (+ standard deviation) of species (individuals) seen during nocturnal censuses at each camp site in the Mantady-Zahamena corridor. Detection distances are perpendicular to the trail; n = number of individuals. Length of | Number | Microcebus | Cheirogaleu Avahi Lepilemur Number | transects s major laniger mustelinus | of species (m) 0.6 (2.0) LOGO) 1 (8.0) 2.4 (3.4+1.6) | 1.0 (4.4#2.8) |0.5 (3.8+0.4) [0.2 (8.0) 14.8.1323) 1171185325) Ve Bde.) 0.6 (0.8+0.6) 1.4 (3.9#1.8) 291 E. f. fulvus, M. rufus and C. major were found at all sites surveyed. H. griseus, V. v. variegata and A. laniger were absent from site 5, and L. mustelinus was censused at all sites except site 4. P. d. diadema was seen from site 2 to 4. E. rubriventer occurred only in the mid-altitude forests of site 2 and 3, but no evidence for its presence was found at site 1 (835 m). D. madagascariensis was only present in the forest of Sandranantitra (site 5). Of particular interest was the extremely high encounter rate of V. v. variegata at site 4, where at least six different groups along a total trail length of 2.5 km were found. The length of transects, number of census walks and the species encountered at the various sites are listed in table 3 for nocturnal censuses and in table 4 for diurnal censuses. During night surveys a total of 35.6 km was walked, and 98.7 km during the day surveys. Table 4: Mean number of sightings per km transect and mean detection distances (+ standard deviation) of species (individuals and groups respectively) seen during diurnal censuses at each camp site in the Mantady-Zahamena corridor. Detection distances are perpendicular to the trail; n = number of groups. Trail | Length | Number | /ndriindri | Propithecus Varecia Eulemur Eulemur Hapalemur Number of diadema variegata rubriventer fulvus griseus species censuses diadema variegata fulvus griseus 0.3 (4.5+3.5) 0.3 (4.5+3.4) 0.1 (1.5+0.7) | 0.1(4.0) |0.2 (5.3#3.3) 0.2 (10.0) 0.1 (5.344.5) | 0.1 (2.0+1.4) 0.04 (12.0) 0.1 (5.0#1.7) 0 0.2 (12.0) 0.4 (9.7+9.1) 1.6 (6.0+3.1) 0 0.1 (1.0) WG: 10.8 (6.2+4.4) 0.3 (8.0+9.9) Overall mean detection distance (m) * species heard or seen outside predefined census walks. Discussion The Mantady-Zahamena corridor represents an important area for lemurs, and both species diversity and relative abundance is remarkable. Eleven species were confirmed in the corridor and in the PN de Mantady: ten species were detected by direct sightings on census walks and the presence of Aye-aye Daubentonia madagascariensis can be inferred from distinctive feeding traces. The forest block surveyed borders to the north on the RNI de Zahamena, and to the south on the PN de Mantady. This corridor contains a complex network of classified forests, commercial forestry zones, local forest exploitation and two other reserves, Mangerivola and Betampona. The entire region within and between 2D, these protected areas is extremely important for biodiversity conservation purposes. According to IUCN criteria, six of the 11 species recorded in the corridor are considered to be globally threatened by extinction (Harcourt & Thornback 1990). Although no quantitative assessment of lemur population densities was made, total number as well as relative abundance of species differed between sites. The number of species decreased from ten species recorded at sites 2 and 3, to eight at sites | and 4, and finally reached a minimum of only six species at site 5. Forest clearance and settlement in the entire corridor are the major threats and might explain relatively the low densities and absence of particular lemur species like P. d. diadema (site 1) or V. v. variegata (sites | and 5). Of particular interest was the low diversity and abundance of lemur species in the lowland forests of Sandranantitra compared to all the other sites surveyed. Numerous clearings and trails have already penetrated into this forest, and lemur traps, although always inactive, were frequently found, indicating that hunting pressure is another major threat to this area and its lemur populations. By contrast, in areas within the Mantady-Zahamena corridor, where deforestation and consequently hunting pressure on lemurs was relatively low (sites 2, 3 and 4), the present diurnal lemur species were remarkably tolerant of humans, tending to approach at close range or to simply ignore them. The result of the overflight indicated that the lowland forest of Andriantantely is an isolated remnant in contrast to the remaining survey sites which were all still connected to adjacent rainforest. With eight lemur species found in Andriantantely, this forest had a higher diversity of primates than the lowland forest of Sandranantitra where only six species were found. Although Sandranan-titra was characterized by a high degree of habitat destruction, it has not been isolated from the mid-altitude mountain forest chain to its west. However, the results of the plant and vegetation survey, which was undertaken simultaneously, indicated that Sandranantitra forest must have suffered a dramatic deforestation and probably defaunation a long time ago. The forest of Sandranantitra was dominated by large trees of Ravenala madagascariensis (Strelitziaceae) indicating that this forest was previously exploited. By contrast, Andriantantely contains ca. 7,000 ha of primary lowland forest that has been isolated for only serveral years, and it still maintains viable populations of its original lemur richness and abundance. Clearly, forest fragmentation does have an impact on the original community, with many species that may persist for long periods after isolation, but other species that may eventually get lost (Kattan et al. 1994, Laurance 1991 Turner sooo: Andriantantely may be in a very early stage of isolation and thus possible consequences are not yet discernible, whereas in Sandranantitra certain lemur species have already retreated to undisturbed adjacent forests. | Mittermeier etal. (1992) list already protected lowland forests as areas which are of highest priority for the immediate establishment of programs for conservation action. However, conservation of protected areas with a notably diverse and dense lemur population, such as Zahamena with 14 lemur species or Mantady with ten lemur species, only makes sense when trans-frontier areas and corridors are included in conservation planning. Exceptional biodiversity cannot be maintained in areas where genetic exchange is impossible due to geographic isolation and disconnection of forest blocks (Ganzhorn et al. 1996/1997, Lindenmayer et al. 293 1993, Rabarivola et al. 1996, Saunders et al. 1991). Since we do not have more detailed information on natural gene flow and genetic differentiation in isolated populations of certain lemur species, fragmentation processes need to be reduced. Experience to date in Madagascar demonstrates that success in preserving and managing protected areas and classified forests is highly dependent on improving the management of critical habitats for biodiversity and on addressing the economic pressures on these ecosystems from neighboring communities (Mittermeier et al. 1987, 1994, Richard & O’Connor 1997). Adjacent regions around the Mantady-Zahamena corridor are of great economic importance, with Moramanga in the south, Toamasina on the eastern coast, and the Lac Alaotra region to the west, which include the biggest lake and principal rice-producing area in Madagascar. Madagascar is one of the poorest countries in the world, and rapid population growth is one of the main causes of natural resource degradation. Thus conservation of Madagascar’s biodiversity can only be successful if it includes economic improvements for local communities. From the point of view of diversity and density of lemur species, the Mantady- Zahamena corridor has to be considered a priority area. The most urgent and basic need is to stop forest fragmentation and habitat destruction, and to place exceptionally high conservation priorities on remaining lowland forests. For this, we suggest the following recommendations aimed at activities for lemur conservation: — Lemur conservation can only be successful if we find ways of allowing the coexistence of humans and lemurs. Tree plantations could provide firewood, fruit, honey and building material for people as well as suitable habitat for lemurs, and would guarantee that the pressure on the natural forests is reduced. — Malagasy primatologists have to be promoted and trained in representing and spreading the word on the importance of lemurs and their natural habitat to the Malagasy public and government institutions. This is of particular importance since there is a strong awareness against conservation projects among governmental and non-governmental organizations. Workshops and seminars undertaken by Malagasy and international experts could train individuals in data collection and analysis, database management, and the geographic information system (GIS). — Community education programs and socio-economic research should be undertaken. — Information on the genetic population structure, population dynamics, possible consequences of disrupted gene flow and inbreeding in small populations is desperately needed for the long-term survival of lemurs.These data are needed to interpret the biological importance of genetic differentiation and variation and their possible value to the survival of lemurs in isolated forest fragments. Acknowledgements Conservation International (CI) conducted a Rapid Assessment Program (RAP) of the biodiversity of Mantady-Zahamena corridor, financed by the Nederlands Comite voor IUCN and the Wolfensohn Family Foundation, USA. Malagasy Government institutions (Association Nationale pour la Gestion des Aires Protégées, Direction des Eaux et Foréts, Office National de l’Environnement), 294 several non-governmental organizations, and the staff of CI in Antananarivo under the guidance of Leon Rajaobelina and Sahondra Radilofe, and CI in Washington D.C. under the guidance of Leeanne E. Alonso, were of indispensible help in the preparation and carrying out of the expedition. I am grateful to Zo Lalaina Randriarimalala Rakotobe for her untiring engagement during the entire RAP, Joanna Fietz for her help during the second expedition, and Lucien Randrianjanaka for his unflagging support in the field. Jörg Ganzhorn made useful suggestions and comments. References Bennett, A.F., K. Henein & G. Merriam (1994): Corridor use and the elements of corridor quality: chipmunks and fencerows in a farmland mosaic. — Biol. Conserv. 68: 155-165. Duckworth 1993 [s.S. 6] Duckworth, J.W., M.I. Evans, A.F.A. Hawkins, R.J. Safford & R.J. Wilkinson (1995): The lemurs of Marojejy Strict Nature Reserve, Madagascar: A status overview with notes on ecology and threats. — Int. J. Primatol. 16: 545-559. Erickson, C.J. (1995): Feeding sites for extractive foraging by the aye-aye, Daubentonia madagascariensis. — Am. J. Primatol. 35: 235-240. Ganzhorn, J.U. (1994): Les lémuriens. — pp. 70-72 in: Recherches pour le Developpement: Inventaire Biologique de la Forét de Zomibtse, vol. special (eds. Goodman, S.M. & O. Langrand). — Centre d’Information et de Documentation Scientifique et Technique, Antananarivo. Ganzhorn, J.U., ©. Langrand, P.C. Wright: S @O7@onmom whe Rakotosamimanana, A.T.C. Feistner & Y. Rumpler (1996/1997): The state of lemur conservation in Madagascar. — Primate Conserv. 17: 70-86. Gascon & Lovejoy 1998 [s.S. 2] Harcourt, C. & J. Thornback (1990): Lemurs of Madagascar and the Comoros. — Gland, Switzerland and Cambridge, U.K. Henderson, M.T., G. Merriam & J. Wegner (1985): Patchy environments and species survival: Chipmunks in an agricultural mosaic. — Biol. Conserv. 31: 95- 105. Jenkins 1987 [s.S. 1] Kattan G.H., H. Alvarez-Lopez & M. Giraldo (1992) orest fragmentation and bird extinctions: San Antonio eighty years later. - Conserv. Biol. 55: 269-281. Langrand & Wilme 1997 [s.S. 1] Laurance, W.F. (1991): Ecological correlates of extinction proneness in Australian tropical rain forest mammals. — Conserv. Biol. 5: 78-89. Lindenmayer, D.B.,R.B. Cunningham & C.F. Donmelly, @993)- ihe conservation of arboreal marsupials in the montane ash forests of the central highlands of Victoria, south-east Australia, IV. The presence and abundance of arboreal marsupials in retained linear habitats (wildlife corridors) within logged 295 forest. — Biol. Conserv. 66: 207-221. Peauemmcerens RA WR, Konstant, M,.E. Nicoll & OF Langrand (1992): Lemurs of Madagascar: An Action Plan for their Conservation. 1993-1999. — IUCN/SSC Primate Specialist Group, Gland, Switzerland. Pameecmuentcr, KR. AL, Lk oH. Rakotovao, V. Randrianasolo; E.J. Suemiime & Do Devitre (1987): Briorites en Matiere de Conservation des Especes de Madagascar. — International Union for Conservation of Nature (IUCN), Gland, Switzerland. Moenimmeten, KoA. 1: Tattersall, WR. Konstant, D.M. Meyers & R.B. Mast (1994): Lemurs of Madagascar. — Conservation International, Washington, D.C. Nelson, R. & N. Horning (1993): AVHRR-LAC estimates of forest area in Madagascar, 1990. — Int. J. Remote Sensing 14:1463-1475. Nicoll, M.E. & O. Langrand (1989): Madagascar: Revue de la Conservation et des Aires Protégées. — World Wide Fund for Nature, Gland. Remoranin.old,@-. 5. Meier, € Langer, W. Scheffrahn & Y. Rumpler (1996): Population genetics of Eulemur macacao macaco (Primates: Lemuridae) on the islands of Nosy-Be and Nosy-Komba and the peninsula of Ambato (Madagascar). — Primates 37: 215-225. Reuehanad. AG. & S. O Connor (1997): Degradation, transformation, and conservation: the past, present, and possible future of Madagascar’s environment. — pp. 406-418 in: Natural Change and Human Impact in Madagascar (eds. Goodman, S.M. & B.D. Patterson). — Smithsonian Institution Press, Washington D.C. Saunders, DA R.J. Hobbs & E.R. Margules (1991): Biological consequences of ecosystem fragmentation: A review. — Conserv. Biol. 5: 18-32. emma & R. Smolker (1998): Lemurs im the Reserve Special d’Anjanaharibe-Sud. — pp. 227-238 in: A floral and faunal inventory of the Réserve Special d’Anjanaharibe-Sud, Madagascar: with reference to elevational variation (ed. Goodman, S.M.). — Fieldiana Zoology 90. Field Museum of Natural History, Chicago. Smith, A.P. (1997): Deforestation, fragmentation, and reserve design in western Madagascar. — pp. 415-441 in: Tropical Forest Remnants, Ecology, Management and Conservation of Fragmented Communities (eds. Laurance, W.F. & R.O. Bieregaard). — University of Chicago Press, Chicago. Smith et al. 1997 [s.S. 1] Sterling, E.J. & M.G. Ramaroson (1996): Rapid assessment of the primate fauna of the eastern slopes of the Réserve Naturelle d’Andringitra, Madagascar. — pp. 293-305 in: A floral and faunal inventory of the Réserve Naturelle Intégrale d’Andringitra, Madagascar: with reference to elevational variation (ed. Goodman, S.M.). — Fieldiana Zoology 85. Ben IJ: Mahoney, AD. Saunders, R.A, Duncan, S.P. Kelly & M.F. Coffin (1995): Timing of hotspot-related volcanism and the breakup of Madagascar and India. — Science 267: 852-855. Tattersall, I. (1982): The primates of Madagascar. - Columbia University Press, New York. 296 Tomiuk, J.,L. Bachmann, M. Leipholdt, S. Atsalis, P.M. Kappeler, J. Schmid & J-U] Ganzhorm ISIS) the impactor cencricsonmine conservation of Malagasy lemur species. — Folia Primatol. (suppl. 1) 69: 121-126. Turner, I.M. (1996): Species loss in fragments of tropical rain forest: a review of the evidence. — J. Appl. Ecol. 33: 200-219. Whitesides, G.H TE. Oates, S.-M: Green eRe] Kubrerdamne (1988): Estimating primate densities from transects in a West African rain forest: a comparison of techniques. — J. Anim. Ecol. 57: 345-367. Author’s addresses: Dr. Jutta Schmid, Institute of Zoology, Hamburg University, Martin-Luther-King- Platz 3, 20146 Hamburg, Germany jutta.schmid @t-online.de 291] Fragmentation effects on reptile and amphibian diversity in the littoral forest of southeastern Madagascar Jean-Baptiste Ramanamanjato Abstract: QMM (QIT Madagagascar Minerals ), a Malagasy company owned by QIT-Fer and Titane Inc. (subsidiary of Rio Tinto plc, UK) and the State of Madagascar intend to exploit deposits of titanium dioxide principally in the form of ilmenite minerals near Tolagnaro (Fort Dauphin) in southeastern Madagascar. Most of the proposed mining area consists of highly degraded ecosystems, but some areas include wetlands and some of the remaining littoral forest of Madagascar. Conservation zones will be established and the mined areas will be rehabilitated progressively after exploitation. In order to assess possible effects of mining, and to arrive at a better understanding of the size and number of conservation zones required to preserve representative components of this ecosystem, a series of intensive impact studies have been carried out over the last few years. The present paper concentrates on the occurrence of reptile and amphibian species in differently sized fragments of the littoral forest. According to the present results the species numbers decline substantially in fragments smaller than 200 — 300 ha. Extinction with declining fragment size is non-random. Therefore a series of small fragments does not provide the biodiversity found in one or a few large fragments. Key words: Fragmentation, reptiles, amphibians, littoral forest, Madagascar Introduction Despite its famous biodiversity and high degree of endemism, there have been surprisingly few studies on the effects of fragmentation on the fauna of Madagascar (Langrand & Wilmé 1997, Ganzhorn et al. 2000, Andrianarimisa et al. in press, Goodman & Rakotondravony in press). In particular, effects of habitat fragmentation on the herpetofauna are unknown. Reptiles and amphibians are well suited for analysing fragmentation effects because of their high diversity (320 species of reptiles and at least 200 species of amphibians) and their specific habitat requirements. In addition, 95% of Madagascar’s reptile species and 99% of the amphibian species are endemic to the island (Blanc 1985, Blommers-Schlosser & Blanc 1991, Glaw & Vences 1994, Raxworthy & Nussbaum 1997). Due to their specific habitat requirements, herps are probably among the most suitable taxa to monitor human-induced or ecological changes which take place everywhere in Madagascar, and in the southeastern region (Anosy) in particular. The remaining forests of Madagascar are subject to a diverse range of threats (Green & Sussman 1990, Gade 1996, Smith 1997). The lowland rain forest has been classified as one of the most threatened vegetation zones on Madagascar (Ganzhorn et al. 1997). Vast areas have already been cleared, particularly on the coastal plain, and only a small fraction of the original cover remains (Du Puy & Moat 1996). The Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 298 vulnerability due to successive attacks by man is exemplified by the forests of Anosy in southeastern Madagascar. At the beginning of the century some 60% of the region was forested. Now, forest cover has become highly fragmented and decreased to about 40% (Rakotoarisoa & Wright 1997). Extensive grasslands are important barriers between the large block of the remaining humid forest stretching along the eastern escarpment and the littoral forest on the coastal plain. In addition, the littoral forest has been broken up by people into more than a hundred fragments. The most important causes of the degradation and fragmentation of littoral forest are charcoal production, timber harvesting, grazing, and grass fires. To save this type of habitat, QIT Madagascar Minerals established a conservation and rehabilitation program. The goal of this program is to define an ecological management plan for the flora and fauna of the coastal plain during and after mining. In order to achieve this goal, the present study addressed the following questions: |. What are the effects of habitat fragmentation and degradation on the reptile and amphibian communities? 2. Which species occur within the project exploration zone that might be at risk of extinction or become vulnerable because of fragmentation? 3. What is the minimum size of fragments that might allow persistence of the native herpetofauna until the Conservation/Rehabilitation Program of QMM becomes effective? Study sites and methods The littoral forests of Mandena and Sainte Luce were selected as study sites (fig.1). Though they are part of the lowland rainforest of Madagascar’s Eastern Domain, the southeastern forests growing on coastal sands are considered a distinct phytogeographic unit (Lowry & Faber-Langendoen 1991). Canopy height is 10 to 15 m, the diameter at breast height (dbh) of the largest emergent trees is generally less than 50 cm (Goodman et al. 1997). Littoral forest appears to be composed of rather dense stands of short evergreen arborescents with stiff, rounded leaves, many of which are covered with a waxy cuticle. Annual precipitation exceeds 1000 mm per year (Goodman et al. 1997). The species composition of sites in the nearby Réserve Naturelle Integrale (RNI) d’ Andohahela (63 100 ha +) was used for comparison since that might reflect the species composition of the littoral forest before fragmentation (Nussbaum et al. 1999). This comparison may not be legitimate as the coastal forests grow on poor sandy soil while the forest of Andohahela grows on lateritic soils (Goodman et al. 1997, Goodman 1999). Nevertheless, this is the best comparable database available for the time being. Surveys were carried out in three phases: (1) fifteen sites of lowland and mid-altitude rainforest were surveyed between August and December 1989 and between September 1990 and January 1991; (2) littoral forest was surveyed by QIT in 1998/99; (3) to complete the regional biodiversity inventory, surveys were also undertaken in many humid and dry forest sites in the Tolagnaro region. This field work was done between 1989 and 1998 by various researchers (Creighton 1992, Ramanamanjato 1993, Raselimanana 1993, Nussbaum & Raxworthy 1994, Raxworthy & Nussbaum 1996, 1997, Nussbaum et al. 1999). 299 Table 1: Species list and distribution within fragments = Sainte Luce; M = Mandena; for the analysis of herpetofauna preference, we used the following broad categories (Goodman et al. 1997) : Habitat: Forest (F)- species restricted to continuous tracts of disturbed or intact forest; generalist (G) — species that use forest and open habitats; Aquatic (A) — species that are found in wetland habitats in littoral zone only (excluding the sea). Lifestyle: T = terrestrial; A = arboreal. Activity: N = nocturnal; D = diurnal ; C = crepuscular; U = unknown Species / Fragment label S9 M15 S17 M3 SI/2 S7 S8 MA M5 M6 M7 Habi- Life Acti- /16 tat style vity Size (ha AS IIEDIAEDDITEDTDIED NEE TEA Amphibians Ptychadena mascareniens. l l l l 1 l | 1 | | G Al € Heterixalus boettgeri l l l l 1 | l l | G A € Mantidactylus domerguei l l l l l l l l | G il (E Mantidactylus ulcerosus 1 1 l 1 1 l l 1 G T C Mantidactylus betsileanus 1 1 1 ] 1 1 l 1 G T @ Mantidactyl. bicalcaratus l l 1 l l l l E A D Mantidactyl. depressiceps l l l l l 1 l E A € Mantidactylus punctatus 1 l l l 1 1 E A (€ Mantidactylus n. sp. 1 1 1 1 1 l 1 1 F 1 C Anodontohyla boulengeri l | 1 1 1 F A @ Plethodontohyla bipunct. l l l l l l F ar @ Plethodontohyla notosticta l l l l l F A € Stumpffia tridactyla l 1 1 1 E a0 @ Aglyptodactyl. madagasca. 1 1 1 1 F T C Mantidactylus biporus 1 1 1 G T € Plethodontohyla alluaudi 1 1 1 F T U Boophis opisthodon l l l G A N Mantidactylus majori 1 je ar € Scaphiophryne calcaratus 1 Is aly C Boophis madagascariensis F A N Paradoxophyla palmata 1 F ili U Reptiles Mabuya elegans l l 1 l 1 l l 1 l 1 1 G JE D Mimophis mahfalensis 1 l i l l l l l l 1 l G 1 D Hemidactylus mercatorius 1 1 1 1 1 1 1 1 1 1 G A N Phelsuma modesta 1 l l 1 l l 1 l 1 l l G A D Furcifer lateralis 1 1 1 1 1 1 1 1 1 1 G A D Furcifer oustaleti l 1 l 1 l l l l | l G A D Liopholidophis lateralis i l 1 l l l l G Di D Mabuya gravenhorsti 1 1 l 1 l 1 1 l 1 | G ar D Madagascarophis colubri. l 1 l l l 1 1 l l l G aT N Phelsuma lineata 1 l l l 1 1 l | l l G A D Amphigloss. melanopleura 1 1 1 1 1 1 1 1 1 l F T D Geckolepis maculata l 1 l l l l 1 l l F A N Phelsuma quadriocellata 1 1 l l l l l l 1 F A D Liophidium torquatus l 1 l 1 l l F i D Lygodactylus tolampyae 1 F A D Typhlops arenarius 1 l l l E ae U Amphiglossus macrocercus 1 l l l 1 l 1 1 F at D Ithycyphus oursi 1 ] 1 1 l 1 E A D Amphiglossus ornaticeps l i l l l l F ar D Boa dumerili l l l 1 l G T N Boa manditra l l l 1 E A N Leioheterodon madagasc. 1 1 l 1 G ar D Liophidium rhodogaster 1 l l B Al D Liophidium vaillanti 1 l l G ar D Lycodryas gaimardi 1 l l F A N Amphiglossus melanurus l l F ir D Langaha madagascariensis 1 | F Al Phelsuma antanosy 1 l 1 F A D EO miu im WS Gl 300 S9 M15 S17 M3 S1/2 S7 S8 M4 M5 M6 M7 Habi- Life Acti- /16 tat style vity 212 Species / Fragment label Size (ha 457 Pseudoxyrhopus kely Amphiglossus astrolabi Brookesia nasus Calumma nasuta Ithycyphus goudoti Liopholidophis stumpffi Liopholidophis sexlineatus Zonosaurus aenus Zonosaurus maximus Lycodryas arctifasciatus Lycodryas betsileanus Lygodactylus miops Uroplatus sikorae a5 a laa} A a > »>>>> + > >>-H ze zzegeedgdg/de@e zZ [ER EEE GE GE EEE EEG Sl 47 36 41 DIE PRS AY DAR 29 21 IG BO IS [Seve lS) Shea cpa 6 Sess Zaalk 20. 71 14 yeas: ©) 7 Total number of species No. of generalist species No. of forest + marsh- dependent species Proportion of generalist species 37% 45% 44 49% 52% 54% 56 62 78% 92% 100 The size of each fragment, shown in table I, was defined using a point-count grid of 0.1 ha. Eleven fragments between 10 and 457 ha were selected for the present analysis. Selected fragments had regularly shaped edges whenever possible. Fragments were separated by grassland, bare sand, or denuded marshes. Most of the fragments had been isolated for at least 50 years. Table 2: Survey efforts MD = man-days spent at each fragment; DH = diurnal search hours; NH = nocturnal search hours; DF = drift fence (pit fall trap days) Year Site MD DH NH DF 1989/90 S9 22) 78 16 1110 1998/99 S9 36 116 30 144 1989/90 M 15/16/20 Des) 78 10 915 1998/99 M 15/16/20 64 256 16 1292 1989/90 S17 7 2 8 0) 1998/99 $17 2 V2 10 144 1998/99 M3 Di 28 14 Dail 1989/90 $1/2 By 128 32 319 1989/90 S7 i 6 0 0 1998/99 S7 7 14 Da 1989/90 S8 4 12 0 0 1998/99 S8 2 8 0 144 1998/99 M4 21 15 2 144 1998/99 M5 21 15 8 144 1998/99 M6 2 15 8 144 1998/99 M7 al 15 8 0) 301 Survey methods include pitfall traps and direct searching (Ramanamanjato 1993; Raselimanana 1993). Direct searching included walking of transects during the day and at night. Possible refuges were excavated with a stump ripper. Pitfall traps consisted of 15 | buckets that were placed in the ground at 10 m intervals along 100 m drift fences. Two to six drift fences were erected in each fragment and monitored daily for three to ten days depending on fragment size. The total sampling effort and pitfall traps are summarized in table 2. Vouchers and reference specimens were collected for taxonomic identification. For each specimen, date and time of capture, activity, locality, microhabitat and biotope were collected. Deviations of species composition in different fragments from random assemblies of species were assessed by using the program “Nestcalc; Nestedness Calculator” provided by Atmar and Patterson (1993). 7 Fort Dauphin (Tolagnaro) Fig.1: Location of study sites; forest areas marked in black 302 Results and discussion General results A total of 296.5 man-days of sampling effort were accumulated in the study, 866 diurnal search hours and 164 search hours at night. A total of 4902 drift fence/pitfall trap-days were accrued. Within the 11 littoral forest fragments, a total of 83 species of reptile and amphibian were recorded. Aquatic and grassland specialist species found only outside the forest were excluded from this list. Thus the final species/area matrix used for this analysis included 62 species (41 reptiles, 21 amphibians) within 11 fragments (table 1). In the large control area of the RNI d’Andohahela, 77 reptile and amphibian species had been recorded (Nussbaum et al. 1999). Apart from the large control area, most species were found in the forest fragments S9 and in M15/16. | According to fig.2, the number of amphibian species seems to decline monotonically with decreasing fragment size, while the number of reptile species seems to decline discontinuously, with a dramatic drop in fragments smaller than around 200 ha, followed by a stabilization, then a final drop at around 50 ha. 40 e Reptiles m Amphibians 30 20 No. of species Fig.2: Species — area realtion- ships between the number of O 100 200 300 400 500 reptile and amphibian species and fragment size in the littoral Fragment size [ha] forests of southern Madagascar The presence/absence matrix of species in different fragments deviates significantly from random species assemblages (table 1). The Nestedness Calculator Program (Atmar & Patterson 1993) assigns a temperature of 5.79° (see also Patterson & Atmar 2000, this volume) to the amphibian assemblages and 13.78° to the reptile assemblages of the 11 fragments of littoral forest. The temperature for the presence/absence pattern of the combined reptile and amphibian species assemblages is 12.11°. Random assemblages, based on 1000 simulations would have temperatures (+ standard deviation) of 53.23° + 6.41°, 57.47° + 4.87°, and 59.52° + 3.98° respectively. Thus, the assemblages of all three groups deviate from random assemblages (p < 0.001 in all cases) and are consistent with nested patterns. 303 Vulnerability and life history characteristics The effects of fragmentation can be seen at several levels of biological organization. At the species level, Mlandenoff (1997) describes three possibilities how a given species can persist in a highly fragmented landscape. First, a species might survive or even thrive in the matrix of human land use; second, species with small home-ranges or otherwise modest area requirements might survive by maintaining viable populations within individual habitat fragments; third, highly mobile species might integrate several patches of suitable habitat, either into individual home-ranges or by exchanging individuals between subpopulations. In the littoral forest, the grassland between patches constitutes an important barrier for forest-dwelling species. This is illustrated by the fact that the decline in reptile and amphibian species seen in fig.2 is paralleled by a shift in the representation of forest-dependent species. The proportion of forest-dependent species decreased from 67% in the 457 ha fragment to 0% in the 10 ha fragment (fig.3). In the smallest fragment (10 ha) only five species of reptile and one species of amphibian were recorded during the survey work. All of them are also found in degraded forests and open areas. The complete lack of forest-dependent species in the 10 ha fragment indicates that a block of forest of this size is either unsuitable for maintaining populations of herpetological species that are large enough to survive for any length of time, or that a block of this size is not perceived as “forest” any more by the forest-dependent herpetological species. Thus, the degree of vulnerability of a given species due to forest fragmentation is likely to be related to its tolerance to habitat change and its capability to use or to bridge the new anthropogenic matrix around the remaining forest fragments (Laurance 1991, Gascon & Lovejoy 1998). 100 -¢ ig. 37 Kereentaee or TEMERAINSE Depele and amphabran species in herpetological communities in littoral forests of different size. The square represents the percentage of generalist 0 Species im ‚the larse forest 0 100 200 300 400 500 (63 000 ha +) of the Réserve INFa te ueneslaler ai netvere mative Fragment size [ha] d’ Andohahela (Parcel 1). % Generalist species Among the categories of species predicted to be most vulnerable are the following: (1) naturally rare species: species with limited or patchy geographic distributions and species with low population densities; (2) wide-ranging species: animals requiring distinct habitats for different phases of their life cycle, such as amphibians, are also vulnerable to barriers; (3) non-vagile species: species with poor dispersal abilities that may not travel far from where they were born, or may be stopped by barriers; (4) species with low fecundity; (5) species with short life 304 cycles; (6) species depending on patchy or unpredictable resources or species that have highly variable population sizes for other reasons; (7) ground-nesters: nesting on or near the ground is another life history trait that seems ill suited to the ecological conditions in fragmented landscapes; (8) large patch or interior species: some species are absent from small patches with little or no true interior habitat; (9) species vulnerable to human exploitation or persecution, such as snakes and frogs that might be killed on sight (for a review see Noss & Csuti 1997). Applying these criteria for vulnerability listed above to the herpetofauna of the littoral forest in the Fort Dauphin region, the following species might be at risk of extinction in the future. Two forest-dwelling species (Phelsuma antanosy and Pseudoxyrhopus kely) that have a very restricted range of distribution and are presumed to be endemic to the littoral forest. Both are found only in very specific microhabitats. Phelsuma antanosy was discovered during the 1990/91 field study at three localities: Petriky (a forest west of Fort Dauphin, not considered here), Sainte Luce S1-S2, S7, and S8. Later, another population was discovered at Ambatororongorongo, a transitional forest about 4-5 km west of Petriky. The species seems to depend on specific palms where they lay their eggs. The exploitation of Pandanus and the endemic palm Dypsis saintelucei affect the population directly or indirectly and constitute a possible cause of extinction (Ramanamanjato & MacIntyre unpubl.). In addition, both species are only found in primary forest with a closed canopy. Between 1990 and 1994, Petriky forest was considerably reduced in size and the population of P. anatanosy disapeared from this site. Pseudoxyrhopus kely was also discovered during the 1990-1992 field work at two sites: M3 of Mandena and S17 of Sainte Luce. Presumed to be endemic to the littoral forest, this burrowing snake seems to occur only in patches of forest above 212 ha. It was never found in smaller fragments or in degraded forests. All of the five individuals encountered so far were seen in closed canopy forest, either under logs or after entering the pitfall traps. This species might be theatened by disruption of the closed canopy. Thus both species belong to the first categories of species vulnerability. Four species with specific habitat requirements might also be a matter of concern. These are Mantidactylus punctatus, Mantidactylus depressiceps, Paradoxophyla palmata, and Amphiglossus astrolabi. The two frog species Paradoxophyla palmata and Mantidactylus punctatus have specialized habitat requirements. The latter relies on streams and the former on Pandanus plants. Amphiglossus astrolabi inhabits wetlands and is classified among the “Wide-Ranging Species”. However, if wetlands become isolated in a matrix of grassland or bare soil, these species will not be able to migrate between patches and are likely to go extinct. Another 17 rain forest species could be localized only in the largest blocks of the littoral forest (M15/16 and S9) with closed canopy. They are absent in small patches with little or no true interior. Human activities continue to reduce the size of M15/16 and blocks S9 and thus increase their risk of extinction as a result of forest fragmentation. The species are: Brookesia nasus, Calumma nasuta, Liopholidophis stumpffi, L. sexlineatus, Stenophis arctifasciatus, S. gaimardi, S. betsileanus, Ithycyphus goudoti, Langaha madagascariensis, Boa manditra , B. dumerilii, 305 Boophis madagascariensis, B. opisthodon, Mantidactylus majori, Aglyptodactylus madagascariensis, Amphiglossus melanurus, Lygodactylus miops, and Scaphiophryne calcaratus. Most of these species are common in the eastern rain forest, such as in the RNI d’Andohahela with 63 100 ha, in Ranomafana National Park with 40 000 ha or in Andringitra Nature Reserve with 31 160 ha (Nicoll & Langrand 1989, Raxworthy & Nussbaum 1996, Nussbaum et al. 1999). In the remaining littoral forests they are restricted to the larger fragments with closed canopy forest. According to the present results, littoral forest fragments smaller than 200-300 ha cannot retain a substantial proportion of the original faunal biodiversity. Fragments below that size, such as M4, M5, M6, and M7, are clearly impoverished. The loss of species with declining fragment size is not random but strongly deterministic. This means that from a conservation point of view one or two large forest fragments cannot be replaced by several smaller ones. It seems crucial that the ratio of forest area to edge remains high, as many species seem to react negatively to opening of the forest canopy and to yet unspecified edge effects. Management plan and long-term monitoring To ensure preservation of the herpetofauna of the littoral forest the following conservation actions are suggested: 1. The few existing large blocks of littoral forest (such as S9 of Sainte Luce and M15/16 of Mandena) should be protected immediately to ensure the survival of the communities listed above. 2. Corridors should be established to link some of the smaller fragments and ultimately increase the surface area of the littoral forest blocks and to help exchange individuals between fragments. QMM is currently establishing experimental corridors with native and exotic tree species to link three small fragments (M4-M5 and M5-M6) and to define options for habitat restoration. 3. In order to be successful, we need to know more about autecological aspects and the life history of various species. Several projects addressing these issues are under way. Acknowledgements The surveys in the Tolagnaro region were funded by QIT Madagascar Minerals. Travel and stay in Germany were funded by the Museum Alexander Koenig in Bonn, the City of Bonn, private donations and QIT Madagascar Minerals SA. I also wish to thank C.M. Naumann, Director of the Museum, for inviting me to attend this symposium, R. van den Elzen for her untiring help during all stages of the Symposium and G. Rheinwald for his skillful editorial work. Thanks are to J. Ganzhorn for his comments on the manuscript. The first phase of survey was made possible by the contributions of R.A. Nussbaum, C.J. Raxworthy, and A.P. Raselimanana. Research permits were issued by the Ministere des Eaux et For£ts. 306 References Andrianarimisa, A.,E Bachmann, J). U. Ganzhorn,Ss Mi Goodman & J. Tomiuk (in press): Effects of forest fragmentation on genetic variation in endemic understory birds in central Madagascar. — J. Ornithol. Atmar, W. & B.D. Patterson (1993): The measure of order and disorder in the distribution of species in fragmented habitats. — Oecologia 96: 373-382. Blanc, E.P. (1985): Priorités en matiere de conservation’ des Teptilesmer amphibiens a Madagascar. — pp. 117-120 in: Priorités en matieres de conservation des especes a Madagascar (eds. Mittermeier, R.A., L.H. Rakotovao, V. Randrianasolo, E.J. Sterling & D. Devitre). - IUCN, Gland. Blommers-Schlösser,R.M. A. &€.P. Blane (1991): Amphibiens Laune de Madagascar Edition. Vol. 75. — Muséum national d’Histoire naturelle, Paris. Creighton, G.K. (1992): Madagascar Minerals Project. Environmental impact assessment. Part I: Natural resources, Appendix IV. — Lewis Environmental Consultants, Montreal. Du Puy, D.J. & J. Moat (1996): A refined classification.or themonmmany, vegetation of Madagascar based on the underlying geology: using GIS to map its distribution and to assess its conservation status. — pp. 205-218 in: Biogeography of Madagascar, (ed. Lourenco, W.R.). - ORSTOM, Paris. Gade, D.W. (1996): Deforestation and its effects in highland Madagascar. — Mountain Research and Development 16: 101-116. Ganzhorn, J.U., B. Rakotosamimanana, lL’ Hannah, J bomen ee Iyer, S. Olivieri, S. Rajaobelina, € Rodstrom 2 Grain (1997): Priorities for Biodiversity Conservation in Madagascar. Primate Report 48-1. - Goltze, Göttingen. Ganzhorn, J.U., S.M. Goodman, J.-B. Ramanamany vor ER Ralison, D. Rakotondravony & B. Rakotosamimanana (2000): Effects of fragmentation and assessing minimum viable populations of lemurs ın Madagascar. — pp.265-272 in: Isolated vertebrate communities in the tropics (ed. Rheinwald, G.) — Bonn. zool. Monogr. 46. Gascon, € & T.E. Lovejoy (1998): Ecological” impacts 07 or fragmentation in central Amazonia. — Zoology 101: 273-280. Glaw, F. & M. Vences (1994): Amphibians and Reptiles of Madagascar, 2nd Edition. — Eigenverlag, Köln. Goodman, S.M. (1999): A Floral and Faunal Inventory of the Réserve Naturelle Integrale d’Andohahela, Madagascar: with reference to elevational variation. — Fieldiana: Zoology, new series. Vol. 94, Field Museum Natural History, Chicago. Goodman, S.M. & D. Rakotondravony (in press): The effects omionrest fragmentation and isolation on mammal communities (Lipotyphla: Tenrecidae) on the Central High Plateau of Madagascar. — J. Zool. Goodman, S.M., M. Pidgeon, A..F. A. Hawkins Seas Schulenberg (1997): The Birds of Southeastern Madagascar. — Fieldiana: Zoology, new series. Vol. 87, Field Museum of Natural History, Chicago. 307 Green, G.M. & R.W. Sussman (1990): Deforestation history of the eastern rainforest of Madagascar from satellite images. — Science 248: 212-215. Langrand, O. & L. Wilme (1997): Effects of forest fragmentation on extinction patterns of the endemic avifauna on the central high plateau of Madagascar. — pp. 280-305 in: Natural Change and Human Impact in Madagascar. (eds. Goodman, S.M. & B.D. Patterson). — Smithsonian Institution Press, Washington, D.C. Paunance, \W.E. (1991): Ecological correlates of extinction proneness in Australian tropical rain forest mammals. — Cons. Biol. 5: 79-89. Lowry, P.P. II. & D. Faber-Langendoen (1991): Madagascar Minerals Project. Environmental impact assessment. Part I: Natural environment, Appendix III: Flora and vegetation study. — Lewis Environmental Consultants, Montreal. Mlandenoff, D. J. (1997): Quantifying landscape pattern and fragmentation. — pp. 279-280 in: Principles of Conservation Biology, 2nd ed., (eds. Meffe, G.K., C.R. Carroll & Contr.). — Sinauer Associates, Inc. Publishers, Sunderland, MA. Nicoll,M. & O. Langrand (1989): Madagascar: Revue de la Conservation et des Aires Protégées. - WWF, Gland, Switzerland. Noss, R.F. & B. Csuti (1997): Habitat fragmentation. — pp. 269-304 in: Principles of Conservation Biology, 2nd ed. (eds. Meffe, G.K., C.R. Carroll & Contributors). — Sinauer Associates, Inc. Publishers, Sunderland, MA. Nussbaum, R.A. & C.J. Raxworthy (1994): The genus Paragehyra (Reptilia: Sauria: Gekkonidae) in southern Madagascar. — J. Zool. 232: 37-59. Nussbaum, R.A., C.J. Raxworthy, A.P. Raselimanana & J.-B. Ramanamanjato (1999): Amphibians and reptiles of the Réserve Naturelle Integrale d’Andohahela, Madagascar. — pp. 155-173 in: A Floral and Faunal Inventory of the Réserve Naturelle Integrale d’Andohahela, Madagascar: with special reference to elevational variation (ed. Goodman, S.M.). — Fieldiana: Zoology, new series. Vol. 94, Field Museum of Chicago, Chicago. Patterson, B.D. & W. Atmar (2000): Analyzing species composition in fragments. — pp. 9-23 in: Isolated Vertebrate communities in the tropics (ed. Rheinwald, G.). - Bonn. zool. Monogr. 46 Rakotoarisoa, J. A. & H.R. Wright (1997): Cultural transformations and their impacts on the environments of Madagascar. — pp. 309-330 in: Natural Change and Human Impact in Madagascar, (eds. Goodman, S.M. & B.D. Patterson). — Smithsonian Institution Press, Washington, D.C. Ramanamanjato, J.-B. (1993): Contribution a l’etude des reptiles et amphibiens de la forét ombrophile du Sud-Est de Madagascar. - DEA Thesis. Université d’ Antananarivo, Antananarivo. Raselimanana, A.P. (1993): Contribution a |’ étude de la batrachofaune et de l’herpetofaune de zones semi arides et des zones littorales de l’extreme Sud-Est de Madagascar. — DEA Thesis. Université d’ Antananarivo, Antananarivo. Raxworthy, C.J. & R.A. Nussbaum (1996): Patterns of endemism for terrestrial vertebrates in eastern Madagascar. — pp 369-383 in: Biogeography of Madagascar, (ed. Lourenco, W.R.). - ORSTOM, Paris. 308 Raxworthy, C.J. & R.A. Nussbaum (1997): Biogeographic patterns of reptiles in eastern Madagascar. — pp. 124-141 in: Natural Change and Human Impact in Madagascar, (eds. Goodman, S.M. & B.D. Patterson). — Smithsonian Institution Press, Washington, D.C. Smith, A.P. (1997): Deforestation, fragmentation, and reserve design in western Madagascar. — pp. 415-441 in: Tropical Forest Remnants, Ecology, Management and Conservation of Fragmented Communities, (eds. Lawrence, W. & O.W. Bierregaard). — University of Chicago Press, Chicago. Author’s address: | Jean-Baptiste Ramanamanjato, QIT Madagascar Minerals S.A., BP 225, Tolagnaro (614), Madagascar, E-mail: jb-rama@dts.mg, Fax: +261 20 92 214 13, Tel: +261 2092218 91 309 Using fossils and phylogenies to understand evolution of reptile communities on islands BeNe Arnold Abstract: In the Mascarene islands, discovery of relict populations and fossils, together with taxonomic revision, have enabled the endemic reptile fauna existing before reduction by human colonization to be substantially reconstructed, showing it to be one of the richest oceanic reptile assemblies, with 33 species of tortoises, lizards and snakes and a further four distinctive populations of lizards. Investigation of the history of such island communities requires phylogenies of the groups concerned. These can be used to estimate original numbers of colonizations, their geographic origin (which may be from more than one direction) and patterns of inter-island spread. They also enable adaptation to island conditions to be more securely recognised and enhance investigation of such common island phenomena as change in body size, shift in niche and adaptive radiation. Relative rates of evolution on mainlands and on different kinds of islands can also be studied and it is possible to ask whether particular island communities turn over rapidly, and why island populations are so vulnerable to incoming predators and competitors. Finally the possibility can be considered that islands are “black holes” from which mainlands are rarely if ever colonized. As well as the Mascarene islands, examples are taken from the Sogotra, Canary and Cape Verde archipelagos. Key words: Mascarene islands, Sogotra, Cape Verde islands, Canary islands, colonization, adaptation, radiation. Introduction Islands and archipelagos are favourite places to study evolution. They usually have the advantages of apparently simple communities that often originated quite recently and some of the selective forces that helped mould them frequently seem evident. Yet it is also clear that what is being studied is likely to be fragmentary, or at least very different from what was present even in the recent past, for the arrival of people on previously uncolonized islands has almost always devastated their biotas. The equilibrium theory of island biogeography suggests there may also have been considerable natural change in species composition in many cases. As the history of lineages affects how evolution and especially adaptation takes place in them, it is necessary to know as much about the past of island communities as possible and this is also important as a basis for designing conservation strategies. While history once tended to be relatively neglected in island studies, the past few decades have seen greater interest and a number of approaches have been employed to obtain a fuller understanding of island communities. These include: 1. searching islands and archipelagos as thoroughly as possible for relict populations of once more widespread forms; 2. revising current taxonomies, often exposing unknown taxa; Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 310 3. looking for fossils; 4. using old accounts and old museum specimens; 5. using phylogenies to reconstruct clade history. Some of these approaches will be illustrated using examples from West Indian Ocean and North Atlantic reptile communities. Reconstructing the reptile fauna of the Mascarene islands The Mascarene islands comprise, from west to east, the islands of Réunion, Mauritius and Rodrigues and lie some 500 km east of Madagascar. They are classic volcanic oceanic islands and had no previous connexion with other areas of land. Some fifty years ago, their known living endemic reptile fauna was quite modest, consisting of perhaps eight recognised species. Since that time additional relict but surviving species have been discovered. These include three species of night gecko on Mauritius, Nactus serpensinsula (Loveridge, 1951), N. durrelli Arnold & Jones, 1995 and N. coindemerensis Bullock et al., 1985, two kinds of day gecko on Réunion, Phelsuma borbonica Mertens, 1966 and P. inexpectata Mertens, 1966, a new Slit-eared skink, Gongylomorphus fontenayi Vinson, 1974 and additional distinctive populations of this species and G. bojeri. The majority of these forms are now confined to one or more small islets around Mauritius, most lying to the north of the main island. One of these islets, Round Island, was already known to have unique forms, including Gunther’s gecko (Phelsuma guentheri), Telfair’s skink (Leiolopisma telfairi) and two species of bolyerine snake assigned to different genera, Casarea dussumieri and Bolyeria multocarinata. Systematic revision has also shown that the small day geckos (Phelsuma) of Mauritius actually consist of four species (Mertens 1963, 1966, 1970, Austin & Arnold, in progress). Fossils, collected mainly in the nineteenth century, confirmed historical records of a giant gecko (Phelsuma gigas) on Rodrigues, which disappeared in the mid- 1800s, and uncovered no less than five endemic species of Mascarene giant tortoises (Cylindraspis). They also revealed the original presence of a giant skink (Leiolopisma mauritiana) on Mauritius that weighed perhaps ten times as much as its surviving sympatric congener, L. telfairi. More recent fossil discoveries include species of Nactus and Leiolopisma from Réunion (Arnold and Bour, in progress), two more Nactus from Rodrigues as well as two further previously unknown extinct geckos, and extensive remains of Phelsuma edwardnewtoni, which disappeared around 1920 (Arnold, Austin & Jones, in progress). Fossils have also been important in extending ranges and many of the relict forms now confined to islets near Mauritius turn out to have once been widely distributed on the main island (Arnold 1980). Cave deposits additionally provide evidence of recent species turnover which is certainly anthropogenic. Those in Rodrigues contain extensive remains of six recently extinct geckos, but of only one of the five gecko species that occur on the island today, confirming that the others are post- colonization introductions (Arnold, Austin, Jones, in progress). The synthetic approach related above has increased the number of known Mascarene endemic reptile species to around 33 with two additional distinctive Table |: Extinct and surviving endemic reptiles of the Mascarene islands. T - extinct. * - relict. # - distinctive populations exist. Réunion Mauritius Rodrigues Giant tortoises Cylindraspis C. indicat C. ineptat C. vosmaerit C. triserratat C. peltastes fF Day geckos Phelsuma P. borbonica# P. guentheri* P. gigast P. inexpectata* P. cepediana # P. edwardnewtonit P. ornata P. guimbeaui P. roseogularis Night geckos Nactus INEESDN 72 N. serpensinsula* INE SiO Ar N. durrelli* N. sp. BF N. coindemerensis* Waif gecko Lepidodactylus L. lugubris Unidentified geckos Two speciest Mascarene skinks Leiolopisma iL, SDo1r Psteliauni- L. mauritiana F Slit-eared skinks Gongylomorphus G. borbonicat — G. bojeri*# G. fontenayi*# Shore skink Cryptablepharus C. boutoni Bolyerine snakes Casarea C. dussumieri* Bolyeria B. multocarinata*/f? populations of Gongylomorphus and two of Phelsuma (table 1). This makes them one of the richest known oceanic island reptile faunas. Of the 33 species, at least 15 (46%) are extinct, 11 (33% ) reduced to small relicts and only seven (21%) retain substantial ranges. Fossil remains are also useful in reconstructing life-modes of extinct forms. The ratio of stable carbon isotopes, C,,/C,;, in the bones of herbivores reflects the kinds of plants that they ate and it is clear from this that the extinct giant tortoise, Dipsochelys grandidieri, from Madagascar consumed a much higher proportion of grass than other species in the southwest Indian Ocean (Burleigh & Arnold 1986). It has been suggested that the fossil giant Mauritian skink was arboreal and vegetarian, like another very large skink, Corucia zebrata, of the Solomon islands, but its teeth are shaped very like the surviving L. telfairi, an omnivore, and lack vegetarian specialisation. The teeth are also scored and pitted by grit, a common 312 phenomenon in forms that spend a lot of time on the ground but usually much less obvious in ones that climb extensively. Old material in museums has also been important in understanding the Mascarene herpetofauna, especially as they are potential sources of DNA for molecular studies and sometimes this is important in identifying them. The type of the tortoise named as Testudo indica is a shell believed to have been in Paris since the 1670s (Vaillant 1899), and that constituting the type of T. graii has been there since at least the early nineteenth century. Although both these shells were thought to come from the Mascarenes, this was unconfirmed as was their specific island of origin. Mitochondrial DNA from these shells has recently been compared with that from fossil bones collected on particular islands and it is clear that both came from the extinct Réunion species (Austin, Arnold, Bour, in progress). This identification helps resolve nomenclature and it is now clear that the name Cylindraspis indica should be applied to the Réunion form. Allocation of these shells also provides more information about the populations concerned. Unlike fossils, such specimens may retain full shell proportions, colouring and surface growth rings, the latter providing some evidence of growth rate, hatching size and consequently egg-size in these extinct animals. Old museum specimens have also proved a useful source of DNA in other cases. Sequence extracted from specimens collected in 1931 of the extinct large lacertid lizard, Gallotia simonyi, of the Roques Salmor in the Canary islands is identical with that of G. s. machadoi on neighbouring Hierro island, validating the recent reintroduction of lizards to the Roques from this source (Carranza et al. 1999). Similarly, 110 year old museum material recently provided DNA elucidating the relationships of the enigmatic giant skink, Macroscincus coctei of the Cape Verde islands (Carranza & Arnold, in progress). Aldabra atoll, an example of longer term historical change Aldabra atoll, which les about 420 km northwest of the northern tip of Madagascar, has a small contemporary reptile fauna consisting of giant tortoises (Dipsochelys) and three lizards widespread in surrounding areas: the geckos Hemidactylus mercatorius and Phelsuma abbotti, and the skink Cryptablepharus boutoni. Fossils are available from two horizons, the Bassin Cabri calcarinites, about 500 000 years old, and cavity fillings in the Aldabra limestone at Point Hodoul less than 100 000 years old (Braithwaite et al. 1973, Taylor et al. 1979). Giant tortoises, crocodiles and iguanids (Oplurus) occur at Bassin Cabri and also at Point Hodoul where five more lizard species are found that do not occur on the island today: geckos of the genera Paroedura, Geckolepis and Phelsuma, and two skinks, an Amphiglossus and Mabuya cf. maculilabris (Arnold 1976). It is probable that Aldabra was completely submerged after the Bassin Cabri calcarinites were laid down and again after the Point Hodoul deposits (Braithwaite et al. 1973, Taylor et al. 1979). If so, giant tortoises must have colonized the atoll at least three times, and crocodiles and iguanians at least twice each. The atoll had substantial faunal change but submergence means there was no continuity between the three horizons for which information is available. Uses of phylogenies The recent availability of estimates of phylogeny for many groups permits speculations about the history of island communities to be tested more rigorously and some of their uses in this context will be outlined here. Numbers of colonizations Knowledge of phylogeny permits the minimum number of colonizations of islands to be estimated. Essentially, if an island taxon has its closest relatives in an assumed source area, rather than on the island itself, it is likely to represent an independent colonization of the island. Cases where island forms have subsequently invaded other areas may complicate such inference but these seem to be rare. On the basis of this approach, the Mascarenes may have evolved their present fauna from just eight original invaders: single ancestral species of the endemic Cylindraspis, Nactus, Phelsuma, Lepidodactylus, Gongylomorphus, Leiolopisma and Cryptablepharus and a single ancestor of the two genera of bolyerine snakes. As on Aldabra, fossils may increase the known number of colonizations, both by showing that more taxa actually arrived and that some of them arrived more than once. Mainland origins of colonizations Phylogenetic information often permits the geographic origin of island populations to be assessed. Origin is most certain where an island taxon has its sister group in a particular geographic area in which successively deeper branches of the clade concerned also occur. It is often assumed that the closest land mass to an island or archipelago is the most likely source area for colonists and this is indeed often the case. Aldabra clearly gained its successive faunas from Madagascar or the Comores islands or both. Similarly the Galapagos got its biota from nearby South America, and the Canary islands from northwest Africa. In contrast the reptile fauna of the Mascarenes came from two different directions. The closest land mass is Madagascar, about 500km away to the east, and phylogeny suggests that Phelsuma and Gongylomorphus reached the islands from this direction. Other taxa are members of more eastern groups based in tropical Australasia and adjoining areas, including Nactus, Lepidodactylus, Cryptablepharus, and Leiolopisma which is a member of the Australasian and Melanesian eugongylid radiation of skinks (Greer 1990); the bolyerine snakes may also have had an eastern origin. All these forms must have made journeys of at least 4500 km to the Mascarenes, a greater distance than that separating the Hawaiian archipelago (usually considered the most isolated in the world) from the nearest mainland. Such huge journeys are less surprising when it is realised that they are greatly facilitated by dispersal agents such as currents and winds. The Equatorial current runs westwards towards the Mascarenes from tropical Australasia and is capable of transporting natural rafts that might carry propagules; the Southeast Trade winds provide additional westward impetus. Such factors are important in other cases of long-distance colonization. A northward current passing Madagascar probably brought many propagules from there to Aldabra and it is estimated that they could have made the 420 km journey 314 in 3-9 days (Taylor et al. 1979). Primarily North African Tarentola geckos travelled 1400 km between the Canary and Cape Verde islands on the Canary current (Joger 1984, Carranza & Arnold, submitted). This runs at about 14-20 km a day (Guppy 1917), so the journey may have taken about ten weeks. Tarentola also reached Cuba, presumably via the Northern Equatorial current, a distance of over 6000 km (Carranza & Arnold, submitted). Sometimes journeys take place against prevailing dispersal agents, for instance those made between Madagascar and the Mascarenes. These may have been promoted by short-term cyclonic storms the tracks of which sometimes cross both areas. Certainly cyclones can move reptiles very rapidly at times, often over substantial distances (Townsend 1936, Censky et al. 1998). Patterns of inter-island spread Not only do taxa reach archipelagos, they also spread through them. The number of possible patterns of colonization increases non-linearly with the number of islands involved. If simple sequences are considered, with spread from one island to another, the number of possibilities for n islands 1s n factorial n!= 1.2.3... .(n-1)n, that is six possibilities for three islands, twelve for four and 604800 for ten. If branching sequences are included, the number may be n n-1, which is six for three islands, 64 for four and a billion for ten! Even this does not exhaust possibilities, for it ignores situations where sequences branch and there may be different orders of colonization. For three islands there are then twelve possibilities and 144 for four islands. Numbers increase yet again if particular islands are colonized more than once, something that 1s often apparent if speciation has taken place. Fortunately, a robust phylogeny of the island populations involved radically reduces the number of possibilities, although even then different colonization sequences are possible. For instance a phylogeny of four island populations, 1(2 (3,4)), could represent sequential colonization from |, thatis | ->2 -> 3 ->4 or | -> 2 -> 4 ->3, or repeated colonization of the other islands from 3 or 4, for instance: 4 ->1,4->2,4->3. Various approaches are helpful for choosing between possibilities, one being distance parsimony where the colonization pattern involving the shortest route is favoured. Another is journey parsimony where the sequence involving the fewest inter-island journeys is accepted. In practice, this can be accomplished using the MacClade program (Maddison & Maddison 1996) and scoring different islands or island groups as separate states of a single character. Dispersal agents are also important in understanding colonization sequences. For instance, the Canary current not only brought Tarentola to the Cape Verde islands but also spread these geckos southward through the archipelago and the same may be true inthe Canaries (Carranza & Arnold, submitted). Dispersal ability of taxa also affects pattern of spread. If possible ability is best judged in a particular case by the performance of arange of relatives, as sometimes a particular taxon may lose dispersal ability once it has reached an island. Some groups like Tarentola clearly disperse well having made numerous, often long journeys. Others like lacertid lizards have more restricted abilities. It is not surprising therefore that Canary island lacertids (Gallotia) have not emulated 215 Tarentola in reaching the Cape Verdes but have made only modest journeys within the Canaries, each of a 100 km or less. These journeys run successively westwards along the archipelago traversing the prevailing winds and currents. Restricted dispersal ability together with island topology has dictated that these lizards are moved by rare storms rather than the prevailing dispersal agents. Recognizing island adaptations The distinctive features of many island species are often assumed to be adaptations by natural selection to the island situation, but phylogenies show this is not always the case. For instance Sogotra and its satellite islands, off the Horn of Africa, has several endemic species of Semaphore gecko (Pristurus) that look at first sight like an adaptive radiation. But phylogenetic analysis shows that they are not a clade and comprise a series of successive branches near the base of the main lineage of Pristurus and most current mainland species have evolved after them (Arnold 1990, 1993). Also, what look like distinctive features are really primitive characteristics of Pristurus. Sogotra is a continental island that broke away from southern Arabia in the mid-Miocene (Beydoun & Bichan 1970) and its Semaphore geckos may represent lineages and morphologies that were widespread in this area at the time, but are now otherwise extinct. There are a number of possible indicators of island adaptation, that is where a trait has actually developed in response to insular conditions. 1. The feature concerned should actually confer performance advantage in the island selective regime, or at least be likely to do so. This is more convincing if the island situation is markedly different from that in the mainland or other area where the organisms concerned originated. 2. The feature is likely to have originated on the island concerned rather than before colonization took place. 3. The feature has developed independently in similar organisms on separate lineages that have also reached islands. Phylogenies can be helpful here in testing points 2 and 3. They can show that features of island forms are in fact derived compared with those of their mainland relatives, rather than being conserved primitive features, and they can confirm that similar animals on different islands are members of independent lineages when this is the case. Giant tortoises have survived in recent times only on isolated island groups. A member of the subgenus Chelonoidis still occurs wild on the Galapagos. Dipsochelys is still found on Aldabra atoll in the southwest Indian Ocean and until about 1800, was on the Seychelles. Finally the Mascarenes had the five species of the endemic genus Cylindraspis which finally disappeared at about the same time. Mascarene tortoises had many distinctive features (Arnold 1979). The shell was often very thin, sometimes being only 2 mm thick in parts and weighing about a third of that of similar mainland tortoises; it also had a very small belly section, the plastron. The two shell openings where the limbs, head and tail projected were huge and there was no thickening of the shell at the borders of these openings. Some species had primitively rounded shells like most tortoises, but two had independently developed a ‘saddle-back’ configuration in which the front of the shell is raised so that it is as high or higher than the mid-back. All these features 316 were poorly developed in mainland giant tortoises when these existed and have also evolved independently to varying extents in the separate Galapagos and Seychelles and Aldabra giant tortoise lineages, suggesting that they are indeed island adaptations. Their development is very probably related to common characteristics of oceanic islands. The island habitats of giant tortoises all lack large competing herbivores and big predators, so tortoises often attained very high densities. Leguat (1708) noted that on Rodrigues dense herds could be 100 paces across and on the small island of Aldabra there were over 150 000 animals in the 1970s (Coe et al. 1979), although numbers are now falling because food supplies have become exhausted. Such population fluctuations are likely to have been made worse at times by climatic irregularity such as that produced by El Nino. Tortoises were consequently probably intermittently subjected to very strong intraspecific competition for food and in this situation modifying the shell had many advantages. The cost of shell maintenance and carriage was reduced and the open shell meant that animals. probably could be more agile in reaching food in the rugged terrain of their often volcanic islands. The high front of saddle-back shells increased upward reach when browsing by raising the base of the neck which was also elongated. The modified shell allowed this extended neck to be housed when the head was withdrawn into it. In the Galapagos such shells occur on dry islands with sparse vegetation where their lack of streamlining does not impede movement and increased browsing range is particularly beneficial. On Mauritius and Rodrigues rounded and saddle-back species occurred sympatrically on the same islands their different browsing abilities perhaps reducing inter-specific competition (Arnold 1979). While adaptive changes appear to have been driven by strong intraspecific competition, another common insular feature facilitated their development, namely the absence of all or most predators. Many of the changes to the shell involve dismantling the usual anti-predator devices of tortoises which were no longer needed, including a heavy rounded shell with small openings that can be blocked by armoured limbs and which have thickened edges that prevent them being enlarged by carnivores. The large size of giant tortoises has also been considered an insular adaptation evolved as tortoises occupied the vacant large-herbivore niche on oceanic islands. But the tortoises of Madagascar, the apparent source area for the Seychelles and Aldabra populationswere equally big. In at least this case, large size may have facilitated transmarine island colonization, as big tortoises float well and can carry their heads clear of the water, but it was not an adaptation to island conditions as such. The Mauritian skink, Leiolopisma telfairi also appears to have reduced ancestral anti-predator mechanisms in response to strong intra-specific competition, for like giant tortoises it reaches very high densities, the population of the 1 km long Round island having been estimated at 100 000. Lizards primitively often evade predators by shedding the tail, which acts as an effective distraction for a pursuer as it often continues to move and contains a substantial proportion of the owner's valuable fat reserves. On small islands where food is frequently sparse, some lizards pull off 317 and eat the tails of conspecifics thereby harvesting a large part of their energy reserves. The evolutionary response to this is sometimes to decrease the fragility of the tail (Arnold 1984), so it is harder to steal, something that has happened in several island populations of Mediterranean wall lizards (Perez-Mellado et al. 1998). The same trend is apparent in L. telfairi where, although the tail is breakable, it is tough and cannot be removed easily. The lizards will also often turn and defend the tail by biting if caught by it. Moreover the tail is slender and contains relatively little fat, reserves now being stored predominantly in the body. Other Mascarene taxa have evolved very distinctive features apparently within the archipelago, but the case for their being adaptations to island conditions is not always broadly supported. As their name suggests, Slit-eared skinks (Gongylomorphus) have modified ear openings. These are dorsoventrally narrowed, protected by a projecting chevaux-de-frise of spines, and are closable. A strange muscle system on the snout appears to be capable of occluding the nasal passages and there are dense batteries of backwardly directed sense organs above the ear openings, along the flanks and on the thighs. These features seem almost paranoid, as if in evolutionary terms the lizards are concerned about something approaching them from behind and penetrating their body openings. An insect parasitoid that deposits eggs or carnivorous larvae in such places could be postulated, yet no such predator has been observed to date. So a case for advantage in the island situation is lacking, and as these features of Gongylomorphus are more or less unique, it is not possible to use independent parallel cases on other islands to infer insular adaptation. Change in body size This is often postulated to occur after reptiles arrive on islands, especially marked increase in size leading to insular gigantism, and phylogenies permit such cases to be tested. They show that the lacertid lizards, Gallotia, have indeed become the largest members of their family since their colonization of the central and western Canary islands and adoption of a vegetarian diet. The extinct Mauritian giant skink (Leiolopisma mauritiana) also reached its large size in situ (Austin & Arnold, in progress) and the same is true of the Cape Verde giant skink (Macroscincus coctei) and gecko (Tarentola gigas) (Carranza & Arnold, in progress). However, size increase is by no means universal in island reptiles. Like at least some giant tortoises, the big Phelsuma of the Mascarenes may be descended from quite large colonizers (Austin & Arnold, in progress). Size has also sometimes fallen. In the Mascarenes, this appears true of one of the Rodrigues species of giant tortoise, Cylindraspis peltastes, of the night gecko, Nactus coindemerensis, and of the ancestor of a clade of six small species of Phelsuma in Mauritius and Réunion (Austin & Arnold, in progress). So, body size of island reptiles often changes, but not in a consistent way. Shift in niche Some island forms occur in ecological situations markedly different from those occupied by their mainland relatives, presumably because in the sparse communities often found on islands such situations are otherwise unoccupied and 318 can be exploited. For instance, there may be a shift in time of activity: Phelsuma is ancestrally diurnal, but P. guentheri of Mauritius and P. gigas of Rodrigues have become substantially night-active (Arnold, in press). The shifts in body size mentioned above may also reflect exploitation of new resources. In some cases there may be a general broadening of niche. This is reflected in several niche parameters of the Mauritian Leiolopisma telfairi. This species can be active both by day and night, its diet includes both smaller animals and vegetation, and its spatial niche involves foraging on rocks and in vegetation as well as hunting on the ground, furthermore young animals occupy different habitats from adults. So a range of niche space is covered that might support a number of species in mainland environments. Adaptive radiation This phenomenon is often associated with islands and classical examples include the Galapagos finches, and the honey creepers (Drepanidinae) and fruit flies (Drosophila) of Hawaii. Yet adaptive radiation is not extensive in some archipelagos and insular taxonomic groups. It is inevitably associated with substantial speciation, so its extent in an archipelago can be partly assessed from the ratio of the number of species known to be present in a taxon to the minimum number of initial colonists. In the case of Mascarene reptiles (33 species and eight colonizations), the ratio is 4.1:1 (compared with about 2.6:1 and 3.3:1 in the Canaries and Cape Verdes). In Mascarene reptiles, the number of speciation events known to have occurred in particular groups varies from none, to nine in Phelsuma. Presumably a high general level of adaptive radiation in islands results partly from a low level of colonization compared with the diversity of the available ecological resources, but variation between individual taxa probably reflects such group specific features as the facility with which speciation occurs. Relative rates of evolution Phylogenies can of course often be used to recognize the mainland sister groups of island forms. As sister groups arise simultaneously and are thus the same age, it is possible to compare how much evolution has occurred in them over the same time span. In the case of oceanic islands where the island lineage enters a new habitat and the mainland one remains in more or less the same environment, it is to be expected that, initially at least, the island lineage will change faster (although in periods of global climatic change both mainlands and islands may alter radically). Certainly, greater morphological and behavioural change often does occur in oceanic island forms compared with their mainland sisters. Galapagos finches have radiated and altered far more than the South American grassquit, which probably shares an exclusive common ancestor with them (Steadman 1982). The same is true of Hawaiian honey creepers compared with their probable sister taxon, a Holarctic grosbeak. Island giant tortoises have also changed more than their likely mainland relatives (Arnold 1979). Greater change on islands than mainlands is not always the case. As already mentioned, Semaphore geckos on Sogotra have maintained similar morphologies perhaps since the mid-Miocene, but the mainland sister group of one of them has 319 produced a radiation of a dozen species including forms that have altered radically, even converging on desert iguanian lizards and being active at higher temperatures than any other geckos (Arnold 1993). Similarly the worm lizard, Pachycalamus, of Sogotra remains monotypic and relatively primitive while its mainland sister clade has speciated and produced different morphologies. Possibly less change in general occurs on continental islands compared with oceanic ones. They are likely to separate from mainlands with relatively large populations of at least some of the species they share with these, so founder effects may be improbable. They are also likely to begin with an established community in place, so there will be less opportunity for species to shift into empty niche space with consequent evolutionary change. Such island communities will also be protected to some extent from incoming taxa likely to affect related mainland ones and they may also sometimes be less subject to climatic change. Not all taxa on a continental island necessarily date back to its separation, for it too is subject to transmarine migration. Forms arriving by this means may be distinguishable by molecular clock data and late appearance in any fossil record the island may have. Do island communities turn over regularly? Species in island populations may change steadily over time, some becoming extinct and being replaced by new colonizers (MacArthur & Wilson 1967). This is certainly true for small islands and low ones subject to at least intermittent habitat loss, or even inundation like Aldabra. With larger, higher, more distant islands many taxa seem to persist over long periods, certainly long enough to undergo repeated speciation as Phelsuma has done in the Mascarenes, Gallotia in the Canaries and Tarentola in the Canary and the Cape Verde islands. Incorporating molecular clocks into phylogenies suggests Gallotia may have been in the Canaries for more than 12 My and Tarentola in the Cape Verdes for more than 4 My (Carranza & Arnold, submitted). Any non-anthropogenic turnover involving such groups must consequently be extremely slow. Why are island populations so vulnerable? Island taxa are very vulnerable to many introduced predators. Herbivorous ones destroy vegetation and thus habitats. People were among the main factors in reducing the Mascarene tortoises by direct exploitation (Bour & North-Coombes 1988) and domestic introductions such as pigs and dogs contributed as well. Inadvertently introduced animals were also important, especially those often associated with people such as rats, but also more exotic introductions like the Indian snake, Lycodon aulicus capucinus which may well have exterminated Slit- eared skinks on Réunion and most of Mauritius in the 1800s (Jones 1993). Exotic competitors may also be important in loss of native forms. In the Mascarenes, endemic night geckos, Nactus, have probably been exterminated on some islands by the introduced Hemidactylus frenatus which has about the same range of body size (Arnold & Jones 1995). Nactus are now known to occur on just five off-shore islands while eight islands, often very close to these, are occupied by introduced Hemidactylus geckos. The probability that the allopatric dove-tailed distributions of these two genera is due to chance is <0.001 (exact VE WES). Ios 320 pattern cannot simply be aresult of mutual competitive exclusion by the two genera as they colonized the islands, because Hemidactylus appears to be a very recent anthropogenic introduction while Nactus has been in the Mascarenes long enough to produce a radiation of about six species. Hemidactylus appears only to occur on islands that have been inhabited by people and, on some of these, recent fossils show that Nactus was originally present. Nor is anthropogenic habitat destruction likely to be the cause of the disappearance of Nactus as it survives on islets where most vegetation has disappeared. Vulnerability to incoming predators and competitors appears to be partly because island taxa have not been in contact with the enemies concerned, at least in recent evolutionary time and cannot adapt to them because on small islands endemics are often rapidly and completely exterminated before this can happen, as has been directly observed in several native birds on Guam after the intoduction of the snake Bioga irregularis (Jaffe 1994). The problem is exacerbated because, as noted in tortoises and Leiolopisma, anti-predator devices present when taxa reach islands are sometimes dismantled in response to intensified intraspecific competition. Although formidable and relatively large lizards, Leiolopisma have been extirpated from nearly all of their original range apparently by rats, even though superficially similar skinks coexist with these mammals on mainlands. Presumably reduction of the tail shedding mechanism and inappropriate defence responses associated with this are the cause. More general features important in countering predators and competitors, such as speed of movement may also be attenuated in island situations. For instance locomotion in the wall lizard, Podarcis atrata, of the Columbretes islands off northeast Spain is markedly slower than in its congeners on the neighbouring mainland (Van Damme 1998). Are islands black holes? The way organisms regularly reach oceanic islands and adapt to their special conditions is impressive, but does such evolution have long-term prospects? Does the vulnerability to introduced predators and competitors characteristic of island forms prevent them from going on to colonize mainlands where such enemies naturally occur? This is in principle testable by using phylogenies to trace the total subsequent history of lineages that have colonized islands. If evolution on islands produces forms that can survive on mainlands, occasional invasions of these are to be expected and would often be reflected by paraphyly of island taxa with respect to related mainland forms. If, on the other hand, islands are like astronomical black holes, so colonizing lineages can enter but rarely if ever escape, such evidence should be lacking. In such investigations, it is necessary to ensure that failure to reach mainlands 1s not just a function of distance. There is evidence that Phelsuma originated on Madagascar and that different lineages independently reached the Mascarenes, the Seychelles, and even the Andaman and Nicobar islands a straight-line distance of 6000km (Austin & Arnold, in progress). In none of these cases have the island forms moved significantly on to mainlands. This is even so for P. parkeri which occurs on Pemba island just 50km from the African mainland. Similarly P. andamensis is a mere 300 km from the coast of Burma, a short distance compared with that from its source area. 32] Obviously, many examples must be examined to test whether this phenomenon is usual. If it is actually the island interlude that prevents insular groups invading mainlands, some cases where one mainland has been colonized from another are to be expected and indeed actually occur. Morphology indicates that several taxa present in Africa have reached tropical America by transmarine migration probably via the South Equatorial current. Included here are skinks, Mabuya, and gekkonids, Lygodactylus and Hemidactylus (Kluge 1969). Also the Brazilian gecko Briba appears to be closely related to the West African Hemidactylus longicephalus group, while the neotropical Bogertia, Phyllopezus and Thecadactylus have similarities to the African and Madagascan Homopholis assemblage (Russell 1975). Invasion of one archipelago from another 1s also known, as already noted Tarentola geckos have reached the Cape Verde islands from the Canaries, but does not seem very common. Concluding remarks Fossils and phylogenies illuminate the history of island communities, add detail to them and enable a range of hypotheses to be tested. Among other things they help reveal the very wide variety of phenomena that occur, particularly the differences between particular taxa and archipelagos. For instance, while colonizing Anolis in the Greater Antilles have speciated to produce a similar range of morphotypes on different islands (Losos et al. 1998), in the Mascarenes Phelsuma has two large species on Rodrigues, two small ones on Réunion and one large and four small species on Mauritius. Acknowledgements I am especially grateful to my colloborators in current studies of Mascarene and Atiantic island reptiles including Jeremy Austin and Salvador Carranza (Natural History Museum, London) who carried out all the molecular work mentioned here on respectively Mascarene, and Canary and Cape Verde taxa; also Carl Jones (Mauritius Wildlife Foundation) and Roger Bour (Museum National d'Histoire Naturelle, Paris). The specific contributions of all of these are acknowledged further in the text. Clive Moncreiff kindly discussed the variety of possible colonization patterns. References Arnold, E.N. (1976): Fossil reptiles from Aldabra atoll, Indian Ocean. — Bull. Brit. Mus. nat. Hist. (Zool.) 29: 83-116. Arnold, E.N. (1979): Indian Ocean giant tortoises: their systematics and island adaptations. — Phil. Trans. R. Soc. Lond. B 286: 127-145. Arnold E.N. (1980): Recently extinct reptile populations from Mauritius and Réunion, Indian Ocean. - J. Zool. Lond. 191: 33-47. Arnold, E.N. (1984): Evolutionary aspects of tail shedding in lizards and their relatives. — J. Nat. Hist. 18: 127-169. 322 Arnold, E.N. (1990): Why do morphological phylogenies vary in quality? An investigation based on the comparative history of lizard clades. — Proc. R. Soc. B 240: 135-172. Arnold, E.N. (1993): Historical changes in the ecology and behaviour of semaphore geckos (Pristurus, Gekkonidae) and their relatives. — J. Zool. Lond. 229: 353-384. Arnold, E.N. (in press): Methods of inferring unknown aspects of organisms and the behavioral origins of bird flight. New Perspectives on the Origins and early evolution of birds. - New Haven: Yale University Press. Arnold, E.N. & C.G. Jones (1995): The night geckos of the genus Nactus in the Mascarene islands with a description of the distinctive population on Round Island. — Dodo, Jersey Wildlife Preservation Trust 30: 119-131. Beydoun, Z.R. & H.R. Bichan (1970): The geology of Socotra Island, Gulf of Aden. - Q. J. Geol. Soc. 15: 413-446. Bour & North-Coombes, Braithwaite, €:J7R (J.D) Vay lone ere Kennedy (1973): The evolution of an atoll: the depositional and erosional history of Aldabra. — Phil. Trans. R. Soc. B 266: 307-340. Bullock, D.J., E.N. Arnold & Q. Bloxham (1985): A new endemic gecko (Reptilia: Gekkonidae) from Mauritius. — J. Zool. Lond. 206: 591-599. Burleigh, R. & E.N. Arnold (1986): Age and dietary differences of recently extinct Indian Ocean tortoises (Geochelone s.lat.) revealed by carbon isotope analysis. — Proc. R. Soc. Lond. B 227: 137-144. Carranza, S. & E.N. Arnold (submitted): Long-distance colonisation and radiation in gekkonid lizards, Tarentola (Reptilia: Gekkonidae), revealed by mitochondrial DNA sequences. Carranza, S., E.N. Arnold, R.H. Thomas, J A Matcoh Spar Löpez-Jurado (1999): Status of the extinct giant lacertid lizard Gallotia simonyi simonyi (Reptilia: Lacertidae) assessed using mtDNA sequences from musem specimens. — Herpetol. J. 9: 83-86. Censky, E.J., K.Hodge & J. Dudley (1998): Oyer water dıspersasos lizards due to hurricanes. — Nature 395: 556. Coe, M.J.,D. Bourn & I.R. Swingland (1979): The biomass, production and carrying capacity of giant tortoises on Aldabra. — Phil. Trans. R. Soc. Lond. B 286: 163-176. Greer, A.E. (1990): The Biology and Evolution of Australian Lizards. — Sydney: Surrey Beatty & Sons. Jaffe, M. (1994): And no birds sing: the story of an ecological disaster in a tropical paradise. — New York: Simon & Schuster. Joger, U. (1984): Die Radiation der Gattung Tarentola in Makronesien. — Courier Forsch.-Inst. Senckenberg 71: 91-111. Jones, C.G. (1993): The ecology and conservation of Mauritian skinks. — Proc. Roy. Soc. Arts Sci. Mauritius 5: 71-95. Kluge, A.G. (1969): The evolution and geographical origin of the New World 323 Hemidactylus mabouia-brooki complex (Gekkonidae, Sauria). — Misc. Pub. Mus. Zool. Michigan 138: 1-78. Leguat, F. (1708): Voyages et Aventures de Francois Leguat & de ses Compagnons en deux iles désertes des Indes Orientales. — London: David Mortier. sere. TER. Jackman, A. Larson, K. de Queiroz, L. Rodriguez-Schettino (1998): Contingency and determinism in replicated adaptive radiations of island lizards. — Science 279: 2115-2118. Loveridge, A. (1951): A new gecko of the genus Gymnodactylus from Serpent Island. — Proc. Biol. Soc. Wash. 64: 91-92. Pete got Ron & E.O. Wilson (1967): The theory of island biogeography. — Princeton: Princeton University Press. Maddison, W.P. & D.R. Maddison (1996): MacClade 3.06. Sunderland, Massachusetts: Sinauer. Mertens, R. (1963): The geckos of the genus Phelsuma on Mauritius and adjacent islands. — Maurit. Inst. Bull. 5: 299-305. Mertens, R. (1966): Die nichtmadagassischen Arten und Unterarten der Geckonengattung Phelsuma. — Senck. Biol. 47: 85-110. Mertens, R. (1970): Neues über einige Taxa der Geckonengattung Phelsuma. — Senek Biol. 51: 1-13. E77 Mellado, V., €. Corti & P. Lo Cascio (1997): Tail sutotomy and extinction in Mediterranean lizards. A preliminary study of continental and insular populations. — J. Zool. Lond. 243: 533-541. Steadman, D.W. (1982): The origin of Darwin's Finches (Fringillidae, Passeriformes). — Trans. San Diego. Soc. nat. Hist. 19: 279-296. Bor ED, ©€.J.R. Braithwaite, J.-F. Peake &E.N. Arnold (1979): Terrestrial faunas and habitats of Aldabra during the late Pleistocene. — Phil. Trans. R. Soc. Lond. B 286: 47-66. Townsend, C.H. (1936): Two giant tortoises were swept twenty miles by hurricane. — Bull. N. Y. zool. Soc. 39: 119-120. Vaillant, L. (1899): Nouveaux documents historiques sur les tortues terrestres des Mascareignes et des Seychelles. — Bull. Mus. nat. Hist nat. Paris 5: 19-23. Vinson, J. (1974): A new skink of the genus Gongylomorphus from the Macabé forest (Mauritius). — Rev. agric. sucr. Ile Maurice 52: 39-40. Author’s address: E. Nicholas Arnold, Natural History Museum, Cromwell Road, London SW7 5BD, United Kingdom NER i eee phos ff ie yee 325 Bird communities on the Comoro islands Michel Louette Abstract: Migrant birds include seabirds, Palaearctic waders, exceptional species (Dromas ardeola from Arabia and Falco eleonorae from the Mediterranean) and occasional vagrants migrating from Africa to Madagascar. Waterbirds are remarkably scarce. Resident terrestrial birds are classified in relation to their presumed period of arrival: 9 (11?) “old” endemic species (non-Malagasy origin); 4 “new” endemic species (Malagasy origin) and 22 species represented by endemic subspecies; 17 non-endemic species, which probably arrived in recent times or still have gene flow. Ngazidja, highest, largest and most isolated island, is a centre of speciation. Smaller and lower Mwali is its satellite. Mayotte is another speciation centre. Ndzuani is poor in endemics. Using standardized point-transect counts, a general relationship between degree of endemism in the birds and preference for forest - presumably the original vegetation on much of the archipelago - was found, with some aberrations. The adaptation of endemics to degraded habitat is variable among species and among islands. This is also true for the adaptation of the non-endemic species to forest. Key words: birds, Comoros, bird communities, endemism, habitat selection ALDABRA =" ® Ä NGAZIDJA Sy u MW nozuanı MWALI \W) « x = ~ > ( N Nez \ N en WB MAYOTTE Fig.i: Position of Comoro islands in the western Indian Ocean, with the 200 m line to show possible extent of land area after a sea level drop. Present-day land mass in black. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 326 Study area The Comoro archipelago lies midway between Madagascar and Africa (approximately 300 km distant from each - fig.1). The islands have never been linked with a continent and are isolated on all sides by seas more than 3000 m deep. Ngazidja is 1024 km? and rises to 2361 m at Mount Karthala; Mwali is 211 km? and rises to 790 m; Ndzuani is 424 km? and rises to 1595 m; Mayotte is 374 km? and rises to 660 m. The climate is humid and hot at sea level, but temperate at higher altitude on Ngazidja. Cyclones, though rarely devastating, occur frequently. The original vegetation is mostly evergreen forest, but above ca. 1500 m on Mount Karthala, the trees give way to stands of giant heath Philippia comorensis and, in drier parts of northern Ngazidja and southern and eastern Mayotte, to scrub (with such typical plants as Baobab trees Adansonia digitata), dating most probably from pre-human times. Mayotte has quite extensive stands of mangrove. The archipelago is of high biological importance (Stattersfield et al. 1998). Man settled about 1000 years ago and affected the environment and the avifauna. From sea level to ca. 600 m (much higher on Ndzuani), shifting cultivation and the development of villages resulted in a patchwork landscape of degraded forest, scrub and plantations of exotics (with many Coconut trees Cocos nucifera) and some badlands. Even at high altitude human influence is seen, due to cattle grazing and burning of the forest. The degradation is uneven: during the period of our visits, relatively intact forests were present on Ngazidja and Mwali (although not necessarily primary); in contrast, on densely populated Ndzuani only small remnants of forest persist, and on Mayotte a very limited area is under natural vegetation. But secondary forests occur on all peaks of Mayotte, dominated by introduced species with a prevalence of Mango trees Mangifera indica (and Litsea glutinosa) in the humid areas, and the deciduous Albizzia lebbeck in the drier ones (Pascal 1997). Locally, isolated forest plantations of exotic trees occur. In the La Grille peak region of Ngazidja, there is a single stand of pure Eucalyptus sp.. In the centre of Mayotte, there are plantations of Teak Tectona grandis and West Indian Almond Terminalia catappa (amongst others) embedded in secondary forest in the northern lowest part of Bénara, and plantations of Peltophorum pterocarpum in the lowest part of Majimbini (Malagnoux 1991). Commercial forest exploitation occurs only on a small scale. Crops are much in evidence. Typical for the Comoros are trees such as Ylang-ylang Cananga odorata and Clove Syzygium aromaticum (the latter especially on Ndzuani and Mwali); Vanilla Vanillia flagrans is still present, but Sugarcane Saccharum officinarum monoculture was abandoned a century ago. Vines and other weeds abound, with Lantana camara dominating in the drier parts. Methods For the taxonomic study, a arge number of specimens was examined in major natural history museums in at Tervuren, London, Paris, Leiden and New York. Taxa were considered as species or subspecies based on morphology (Louette 1988). Birds were observed by a team during near-annual visits of several weeks per year in 1983-1998. Furthermore, essentially in order to compare bird diversity and the habitat choice of the endemic (forest) species, standardized point-transects were 327 counted (according to the method described by Hustings et al. 1985) by two observers, on Ngazidja at 33 stations, during the years 1985 and 1989; on Mayotte at 9 stations during each of three consecutive years 1992-1994: on Mwali at 3 stations in 1985 and at 2 stations by one observer in 1989 (complete descriptions in Louette et al. 1988b, Stevens et al. 1995, Stevens & Louette 1999); incomplete counts at 6 stations on Ndzuani in 1992 by the author alone are also available. Here, these data are reanalysed in order to compare communities on the different habitats on the islands and to discuss the adaptation of bird species to available habitat. Simply considering an altitudinal gradient would not be appropriate, because the amount of degradation differs among the islands and Ngazidja is unique in having a montane zone above the tree line. Habitat types for the present analysis are listed in the legend of figs.2-9. Number of birds seen per unit time and per habitat type (a sum of stations) is used as a parameter in the ecological profile. Nocturnal birds and migratory water birds are not considered. Status and origin of the birds Faunistics, taxonomy, origin and evolution were studied by Benson (1960) and by Louette (1988, 1992, 1996), which are the main sources for the following discussion. a. Migrants This archipelago is at the crossroads of several bird migration routes, although there are almost no passerines, except rarely Hirundinidae: — Waders from Palaearctic breeding grounds, wintering on the coasts of the southern oceans in general; the most common ones in the Comoros are: Arenaria interpres, Pluvialis squatarola, Charadrius leschenaultii, C. hiaticula, Numenius phaeopus, Actitis hypoleucos, Tringa nebularia, Calidris ferruginea. It is doubtful however if any one of them ever attains numbers of “Ramsar Convention” importance (1% of world population). — The enigmatic Dromas ardeola from the Horn of Africa and Arabia (significant because it is a localized species: Hockey & Aspinall 1996). This is an important wintering area: several tens of individuals can be found together at favourite sites. — The specialized Falco eleonorae from the Mediterranean (important because a localized species). Several individuals (up to 4) were observed together at particular sites. — An array of species represented by a small number of individuals: Ardeola idae, rarely Phoenicopterus ruber, Glareola ocularis, Eurystomus glaucurus, and Coturnix delegorguei which was only recently discovered! All these are on the passage from Madagascar to Africa. No migrants from Asia to Africa have been recorded so far. — Terns Sterna spp. in hundreds, especially S. bengalensis and S. bergii, also Anous stolidus (plus a few Fregata spp. and Sula spp.), from the Indian Ocean region. — Shearwaters Procellariidae spp. and Storm Petrels Hydrobatidae spp. from the Indian Ocean (and possibly even the Pacific Ocean). b. Breeding waterbirds Very few breeding pairs are present: — Phaethon lepturus at favourite localized spots on the coast of the four islands; 328 — Shearwaters Puffinus lherminieri and Gannets Sula dactylatra, only on Mwali; — Grebes Tachybaptus ruficollis on all the freshwater lakes of which each island has one; — Herons Bubulcus ibis, Ardea cinerea, Egretta alba (very few pairs). c. Breeding land birds The Comoro land avifauna is a rather impoverished mixture of Madagascan and African elements with a high proportion of endemic species and subspecies. Endemism is favoured by geography (separate islands, pronounced relief and sufficient size), the originally rich habitats of the islands and (especially) the distances between the islands and also from other regions restricting gene flow. It is debatable if the Comoro avifauna is impoverished or on the contrary fairly complete. None of the surrounding areas are really comparable, because Madagascar is very large and was separated a long time ago from other continents, the Mascarenes are truly oceanic, Pemba is much closer to Africa and colonized to a very large extent (but not completely!) from there. Other archipelagos in the area consist of much smaller or flat islands (such as the granitic Seychelles or the atolls of the Aldabra group) but their bird communities are similar to that of the Comoros, although more impoverished. Affinities and estimated age of arrival are diverse. Three groups, related to their apparent period of arrival, can be recognized (for a complete list see Louette 1988): — A first wave: 9 ancient endemic species of non-Malagasy, probably African, origin present on one or more islands. The recently described Otus moheliensis from Mwali (Lafontaine & Moulaert 1998) and O. capnodes from Ndzuani seem to be good species and different from O. pauliani from Ngazidja, bringing the total number of species in this group to 10 or even 11. — A second wave: 4 modern endemic species and 22 species represented by endemic subspecies, probably separated from the source population in Madagascar more recently. — Non-endemic species that arrived presumably in modern times, or which have regular gene flow with their source area. They are indistinguishable from forms in Madagascar (5), Africa (4) or both areas (8). Some of them arrived unaided by man, but their establishment was made possible by his activities, while others were deliberately introduced. How can the composition of the first and second waves be explained? Louette (1996) gave the following reasoning. The Comoros are roughly equidistant from the present-day African and Madagascar coasts. However, the prevailing winds nowadays are from the direction of Madagascar for much of the year, which would favour colonizers from that direction (although winds may have been different in the distant past). Colonization from Madagascar may have been particularly favoured at certain periods by the presence of ‘stepping-stones’ towards the Comoro archipelago. During episodes of glaciation in the Pleistocene, the sea level was apparently several tens of metres lower than at present. As a result the north-western Madagascar coast would have been greatly extended, as would the Glorieuses, and two supplementary islands would have existed east of the Comoros (see model, based on 200 m sea level lowering, in fig.1). The distance to the African coast, however, would have been very little changed from what is today. These factors may have decreased the chance of a 329 definitive isolation of colonists from Madagascar compared to those from Africa, so that endemic species originating from Madagascar would appear to be younger. A general rise in sea level would have had little effect on the size of the four Comoros, although perhaps most on Mayotte. The peculiarities of the four islands are as follows: Ngazidja, highest, largest and most distant in the archipelago, is a centre of speciation; smaller and lower, Mwali is its satellite. Mayotte is another, smaller speciation centre (somewhat surprisingly, considering its proximity to Madagascar). The central island, Ndzuani, is poor in endemics; it shows affinities with Aldabra, sharing a few species of direct Madagascan origin not found elsewhere on the archipelago. Was island-hopping or independent colonization from the mainland prevalent in this archipelago? Multiple colonizations from one source have been demonstrated in several genera (Louette & Herremans 1985, Louette et al. 1988a). These were separated by large time intervals, resulting in a different faunal composition on each island. Nevertheless, the checkerboard pattern of forms in Nectarinia on the islands proves that colonizations of birds belonging to one genus may have arrived from different directions. The hypothesized histories of birds are based on their characteristics on each island, their interrelationship and the relationship with neighbouring areas; unfortunately this exercise deals only with surviving species because no important fossils are yet known. Adaptation to the environment It is possible that substantial parts of the avifauna have disappeared since the arrival of man. Strict forest specialists come to mind as possible extinctions. It is likely that only those species successful in adapting to a new situation survive on such small islands. Indeed, some endemics now live outside forest habitats on some or on all islands. The capacity of the original birds to survive in the new anthropogenic habitats is analyzed below. A general correlation of degree of endemism in bird taxa with preference for forest (presumably the original vegetation on much of the archipelago) was demonstrated on Ngazidja (Stevens et al. 1995), on Mwali (Louette et al. 1989) and on Mayotte (Stevens & Louette 1999). The present paper also documents habitat selection by recent arrivals, after man-altered habitats became available. This analysis is mainly based on the comparison of the two most extreme and best explored islands: Ngazidja and Mayotte. Ngazidja (= Grande Comoro) This is the largest island, and alone of the islands has large areas of high altitude terrain. It also has most primary forest and the highest level of endemism (Louette 1988). Generally speaking, endemism increases with altitude, so that above 1000 m nearly all birds belong to endemic species or subspecies. In the main forest on Mount Karthala, endemic species reach their highest densities in the undisturbed areas (Stevens et al. 1995) and diversity of endemism is highest in the wettest area (Stevens et al. 1993). A major shift occurs at medium altitude, where there is an edge effect. The density of endemic species increases up to about 1000 m and remains numerous to 1700 m, after which it decreases. Endemic subspecies are found from sea level to 330 the highest peaks with largest numbers at 400-1500 m. Non-endemic species are most numerous below 800 m and uncommon at higher altitudes. In the isolated forest of La Grille (on the smaller volcanic peak of Ngazidja) and in pure Eucalyptus plantation, at 700-900 m, a large proportion of non-endemics compensates for the absence of endemic species. It is well known that species richness decreases with altitude (Brown & Gibson 1983) due to less diverse vegetation and a smaller available surface at higher altitudes. Such a model predicts more species at lower altitudes. Since the data on Ngazidja do not fit this general model it can be supposed that a number of stenotopic endemic species from lower altitudes are extinct, or that the endemic species now found at mid-altitudes preferred originally low-altitude forest (Louette & Stevens 1992). Presumably, as a result of human activities at lower altitudes on Mount Karthala and the consequent disappearance of the lowland forests, some Foudia madagascariensis Acridotheres tristis birds/h OMayotte Z6Grande Comore faMoheli montane’ main forest isolated tree degraded open a T ZA r T forest plantation montane mainforest tsolated tree degraded forest plantation Nectarinia humbloti & N. coquereli Foudia eminentissima EB Mayotte Grande Comore Moheli Mayotte Grande Comore @Moheli 7 T T montane mainforest isolated tree degraded montane mainforest isolated tree degraded forest plantation forest plantation Figs.2-5: Mean number of individual birds counted per hour in 6 habitat types for 4 selected species - On Ngazidja (= Grande Comoro) six habitat types: montane (J) - main forest (A-F) - isolated forest (G) - tree plantations (H) - degraded mosaic (K-L) - open terrain (I) (letters in brackets stand for the restricted habitat types in Stevens et al. 1995) - On Mayotte five habitat types: main forest (all stations except the following) - isolated forest (Choungui) - tree plantation (Bénara low and Majimbini low) - degraded mosaic (Combani and Ochoungui low) - open terrain, including mangrove (Mirereni and Dzoumogné) (see Stevens & Louette 1999) - On Mwali, along the transect from Miringoni towards Chalet St Antoine two habitat types: forest (upper) - degraded mosaic (middle and lower altitudes) (see Louette et al. 1989) 33] endemic species unable to move upwards became extinct or nearly so, such as Dicrurus fuscipennis, which is very localized in lowland forest. The fact that the non- endemics — the latest arrivals on the island — are also restricted in range could be due primarily to altitude, and — despite the supposed disappearance of some endemics — to unsuitable vegetation structure. At 700-900 m on La Grille non-endemics such as Foudia madagascariensis (fig.2) and Acridotheres tristis (fig.3) readily enter isolated forest remnants and the Eucalyptus plantation. Mayotte Mayotte is the least elevated island and consequently all forests are situated lower than the lowest forests on Ngazidja. The forests are secondary and the rainfall gradient is moderate. Endemic species are more common in the forest, except Nectarinia coquerelii (fig.4). Forest is still utilized, but to a lesser extent, by several endemic subspecies, others utilize different habitats, such as Foudia eminentissima (fig.5), a forest bird elsewhere in the archipelago and on Madagascar (Adriaensen et al., in press). Strikingly, the abundant predator Accipiter francesii shows no preference for forest. All non-endemic species are more common in non-forest habitats, but to varying degrees. Thus, Acridotheres tristis (fig.3) readily enters forest here. Often this process is called habitat choice, but in fact on such small islands Columba polleni Alectroenas sganzini 7 ‘ BA : ID May otte y a ea za | |aGrande Comore 2 U + T T T montane mainforest isolated tree degraded open ZMoheli montane mainforest isolated tree degraded open forest plantation forest plantation 2 (a [2 [® c= om Ca Tem m Terpsiphone mutata Corvus albus DMayotte 14 E = o Mayotte @Grande Comore A : ‘| Moheli @ Grande Comore BE 2 fg Moheli + = T T T za montane main forest isolated tree degraded forest plantation montane mainforest isolated tee degraded open forest plantation eB cz ce CZAR Coe [77] Ban) Ca BA EA 73 Figs.6-9: Mean number of individual birds counted per hour in 6 habitat types for 4 selected species; sites and habitat types as in figs.2—5. SS) Oo i) there is simply no choice, and tolerance of the existing habitat is a precondition of survival. What do we know about the situation on the two other islands? On Mwali, endemic diversity is high and several endemic birds adapt better to secondary vegetation than on Ngazidja. On Ndzuani, few forests remain, but some endemics survive, showing even stronger capacity for adaptation to changed conditions, the most remarkable case being that of Alectroenas sganzini. Species groups Old endemic species A few of these are narrowly restricted to high altitude and forest. Others show adaptation: Columba polleni (fig.6) prefers forest (main block rather then isolated remnants) everywhere, but on Mayotte it accepts secondary forest and tree plantations. Turdus bewsheri adapts rather well to degraded habitat. Nectarinia spp. are best adapted, probably because flowers, their food source, became more abundant and diverse after man’s arrival. On Mayotte, the Dicrurus species, although not omnipresent, adapted to non-forest, while the equivalent species on Ngazidja was unable to do so. New Endemic species Alectroenas sganzini (fig.7) evolved on the Comoros from A. madagascariensis parent stock, a stenotypic forest bird. It prefers forest when this habitat is available and on Ngazidja it lives in logged forest (Stevens et al. 1995). It occurs rarely in non- forest habitats on Mayotte and Mwali, the islands where forest is scarce nowadays, but on Ndzuani it was readily observed in degraded habitat, even at sea level. It has adapted to non-forest on Aldabra which it colonized from the Comoros, evidently because no real evergreen forest is present on this atoll. Other new endemics are restricted to forest. On Ngazidja and Mwali, the first arrival, now a separate species, Hypsipetes parvirostris excludes the second arrival H. madagascariensis from the forest at higher altitude (Louette & Herremans 1985). In Nesillas, two separate invasions led to sympatry on Mwali (N. mariae and the endemic race of the Madagascan N. typica, which occupy different niches). N. brevicaudata — possibly the precursor of N. mariae of Mwali — lives at medium and high altitude on Ngazidja. N. typica longicaudata, presumably an independent and more recent colonist, is widespread on Ndzuani. It is surprising that some Nesillas taxa are restricted to higher zones while others occupy the whole island (Louette et al. 1988a). Endemic subspecies Accipiter francesii lives on Mayotte at very high breeding densities in degraded forest; in the non-breeding season it occurs in the drier parts of the island. On Ngazidja it is restricted to, and rather uncommon in, the well forested zone, but on Ndzuani it is generally uncommon and it is absent on Mwali. Streptopelia picturata is most numerous in forest, but it does occur in numbers outside; the trend is reversed on Mwali. From its morphology, the race of Terpsiphone mutata (fig.8) on Mayotte is a recent arrival and is ubiquitous. The older races on the other islands are scarcer, the one on Ngazidja preferring forest. On Mwali the density trend is reversed. Zosterops maderaspatanus is more numerous outside forest on Ngazidja. Foudia eminentissima (fig.5) is widespread on Ngazidja, where it apparently has a specific niche (Louette 1988) while on Mayotte it is curiously restricted to open places (Adriaensen et al., in press). Non-endemic species Circus maillardi is well adapted to forest and, surprisingly, does not occupy open terrain. It occurs on the three western islands, but is absent on Mayotte. Milvus migrans, a commensal raptor in Africa, should seemingly benefit from the man- altered habitat where it used to occur, but recently became extinct (Louette 1993). Numida meleagris is probably restricted by hunting and Agapornis cana possibly by trapping. Streptopelia capicola occurs commonly only in dry habitat, as in Africa, but Turtur tympanistria avoids high altitudes and evergreen forest and is rare in the Comoros, although there is no apparent competitor. This is a forest species in Africa, living in high rainfall areas and going readily up to 2500 m in Kenya (Zimmerman et al. 1996). Corvus albus (fig.9) is rather uncommon on the four islands, where it has been present since immemorial times, so it is difficult to understand why it should be so rare on Mayotte. Likewise, Lonchura cucullata is a self-introduced species, restricted to grasslands and and localized to gardens. It may be asked why Foudia madagascariensis (fig.2) is not more common. On Réunion, Barré (1983) notes it represents 35% of the bird population in “savane boisée” and is less common but present in all habitats. On that island it is replaced by Passer domesticus in urban environments, which it occupies in the Seychelles where the latter species is lacking. In Mauritius, F. madagascariensis is common, nesting even at greater altitude but at lower densities than in the lowlands and is not really a competitor of the endemic Fody (Foudia eminentissima) (Cheke 1987). Passer domesticus is almost limited to coastal areas with western-style buildings. It used to be widespread in Mwali and became extinct in Ndzuani, which is the most degraded island! This is a surprisingly ‘poor performance’ for such an opportunistic generalist species that has conquered all continents and many islands. Acridotheres tristis (fig.3), introduced in the Comoros at the beginning of the century, is the most successful non-endemic species. Nowadays it is the commonest bird, alongside species of Hypsipetes, Nectarinia and Zosterops. It seems also the only one that has occupied a new niche (forest). It is clear that the recently arrived species have only adapted to a variable and limited degree to the Comoro habitats. Conservation Tropical forests on islands are among the most threatened habitats in the world (Vickery 1998). The Comoros are still relatively intact, but deforestation and introduction of exotics have taken their toll. The impact of rats is very important and, when these are present, longevity becomes a more important factor than breeding success for birds. Relative numbers of specialist forest species are often reduced in disturbed forest (Bennun & Fanshawe 1997) but this is not very apparent in the Comores. It is the case though on Ndzuani, the most disturbed island, which nevertheless still has some endemics (although the populations may not be viable over long periods). Mwali is the most intact island: there are low numbers of 334 Acridotheres tristis or Foudia madagascariensis in the forest and all endemics are in the “correct” forest to degraded forest habitat, except the well adapted Terpsiphone mutata (and also Foudia eminentissima, which is unaccountably scarce). The lowlands of Ngazidja are badly degraded, but the forest at higher altitudes on Mount Karthala was relatively intact at the time of our study; the isolated forest on La Grille, however, is doomed. Mayotte is surprising: the forest has been settled by Acridotheres tristis, but why are other invasive species such as Foudia madagascariensis and Corvus albus so scarce? The isolated forest is less acceptable to Columba polleni and avoided by Dicrurus waldenii. In general it is clear that tree plantations are a very poor replacement habitat for forest, but the spontaneous “Mango” forests of Mayotte at least retain some endemic elements and enable the survival of part of the original avifauna, already adapted to a less luxuriant environment. Rapid destruction of rain forests on the islands of Mwali and Ngazidja would probably eliminate the less flexible endemic bird taxa. As to the migrants, the “overwintering” Dromas ardeola population deserves to be monitored; specific research is needed on its ecology and its mortality should be monitored. Acknowledgements I thank the curators of museums for access to their facilities. J. Stevens, M. Herremans, F. Néri, L. Bijnens and innumerable other companions were helpful with the data gathering and with the discussion of the results. A. Reygel and W. Tavernier gave technical assistance. References Adriaensen, E.,:-M. Louette, J. Stevens, W. Blompensr22362 Verheyen (2000, in press): Distribution of the Forest Fody Foudia eminentissima algondae on the island of Mayotte. — Ostrich. Barre, N. (1983): Distribution et abondance des oiseaux terrestres de l’1le de la Reunion (Océan Indien). — Rev. Ecol. (Terre Vie) 37: 37-85. Bennun, L., & J. Fanshawe (1997): Using forest birds to evaluate forest management: an East African perspective. - www.uk.earthwatch.org/africa. Benson, C.W. (1960): The birds of the Comoro islands: results of the British Ornithologists’ Union Centenary expedition. — Ibis 103b: 5-106. Brown, J.H., & A.C. Gibson (1983): Biogeography. — C.V. Mosby, Saint Louis. Cheke, A.S. (1987): The ecology of the smaller land birds of Mauritius. — pp 151- 207 in: Studies of Mascarene land birds. (ed. Diamond, A.W.). — Cambridge University Press, Cambridge. Hockey, Ph., & S.J. Aspinall (1996): The Crab Plover, enigmatic wader of the desert coasts. — Africa Birds & Birding 1: 60-67. Hustings, M.F.H., R.G.M. Kwak, P.G.M. Opdam &M27 SEN. Reynen (1985): Vogelinventarisatie. Achtergronden, richtlijnen en verslaggeving. - PUDOC, Wageningen, Netherlands. 318,8) Lafontaine, R.-M., & N. Moulaert (1998): Une nouvelle espéce de petit- duc (Otus, Aves) aux Comores: taxonomie et statut de conservation. — J. Afr. Zool. P12 163-169. Bouette, M. (1988): Les Oiseaux des Comores. — Ann. Mus. r. afr. Centrale 0019255: 1-192. Louette, M. (1992): The origin and evolution of Comoro land birds. — Proc. VII Pan-Afr. Orn. Congr. (Nairobi): 207-215. Louette, M. (1993): Decrease of the Black Kite Milvus migrans and other commensal birds in the Comoros. — Scopus 17: 14-18. Louette, M. (1996): Biogéographie, origine et Evolution des oiseaux aux Comores. — pp. 337-348 in: Biogéographie de Madagascar. (ed. Lourenco, W.R.). — ORSTOM, Paris. mowectuen N... & M. Herremans (1985): Taxonomy and evolution in the bulbuls (Hypsipetes) on the Comoro Islands. — Proc. Intern. Symp. African Vertebr., Bonn: 407-423. Beeren Mo Herremans, E. Bijnens & LE. Janssens (1988a): Taxonomy and evolution in the brush warblers Nesillas on the Comoro Islands. — Tauraco 1: 110-129. Bee Vi. I Stevens, L. Bijnens & L. Janssens (1988b): A survey of the endemic avifauna of the Comoro Islands. Study Report, 25. — ICBP, Cambridge. momemecr Me. TE, Bijmens, J. Stevens & L. Janssens (1989): Comparison of forest bird communities on Ngazidja and Mwali (Comoro Islands). — Ostrich Suppl. 14: 33-37. monomer Vin. 6c J. Stevens (1992): Conserving the endemic birds on the Comoro islands. I: general considerations on chances for survival. — Bird Conserv. Int. 2: 61-80. Malagnoux, M. (1991): Mission d’appui au secteur forestier. — Report to Service des Eaux et Foréts, Mayotte. Pascal, ©. (1997): La végétation naturelle a Mayotte. — Direction de I’ Agriculture et de la Forét (Mayotte). Based, ALJ, Mo). Crosby, A.J. Longe & D.C. Wege (1998): Endemic bird areas of the world. Birdlife Conservation series N°7. — Birdlife, Cambridge. Besen, Mo Lowette, L. Bijnens & M. Herremans (1993): Bird diversity on Ngazidja, Comoro Islands. — Proc. VIII Pan-Afr. Orn. Congr. (Bujumbura),: 179-186. Pause i. Ms Louette, L. Bijnens & M. Herremans (1995): Conserving the endemic birds on the Comoro Islands, III: bird diversity and habitat selection on Ngazidja. — Bird Conserv. Int. 5: 463-480. Stevens, J., & M. Louette (1999): Land bird abundance and the conservation of endemism on the island of Mayotte (Indian Ocean). — Alauda 67: 123-139. Vickery, J. (1998): Report on the BOU’s Tropical Forests and Islands Conference. — www.bou.org.uk/meetrep. 336 Zimmerman, D.A.,D.A. Turner & D.J. Pearson (1996): Birds of Kenya and northern Tanzania. — C. Helm, London. Author’s address: Dr Michel Louette, Royal Museum for Central Africa, B 3080 Tervuren, Belgium 3371 The reptile fauna of the Sogotra archipelago Ulrich Joger Abstract: During a four-week faunal survey of Sogotra and its smaller satellites Semha and Abd El Kuri, as part of the UNDP/GEF project “Conservation and Sustainable Use of the Biodiversity of Sogotra Archipelago’, most of the resident reptile species were collected, including at least 4 undescribed taxa. The vertebrate fauna of Sogotra is dominated by lizards, especially geckos, this family making up two-thirds of the approximately 30 reptile species, 27 of them being endemic. On Semha and Abd El Kuri, local endemic gecko species occur. Three lizard genera and one snake genus are endemic to the archipelago. The predominant biogeographic affiliation of the Sogotran reptile fauna is Afrotropical. Key words: Sogotra archipelago, herpetofauna, endemism Introduction The Sogotran archipelago in the northwestern Indian Ocean, politically part of the Republic of Yemen, consists of three inhabited and a number of uninhabited islands. Geologically, they comprise the peaks of a submarine ridge in continuation of the northern Somalian highlands. However sea depth between Cap Guardafui on the Horn of Africa and Abd EI Kuri, the nearest island of the archipelago about 90 km to the east, is between 200 and 1000 m, thus beyond Pleistocene sea level changes. If the separation of these islands was associated with the tectonic processes that lead to the formation of the Gulf of Aden and the Red Sea, it could be dated, with some reservation, to the first half of the Tertiary (older than 30 million years b.p.; see Girdler 1984). Therefore there should have been both a basic continental (African) faunal stock as well as plenty of time for the considerable evolution of a specific island fauna. The individual islands are geologically similar (mainly limestone plateaus with coastal strips of limited extension) but different in size, climate and vegetation. Sogotra, the major island, has an area of approximately 3600 km’ (3625 km’ according to Wranik 1998a), mountain peaks (the central Haghier range reaches 1500 m and frequent fogs are present), abundant water and a rich savannah-like vegetation dominated by trees or shrubs of the genera Croton, Boswellia, Adenium, Dendrosicyos, Euphorbia and Dracaena (fig.1). The total flora of the archipelago amounts to at least 850 species, of which many are endemic (Miller & Bazara’a 1998). Abd El Kuri (162 km?) and Semha (45 km’) also reach heights of up to 800 m, but are considerably drier, with no permanent streams and a succulent flora largely devoid of trees except a few Euphorbia (fig.2). Traditional agriculture is based on livestock (Sogotra and Semha) and date plantations (mainly Sogotra). Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 Fig.l: Slope with Dracaena trees to the north of Kilisan, central eastern, Soqgothar Habitat of Haemodracon riebecki and Hemidactylus sp. (description in preparation). Until recently, the faunal exploration of the archipelago has scratched only the surface of the total fauna. After a number of expeditions at the end of the 19" and the beginning of the 20" centuries (Günther 1881, Peters 1882, Dixey et al. 1898, Forbes 1903, Steindachner 1903), the islands were tenored by )Zoolosical researchers. This was mainly due to geographical as well as political isolation. Only limited field work was possible during the time of southern Yemeni independence (Wranik et al. 1986). An overview of animal taxa known from Soqotra prior to our expedition was given by Wranik (1998a). This report is based on a four-week stay on Sogotra, Semha and Abd EI Kuri during the whole of February 1999, as participants of the UNDP/GEF project ‘Conservation and Sustainable Use of the Biodiversity of Soqotra Archipelago’, which aimed at compiling an inventory of fauna and flora of the archipelago in order to design a master plan for conservation (Joger et al. 1999). 339 Fig.2: Euphorbia tree in the interior of Semha. Results and Discussion Endemism and biogeographic relations The vertebrate fauna of the Soqotran archipelago is dominated by reptiles and birds. All freshwater fishes found so far belong to introduced species or represent secondary invaders from the marine environment. Amphibians do not exist. This paucity of freshwater vertebrates may be attributed to a combination of varying aridity in the geological past and the ubiquitous presence of voracious freshwater crabs (Potamidae). Most wild mammal species have been introduced, such as mice (Mus musculus), rats (Rattus rattus), and civet cats (Viverricula malaccenis). However, the speculation of Wranik (1998a) that the Sogotran fauna originated “before mammals appeared on earth” must be strongly rejected. This would imply a Separation in Triassic times (approximately 200 million years b.p.) — which is not in accordance with tectonic evidence. Possible candidates for endemic mammal taxa are bats (Rhinopoma sp.) and a shrew (Suncus sp.) which we found on Sogotra. The specimens are under study. The avifauna of the Sogotra archipelago is comparatively well known and comprises 31 breeding species, of which 6 species and 10 or 11 subspecies are considered endemic (Al Sagheir & Porter 1998). This rather low number of endemics could have been expected in animals capable of flying, and is found to be Similar in dragonflies (Schneider & Dumont 1998). In contrast, among approximately 30 reptile species present, only 3 geckos of the genus Hemidactylus 340 Sogotran Reptile Genera Fig.3: Genus level endemism in Sogotran reptiles. ED] Sogotran endemic Ei regional endemic KO non-endemic (10%) also occur outside Sogotra and are likely to have been introduced by man. At least two of them (Hemidactylus turcicus and H. flaviviridis) frequently occur in and around human habitations. Thus the indigenous reptile fauna can be considered 100% endemic. Even one third of genera are endemic (fig.3), namely the gekkonid Haemodracon Bauer, Good & Branch, 1997, the scincid Parachalcides Boulenger, 1899, the trogonophid amphisbaenian Pachycalamus Günther, 1881 (fig.4), and the colubrid Ditypophis Günther, 1881 (fig.5). A regional or subendemic genus is the diurnal gekkonid Pristurus, which occurs in the countries bordering the Red Sea and the Gulf of Aden. Its sister genus appears to be Quedenfeldtia from southwestern Morocco, a relict area where a pre-Saharan fauna, including Trogonophis, arelative of Pachycalamus, occurs'. Haemodracon may be part of Fig.4: Pachycalamus brevis, the endemic So@qotran amphisbaenid. ' The remaining genera of the Trogonophidae are Diplometopon from eastern Arabia and Agamodon from Yemen and Somalia. 34] Fig.5: Ditypophis vivax, one of the two endemic Sogotran colubrids. an Afro-Malgasian group of leaf-toed geckos (Bauer et al. 1997). Ditypophis was considered related to the Afro-Malgasian genera Pythonodipsas, Geodipsas or Dipsadoboa (= Thamnodynastes) (Giinther 1881, Bogert 1940, Parker 1949). Parachalcides resembles the South African genus Sepsina (Boulenger 1903). Although the real evolutionary relationships of these genera will be analyzed using molecular techniques, it can be stated that the fauna is of predominantly Afro- Malgasian or Afro-Arabian (pre-Saharan) origin, and the disjunct distribution of groups and the high diversification point to an ancient separation of these taxa (Joger 1987). Among endemic species of non-endemic genera, the Soqotran chameleon (Chamaeleo monachus Gray, 1864) clearly belongs to the mainland African/west Asian genus and subgenus Chamaeleo s.str., not to the Malgasian radiation (Klaver & Bohme 1986). Its type locality “Madagascar” must therefore be erroneous, and Wranik’s (1998a) consideration that it might have been imported by man is also contradicted by the fact that it does not occur elsewhere. The endemic nocturnal geckos of the genus Hemidactylus are rather diverse, but most of them resemble one or the other Somalian species. Somalia can be considered an area of primary radiation of the genus (Joger 1985). It should be noted that Abd El Kuri, the island closest to Somalia, harbours two Hemidactylus species of its own (H. forbesii Boulenger, 1899 and H. oxyrhinus Boulenger, 1899). The scincid Mabuya socotrana (Peters, 1881) also appears related to African congeners. The lacertid lizard Mesalina balfouri (Blanford, 1881) is similar to M. guttulata, a Saharo-Sindian species (Boulenger 1903). The colubrid snake Coluber socotrae (Günther, 1881) is probably allied to the Afro-Arabian C. florulentus group (Parker 1949). Little can be said about the affiliations of the three blind snakes of Sogotra: Leptotyphlops filiformis (Boulenger, 1899), L. macrura (Boulenger, 1899), and Typhlops socotranus Boulenger, 1889. The former is similar to the Saharo-Arabian L. macrorhynchus, the latter has been considered a relative of 7. cuneirostris from Somalia (Parker 1949). 342 Missing families and genera; erroneous records Land tortoises were mentioned in a Greek report from the 1" century A.D. on “Dioscurida” - an island later identified as Sogotra (Schoff 1912, from Schneider 1996). Today only sea turtles nest on the archipelago. Taking into account that large tortoises used to live on many Indian Ocean islands such as the Seychelles, Aldabra, Madagascar and the Mascarenes, where they were eradicated by sailors, settlers and introduced predators (Arnold 1981), there is some likelihood for this (although no tortoise bones or shells have been found so far on Sogotra). As the same report also mentions large lizards, Wranik (1998a, b) speculates about the occurrence of a now extinct Varanus. However this is unlikely. Contrary to land tortoises, varanids have never been reported from other western Indian Ocean islands. Instead, large scincids and gekkonids are known from some of them (Arnold 2000, this volume). The explanation for their absence is that varanids originated in southeast Asia or Australia (where they still have their greatest diversity) and invaded Africa and Arabia only during the Miocene (18-20 million years b.p.), when the Tethys Sea had been interrupted by tectonic upheavals associated with the collision of the Afro-Arabian plate with continental Asia. The few Varanus species living in Africa and Arabia today have been found to be closely related to each other (Böhme et al. 1989) and can therefore be regarded as descendants of a rather recent invader - after Sogotra had been separated from the African continent. A similar biogeographical history can be assumed for the Agamidae — a family present everywhere in Australia, Asia and Africa, except on Madagascar and the islands of the western Indian Ocean. On Madagascar their ecological niches are occupied by the otherwise New World iguanids. A fossil iguanid has been reported from Aldabra (Arnold 1981). Joger (1986, 1991) reconstructed the phylogeny of the two agamid subfamilies, Agaminae and Uromastycinae, and found that in both, Arabian and African radiations are much more recent than Oriental ones. They also reached East Africa after the separation of Soqotra. A record of Uromastyx ocellatus from Sogotra (Peters 1882) can be regarded as a locality error, since a chameleon from mainland Yemen (Chamaeleo calyptratus) was recorded from Soqotra in the same article. Among the Gekkonidae, two names have been deleted from the faunal list of Sogotra. Ptyodactylus sokotranus Steindachner, 1902, described from Sogotra, was later found to be a synonym of Phyllodactylus (= Haemodracon) riebecki (Eiselt 1962), a fact that was overlooked by Wermuth (1965). The Soqotran population of Pristurus rupestris, a mainland gecko recorded from Sogotra by Peters (1882) and Boulenger (1903), was later described as P. sokotranus Parker, 1938. The saw-scaled viper Echis coloratus was recorded from Sogotra by Balfour (Günther 1881) and controversially discussed by Parker (1949), Corkill & Cochrane (1965), Joger (1984) and Wranik (1998a, under the wrong name Echinus coloratus). During our survey, large black snakes resembling a cobra (Naja sp.) were observed at higher altitudes on Sogotra and Semha, but not caught. Although Echis and Naja are of African origin (Joger 1987) and thus possible original inhabitants of Soqotra, there are no records of venomous snakes or fatal snake bites among the human population. 343 Echis coloratus, like Chamaeleo calyptratus and Uromastyx ocellatus, occurs in Yemen and its Soqotran record can be attributed to a locality error, while the black snakes are most likely old melanistic individuals of Coluber socotrae. Comparison between individual islands Table | and fig.6 show that the herpetofauna of the smaller islands is much poorer than that of Sogotra. This is due to the well-known influence of both island size and habitat diversity on species numbers. Species of high altitude and a humid climate, such as Parachalcides and certain Hemidactylus, as well as arboreal Pristurus, are unlikely to occur on the arid satellite islands. However it should be noted that the fauna of the lesser islands has not been sufficiently explored. During a two-day Table 1: Terrestrial reptile species known from the three main islands of the Soqotran archipelago. Non-endemic (probably allochthonous) species are denoted by an asterisk (*). Sogotra Semha Abd el Kuri Gekkonidae Haemodracon riebecki (Peters) Haemodracon sp. (Joger, in prep.) H. trachyrhinus (Boulenger) Hemidactylus flaviviridis Rüppell* Hemidactylus forbesii Boulenger H. granti Boulenger H. homoeolepis Blanford Hemidactylus. cf. homoeolepis H. oxyrhinus Boulenger H. pumilio Boulenger HA. turcicus Linnaeus* H. sp. (Joger, in prep.) Pristurus abdelkuri Arnold* Pristurus abdelkuri Arnold P. guichardi Arnold Pristurus sp. (Rösler & Wranik, in press) P. insignis Blanford P. insignoides Arnold P. sokotranus Parker P. sp. (Rösler & Wranik, in press) Chamaeleonidae Chamaeleo monachus Gray Lacertidae Mesalina balfouri (Blanford) Mesalina cf. balfouri Mesalina cf. balfouri Scincidae Mabuya socotrana (Peters) Mabuya cf. socotrana Mabuya cf. socotrana Parachalcides socotranus Boulenger Trogonophidae Pachycalamus brevis Giinther Leptotyphlopidae Leptotyphlops filiformis (Boulenger) L. macrura (Boulenger) Typhlopidae Typhlops socotranus Boulenger Colubridae Coluber socotrae (Günther) Coluber cf. socotrae Ditypophis vivax Ginther 344 @ Colubridae EB Typhlops EM Leptotyphlops EZ Trogonophidae Chamaeleonidae Lacertidae D Scincidae OD] Haemodracon D Pristurus DU Hemidactylus Sogotra Abd al Kuri 3549 km? 2 162 km? Fig. 6: Comparison of the reptile fauna ot the three main ıslands of the Sogotran archipelago stay on Semha, we discovered a new Haemodracon (fig. 7) and a new Pristurus (fig. 8), and further taxa may have escaped our attention. For instance, the wormlike underground reptiles Pachycalamus, Typhlops and Leptotyphlops cannot be excluded. Other taxa, such as the Mesalina and Coluber of Semha, appear slightly different from their Sogotran congeners. Molecular investigations have been initiated to find out whether speciation has aleady occurred in these genera. The reptile fauna of Abd El Kuri, the westernmost and most arid island, appears even more reduced in number of species, although its size 1s much larger than Semha. Nevertheless it contains three locally endemic gecko species, indicating an independent evolution as a result of long geographical isolation. Sea depth between Abd El Kuri and the other islands is more than 200 m, making any Pleistocene land connection unlikely (whereas Semha could have been connected to Sogotra via the smaller islet Darsa during times of lowest sea level). Fig.7: Haemodracon sp., a new species (Joger, submit- ted) of leaf-toed gecko from Semha. 85) IS n Figo: Pristwrus sp. trom Semha, described as a new species (Rösler & Wranik, submitted) So far, no snake species have been collected on Abd El Kuri. However, we could only explore a small part of that island, and its faunal list may be not complete. Radiations within the archipelago Two-thirds of the reptile species of the Sogotra archipelago are geckos. Of them, two genera have undergone major radiations and multiple speciations: — A diurnal radiation is represented by the genus Pristurus. The definite number of species and subspecies in the archipelago has yet to be determined, but there are at least five endemic taxa on Sogotra and one each on Semha and Abd El Kuri’. Niche separation between them is mainly achieved by different microhabitats: one species is more or less ground dwelling (P. socotranus), two are saxicolous (P. insignis complex), and two are arboreal (P. guichardi complex). — A nocturnal radiation is made up of Hemidactylus species. At least six species (three of them introduced) occur on Sogotra, one on Semha, and two on Abd El Kuri. Two Sogotran species (one of them new) are confined to the highlands. All live on rocks, but sympatric species are distinguished by size difference, so apparently occupying different food niches. A similar situation is found in the likewise nocturnal Haemodracon, a genus of which two species (H. riebecki on Sogotra and H. sp., fig.7, on Semha) exceed sympatric Hemidactylus in size, whereas a third one (H. trachyrhinus on Sogotra) is very small (Rösler & Wranik). Exploring speciation within these species complexes is an important task for future research. * ) Pristurus abdelkuri also occurs as an introduced population in coastal areas of Sogotra. Fig. 9: One of many dead Green turtles (Chelonia mydas) near a fishermen’s landing point on Abd El Kuri. Also habitat of Hemidactylus forbesii. Conservation The only reptiles that are threatened by direct human exploitation are sea turtles, which are caught by fishermen off the coast and probably also on their nesting beaches (fig.9). They are locally consumed and therefore their protection by CITES is not effective on Sogotra. Wranik (1998a) expresses concern that Chamaeleo monachus might be threatened by collecting for pet keepers. This fear is, however, not justified, as chameleon trade is controlled by CITES, and commercial pet traders have more accessible places elsewhere to supply themselves with chameleons. Moreover, the species is widespread and abundant on the island but difficult to find, due to its effective camouflage. There is, however, an increasing pressure on reptile habitats by construction and development projects. Most reptiles occur in abundant numbers, but the large leaf- toed geckos of the genus Haemodracon appear to be less numerous and more vulnerable than other species. They lead a more or less sessile life in Dragon blood trees (Dracaena cinnabari) and Date palms (Phoenix dactylifera), but only old large trees with holes appear suitable. Tree cutting could adversely affect their populations. However they were also seen on vertical rocks with crevices. The population of the new species on Semha appears especially endangered due to its small total population size. Other potentially threatened species are those restricted to special biotopes, such as mangrove forests (Pristurus sp., fig.10) and higher altitude habitats (some Hemidactylus species, Parachalcides socotranus). 347 Fig.10: A mangrove-dwelling Pristurus, described as a new species (Rösler & Wranik, submitted). An additional danger, the devastating effect of which is known from many island faunas, is introduction of foreign species. The Sogotran capital, Hadibo, is already devoid of indigenous Hemidactylus due to the presence of introduced A. flaviviridis and H. turcicus which have apparently outcompeted the original species. Introduced rats (Rattus rattus) are also very abundant at Hadibo. At present there are no dogs on the island, but if there were as many wild dogs as in mainland Yemen they would present a threat to the colubrid snakes. Cats are already abundant on Sogotra. Conclusions The current reptile fauna of the Sogotran archipelago is unique and results from a long evolutionary diversification of an ancient African or Afro-Arabian stock. Future studies should aim at the reconstruction of the phylogenies of these and other groups of Sogotran organisms. A comparative estimation of genetic distances from mainland Arabian, African and Socotran taxa of several genera is in preparation. Such investigations promise to provide insights into the process of speciation, as well as into an understanding of the evolutionary and biogeographic history of East Africa, Arabia and the western Indian Ocean. 348 Acknowledements I would like to thank Tony Miller, Wolfgang Schneider and the Yemeni Environmental Protection Council (EPC) for providing me with the opportunity to do zoological research in the Sogotran archipelago. For help in the field, I am indebted to Masaa Al-Gumaili, Abdul Kareem Nasher, Jamil Salim Mubarak, Hans Pohl and Kay van Damme. References Al Sagheir, ©. &R. E. Porter (1998)2 The bird diversity, ot Sogonz pp.199-212 in: Proceedings of the first international symposium on Sogotra island: Present and future (ed. Dumont, H.J.). — United Nations Publications, New York. Arnold, E.N. (1981): Recently extinct reptile faunas from southwest Indian Ocean. - BMNH Triennial Rep. 1978-80: 52-56. Arnold, E.N. (2000): Using fossils and phylogenies to understand evolution of reptile communities on islands. — pp. 309-323 in: Isolated Vertebrate Communities in the Tropics. Proccedings of the 4th international Symposium of the Research Institute and Zoological Museum Alexander Koenig, Bonn. (ed. Rheinwald, G.) — Bonn. zool. Monogr. 46 Bauer, A., D.A. Good & W.R. Branch (1997): The taxonomy ofsche southern African leaf-toed geckos (Squamata: Gekkonidae), with a review of Old World “Phyllodactylus” and the description of five new genera. — Proc. Calif. Acad. Sci. 49 (14): 447-497. Bogert, C.M. (1940): Herpetological results of the Vernay Angola expedition. Part I, including an arrangement of the African Colubridae. — Bull. Am. Mus. Nat. Häst. 77: 1.107. Böhme, W.,U.Joger & B. Schätti (1989): A new monitor lizard (Reptilia: Varanidae) from Yemen, with notes on ecology, phylogeny and zoogeography. — Fauna Saudi Arabia 10: 433-448. Boulenger, G.A. (1903): Reptiles. — pp.75-105 in: Natural History of Socotra and Abd-el-Kuri (ed. Forbes, H. O.). — Bull. Liverpool Mus. (Spec. Bull.): 75- LOS: Corkill, N.L. & J.A. Cochrane (1966): The snakes of the Arabian Peninsula and Socotra. — J. Bombay Nat. Hist. Soc. 62: 475-506. Dixey, F.A.,M. Burr & ©. Pickard-Cambridge (1898): Onacolllecnon of insects and arachnids made by Mr. E. N. Bennett in Socotra, with descriptions of new species. — Proc. Zool. Soc. London 1893: 372-391. Eiselt, J. (1962): Ergebnisse der zoologischen Nubien-Expedition 1962, Teil II: Amphibien und Reptilien. — Ann. Naturh. Mus. Wien 65: 281-296. Forbes, H.O. (1903): Natural History of Socotra and Abd-el-Kuri. — Bull. Liverpool Mus. (Spec. Bull.): 1-598. Girdler, R.W. (1984): The evolution of the Red Sea and the Gulf of Aden in space and time. — pp. 747-762 in: Marine Science of the North-West Indian Ocean and adjacent waters. Deep Sea Research Part A: Oceanographic research 349 papers 31. Oxford. Günther, A. (1881): Descriptions of the amphisbaenians and ophidians collected by Prof. I. Bayley Balfour. — Proc. Zool. Soc. London 40: 461-463. Joger, U. (1984): The venomous snakes of the Near and Middle East. — Tübinger Atlas des Vord. Orients, Beih. A 12, 115 pp. Joger, U. (1985): The African gekkonine radiation — preliminary phylogenetic results, based on quantitative immunological comparisons of serum albumins. — pp. 479-494 in: Proceedings of the International Symposium on African Vertebrates (ed. Schuchmann, K.-L.). Bonn 1985. Joger, U. (1986): Phylogenetic analysis of Uromastyx lizards, based on albumin immunological distances. — pp. 187-192 in: Studies in Herpetology. Societas Europaea Herpetologica (ed. Rocek, Z.). Prague. Joger, U. (1987): An interpretation of reptile zoogeography in Arabia, with special reference to Arabian herpetofaunal relations with Africa. — pp.257-27] in: Proceedings of the Symposium on the Fauna and Zoogeography of the Middle East, Mainz 1985 (eds. Kinzelbach, R., F. Krupp & W. Schneider): Tiibinger Atlas des Vorderen Orients, Beiheft A 28. Joger, U. (1991): A molecular phylogeny of agamid lizards. — Copeia 1991 (3): 616-622. Toe ot Pohl &@ W. Schneider (1999): Kurzbericht über eine Expedition auf das Sokotra-Archipel. — Informationen aus dem Hessischen Landesmuseum Darmstadt 99/1: 30-33. Ken ea & W. Böhme (1986): Phylogeny and classification of the Chamaeleonidae (Sauria) with special reference to hemipenis morphology. — Bonn. zool. Monogr. 22: 1-64. Miller, A.G. & M. Bazara’a (1998): The conservation status of the flora and fauna of the Sogotran Archipelago. — pp. 15-34 in: Proceedings of the first international symposium on Sogotra island: Present and future (ed. Dumont, H.J.). — United Nations Publications, New York. Parker, H.W. (1949): The snakes of Somaliland and the Sokotra islands. — Zool. Verh., Rijksmus. Natuurl. Hist. Leiden 6: 1-115. Peters, W. (1882): Mittheilung über die von Herrn Dr. E. Riebeck auf Socotra gesammelten Reptilien. — Sitzungsber. Ges. naturforsch. Freunde Berlin 1882: 42-46. Rösler, H. & W. Wranik (1998): Beiträge zur Herpetologie der Republik Jemen. 3. Geckos des südlichen Jemen und der Insel Sokotra (Reptilia: Sauria: Gekkonidae). — Faun. Abh. Staatl. Mus. Tierkde. Dresden 21: 113-132. Schneider, W. (1996): Sokotra - die Insel der Dioskuriden. — Informationen aus dem Hessischen Landesmuseum Darmstadt 96(1): 25-28. Schneider, W. & H.J. Dumont (1998): Checklist of the Dragonflies and Damselflies of Sogotra island (Insecta: Odonata). — pp. 219-232 in: Proceedings of the first international symposium on Sogotra island: Present and future (ed. Dumont, H. J.). — United Nations Publications, New York. Schoff, W.H. (1912): The Periplus of the Erythraean Sea. — New York. 350 Steindachner, F. (1903): Batrachier und Reptilien aus Südarabien und Sokotra, gesammelt während der südarabischen Expedition der kaiserlichen Akademie der Wissenschaften. — Sitzber. Kais. Akad. Wiss. Wien, Math-nat. Kl., 112: 7-14. Wermuth, H. (1965): Liste der rezenten Amphibien und Reptilien. Gekkonidae, Pygopodidae, Xantusiidae. — Das Tierreich, Berlin 80. Wranik, W. (1998a): Faunistic notes on Sogotra island. — 135-198 in: Proceedings of the first international symposium on Sogotra island: Present and future (ed. Dumont, H. J.). — United Nations Publications, New York. Wranik, W. (1998b): Contributions to the herpetology of the Republic of Yemen. 4. Socotra Island and southern Yemen mainland. — Faunist. Abh. staatl. Mus. Tierk. Dresden 21/Suppl.: 163-182. Wranik, W., K.B. Reddy & A.R. Thabet (1986): Socotra island Contributions to the flora and fauna. — Aden University, 150 pp. Author’s address: PD Dr. Ulrich Joger, Hessisches Landesmuseum, Friedensplatz 1, D-64283 Darmstadt, Germany. E-mail: u.joger@hlmd.tu-darmstadt.de Spill Preliminary chromosomal results of Niviventer Marshall, 1976 (Mammalia, Rodentia, Muridae) | from the Dalat plateau in southern Vietnam Me be Baskevich’& G.VeoKuznetsov Abstract: Results of chromosomal studies (routine staining, C-banding) of 33 specimens of Niviventer rats collected in two localities (Cong Troi and Phi Lieng) of Dalat Plateau in Southern Vietnam are presented. The rats were identified on the basis of preliminary cranial and electrophoretic data (Likhnova et al. 1998) as N. tenaster, N. fulvescens and N. langbianis. All specimens have 2n=46 chromosomes. The most distinctive was the karyotype of N. langbianis (1 female) collected in tropical forest at about 1000 m above sea level in Phi Lieng. The karyotypes of N. tenaster (21 specimens) and N. fulvescens (11 specimens) were similar at the level of routine staining of chromosomes, and distinctive at C-banding in heterochromatin quantity and intensity of Y-chromosome’s staining. The possible role of chromosomal peculiarities in forming and supporting the isolating mechanisms among these sympatric but not syntopic species is discussed. Key words: Chromosomes, Rodentia, Muridae, Niviventer rats, Dalat Plateau, Vietnam. Introduction Dalat Plateau is situated in Lambdong Province (southern Vietnam) where a special ecologo-faunistic group of rodents is suspected (Cao Van Sung & Kuznetsov 1992a). The distribution of the Vietnam endemic Rattus osgoodi at least is limited to this area. Therefore it is of interest to gain more information about the composition and distribution of rodents from this region including that of the Dalat Plateau. The data on rodents from Dalat Plateau so far are fragmentary and based only in a few cases on modern genetic approaches (Baskevich & Kuznetsov 1998, Likhnova et al. 1998). In this report we present the results of karyological studies of Niviventer rats - important components of the vertebrate community in the pine and tropical forests of the Dalat Plateau. Material and methods The specimens were collected in December 1995 at two localities of the Dalat Plateau: Cong Troi and Phi Lieng. The 33 specimens were identified on the basis of preliminary cranial and electrophoretic data (Likhnova et al. 1998) as N. tenaster, N. fulvescens and N. langbianis. Preparations of mitotic chromosomes were obtained by means of the general air-drying technique. The C-banding staining procedure was carried out according to Sumner (1972). Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 352 Results The most distinctive was the karyotype of N. langbianis (one female, collected in tropical forest at about 1000 m above sea level in Phi Lieng). Its 2n=46-karyotype consists of one pair of large subtelocentric, 18 pairs of acrocentric, three pairs of small metacentric autosomes and acrocentric X-chromosomes. At C-band-staining small pericentromeric heterochromatin blocks were distinguished in almost all autosomes and X-chromosomes (fig.1). The karyotype of this endemic to Indochina is studied here for the first time. It is noteworthy that on the level of routine staining its karyotype is close to that of N. cremoviventer, described earlier from the Malay Peninsula (Yong 1969a) and from Tai-Nguen Plateau in Vietnam (Chan Van Minh 1988). This may point to close phylogenetic relationships between these two species on the South-East Asian mainland. On the other hand it is known that the karyotype of N. cremoviventer of the Sunda Shelf (Java) has a specific karyotype (2n=46, NF=60) (Duncan et al. 1974). This last fact may demonstrate the effect of long geographical isolation on the chromosomal differentiation of Niviventer rats. - The 2n=46-karyotypes of N. tenaster (21 specimens) and N. fulvescens (11 specimens) are similar on the level of routine staining of chromosomes: the karyotypes are composed of 19 pairs of gradually decreased acrocentric and three small pairs of metacentric autosomes, medium-sized submetacentric X- and small acrocentric Y-chromosomes. At C-band-staining one can observe significant interspecific differences between their chromosomal sets. In N. tenaster, large strongly stained heterochromatine blocks were observed in all autosomes. The medium-sized X-chromosome has an additional heterochromatin in the short arm, the Y-chromosome is fully heterochromatinized, but the intensity of its hetero- Fig.1: C-band-stained karyotype of female Niviventer langbianis from Phi Lieng 333 G2 4h as Ns Fig.2: C-band-stained karyotype of male N. renaster from tropical forest in Cong Troi chromatin staining is less than that of the autosomes (fig.2). Using C-band-staining in the karyotype of N. fulvescens (fig.3), however, only small, mainly slightly stained heterochromatine blocks of pericentromeric localization in all the auto- somes, that of intercalaric localization in a few autosomes, and more intensively stained heterochromatin of heterochromosomes were revealed. e* se ‘@ | ye Fig.3: C-band-stained karyotype of male N. fulvescens from pine forest of Phi Lieng W n TS Discussion When we compare our results we can say that the chromosomal peculiarities, in addition to morphological and electrophoretic ones (Likhnova et al. 1998), allow us to distinguish these three sympatric but not syntopic Niviventer species in Dalat Plateau. They are not syntopic because all studied specimens of N. tenaster were collected in tropical forest at 1600 m a.s.l. in Cong Troi, whereas all specimens of N. fulvescens were collected in pine forest at both localities (Phi Lieng and Cong Troi) of the Dalat Plateau. N. langbianis on the other hand belongs to the arboreal habitus type; the single female of this species was trapped by us in the tropical forest in Phi Lieng, so we can see a clearly expressed biotopic isolation between these three sympatric Niviventer species in Dalat Plateau. N. longbianis and N. tenaster are Indochinese endemics while N. fulvescens has a wider distribution (it is the only member of the genus with a geographic distribution encompassing the SE Asian mainland and some islands of the Sunda Shelf; Musser 1981, Musser & Carleton 1993). It is obvious that the ranges of the three species overlap to some degree. Therefore it is of interest to study their isolation mechanisms on various levels (ecological, chromosomal and others). We believe that chromosomal peculiarities do not play an important role in forming and supporting the isolating mechanisms among these sympatric but not syntopic species because interpopulational chromosomal variability was marked in the widely distributed N. fulvescens (Yong 1969a,b, Duncan et al. 1970, 1974, Markvong et al. 1973, Cao &Chan 1984, Cao & Kuznetsov 1992, our data). The high variability may indicate the taxonomical complexity of N. fulvescens. Table I: Chromosome variability in populations of Niviventer fulvescens Country, locality 2 Autosomes Sex (pairs) chromosomes M S ST A x Ne NF M 54 Vietnam, Con Son Is. / 8 A Reference Duncan et al. 1970 Vietnam, Dalat Plateau 2 [ our data Vietnam & - Cao & Chan 1984 Vietnam - g Cao & Kuznetsov 1992b Thailand / 2 Yosida 1973 Thailand / £ 54 Markvong etal. 1973 Malaja Yong 1969a Java Duncan et al. 1974 Hong Kong Yong 1969b Notes: 2n = diploid number of chromosomes, NF = fundamental number, M = metacentric, SM = submetacentric, ST = subtelocentric, A = acrocentric 333 On the other hand it is known that the heterochromatin quantity can have an effect on the behavior of chromosomes at the meiotic prophase (Borodin et al. 1995). The positive correlation between the heterochromatin quantity and the asinapsis of heterochromosomes at the meiotic prophase may support the idea of possible contributions of chromosomal peculiarities to the forming and support of isolation mechanisms between these sympatric Niviventer species whose karyotypes are characteristic in their different heterochromatin patterns. The results obtained allow us to discuss the phylogenetic relationships between the Niviventer taxa on a new basis. Our data support the division of Niviventer suggested earlier by Musser (1981) and which was based recently on electrophoretic results (Likhnova et al. 1998). Acknowledgements We thank our colleagues Dr. V.V. Suntsov, Dr. N.I. Suntsova, Dr. O.P. Likhnova and A.P. Mikhailin who helped us to provide the Niviventer rats material. This work was supported by Russian Fund of Fundamental Investigations (grant N 98-04- 49210). References Baskevich, M.I. & G.V. Kuznetsov (1998): To the question of cytogenetic differentiation of the Rattus rats (Rodentia, Muridae) from the Dalat Plateau in South Vietnam. — Zool. Zhurn. 77: 1521-1528. Borodin, P.M.,O.V. Sablina & M.I. Rodionova (1995): Pattern of X-Y chromosome pairing in microtine rodents. — Hereditas 123: 17-23. Cao, Van Sung & Van Minh Chan (1984): Karyotypes et systematique des rats (genre Rattus Fischer) du Vietnam. - Mammalia 48: 557-564. Cao, Van Sung & G.V. Kuznetsov (1992a): About ecological and faunistic groups of rodents in the fauna of Vietnam. — pp. 198-204 in: Zoological studies in Vietnam (ed. Sokolov, V.E.). — Nauka, Moscow. Cao, Van Sung & G.V. Kuznetsov (1992b): About the taxonomic position of the rats of the genus Rattus (Muridae, Mammalia) in Vietnam. — pp. 74-81 in: Zoological studies in Vietnam (ed. Sokolov, V.E.). -— Nauka, Moscow. Chan, Van Minh (1988): Fauna and ecological pecularities of rodents from plateau Tai-Nguen, 22pp. — Abstract of Thesis, Moscow. Duncanwiot- FD. van Peenen & P.F. Ryan (1970): Somatic chromosomes of eight mammals from Con Son Island, South Vietnam. — Caryologica 23: 173-181. Duncan, J.F., R. Irsiana & A. Chang (1974): Karyotypes of five taxa of Rattus (Rodentia, Muridae) from Indonesia. — Cytologia 39: 295-302. tivhmwova. O-.P.. G. Musser, G. Kuznetsov & J. Pavlinov (1998): Preliminary results of allozymic and cranial analyses of Niviventer (Rodentia: Muridae) from Southern Vietnam. — p.356 in: Abstracts of Euro-American Mammal Congress (ed. Santiago Reig). — Univ. Press, Santiago de Compostela. Markvong, A., J.T. Marshall, jr. & A. Gropp (1973): Chromosomes of 356 rats and mice of Tailand. — Nat. Hist, Bull. Siam. Soc. 25: 23-40. Musser, G.G. (1981): Results of the Archbold expeditions (no. 105): Notes on systematics of Indo-Malayan murid rodents, and description of new genera and species from Ceylon, Sulawesi, and Philippines. — Bull. Mus. Nat. Hist. 168: 229- 334. Musser, G.G. & D. Carleton (1993): Family Muridae. — pp.501-755 in: Mammal species of the world. A taxonomic and geographic reference. 2. edit. /eds: Wilson, D.E. & D.M. Reeder). — Smithsonian Institution Press, Washington & London. Sumner, A.T. (1972): A simple technique for demonstrating. centromente heterochromatin. — Exp. Cell Res. 75:304-306. Yong, H.S. (1969a): Karyotypes of Malay rats (Rodentia, Muridae, genus Rattus Fischer). — Chromosoma 34: 245-267. Yong, H.S. (1969b): Karyotypes of three species of rats from Hong Kong and Tailand (Muridae, genus Rattus Fischer). — Cytologia 34: 394-398. Yosida, T.H. (1973): Evolution of karyotypes and differentiation in 13 Rattus species. - Chromosoma 40: 285-297. Authors’ address: Dr. Maria Baskevich and Dr. German Kuznetsov, Severtsov Institute of Ecology and Evolution, Russian Academy of Sciences, Leninsky prospect 33, Moscow 117071, Russia, e-mail: sevin@glas.apc.org. 93 Mammals of coastal islands of Vietnam: zoogeographical and ecological aspects German Kuznetsov Abstract: The faunistic composition of the coastal islands of Vietnam changes according to their geographic position and landscape conditions. For example, along the 1500 km from north to south different species of Callosciurus can be found on the islands: Dangkho is inhabited by C. erythraeus, Cham by C. flavimanus, and the southern islands by C. finlaysonii with different subspecies. Similar successions and relations were reported in birds, reptiles, insects and other groups of animals. The number of individuals of the genus Callosciurus changes accordingly: it decreases from south to north. Obviously the environmental conditions for squirrels improve towards the south. The islands have higher numbers of mammals, chiefly of rodents, than the continent. For instance, the total number of all species of rats in natural biotopes on the islands is about twice as high as on the continent in the natural communities of primary tropical forests. A similar rule is observed among species of squirrels of the genus Callosciurus. The numbers of rats, however, on different islands varied from 0.9 to 10.0 per 100 trap-nights. We did not find any clear correlation between the species diversity of rats and their abundance. Key words: coastal islands of Vietnam, mammals, diversity, ecology, biogeography Introduction It is noteworthy that zoologists and biogeographers find it particularly interesting to examine oceanic islands, regarding them as the simplest and most convenient models of ecological relationships between animals as well as their evolution (Voronov 1976,Chernov 1982, Guilleume 1983, Mein 1983, Yes’ kov 1984, Lawlor 1986). At the same time, studies of coastal islands allow a more detailed tracing of colonization, extinction, immigration, as well as of changes in the structure of communities, when compared with the continental fauna. Such studies also help to test various hypotheses on the initial formation of insular faunas (Crowell 1986, Istomin 1989). It is also of interest to clarify peculiarities of species diversity and to design models of the colonization of such islands by continental species. Most likely such studies were primarily driven by the concept of equilibrium zoogeography which was advanced by MacArthur & Wilson (1967) over 30 years ago. Their theory demonstrates the dynamic pattern of equilibrium in insular biota between immigration from sources of colonization and the extinction of colonists. Terrestrial mammals have specific features that are useful in colonizing islands and in distributing over them. There are also characteristic traits in the adaptational Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 358 evolution of mammals remaining on such islands because of their sudden isolation from other animals. In this connection the study of mammals on the coastal islands of Vietnam is interesting. This is even more important because practically no theriological studies of these islands had ever beenmade. It should be mentioned that insular fauna is usually approached from both ecological and biogeographical points of view (Levushkin 1982). Material and methods We collected our materials when working within the Zoogeographical Team of the Research Vessel “Academician Aleksander Nesmeyanov”. The aim was to study the insular ecosystems of the coastal shelf of Vietnam from the Archipelago of Bai Tu Long in the Gulf of Tonkin down to the islands in the Gulf of Siam. In total, we examined seven islands (Dangkho, Phoung Hoang, Cham, Bai Canh, Con Dao, Tho Chu‘, Thom) during the period from 20.03. to 13.04.1987. We processed 1860 trap- nights and determined the numbers of squirrels of the genus Callosciurus; all transect censuses equaled 65 km. By shooting, Gero traps and nets we collected about 200 individuals, chiefly Chiroptera, rats and squirrels. Our main purpose was the study of the taxonomy of the mammalian fauna of the islands examined. We also recorded some ecological parameters of the populations of rats and squirrels (abundance, diurnal activity, distribution in the habitat, reproduction, etc.). For other islands of the Vietnam shelf (Kaitien, Ban Sen, Quan Lan, Cat Ba and Phu Quoc) we used material from the expedition carried out by the National Central Research Institute of Vietnam during the period 1988-1998. We also considered literature relevant to this region (Kloss 1926, Van Peenen et al. 1970, Kuznetsov & Fam Chong An’ 1992) and investigated mammalian collections in different museums of Vietnam, the Moscow State University and the private collection of Fam Chong An’. Taxonomic list of mammalian species on the coastal islands of Vietnam To characterize mammalian fauna on the coastal islands of Vietnam we used data from 12 islands: from the Archipelago of Bai Tu Long in the Gulf of Tonkin in the north to the Tho Chu‘ island in the Gulf of Siam in the south, i.e. we considered almost the entire area of large coastal islands of Vietnam. We identified 78 mammalian forms inhabiting the coastal islands of Vietnam obtained by trapping, direct observations, life records, and questionnaires. The coastal islands of Vietnam are inhabited by representatives of almost all mammalian orders. On the islands of the Archipelago of Bai Tu Long in the Gulf of Tonkin small bats (Hipposideros, Rhinolophus, Pipistrellus, Scotophilus) are most numerous; among monkeys the most frequent species is Macaca mulatta, among rodents Rattus, Niviventer, Leopoldamys; among squirrels Callosciurus, Tamiops; among Carnivora Viverricula indica, Prionailurus bengalensis, Lutra lutra; among >59 Ungulata Sus scrofa and Muntiacus muntjak. Cervus unicolor was widespread on the islands of Bai Tu Long. Immigration from the continent may have started on the northern islands of Vietnam, the Archipelago of Bai Tu Long; here potential ecological conditions permitted the new species to exist. However, there is no doubt that in the historical past ungulates also inhabited many southern islands of the country. Until recently, the island of Con Dao was inhabited by such a large species as Bubalus bubalis and the Bai Tu Long Archipelago was inhabited by mammalian species which are typical of the mainland of northern Vietnam, meaning some species have survived better on the islands than on the mainland. For example, the island of Cat Ba is inhabited by such species as Capricornis sumatraensis, Macaca arctoides, Presbytis francoisi. We must admit that the habitat is of greatest importance for the formation of the fauna on these islands Chernov (1982) was right in saying that the processes of colonization of these islands are essentially determined by biological relationships and by the specific insular habitats. To illustrate this idea we can point to the habitat of Capricornis sumatraensis on Cat Ba: this species could survive only due to the hard rocky biotopes of the local limestone mountains. Indeed, as Chernov says, “...the essence of the question is not how it may come to the island but how it can live on it” (Chernov 1982: 36). This remark is valid for other continental islands as well. Therefore biogeographical and ecological factors determine the distribution of many animal groups on the islands. Of particular interest in terms of mammals is the distribution of Viverridae on the coastal islands of Vietnam. Arboreal viverrid species seem to be absent from the northern islands, but quite common on some southern ones (Con Dao island). Viverrid species with a broader nutritional specialization, for instance Viverricula indica, have some advantages over narrow- diet viverrids widespread in terrestrial biotopes including the sea coast, while V. indica 1s spread over many islands of the Bai Tu Long Archipelago. On some of them (Kaitien, Phoung Hoang) the species 1s quite common. It is interesting that V. indica is less numerous in areas with other carnivores, for instance on Cat Ba island where Herpestes javanicus, Viverra zibetha and Melogale moschata live. By contrast, on the islands without competition from other carnivores, Viverricula indica largely increases its population (on Kaitien island). The human influence has proved to be the most powerful today in its impact upon the distribution and spread of viverrid species on the coastal islands of Vietnam. It is likely that anthropogenic factors also explain the lack of viverrids on the island of Thom in the Gulf of Siam. Immigration of some viverrid species from the island of Phu Quoc seems possible, since the distance between these two islands is only 4 km. The same factor may explain the absence of Hylopetes lepidus on the island of Thom, because this island is very small and, instead of tropical forest, bears large plantations of mango and other crops, preventing Hylopetes lepidus from recolonizing the island (pers. comm. Cao Van Sung). Analysis of the distribution of various animals on the islands - mammals (including bats, see table 1), reptiles, locusts, beetles and other insects - shows that we can divide them into two groups: the Archipelago of Bai Tu Long in the north 360 and the other islands (Darevskii et al. 1991). The principal zoogeographical border on the islands coincides with that on the mainland of Vietnam (fig.1), drawn somewhat north of Da Nang (Dao Van Tien 1978). Though the climatic regimes differ essentially, the coincidence of borders points to the predominant influence exerted by the neighbouring land upon the composition of the insular faunas, and hence to the great role animal migrations play here. Vietnam 7 16° Thailand Central Annam Fig.1: Map of Vietnam with the sites of studied islands: 1 — Kaitien 2 — Dangkho 3 — Ban Sen Cambodia 4 — Quan Lan 5 — Phoung Hoang 6 — Cat Ba 7 —Cham 8 — Bai Canh 9 — Con Dao 10 — Tho Chu‘ Cochinchina 9 e 9 11 - Thom 12 - Phu Quoc Insular faunas result from a balance between immigration and extinction; the proximity to the coast does not determine any specific insular anımal populations. The chiefly continental pattern of the fauna could be associated not only with the relatively late (Late Pleistocene) separation of the islands from the mainland but also with the animals’ considerable migration ability. Immigration could be areason behind the patchy pattern of many populations on the islands, but the picture observed today can be better explained by the theory of “equilibrium” insular biogeography (Darevskii et al. 1991). Table |: Primary list of mammals of the continental islands of Vietnam Se seTeTe Ten Te] AN 1) Sea +L Crocidura fuliginosa +L Suncus murinus +C +L Tupaia glis +L +L +L +C | +LV Cynopterus sphinx +L +L Macroglossus minimus +CL Pteropus hypomelanus ++V | +4+LV | +V +V +V P. vampyrus +#V I +#V I EV EV ler Rousettus leschenaulti +V Taphozous melanopogon +V +P Megaderma spasma + +V +P Hipposideros armiger +L |++V | +4+L | +L +4+C | ++V | +V ++C V V H. bicolor +V +V ARC || ap \y +L teE2 lH FE H. diadema +V +V +P H. larvatus ++V ++C I ++€CL 1 +V | +V | ++CE +€ Rhinolophus borneensis +VL | +V R. luctus | +LV R. thomasi +V +V +L Pipistrellus abramus +V +V P. ceylonicus +V P. javanicus +V Scotophilus heathi +V Macaca arctoides +L +L M. mulatta +V I+C| +C | +V +V. 21 2+@IE M. nemestrina +L M. fascicularis ++V [| ++V [| ++VL | ++V ++L Presbytis francoisi poliocephalus ++CL Hylobates lar +L Lutra lutra +4+V I +V I +4+V | +V ++C +L Aonyx cinerea +P Melogale moschata +L Paradoxurus hermaphroditus +P ++VP | +L Viverricula indica +4+4+C] +V | +4+C | +4+C +C | +4+CL +V Viverra zibetha +P +P +VL Herpestes urva +L +P H. javanicus +V Prionailurus bengalensis +C | +C +L +L Panthera pardus +L Capricornis sumatraensis +P | ++V +P I++VL Sus scrofa 4+4+V 14+4+C ] ++V | +V | +4+V I +V Cervus unicolor +C +L Muntiacus muntjak +V [| +C [| +4+C | +4+V [| +4C [4+4VL Bubalus bubalis +L Manis pentadactyla +V +P +P Callosciurus erythraeus castaneoventrig ++C | ++C | ++C | +C ++V | +4+C C. flavimanus C. finlaysonii germaini C. finlaysonii pierrei C. finlaysonii harmandi Dremomys rufigenis Tamiops macclellandi Tamiops swinhoei Ratufa bicolor hainana R. bicolor condorensis Hylopetes spadiceus H. lepidus Menetes berdmorei Rhizomys pruinosus Bandicota indica Berylmys berdmorei Leopoldamys sabanus Maxomys surifer Mus caroli Niviventer bukit N. confucianus | N. cremoriventer N. niviventer N. fulvescens Rattus argentiventer . exulans . nitidus . norvegicus . molliculus . koratensis . flavipectus . germaıni Hystrix brachyura Atherurus macrourus Lepus peguensis Legend | — 12: names of the islands: 1 — Kaitien; 2 — Dangkho; 3 — Ban Sen; 4 — Quan Lan; 5 — Phoung Hoang; 6 — Cat Ba; 7 — Cham; 8 — Bai Canh; 9 — Con Dao; 10 — Thö Chu‘; 11 — Thom; 12 — Phu Quoc. + — rare species, ++ — uncommon species, +++ — common species, C — the species was collected. L - published data, V — visual observations, P — information by local people. Ecological peculiarities and the evolution of insular mammals The mammals of the coastal islands of Vietnam illustrate some of the ecological regularities of insular biogeography. For example, on some islands (Cham, Tho Chu‘) we found a compensation effect, i.e., when the diversity of rat species on an island decreases, the number of individuals increases. However, this effect was not manifested in all cases, suggesting more intricate relations in multi-species communities. Obviously, the mechanisms governing the 363 formation of insular communities are still unclear. The hypothesis has been advanced that during the succession of insular invertebrate faunas, “early” species with high reproduction rates are replaced by “late” species with low reproduction rates. In terms of population ecology this can be explained by the turnover from r-strategy to K-strategy, and by the increasing complexity in the trophic structure of the community. In terms of biogeography this can be explained by the growing importance of the “late” species with their slower rates of colonization (Sukhanov 1982: 74). It is possible that to some extent mammals use the same strategy. However, there are no reliable data on the variations in the species composition of every insular fauna type, making it difficult to apply this hypothesis to all mammal species. It is obvious, however, that in all species of different genera of rats dispersion plays an important role. According to the studies of Kucheruk & Lapshov (1989) - using “competitiveness” among representatives of local Rattus species as a factor in distribution - we can explain the poor penetration of the Brown rat into many islands between South-East Asia and Australia. In our analysis Rattus norvegicus was caught on the island of Cham not far from the port of Da Nang in the central part of Vietnam. Except for sightings on Dangkho, this species was not found on the other islands. Half of the islands studied are inhabited by 6 to 7 rat species (except the islands of Kaitien, Ban Sen and Quan Lan where rats were not studied). On the different islands different species are dominant: on Dangkho 72% of all rats caught are Rattus flavipectus, on Cham the most abundant is R. nitidus (42.1%), prior to R. norvegicus (26.3%). On the southern islands other species are predominant. Among the dominants on Con Dao we found Leopoldamys sabanus (66,6%); on Bai Canh R. germaini (55.5%) and L. sabanus (33%); on Thom Maxomys surifer (100%); on Thö Chu‘ R. koratensis (42.1%) and R. germaini (26.3%). Taking into account the predominant colonization of small islands by the House mouse Mus musculus (Kucheruk & Lapshov 1989), mice of the genus Mus must undoubtedly be present on many coastal islands of Vietnam. However, only Mus caroli has so far been recorded, solely on Cat Ba. There are considerable differences in population numbers of rats between different islands, caused mainly by human influence on the natural ecosystems of the islands. These impacts enhance the reproduction of the most competitive species of rats, which are well adapted to changes in the local climatic conditions and habitats. Despite the generally low altitudes of the islands (the maximum height above sea level is 600 m), many of them exhibit altitudinal differentiation in mammalian populations. For example on Dangkho (the highest site above sea level is 140 m), the mammalian population is altitudinally differentiated both in terms of species and numbers. The maximum population density among rats, mainly Rattus flavipectus (72% of all individuals), was registered at the coast near human dwellings (26.6 individuals per 100 trap-nights) and in anthropogenic landscapes (5.3 individuals per 100 trap-nights) up to the altitude of 20 m a.s.l. Above that height the plant communities, and accordingly the rat species, undergo a successional change. Callosciurus erythraeus, Macaca mulatta and Niviventer bukit inhabit the tropical forest from 60 to 140 m a.s.l., i.e. the upper part of the island. Muntiacus muntjak and Cervus unicolor are encountered in almost all biotopes of the island, except its rocky coast and the narrow anthropogenic zone. Lutra lutra colonizes only the marine coastal zone, feeding on marine invertebrates (mollusks, crabs, etc.), up to 2 m a.s.l. Viverricula indica and Prionailurus bengalensis are more frequently found on the sea coast, but sometimes also on mountain slopes. A strict vertical zonation was found in the distribution of the herpetofauna, for instance on the island of Con Dao (pers. comm. I.S. Darevskii). 364 According to geographical position and landscape conditions, some specific altitudinal distribution patterns of mammals were also observed on the other islands studied, such as Con Dao, Bai Canh, and Tho Chu‘. Their animal populations change from north to south (from the Gulf of Tongking to the Gulf of Siam). Along the more than 1500 km of islands we find a succession of species of the genus Callosciurus. Dangkho is inhabited by C. erythraeus, Cham by €. flavimanus, and the southern islands by C. finlaysonii with different subspecies. This succession may be related to some pecularities of zoogeographical regions on the mainland. A similar succession and relationship were found in birds, reptiles, insects and other groups of animals. Squirrels of the genus Callosciurus occur on the southern islands in greater numbers (e.g., 2.16 individuals per km of census on Thö Chu‘) than on the northern ones (e.g., 0.77 individuals per km on Dangkho). Apparently the south offers a more favorable environment for squirrels, permitting them to realize their biotic potential. Compared to the mainland, the islands have higher numbers of mammals, mainly of rodents. For instance, the total numbers of all rat species in natural habitats on the islands are about twice as high as in primary tropical forests on the mainland of Vietnam. A similar relationship can be observed among species of Callosciurus squirrels. With the number of rats on different islands varying from 0.9 to 10.0 trap-nights, we did not find any correlation between the diversity of rats and their abundance. This shows that the numbers of individual animals and their species numbers are formed by camplex mechanisms. It is possible that a certain role in the process is played by the size of an island and its habitat diversity. It is interesting that the highest numbers of rats were recorded on Cham and Thö Chu‘, with areas of 11.9 - 13.8 km’. On small islands we noted the phenomenon of “‘faunistic deficiency”. This can be accounted for by the fact that the area of such islands - as a rule - is smaller than the critical territory necessary for certain species of mammals, birds or other groups of animals. At the same time, these islands are under heavy human pressure (Cham, Thom, Th6 Chu‘). Insular ecosystems are known to be less resistent to anthropogenic impacts than continental ones. However, there are cases when “faunistic deficiency” can be largely attributed to other reasons, such as ecological or landscape factors (habitat equipment, terrain features, lack of water bodies, etc.) or isolation by strong sea currents between islands. The latter is the case between Con Dao and Bai Canh, where straits of just 800 m in width separate the islands, thus forming a barrier for many animals. This is obviously the reason why the mammalian fauna on Bai Canh is 1.5 to 2 times poorer than that on Con Dao. Since this also applies to the insect fauna (pers. comm. A. Ponomarenko), the diversity of ants on Bai Canh is much lower than on the island of Con Dao. However, there are some ant species which are found only on Bai Canh. Therefore the difference in the faunistic composition cannot be explained by the different size of the islands alone. The presence of different species on Bai Canh and on Con Dao favors the suggestion that their fauna is rather transient, which corresponds to the predictions of the equilibrium theory of insular biogeography. On the other hand, the essentially lower diversity of animals on Bai Canh island compared with Con Dao, despite the anthropogenically impoverished fauna of the latter, points to the primary significance of the dimensions of islands and also favors the theory of equilibrium biogeography. Habitat equipment does not only influence the colonization process on the islands (Chernov 1982) but also the survival of the species there. It governs the conditions of habitat diversity the moment the islands become isolated and then maintains this diversity of plant communities. We believe that this can be demonstrated by the existence of Lepus peguensis (a xerophylous 365 species) on Con Dao, or of some strictly specialized species (such as Tupaia) inhabiting the tropical forest on Bai Canh, or Capricornis sumatraensis living on the limestone residual rocks of Cat Ba. At the same time isolation of the islands promoted the divergence of various groups of animals. So, the mammalian fauna on the coastal islands of Vietnam characteristically contains endemics of mainly subspecies rank. For example, on Con Dao more than 13 subspecies of mammals were recorded. We find similar examples in birds and reptiles. All this demonstrates the rate of insular divergence within the last 10 - 15 thousand years. Today the rate of mammal diversification on the coastal islands is basically limited by the rate of extinction (Lawlor 1986). Acknowledgment I am grateful to Fam Chong An’, Dang Huy Huynh, Cao Van Sung and A.G. Ponomarenko for their assistance in my work. References Chernov, Yu. I. (1982) : On the ways and sources for the formation of fauna of small islands in the Oceania. — Zhurn. obshch. biol. 43: 35-47. Crowell, K.L. (1986) : A comparison of relict versus equilibrium models for insular mammals of the Gulf of Maine. — Biol. J. Linn. Soc. 28: 37-64. Dao Van Tien (1978) : An experience in zoogeographical regionalisation of Vietnam. — Zool. zhurn. 57: 582-586. Pagel ES. A.G. Ponomarenko & G.V. Kuznetsov (1991): Zoogeography of coastal islands of Vietnam. — Priroda 9: 42-48. Guilleume, C.P. (1983): Les reptilas insulaires mise en place et evolution. — Colloq. Intern. CNRS 330: 281-287. Heaney, L.R. (1986): Biogeography of mammals in SE Asia: estimates of rates of colonization, extinction and speciation. — Biol. J. Linn. Soc. 28: 127-165. Istomin, A.V. (1989) : Formation of insular communities and populations as illustrated by the example of small mammals from the Kandalaksha Bay of White Sea. — Author. abstr. dissent. = p: 21. Kloss, C.B. (1926) : Mammals from Pulo Condore, with descriptions of two new subspecies. — J. Nat. Hist. Soc. Siam 4 : 357-359. Kucheruk, V.V.& A.V. Lapshov (1989) : Synanthropic rodents of oceanic islands. — Ekologia 2: 41-49. Kuznetsov, G.V. & Fam Chong An’ (1992) : Mammals of the coastal islands of Vietnam ( biogeographical and ecological aspects). — pp. 182-197 in: Zoological investigations in Vietnam (ed. Sokolov, V.E.). — Moscow, Nauka. Lawlor, T.E. (1986) : Comparative biogeography of mammals on islands. — Biol. J. Linn. Soc. 28: 99-125. Levushkin, S.N. (1982). : The problem of insular faunas in the light of relations between biogeography and ecology. — pp. 26-52 in : Morskaya biogeog. : predmet, meth., printsipy raionir. — Moscow, Nauka. MacArthur, R.H. & E.O. Wilson (1967) : The theory of island biogeography. — Princeton, Univ. Press. 366 Mein, P. (1983) : Particularités de l’Evolution insulaire chez les petits mammiferes. — Colloq. intern. CNRS 330: 189-193. Sukhanov, V.V. (1982): A model of species abundance distribution over the littoral of an insular ridge. — pp. 52-75 in: Morskaya biogeog. : predmet, meth., printsipy raionir. — Moscow, Nauka (1982). Van Peenen, P.F.,M.L. Cunningham & J.F. Duncan (1970): A collection of mammals from Con Son island, Vietnam. — J. Mammal. 51: 419-424. Voronov, A.G. (1976) : General problems of biogeography and its main trends at the present-day stage. — pp. 8-97 in: Itogi nauki I tekniki. Biogeografiya. - Moscow. VINITI (1978). Yes’kov, K.Y. (1984): The drift of continents and problems of historic biogeography. — pp. 24- 92 in: Faunagenesis and phylogenesis (ed. Chernov, Y.I.). — Nauka, Moscow. Author’s address: Dr. German Kuznetsov, Severtsov Institute of Ecology and Evolution RAS, Leninsky prospect 33, 117071 Moscow, Russia 367 On the herpetofauna of the Sultanate of Oman, with comments on the relationship between the Afrotropical and Saharo-Sindian faunas Poorman Wrihinis ee FSi Biukbrert Abstract: At present a total of 90 species of amphibians and reptiles is known to occur in the Sultanate of Oman. During a field trip to the Sultanate a total number of 272 specimens representing 32 species of reptiles and amphibians was collected. Six additional species were observed but not collected. Some recently collected material, comprising 16 species and 30 specimens from the ZFMK collection, is included in the commented list provided, resulting in a total of 42 species listed. It is followed by a list of localities visited with their respective herpetofaunal communities. The zoogeographic relationships of the Omani herpetofauna with the Palearctic (Iran) and Afrotropical (Somalia) faunal realms is briefly discussed. Key words: Oman, Amphibia, Reptilia, zoogeography Introduction At present a total of 90 species of reptiles and amphibians are known to occur in the Sultanate of Oman. The largest group are the lizards (including amphisbaenians) with 55 species. The snakes have 28 species, the turtles five species (only marine), and the anurans (toads) two species. During a field trip to the Sultanate of Oman (fig.1) in November and December 1998 a total of 272 reptiles and amphibians, representing 32 species, where collected. Six additional species were observed but not collected. Most of the material is now kept in the collection of the Museum Koenig, Bonn, Germany (ZFMK). A considerable number of specimen were kept alive to establish breeding colonies in captivity to gain additional information on reproductive biology and behaviour. The aim of this paper is to provide a preliminary commented list of the material collected during the 1998 expedition to the Sultanate of Oman (followed by a list of localities visited with their respective herpetofaunal communities) and of some material in the ZFMK. The following literature was used for identification of the material: Arnold 1975, Arnold 1980, Arnold & Gallagher 1975, Leviton et al. 1992 and Salvador 1982. Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 Saudi- Arabia Fig.1: Map of the Sultanate of Oman with collecting sites (circles: collected by Wilms & Hulbert; other material in ZFMK, Bonn. 369 Commented list Amphibia Bufonidae Bufo arabicus Heyden, 1827 Type locality: “Das petraeische Arabien” (= Sinai Peninsula), Egypt Localities and collection numbers: ZFMK 71000-003, locality no.17, 28.11.1998; Re 10820028. 0412271998) still alive: ZEMK 70968-972, locality no.30, 04.12.1998 Distribution: Arabia We found Bufo arabicus in canals used to irrigate the gardens on the edge of an oasis (locality no.17). There we found only adult specimens. The other localities are located in the Wadi bani Awf. In that mesic region tadpoles as well as semiadult and adult specimens were found. Bufo dhufarensis Parker, 1931 Type locality: Milwah al Aud, Dhofar, Oman Locality and collection numbers: ZFMK 70973-976, locality no.15, 25.11.1998 Distribution: This species lives in peripheral Arabia from the vicinity of Mecca south to Yemen and eastward to northern Oman. The toad Bufo dhufarensis was found in a Wadi on Masirah Island. According to Meinig & Kessler (1998), spawn and tadpoles were not found in late November and early December. This was confirmed by our observations. Reptilia Agamidae Acanthocercus adramitanus (Anderson, 1896) Type locality: Hadramaut, Yemen Locality and collection number: ZFMK 70999, locality no.8, 20.11.1998 Distribution: western and southern Arabia This specimen was found in Wadi Hinna at the edge of a small clear stream. Phrynocephalus arabicus Anderson, 1894 Type locality: Plateau of the Hadramaut, Yemen Localities and collection numbers: ZFMK 41347, locality no.20; ZFMK 66643 & ZFMK 66555-556, locality no.36; ZFMK 70551-552, locality no.37 Distribution: widely distributed in Arabia. Pseudotrapelus sinaitus (Heyden, 1827) Type locality: Sinai, Egypt Localities and collection numbers: ZFMK 70550, locality no.42; sight record, localıty 00.3, 13.11.1998: ZEMK 70923, locality no.6, 19.11.1998; ZEMK 70998, locality no.9, 20.11.1998; still alive, locality no.10, 20.11.1998 Distribution: Egypt, Sudan, southeastern Libya, Palestine, Jordan, and northern, western and southern parts of Arabia 370 Uromastyx aegyptia microlepis Blanford, 1874 Type locality: vicinity of Basrah, Iraq Locality: killed by predator, locality no.5, 17.11.1998 Distribution: widespread in Arabia, in the north from Wadi Araba eastwards to western Iran An old carcass without head was found at the border of a dry Wadi. We did not see any living specimen during the whole trip. Therefore we suppose that this species is hibernating during wintertime. Uromastyx benti (Anderson, 1894) Type locality: Mukalla, Hadramaut, Yemen Locality: still alive, locality no.10, 20.11.1998 Distribution: Yemen and southwestern Oman Uromastyx benti was recorded for the first time from the Sultanate of Oman in 1998 (Seufer et al. 1998). We have found this species on the Mirbat plain. The adult specimens inhabit rocky outcrops with a sparse vegetation and retreat into cracks and crevices. The ground was stony. The semiadult specimens were not found in the same area. They inhabited a rocky habitat nearby with a sandy substrate. Uromastyx thomasi Parker, 1930 Type locality: Bu Ju’ay, Rub’al Khali, Oman Locality and collection number: ZFMK 70413 and alive, locality no.12, 23.11.98- DS MATOS Distribution: coastal area of southern Oman; Dhofar up to Masirah Island A dense population of Uromastyx thomasi was found near Ras Hilf, Masirah Island. The habitat was a stony, in some parts sandy plain with sparse vegetation. The specimens were active from 11.00 to 16.00 h. The temperature of the surface and air temperature one meter above ground was measured. The surface temperature ranged from 37.5 to 51.1°C, while the air temperature was 29.2 - 35.8°C. The body temperature of 24 specimens ranged from 33.1 to 39.4°C. The temperature was measured within 5 minutes of capturing the animal. The temperature in the burrows was between 30.3 and 33.6°C at a depth of 20-41cm. In four burrows temperature at a depth of more than 53 cm was measured. The data are 34°C at 53.5 cm, 31.4°C at 47.5 cm, 29.7°C€ at 67 cm and 29°6-© at 772s) cme length of the burrows varied between 45 and 165 cm. One specimen played dead after capturing. Gekkonidae Asaccus platyrhynchus Arnold & Gardner, 1994 Type locality: Tanuf, Oman Localities and collection number: sight record, locality no.30, 04.12.1998; ZFMK 70967, locality no.31, 04.12.1998 Distribution: Jebel Akhdar region, northern Oman 3 One specimen was observed in a small crack together with five Ptyodactylus hasselquistii and two Hemidactylus persicus. In captivity these animals lay only one single egg per clutch. Bunopus spatalurus hajarensis Arnold & Gallagher, 1977 Type locality: Wadi Ham, Masafi (25°18.5'N 56°10'E), Oman Localities and collection numbers: ZFMK 71019, locality no.1, 12.11.1998; ZFMK 71020, locality no.23, 01.12.1998; ZFMK 71018, locality no.25, 01.12.1998 Distribution: eastern UAE, northern Oman south to Masirah island Bunopus tuberculatus Blanford, 1874 Type locality: Baku Kalat, Bampur and Mand (Baluchistan) Localities and collection numbers: ZFMK 70888-890, locality no.4, 15.11.1998; still alive, locality no.20, 29.11.1998 Distribution: Palestine, Jordan, southern Iraq, Arabia, Iran, Afghanistan and Pakistan Cyrtopodion scaber (Heyden, 1827) Type locality: near Tor, Sinai Peninsula, Egypt Localites and collection numbers: ZFMK 53524-526, locality no.3; ZFMK 71014- 017 and alive, locality no.25, 01.12.1998 Distribution: from Egypt, south to Ethiopia and east to Pakistan Hemidactylus flaviviridis Rüppell, 1835 Type locality: Massawa, Eritrea Localities and collection number: ZFMK 53526, locality no.3; sight record, localities 25 and 26, 01.12.1998 and 02.12.1998 Distribution: coastal areas of the Red Sea, Somalia, Arabia; Iraq, Iran, Pakistan, northern India Hemidactylus homoeolepis Blanford, 1881 Type locality: Socotra, Yemen Localites and collection numbers: ZFMK 66562, locality no.35; ZFMK 70883, locality no.5, 16.11.1998 Distribution: Socotra, southern Arabia from Shugra to Ad Daffah near Ras al Hadd (Gardner pers. comm.) Hemidactylus turcicus parkeri Loveridge, 1936 Type locality: Zanzibar, Tanzania Locality and collection numbers: ZFMK 70415, ZFMK 70952-953, locality no.13, 23.11.1998 Distribution: coastal areas of the Red Sea, Somalia, southern Arabian littoral as far as eastern UAE, southern Iran, Pakistan Ss men un Sa AR N > x Above left: Uromastyx thomasi from Masirah Island Photo: Thomas Wilms Above right: Echis cf. pyramidum from 3km west of Mirbat Photo: Felix Hulbert Bottom left: Pristurus carteri from 16°58 N, 53°50 E (west of Salalah) Photo: Felix Hulbert Hemidactylus persicus Anderson, 1872 Bypielocalteyz: Ican Locality: still alive, locality no.30, 04.12.1998 Distribution: Iraq, southern Iran, Pakistan, northeastern Saudi Arabia, northern Oman, Bahrain Hemidactylus yerburii Anderson, 1895 Type locality: Haithalhim and Aden, Yemen Locality: still alive, locality no.5, 16.11.1998 Distribution: southern Arabia, northern Somalia Pristurus carteri (Gray, 1863) Type locality: Masirah Island, Oman Localities and collection numbers: ZFMK 66554, locality no.36; ZFMK 70899, locality no.2, 12.11.1998; ZFMK 70875-8800, ZFMK 70990-992, locality no.5, 16.11.1998; ZEMK 70927-928, locality no.6, 19:11. 19938. ZEMRKU IST ZFMK 70988-989, locality no.10, 20.11.1998; ZFMK 70934-941, locality no.12, 23.11.1998-26.11.1998; ZEMK 709322933, locality no. 14, 2593998 Distribution: southern Arabia, from northern Yemen to Seeb, northern Oman (Gardner pers. comm.), Masirah island During the field trip, between 29.11 and 01.12.1998, one female from the vicinity of Thumrait laid one single egg. The juvenile hatched after 70-72 days at 29-30°C. Pristurus gallagheri Arnold, 1986 Type locality: Rostaq, Jebel Akhdar, northern Oman Locality: still alive, locality no.30, 04.12.1998 373 Distribution: Jebel Akhdar We found P. gallagheri on trees and on rocks. The species is arboreal (Gardner 1994). Pristurus minimus Arnold, 1977 ieapiesocality: Jazincoast, (18° 30'N, 56° 30'E), Oman Localities and collection numbers: ZFMK 68500, locality no.31; ZFMK 70891- 898, locality no.2, 12.11.1998; ZFMK 70942-943, locality no.12, 26.11.1998; sight record, locality no.14, 25.11.1998; ZFMK 70903, locality no.20, 29.11.1998 Distribution: southeastern Arabia Pristurus rupestris Blanford, 1874 Type locality: Muscat, Oman Localities and collection numbers: ZFMK 60558, locality no.33; ZFMK 70900- PUeemocadiity now, 12771998, sieht record, locality no.3, 13.11.1998; ZEMK 70885-886, locality no. 5, 16.11.1998; ZFMK 70924-926, locality no.6, 19.11.1998; ZFMK 70994-997, locality no.8, 20.11.1998; sight record, locality nO OS 2, 1998 Distribution: North-east Africa, Jordan, Arabia and coastal Iran One hatchling of the Wadi Hinna population measured 12,5 mm (Snout-vent length) and 13 mm (Tail length). Ptyodactylus hasselquistii (Donndorff, 1798) ivpe toe ality: Cairo, Egypt Localities and collection numbers: ZFMK 66561, locality no.35; ZFMK 70961- 962, locality no.3, 13.111998; ZFMK 70881-882, locality no.5, 16.11.1998; ZFMK 70963-966 and alive, localities no.30 & 31, 04.12.1998 Distribution: North-east Africa, Arabia, north to Syria, Iraq and southwestern Iran At locality no.3 three clutches of P. hasselquistii were found in a cave. The diameter of all eggs was 15.8 mm. Stenodactylus arabicus (Haas, 1957) Type locality: Abqaiq, near Dharan, Saudi Arabia Localities and collection numbers: ZFMK 66557-559, locality no.36; ZFMK 66563-564, locality no.34; ZFMK 70979, locality no.18, 28.11.1998 Distribution: Saudi Arabia, UAE, southern Yemen, Oman ZFMK 70979 was found in sand dunes (Wahiba) with very sparse vegetation at IGS Orle Stenodactylus doriae (Blanford, 1874) Type locality: Bandar Abbas, Iran Localities and collection numbers: ZFMK 42492, locality no.20; ZFMK 70904- 905, ZFMK 70977-978 and alive, locality no.20, 29.11.1998 Distribution: widespread in Arabia, Palestine, Jordan, Iraq, southwestern Iran 374 Stenodactylus leptocosymbotes Leviton & Anderson, 1967 Type locality: between Dubai and Abu Dhabi (UAE) Localities and collection numbers: ZFMK 70887, locality no.4, 16.11.1998; ZFMK 70944-950, ZFMK 70414, locality no.12, 24.11.1998; ZFMK 70954-960, locality no.16, 27.11.1998; ZFMK 71004-013, locality no.19, 28.11.1998; ZFMK 70906- 922, locality no.20, 29.11.1998 Distribution: southeastern and southern Arabia; eastern UAE and Oman; westwards to southern Yemen Normally Stenodactylus leptocosymbotes prefers harder substrates than St. doriae (Arnold 1980). We found both species sympatric in an area with soft sand. Two females from Masirah Island laid two clutches on 31.01.99. The eggs measured 13.7 x 9.7 mm (two eggs), 12.3 x 10 mm and 12.3 x 10.1 mm. Tropiocolotes scorteccii Cherchi & Spano, 1963 Type locality: El Safa, Yemen Locality and collection number: ZFMK 66560, locality no.10; still alive, locality no.5, 17.11.1998 Distribution: southern Arabia, Hadramaut and Dhofar Lacertidae Acanthodactylus blanfordii Boulenger, 1918 Type locality: “Perse et Béloutchistan” Locality: still alive, locality no.24, 01.12.1998 Distribution: southeastern Iran, western Pakistan, southern Afghanistan and northern coast of the Sultanate of Oman Acanthodactylus boskianus (Daudin, 1802) Type locality: Egypt Locality: sight record, locality no.17, 28.11.1998 Distribution: North Africa, Arabia northwards to the Turkish border Acanthodactylus felicis Arnold, 1980 Type locality: Hadhramaut, southern Yemen Locality and collection number: ZFMK 70981, locality no.10, 20.11.1998 Distribution: southern Yemen and Dhofar The only character given by Arnold (1986) to distinguish A. felicis from A. yemenicus is the number of ventral scales across the belly (usually 8 in felicis and usually 10 in yemenicus). The specimen in question has 8 scales in longest row across the belly. Acanthodactylus felicis occupies sandy substrates (Arnold 1980). We found this species on a harder substrate in association with Uromastyx benti, Pseudotrapelus sinaitus, Pristurus carteri, Mabuya spec. and Echis cf. pyramidum. Acanthodactylus schmidti Haas, 1957 Type locality: Dharan, Saudi Arabia Locality and collection number: ZFMK 70980, locality no.20, 30.11.1998 375 Distribution: Arabia north to Jordan; absent from the western and southwestern parts of Arabia and from the Batina (Oman) Mesalina adramitana (Boulenger, 1917) Type locality: Hadhramaut, Yemen Localities and collection numbers: ZFMK 70884, locality no.5, 17.11.1998; ZFMK 70951, locality no.12, 26.11.1998 Distribution: southern and southeastern Arabia Mesalina cf. ayunensis Arnold, 1980 Type locality: Ayun, Dhofar, Oman Locality and collection number: ZFMK 66566, locality no.32 Distribution: Dhofar, Sultanate of Oman Omanosaura cyanura (Arnold, 1972) Type locality: Wadi Shawkah (25°06'N 56°03'E), Trucial States (= United Arab Emirates) Localities and collection numbers: ZFMK 56397, locality no.38; ZFMK 56398, locality no.39 Distribution: northern Oman (Musandam, Jebel Akhdar), eastern UAE Omanosaura jayakari (Boulenger, 1887) Type locality: Muscat, Oman Localities and collection numbers: ZFMK 63696-697, locality no.40; sight record, locality no.29, 04.12.1998 Distribution: northern Oman and UAE Scincidae Chalcides ocellatus (Forskal, 1775) ioypie locality: Egypt Locality: sight record, locality no.11, 21.11.1998 Distribution: North Africa, coastal area of Arabia, South-west Asia to Pakistan Mabuya spec. Locality: sight record, locality no.10, 20.11.1998 Scincus mitranus Anderson, 1871 Type locality: Arabia Localities: sight record, locality no.5, 17.11.1998; still alive, locality no.18, 28.11.1998 Distribution: Arabia with exception of the western and northwestern parts Colubridae Coluber rhodorachis (Jan, 1865) iype hocality : Persia Localities and collection numbers: ZFMK 70930, locality no.7, 19.11.1998; ZFMK 70929, locality no.21, 30.11.1998; sight records, localities no.29 & 30, 04.12.1998 Distribution: North-east Africa to northern India, western and southern Arabia 87/6 Lytorhynchus diadema (Dumeril, Bibron & Dumeril, 1854) Type locality: Algeria Locality and collection number: ZFMK 70931, locality no.41 Distribution: North Africa and Arabia Viperidae Echis carinatus sochureki Stemmler, 1969 Type locality: Ban-Kushdil-Khan, Pishin, Pakistan Locality: still alive, locality no.1, 12.11.1998 Distribution: Arabia, the UAE and northern Oman Echis coloratus Günther, 1878 Type locality: Jebel Sharr [country ?] Locality: still alive, locality no.5, 16.11.1998 Distribution: Egypt, Sinai, Israel, Jordan, western rocky mountain areas of Arabia, south to the Hadramaut; northern and central Oman Echis cf. pyramidum (Geoffroy St. Hilaire, 1827) Lypellocalaty = Beypı Locality: still alive, locality no.10, 20.11.1998 Distribution: southwestern Arabia, Yemen and Oman Zoogeographical discussion Because of its position at the border between the Afrotropical and Palearctic biogeographic regions, southern Arabia is of particular interest. Arabia is mainly classified as part of the Saharo-Sindian Subregion of the Palearctic Region. Only the southwest (Yemen and southern Oman) is part of the Afrotropical subregion of the Paleotropical region. Arabia is separated from Somalia in eastern Africa only by the Gulf of Aden. Despite the narrowness of the water barrier between Arabia and the African mainland it marks a very abrupt faunal change. This is especially surprising because of the supposed land connection between Arabia and Somalia during the Pleistocene eustatic sea level changes (Arnold 1987, Joger 1987, Lanza 1988). For example, only 25 reptilian species of approximately 197 non-marine species in Somalia can be found in Arabia, of which only 14 species inhabit the Sultanate of Oman (Bitis arietans, Chalcides ocellatus, Coluber rhodorachis, Echis cf. pyramidum, Hemidactylus yerburii, H. turcicus parkeri, Leptotyphlops macrorhynchus, Mabuya brevicollis, Naja haje, Pristurus rupestris, Ptyodactylus hasselquistii, Psammophis schokari, Ramphotyphlops braminus and Telescopus dhara). Of the 29 amphibian species of Somalia, none is known to occur in Oman. Besides the above mentioned genera, Somalia and Oman share species of the genera Atractaspis, Acanthocercus, Bufo, Chamaeleo, Eryx, Mesalina, Spalerosophis, Tropiocolotes, Uromastyx and Varanus. Of those, Acanthocercus, Atractaspis, Bitis, Chamaeleo and Mabuya are of Afrotropical origin. ST Table |: List of localities 2SS SUN Vicinity of Seeb Airport Echis carinatus sochureki, Bunopus Seve Oe spatalurus hajarensis 2 Pea IN Ras al Hadd sea-turtle protecto- 59249 E rate 23° 34 N Vicinity of Al-Khoud / Univer- 8.08 E sity Pristurus minimus, P. rupestris, P.car- teri Ptyodactylus hasselquistii, Pristurus rupestris, Pseudotrapelus sinaitus Hemidactylus flaviviridis, Cyrtopodion scaber Bunopus tuberculatus, Stenodactylus leptocosymbotes Al-Khoud Echis coloratus, Ptyodactylus hassel- quistil, Hemidactylus homoeolepis, H. yerburyil, Pristurus rupestris, P. carte- ri, Mesalina adramitana, Uromastyx . MO aN 67 km from Haima in direction of 56° 04'E Marmul 5) 17° 40 N 27 km fromThumrait in direction 54° 09 E aegyptia microlepis, Tropiocolotes scortecci, Scincus mitranus of Marmul 16° 58 N West of Salalah Pristurus rupestris, P. carteri, Pseudo- 53° Dis trapelus sinaitus 4 km before Mushsayl Coluber rhodorachis Wadi Hinna, pool below the re- Acanthocercus adramitanus, Pristurus sting place rupestris 1 km from Wadi Hinna in directi- | Pseudotrapelus sinaitus on of Mirbat 3 km west of Mirbat Uromastyx benti, Acanthodactylus feli- cis, Mabuya spec., Echis cf. pyrami- dum, Pseudotrapelus sinaitus, Pristurus carteri 17°00 N 54°42°E near Mirbat Tropiocolotes scorteccii City of Mirbat Chalcides ocellatus Northern Masirah Uromastyx thomasi thomasi - habitat Pristurus carteri, P. minimus, Steno- dactylus leptocosymbotes, Mesalina adramitana Masirah Hotel Hemidactylus turcicus parkeri 207 DAN Central Masirah Pristurus carteri, P. minimus 58° 47 E 20° 30 N Wadi Blad Central Masirah 58° 54E BED BSW | So ls i 22 SS N en Bufo arabicus, Acanthodactylus boskia- 38 ua nus 18 DRERSSN Wahiba Sands Scincus mitranus, Stenodactylus arabi- 58°45E cus | 58° 47 E of Muscat, DZ SON yon 30E Bufo dhufarensis Wahiba Sands Stenodactylus doriae, S. leptocosymbo- tes, Bunopus tuberculatus, Acantho- dactylus schmidti, Pristurus minimus Phrynocephalus arabicus, Stenodacty- lus doriae approximately 100 km inland of Muscat Sur — Muscat Coluber rhodorachis 106 km to Muscat Barka, in direction of Al Rustaq 23° 32 N 57° 51E 23° 29N 6 km NE of Afi Bunopus spatalurus hajarensis SOE 4 dunes behind Al-Naseem park Acanthodactylus blanfordi SSH NDIE Echis carinatus sochureki DD N (99) i) i) Sultan Qaboos University Cyrtopodion scaber, Bunopus spata- Campus lurus hajarensis, Hemidactylus flavi- viridis 07 Muscat ee linenidacis av und EN 27 National Museum of Natural Pristurus rupestris History - Botanical Garden Seeks Far rachis Wadi Bani Awf near Bimah Ptyodactylus hasselquistii, Hemidac- tylus persicus, Asaccus platyrhynchus, Bufo arabicus, Coluber rhodorachis, Pristurus gallagheri eS) [89] N Oe IN i) >) \o (oe) ON Nn Wadi Bani Awf Asaccus platyrhynchus, Ptyodactylus hasselquistii, Hemidactylus persicus, Bufo arabicus Dhofar between Duqm and Pristurus minimus Surayr Dhofar/ Jebel Samhan Khutwa Pristurus rupestris Umm az Zumaim Stenodactylus arabicus tylus homoeolepis Pristurus carteri, Phrynocephalus ara- bicus, Stenodactylus arabicus Phrynocephalus arabicus Wadi near Al Hammah Omanosaura cyanura between Barka and Al Rustaq Lytorhynchus diadema Wadi Bani Khalid / Wahiba Pseudotrapelus sinaitus Sands In the east, southern Arabia is separated from Iran by the Gulf of Oman and the Arabian Gulf. Of approximately 168 taxa of non-marine reptiles found in Iran (Anderson 1974, Latifi 1991), 24 taxa also occur in Oman: Acanthodactylus blanfordii, Acanthodactylus schmidti, Ablepharus pannonicus, Bunopus tuberculatus, Chalcides ocellatus, Calotes versicolor (presumably introduced), Coluber rhodorachis, Diplometopon zarudnyi, Echis carinatus sochureki, Eryx jayakari, Hemidactylus flaviviridis, Hemidactylus persicus, Leptotyphlops macrorhynchus, Mabuya aurata, Malpolon moilensis, Phrynocephalus maculatus, Pristurus rupestris, Ptyodactylus hasselquistii, Psammophis schokari, KHIR] BR Jw] WwW Jw eS) w POF WW Joo Oo ice) DD I | CO Jo], on ON Nn [BY] IND u >) 3719 Pseudocerastes persicus, Spalerosophis diadema, Stenodactylus doriae, Uromastyx aegyptia microlepis and Varanus griseus. In addition, on a generic level Oman and Iran share species of the genera Asaccus, Lytorhynchus, Tropiocolotes, Mesalina and Scincus. In general, northern Oman has strong affinities with Iran and southern Pakistan (Baluchistan) while southern Oman (Dhofar) has affinities with the Horn of Africa. Acknowledgements We would like to thank Mr. Ali bin Amer al Kiyumi (Director-General of Nature Reserves) and Mr. Salim M. Al-Saady (Director of Research), Muscat, Oman for their hospitality and help during our field trip and for all required permits. Furthermore we want to thank Dr. A. S. Gardner and his wife Dr. K. Oakley (Sultan Qaboos University, Al-Khoud, Muscat/ Oman) for their hospitality and for the support given. We are grateful to Dr. D. Dix (Arlington/ USA) and Dr. A.S. Gardner (Muscat/ Oman) for their linguistic help and to Prof. Dr. W. Böhme and Dr. I. Ineich, Paris, for comments on the manuscript. Zusammenfassung Zur Herpetofauna des Sultanats von Oman, mit Hinweisen zur Beziehung zwischen afrotropischer und saharosindischer Fauna Heute sind insgesamt 90 Arten von Amphibien und Reptilien aus dem Gebiet des Sultanats von Oman bekannt. Während einer Forschungsreise in das Sultanat konnten 272 Exemplare von Reptilien und Amphibien aus 32 Arten gesammelt werden. Sechs weitere Arten wurden nachgewiesen aber nicht gesammelt. Weiteres Material aus der Sammlung des ZFMK wurde einbezogen, so daß insgesamt 42 Arten von Amphibien und Reptilien in einer kommentierten Liste aufgeführt sind; es folgt eine Liste der besuchten Lokalitäten mit ihren entsprechenden Herpetozönosen. Die Beziehungen der omanischen Herpetofauna zur paläarktischen Herpetofauna des Irans und der afrotropischen Herpetofauna Somalias werden aufgezeigt. References Anderson, S. (1974): Preliminary Key to the Turtles, Lizards and Amphis- baenians of Iran. — Fieldiana Zoology 65(4): 27-44. Arnold, E. (1975): Little-known Geckoes (Reptilia: Gekkonidae) from Arabia with description of two new species from the Sultanate of Oman. — J. Oman Stud. Spec. Rep. 1: 81-110. Arnold, E. (1980): The Reptiles and Amphibians of Dhofar, Southern Arabia. — J2O@man Stud. Spec. Rep. 2: 273-332. Arnold, E. (1986): A Key and Annotated Check List to the Lizards and Amphisbaenians of Arabia. — Fauna of Saudi Arabia 8: 385-435. Arnold, E. (1987): Zoogeography of the Reptiles and Amphibians of Arabia. — Proc. Symp. Fauna Zoogeogr. Middle East: 246-256. 380 Arnold, E. & M. Gallagher (1975): Reptiles and Amphibians from the Mountains of Northern Oman. — J. Oman Stud. Spec. Rep. 1: 59-80. Gardner, A.S. (1994): Pristurus gallagheri, an arboreal semaphore gecko from Oman, with notes on its tail flagging behaviour. — Dactylus 2: 126-129. Joger, U. (1987): An Interpretation of Reptile Zoogeography in Arabia, with Special Reference to Arabian Herpetofaunal Relations with Africa. — Proc. Symp. Fauna Zoogeogr. Middle East: 257-271. Lanza, B. (1988): Amphibians and reptiles of the Somali Democratic Republic: check list and biogeography. — Biogeographia 14: 407-465. Latifi, M. (1991): The snakes of Iran. — Society for the Study of Amphibians and Reptiles 159 pp. Leviton, A.,S. Anderson, K. Adler & A. Minton (1992): Handbookte Middle East Amphibians and Reptiles. — Society for the Study of Amphibians and Reptiles 252 pp. Meinig, H. & H. Kessler (1998): Herpetologische Beobachtungen im Rahmen einer Nationalparksplanung: Barr al Hikman und Masirah Island, Sultanat von Oman (Amphibia; Reptilia). — Faun. Abh. 21 (Suppl.): 89-97. Salvador, A. (1982): A Revision of the Lizards of the Genus Acanthodactylus (Sauria: Lacertidae). - Bonn. zool. Monogr. 16, 167 pp. Seuter, H., T. Kowalski, U. Prokoph & H. Zilser 1998) ersinachwers von Uromastyx benti (Anderson, 1894) fiir den Oman (Provinz Dhofar). — Herpetofauna 20(114): 22-23. Authors’ addresses: Dipl. Biol. Thomas Wilms, Zoologisches Forschungsinstitut und Museum A. Koenig, Sektion Herpetologie, Adenauerallee 160, 53113 Bonn, Germany Felix Hulbert, Worthstr. 29, 65343 Eltville, Germany 38 | The evolution of Microtia Freudenthal, 1976 (Mammalia, Rodentia), an endemic genus from the neogene of the Gargano island, southern Italy Virginie Viniine no) ania Abstract: The Gargano area, considered as a Late Neogene paleoisland, has yielded many micromammal remains from numerous karstic fissures. All described species show extraordinary morphological peculiarities, due to island evolution. Among the seven murid rodent genera recognized, Microtia is the most abundant and is represented by three main lineages that differ in size. These three lineages were evidenced through a study of molar size and morphology, the size evolution of the first lower molar being the basis for a biostratigraphic time scale of the Gargano fissures. Isolated endemic rodents often evolve towards gigantism, but many recent studies have pointed out the role of interspecific competitive interactions on community size structure, which can also greatly influence species size evolution. The occurrence of numerous karstic localities of different ages in the Gargano paleoisland, and the abundance of Microtia in each of these localities, provided a unique opportunity to investigate size structure of the Microtia community and its evolution. Microtia body size was inferred from the lower incisor size which is known to be highly correlated to body size among extant rodent species. The three lineages did not follow the same evolutionary size trend: two of them evolved towards gigantism, while the third tended towards a smaller size. In addition, I characterized the evolution of the lower incisor curvature and concluded that Microtia was characterized by a specialization for burrowing which was accompanied by both an increase or a decrease in size. Finally, the evolutionary size changes among the three sympatric Microtia lineages allowed the minimization of competition between them. Key words: isolated rodent, incisor, Gargano, Neogene, size evolution Introduction The Gargano fossil localities and their mammalian fauna The Gargano peninsula is located in southern Italy, in Foggia province. In addition to the geological evidence, the highly differentiated and impoverished fauna of the Gargano clearly indicates that the Gargano is a paleoisland that remained isolated during several million years (fig.1). The Gargano vertebrate remains were discovered by M. Freudenthal, in the early 1970s, where the fossils were found in numerous karstic localities (fig.2) that span an imprecise time interval between the Late Miocene and the Early Pliocene (Freudenthal 1976, Abbazzi et al. 1996). The Gargano mammalian fauna is impoverished, with only some representatives of four orders: the insectivores, rodents, lagomorphs, and artiodactyls. Carnivores, perissodactyls and proboscidians are missing. Most of the taxa described to date show extraordinary morphological peculiarities, and, in general, a remarkable tendency towards the gigantism that has been associated with island evolution Rheinwald, G., ed.: Isolated Vertebrate Communities in the Tropics Proc. 4th Int. Symp., Bonn Bonn. zool. Monogr. 46, 2000 E | Emerged areas —— Present day shorelines Fig.l: Paleogeographic reconstruction of the Gargano area during the late Neogene Mite Gargano peninsula is a paleoisland that remained isolated during several After De Giuli et al., 1986 million years. (modified, after De Giullesar 1986). (Freudenthal 1972, Ballmann 1973, 1976, Willemsen 1983, Leinders 1984). The Gargano fauna all became extinct after the establishment of a connection of the area to the Italian mainland in the early Pleistocene. Within these endemic forms, two taxa of mammals show tremendous morpho- logical specializations. The first one, Deinogalerix, is a giant insectivore, but due to its large size it probably included small mammals in its diet, and was thus probably closer to a scavenger in its habit (Freudenthal 1972). This taxon is probably the largest insectivore ever found, with a skull of approximately 20 cm in length. The second taxon, Hoplitomeryx, belongs to the Ruminantia and was so differentiated that it is considered as a familiy-level endemic to the Miocene of southern Italy (Leinders 1984). The five horns on the skull in this taxon is a very good example of great morphological innovation. BEE Miocene ''Terre rosse"' L__] Late Plio./Early Pleist. siliciclastics After Abbazzi et al., 1996 Fig.2: The Gargano fossils were discovered in numerous karstic fissures that span an imprecise time interval between the late Miocene and the early Pliocene (modified, after Abbazzi et al. 1996). 383 3 mm After Abbazzi et al., 1993 and Freudenthal, 1976 Fig.3: The evolution of the M/1 from Microtia. This evolution is characterized by an increase in size and by the development of additional lamellae at the anterior end of the M/1. (modified, from Freudenthal 1976 and Abbazzi et al. 1993). Another very peculiar taxon from the Gargano fauna is the rodent genus Microtia, which is the most abundant rodent among the seven rodent species described to date, and is represented by three main lineages that differ in size. Microtia molars have been previously described (fig.3), and the evolution of the main middle-sized lineage is the basis for a biostratigraphic time scale of the Gargano fissures (Abbazzi et al. 1993, Freudenthal 1976). The evolution of this lineage is characterized by an increase in molar size and by the development of additional lamellae at the anterior end of the first lower molar, M/1, and at the back of the third upper molar, M3/ (Freudenthal 1976). In a recent study of the Microtia skull (Parra et al. 1999), we proposed that some of its morphological characteristics indicated an adaptation for a burrowing life, which is unique for a murine rodent. Some of these characteristics are, for example, an elongated snout, highly proodont upper and lower incisors, and the presence of a huge bony apophysis on the squamosal bone for the insertion of the temporal muscle (fig.4). Here, I propose to review the morphological and size evolution of the lower incisor of this very peculiar rodent genus. The information yielded from incisors Rodent species are characterized by one upper and lower incisor pair with continuous growth that display a great variation in size, shape and morphology. However, the information contained in these incisors is usually not considered in paleontological studies. The main reason for using incisors 1s that microfossil layers contain numerous incisors, isolated, or still connected to the upper or lower jaws. Secondly, both upper and lower incisors are accurate estimators of body weight among rodents (Parra & Jaeger 1998). The following equation describes the relationship between the antero-posterior diameter of the lower incisor (AP, in mm) and body weight (W, in g) for species of extant rodents: log = Ooo OBA) se Dole ere. Ih] 384 Strong postorbital apophysis on the squamosal Elongated snout es Proodont incisor Enlarged zygomatic arch Large zygomatic plate Fig.4: The skull of with a strong masseteric tubercule Microtia, in lateral view. An elongated Marked masseteric ridges snout, highly pro- Elongated symphysis odont upper and lower ae incisors and the zz Er presence of a huge bony apophysis on the ie squamosal are some of its morphological characteristics (modi- Prominent symphysal fied, from Parra et al. eminence lcm in press) This relation is highly significant (r = 0.93, p < 0.001), and, applying the hypothesis that it has not changed through time, we can apply it to fossil rodent species. The evolution of rodent species on islands The general pattern of insular evolution amongst mammals indicates that small species become larger on islands, and large species become smaller (Foster 1964, van Valen 1973). Several hypotheses have been proposed to explain the increase in the size of isolated rodent species. The most important ones are: (1) Resource limitation on islands favors K-strategies among rodents, and consequently an increase in their body size; (2) Decreased predation (Heaney 1978), or the absence of predators on islands (Valverde 1964) allows rodent species to reach a larger size. Under predation pressure on the mainland rodent species stay small in order to escape from predators; (3) The existence of a physiological optimal body size that can be reached by isolated mammals (Brown et al. 1993, Damuth 1993); (4) Finally, the decreased number of competing species on islands allows isolated species to increase their range of exploitable resources, and consequently their body size (Lomolino 1985). I will test each of these hypotheses using the Microtia assemblage and its evolution. In addition, I will indicate adaptative changes in behaviour for Microtia. 385 The case of Microtia The evolution of size The hypothesis of resource limitation is difficult to assess in the case of the Gargano fauna, because of insufficient data on the paleoenvironment of this area. Fig.5 provides a schematic representation of the evolution of the three Microtia lineages. It is based on almost 400 specimens that were collected in 11 different fissures. Complete data measurements are given elsewhere (Parra & Jaeger, manu- script). The size character used is the antero-posterior diameter of the lower incisor, which is used as a body size estimator (Eq. 1). The first observation is that the evolutionary size trends are different for the three lineages: the two largest lineages evolving towards larger size, the smallest one tending towards smaller size. The evolution towards a larger size is in accordance with the general observations made on other isolated rodents (Foster 1964, Sondaar 1977, Case 1978, Heaney 1978, Lawlor 1982, Adler & Levins 1994), but it is difficult to explain the size trend observed for the smallest lineage. Fissures (time) Fig.5: Schematic repre- sentation of the evolu- (MOMAny Give ieemGls observed for the three Microtia lineages, based on the lower incisor size (modified from Parra & Jaeger, manuscript). Grey areas represent the total range and lines AP (mm indicate the mean values of the lower incisor size (antero-posterior dia- meter, AP, in mm). 0.95 1533 1273 215 255 2.95 Two approaches have led to the hypothesis of an existence of an optimal body size among mammalian species. The first one was based on the relation between body mass and population density among mammals (Damuth 1993). Here it is proposed that there is an optimal body size for intermediate body size species, around | kg, these species being more efficient at energy acquisition. The second approach was described by Brown et al. (1993). The species’ body size is affected by both the rate of acquisition of environmental resources and the rate of conversion of these resources for reproduction. This last model predicts that there is an optimal body size for small body species of around 100 g. In these two studies it is claimed that the theory of an optimal body size provides an explanation for the island-rule, i.e. evolution towards a larger size of small mammals and a smaller size of large species (Brown et al. 1993, Damuth 1993). However, the two values proposed are very different — a difference of an order of magnitude. The available data on isolated rodent species do not favour either one of these hypotheses. However, a tabulation of all known isolated species body sizes may be helpful for further clarification of this point. Importantly, the results 386 obtained for Microtia do not provide any evidence for the existence of an optimal body size, because of the different directions of evolutionary size trends observed for each Microtia lineage. The hypothesis of reduced predation does not find support here, because many potential predators for Microtia were discovered in the Gargano fossil remains. One of these predators is the giant insectivore, Deinogalerix, which probably included small mammals in its diet (Freudenthal 1972). But the most interesting one is a giant owl that was likely to have preyed heavily on Microtia. This owl genus, Tyto, like Microtia, is also represented by three lineages that differ in size (Ballmann 1976). The most interesting observation that can be made concerning the size evolution of the two genera Microtia and Tyto is that their evolutionary size trends are very similar, the two largest lineages evolving towards a larger size while the smallest lineage tends towards a smaller size. This could suggest predator/prey coevolution, or, at least, that owl size was tracking rodent size. The last hypothesis proposed was reduced competition between species on islands. In the case of Microtia, however, I am led to suggest that competition between the different species was still strong. Indeed we can observe an increase in the size differences between species through time, which can be explained by the existence of competition between them. Interspecific competition is therefore the only hypothesis that can explain the different size trends observed for the Microtia lineages. Changes in behaviour The evolution of the shape of the incisor can be indicative of some adaptive changes in Microtia. As a first approximation, I used the radius of incisor curvature to characterize the lower incisor shape. The evolutionary patterns observed for the lower incisor size and shape of Microtia are always characterized by an increase in the radius of incisor curvature, which can be accompanied by an increase or a decrease in size. Procumbent incisors are one of the morphological characteristics of burrowing rodents (Dubost 1968, Gasc et al. 1985, Lessa 1990). Hence this result can be interpreted as an improvement of the burrowing adaptation, whether with an evolution towards nanism or gigantism. The evolution of body weight In addition, I used the relationship established for extant rodent species between the size of the lower incisor and body weight (Eq. 1). Body weight was estimated for the main middle-sized lineage at its oldest and youngest fissures of occurrence. If Microtia weighed a little more than 40 g in the oldest fissure, there is a huge increase in weight through time, since it reaches more than 140 g in the youngest fissure, an increase of more than 230%. Such an increase in weight implies great changes in physiology accompanying the increase in size. Such a large increase in body weight might be related, for example, to a reduction in reproductive output, population density and energetic expenditure per unit weight (Peters 1983, Adler & Levins 1994), all of which would enhance the probability of extinction of the Microtia fauna on reconnection with the mainland in the early Pleistocene. 387 Conclusion We proposed that the evolutionary size patterns observed for Microtia were mainly the result of competition between species, which is in accordance with previous results from some mainland rodent communities (Dayan et al. 1994, Parra et al., 1999). However, other factors, such as predation pressure or evolution towards an optimal body size, do not appear to be important in the particular case of Microtia. In conclusion, the evolution in the rodent genus Microtia brought about changes not just in size but, with the improvement of the adaptation to a burrowing life, also in behaviour. The large increase in body weight, concurrent with body size change, was also likely to have led to substantial changes in physiology. Acknowledgments The material used in this study was collected by M. Freudenthal and is deposited in the National Museum of Natural History, Leiden, The Netherlands. I am very grateful to M. Freudenthal for providing me with his material, and for helpful discussions on the Gargano fauna during the last three years. Thanks to A. Gonzalez for his help in improving this manuscript. I am very grateful to B. Patterson for his helpful suggestions and comments on a previous version of this manuscript. This 1s publication ISEM no 99-010. References az Er Masini é& D. Torre (1993): Evolutionary patterns in the first lower molar of the endemic murid Microtia. — Quaternary Int. 19: 63-70. bar owe MM. Benvenuti, G. Boschian, S. Dominici; F. Masini, em Mezzrabotta, E Piccini, L- Rook, G. Vallerı &D. Torre (1996): Revision of the Neogene and Pleistocene of the Gargano region (Apulia, Italy). The marine and continental successions and the mammal faunal assemblages in an area between Apricena and Poggio Imperiale (Foggia). — Memorie della Societa Geologica Italiana 51: 383-402. dier GH. & R. Levins (1994): The Island Syndrome in Rodent Populations. — Quatern. rev. Biol. 69: 473-490. Ballmann, P. (1973): Fossile Vögel aus dem Neogen der Halbinsel Gargano (Italien). — Scripta Geol. 17: 1-75. Ballmann, P. (1976): Fossile Vögel aus dem Neogen der Halbinsel Gargano (kalten), 2. Teil. - Scripta Geol. 38: 1-39. Bow VE, P A. Marquet & M.L. Taper (1993): Evolution of Body Size: Consequences of an energetic definition of fitness. - Amer. Nat. 142: 573- 584. Case, T.J. (1978): A general explanation for insular body size trends in terrestrial vertebrates. — Ecology 59: 1-18. Damuth, J. (1993): Cope's rule, the island rule and the scaling of mammalian population density. — Nature 365: 748-750. 388 Dayan, T., & D. Simberloff (1994): Morphological relationships among coexisting heteromyids: an incisive dental character. — Amer. Nat. 143: 462-477. Dubost, G. (1968): Les mammiferes souterrains. — Rev. Ecol. Biol. Sol V: 99- 183: De Giuli, €, E Masini & D- Torre (1986), Endemismrandepie- chronological reconstructions: the Gargano case history. — Boll. Soc. Paleontol. Itale25: 269-2719: Foster, J.B. (1964): Evolution of mammals on islands. — Nature 202: 234-235. Freudenthal, M. (1972): Deinogalerix koenigswaldi nov. gen., nov. spec., a giant insectivore from the Neogene of Italy. — Scripta Geol. 14: 1-11. Freudenthal, M. (1976): Rodent stratigraphy of some Miocene fissure fillings in Gargano (prov. Foggia, Italy). — Scripta Geol. 37: 1-20. Gasc, J.-P., S: Renous, A. Casinos, E. Laville & IB 002 935% Comparison of diverse digging patterns in some small mammals. — pp. 35-38 in: Vertebrate Morphology (eds. Duncker, H.-R. & G. Fleischer). — Gustav Fisher Verlag, Stuttgart, New York. Heaney, L.R. (1978): Island area and body size of insular mammals: evidence from the tri-colored squirrel (Callosciurus prevosti) of southeast Asia. — Evolution 32: 29-44. Lawlor, T.E. (1982): The evolution of body size in mammals: evidence from insular populations in Mexico. — Amer. Nat. 119: 54-72. Leinders, J. (1984): Hoplitomerycidae fam. nov. (Ruminantia, Mammalia) from Neogene fissure fillings in Gargano, Italy. Part 1. - Scripta Geol. 70: 1-51. Lessa, E.P. (1990): Morphological Evolution of Subterranean Mammals: Integrating Structural, Functional, and Ecological Perspectives. — pp. 211-230 in: Evolution of Subterranean Mammals at the Organismal and Molecular level (eds. Nevo, E. & O.A. Reig). — Wiley-Liss, New-York. Lomolino, M.V. (1985): Body size of mammals on islands: the island rule reexamined. — Amer. Nat. 125: 310-316. Parra, V. & J.-J. Jaeger (1998): Estimation de la taille et du poids corporel chez les rongeurs (Rodentia, Mammalia) a partir de la taille des incisives. —C. R. Acad. Scien; Serie Ma; Scien: Herre Planetes 320279235: Parra, V.,M. Loreau & J.-J. Jaeger (1999): Incisor size and community structure in rodents: Two tests of the role of competition. — Acta Oecol. 20: 93- 101. Parra, V., J.-J. Jaeger & H. Bocherens (in press): The skullotVienema (Freudenthal, 1976), an extinct burrowing murine rodent genus of the late Neogene of Gargano, South Italy. — Lethaia. Peters, R.H. (1983): The ecological implications of body size. — Cambridge University Press, London. Sondaar, P.Y. (1977): Insularity and its effect on mammal evolution. — pp. 671- 705 in: Major Patterns in vertebrate Evolution (eds. Hecht, M.K. & P.C. Goody). — Plenum Press, New-York, London. 389 Mamvende. sd ear.) (1964): Remarques” sur la structure et l’evolution des communautes de vertebres terrestres I. Structure d'une communauté II. Rapports entre prédateurs et proies. — Terre et la Vie 2: 121-154. van Valen, L.M. (1973): A new evolutionary law. — Evolutionary Theory 1: 1- 30. Willemsen, G.F. (1983): Paralutra garganensis sp. nov. (Mustelidae, Lutri- nae), a new otter from the Miocene of Gargano, Italy. — Scripta Geol. 72: 1-8. Author’s address: Dr. Virginie Millien-Parra, Institut des Sciences de l’Evolution, UMR 5554, Université Montpellier II, cc 64 place Eugene Bataillon, 34095 Montpellier cedex 5, France, e-mail: parra@isem.univ-montp2.fr 390 ABSTRACTS Arboreal Squamates in Tropical Forests of Vietnam: Species Richness, Natural History and Conservation N.B. Ana nj eva & N.L.Orlov, Department of Herpetology, Zoological Institute, Russian Academy of Sciences, St.Petersburg 199034, Russia Herpetofauna of the tropical montane forests in Tonkin and Annam includes about 90 species of lizards and 140 species of snakes. For squamate assemblages of tropical forests a considerable percentage of arboreal forms is typical. In the fauna of Vietnam about 50 lizard species and 70 snake species demonstrate the adaptations to the arboreal mode of life developed to different extent. It is well known that the spatial structure of the herpetological assemblages of tropical forests is very complicated. Squamate reptiles inhabit different altitudes in the poly-dominant primary forests, i.e different types of vegetation from small bushes and small woods to the canopy of the highest belt of this forest. Most squamates present in the fauna of Vietnam are species with arboreal adaptations, belonging to the families of Agamidae, Gekkonidae, Dibamidae, Scincidae, Varanidae, snakes of the families of Natricidae, Colubridae, Lampophiidae, Elapidae, Viperidae. It is necessary to remark the seasonal changes of the type of belt occupied by species and its association with the physiological cycle of the animals. It can be explained by examples of arboreal agamids (Acanthosaura crucigera) from the southern and central Annam mountains. During the dry and cool winter season it goes down to the ground demonstrating terrestrial mode of life and even "hibernation" under the trunks of the falling trees. The females of the arboreal Acanthosaura capra in the same regions of the tropical forests go down to the ground to lay eggs. The complicated spatial structure of primary tropical forests with five layers of vegetation (ferns: 0-2 m; bush: 2-8m; undercrown: 8-15; canopy: 20-34; overcrown: 40-50 m and above) is optimal for high species richness and coexistence of closely related species with similar ecological requirements. The number of sympatric species is rather high and reaches 8 species of Elaphe, 5 species of Boia, 10 species of arboreal agamids. The development of different morphological adaptations to arboreal life in different squamates and the origin of different groups of arboreal lizards and snakes of Vietnam is discussed. Does frugivore diversity influence seed dispersal and seedling establishment of fruiting plants? B. Bleher & K. Böhning-Gaese, Institut für Biologie II, RWTH Aachen, Kopernikusstr. 16, 52074 Aachen, Germany, tel. +(049)-0241-807517, fax +(049)-0241-8888133, e- mail bleher@bio2.rwth-aachen.de Seed dispersal plays an important role in the regeneration and conservation of ecosystems. To address the question whether a reduced frugivore diversity due to anthropogenic (e.g. habitat fragmentation), ecological or historical processes 39] affects seed dispersal and seedling establishment of fruiting plants, we compared seed dispersal in two tree species of the genus Commiphora in two areas with different frugivore diversity, South Africa and Madagascar. South Africa has a high frugivore diversity comparable to other continents, Madagascar is strikingly depauperate. In both areas, tree visitation rates, seed handling and crop removal rates were quantified by tree observations. Fruit production and dispersal rates were quantified using fruit traps. Seedlings were mapped along transects, the spatial distribution of adult trees was measured using the T?-method. Trees in South Africa were visited by 13 bird species with Crowned Hornbills (Tockus alboterminatus) being the main visitors (29,9%) and dispersers (43,1%). In Madagascar, trees were visited by four bird species with Lesser Vasa Parrot (Coracopsis nigra) as main visitor (57,5%) and disperser (100%). Total dispersal rates in South Africa were considerably higher than in Madagascar (66,3% versus 9,0%) which was due to differences in frugivore diversity and especially handling behavior. Whereas in South Africa most visitors dispersed seeds by swallowing them (11 of 13), in Madagascar dispersal only took place when parrots left with a seed in their beak. Correlating with different seed dispersal rates, seedling distributions in South Africa showed a higher density away from the closest Commiphora tree. Furthermore, adult trees in South Africa were uniformly distributed, in Madagascar they showed a clumped distribution. Results show that regional differences in frugivore diversity and especially seed handling behavior have a strong impact on the seed dispersal rates of a plant, on seedling establishment and possibility on the spatial distribution of adult trees. The Ichthyological Section of the ZFMK: Topics of research in biodiversity BenBmisscer i siireynor, A. Nolte, R. Sonnenberg & I. Steinmann, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany Biodiversity initiatives have hitherto tended to focus on terrestrial systems or, when aquatic systems are targeted, on coral reefs or wetlands. Freshwater ecosystems have not attracted much attention but they are under serious threats and are disappearing as a result of human induced impacts. Many poor communities take their benefit from the high economic value of freshwater biodiversity. Therefore, mechanisms forming the species diversity in freshwater ecosystems are the main research topics in the fish section of the ZFMK. At the moment, the ZFMK has six projects on freshwater biodiversity. They are: — Biodiversity and perspectives for sustainable utilization of fish resources in mountain regions of Central Vietnam. — Changes of distribution and ecological adaptation of fish species in the massively disturbed ecosystem of the river Rhine. — Molecular phylogeny of Cichlids in the coastal drainages of Tanzania and Kenya. — Patterns of ecology and intra-fluvial migrations of the European Eel Anguilla anguilla. 392 — Phylogeny and distribution of rainforest restricted Cyprinodontiforms and Cichlids in West- and Central Africa. — Postglacial migration patterns and speciation of the european cold-stenotherm freshwater fishes in the Cottus gobio-group. Influence of tree species and habitat structure on the canopy beetle fauna in Semliki Forest, Uganda M.Cousin& T. Wa gner, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany The arthropod fauna of 24 trees belonging to Cynometra alexandri (Caesalpiniaceae), Ficus capensis (Moraceae), and Elaeis guineensis (Arecaceae), was collected in Semliki Forest, a lowland rain forest in Uganda, using insecticidal fogging technique. Altogether 51410 arthropods, including 2377 beetles were found, the latter were additionally assigned to 487 morphospecies. Chrysomelidae, Curculionidae, Corylophidae and Staphylinidae were the most abundant taxa. Alpha-diversity of the beetle fauna, calculated by several indices and rarefaction, is very similar on the different tree species. Beta-diversity was expressed by Morisita-Horn-Indices. Faunal overlap among conspecific trees was significantly higher than between different tree species. Specialisation of beetles on the distinct tree species, and the influence of microhabitats on the distribution patterns are discussed. Geckos milk honeydew produced by Cicada in Madagascar M. Follin ge" & W. Böhme+, *Universität Bielefeld, Abt. Morphologie/Systematik, +Museum A. Koenig, Sektion Herpetologie During studies in a West Madagascan dry forest, data on the behaviour and feeding ecology of geckos were systematically collected, mainly from Phelsuma madagascariensis kochi in the rainy season from December 1996 to March 1997. In this time mutual correlations between geckos and cicada have been observed, which are described here for the first time. Daygecko species of the genus Phelsuma and Lygodactylus as well as the nocturnal gecko Homopholis sakalava show a specific cicada milking behaviour in which cicada excrete a drop of honeydew as an interspecific key stimulus reaction. Milking behaviour was also observed in Phelsuma abotti on Nosy Be in northern Madagascar. Similar mutual behaviour patterns are well known between ants and green flies but have not been reported for vertebrates yet. Cicada of the family Flatidae feed on the phloem fluids of different tree species in Madagascar. The insects pierce the bark of the trunk; here the geckos seek for the insects and stimulate the honeydew secretion. During the dry season in August, 1998, the individually known ‘cicada trees' were checked again but neither milking behaviour nor cicadas could be observed during that period (Knogge pers. comm.). Possibly the use of honeydew compensates a seasonal lack of floral nectar or tree gums during the rainy season. 393 An exact analysis of the ingredients of the honeydew and their concentrations is still outstanding. In greenflies, previous investigations on honeydew show a constistence of 90-95% of carbohydrates in its dry substance and 0,2-1,8% nitrogen in form of amino acids and amids. Trophobiotic relationships lead to particular secretion behaviour. For example, Homoptera living associated with ants display an interspecific communication pattern at which the honeydew secretion of the green flies is induced by open jaws and soft stroking movements of the antennae of the ants. According to these observations of key stimuli, the similar gecko-cicada relationship must be regarded as a developmental trophobiotic stage. As these insects usually occur locally in high densities the high sugar concentrations of their excretions may cause severe problems by reducing their mobility and breathing abilities. The reported disposal of their excrements obviously is an advantage for the cicada. It has never been observed that geckos mistook cicada for prey. Morphological age-dimorphism and its eco-ethological consequences illustrated for the White-tailed Hawk (Buteo albicaudatus) ) A. Gamauf, Museum of Natural History Vienna, Department of Vertebrate Zoology — Bird Collection, Burgring 7, A-1014 Vienna, Austria. Tel.+43-1-52177/499; Fax +43-1-5235254; e-mail: anita.gamauf@nhm-wien.ac.at The White-tailed Hawk (Buteo albicaudatus) lives in southern Venezuela in isolated savannas within the tropical rain forest. In the course of a study of the "Surumoni Project”, clear differences in the flight silhouettes of juvenile and adult White-tailed Hawks were noticed. This was confirmed by measurements of studied skins, which revealed significant differences at the hawks’ wings and tails. The wings of juvenile hawks are characterized by their greater length, larger Kipp's distance, deeper notches and longer alulae. In addition, the juvenile hawks have longer and more rounded tails and a lower wing load than the adults. Initial observations in November/December 1997 and 1998 showed that these morphological differences are reflected in the hawks' hunting behavior and habitat selection. Significant differences in the hawks’ hunting behavior, especially regarding 6 distinct hunting modes as well as flight altitude, were also recorded. The lighter juvenile hawks used only 2 flight styles (gliding and soaring, equally frequent) at a lower altitude, whereas adults used 4 different flight styles (gliding, soaring, hovering and flap gliding, in descending order) at a higher altitude. With respect to habitat selection, juveniles tend to prefer the hilly grass savannah, as opposed to the wooded cliffs and plains of the savannah preferred by the adults. “supported by the Federal Ministry for Education and Cultural Affairs (BMUuK) and the Austrian Science Foundation (FWF) 394 Centers of endemic terrestrial vertebrates in Colombia: a methodological approach N.A. Guzman Ballesteros, c/o Museum für Naturkunde Stuttgart, Schloß Rosenstein, 70191 Stuttgart, Germany The distribution of endemic species is increasingly used as a tool for the identification of key sites for nature conservation in a global scale. Additionally, endemic species are important indicators for the evaluation of the resilience and complexity of a regional landscape. More than 450 species of terrestrial vertebrates are endemic to Colombia, another 300 are also common to neighboring countries (Guzman 1997, preliminary data). Although most of these species are well-known since the last century, there are still doubts concerning their taxonomic identity and some details of their population ecology, the risk of extinction and their real distribution. Nevertheless, if we use the existing data on Colombia‘s fauna, it is possible to define 42 areas in northern, western and Andean regions, which show that various centers of endemism exist in Colombia. This investigation intends to supply informations on the zoogeography of Colombia by reviewing all records on endemic terrestrial vertebrates of Colombia, scattered all over the world. In a first step, the informations on amphibians, reptiles, birds and mammals of Colombia and the influence of similar geomorphological, ecological and historical processes on their endemicity are analysed. An „ENDEMIC SPECIES CATALOGUE will be compiled, specifying distribution, density maps and taxonomy. Scientific and photographic informations will be obtained from published accounts, additionally supported by data from museums and universities in Colombia, the USA and Europe. A "BIOGEO- GRAPHICAL ANALYSIS" of the centers of endemism will result in a comparison of their natural conditions as regional unities in Colombia and its vicinity. As a case study, the present situation of habitats in a center of endemism in the Sierra Nevada de Santa Marta will be evaluated and the ecological viability of an endemic population will be analysed (by GIS). This work has been guided since 1995 by the University of Stuttgart and the Museum of Natural History in Stuttgart Mammalian biogeography of the Japanese Archipelago and rodent communities size structure V.Millien-Parra, Institut des Sciences de l‘Evolution, UMR 5554, Université Montpellier II, cc 64 place Eugene Bataillon, 34095 Montpellier cedex 5, France The Japanese archipelago is located at the eastern coast of Asia and encloses more than 3900 islands of widely different areas, ranging from less than one sq. km to more than 230 000 sq. km. All these islands remained isolated from the adjacent Asiatic mainland since the last sea lowering during Quaternary, and the extant mammalian fauna is assumed to originate from Miocene. 395 The distribution pattern of terrestrial mammal species in Japan is in accordance with the first prediction of the MacArthur & Wilson theory, i.e. the diversity on islands is highly related to the island size. However, this study reveals the importance of non-equilibrium effects. At a large scale, the current distribution of mammals in Japan seems to be due to selective post-glacial extinction processes. A large proportion of the Japanese mammals is endemic, and extinctions were not balanced by the colonization of species from the Asiatic mainland. Additionally, we show the importance of inter-islands dispersal processes, in particular from larger islands to smaller ones. Deep marine channels between islands affect as well the proportion of endemic species present on islands as the inter-island colonization. Additionally, we investigated the intraspecific patterns of morphology of the size variations among muridae from Japan, using the size of the lower incisor as an estimate of body size. Different geographical trends can be demonstrated, where the size was mainly affected by a latitudinal gradient as well as by island area. Generally, small mammals tend to evolve gigantism on islands. However, many recent studies reveal the importance of competitive interactions between sympatric species that influence the community size structure. We used three different approaches to test the influence of interspecific competition on murid community structure. We can not present any regular size structure. However, evolution on islands seems to reduce competition between species by minimizing size overlaps between them. These results may indicate that interspecific competition is higher on island communities, when compared to mainland ones, mainly due to limited resources on island environments. The real biodiversity is inside D. Modry*, J.R. Slapeta*, P. Necas# & B. Koudela°, *Department of Parasitology, University of Veterinary and Pharmaceutical Sciences, 61242 Brno, Czech Republic, # Sportovni 11, 10200 Brno, Czech Republic, ° Institute of Parasitology of Czech Academy of Sciences of the Czech Republic, Ceské Budéjovice, Czech Republic Parasites are often suggested to be the most prevalent organısms on earth and parasitism itself is considered to be the most common lifestyle within eukaryotic organisms. Most biological knowledge therefore stems from studying the nonparasitic minority of species and knowledge about parasites of wild living vertebrates is more or less fragmentary. Coccidia (sensu stricto) are usually termed protozoan apicomplexan parasites of the family Eimeriidae. Members of this family parasitise all vertebrate groups and some of them are also known from invertebrates. Due to their high host specificity, coccidia represent a highly diversified group of protozoans. Probably the only estimation of coccidian diversity is that published by Levine in 1965, expecting more than 2,5 x 10° species just within the single genus Eimeria. Although data on host specifity of coccidian species of reptiles are lacking, the specifity on the level of host species or genus can be expected. Our comparison of the diversity of African reptiles and their coccidia leads to the evidence that these hosts lack coccidia (probably similar to the situation in other groups of parasites). 396 The number of species of reptiles inhabiting the African continent, reported by Welch (Herpetology of Africa, 1982) is about 990 (surely deeply underestimate) and so far, there are only 33 coccidia of the family Eimeriidae reported from reptile hosts (Agamidae: 2; Boidae: 1, Chamaeleonidae: 1; Colubridae: 5, Cordylidae: 1; Elapidae: 1; Gekkonidae: 5; Lacertidae: 4; Scincidae: 8; Varanidae: 1; Viperidae: 5). The only encouraging is the fact that 10 of these species were described within the last decade. Coccidian diversity is documented on results of our research on East African members of the family Chamaeleonidae, which revealed 6 new species (3 Eimeria and 3 Isospora) within 8 examined chameleon species. Main reasons for simultaneous research on vertebrates and their parasites are the following: a) speciation of parasites and their hosts 1s approximately simultaneous and data on parasite phylogeny could elucidate the phylogeny of hosts (and vice versa); b) some of coccidian species are pathogenic and can represent a serious threat for programs of captive or semi-captive propagation of endangered vertebrate taxa; c) some of the coccidia of wildlife represent potential pathogens of domestic animals and man. The amphibia of an isolated archipelago: the Eastern Arc forests, Tanzania A. Schi@gtz, Humlehaven 2, Atterup, DK - 4571 Grevinge, Denmark The Eastern Arc Mountains, from Teita Hills, Kenya to Udzungwas Mountains in Tanzania, have a natural vegetation of moist evergreen forest from low altitude up to the limit of forest vegetation at about 2400 m a.s.l. Based on studies of amphibians, the zoogeography of these forests — as a sequence of forest islands in a sea of savanna — is described. Due to a very long isolation the fauna in these forests is unique. In spite of the minute size of the forests, the level of endemism for those species living in the moist forest is 100% when compared with the central forest block in Africa, and very high when compared with the other forests in Eastern and Southern Africa. The fauna of these forests is described, their uniqueness discussed and comparison made with other groups of animals, and with forests on isolated mountains elsewhere in Eastern and Southern Africa. The forests on the Eastern Arc are small and dwindle fast. Conservation of the forest flora and fauna is the most urgent conservation priority in Africa. The Alcolapia Species-Flock (Teleostei: Cichlidae) of Lakes Natron and Magadi, Kenya and Tanzania: Another Piece in the Puzzle of Cichlid Evolution? R. Sonnenberg‘, EL. Seegers**® &R. (Yamamoto > 4 3) Zooloseeh Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany, ** Grenzstrasse 47 b, 46535 Dinslaken, Germany; *** Gemeinschaftspraxis für Laboratoriumsmedizin, Abteilung fiir Molekulargenetik, Brauhausstrasse 4, D-44137 Dortmund, Germany yy Recently a small species flock of tilapiine cichlids (,Soda tilapias‘) was collected from East Africa, including Alcolapia alcalicus (Hilgendorf, 1905), formerly known as Oreochromis a. alcalicus, from Lake Natron, Tanzania and Kenya, and A. grahami (Boulenger, 1912), formerly known as O. a. grahami, from lake Magadi, Kenya. Two further, morphologically highly deviating species were described, Alcolapia ndalalani (Seegers & Tichy, 1999) and A. latilabris (Seegers & Tichy, 1999). Geological studies have shown that Magadi and Natron are remnants of a former paleolake, called ,Orolonga‘, which split 8,000 - 9,000 years bp into the modern lakes. Molecular and fossil data indicate that the ‚Soda tilapias‘ and especially the species from Lake Natron radiated after the separation of the lakes. The different morphotypes resemble species in head morphology and oral dentition known from lakes Malawi (mbunas: rock dwelling species) and Tanganyika, and especially A. Jatilabris looks like some morphotypes believed to be ,perfectly adapted* to the rocky shores of these lakes. A molecular study of 61 specimens shows only small differences between 18 haplotypes found, thus indicating the monophyly of the ‚Soda tilapıas‘. The sequence data clearly represent a gene tree and not a species tree. An outgroup comparison with the four morphologically closest and geographically nearest Oreochromis species shows a close relationship of all ‚Soda tilapıas‘ while Oreochromis is relatively more distantly related. We therefore consider Alcolapia, formerly a subgenus of Oreochromis, as a valid genus. Since the ‚Soda tilapıas‘ are not directly related to the cichlid flocks from lakes Malawi and Tanganyika this would imply that the striking similarity in head morphology and dentition evolved at least three times. However, we assume that these similarities are not the result of independent and at least threefold adaptive radiations, especially since the rocky shores of the other two lakes are not present in Lake Natron. Instead, we believe that the morphotypes are caused by ,re- awakening’ of early aquired developmental mechanisms by genes leading to similar morphotypes as a result of a relatively high mutation pressure in Lake Natron as indicated by the high number of haplotypes. Comparative adaptive behaviour of Derby Eland in the zoo of the Wildlife College, Garoua and the Benoue National Park Zoo C. Tangwing-Njoke, P.O.Box 5074, Bamenda, Kamerun Study period: June, July, August, September 1986 Classification: Phylum: Vertebrata; Class: Mammalia; Order: Artiodactyla; family: Bovidae; genus: Taurotragus (Crandell, 1964) Five races were recognized by Allen (1939). There are two outstanding species, namely the Common Eland (Taurotragus oryx) and the Giant Eland (Taurotragus derbianus), the former being smaller in size and the latter being more robust and having larger horns. Distribution: Taurotragus derbianus is found from Senegal eastwards across Nigeria, Northern Equatorial Guinea, and North Eastern Congo to Sudan. Taurotragus oryx is found in the plain areas of Africa from the Cape north to Angola and Kenya, although in the South it now exists only in preserves. Findings: 398 1. Feeding: The eland is more a browser than a grazer in both localities. The food intake had a large moisture content, high percentage of crude protein and high percentage of nitrogen free extract. The five major food types eaten by the Eland in the Garoua zoo had high digestibles on dry matter basis. The nutrient composition of food types of Eland in both zones was very similar. That in the school zoo ate more water yam in the dry season than in the wet season, more vegetation in the wet season than in the dry season and it drank similar quantities of water in both seasons. 2. Time budgetting: The female eland rested more than the male in the school zoo and fed less than the male. In the Benoue environment there was no real difference in the activities of the male and female eland. The most prominent daily activities in both zones were feeding, resting and walking; while fighting and courtship were the least. The Eland in the school zoo fed most in the afternoon, rested most in the evening and walked most in the morning. In the Benoue environment it fed most in the evening, rested most in the morning and walked most in the evening. Climatic conditions greatly influenced their activities. Phylogeny of the anuran genus Mantella from Madagascar M. Vences, F. Glaw, A. Hille & W. Böhme, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany Malagasy poison frogs of the genus Mantella (Ranidae: Mantellinae) are among the most prominent and attractive amphibians from Madagascar. Studies on syste- matics and distribution of Mantella can contribute important data to assess the conservational status of species. This is of special interest since large numbers of specimens are yearly exported for pet trade. We combined a systematic revision of the genus with phylogenetic studies using osteological, morphological, bioacoustical, caryological and allozyme characters. All data sets agree in defining several species groups within the genus. Recent molecular studies carried out by M. Veith and C. Schafer (Mainz University) corroborated the monophyly of most of these. Geographic distribution of species within species groups is largely allopatric or parapatric; all cases of syntopy regard species of different groups. Nei's genetic distance ranges from 0.04 to 0.4 within species groups and from 0.18 to 0.75 between species of different groups. Mantella laevigata and the M. betsileo group are considered as the most basal lineages within the genus. M. madagascariensis and M. baroni, which are extremely similar to each other by dorsal colouration, are no sister species and belong to different species groups, indicating that Müllerian mimicry could partly be responsible for the pattern similarity. These examples show that the phylogeny of Mantella can contribute important new aspects both to the still poorly understood patterns of Madagascan biogeography, and to the general evolutionary mechanisms related to aposematism. 399 Distribution patterns and specificity of Chrysomelids (Coleoptera) in central African forests T. Wagner, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, 53113 Bonn, Germany Chrysomelidae were collected in forests of Kivu, Rwanda, and predominately Uganda, ranging from lowland rain forests (670 m a.s.l.) to upper montane forests (2950 m a.s.l.). Data were collected during dry and rainy season. In Budongo Forest, Uganda, trees were fogged in secondary and adjacent primary forest plots. Chrysomelid abundance is strongly influenced by foraging ants in the tree crown. Faunal overlap of leaf beetles is lower on conspecific trees between primary and secondary forest than on different tree species in one of these forest types. The community structure of chrysomelids shows high differences between seasons in Budongo Forest, while the fauna shows much higher similarities between different forests during dry season. Especially small Alticinae were highly abundant in all forests investigated during dry season. These beetles were possibly aggregated along a gradient of humidity in the tree crowns. Data underline low specificity to plant species, and high influence of predation and abiotic factors on the composition of arboreal tropical chrysomelid communities. Lessons on evolutionary processes from an unwanted experiment in a tropical fish community F. Witte & C.D.N. Barel, Institute of Evolutionary and Ecological Sciences, Leiden University, Netherlands Lacustrine East-African cichlid fishes provide outstanding examples of isolated vertebrate communities in the tropics. In spite of differences in age, origin and limnology, each of the three great lakes (Tanganyika, Malawi and Victoria) contains several hundreds of endemic fish species of the family cichlidae. Though the cichlid flocks in these lakes evolved independently, there is a considerable anatomical and ecological resemblance between species from different lakes. The diversity in Lake Tanganyika is greatest due to its older age and the polyphyletic origin of its cichlid species. The monophyletic flock of Lake Victoria, comprising more than 500 species, is extraordinarily young since 12 400 years ago the lake was dry. The fast evolution of the cichlids in Lake Victoria can be explained by disruptive sexual selection upon male coloration allowing speciation even in sympatry. The cichlid bauplan apparently facilitated subsequent adaptive radiation into virtually all available habitats and food sources of the lake. The anatomical difference between closely related sympatric species is generally small, sometimes even smaller than environmentally induced intraspecific differences in allopatry or allochrony. The young species are genetically isolated by premating barriers which can be disturbed by environmental changes. In the 1980s approximately 200 cichlid species vanished from Lake Victoria, particularly in the sublittoral and offshore waters which cover by far the greatest 400 part of the lake. The main cause of this decline was an explosive population increase of Nile perch, a huge predator introduced into the lake in the 1950s. Concomitant eutrophication and algal blooms resulted in a decrease of water transparency and of oxygen concentrations which in turn may have contributed to a further decline of the cichlids. However, unexpectedly, recent surveys demonstrated the recovery of some species, while closely related, ecologically similar species fail to cope with the new ecosystem. The dramatic perturbations in Lake Victoria provide us with natural experiments on the ecology and evolution of vertebrate communities in the tropics and warn us for the vulnerability of these unique ecosystems to human interference. bo 113, 14. 15 16. 17. 18. > 20. 2: 22. 23: In der Serie BONNER ZOOLOGISCHE MONOGRAPHIEN sind erschienen: aumann, C.M.: Untersuchungen zur Systematik und Phylogenese der holark- tischen Sesiiden (Insecta, Lepidoptera), 1971, 190 S., DM 48,— Ziswiler, V., H.R. Güttinger & H. Bregulla: Monographie der Gattung Erythrura Swainson, 1837 (Aves, Passeres, Estrildidae). 1972, 158 S., 2 Tafeln, DM 40,- . Eisentraut, M.: Die Wirbeltierfauna von Fernando Poo und Westkamerun. Unter besonderer Berücksichtigung der Bedeutung der pleistozänen Klimaschwankungen für die heutige Faunenverteilung. 1973, 428 S., 5 Tafeln, DM 106,- . Herrlinger, E.: Die Wiedereinbürgerung des Uhus Bubo bubo in der Bundesrepu- blik Deutschland. 1973, 151 S., DM 38,— . Ulrich, H.: Das Hypopygium der Dolichopodiden (Diptera): Homologie und Grundplanmerkmale. 1974 , 60 S., DM 15,— . Jost, O.: Zur Ökologie der Wasseramsel (Cinclus cinclus) mit besonderer Berück- sichtigung ihrer Ernährung. 1975, 183 S., DM 46,— . Haffer, J.: Avifauna of northwestern Colombia, South America. 1975, 182 S., DM 46,— . Eisentraut, M.: Das Gaumenfaltenmuster der Säugetiere und seine Bedeutung für stammesgeschichtliche und taxonomische Untersuchungen. 1976, 214 S., DM 54,— . Raths, P, & E. Kulzer: Physiology of hibernation and related lethargic states in mammals and birds. 1976, 93 S., 1 Tafel, DM 23,- . Haffer, J.: Secondary contact zones of birds in northern Iran. 1977, 64 S., 1 Falt- tafel, DM 16,- . Guibé, J.: Les batraciens de Madagascar. 1978, 144 S., 82 Tafeln, DM 36,— . Thaler, E.: Das Aktionssystem von Winter- und Sommergoldhähnchen (Regulus regulus, R. ignicapillus) und deren ethologische Differenzierung. 1979, 151 S., DM 38,- Homberger, D.G.: Funktionell-morphologische Untersuchungen zur Radiation der Ernährungs- und Trinkmethoden der Papageien (Psittaci). 1980, 192 S., DM 48,- Kullander, S.O.: A taxonomical study of the genus Apistogramma Regan, with a revision of Brazilian and Peruvian species (Teleostei: Percoidei: Cichlidae). 1980, 152 S., DM 38,- Scherzinger, W.: Zur Ethologie der Fortpflanzung und Jugendentwicklung des Habichtskauzes (Strix uralensis) mit Vergleichen zum Waldkauz (Strix aluco). 1980, 66 S., DM 17,— Salvador, A.: A revision of the lizards of the genus Acanthodactylus (Sauria: Lacertidae). 1982, 167 S., DM 42,- Marsch, E.: Experimentelle Analyse des Verhaltens von Scarabaeus sacer L. beim Nahrungserwerb. 1982, 79 S., DM 20,- Hutterer, R., & D.C.D. Happold: The shrews of Nigeria (Mammalia: Soricidae). 1983, 79 S., DM 20,— Rheinwald, G. (Hrsg.): Die Wirbeltiersammlungen des Museums Alexander Koenig. 1984, 239 S., DM 60,— Nilson, G., & C. Andrén: The Mountain Vipers of the Middle East — the Vipera xanthina complex (Reptilia, Viperidae). 1986, 90 S., DM 23,— Kumerloeve, H.: Bibliographie der Säugetiere und Vögel der Türkei. 1986, 152S., DM 33,—= Klaver, C., & W. Böhme: Phylogeny and Classification of the Chamaeleonidae (Sauria) with Special Reference to Hemipenis Morphology. 1986, 64 S., DM 16,- Bublitz, J.: Untersuchungen zur Systematik der rezenten Caenolestidae Trouessart, 1898 — unter Verwendung craniometrischer Methoden. 1987, 96 S., DM 24,— 24. 25: 26. 2k 28. isk 30. 31. 32. 33: 34. 39, 36. 1. 38. 30; AO. 41. 42, 43. 44, as 46. (Pisces: Cyprinidae). 1996, 227 S., DM 57,- Tees. Be) Kirahelah taxonomy and phylogenetic impliations 120 S., DM 30 - i 1987, 322 S. DM 81,- Löhrl, H.: Etho-ökologische Untersuchungen. an vr Kleiberarten 6 tidae) — eine en PSOE sae 208 S., DM 52,— a schichtliche A 1988, 175 S., DM 44. = 5 Lang, M.: Phylogenetic and biogeographic patterns of Basiliseine I ue (Reptilia: Squamata: "Iguanidae"). 1989, 172 S., DM 43,— Hoi-Leitner, M.: Zur Veränderung der Säugetierfauna des Neusiedlersee- 2 im Verlauf der letzten drei Jahrzehnte. 1989, 104S$.,DM26— ae Bauer, A. M.: Phylogenetic systematics and Biogeography of the Carphoda (Reptilia: Gekkonidae). 1990, 220 S., DM 55,— Fiedler, K.: Systematic, evolutionary, and ecological implications is yl phily within the Lycaenidae wi Lepidoptera: a Ce we DM 53,— Be Arratia, G.: Development and variation of the suspensorium of en ei (Teleostei: Ostariophysi) and their phylogenetic relationships. 1992, 148 S. Kotrba, M.: Das Reproduktionssystem von Cyrtodiopsis whitei Curran ( Diptera) unter besonderer Berücksichtigung der inneren weiblichen organe. 1993, 115 S., DM 32,— a Blaschke-Berthold, U.: Anatomie und Phylogenie der Bibionomorpt Diptera). 1993, 206 S., DM 52,- Hallermann, J.: Zur Morphologie der Ethmoidalregion der Iguania (S eine vergleichend-anatomische Untersuchung. 1994, 133 S., DM 33,- Arratia, G., & L. Huaquin: Morphology of the lateral line system and of of Diplomystid and certain primitive Loricarioid Catfishes and yey logical considerations. 1995, 110 S., DM 28,- — Hille, A.: Enzymelektrophoretische Untersuchung zur ie Popul struktur und geographischen Variation im Zygaena-transalpina-Superspezi plex (Insecta, Lepidoptera, Zyeaenidae). 1995, 224 S., DM 56,- ae Martens, J., & S. Eck: Towards an Ornithology of the Himalayas: ecology and vocalizations of Nepal birds. 1995, 448 S., 3 Farbtafeln, DM Chen, X.: Morphology, phylogeny, biogeography and systematics of : Browne, D.J., & C.H. Scholtz: The morphology of the hind wing Ae wing base of the Scarabaeoidea (Coleoptera) with some phylogenetic tions. 1996, 200 S., DM 30,— | oe Bininda-Emonds, O. R. P, & A. P. Russell: A morphological perspect the phylogenetic relationships of the extant phocid seals (Mammalia Carn: Phocidae). 1996, 256 S., DM 64,- Klass, K.-D.: The external male genitalia and the phylogeny of Br Mantodea. 1997, 341 S., DM 85, — | Hörnschemeyer, T.: Morphologie und Evolution des Hiigelesienie optera und Neuropterida. 1998, 126 S., DM 32,— a Solmsen, E.-H.: New World nectar-feeding bats: biology, ner ä metric approach to systematics. 1998, 118 S., DM 30,— Berendsohn, W.G., C.L. Hauser & K. H. Lampe: Biodiversitatsinformati Deutschland: Bestandsaufnahme und Perspektiven. 1999, 64.S., DM 16, Rheinwald, G. (Hrsg.): Isolated Vertebrate Communities in the Tropics. Pr ings of the 4" International Symposium, Bonn May 13-17, 1999. ar S., 4 tafeln, DM 100.—