UNlvtR- OF ILLINOIS RY «URBANA 1PAIGN BIC _' yvul? ^or i^o.^S" l*\s«©i* boelb £«vJer l Floral and Faunal he Eastern Slopes c Jaturelle [ntegrale d Vith Refer* ice to teven A A Floral and Faunal Inventory of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar & WWF FIELDIANA Zoology NEW SERIES, NO. 85 A Floral and Faunal Inventory of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar: With Reference to Elevational Variation Steven M. Goodman, Editor Department of Zoology Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60615 U.S.A. World Wide Fund for Nature Aires Protigies B. P. 738 Antananarivo (101) Madagascar Accepted February 28, 1996 Published September 30, 1996 Publication 1480 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 1996 Field Museum of Natural History ISSN 0015-0754 PRINTED IN THE UNITED STATES OF AMERICA Contents Preface vii 1. Description of the 1993 Biological Inventory of the Reserve Naturelle Integrate d'Andrin- gitra, Madagascar 1 Steven M. Goodman Appendix 1-1. Participants in the Project (Field and Laboratory) 5 2. Description of the Reserve Naturelle Integrale d'Andringitra, Madagascar 7 Steven M. Goodman and Beverley A. Lewis 3. Meteorology 20 Steven M. Goodman and Aristide Andrianarimisa 4. A Study of the Botanical Structure, Composition, and Diversity of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar 24 Beverley A. Lewis, Peter B. Phillipson, Michele Andrianarisata, Grace Rahajasoa, Pierre J. Rakotomalaza, Michel Randriambololona, and John F. McDonagh Appendix 4-1. Checklist of Vascular Plant Species Recorded from Andringitra 55 Appendix 4-2. Vertebrate Food Plants on the Eastern Slopes of the RNI d'Andringitra 72 5. The Pteridophytes of the Eastern Slopes of the Reserve Naturelle Integrate d'Andrin- gitra 76 France Rakotondrainibe and Fidele Raharimalala 6. The Utilization of Canarium (Burseraceae) Seeds by Vertebrates in the RNI d'Andringitra, Madagascar 83 Steven M. Goodman and Eleanor J. Sterling 7. Millipedes (Diplopoda) from the Eastern Slopes of the Reserve Naturelle Integrale d'An- dringitra, Madagascar 90 Henrik Enghoff 8. Ant Diversity Patterns Along an Elevational Gradient in the Reserve Naturelle Integrate d' Andringitra, Madagascar 93 Brian L Fisher 9. Spatial Distribution of Some Aquatic Insects in the Reserve Naturelle Intdgrale d'Andrin- gitra, Madagascar 1 09 Francois-Marie Gibon, Jean-Marc Elouard, and Michel Sartori 10. New Heptageniidae (Insecta: Ephemeroptera) from the Reserve Naturelle Integrale d'An- dringitra, Madagascar 121 Michel Sartori and Jean-Marc Elouard 11. Two New Species of Simulium (Diptera: Simuliidae) from the Reserve Naturelle Integrale d'Andringitra, Madagascar 131 Jean-Marc Elouard, Theogene Pilaka, and Fabienne Ranaivoharindriaka 12. Parasitic and Commensal Arthropods of Some Birds and Mammals of the Reserve Naturelle Integrate d'Andringitra, Madagascar 136 Barry M. OConnor 13. Blood Parasites from Birds in the Reserve Naturelle Integrale d'Andringitra, Madagascar 142 Ellis C. Greiner, Michael S. Putnam, and Steven M. Goodman 14. Elevational Variation in Soil Macroinvertebrates on the Eastern Slopes of the Reserve Na- turelle Integrale d'Andringitra, Madagascar 144 Steven M. Goodman, Philip P. Parrillo, Sam James, and Petra Sierwald 15. Shrimps (Crustacea: Decapoda: Atyidae) of the Reserve Naturelle Integrale d'Andringitra, Madagascar 152 Mahefason Richard Andriamihaja 16. Crayfish (Parastacidae) and Crabs (Potamonidae) of the Reserve Naturelle Integrate d'An- dringitra, Madagascar 155 Bako Rabeharisoa 17. Amphibians and Reptiles of the Reserve Naturelle Integrate d'Andringitra, Madagascar: A Study of Elevational Distribution and Local Endemicity 158 Christopher J. Raxworthy and Ronald A. Nussbaum 18. The Birds of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagas- car 171 Steven M. Goodman and Michael S. Putnam 19. The Shrew Tenrecs (Microgale) (Insectivora: Tenrecidae) of the Reserve Naturelle Integrate d'Andringitra, Madagascar 191 Paulina D. Jenkins, Steven M. Goodman, and Christopher J. Raxworthy Appendix 19-1. Key to the Species of Microgale Occurring in RNI d'Andringitra 217 20. Insectivore Ecology in the Reserve Naturelle Integrate d'Andringitra, Madagascar 218 Steven M. Goodman, Christopher J. Raxworthy, and Paulina D. Jenkins 21. Systematic Studies of Madagascar's Endemic Rodents (Muroidea: Nesomyinae): A New Genus and Species from the Central Highlands 231 Michael D. Carleton and Steven M. Goodman 22. The Rodents of the Reserve Naturelle Integrate d'Andringitra, Madagascar 257 Steven M. Goodman and Michael D. Carleton Appendix 22-1. Comparative Material 283 23. Results of a Bat Survey of the Eastern Slopes of the Reserve Naturelle Integrate d'Andrin- gitra, Madagascar 284 Steven M. Goodman 24. The Carnivores of the Reserve Naturelle Integrate d'Andringitra, Madagascar 289 Steven M. Goodman 25. Rapid Assessment of the Primate Fauna of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar 293 Eleanor J. Sterling and Marie Gisele Ramaroson Gazetteer of Localities Mentioned in the Text 306 Index (Common Names, Scientific Names, and Subject) 309 VI Preface In late 1992, Olivier Langrand and Sheila O'Connor, of World Wide Fund for Nature (WWF), Madagascar, and I discussed the need to conduct systematic multidisciplinary biological inventories of poorly known forested areas of Madagascar. These surveys would provide critical information on the biological diversity of re- serves, basic data on the natural history of a wide variety of organisms, baseline information to as- sess changes in the biota of these areas, and the means to get Malagasy and foreign researchers together for collaborative field projects. Further, it was considered critical that the initial results of each survey be published in monograph form and as quickly as possible, so that the information could be made available to Malagasy officials and the conservation community. For decades, researchers have been conducting biological inventories on Madagascar. Often these surveys consist of relatively small groups of field- workers, encompassing one or two disciplines, and are conducted in areas with permanent infra- structure. Further, methodologies are often incon- sistent between studies, thus limiting comparisons between sites. Our main questions were whether it was feasible and practical to conduct surveys with a large multidisciplinary and multinational group of biologists in remote areas and whether the data that could be gathered during relatively rapid surveys would be useful. As a test case, in April 1993 a biological in- ventory of the Zombitse Forest, near Sakaraha, was organized. This site was relatively easy to survey because of its proximity to roads and other infrastructure. The group consisted of 18 biolo- gists— nine from Madagascar and the balance from another five countries. In general, the results of this survey, which have already been published elsewhere,1 were satisfactory. After the successful mission to the Zombitse Forest, it was clear that such projects were feasi- ble and generated useful information. We decided to conduct inventories along altitudinal gradients in mountainous areas of the island, particularly where WWF has ongoing conservation projects. The months from September to December, tucked in between the end of the austral winter and the 'Goodman, S. M., and O. Langrand, eds. 1994. In- ventaire biologique foret de Zombitse. Recherches pour le Developpement, Serie Sciences biologiques, No. Spe- cial. Centre d' Information et de Documentation Scien- tifique et Technique, Antananarivo, Madagascar. beginning of the rainy season, are the best period to conduct such surveys. It is the only time of the year when more or less optimal conditions exist for multidisciplinary groups to conduct biological inventories. For example, this is the period in which many plants begin to flower and fruit; birds are breeding, singing, and conspicuous; and the activity of reptiles and amphibians increases ow- ing to warming days and greater rainfall. The first transect, in 1993, was conducted on the eastern slopes of the Reserve Naturelle Inte- grate (RNI) d'Andringitra. The results of this tran- sect are reported in this volume. The 1994 survey was conducted in the Reserve Speciale d'Anja- naharibe-Sud in the northeast (the results of that survey are now being organized as a parallel monograph to this one), and the 1995 transect was conducted in the RNI d'Andohahela in the south- east. The 1993 biological inventory of the RNI d'Andringitra was made possible by funding from the German government through Kreditanstalt fur Wiederaufbau (KfW) to WWF as part of an in- tegrated conservation and development project undertaken to protect the Andringitra and Pic d'lvohibe areas. The success of the 1993 mission rests largely with the people of Ambalamanenjana. The local village committee was extremely congenial to work with and provided remarkable assistance in arranging guides, camp assistants, and porters. We extend our sincere thanks to the people of Am- balamanenjana. We are indebted to WWF staff in both Anta- nanarivo and Ambalavao for their help in orga- nizing this mission; in particular, we thank O. Langrand, S. O'Connor, H. Rabetaliana, P. Schachenmann, and C. Wilm6. For permits to work in the reserve we are grateful to the Direc- tion des Eaux et Forets and the Association Na- tional pour la Gestion des Aires Protegees, es- pecially G. Rakotonarivo, C. Ravaoarinoromanga, and M. H. Faramalala. A mission such as the RNI d'Andringitra in- ventory is never simple or easy, and the compan- ionship and endurance of the participants are greatly acknowledged. The relatively rapid deter- mination of material gathered during the mission, analysis of data, and assemblage of this report by both field and laboratory researchers attest to the perseverance of those involved. We owe a great debt to the reviewers, who crit- ically evaluated the chapters presented in this vol- ume, in several cases multiple papers, and often VII under considerable time constraints. The endpaper itor of Fieldiana, and M. Pannell, Managing Ed- map was drawn by MYE, Antananarivo, using a itor of the Field Museum Press, took on the task modified base map "Carte de vegetation de la of producing this volume with enthusiasm and R.N.I, no. 5 d'Andringitra," prepared by T. An- skill. Partial subsidy for publication of this mono- driamanga and J. Rahantamalala. M. Razafimpa- graph was provided by KfW. hanana drafted many of the French-language re- sumes, which were subsequently edited by O. S. M. Goodman Langrand. J. Sedlock, J. Weinstein, and D. White helped immensely in preparing the illustrations July 1996 for this volume. Finally, W. Burger, Scientific Ed- Chicago, Illinois Vlll Chapter 1 Description of the 1993 Biological Inventory of the Reserve Naturelle Integrate d'Andringitra, Madagascar Steven M. Goodman The Reserve Naturelle Integrate (RNI) d'An- dringitra is located in south-central Madagascar (endpaper map) between 22°07'-22°2rS latitude and 46°47'-47°02'E longitude. It comprises an area of approximately 31,160 ha over the eleva- tional range of 650-2658 m. The reserve was cre- ated in 1927, and the limits were demarcated in 1966 (Nicoll & Langrand, 1989). The main portion of the reserve is a flank of high mountains that skirts the Central High Plateau and the eastern escarpment leading down to the Indian Ocean. The notable peaks within the reserve are Pic Boby (2658 m) and Pic Bory (2630 m). On the basis of the combination of elevational range, size, and the remarkable diversity of habitats and plant communities, the RNI d'Andringitra has been considered an area of significant biological diver- sity (Nicoll & Langrand, 1989) and identified as a priority site under the National Environmental Ac- tion Plan. It is a region with numerous types of habitats — large continuous and mostly undisturbed rain forests to the east, dry forest to the west, and magnificent crystalline mountain ranges, mountain meadows, heathlands, and escarpments on the high plateau. Most previous scientific studies in the re- gion have been conducted at the upper elevational limit of the humid forest and on the high plateau. These two biomes are different from the lower- lying humid forest on the eastern slopes of the massif. The latter area, which until now had re- mained unstudied, was the focus of a biological inventory at the end of 1993, the results of which are presented in this volume. A multinational and multidisciplinary group of biologists studied the fauna and flora of the east- ern slopes of the reserve between September and December 1993. Our main interests were to doc- ument the biological richness of the area and to study species distribution and turnover as a func- tion of elevation. Further, we wanted to test the utility of rapid assessment methods for a wide va- riety of organisms. Four camps were placed at different elevations (720, 810, 1210, and 1625 m), and transects were centered around these camps at an altitudinal limit of ±75 m elevation. Two separate field groups worked these elevational zones; each group generally moved between camps in unison and worked a respective zone for the same period. The only exception was the aquatic insect specialists, who needed substantial- ly less time per elevational transect. We also gathered data to test one additional hy- pothesis within the context of a general inventory. The 720 m zone was in a forested area with con- siderable human disturbance, and the 810 m zone was mostly undisturbed. A comparison of these two zones allowed us to test the hypothesis that forest degradation within the same altitudinal swath has little effect on the various organisms studied. Further, the data from the 810, 1210, and 1625 m sites allowed examination of the distri- bution and density of organisms along an eleva- tional gradient. Abbreviations Used in the Text ANGAP Association Nationale pour la Gestion des Aires Prot6g6es FIELDIANA: ZOOLOGY, N.S., NO. 85, SEPTEMBER 30, 1996, PP. 1-319 CNRS Centre National de la Recherche Scientifique DEF Direction des Eaux et Forets FTM Foiben-Taosarintanin'i Madagasikara (Institut National de Geodesie et Car- tographic) FMNH Field Museum of Natural History KfW Kreditanstalt fur Wiederaufbau LRSAE Laboratoire de Recherche sur les Sys- temes Aquatique et leur Environne- ment MBG Missouri Botanical Garden MNHN Museum National d'Histoire Natu- relle, Paris ORSTOM Office de la Recherche Scientifique et Technique Outre-mer PBZT Pare Botanique et Zoologique de Tsimbazaza PN Pare National RB Reserve de Biosphere RCP La Recherche Cooperative sur Pro- gramme No. 225, under the Centre National de la Recherche Scientifique RNI Reserve Naturelle Integrate RS Reserve Speciale UMMZ University of Michigan Museum of Zoology WWF World Wide Fund for Nature Transect Sites Coordinates were determined with the use of a Geographical Position System, the geographical names from various maps (Foiben-Taosarintanin'i Madagasikara [Institut National de Geodesie et Cartographie]; FTM, 1979) and consultation with local people. The altitude was determined with al- timeters. The positions of each camp and the major trail systems are shown on the endpaper map. 720 m (camp 1) — Fianarantsoa Province, approx- imately 45 km S of Ambalavao, east bank Ian- tara River, along Ambalamanenjana-Ambatom- boay trail, edge of RNI d'Andringitra, 22°13'20"S, 47°01'29"E. 810 m (camp 2) — Fianarantsoa Province, approx- imately 43 km S of Ambalavao, RNI d'Andrin- gitra, junction of the Sahanivoraky and Saha- vatoy rivers, 22°13'40"S, 47°00'13"E. 1210 m (camp 3) — Fianarantsoa Province, ap- proximately 40 km S of Ambalavao, RNI d'Andringitra, along Volotsangana River, a trib- utary of the Sahavatoy River, 22°13'22"S, 46°58'18"E. 1625 m (camp 4) — Fianarantsoa Province, ap- proximately 38 km S of Ambalavao, RNI d'Andringitra, on ridge above and east of Vo- lotsangana River, 22°11'39"S, 46°58'16"E. Itinerary of the 1993 Expedition Reconnaissance — Reconnaissance (10-15 Sep- tember 1993) was conducted by Goodman to lo- cate the zone on the eastern slope appropriate for an elevational transect. He was accompanied by Schachenmann and Rabetaliana for a portion of the trip. First Field Trip — During the first field trip (23 September- 1 November 1993), scientific mem- bers included Fisher (ants), Goodman (birds and bats), Putnam (birds), Ramaroson (lemurs), and Sterling (lemurs). The dates for each transect fol- low: 23 September-4 October, 720 m transect; 5- 13 October, 810 m transect; 14-21 October, 1210 m transect; and 22-31 October, 1625 m transect. Second Field Trip — During the second field trip (14 November- 18 December 1993), scientific mem- bers included Andrianarisata (botany), Elouard (aquatic insects), Gibon (aquatic insects), Goodman (small mammals and carnivores), Lewis (botany), McDonagh (botany), Rabibisoa (reptiles and am- phibians), Rahajasoa (botany), Raharilala (botany), Rakotomalaza (botany), Randriambololona (bota- ny), Raxworthy (reptiles and amphibians), and Ra- zafimanantsoa (reptiles and amphibians). The dates for each transect follow: 14-21 November, 720 m transect; 22-29 November, 810 m transect; 30 No- vember-7 December, 1210 m transect; and 8 De- cember-17 December, 1625 m transect. The only survey reported here that was not con- ducted in the latter portion of 1993 is the pteri- dophyte study, conducted by Rakotondrainibe and Raharimalala (Chapter 5). In May and June 1995, these two researchers visited the eastern slopes of the reserve and conducted an elevational transect. Their camp and research sites were the same as those used by the 1993 group. A list of all field and laboratory participants in this project is pre- sented in Appendix 1-1. Logistics and Trail Systems The eastern slopes of the RNI d'Andringitra are isolated and not directly accessible by motor ve- FIELDIANA: ZOOLOGY Fig. 1-1. Camp 1, at 720 m, along the Iantara River. A view from the small settlement of Ambarongy northwest across the Iantara River. The river marks the eastern boundary of RNI d'Andringitra. The forest clearing in the foreground has been converted into irrigated terraces. Across the river and within the reserve is a recent tavy. In the lower right corner is camp 1 . (Photograph by R Schachenmann.) hide. We drove from Ambalavao along the road leading to Analafolaka (endpaper map). About 40 km from Ambalavao and a few kilometers before Miarinarivo there is a road that ends in the village of Ambalamanenjana. This was the village "at the end of the road" where we stored supplies and from which most of our guides, assistants, and porters originated. From Ambalamanenjana we walked along a more or less southerly trail that for the first 3-4 hours passed through deforested swamps, wet meadows, and open savannah, then entered forest of varying levels of disturbance, crossed the Col de Vohitsatsiva just west and slightly below Pic Vohipia (1761 m), and then dropped into the Iantara River valley near its con- fluence with the Korokoto River; the latter point is the northeastern boundary of the RNI d'An- dringitra. The trail continues south along the Ian- tara River, marking the eastern limit of the re- serve, to Ambatamboay and then to villages fur- ther south. Our first camp (720 m) was along this trail. The direct distance from Ambalamanenjana to the northeastern reserve limit is about 12 km; it was another 6 km south to our first camp. Our first camp was just north of the junction of the Sahavatoy and Iantara rivers, at the edge of a 4-6 ha clearing associated with the small settle- ment of Ambarongy, a community of two or three families of slash-and-burn (tavy in Malagasy) ag- riculturalists (Fig. 1-1). Between the Col de Vo- hitsatsiva and our 720 m camp, the trail passed through more or less continuous forest of various levels of degradation. The forested area along the Iantara River and the low hills just to the east contained an assortment of trails that were suit- able for most of our transects. Between Ambaron- gy and Ambatamboay, the main Iantara trail leaves the forest and passes through areas of tavy, settlements, and secondary forest. We used the Sahavatoy Valley as our access into the higher elevations of the reserve; the val- ley has its origin near Pic Bory (2630 m). A trail near our first camp traversed the Iantara River and passed obliquely toward the Sahavatoy River. Our GOODMAN: DESCRIPTION OF THE 1993 BIOLOGICAL INVENTORY second camp (810 m) was on the north bank of the Sahavatoy River at the junction of the Saha- nivoraky River, about 2.5 km (trail distance) away from camp 1. The trail continued as a poorly de- fined cattle path for another 1 .5 km along the Sa- havatoy River, all within the elevational transect zone. The area around the 810 m camp contained a trail system sufficient for the inventory. The for- est in the vicinity of the camp was relatively un- disturbed except for a clearing along the main trail about 1 km below the camp. Further, in various areas, mostly below the camp, there were clear signs of cattle browsing. Within the reserve, on the southern bank of the Sahavatoy River and closer to the confluence with the Iantara River, there was an active tavy of several hectares within the reserve limits. Together with several local people, and with permission from the Direction des Eaux et Forets (DEF), we cut a trail leading from the end of the Sahavatoy River trail, above the 810 m camp, along a projectory paralleling the north side of the river valley. The new trail, which was just wide enough for people to pass single file, climbed steeply up a ridge to about 1 1 00 m and passed through an area with numerous hills and deep val- leys. The trail continued until it crossed the Vo- lotsangana River, where we established the third camp, at 1210 m, in a small hollow along the west bank. There was no sign of human disturbance in this area, and it was necessary to establish all of our own trails within this transect zone. Below our camp, the Volotsangana River plunged dramati- cally into the Sahavatoy River — within less than lA km of ground distance it dropped about 200 m in elevation. Our fourth camp, at 1625 m, was established along a north-bearing trail that led up a steep ridge on the eastern side of the Volotsangana Riv- er, not far from our 1210 m camp. This trail, which we cut, climbed steeply; within about 2 km of the bifurcation with the trail leading between camps 2 and 3 it leveled off to about 1550 m, which was within the lower limit of the 1625 m transect. This main ridge trail continued for an- other 1.5 km past our fourth camp and was the main path of the 1625 m transect. We were able to cut a route that climbed up to the high plateau region of the reserve. This path passed along a saddle ridge, above the origin of the Volotsangana River, to the west of Ampasi- potsy, and allowed access to the upper Kimora River valley and Pic Bory. The herpetologists used this trail to explore the upper regions of the mountain and establish a fifth camp along the Ki- mora River at 2075 m. Other researchers also vis- ited the high plateau area of the reserve. Previous Research Conducted in the Reserve Little scientific research has been conducted on the Andringitra Massif (Paulian, 1958). The no- table exception is the 1970-1971 Recherche Cooperative sur Programme (RCP) No. 225, un- der the Centre National de la Recherche Scienti- fique (CNRS), mission to the reserve (Paulian et al., 1971). The principal studies conducted during this expedition were on the geomorphology, cli- mate, and vegetational structure of the high moun- tain portion of the reserve. Several zoologists ac- companied the group and made collections (see subsequent chapters in this monograph for a re- view of this information). Because much of the zoological information from the RCP expedition has not been formally published (e.g., that on rep- tiles, rodents, and insectivores) but is critical to our understanding of the high mountain fauna, de- tails on the positions of the RCP camps (Paulian et al., 1971) are presented here (endpaper map). Camp 1 (1650 m) — Manjarivolo, at the southern end of the Andringitra chain and to the north of Pic d'lvohibe. The area covered included dense humid middle-altitude forest at 1600 m and dense humid montane forest between 1 600 and 1850 m. The expedition occupied this site between 23 October and 4 November 1970. Camp 2 (2030 m) — Plateau d'Andohariana, in the northern sector of the mountain complex, be- low and north of the main flank of high peaks and above Antanifotsy. The vegetation of this zone is largely herbaceous and ericoid. This camp was occupied between approximately 8 and 23 November 1970. Camp 3 (2470 m) — Cuvette Boby, just below Pic Boby (2658 m). Mostly rock formations with rupicolous plant communities are found there. Workers stayed at this site between 23 and 29 November 1970. During the same period the area around Varavarana, close to Pic Bory, and Pic Ivangomena (2556 m), in the same contour as Pic Boby, were explored. Camp 4 (2000 m) — At Marositry in the north- eastern portion of the high mountain complex. The vegetational cover was mostly ericoid, with FIELDIANA: ZOOLOGY vestigial areas of dense sclerophyllous plants, largely composed of Agauria. This camp was occupied between 30 November and 5 Decem- ber 1970. Camp 5 (1995 m) — Anjavidilava, just below the "signal" d'Anjavidilava (2030 m), and along the Eaux et Forets weather station trail. Several types of vegetation occur in this area — near the summit of Anjavidilava, sclerophyllous forest largely composed of Philippia, ericoid bush, and rupicolous vegetation; to the south and in the Kimora Valley, the upper portion of dense middle-altitude humid forest; on the slopes northwest of Anjavidilava, a combination of montane and middle-altitude dense humid for- est; and to the northeast of the massif of Vo- hidray, a combination of ericoid bush and ru- picolous vegetation. This zone and RCP camp 6 were occupied between 17 December 1970 and 16 January 1971. Camp 6 (1530 m) — Ambalamarovandana, on the west side of the Vohidray flank of the main mountain chain. The forest in this area was classified as middle-altitude humid forest. Literature Cited FTM. 1979. Sendrisoa no. 55, precarte, 1:100,000. Foi- ben Taosarintanin'i Madagasikara, Antananarivo. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la conservation et des aires protegees. World Wide Fund for Nature, Gland, xvii + 374 pp. Paulian, R. 1958. L'Andringitra. Revue de Madagascar, troisieme trimestre, nouvelle s6rie, 3: 51-54. Paulian, R, J. M. Betsch, J. L. Guillaumet, C. Blanc, and P. Griveaud. 1971. RCP 225. Etudes des eco- systemes montagnards dans le region malgache. I. Le massif de l'Andringitra. 1970-1971. Geomorphologie, climatologie et groupements veggtaux. Bulletin de la Soci<40°; and (5) plot topography: classified as valley, mid-slope, or ridge. The analysis was conducted using the CANOCO computer program. Results Environmental Data The environmental data recorded for each sam- ple plot are given in Table 4- 1 . Structural Characteristics — Trees Data on the structural characteristics of the trees recorded in the plots at each transect, with analyses of variance, are given in Table 4-2. Numbers of Trees — The mean number of trees with dbh >10 cm in the plots for each transect 30 zone is given in Table 4-2. It ranged from a min- imum of 21.4 stems at transect 2 to 28.0 stems at transect 3, equivalent to 1,070 and 1,400 stems/ha, respectively. There were no significant differences in number of trees either between transects 1 and 2 or between transects 2, 3, and 4. The number of trees differed considerably among the individual plots, ranging from 16 trees (in plot 2 of transect 1) to 34 trees (in plot 13 of transect 3). dbh Measurements — Trees with dbh measure- ments >25 cm are listed for each transect in Ta- bles 4-3 through 4-6. The dbh values were sorted into size classes at intervals of 5 cm, and histo- grams of frequency were plotted for each transect (Fig. 4-1). This shows that the majority of the trees at all four transect zones had values in the smallest class, falling between 10 and 14.9 cm. The numbers of trees falling within the larger size classes decreased sharply at all elevations sam- pled within the reserve; only a few trees were in each size class with in which the dbh was >35 cm. Clear differences in tree dbh were found be- tween transect zones (Fig. 4-1). Forest outside the reserve boundaries, in the plots at the 720 m tran- sect zone, had no trees with dbh >35 cm and only 1 1 trees with dbh >25 cm. Forest within the re- serve, in the 810 m transect zone, had 21 trees with a dbh >25 cm, nine of which had a dbh >35 cm. Indeed, the largest tree recorded in the study, Sloanea rhodantha var. rhodantha form "quadri- FIELDIANA: ZOOLOGY § St r -«5 - s Tt Os n 00 SO SO l I II r-- p so r-; r^ od od (N -MMN '00 00 r^ po oo o r~ rJ Tt ^^ y y r-; © «ri oo od r~ ■* r^ - so r» r~ rn c«S ■** +1 +1 +1 +1 00 rf O 00 ^ -. 6< P*> — EG-* E E O «/-> < < = oo^>>^ — oj o •<* < < J E I 3 I 'E I u 60 « ■ "3 S o- «= «s w Ji C u u Jg £ w C E £ i) ii 5 J C S.g c c § +i S JS js -S <« ° 25 cm. Transect 4 (1625 m) had 65% of trees in the smallest size class (10-14.9 cm dbh) and few that exceeded 25 cm dbh, although this zone had the largest number of trees (five) with a dbh >50 cm. There were no significant differences in dbh values between transect 1 and 2 or between tran- sects 2, 3, and 4 (see Table 4-2). It should, how- ever, be noted that the intra-transect variability of this measurement was high, particularly at tran- sect 2, and the differences between dbh measure- ments at transects 1 and 2 were almost significant at the 5% probability level. Basal Area — The lowest total basal area was calculated for transect 1 at 26.5 m2/ha. The basal areas for transects 2 through 4 ranged from 43.2 to 49.1 m2/ha (Table 4-2). There were no statis- tically significant differences between transects among the total basal area values; this result was expected because they are calculated from the dbh measurements. Tree Height — Tree height was segregated into five size classes for each of the transects (Fig. 4-2). Although no attempt has been made to sta- tistically examine the correlation between tree height and tree dbh, some general trends can be drawn from the data. For example, in transect 4 the trees of a particular dbh class tend to be short- er than trees of similar dbh in the lower transects (compare Fig. 4- 1 with Fig. 4-2). In transect 3 the same effect is apparent, although it is less well defined. Calculating the mean height/dbh values for all trees with dbh >25 cm for transects 1 through 4 (values from Tables 4-3 through 4-7), values of 57.3, 58.8, 49.0, and 37.1, respectively, are obtained, confirming that at higher elevations these trees tend to have a lower height-to-dbh ra- tio. Tree height was significantly greater in transect zone 2, within the reserve boundaries, than out- side the reserve in transect zone 1 (P ^ 0.05). Furthermore, the number of tall trees (>20 m) was lower in transect zone 1 than zone 2. Seventy-five percent of the trees in transect 4 were within the 5-10 m height range, indicating a much lower canopy. Tree height decreased sig- nificantly (P ^ 0.05) from transect zone 2 to 3 and from transect 3 to 4, indicating a clear de- crease in tree height with increasing elevation. Mean tree height as a function of elevation in transects 2 through 4 showed that a significant LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 31 Table 4-3. Transect 1 (720 m). Trees with dbh >25 cm arranged in decreasing size. dbh Basal area Height Family Genus/species (cm) (m2) (m) Lauraceae sp. 3 34.8 0.095 20 Lauraceae sp. 10 31.3 0.077 12 Clusiaceae Mamtnea sp. 6 30.3 0.072 15 Not recorded 30.0 0.071 20 Rubiaceae Gaertnera sp. 3 29.7 0.069 14 Lauraceae Cryptocarya sp. 3 28.4 0.063 16 Lauraceae sp. 16 27.2 0.058 15 Lauraceae Cryptocarya madagascariensis 25.5 0.051 18 Burseraceae Canarium madagascarien.se 25.2 0.050 18 Lauraceae sp. 18 25.0 0.058 15 negative relationship existed (R2 = 0.79, P < 0.001; Fig. 4-3). Crown Diameter — Mean values for crown di- ameter ranged from 3.3 (transects 1 and 4) to 4.1 m (transect 2). Crown diameter was significantly greater in transect zone 2 than in zone 4 (P < 0.05), but otherwise none of the differences be- tween transects were statistically significant based on the two ANOVA tests (Table 4-2). Structural Characteristics — Lianas and Epiphytes Number of Lianas — The mean number of li- anas per tree ranged from 0.4 to 1.2 in the four transect zones (Table 4-7). None of the individual lianas was greater than 10 cm dbh. The difference in mean number of lianas per tree between tran- sects 1 and 2 or between 2, 3, and 4 was not significantly different. As with tree height, obser- vations suggest that this parameter is influenced by microhabitat. Epiphytes — Using the visual scale in tree crowns, mean values of vascular epiphytes ranged from 1.2 (transect 1) to 2.2 (transect 4) and on tree trunks from 0.6 (transects 1 and 3) to 1.0 (transects 2 and 4) (Table 4-7). There were gen- erally more epiphytes in the crowns of trees than on the trunks. No statistically significant differ- ence was observed in the numbers of vascular ep- iphytes between transects. Table 4-4. Transect 2 (810 m). Trees with dbh >25 cm arranged in decreasing size. dbh Basal area Height Family Genus/species (cm) (m2) (m) Elaeocarpaceae Sloanea "quadriloba" 81.1 0.517 28 Sapindaceae sp. 4 75.8 0.451 27 Lauraceae Cryptocarya sp. 3 61.5 0.297 30 Clusiaceae Calophyllum drouhardi 54.8 0.236 25 Elaeocarpaceae Sloanea "quadriloba" 48.8 0.187 18 Elaeocarpaceae Sloanea "quadriloba" 42.0 0.139 16 Lauraceae Cryptocarya sp. 3 36.2 0.103 28 Fabaceae Dalbergia sp. 1 36.1 0.102 20 Myrtaceae Syzygium sp. 2 35.8 0.101 22 Cunoniaceae Weinmannia sp. 6 34.7 0.095 25 Clusiaceae Calophyllum drouhardi 31.8 0.079 18 Sterculiaceae Dombeya sp. 3 31.7 0.079 16 Monimiaceae Tambourissa thouvenotii 30.9 0.075 12 Lauraceae sp. 10 30.5 0.073 20 Cunoniaceae Weinmannia sp. 1 29.2 0.067 20 Cunoniaceae Weinmannia sp. 1 28.8 0.065 18 Moraceae Streblus dimepate 28.2 0.062 20 Lauraceae Cryptocarya sp. 3 26.5 0.055 25 Lauraceae Cryptocarya sp. 3 25.6 0.051 22 Violaceae Rinorea cf. arborea 25.0 0.049 14 Lauraceae Cryptocarya sp. 3 25.0 0.049 20 Note: records of Sloanea quadriloba refer to Sloanea rhodantha var. rhodantha form "quadriloba. 32 FIELDIANA: ZOOLOGY Table 4-5. Transect 3 (1,210 m). Trees with dbh >25 cm arranged in decreasing size. dbh Basal area Height Family Genus/species (cm) (m1) (m) Euphorbiaceae Cleistanthus boivinianus 70.0 0.385 20 Clusiaceae Calophyllum sp. 1 57.5 0.260 20 Lauraceae sp. 1 54.2 0.231 20 Clusiaceae Calophyllum drouhardi 47.0 0.173 22 Clusiaceae Calophyllum drouhardi 44.5 0.156 20 Aquifoliaceae Hex mitis 38.6 0.117 22 Euphorbiaceae Mallotus capuronii 35.4 0.098 16 Myrtaceae Syzygium sp. 2 35.2 0.097 17 Euphorbiaceae Cleistanthus boivinianus 34.0 0.091 16 Euphorbiaceae Cleistanthus boivinianus 31.4 0.077 16 Aquifoliaceae Ilex mitis 31.3 0.077 14 Myrtaceae Syzygium sp. 16 31.0 0.075 18 Araliaceae Schefflera myriantha 30.9 0.075 14 Annonaceae Polyalthia humbertii 30.5 0.073 20 Euphorbiaceae Cleistanthus boivinianus 30.1 0.071 17 Araliaceae Polyscias sp. 4 29.9 0.070 16 Euphorbiaceae Mallotus cf. capuronii 29.5 0.068 16 Rutaceae Evodia sp. 1 29.2 0.067 12 Icacinaceae sp. 3 27.9 0.061 14 Lauraceae sp. 2 27.9 0.061 15 Lauraceae Cryptocarya crassifolia 27.7 0.060 18 Mean values for estimates of bryophytes and lichens in tree crowns ranged from 0.6 (transect 1) to 3.0, the highest value possible, in transect 4, and on tree trunks from 0.4 (transects 1 and 3) to 2.6 (transect 4). Differences in these values were not statistically significant between transects 1 and 2, but the value for tree trunks was signifi- cantly higher for transect 4 than for transects 2 and 3 (P ^ 0.01). For tree crowns the value for transect 4 was significantly higher than that for transect 2 (P < 0.05) (Table 4-7). The abundance of epiphytic bryophytes and lichens increased with elevation. Casual observations suggested that epiphytes were locally abundant in certain areas of forest in all four transect zones. Local variation in micro- environmental factors such as shelter, humidity, and host plant characteristics also influence epi- phyte distribution. Forest Floristic Composition Diversity — The number of plant families and species found in each plot and the number of spe- cies in each plot new to the transect are given in Table 4-8; the statistical analysis of this informa- tion is given in Table 4-9. Forty-four plant species (dbh ^ 10 cm) were found in the plots in transect 1 (Table 4-10), 54 species in transect 2 (Table 4-11), 49 species in transect 3 (Table 4-12), and 51 species in transect 4 (Table 4- 1 3). A smaller number of families were also represented in transect 1 (20) than at any of the other three transects (26-28). However, the lower numbers of species and families represented in transect 1 compared to transect 2 were not sta- tistically significant (Table 4-9), due to the high level of variation among the plots. There was also no statistical difference in species or family di- versity with increasing elevation (Table 4-9). In order to determine whether the five 10 X 20 m plots were effective in sampling the flora of each elevational zone, the total cumulative num- bers of species per transect were plotted against plot number (Fig. 4-4). The extent to which the resultant species accumulation curves reach an as- ymptote indicates how adequately the five plots represent the entire floristic diversity of each tran- sect zone. On the basis of these plots, it is clear that the survey does not adequately represent the species diversity in any of the elevation zones, because all of the curves fail to show any sign of approaching a plateau at the fifth plot. Floristics — The contribution of each plant family to the vegetation in terms of numbers of stems and total basal area at each transect is given in Tables 4-10 through 4-13. Figures 4-5 and 4-6 show the relative abundance of plant families cal- culated from the number of stems present in a LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 33 Table 4-6. Transect 4 (1,625 m). Trees with dbh >25 cm arranged in decreasing size. Family Genus/species dbh (cm) Basal area (m2) Height (m) Lauraceae Violaceae Myrtaceae Lauraceae Lauraceae Lauraceae Araliaceae Sapotaceae Canellaceae Cunoniaceae Lauraceae Verbenaceae Lauraceae Rutaceae Lauraceae Sterculiaceae Lauraceae Lauraceae Ocotea sp. 2 Rinorea sp. 2 Syzygium sp. 9 Ocotea sp. 2 sp. 2 Ocotea sp. 2 Schefflera myriantha Faucheria sp. 1 Cinnamosma fragrans Weinmannia sp. 7 Ocotea sp. 2 Wtex sp. 1 Cryptocarya crassifolia Evodia sp. 2 Ocotea sp. 2 Dombeya sp. 2 Ocotea sp. 2 Cryptocarya sp. 1 67.5 0.358 14 57.5 0.260 16 56.0 0.246 11 54.7 0.235 15 51.4 0.207 15 47.0 0.173 11 35.0 0.096 12 33.5 0.088 12 33.4 0.088 13 32.5 0.083 12 31.3 0.077 12 30.6 0.074 14 29.8 0.070 13 29.5 0.068 14 27.8 0.061 14 27.0 0.057 15 26.0 0.053 12 26.0 0.053 12 100 I PL, Transect 1 (720 m) Transect 2 (810 m) Transect 3 (1210m) LZZ3 Transect 4 (1625 m) 10-14.9 15-19.9 20-24.9 25-29.9 30-34.9 35-39.9 40-44.9 45-49.9 >50.0 DBH size class range (cm) Fig. 4-1. Frequency by dbh (diameter at breast height) size class for trees of > 10 cm dbh in each transect. 34 FIELDIANA: ZOOLOGY u 120 100 80 60 40 20 I— "^ Transect 1 (720 m) LZH Transect 2 (810 m) H Transect 3 (1210 m) EI3 Transect 4 (1625 m) = LB <5 5-10 10-15 15-20 Tree height class (m) >20 Fig. 4-2. Frequency of trees of s 10 cm dbh in each height class in each transect. Table 4-7. Structural characteristics of the lianas and epiphytes in each transect zone, representing mean values of five plots per transect (mean values given ± standard deviation, with the minimum and maximum values in parentheses), with analyses of variance. Transect No. of lianas per tree Epiphytes in tree crown* Epiphytes on tree trunk* Moss and lichens Moss and lichens in tree crown* on tree trunk* 1 (710 m) 2 (810 m) 3 (1,210 m) 4(1,625 m) ANOVA 1/2+ ANOVA 2/3/4$ LSD§ 0.8 ±0.21 (0.6-1.1) 1.2 ±0.85 (0.1-2.2) 0.4 ±0.18 (0.2-0.6) 0.9 ±0.49 (0.3-1.4) NS NS 1.2 ±0.45 (1-2) 1.6 ±0.90 (1-3) 2.0 ±0.71 (1-3) 2.2 ± 0.45 (2-3) NS NS 0.6 ±0.55 (0-1) 1.0 ±0.71 (0-2) 0.6 ±0.55 (0-1) 1.0 ±0 (1-1) NS NS 0.6 ± 1.3(0-3) 1.2 ± 1.3(0-3) 2.0 ± 0 (2-2) 3.0 ±0 (3-3) NS ** 1.1 0.4 ± 0.89 (0-2) 0.8 ± 1.10(0-2) 0.4 ±0.55 (0-1) 2.6 ± 0.55 (2-3) NS *** 1.3 * Values attributed on a 0-3 scale (see text for explanation). + Results of the analysis of variance between transects 1 and 2: NS, not significant; **, significant at P ^ 0.05 level. t Results of the analysis of variance between transects 2, 3, and 4: NS, not significant; **, significant at P ^ 0.05 level; ***, significant at P ^ 0.01 level. § LSD, least significant difference between means (ANOVA 2/3/4). LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 35 1000 1200 Elevation (m) Fig. 4-3. Regression line showing tree height re- gressed against elevation for plots in transects 2 through 4. Calculated using mean tree height values for each plot. transect zone and the basal areas, respectively. In the forest outside the reserve, at 720 m (transect zone 1), the plant family with the largest number of stems was the Violaceae, comprising a single species, Rinorea cf. arborea, an understorey or canopy tree. Lauraceae (Cryptocarya and Ocotea) was the next most common family, followed by Myrtaceae (Syzygium), and Rubiaceae (Gaert- nera, Psychotria, and Tricalysia) (Table 4-10). The basal area data (Fig. 4-6) from transect zone 1 clearly reflect the status of Lauraceae as large canopy trees because they form 30.6% of the total basal area (Table 4-10). Although not found in any of the plots, Rav- enala madagascariensis Sonn. (Strelitziaceae) was frequent in the forest outside of the reserve. This species is common in Malagasy rain forest up to 1000 m elevation. It occurs primarily in ar- eas with significant rainfall, both in primary rain forest and open secondary growth. It invades dis- turbed forest and cleared areas (Kress et al., 1994). At 720 m (transect 1), R. madagascariensis was abundant in disturbed areas of forest. The forest inside the reserve boundaries at 810 m (transect 2) was similar in floristic composition to that found outside at 720 m (transect 1). The greatest number of stems were Violaceae (Rino- rea), followed by Lauraceae (Cryptocarya and Ocotea), Myrtaceae (Syzygium), and Rubiaceae (Gaertnera, Psychotria, and Tricalysia) (Table 4-11). The basal area data (Fig. 4-6) show that Elaeocarpaceae (Sloanea) formed nearly 20% of the total basal area. Sloanea are large, emergent, and often buttressed trees. Lauraceae were the second most important family, as estimated by to- tal basal area, forming almost 19% of the total. Sapindaceae formed 11% of the total basal area, though this can be traced to the occurrence of a single 75.8 cm dbh tree. Twelve families repre- sented in zone 2 were not found in transect zone Table 4-8. Family and species diversity at each plot. Percent of No. of No. of No. of species species new Transect Plot families species new to transect to transect 1 (720 m) 1 6 8 8 100 2 9 11 8 72.7 3 9 10 9 90.0 4 9 14 8 57.1 5 7 13 10 76.9 2 (810 m) 6 10 10 10 100 7 15 18 16 88.9 8 10 15 14 93.3 9 10 11 7 63.6 10 9 12 7 58.3 3 (1,210 m) 11 11 12 12 100 12 9 10 7 70.0 13 13 21 14 66.7 14 15 21 12 57.1 15 11 12 8 66.7 4(1,625 m) 16 13 14 14 100 17 10 16 12 75.0 18 15 21 13 61.9 19 10 14 5 35.7 20 14 14 8 57.1 36 FIELDIANA: ZOOLOGY Table 4-9. Family and species diversity of the trees in each transect zone, representing mean values of five plots per transect, with analyses of variance. Transect No. of families per plot* No. of species per plot*t 1 (720 m) 2 (810 m) 3 (1,210 m) 4(1,625 m) ANOVA 1/2$ ANOVA 2/3/4§ 8.2 ± 1.30(6-9) 10.2 ±2.78 (8-15) 11.0 ±2.12 (8-14) 12.4 ± 1.82(10-14) NS NS 11.2 ±2.39 (8-14) 13.2 ± 3.27(10-18) 15.2 ±5.36(10-21) 15.8 ±3.03 (14-21) NS NS * Mean values given ± standard deviation, with minimum and maximum values in parentheses. t Includes morphospecies. $ Results of the analysis of variance between transects 1 and 2: NS, not significant. § Results of the analysis of variance between transects 2, 3, and 4: NS, not significant. 1 , whereas only four that were represented in tran- sect zone 1 were absent in zone 2. In transect 3 (1210 m), Clusiaceae (Mammea, Symphonic and Calophyllum) had the largest number of stems; together with Euphorbiaceae (Croton, Cleistanthus, and Mallotus) they ac- counted for 43% of the trees sampled (Table 4-12, Fig. 4-3). Other important families included Lau- raceae (Ocotea and Cryptocarya), Myrtaceae (Sy- zygium), Rubiaceae (Schismatoclada, Gaertnera, Canthium, and Psychotria), and Araliaceae (Po- lyscias and Schefflera). The basal area data and stem data show the same trend, with Clusiaceae and Euphorbiaceae forming nearly 43% of the to- tal basal area. Elaeocarpaceae form 10% of the basal area, but only 4% of the total number of stems. Five families were represented that were not recorded in transect zones 1 and 2, whereas nine families were absent that were found at lower elevation. At transect 4 (1625 m) there was a dramatic shift in forest composition. Lauraceae formed 1 1 % of the total number of stems sampled. The other most important families were Podocarpa- ceae (Podocarpus), Araliaceae (Polyscias and Schefflera), Cunoniaceae (Weinmannia), and Pan- danaceae (Pandanus) (Table 4-13, Fig. 4-3). Eu- phorbiaceae were distinctly less abundant than at Table 4-10. Plant families in transect 1 (720 m) listed in descending order of stem number. Information on basal area and no. of species present is also provided. Percent Basal Percent Ranked by No. of of total Ranked by area of total No. of Family stem no. stems stem no. basal area (m2) basal area species Violaceae 1 34 28.6 2 0.47 17.7 1 Lauraceae 2 26 21.8 1 0.86 32.5 9 Myrtaceae 3 15 12.6 3 0.34 12.8 6 Rubiaceae 4 7 5.9 4 0.17 6.4 5 Clusiaceae 5 6 5.0 5 0.15 5.7 4 Monimiaceae 6 6 5.0 6 0.10 3.8 2 Elaeocarpaceae 7 4 3.4 7 0.08 3.0 1 Sapindaceae 8 4 3.4 9 0.06 2.3 1 Oleaceae 9 4 3.4 10 0.05 1.9 2 Sapotaceae 10 2 1.7 12 0.05 1.9 2 Apocynaceae 11 2 1.7 17 0.03 1.1 2 Not determined 12 0.8 8 0.07 2.6 1 Burseraceae 13 0.8 11 0.05 1.9 1 Tiliaceae 14 0.8 13 0.04 1.5 1 Cunoniaceae 15 0.8 14 0.04 1.5 1 Fabaceae 16 0.8 15 0.03 1.1 1 Sterculiaceae 17 0.8 16 0.03 1.1 1 Annonaceae 18 0.8 18 0.01 0.4 1 Melastomataceae 19 0.8 19 0.01 0.4 1 Araliaceae 20 0.8 20 0.01 0.4 1 119 100 2.65 100 44 LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 37 Table 4-11. Plant families in transect 2 (810 m) listed in descending order of stem number. Information on basal area and no. of species present is also provided. Percent Percent Ranked by No. of of total Ranked by Basal of total No. of Family stem no. stems stem no. basal area area (m2) basal area species Violaceae 1 17 15.9 7 0.20 4.5 1 Lauraceae 2 16 15.0 2 0.82 18.9 9 Myrtaceae 3 10 9.4 6 0.26 6.0 4 Rubiaceae 4 9 8.4 10 0.10 2.3 6 Clusiaceae 5 7 6.5 4 0.38 8.7 4 Cunoniaceae 6 6 5.6 5 0.35 8.0 2 Eleocarpaceae 7 4 3.7 1 0.86 19.8 1 Euphorbiaceae 8 4 3.7 12 0.07 1.7 2 Sapindaceae 9 3 2.8 3 0.47 10.8 3 Moraceae 10 3 2.8 11 0.09 2.1 1 Araliaceae 11 3 2.8 14 0.07 1.6 2 Anacardiaceae 12 3 2.8 15 0.07 1.6 1 Dracaenaceae 13 3 2.8 16 0.06 1.4 1 Sterculiaceae 14 2 1.9 8 0.12 2.8 2 Fabaceae 15 2 1.9 9 0.11 2.5 1 Cyathaceae 16 2 1.9 20 0.02 0.5 2 Icacinaceae 17 2 1.9 22 0.02 0.5 1 Monimiaceae 18 0.9 13 0.07 1.6 1 Sapotaceae 19 0.9 17 0.04 2.1 1 Rutaceae 20 0.9 18 0.04 2.1 1 Annonaceae 21 0.9 19 0.03 0.7 1 Tiliaceae 22 0.9 21 0.02 0.5 1 Dichapetalaceae 23 0.9 23 0.02 0.5 1 Myrsinaceae 24 0.9 24 0.02 0.5 1 Rhopalocarpaceae 25 0.9 25 0.01 0.3 1 Aquifoliaceae 26 0.9 26 0.01 0.3 1 Melanophyllaceae 27 0.9 27 0.01 0.3 1 Burseraceae 28 0.9 28 0.01 0.3 1 107 100 4.35 100 54 lower elevations. With regard to basal area, Lau- raceae formed 34%, reflecting their stature as can- opy species (Table 4-13). Four of the six largest trees found in transect zone 4 were Lauraceae (Ta- ble 4-6). Cyatheaceae, Pandanaceae, and Podo- carpaceae were comparatively unimportant in terms of basal area (Table 4-13, Fig. 4-6), reflect- ing their small size as understorey trees, even though they were relatively common in the plots. Five families were recorded in zone 4 that had not been found at lower elevations, and eight fam- ilies that were present in transect zone 3 were ab- sent from zone 4. TWINSPAN Canopy Data — The dendrogram obtained from the crown diameter data with TWINSPAN is giv- en in Figure 4-7a. The first-level division sepa- rated the low-altitude plots (transects 1 and 2) from the high/mid-altitude plots (transects 3 and 4), with the exception of plot 7, which was grouped with the mid-altitude plots. The second- level division within the mid/high-altitude cluster (including plot 7) partitioned the mid-altitude (transect 3) plots and plot 7 from the high-altitude (transect 4) plots. The third-level division divided plot 7 from the five plots of transect 3, and it separated plots 16 and 17 from plots 18, 19, and 20 of transect 4. A fourth-level division separated plots 1 1 and 12 from plots 13, 14, and 15 of tran- sect 3. Within the low-altitude plots (excluding plot 7), plots 2 and 1 were each separated from the rest by second- and third-level divisions, re- spectively. Plots 4 and 5 were separated from the rest by a fourth-level division, whereas plots 3, 6, and 10 were separated from plots 8 and 9 by a fifth-level division. dbh Data — The dendrogram obtained from the dbh data using TWINSPAN is given in Figure 4-7b. The first-level division separated perfectly the low-altitude plots (transects 1 and 2) from the high/mid-altitude plots (transects 3 and 4). The 38 FIELDIANA: ZOOLOGY Table 4-12. Plant families in transect 3 (1210 m) listed in descending order of stem number. Information on basal area and no. of species present in also provided. Percent Basal Percent Ranked by No. of of total Ranked by area of total No. of Family stem no. stems stem no. basal area (m1) basal area species Clusiaceae 1 31 22.3 1 1.18 24.0 4 Euphorbiaceae 2 29 20.9 2 0.92 18.7 3 Lauraceae 3 9 6.5 4 0.45 9.1 4 Myrtaceae 4 9 6.5 5 0.40 8.1 4 Rubiaceae 5 8 5.8 11 0.10 2.0 4 Araliaceae 6 7 5.0 6 0.26 5.3 4 Eleocarpaceae 7 5 3.6 3 0.50 10.2 2 Monimiaceae 8 5 3.6 10 0.13 2.6 1 Aquifoliaceae 9 4 2.9 7 0.25 5.1 1 Icacinaceae 10 4 2.9 9 0.13 2.6 1 Ebenaceae 11 4 2.9 12 0.07 1.4 1 Annonaceae 12 3 2.2 8 0.14 2.8 2 Sterculiaceae 13 3 2.2 14 0.06 1.2 1 Oleaceae 14 3 2.2 15 0.05 1.0 2 Sapindaceae 15 2 1.4 16 0.04 0.8 2 Cunoniaceae 16 2 1.4 17 0.03 0.6 2 Erythroxylaceae 17 2 1.4 18 0.02 0.4 2 Myrsinaceae 18 2 1.4 20 0.02 0.4 1 Rutaceae 19 1 0.7 13 0.07 1.4 1 Rhamnaceae 20 1 0.7 19 0.02 0.4 1 Loganiaceae 21 1 0.7 21 0.01 0.2 1 Flacourtiaceae 22 1 0.7 22 0.01 0.2 1 Anacardiaceae 23 1 0.7 23 0.01 0.2 1 Cyathaceae 24 1 0.7 24 0.01 0.2 1 Not determined 25 1 0.7 25 0.01 0.2 1 Melastomataceae 26 1 0.7 26 0.01 0.2 1 140 100 4.92 100 49 second-level division within the mid/high-altitude cluster divided perfectly the mid-altitude (transect 3) plots from the high-altitude (transect 4) plots. The third-level division separated plot 15 from plots 11, 12, 13, and 14 of transect 3, and it sep- arated plots 16 and 17 from plots 18, 19, and 20 of transect 4. Within the low-altitude plots, a sec- ond-level division divided plots 1, 2, and 5 from the rest. A third-level division separated plot 7, and a fourth-level division clustered plots 3, 4, and 6 together and plots 8, 9, and 10 together. Differential and Diagnostic Species — No dif- ferential species were recorded that separated the low-altitude plots (transects 1 and 2) from the mid/high-altitude plots (transects 3 and 4), even if the anomalous plot 7 was excluded. However, Ri- norea cf. arborea may be regarded as a diagnostic species, being present in seven of the 10 plots of transects 1 and 2 and absent from plots at tran- sects 3 and 4. Among plot clusters generated by TWINSPAN at lower division levels with either the canopy or the dbh data for the low-altitude transects, only the plot 8-9 cluster (crown data) had a differential species (Weinmannia sp. 1). Other plot clusters with diagnostic species were: plot cluster 3-4-10 (canopy data) Antidesma petiolare and Gaertnera sp. 1; plot cluster 1-2-5 (dbh data) Cryptocarya sp. 3; and plot cluster 8-9-10 (dbh data) Dalber- gia sp. 1, Syzygium sp. 2, and Weinmannia sp. 1. Differential and diagnostic species for the plots in transect 3 were Cleistanthus boivinianus (all five plots), Mallotus cf. capuronii (four plots), and Lauraceae sp. 1 (three plots). Among the plot clusters at higher division levels, only the plot 1 1- 13 cluster (canopy data) had differential species (Croton sp. 2 and Schismatoclada sp. 2). Differential and diagnostic species for the plots of transect 4 were Pandanus sp. 2 (all five plots), Weinmannia sp. 7 (four plots), Ocotea sp. 2 (three plots), Polyscias sp. 2, and the two diagnostic spe- cies of plot cluster 18-19-20 (listed below). Dif- ferential species for the two plot clusters that were obtained from both the canopy and the dbh data are: plot cluster 16-17 Casearia sp. 1, Croton sp. 1, and Pandanus sp. 2; and plot cluster 18-19-20 Podocarpus madagascariensis and Syzygium sp. 4. The numbers of differential species for the plots LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 39 Table 4-13. Plant families in transect 4 (1625 m) listed in descending order of stem number. Information on basal area and no. of species present is also provided. Percent Basal Percent Ranked by No. of of total Ranked by area of total No. of Family stem no. stems stem no. basal area (m2) basal area species Lauraceae 1 16 11.0 1 1.49 34.5 5 Podocarpaceae 2 13 9.0 9 0.16 3.7 1 Araliaceae 3 11 7.6 2 0.37 6.3 4 Cunoniaceae 4 11 7.6 4 0.27 6.3 3 Pandanaceae 5 10 6.9 12 0.10 2.3 2 Flacourtiaceae 6 9 6.2 7 0.19 4.4 2 Cyathaceae 7 8 5.5 14 0.09 2.1 1 Myrtaceae 8 7 4.8 3 0.33 7.6 3 Rubiaceae 9 7 4.8 10 0.16 3.7 6 Sterculiaceae 10 6 4.1 6 0.21 4.9 2 Euphorbiaceae 11 6 4.1 11 0.11 2.5 3 Asteraceae 12 5 3.5 17 0.06 1.4 1 Sapotaceae 13 3 2.1 8 0.17 3.9 1 Elaeocarpaceae 14 3 2.1 16 0.07 1.6 2 Icacinaceae 15 3 2.1 19 0.04 0.9 1 Monimiaceae 16 3 2.1 20 0.03 0.7 2 Rutaceae 17 2 1.4 13 0.10 2.3 2 Verbenaceae 18 2 1.4 15 0.08 1.9 1 Myrsinaceae 19 2 1.4 21 0.02 0.5 2 Dichapetalaceae 20 0.7 5 0.26 6.0 1 Aquifoliaceae 21 0.7 18 0.04 0.9 1 Clusiaceae 22 0.7 22 0.02 0.5 1 Erythroxylaceae 23 0.7 23 0.02 0.5 1 Melanophyllaceae 24 0.7 24 0.02 0.5 1 Canellaceae 25 0.7 25 0.01 0.2 1 Ericaceae 26 0.7 26 0.01 0.2 1 134 100 4.32 100 51 that were separated from all other plots in the analyses were as follows: plot 1 (canopy data), four species; plot 2 (canopy data), six species; plot 7 (canopy and dbh data), 10 species; and plot 15 (dbh data), three species. C 30- 3 2 u 20 'O transect 1 (720 m) transect 2 (810 m) transect 3 (1210 m) transect 4 (1625 m) 0 1 2 3 4 5 6 Plot number (replicate) Fig. 4-4. Species accumulation curves for transects 1 through 4 in the RNI d'Andringitra. DECORANA The ordinations generated by DECORANA for the dbh data are given in Figure 4-8a (plots) and 4-8b (species). Similar ordinations were obtained with the canopy (crown diameter) data. In both cases the five plots of transects 3 and the five plots of transect 4 are revealed as closely clustered points, but the 10 plots of transects 1 and 2 are not clearly separated and are much more scat- tered. Plots 7 and 8 are close to each other but are relatively far from the remainder of the tran- sect 1 and 2 plots. Plot 1 is shown to be an ex- treme outlier, although its spatial separation from the rest of the transect 1 and 2 plots is primarily with respect to the Y-axis; this is considered to be much less significant in DECORANA than separation against the X-axis (Gauch, 1982). The results of the canopy and dbh data differ only in subtle details. The canopy data placed plot 2 as an outlier among the transect 1 and 2 plots, whereas the dbh data placed plot 9 in an outlying position. The clustering of samples shows a close correspondence with the TWINSPAN results. 40 FIELDIANA: ZOOLOGY Olea SaPJ >< Myrt 15 others 25 720 m (119 trees) 810 m (107 trees) others 36 1210 m (140 trees) 1625 m (134 trees) Fig. 4-5. Number of stems for each family. Calculated on a per transect basis. The characteristic species for each cluster de- termined from DECORANA also show little dif- ference between the canopy and dbh data. Char- acteristic species for the transect 1 and 2 plot clus- ter (excluding the outlier plots 2 and 9 and the extreme outlier plots 1 , 7, and 8) are given below (differences between the results from canopy data and dbh data are indicated): Albizia gummifera Antidesma petiolare Calophyllum paniculatum LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 41 Olea others others 0.05 v 0.30__- 0.60^-n bbb^ Eleo fc&'.JJs mL'A ^k Laur Ik 0.81 v u Rubiy Faba o.iUfl 0.11 ^^^ MM ^^0.86 0.08 — -M *%kJ Ster /^ fc^\\// ^ Moni__ — — fl 0,12 feeLi sSS^ HHHffii^^ o.io K ^r. Myrt 1 ^^^mwi K^ Clus v= a^^H WW* ^ 0.26 HBB L ^H o.i5 yt ^r~ ' Euph^ 1 \^H ■ / ' ' V ^^B ••4' Rubi 0.27 ^ 1 ' ^'^^^ 0.17 X^£T^ ^^ Euph Cunc i >^3B 1- I 'V zS Myrt 0.34 ^^^ 0.51 0.35 Clus 0.38 *^ — Sapi 0.47 720 m 810 m Icac Moni 0.14 ^ 0. 1210 m Flac 0.19 Dich Myrt 0.26 Cuno 3'4 0.27 1625 m Fig. 4-6. Dominant plant families. Calculated from basal areas (mVO.l ha). Canarium madagascariense Cryptocarya madagascariensis Cryptocarya sp. 3 Cyathea sp. 2 Dichapetalum sp. 1 Ephippiandra sp. 1 Gaertnera sp. 1 Chrysophyllum sp. 1 Grewia sp. 2 Lauraceae sp. 3 Lauraceae sp. 5 Lauraceae sp. 1 1 Lauraceae sp. 12 Lauraceae sp. 13 Lauraceae sp. 14 Mammea sp. 3 (dbh data only) 42 FIELDIANA: ZOOLOGY Fig. 4-7a (top). Dendrogram obtained from TWIN- SPAN using canopy data. Fig. 4-7b (bottom). Dendrogram obtained from TWINSPAN using dbh data. Mammea sp. 4 Mammea sp. 5 Noronhia sp. 7 Oncostemum sp. 13 Rinorea cf. arborea Sapindaceae sp. 1 Syzygium sp. 5 (dbh data only) Syzygium sp. 6 Syzygium sp. 8 Syzygium sp. 1 1 Syzygium sp. 14 Syzygium sp. 17 Weinmannia sp. 5 The characteristic species of plot 2 are: 1000 800 000 400 200 0 -200 PI 1-15 PI6-20 ~&t (2\ W P2A10 -000 -400 50 cm dbh) in the plots in transect zone 4 than at any other zone, and that LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 47 the numbers of trees in many of the intermediate dbh size classes were comparable to those at tran- sect zone 2. The low numbers of large trees and the small size of the plots sampled suggest that there is a large random component involved. A single large tree could introduce a large bias into the results. It is also possible that selective felling or other disturbance has taken place in one or more of the transect zones 2, 3, and 4. Tree Height — Tree height was significantly lower outside the reserve at 720 m (transect 1) than inside the reserve at 810 m (transect 2). This also suggests that the largest trees have been se- lectively removed from the forest outside the re- serve. Tree height is one of the physical characteristics thought to be most affected by increasing eleva- tion. Significant differences were found between transect zones 2, 3, and 4, with mean height de- creasing from 15 m at 720 m to 8 m at 1625 m. The heights of the tallest trees within each tran- sect zone show similar trends; trees over 22 m were only found in transect 2, where they consti- tuted 6% of the trees recorded. In transect 4, the tallest tree was 16 m in height. Within a particular area, tree height is strongly influenced by envi- ronmental factors such as degree of exposure and substrate characteristics. For example, tall trees were found where the land was flat, whereas the trees on ridge tops were always somewhat stunted compared to those in the surrounding vegetation, regardless of elevation. Exposed and unstable ar- eas are more susceptible to wind damage and land slips; these phenomena will reduce the average age of the vegetation on these areas. The differ- ences in tree heights at the different elevations correspond closely with the characteristic main canopy levels of lowland, montane, and sclero- phyllous forest types, respectively (White, 1983). Crown Diameter — The visual estimate of this parameter could be made rapidly, but it was not precise. No decrease in crown diameter was de- tected with increasing elevation. This may be due to two things. First, crown diameter is extremely difficult to estimate from the ground because it is often obscured by understorey trees (our results may simply reflect inaccuracies). Second, there was also considerable variation within transect zones that may have masked statistically signifi- cant differences. Numbers of Lianas — There were no significant differences in the numbers of lianas found in the different elevation zones. From our observations it would appear that the presence of lianas is more a function of microhabitat than elevation. Epiphytes — The forest in transect zone 4 (1625 m) was particularly rich in epiphytic bryophytes and lichens, having a significantly greater quantity than the forests at lower elevations in both the tree crowns and on the tree trunks. This is a charac- teristic of sclerophyllous montane forest (White, 1983) that occurs at elevations where the plants are almost continually wet and the air is humid. However, there was also forest rich in epiphytic bryophytes and lichens in transect zones 2 and 3 in areas close to streams. The high humidity as- sociated with montane and cloud forest can also occur locally near streams at lower elevations. The effects of exposure related to elevation were in some cases similar to those at lower elevation due to aspect or site. For example, exposed low- altitude ridges often have flora and structure sim- ilar to that of high-altitude sheltered forest. Comparison with Other Forests — An assess- ment of structural differences between RNI d'Andringitra and other humid forests on the is- land can be done using a comparison of the den- sities and relative frequencies of different size classes of trees. The combined plots in each ele- vation zone in the RNI d'Andringitra total 0.1 ha, and by extrapolation there was an estimated range from 1,070 (at 810 m) to 1,400 (at 1210 m) stems per ha. This appears to be consistent with other Malagasy rain forests. At Vohiparara, PN de Ran- omafana (elevation 1100-1200 m), there was a total of 1,093 stems/ha (S. Malcomber, pers. comm.). Extrapolating from the results of the 0. 1 ha surveys done by Gentry (1993), there were about 1,220 stems per ha at Peri net and about 1,600 stems per ha at Nosy Mangabe. This con- trasts with much lower numbers in African moist forest, where an average of 590 trees per ha are found, and the Neotropics (moist and wet forest), which have an average of 640 trees per ha (Gen- try, 1993). Our results for basal area at 810, 1210, and 1625 m are higher than the 35 m2/ha found at Vatoharanana (1000-1100 m) in the PN de Ran- omafana (Schatz & Malcomber, in press), al- though the tree size distribution is consistent be- tween the two sites. Gentry (1993) reports that basal area values in Africa average 34 m2/ha and 35-36 m2/ha in the Neotropics. Gentry (1993) also noted that there was a comparatively high density of large trees (>1 m dbh) in Gabon, and such trees, even if they occur in low numbers, make up a large proportion of the basal area and 48 FIELDIANA: ZOOLOGY biomass of the forest. In the RNI d'Andringitra the maximum dbh of any tree measured in the plots was 81.1 cm. Thus, there were few very large trees in this forest relative to rain forests in mainland Africa. This was supported by casual observations outside the sample plots, and it may indicate a faster turnover time, due perhaps to high winds or substrate instability caused by high rainfall and/or shallow soil, or a lack of species capable of reaching very large proportions in the RNI d'Andringitra. No lianas with stems greater than 10 cm dbh were found in the RNI d'Andringitra. In the PN de Ranomafana, only two lianas with stems great- er than 10 cm dbh were encountered in the per- manent 1 ha plot (Schatz & Malcomber, in press). This compares with an average of 9.25 lianas of greater than 10 cm dbh in Gabonese plots (Reits- ma, 1988). Our results confirm the observation of Schatz & Malcomber (in press) that Malagasy for- est lacks the large lianas of some African rain forests. Gentry (1993) also noted unusually high densities of small dbh (^2.5 cm) lianas elsewhere in Malagasy forests (P6rinet and Nosy Mangabe). Structural differences between forests on dif- ferent continents are only presently being studied and analyzed, but they may have a profound im- pact on other organisms that occur in these habi- tats. For example, differences in liana size and density may have been a critical factor selecting for a variety of locomotor adaptations among can- opy vertebrates on different continents (Gentry, 1988). Diversity There was an overall reduction in numbers of tree species and there were fewer families in the forest outside the reserve (transect zone 1) in comparison to the forest within the reserve at nearly the same elevation (transect zone 2). How- ever, the high degree of variation between plots in each transect zone suggests that different com- munities exist with different levels of diversity at this elevation. This finding has important conser- vation implications, and we recommend that fol- low-up work be done to confirm the validity of this observation. The average number of species over 10 cm dbh in a 0.1 ha sample of primary forest at RNI d'Andringitra (transects 2-4) was 74 (Table 4-7). This compares with 76 at P6rinet and 80 at Nosy Mangabe (Gentry, 1993). Gentry (1993) also not- ed that African moist forests have an average of 33 species of trees of more than or equal to 10 cm dbh in a 0. 1 ha sample; in African dry forests there is an average of 21 species, and Malagasy dry forest (Ankarafantsika) has 28 species. This underlines the high diversity of Malagasy forests in comparison with those in Africa. Between 720 and 1625 m there was no appre- ciable reduction in floristic diversity with increas- ing elevation. In a study on Andean forests, Gen- try (1988) obtained a similar result for forests up to 1700 m; a clear decrease in diversity occurred above this elevation. Floristics In the forest at transect zones 1 and 2, Elaeo- carpaceae (Sloanea rhodantha var. rhodantha form "quadriloba") and Burseraceae (Canarium madagascariense ssp. obtusifolium) formed large forest emergents, and members of the Lauraceae, Myrtaceae, Violaceae (Rinorea), Monimiaceae (Tambourissa, Ephippiandra, and Decaryden- drori) were the most common canopy and under- storey trees. With increasing elevation there was a change in floristic composition. For example, Canarium reached its upper limit at approximately 1000 m; similarly, Sloanea rhodantha var. rhodantha form "quadriloba" was replaced by S. rhodantha var. rhodantha form "quercifolia" between 810 and 1210 m. This form of Sloanea rhodantha is readily distinguished from the form "quadrilo- ba" by its smaller leaves with serrate margins and acute acuminate apices, whereas the latter has larger, entire leaves with obtuse to emargin- ate non-acuminate leaves. Tirel (1985) reported the widespread occurrence in the eastern Mala- gasy region of a boundary at about 1200 m be- tween different forms of S. rhodantha var. rho- dantha. Although these forms appear to be suf- ficiently distinct to merit specific rank, Tirel not- ed the existence of intermediates in some areas and has discussed the complex variation in the species (1985). Pending a more detailed taxo- nomic investigation, it is advisable to provision- ally accept Tirel's conclusions. Between 810 and 1210 m there were also some changes in family composition, with Clu- siaceae becoming more prominent above 1000 m; in particular, the genera Symphonia and Gar- cinia beame more abundant at higher elevations. These findings are consistent with those of Gen- LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 49 try (1988), who speculated that 1000 m marks a transition from low- to mid-elevation forest in Madagascar. In their study of a 1 ha plot at Vatoharanana, PN de Ranomafana (elevation 1000-1100 m), based on numbers of stems, Schatz and Malcom- ber (in press) found the Monimiaceae to be the most abundant family; together with the Laura- ceae it accounted for 25% of the stems. Other prominent families at Vatoharanana include the Myrtaceae, Elaeocarpaceae, Sterculiaceae, and Clusiaceae. In comparison, the floristic composi- tion of the forests on the eastern slopes of the RNI d'Andringitra at 800-1200 m (transect zones 2 and 3) differs in having fewer Monimiaceae but more Euphorbiaceae; otherwise it is similar, with the Lauraceae well represented. Schatz and Mal- comber (in press) also found that over half of the largest trees (>50 cm dbh) belonged either to Elaeocarpaceae (Sloanea) or Lauraceae (Crypto- carya and Ocotea). This agrees with our results from the 810 m zone. Between 1210 and 1625 m another shift in flo- ristic composition was detected, perhaps marking a change from mid- to high-elevation forest. This higher elevation forest (surrounding transect zone 4) is characterized by Podocarpus, Weinmannia, Pandanus, Cyathea, and Asteraceae. The lowest elevational record of Vaccinium (Ericaceae) was in this zone. According to White (1983), the following flo- ristic characteristics help to differentiate lowland, montane, and sclerophyllous forests: (1) Rinorea and Canarium are common in lowland forest. (2) Melastomataceae are poorly represented in scle- rophyllous forest. (3) Araliaceae, Asteraceae, Er- ythroxylaceae, Verbenaceae, and Podocarpus are usually more common at higher altitudes. (4) Clu- siaceae, Fabaceae, Myrtaceae, Rubiaceae, and Sapindaceae are generally rarer in sclerophyllous forest. (5) Ephippiandra (Monimiaceae) is en- demic to moist montane and sclerophyllous for- ests. In general, forests within the three elevational zones sampled in the RNI d'Andringitra, corre- spond to the three forest types in White's (1983) classification. However, it should be noted that transects 1 and 2 lie near the upper elevational limit of the lowland forest and transect 4 lies near the lower elevational limit of the sclerophyllous forest. Additionally, we propose the following modifications: (1) Ephippiandra is not restricted to Malagasy high elevation forest, because it was found at 800 m (transect zone 2) and has been recorded at 300 m on the Masoala Peninsula (G. Schatz, pers. comm.). (2) Melastomataceae, which are underrepresented in the plot data because the majority of species do not produce woody stems of more than 10 cm diameter, were as well rep- resented in transect zone 4 as elsewhere. (3) There was no appreciable decrease in the quantity of Myrtaceae and Rubiaceae in transect zone 4. The following species were recorded at plots in all three altitudinal bands; they are therefore eco- logical transgressors in the forests of the eastern slopes of the RNI d'Andringitra: Calophyllum drouhardii Ilex mitis Lauraceae sp. 2 Schefflera myriantha Sloanea rhodantha As in other Malagasy rain forests, few Faba- ceae occur in the forests of the RNI d'Andringitra. In the sample plots, the family comprised only 0.6% of the stems and 0.86% of the total basal area. This is in marked contrast to African and Neotropical moist forests, and Malagasy dry for- ests, where Fabaceae are virtually always the dominant family (White, 1983; Gentry, 1988). In southeast Asian rain forests, Dipterocarpaceae are the dominant family. In Malagasy rain forest, the situation regarding dominance is not so clear, due to floristic variability along elevational gradients. However, taken as a whole, Lauraceae is the dom- inant family, both in terms of numbers of stems and total basal area. In the forests of the eastern slopes of the RNI d'Andringitra, 13.4% of the stems belonged to species of Lauraceae, and these comprised 22.29% of the total basal area. The eastern slopes of the RNI d'Andringitra are poor in palm species. Environmental Influences Elevation — All three methods of multivar iate analysis showed a clear correlation betweei the plot data and plot elevation. Although th< plots at 720 and 810 m, both of which include( both disturbed and undisturbed forest, exhibited a wide range of variation in composition, the? were in general quite distinct from those a higher altitudes. Plots at 1210 and 1625 m wer< strongly clustered, indicating relatively unifom composition in both zones, particularly at 162 m. 50 FIELDIANA: ZOOLOG' These results confirm elevation as the most important natural environmental variable in the forests of the study area. The strong clustering of the data is not surprising, however, because plot elevation was the primary basis for sample stratification. If plots were to be sampled at in- termediate elevations, the discrete clusters of points obtained for each transect zone would probably be linked. Disturbance — Disturbance was revealed by CANOCO to be the most significant environmen- tal variable influencing the composition of the for- ests. Clearly, disturbance is a complex phenome- non; its nature and history are both critical factors in correctly interpreting its consequences. Reduc- ing disturbance to a single binary character — "disturbed" or "not disturbed" — based on sub- jective visual assessment is a gross oversimplifi- cation. The results of TWINSPAN and DECOR- ANA confirm that the plots at 720 and 810 m do not clearly fall into two clear categories based on their perceived disturbance, and in some cases clustering of points between the two zones is stronger than within each zone. Three factors should be considered in this regard: (1) the extent and nature of the disturbance may vary; (2) plots at 810 m did show some signs of disturbance; and (3) the analyses may have detected different plant communities that exist at that altitude and have remained somewhat similar in composition, de- spite disturbance to some. The importance of each of these factors could only be determined by more detailed study. Other Environmental Variables — The envi- ronmental variables aspect, slope angle, and to- pography appear to be much less influential to the vegetation than elevation and disturbance. Sample plots that are similar with respect to these vari- ables do not appear to be particularly similar bo- tanically (compare plots 5, 12, 16, 18, and 20). A more structured approach to sampling within each elevational band with respect to selection of plots with similar aspect, slope, and topography would permit a better assessment of the influence of these variables. It must be presumed that aspect, slope, and topography may be partly responsible for the differences between plots within each tran- sect zone, but more plots would need to be sam- pled in each transect to test this adequately. Other unmeasured environmental variables, such as mi- croclimate and soil characteristics, may be partly responsible for differences within and between the transect zones. Plant Communities The effects of increasing elevation on local microclimate are largely responsible for the dif- ferent vegetation types found on a mountain. With increasing elevation, mean temperatures decrease, and humidity, mist, and cloud cover in- crease; the latter, in turn, reduce the amount of light reaching the vegetation. One of the aims of multivariate analysis in plant ecology is to rec- ognize distinct associations and communities of species and the hierarchical relationship between them, based on the similarity between sample plots. A second aim is to assess the correlation of measured environmental variables with these communities. At present no standard systematic classification of the vegetation of Madagascar at the community level is available, in large part owing to the pau- city of basic research for most parts of the island. In the present study the analyses have clearly re- vealed clusters of similar plots among the 20 plots sampled, and the characteristics of each cluster may represent specific plant communities. The sample plots tended to cluster most closely with other plots in the same elevational band, suggest- ing that different plant communities occur in each elevational zone. The present analysis suggests that these communities are quite distinct in nature, but they may simply represent sets of samples along an elevational gradient, rather than truly distinct communities. Plots at 720 and 810 m have produced some- what confusing results; the effects of disturbance have distorted the pattern of primary vegetation communities. The results of the multivariate anal- yses of associations of plant species on the eastern slopes of the RNI d'Andringitra can be summa- rized as follows. ( 1 ) Rinorea cf. arboreal Sloanea rhodantha var. rhodantha form "quadriloba" forest (plots 3, 4, 5, 6, 9, and 10; 750-820 m). This forest is char- acterized by the dominance of Rinorea cf. arbo- rea. It is the most important species in terms of numbers of individuals (12 out of 28 trees in plot 4), and it made the highest total contribution to the plot with respect to crown diameter and dbh. Individual trees of this species were never large, with crown diameter rarely exceeding 3 m and dbh less than 20 cm. Sloanea rhodantha form "quadriloba" is associated with Rinorea cf. ar- borea, although it was absent from plots 9 and 10. In all plots, Sloanea was less important in terms of number of individuals and total crown LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 51 and dbh measurements than Rinorea, except in plot 6, where it waas codominant. In plot 9, Wein- mannia sp. 6 is codominant with the Rinorea; elsewhere this species is only found in plot 8. In plot 5, Syzygium sp. 6 was numerous and was the second most important species; it was not record- ed in any other plots at any altitude. This forest type occurred in a range of situations, and this environmental diversity may account for some of the differences recorded between each of the sam- ple plots. The results of the CANOCO analysis suggest that this community corresponds to some degree of disturbance. In plots within the same elevational band that the analysis reveals as un- disturbed, Rinorea cf. arborea is absent (plot 7), infrequent (plot 8), or only codominant (plot 9). It may be a species that proliferates after mild disturbance. (2) Lauraceae sp. 16 forest (plot 1; 715 m). The vegetation at plot 1 was overwhelmingly domi- nated by this species, which was not recorded elsewhere. In plot 1 there was Rinorea/Sloanea, the widespread Xylopia sp. 1, and three other spe- cies unique to the plot. Species diversity was the lowest of all the plots, with only eight species recorded; this result may be the result of distur- bance. The CANOCO analysis supports this con- clusion, although plot 1 contained some relatively large individual trees. The Lauraceae sp. 16 forest occurred on a southwest-facing 35° slope at the bottom of a valley, an environmental setting sim- ilar to that of plot 2. (3) Lauraceae sp. 10 forest (plot 2; 740 m). The vegetation at plot 2 was dominated by this spe- cies, with Tambourissa thouvenotii and Syzygium sp. 5 of secondary importance. These and some of the other species also occur in Rinorea cf. ar- borealSloanea forest, although six species were unique to plot 2. The dominant Lauraceae sp. 10 was found only in plots 2 and 7, both valley bot- tom plots. (4) Sloanea rhodantha var. rhodantha form "quadriloba"/Lauraceae sp. 10 forest (plot 7; 810 m). The vegetation of plot 7 was dominated by Sloanea rhodantha form "quadriloba" and Lau- raceae sp. 10. However, few other species of ei- ther the Rinorea/Sloanea or the Lauraceae sp. 10 were recorded. Calophyllum drouhardii was of secondary importance; it occurred mainly in for- ests at higher elevations. Three other species found at higher altitudes, Ilex mitis, Schefflera my- riantha, and Sapindaceae sp. 4, were also present, but 10 of the species recorded at plot 7 did not occur elsewhere. Plot 7 was valley bottom forest on level ground, and although its environmental setting was similar to that of plot 10, no species were common to both plots. (5) Syzygium sp. 2 forest (plot 8; 860 m). The vegetation in plot 8 was dominated by Syzygium sp. 2, although Dalbergia sp. 1, Syzygium sp. 3, and Calophyllum drouhardii were also important. Syzygium sp. 2 and C. drouhardii are important components of the forest in transect zone 3, whereas Dalbergia sp. 1 and Syzygium sp. 3 were also found only in Rinorea/Sloanea forest in plot 6. Four of the species at plot 8 were not recorded elsewhere. The Syzygium sp. 2 forest was on a southwest-facing ridge forest; in this exposed sit- uation it would be expected that some species of higher altitudes might be present. Although plot 3 is in a similar situation, plots 3 and 8 only had two species in common, Sapindaceae sp. 1 and Syzygium sp. 14. (6) Cleistanthus boivinianus forest (plots 11, 12, 13, 14, and 15; 1220-1310 m). The plots at 1210 m were characterized by the presence of Cleistanthus boivinianus in all five plots, Mallotus cf. capuronii in four of the five plots, and Lau- raceae sp. 1 in three of the five plots. These spe- cies were absent from plots at all other elevations. Cleistanthus boivinianus was dominant only in plot 14; it was codominant with Ilex mitis in plot 15, whereas the other two species were never present in more than moderate quantity. In plots 11 and 12, Garcinia sp. 1 was dominant, com- prising 18 of the 52 trees recorded in the two plots. In plot 13, Calophyllum drouhardii was dominant, but the Cleistanthus was also impor- tant. Calophyllum drouhardii appears to be a common species with a wide environmental tol- erance, occurring in forest at each altitude of dif- ferent aspect, slope, and topography, and also in disturbed forest. (7) Weinmannia sp. 1/Ocotea sp. 2/Podocarpus madagascariensis forest (plots 16, 17, 18, 19, and 20; 1550-1660 m). The forest around 1625 m ap- pears to be a mosaic of species. Only Pandanus sp. 2 occurred in all five plots, but it was never prevalent. The following species are also impor- tant: Weinmannia sp. 7 (four plots), Ocotea sp. 2 (three plots), and Podocarpus madagascariensis (three plots). Each of these species is dominant or codominant in each of the plots except plot 16, where the Weinmannia occurs, but is much less important than certain other species, notably Ma- caranga echinocarpa and Asteraceae sp. 1. Al- though this plot has seven species that were not recorded elsewhere, the analyses suggest that it 52 FIELDIANA: ZOOLOGY should not be separated from the other plots at 1625 m. Methodological Improvements On the basis of the results obtained from the eastern slopes of the RNI d'Andringitra, several recommendations can be made to improve the methodology for botanical rapid assessment stud- ies. The high degree of variation between plots in each transect zone, and the high proportion of pre- viously unrecorded species found in the final plot sampled in each zone, indicate that plot size and/ or the number of plots sampled were inadequate. The minimum plot size and sample number that will provide an acceptable result for a particular type of vegetation depends on the structural and floristic variability of the vegetation and the type of study that is being undertaken. As discussed in the introduction, it was unreasonable to aim for complete or near complete sampling of plant communities as diverse and complex as a Mal- agasy rain forest during a rapid assessment. Rather, a relative measure of diversity and de- scription of the main communities, at least for comparative purposes, can be obtained by sam- pling a proportion of the vegetation. For the pur- poses of rapid assessment, the standardization of methods is desirable, and it is recommended that the exercise of determining optimum plot size using the established method of sampling nested quadrats would improve the reliability of future attempts at rapid assessment of forest commu- nities in Madagascar. This approach should be coupled with a study to determine the minimum number of plots required to provide an adequate vegetation survey. The stratification of sampling on the basis of disturbance and elevation served to highlight the importance of these environmental variables in the forest of the Eastern Malagasy Region. How- ever, a more structured approach to the sampling of vegetation on different aspects, slopes, and to- pography (and other environmental variables) would have yielded more complete information about the influence of these variables. Thus, every I effort should be taken to sample an equivalent set of sites, within each elevational zone, subject to the constraints of the landscape. Summary The vegetation in the study area (720-1625 m) in the RNI d'Andringitra is a complex mixture of forest types and individual plant communities; this study confirms the exceptional diversity of forest in the Eastern Malagasy Region. Structural and floristic differences were found in forests out- side (720 m) and inside the reserve (810 m). These differences are probably due to disturbance, although the results suggest that disturbance of the forest has occurred within the reserve bound- aries, at least within some plots. The forest out- side the reserve had fewer large trees, indicating that the largest individuals had been selectively removed, and a lower floristic diversity than plots within the reserve. In the three elevation bands sampled (810, 1210, and 1625 m), the forest had distinctive structural and floristic features. With increasing elevation the canopy height decreased and the amount of hanging moss and lichen increased. These were the only structural parameters mea- sured that were significantly correlated with in- creasing elevation. There was no indication that overall floristic diversity decreased with elevation within the range sampled, although there were marked changes in the floristic composition with increasing elevation. The forest, at ±800, ±1200, and ±1600 m, corresponded closely with White's (1983) lowland, montane, and sclerophyllous montane forest types, respectively. Although un- doubtedly influencing the vegetation at each sam- ple plot, the slope, aspect, and topography showed no correlation with the composition of the forest across the elevational gradient, based on crown and dbh measurements. Our results on the floristic diversity of the for- ests on eastern slopes of the RNI d'Andringitra agree with those for certain other rain forest ar- eas in Madagascar in demonstrating extremely high species richness compared with similar for- est types in Africa. It is hoped that this study will stimulate further work on the diversity and structure of Malagasy rain forests, and that it will contribute toward the development of a system- atic classification of the vegetation at a commu- nity level. We also hope that this study will add information to the larger picture of the structural differences and similarities between forests on different continents. Additionally, we hope to stimulate discussion as to the best methodology for rapid assessment studies of rain forest areas. LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 53 Acknowledgments We thank DEF and ANGAP for permission to work in the RNI d'Andringitra, and we thank WWF for providing logistical support. We are grateful to Albert Randrianjafy (Director, PBZT) and Philippe Morat (Director, Museum National D'Histoire Naturelle, Paris) for providing research facilities. We also thank Jeanine Raharilala (PBZT) for her work in the field and help with field identifi- cations. We are grateful to Tony Palmer (Range and Forest Institute) and Ted Avis and Rachel Judd (Rhodes University) for their advice and as- sistance with the multivariate analyses. We are particularly indebted to Steven Goodman for ex- tensive comments on earlier drafts, and to Pete Lowry, George Schatz, Simon Malcomber, Mich- elle Zjhra and three anonymous reviewers for their comments and suggestions on the manu- script. The work was supported through a grant from KfW, the Claiborne/Ortenberg Foundation, and the John D. and Catherine T MacArthur Foundation. Literature Cited Barbour, M. G., J. H. Burk, and W. D. Pitts. 1980. Terrestrial plant ecology. Benjamin/Cummings Pub- lishing Company, Menlo Park, California. Brummitt, R. K. 1992. Vascular plant families and gen- era. Royal Botanic Gardens, Kew. Du Puy, B., J. P. Abraham, and A. J. Cooke. 1994. Les plantes, pp. 15-29. In Goodman, S. M., and O. Langrand, eds., Inventaire Biologique Foret de Zom- bitse. Recherches pour le Developpement, Serie Sci- ences Biologiques no. Special. Centre d' Information et de Documentation Scientifique et Technique, An- tananarivo. Gauch, H. G. 1982. Multivariate analysis in community ecology. Cambridge University Press, Cambridge, U.K., 298 pp. Gentry, A. H. 1982. Patterns of neotropical plant spe- cies diversity. Evolutionary Biology, 15: 1-84. . 1988. Changes in plant community diversity and floristic composition on environmental and geo- graphical gradients. Annals of the Missouri Botanical Garden, 75: 1-34. . 1993. Diversity and floristic composition of lowland tropical forest in Africa and South America, pp. 500-547. In Goldblatt, P., ed., Biological relations between Africa and South America. Yale University Press, New Haven, Connecticut. Guillaumet, J.-L. 1983. Forets et fourres de montagne a Madagascar. Candollea, 38: 481-502. Hell, M. O. 1979a. TWINSPAN— A Fortran program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. Cornell University, Ithaca, New York, 48 pp. 1979b. DECORANA— A Fortran program for detrended correspondence analysis and reciprocal averaging. Cornell University, Ithaca, New York, 30 pp. Humbert, H. 1955. Les territoires phytogeographiques de Madagascar. In Les divisions ecologiques du mon- de. Colloques Internationale du Centre National de Recherche Scientifique LIX. Annee biologique, sen 3, 31(5-6): 439-448. Humbert, H., and G. Cours Darne. 1965. Notice de la carte Madagascar. Centre National de la Recherche, Scientifique et de l'Office de la Recherche Scienti- fique et Technique Outre-Mer, 164 pp. Jenkins, M. D., ed. 1987. Madagascar. An environmen- tal profile. International Union for the Conservation of Nature, Gland, 374 pp. Koechlin, J., J. L. Guillaumet, and P. Morat. 1974. Flore et v6getation de Madagascar. J. Cramer, Vaduz, Liechtenstein, 687 pp. Kress, W. J., G. E. Schatz, M. Andrianifahanana, and H. Simons Morland. 1994. Lemur pollination in Madagascar. American Journal of Botany, 8: 542- 551. Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, and P. Griveaud. 1971. Etude des ecosystemes mon- tagnards dans la region Malgache. I. — Le massif de l'Andringitra. 1970-1971. G6omorphologie, climato- logie et groupements vegetaux. Bulletin de la Societe d'Ecologie, 11(2-3): 198-226. Perrier De La Bathie, H. 1921. La vegetation mal- gache. Annales du Musee Colonial de Marseille, ser. 3, 9: 1-268. . 1927. Le Tsaratanana, l'Ankaratra et l'Andrin- gitra. Memoires de l'Academie Malgache, 3: 1-71. Phillipson, P. B. 1994. Madagascar. In Davis, S. D., V H. Heywood, and A. C. Hamilton, eds. Centres of plant diversity: A guide and strategy for their conser- vation. Volume 1: Europe, Africa, South West Asia and the Middle East. IUCN Publications Unit, Cam- bridge, U.K. Reitsma, J. M. 1988. Vegetation forestiere du Gabon. Tropenbos Technical ser. 1: 1-142. Schatz, G. E., and S. T. Malcomber. In press. Floristic composition of one hectare plots in Ranomafana Na- tional Park. In Wright, P. C, ed., Biodiversity in Ran- omafana National Park, Madagascar. State University of New York Press, Albany. Ter Braak, C. J. F 1986. Canonical correspondence analysis: A new eigenvector technique for multivariate direct gradient analysis. Ecology, 67: 1167-1179. Tirel, C. 1985. Elaeocarpaceae. Flore de Madagascai et des Comores. 125. Museum National d'Histoire Na- turelle, Paris, 53 pp. Watt, T. A. 1993. Introductory statistics for biolog) students. Chapman & Hall, London, 185 pp. White, F 1983. The vegetation of Africa. UNESCO Paris, 356 pp. Whittaker, R. H. 1977. Evolution of species diversity in land communities. Evolutionary Biology, 10: 1-87 54 FIELDIANA: ZOOLOGY Appendix 4-1. Checklist of Vascular Plant Species Recorded from Andringitra This checklist is arranged in three parts: Ferns and Fern Allies, Gymnosperms, and An- giosperms, each ordered alphabetically by plant family (following Brummitt, 1992, with irreg- ular names corrected). Localities given in pa- rentheses refer to information accompanying the relevant taxonomic descriptions of taxa. In- formation obtained from Perrier de la Bathie (1927) on endemics recorded from Andringitra is cited, and species mentioned by Paulian et al. (1971) are cited with relevant page numbers. Nomenclature is brought up to date as far as possible. Lewis specimens with numbers in pa- rentheses preceded by "s" are sterile voucher specimens. Voucher specimens are generally lodged at TAN and MO. Morphospecies without generic allocation are placed immediately be- low each family name. This list was generated with the help of the Conspectus database pro- ject, coordinated by George Schatz, Missouri Botanical Garden, St. Louis, Missouri. Several specimens are cited from the Reserve Naturelle (RN) series. For the sake of conciseness the 1993 collections are designated by camp num- ber. The precise localities for each transect zone are: Camp 1: Approximately 45 km S Ambalavao, east bank Iantara River, along the Ambala- manenjana-Ambatamboay trail, 22°13'20"S, 47°0r29"E, 720 m. Camp 2: Approximately 43 km S Ambalavao, junction of the Sahanivoraky and Sahavatoy rivers, 22°13'40"S, 47°00'13"E, 810 m. Camp 3: Approximately 40 km S Ambalavao along tributary of Sahavatoy River, 22°13'22"0S, 46°58'18"E, 1210 m. Camp 4: Approximately 38 km S Ambalavao, on ridge above headwaters of Sahavatoy Riv- er, 22°H'39"S, 46°58'16"E, 1625 m. Part 1: Ferns and Fern Allies Adiantaceae Pityrogramma humbertii C. Chr. — "Rochers de la zone sommitale 2300-2650 m" (Paulian et al., 1971, p. 248). Pteris cretica L. — Approximately 5 km SE An- tanifotsy, the reserve headquarters, 1500 m: Schatz 2662, with P. Goldblatt, A. Rakotozafy, and J. Randrianasolo. Aspleniaceae Asplenium marojejyense Tardieu — 50 km S Am- balavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 257. Asplenium poolii Baker — 50 km S Ambalavao, near abandoned meteo station above Ambala- marina, high altitude moss/lichen rain forest- ericaceous bush, Philippia dominant, 1975 m: Nicoll 258. Asplenium preussii Hieron. ex Brause — Approxi- mately 5 km SE Antanifotsy, the reserve head- quarters, 1500 m: Schatz 2661, with P. Gold- blatt, A. Rakotozafy, and J. Randrianasolo. Asplenium rutifolium (Berg.) Kunze — 50 km S Ambalavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia domi- nant, 1975 m: Nicoll 254. Asplenium sandersonii Hook. — Approximately 5 km SE Antanifotsy, the reserve headquarters, 1500 m: Schatz 2657, with P. Goldblatt, A. Rakotozafy, and J. Randrianasolo. Cyatheaceae Cyathea dregei Kuntze — "V6g6tation liee a la presence d'eau 2000-2300 m" (Paulian et al., 1971, p. 243). Cyathea sp. 1 — Camp 4: Lewis (s 407). Cyathea sp. 2 — Camp 2: Lewis (s 127). Cyathea sp. 3 — Camp 2: Lewis (s 184). Grammitidaceae Cheilanthes horizontalipinnata. Bonap. — Rocail- les pres de la cime (Perrier de la Bathie, 1927). Xiphopteris hildebrandtii (Hieron.) Tardieu- Camp 3: Lewis 971. LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSfTION, AND DIVERSITY 55 Hymenophyllaceae Hymenophyllum compactum Bonap. — Foret a mousses, vers 2000 m, versant Est (Perrier de la Bathie, 1927). Hymenophyllum sp. 1 — Foret a mousses, vers 2000 m, versant Est (Perrier de la Bathie, 1927). Trichomanes angustilaciniatum Bonap. — Rocail- les vers 2000 m versant Est (Perrier de la Bath- ie, 1927). specified) — "Foret a strate herbacee [below] 1970 m" (Paulian et al., 1971, p. 254). Podocarpus madagascariensis Baker var. mada- gascariensis — Camp 3: Lewis 941. Podocarpus madagascariensis Baker var. rotun- dus Laurent — Camp 4: Lewis (s 432). Part 3: Angiosperms Acanthaceae Lomariopsidaceae Elaphoglossum achroalepis (Baker) C. Chr. — Epiphytic on riverine trees, Camp 2: Lewis 877. Lycopodiaceae Lycopodium cernuum L. — 50 km S Ambalavao, near abandoned meteo station above Ambala- marina, high altitude moss/lichen rain forest- ericaceous bush, Philippia dominant, 1975 m: Nicoll 264. Lycopodium clavatum L. — "Hauts fourre arbus- tifs 2000-2300 m" (Paulian et al., 1971, p. 241); "Foret dense sclerophylle de montagne 2000 m" (Paulian et al., 1971, p. 250). Polypodiaceae Loxogramme lanceolata (Sw.) C. Presl — 50 km S Ambalavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia domi- nant, 1975 m: Nicoll 256; approximately 5 km SE Antanifotsy, the reserve headquarters, 1500 m: Schatz 2660, with P. Goldblatt, A. Rakoto- zafy, and J. Randrianasolo. Pleopeltis excavata (Bory ex Willd.) Sledge — 50 km S Ambalavao, near abandoned meteo sta- tion above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Phi- lippia dominant, 1975 m: Nicoll 260. Part 2: Gymnosperms Podocarpaceae Podocarpus madagascariensis Baker (var. not Mendoncia flagellaris Benoist — Camp 1: Lewis 759. Mendoncia sp. 1 — Camp 3: Lewis 944. Aloaceae Aloe andringitrensis H. Perrier (Madagascar: tres commune, aussi bien sur le versant Ouest, que sur le versant Est); Rocailles, de 2000 mala cime (Perrier de la Bathie, 1927); "Vegetation rupicole 2000-2300 m" (Paulian et al., 1971, p. 242). Aloe capitata Baker "Vegetation rupicole 2000- 2300 m" (Paulian et al., 1971, p. 242). Aloe decorseii H. Perrier (Madagascar: massif d'Andringitra) — Rocailles, vers 1700 m, ver- sant Sud-Ouest Est (Perrier de la Bathie, 1927). Aloe haworthioides Baker var. aurantiaca H. Per- rier— Rocailles, vers 2000 m, versant Est (Per- rier de la Bathie, 1927). Aloe macroclada Baker "Vegetation rupicole 2000-2300 m" (Paulian et al., 1971, p. 242). Anacardiaceae Micronychia macrophylla H. Perrier — Camp 1: Lewis 803. Micronychia madagascariensis Oliv. — Camp 3: Lewis (s 311), 961. Micronychia tsiramiramy H. Perrier — Camp 3: Lewis 778, 976. Protorhus sp. 1 — Camp 2: Lewis (s 138, 140) 881, 912. Rhus taratana (Baker) H. Perrier — Camp 1 : Lew- is 805. Rhus thouarsii (Engl.) H. Perrier — Camp 1 : Lewi; 778; Camp 3: Lewis 976. 56 FIELDIANA: ZOOLOGY Annonaceae Artabotrys mabifolius Diels — Camp 2: Lewis 844. Polyalthia humbertii Cavaco & Keraudren — Camp 3: Lewis (s 341, 343). Xylopia sp. 1 — Camp 1: Lewis (s 7); Camp 2: Lewis (s 190); Camp 3: Lewis (s 259). Anthericaceae Arthropodium caesioides H. Perrier — "Prai- ries/prairies altimontaines 1700-2000 m" (Pau- lian et ah, 1971, pp. 236, 256); "Foret dense scleYophylle de montagne [above] 1950 m" (Paulian et ah, 1971, p. 256). Apiaceae sp. 1 — Camp 3: Lewis 967. Phellobium madagascariensis Baker "Prairies al- timontaines 2000-2300 m" (Paulian et ah, 1971, p. 241), "V6g6tation liee a la presence d'eau 2000-2300 m" (Paulian et ah, 1971, p. 243). Sanicula europaea L. "Foret dense de moyenne altitude 1500-1650 m" (Paulian et ah, 1971, p. 250). Apocynaceae sp. 1 — Camp 1: Lewis (s 30). sp. 2 — Camp 1 : Lewis (s 27). Landolphia myrtifolia (Poir.) Markgraf — Camp 1 : Lewis 795. Mascarenhasia arborescens DC. — Camp 1: Lew- is 760; Camp 2: Lewis 873. Tabernaemontana ciliata Pichon — Camp 2: Lew- is 919. Aquifoliaceae Ilex mitis (L.) Radlk. — Camp 2: Lewis (s 149), 893; Camp 3: Lewis (s 344, 352); Camp 4: Lewis (s 463). Araceae Pothos scandens L. — Camp 1: Lewis 777. Araliaceae Polyscias sp. 1 — Camp 4: Lewis (s 447, 464), 1047. Polyscias sp. 2 — Camp 4: Lewis (s 387, 419, 434), 843. Polyscias sp. 3 — Camp 1: Lewis (s 34), 835. Polyscias sp. 4 — Camp 3: Lewis (s 353), 965. Polyscias sp. 5 — Camp 4: Lewis (s 399, 427). Polyscias sp. 6 — Camp 2: Lewis 926; Camp 3: Lewis (s 336), 979. Polyscias sp. 7 — Camp 2: Lewis (s 180); Camp 3: Lewis (s 295, 291). Polyscias sp. 8 — Camp 3: Lewis (s 229). Polyscias sp. 9 — Camp 4: Lewis (s 423). Schefflera longipedicellata (Lecomte) Bernardi — Approximately 5 km SE Antanifotsy, the re- serve headquarters, 1500 m: Schatz 2674, with P. Goldblatt, A. Rakotozafy, and J. Randriana- solo. Schefflera monophylla (Baker) Bernardi — Ap- proximately 5 km SE Antanifotsy, the reserve headquarters, 1500 m: Schatz 2677, with P. Goldblatt, A. Rakotozafy, and J. Randrianasolo. Schefflera myriantha (Baker) Drake — Camp 2: Lewis (s 139); Camp 3: Lewis (s 317); Camp 4: Lewis (s 497), 1028. Schefflera sp. 1 — Camp 3: Lewis 981. Schefflera sp. 2— Camp 4: Lewis 1099. Arecaceae Dypsis linearis (Becc.) Beentje & J. Dransf. — Camp 1: Lewis 774; Camp 2: Lewis 845. Dypsis sp. 1 — Camp 2: Lewis 864. Ravenea glauca Jum. & H. Perrier — Bois des pentes occidentals, de 1600 a 2000 m (Perrier de la Bathie, 1927). Asclepiadaceae Asclepias fruticosa L. — S Ambalavao above vil- lage of Antanifotsy, partially disturbed, proba- LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 57 bly burned areas dominated by grasses, 1400 m: Lowry 4513. Ceropegia perrieri Choux — Foret a mousses, vers 1600 metres, versant Est (Perrier de la Bathie, 1927). Cynanchum andringitrense Choux — "Foret d'Ambohiby, alt. 1800 m", Leandri 761. Cynanchum pycnoneuroides Choux — Rocailles, 1 800 m, versant Sud-Oeust (Perrier de la Bath- ie, 1927). Cynanchum lineare (Bello) Alain — Andringitra, 1800 m: Liede 2863, with J. Conrad. Cynanchum obovatum Choux — Camp 2: Lewis 896. Cynanchum papillatum Choux — Andringitra, 2000 m: Liede 2862, with J. Conrad. Tylophora sylvatica Decne. — Camp 2: Lewis 909. Asteraceae sp. 1 — Camp 4: Lewis (s 384). sp. 2 — Open ericaceous vegetation, on E side Ampasipotsy, above Camp 4, 2000 m: Lewis 1074. Apodocephala pauciflora Baker — Camp 2: Lewis 886. Brachylaena ramiflora (DC.) Humbert — 50 km S Ambalavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Phi- lippia dominant, 1975 m: Nicoll 239; "Vege- tation rupicole 2000-2300 m" (Paulian et al., 1971, p. 242). Conyza andringitrana Humbert — Bords de tor- rents, vers 1200 m, versant Est (Perrier de la Bathie, 1927). Helichrysum abeitifolium Humbert ssp. gracilifol- ium (Humbert) Humbert — Rocailles de torrents, vers 1600 m (Perrier de la Bathie, 1927). Helichrysum adhaerens (DC.) R. Vig. & Humbert var. subulatum Humbert (Centre: massif de 1' Andringitra). Helichrysum aff. cryptomerioides Baker — "La cuvette de Boby 2500 m" (Paulian et al., 1971, p. 247). Helichrysum attenuatum Humbert — Rocailles, de 1500 a 2500 m (Perrier de la Bathie, 1927). Helichrysum cf. baronii Humbert — Growing on rocks in open ericaceous vegetation, on E side Ampasipotsy, above Camp 4, 2000 m: Lewis 1072. Helichrysum barorum Humbert (massif d' Andrin- gitra, au pied de l'escarpement a Tangle NE du massif a Test de la riviere Antsifo). Helichrysum bracteiferum (DC.) Humbert var. an- dringitranum Humbert — "Praires altimontaines 2000-2300 m" (Paulian et al., 1971, p. 241). Helichrysum calocladum Humbert — Tourbieres, vers. 2000 m (Perrier de la Bathie, 1927); "Ve- getation liee a la presence d'eau 2000-2300 m" (Paulian et al., 1971, p. 242). Helichrysum cryptomerioides Baker — Growing on rocks in open ericaceous vegetation, on E side Ampasipotsy, above Camp 4, 2000 m: Lewis 1067. Helichrysum danguyanum Humbert — Rocailles, de 2000 a 2500 m (Perrier de la Bathie, 1927); growing on rocks in open ericaceous vegeta- tion, on east side of Ampasipotsy, above Camp 4, 2000 m: Lewis 1073. Helichrysum hypnoides (Bojer ex DC.) A. Ca- mus— "Praires altimontaines 2000-2300 m" (Paulian et al., 1971, p. 241). Helichrysum minutiflorum Humbert — Rocailles, vers 2500 m (Perrier de la Bathie, 1927). Helichrysum mirabile Humbert — Rocailles, de 1600 a 2400 m (Perrier de la Bathie, 1927). Helichrysum retrorsum DC. — Growing on rocks in open ericaceous vegetation, on E side Am- pasipotsy, above Camp 4, 2000 m: Lewis 1066. Helichrysum stilpnocephalum Humbert — Rocail- les, de 2000 a 2500 m (Perrier de la Bathie, 1927). Helichrysum tomentosum Humbert — Rocailles, de 2200 a 2600 m (Perrier de la Bathie, 1927). Rochonia aspera Humbert — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Senecio andringitrensis Humbert — Rocailles, de 1600 mala cime (Perrier de la Bathie, 1927). Senecio caniculatus Bojer — "Vegetation rupico- le" (Paulian et al., 1971, p. 257). Senecio denisii Humbert — Rocailles, de 2000 m a la cime (Perrier de la Bathie, 1927). Senecio emirnensis DC. var. angavonensis (Bojer ex DC.) Humbert — 50 km S Ambalavao, near abandoned meteo station above Ambalamari- na, high altitude moss/lichen rain forest-eri- caceous bush, Philippia dominant, 1975 m: Nicoll 247. 58 FIELDIANA: ZOOLOGY Senecia latibracteatus Humbert — Rocailles, au- dessus de 2400 m (Perrier de la Bathie, 1927). Senecio melastomaefolius Baker — "Vdgdtation rupicole" 1800 m" (Paulian et al., 1971, p. 237). Senecio myricaefolius (Bojer ex DC.) Humbert — Camp 4: Lewis 1076. Senecio vaingaindrani Scott-Elliott — Camp 1: Lewis 818. Stoebe cryptophylla Baker — "Hauts fourre arbus- tifs 2000-2300 m" (Paulian et al., 1971, p. 238). Stoebe pachyclada Humbert — Broussailles 6ricoi- des, de 2000 a 2400 m, versant Est (Perrier de la Bathie, 1927). "Hauts fourre' arbustifs 2000-2300 m" (Paulian et al., 1971, p. 238). Syncephalum arbutifolium (Baker) Humbert — Broussailles 6ricoides, entre 1600 et 2400 m (Perrier de la Bathie, 1927). Syncephalum candidum Humbert — Rocailles, vers 2500 m (Perrier de la Bathie, 1927). Vernonia alleizettei Humbert var. rienanensis Humbert — Camp 2: Goodman 6402. Vernonia rubicundus Klatt. — Ericaceous vegeta- tion, on E side Ampasipotsy, above Camp 4, 1800 m: Lewis 1061. Vernonia sp. 1 — Camp 4: Lewis 1048. Vernonia sp. 2 — Camp 3: Lewis 954. Balsaminaceae Impatiens sp. 1 — Camp 2: Lewis 855. Impatiens sp. 2 — Camp 1: Lewis 753; Camp 3: Lewis 937. Impatiens sp. 3 — Camp 4: Lewis 1112; ericaceous vegetation, on E side Ampasipotsy, above Camp 4, 1800 m: Lewis 1058. Bignoniaceae Ophiocolea floribunda H. Perrier — Camp 1: Lew- is 771; Camp 3: Lewis 990, 1015. Brassicaceae Nasturtium sp. 1 — Rocailles, de 2000 m a la cime (Perrier de la Bathie, 1927). Burseraceae Canarium madagascariense Engl. ssp. obtusifol- ium (Scott-Elliott) Leenhouts — Camp 1: Lewis (s 86); Camp 2: Lewis (s 134). Cactaceae Rhipsalis baccifera (Sol. ex J.M. Mill.) Stearn — "Foret a strate herbacee 1695-1840 m" (Pau- lian et al., 1971, p. 254); Camp 2: Lewis 914. Campanulaceae Lobelia agrestis E. Wimm. — Camp 1 : Lewis 788. Cannellaceae Cinnamosma fragrans Baill. sp. 1 — Camp 4: Lewis (s 450, 496), 1077. Clusiaceae Calophyllum drouhardii H. Perrier — Camp 2: Lewis (s 146, 179); Camp 3: Lewis (s 284, 289, 296, 306, 307, 309, 316, 345). Calophyllum paniculatum Stevens — Camp 1: Lewis (s 35, 39), 775. Calophyllum sp. 1 — Camp 3: Lewis (s 279). Garcinia sp. 1 — Camp 2: Lewis 195; Camp 3: Lewis (s 230, 232, 238, 246, 254, 257, 260, 271, 274, 278, 280). Mammea sp. 1 — Camp 4: Lewis 1088. Mammea sp. 2 — Camp 4: Lewis 1063. Mammea sp. 3 — Camp 1: Lewis (s 20); Camp 2: Lewis (s 219). Mammea sp. 4 — Camp 1: Lewis 792; Camp 2: Lewis (s 129), 882. Mammea sp. 5 — Camp 1: Lewis (s 46), 766; Camp 2: Lewis 875. Mammea sp. 6 (Ochrocarpus cerasifer H. Per- rier)— Camp 1: Lewis (s 11), 782. Psorospermum sp. 1 — Camp 3: Lewis 983. Psorospermum sp. 2 — Camp 1: Lewis 817. Psorospermum sp. 3 — Camp 1: Lewis 815. Symphonia microphylla (Cambess.) Benth. & LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 59 Hook. f. ex Vesque ssp. microphylla — 50 km S Ambalavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 249. Symphonia sp. 1 — Camp 3: Lewis (s 262, 273, 321). Commelinaceae Coleotrype goudotii C.B. Clarke — Camp 1: Lewis 839. Convolvulaceae balamarina, high altitude moss/lichen rain for- est-ericaceous bush, Philippia dominant, 1975 m: Nicoll 240. Weinmannia mammea Bernardi — Camp 1: Lewis 781; Camp 3: Lewis 960. Weinmannia sp. 1 — Camp 2: Lewis (s 176, 202). Weinmannia sp. 2 — Camp 3: Lewis (s 298); Camp 4: Lewis (s 413). Weinmannia sp. 3 — Camp 4: Lewis (s 465). Weinmannia sp. 4 — Camp 2: Lewis (s 224). Weinmannia sp. 5 — Camp 1: Lewis (s 53). Weinmannia sp. 6 — Camp 2: Lewis (s 181, 204). Weinmannia sp. 7 — Camp 4: Lewis (s 386, 428, 451,453,473,491), 1022. Hubertia andringitrensis (Humbert) C. Jeffrey — Perrier de la Bathie 2784, 13687. Crassulaceae Crassula sp. 1 — Growing on rocks, Camp 3: Lewis 973. Kalanchoe bergeri Raym.-Hamet & H. Perrier — de 2000 m a la cime (Perrier de la Bathie, 1927); "Fissures et corniches 2300-2650 m" (Paulian et al., 1971, p. 247). Kalanchoe gracilipes (Baker) Baill. — Camp 2: Goodman 6380. Kalanchoe jongmansi Raym.-Hamet & H. Per- rier— de 2000 mala cime (Perrier de la Bathie, 1927). Kalanchoe mangini Raym.-Hamet & H. Perrier — de 2000 m a la cime (Perrier de la Bathie, 1927). Kalanchoe miniata Hils. & Bojer var. andringi- trensis— H. Perrier. Bois vers 1700 m, versant Ouest (Perrier de la Bathie, 1927). Kalanchoe sp. 1 — Camp 1: Lewis 800. Sedum madagascariense H. Perrier — de 2000 m a la cime (Perrier de la Bathie, 1927); "Flore xerophile d'altitude 2300-2650 m" (Paulian et al., 1971, p. 247). Cyperaceae Carex andringitrensis Cherm. — Rocailles, vers 2000 m jusqu' a la cime (Perrier de la Bathie, 1927). Coleochloa setifolia (Ridley) Gilly — "Vegetation rupicole 1800 m" (Paulian et al., 1971, p. 237). Costularia cf. purpurea Cherm. — Camp 4: Lewis 1090. Cyperus nemoralis Cherm. — Rocailles, vers 2000 m (Perrier de la Bathie, 1927). Cyperus calochrous Cherm. — Rocailles, vers 2000 m (Perrier de la Bathie, 1927). Cyperus micrantherus Cherm. — Rocailles, vers 2000 m versant Ouest (Perrier de la Bathie, 1927). Mariscus andringitrensis Cherm. — Rocailles, vers 2000 m (Perrier de la Bathie, 1927). Mariscus viguieri Cherm. var. condensatus Cherm. — Bords des ruisseaux, vers 2000 m versant Est (Perrier de la Bathie, 1927). Pycreus reductus Cherm. — Marais, au-dessus de 1600 m, versant Est (Perrier de la Bathie, 1927). Scleria andringitrensis Cherm. — Rocailles, vers 2000 m versant Est (Perrier de la Bathie, 1927). Cunoniaceae Weinmannia eriocarpa Tul. — 50 km S Ambala- vao, near abandoned meteo station above Ara- Dichapetalaceae Dichapetalum sp. 1 — Camp 2: Lewis (s 123), 862. Dichapetalum sp. 2 — Camp 1: Lewis 762. 60 FIELDIANA: ZOOLOGY Dracaenaceae Dracaena xiphophylla Baker — Camp 2: Lewis (s 188), 913. Ebenaceae Diospyros sp. 1 — Camp 2: Lewis 908; Camp 3: Lewis 1019. Diospyros sp. 2 — Camp 1: Lewis 831; Camp 2: Lewis 871. Diospyros sp. 3 — Camp 1 : Lewis 799. Diospyros sp. 4 — Camp 3: Lewis 301, 312, 313. Elaeocarpaceae Elaeocarpus capuronii Tirel — Camp 1: Lewis 830. Elaeocarpus hildebrandtii Baill. — Camp 4: Lewis 1045; Andringitra, canton de Sendirisoa: Rak- oto RN6499. Elaeocarpus subserratus Baker — Camp 3: Lewis (s 325, 364); Camp 4: Lewis (s 431), 1052. Sloanea rhodantha (Baker) Capuron var. rhodan- tha form "quadriloba" — Camp 1 : Lewis (s 44, 68, 114); Camp 2: Lewis (s 141); Andringitra: Leandri et al., 3383. Sloanea rhodantha (Baker) Capuron var. rhodan- tha form "quercifolia" — Camp 3: Lewis (s 286, 347, 360), 1006; Camp 4: (s 448), 1104, 1 105; Massif de 1' Andringitra, foret d'Imaiso 1500-2000 m: Leandri et al. 3383 (form un- certain). Ericaceae Agauria salicifolia Hook. f. — "V6g6tation rupi- cole 2000-2300 m" (Paulian et al., 1971, p. 242); "Foret dense scleYophylle de montagne 2000 m" (Paulian et al., 1971, p. 250). Agauria sp. 1 — Camp 1: Lewis 761. Philippia andringitrensis H. Perrier — Broussailles encoides de la cime, depuis 2400 m (Perrier de la Bathie, 1927). Philippia floribunda Benth. ssp. pilulifera (Hum- bert) H. Perrier — Broussailles encoides de la cime, depuis 2400 m (Perrier de la Bathie, 1927). Philippia pilosa Baker — "V6g6tation liee a la pr6sence d'eau 2000-2300 m" (Paulian et al., 1971, p. 243). Philippia spinifera H. Perrier — Broussailles eYi- coi'des de la cime, depuis 2400 m (Perrier de la Bathie, 1927). Philippia viguieri H. Perrier — Broussailles 6ricoi- des de la cime, depuis 2400 m (Perrier de la Bathie, 1927). Philippia sp. 1 — Ericaceous vegetation above the tree line, on east side of Ampasipotsy, above Camp 4, 1800-2000 m: Lewis 1051, 1065. Vaccinium emirnense Hook. — Camp 2: Lewis 905. Vaccinium secundiflorum Hook. — Ericaceous vegetation above the tree line, on E side Am- pasipotsy, above Camp 4, 1800-2000 m: Lewis 1062, Goodman 6450; 50 km S Ambalavao, near abandoned meteo station above Ambala- marina, high altitude moss/lichen rain forest- ericaceous bush, Philippia dominant, 1975 m: Nicoll289, 290, and 291. Vaccinium sp. 1 — Camp 4: Lewis (s 383), 1079. Eriocaulaceae Eriocaulon fenestratum Bojer — "Veg&ation liee a la presence d'eau 2000-2300 m" (Paulian et al., 1971, p. 242). Erythroxylaceae Erythroxylum capitatum Baker — Camp 2: Lewis 894; Camp 3: Lewis (s 241). Erythroxylum sp. 1 — Camp 4: Lewis 1102. Erythroxylum sp. 2 — Camp 3: Lewis (s 265); Camp 4: Lewis (s 478). Euphorbiaceae sp. 1 — Camp 4: Lewis (s 439). Acalypha andringitrensis Leandri (Madagascar: Centre, massif de 1'Andringitra, vallees de la Riambava et de 1'Antsifotra). Amyrea humbertii Leandri — Camp 1 : Lewis 826. Antidesma petiolare Tul. — Camp 2: Lewis (s 124, 216), 824. Blotia tanalorum Leandri — Camp 1: Lewis 824. LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 61 Claoxylon sp. 1 — Camp 2: Lewis 861. Claoxylon sp. 2 — Camp 2: Lewis 863. Cleistanthus boivinianus (Baill.) Mull. Arg. — Camp 3: Lewis (s 240, 251, 267, 268, 269, 270, 277, 283, 318, 328, 334, 339, 348, 350), 994. Croton myriaster Baker var. austromadecassa Leandri (Centre: Andringitra, Ikongo, Ando- hahelo, Beampingaratra). Croton nitidulus Baker var. tandrokensis Leandri (Massif d' Andringitra, col du Tandroka, versant Est; 1200 m). Croton sp. 1 — Camp 3: Lewis 987; Camp 4: Lew- is (s 418), 1023. Croton sp. 2— Camp 3: Lewis (s 234, 239, 250). Croton sp. 3 — Camp 2: Lewis 878, 931. Drypetes madagascariensis (Lam.) Humbert & Leandri — Camp 2: Lewis 917. Euphorbia emirnensis Baker — Open ericaceous vegetation: Lewis 1068. Macaranga ankafinensis Baill. — Camp 1: Lewis 752; Camp 2: Lewis 849; Camp 3: Lewis 939, 993. Macaranga cf. decaryana Leandri — Camp 2: Lewis 859; Camp 3: Lewis 963; Camp 4: Lewis 1110. Macaranga cuspidata Baill. — Camp 3: Lewis 1009. Macaranga echinocarpa Baker — Camp 3: Lewis 1017; Camp 4: Lewis (s 369, 391, 392). Macaranga oblongifoia Baill. — Camp 4: Good- man 6458. Macaranga sp. 1 — Camp 2: Lewis 925. Macaranga sp. 2 — Camp 1: Lewis 809; Camp 2: Lewis 918. Mallotus cf. capuronii Leandri — Camp 2: Lewis 927; Camp 3: Lewis 287, 276, 323, 338, 952. Phyllanthus fusco-luridus Mull. Arg. ssp. villosus Leandri (Andringitra). Phyllanthus iratsiensis Leandri (Massif de 1' An- dringitra, Iratsy, vallees de la Riambava). Phyllanthus monticola Leandri (Andringitra, foret a mousses). Phyllanthus sp. 1 — Camp 2: Lewis 891. Phyllanthus sp. 2 — Ericaceous vegetation, on E side Ampasipotsy, above Camp 4, 1800 m: Lewis 1056. Phyllanthus sp. 3 — Camp 1: Lewis 822. Savia andringitrana Leandri (base nord du Pic d'lvohibe). Suregada baronii (Moore) Croizat — Camp 1: Lewis 833. Uapaca sp. 1 — Camp 2: Lewis (s 192). Fabaceae Abrus precatorius L. — Camp 1 : Lewis 779. Abrus sp. 1 — Camp 1: Lewis 802. Albizia gummifera (J.E Gmel) C.A. Sm. — Camp 1: Lewis (s 90), 767. Albizia sp. 1 — Camp 1: Lewis 798. Crotalaria andringitrensis R. Vig. — (Andringitra range). Dalbergia sp. 1 — Camp 2: Lewis (s 182, 217). Desmodium repandum DC. — "Foret dense de moyenne altitude 1500-1650 m" (Paulian et al., 1971, p. 250). Dolichos minutiflorus R. Vig. (Massif de 1' An- dringitra). Indigofera andringitrensis R. Vig. (Massif de F Andringitra; 1400-2000 m). Mimosa andringitrensis R. Vig. (Massif de 1' An- dringitra; 1400-2000 m)— Bois, vers 1600 m (Perrier de la Bathie, 1927). Mimosa descarpentriesii R. Vig. (Massif de I' Andringitra; 1800-2300 m). Mundulea andringitrensis R. Vi< . (Massif de l'Andringitra) — "Hauts fourre arbustifs 2000- 2300 m" (Paulian et al., 1971, p. 238); "Ve- getation rupicole 2000-2300 m" (Paulian et al., 1971, p. 242). Mundulea splendens R. Vig. (environs d'Ambos- itra a l'Andringitra; 1400-2000 m). Smithia chamaecrista Benth. var. stipulata R. Vig. (versant Ouest de l'Andringitra, 800-1600 m). Strongylodon sp. 1 — Camp 2: Lewis 858. Flacourtiaceae Aphloia theiformis Benn — Camp 2: Lewis 785, 879; Camp 3: Lewis (s 275); Camp 4: Lewis (s 410, 414), 1027. Casearia elliptica Tul. — Camp 3: Lewis 1018. Casearia nigrescens Tul. — Camp 2: Lewis 930. Casearia nigrescens Tul. var. lucida (Tul.) Sleu- mer — Approximately 5 km SE Antanifotsy, the reserve headquarters, 1500 m: Schatz 2678, 62 FIELDIANA: ZOOLOGY with P. Goldblatt, A. Rakotozafy, and J. Ran- drianasolo. Casearia sp. 1 — Camp 4: Lewis (s 371, 377, 398), 1024. Scolopia sp. 1 — Camp 3: Lewis 1013. Scolopia sp. 2 — Camp 4: Lewis 1080. Gentianaceae Exacum dipterum Klack. — Versant est du massif de I'Andringitra, 1200 m: Perrier de la Bathie 9069; Andringitra, Col du Tandroka, versant E du massif d' Andringitra, 1600 m: Perrier de la Bathie 9067. Exacum emirnense (Baker) Schinz — Andringitra, Sendrisoa: Rakoto RN6496. Exacum naviculare Klack. — Andringitra, 2000 m: Perrier de la Bathie 9080bis. Ornichia madagascariensis (Baker) Klack. ssp. pubescens (Baker) Klack. — Andringitra: Raza- findrakoto, RN3501; Camp 1: Lewis 811. Swertia rosulata (Baker) Klack. Andringitra, can- ton de Sendrisoa: Randriamiera RN5579. Tachiadenus platypterus Baker ssp. platypterus — N. du massif d' Andringitra 1600 m: Rakoto RN6475; Andringitra, canton de Vohitsaoka; Perrier de la Bathie 14468. Streptocarpus hilsenbergii R. Br. — Growing on rocks, Camp 3: Lewis 1008. Streptocarpus macropodus B.L. Burtt — Camp 4: Lewis 1038, 1043. Streptocarpus suffruticosus Humbert "Foret a strate herbac£e [below] 1970 m" (Paulian et al., 1971, p. 254); Growing on rocks, Camp 3: Lewis 955, 962. Hyacinthaceae Rhodocodon urgineoides Baker — Camp 2: Lewis 876. Icacinaceae sp. 1 — Camp 3: Lewis 982, 992; Camp 4: Lewis 1035. sp. 2 — Camp 1: Lewis 836. sp. 3 — Camp 3: Lewis (s 248). Apodytes sp. 1 — Camp 2: Lewis (s 155). Apodytes sp. 2 — Camp 4: Lewis 1082. Cassinopsis tomentosa H. Perrier — Camp 3: Lew- is (s 297, 288), 945. Iridaceae Geraniaceae Geranium andringitrense H. Perrier — Rocailles, de 2000 m a la cime (Perrier de la Bathie, 1927); "Foret dense sclerophylle de montagne 2000 m" (Paulian et al., 1971, p. 250); erica- ceous vegetation above the tree line, on E side Ampasipotsy, above Camp 4, 1800-2000 m: Lewis 1069. Pelargonium madagascariense Baker — Rocailles, de 2000 a 2400 m (Perrier de la Bathie, 1927). Gesneriaceae Didymocarpus madagascariensis C.B. Clarke — Growing on rocks, Camp 3: Lewis 972. Didymocarpus vestitus Baker var. lanceolatus Humbert ex B.L. Burtt (versant sud du massif de I'Andringitra, riviere Ihovika, bassin de la Matitanana). Aristea angustifolia Baker — Slopes of massif, be- low steep cliffs, 6-8 km S Antanifotsy, along path to Pic Boby, 1900-2050 m: Goldblatt and Schatz 8971; Massif d' Andringitra, 1800-2200 m: Perrier de la Bathie 14410. Aristea cladocarpa Baker — Camp 1: Lewis 804; 50 km S Ambalavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Phi- lippia dominant, 1975 m; grassy/rocky summit with orchids, 1975 m: Nicoll 241. Aristea humbertii H. Perrier — Massif de I'Andrin- gitra, Iratsy: vallees de la Riambava et de 1'Antsifotra et montagnes environnantes 2000- 2500 m; Humbert 3790. Aristea madagascariense Baker — Andringitra, Canton de Sendrisoa: Henri RN4827; Andrin- gitra: Razafindrakoto RN2408. Gladiolus andringitrae Goldblatt — Andringitra, Boby: Homolle 1189; Massif de I'Andringitra 2400 m: Perrier de la Bathie 13655; Massif de LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 63 1' Andringitra 2000-2400 m; Perrier de la Bath- ie 14483; Andringitra Forest Reserve, slopes of massif, below steep cliffs, 6-8 km S Antani- fotsy, along path to Pic Boby, 1900-2050 m: Goldblatt and Schatz 8974. Gladiolus bojeri (Baker) Goldblatt — Plateau d'Andohariana. Andringitra 2000-2100 m: Guillaumet 3754; Massif de l'Andringitra, Ir- atsy, vallees de la Riambava et de l'Antsifotra et montagnes environnantes 1600-2000 m: Humbert 3655; Andringitra Forest Reserve, slopes of massif, below steep cliffs, 6-8 km S Antanifotsy, along path to Pic Boby, 1900- 2050 m: Goldblatt and Schatz 8970. Gladiolus dalenii Geel — Andringitra, Canton de Sendrisoa, Henri RN5554; Andringitra: Raza- findrakoto RN2412. Lamiaceae sp. 14 — Camp 2: Lewis (s 226). sp. 15— Camp 3: Lewis (s 243, 329, 332); Camp 4: Lewis (s 461). sp. 16 — Camp 1: Lewis (s 3). Cryptocarya crassifolia Baker — Camp 3: Lewis 340; Camp 4: Lewis (s 374, 503), 1081. Cryptocarya madagascariensis (Baill.) Kos- term. — Camp 1: Lewis (s 99). Cryptocarya oppositifolia Kosterm. — Camp 4: Lewis 1085. Cryptocarya sp. 1 — Camp 4: Lewis 1031. Cryptocarya sp. 2 — Camp 1: Lewis (s 19, 57); Camp 2: Lewis (s 213). Cryptocarya sp. 3 — Camp 1: Lewis (s 8, 112); Camp 2: Lewis 911. Ocotea sp. 1 — Camp 3: Lewis 1005. Ocotea sp. 2 — Camp 2: Lewis 889; Camp 4: Lew- is (s 406, 411, 412, 435, 466), 1026. Micromeria sp. 1 — Ericaceous vegetation, on east side of Ampasipotsy, above Camp 4, 2000 m: Lewis 1064. Orthosiphon sp. 1 — Camp 4: Lewis 1108. Orthosiphon sp. 2 — Camp 3: Lewis 933. Perrierastrum oreophilum Guillaumin (Massif d' Andringitra). Plectranthus sp. 1 — Camp 3: Goodman 6408. Solenostemon bojeri (Benth.) Guillaumet & Cor- net— Camp 4: Goodman 6455. Lauraceae Lentibulariaceae Utricularia humbertiana H. Perrier var. andrin- gitrensis H. Perrier (Massif de l'Andringitra, 2500 m). Utricularia livida E. Mey. — 50 km S Ambalavao, near abandoned meteo station above Ambala- marina, high altitude moss/lichen rain forest- ericaceous bush, Philippia dominant, 1975 m: Nicoll 243. Utricularia mauroyae H. Perrier (Massif de l'Andringitra, sur les cimes, vers 2500 m). sp. 1— Camp 3: Lewis (s 261, 302, 314, 342). sp. 2 — Camp 2: Lewis (s 165); Camp 3: Lewis (s 300, 367); Camp 4: Lewis (s 420). sp. 3 — Camp 1: Lewis (s 106). sp. A — Camp 2: Lewis (s 173). sp. 5 — Camp 2: Lewis (s 212). sp. 6 — Camp 2: Lewis (s 162). sp. 7 — Camp 4: Lewis (s 401). sp. 8 — Camp 2: Lewis (s 137). sp. 9— Camp 2: Lewis (s 150, 159). sp. 10 — Camp 1: Lewis (s 22); Camp 2: Lewis (s 144). sp. 11 — Camp 1: Lewis (s 65). sp. 12 — Camp 1: Lewis (s 66). sp. 13 — Camp 1: Lewis (s 72). Loganiaceae Anthocleista madagascariensis Baker — Camp 2: Lewis 897; Camp 3: Lewis 337; Camp 4: Lewis 1089. Buddleja indica Lam. — Camp 1: Lewis 821. Nuxia involucrata A.DC. — Camp 1: Lewis 773. Strychnos sp. 1 — Camp 3: Lewis 975. Loranthaceae Bakerella clavata (Desrouss.) Balle (var. not de- termined)— Camp 3: Lewis 950; Camp 4: Lew- is 1025. Bakerella clavata (Desrouss.) Balle var. lenticel- 64 FIELDIANA: ZOOLOGY lata (Baker) Balle — Camp 4: Lewis 1029, 1041, Goodman 6453; ericaceous vegetation above the tree line, on E side Ampasipotsy, above Camp 4, 1800-2000 m: Lewis 1057. Bakerella clavata (Desrouss.) Balle var. perala- ta — Camp 1: Lewis 801. Bakerella tandrokensis (Lecomte) Balle (Andrin- gitra) — Ericaceous vegetation above the tree line, on east side of Ampasipotsy, above Camp 4, 2050 m: Goodman 6459. Lythraceae Rotala sphagnoides H. Perrier (Massif d'Andrin- gitra). Malpighiaceae Tristellateia madagascariensis Poir. — Camp 2: Lewis 902. Malvaceae Kosteletzkya macrantha Hochr. — Savoka a Philip- pia, ravins de 1500 a 2000 m, versant Ouest (Perrier de la Bathie, 1927). Kosteletzkya malvocaerulea Hochr. — Savoka a Philippia, ravins de 1500 a 2000 m, versant Ouest (Perrier de la Bathie, 1927). Gravesia calliantha Jum. & H. Perrier — vers 1700 m Rocailles, versant SE. (Perrier de la Bathie, 1927); Camp 2: Goodman 6409. Gravesia cf. vestita (Baker) H. Perrier — Camp 2: Lewis 884. Gravesia inappendiculata H. Perrier — Camp 4: Goodman 6456. Gravesia sp. 1 — Camp 3: Lewis 996. Gravesia sp. 2 — Camp 3: Lewis 935. Gravesia sp. 3 — Camp 1: Lewis 837. Medinilla ericarum Jum. & H. Perrier — Camp 2: Goodman 6407. Medinilla sp. 1 — Camp 2: Lewis 890; Camp 3: Lewis 953; Camp 4: Lewis 1059, 1093. Medinilla sp. 2 — Camp 3: Lewis 977. Medinilla sp. 3 — Camp 2: Lewis 856. Medinilla sp. 4 — Camp 1: Lewis 813. Medinilla torrentium Jum. & H. Perrier — Bois, vers 1 700 m, versant SE. (Perrier de la Bathie, 1927); Camp 4: Goodman 6429. Memecylon cf. ivohibense Jacq.-FeL — Camp 3: Lewis (s 346) 892. Memecylon cf. longipetalum H. Perrier — Camp 1 : Lewis (s 21); Camp 2: Lewis 868. Memecylon cf. perangustum Jacq.-FeL — Camp 1 : Lewis 806. Osbeckia andringitrensis H. Perrier — Rocailles, vers la cime (Perrier de la Bathie, 1927). Osbeckia bicolor H. Perrier — Bois, vers 2000 m, versant Est (Perrier de la Bathie, 1927). Osbeckia mandrarensis H. Perrier — Camp 4: Lewis 1097. Melanophyllaceae Melanophylla cf. alnifolia Baker — Camp 2: Lewis 847, 860. Melanophylla cf. humbertiana Keraudren — Camp 2: Lewis 895. Melanophylla sp. 1 — Camp 4: Lewis (s 422). Melastomataceae Meliaceae Turraea sp. 1 — Camp 1: Lewis 838. Menispermaceae Burasaia madagascariensis Thouars — Camp 1: Lewis 829. Cyclea sp. 1 — Camp 1: Lewis 765. Dichaetanthera matitanensis Jum. & H. Perrier — Foret a mousses, vers 1 800 m, versant Est (Per- rier de la Bathie, 1927). Dichaetanthera sp. 1 — Camp 1: Lewis 763. Dichaetanthera sp. 2 — Camp 2: Lewis 854. Menyanthaceae Limnanthemum aff. indicum Griseb. — "Vegeta- tion liee a la presence d'eau 2000-2300 m" (Paulian et al., 1971, p. 242). LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 65 Monimiaceae Decaryodendron perrieri Cavaco — Camp 2: Lew- is 901. Ephippiandra microphylla (Perk.) Cavaco — Camp 4: Lewis 1100. Ephippiandra sp. 1 — Camp 1: Lewis (s 40). Tambourissa cf. parvifolia Baker — Camp 3: Lew- is (s 315), 942, 951, 964. Tambourissa sp. 1 — Camp 4: Lewis (s 452). Tambourissa sp. 2 — Camp 4: Lewis (s 402). Tambourissa sp. 3 — Camp 1: Lewis 816; Camp 2: Lewis 874, 898. Tambourissa thouvenotii Danguy — Camp 1 : Lew- is 789, 834; Camp 2: Lewis (s 214). Moraceae Ficus brachyclada Baker — Camp 1: Lewis 825; Camp 2: Lewis 841, 888. Ficus lutea Vahl — Camp 1: Lewis 820. Ficus politoria Lam. — Camp 3: Lewis 966; Camp 4: Lewis 1103. Ficus reflexa Thunb. — Camp 1: Goodman 6353. Ficus rubra Vahl — Camp 1: Lewis 724; Camp 2: Lewis 866. Ficus tiliifolia Baker — Camp 1 : Lewis 790; Camp 2: Lewis 904. Ficus trichopoda Baker — Camp 1: Lewis 769. Ficus sp. 1 — Camp 3: Lewis 988. Streblus dimepate (Bureau) C.C. Berg — Camp 2: Lewis (s 133, 145, 200). Myrothamnaceae Myrothamnus moschatus (Baill.) Baill. — Andrin- gitra, canton de Vohitsaoka: Rakotovao RN6555. marina, mid-altitude rain forest with high stem density and canopy at 25-30 m, 1700 m: Nicoll 305. Monoporus simplex H. Perrier — 50 km S Amba- lavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 233. Monoporus sp. 1 — Camp 2: Lewis 842. Monoporus sp. 2 — Camp 2: Lewis 870. Oncostemon ankifiense Mez. — Camp 4: Goodman 6430. Oncostemon botryoides Baker — Camp 1: Lewis 783. Oncostemon microsphaerum Baker — Camp 3: Lewis 999. Oncostemon racemiferum Mez. — Camp 1: Good- man 6352. Oncostemon sp. 1 — Camp 2: Lewis 923. Oncostemon sp. 2 — Camp 3: Lewis 1003. Oncostemon sp. 3 — Camp 2: Lewis 929. Oncostemon sp. A — Camp 3: Lewis 1004. Oncostemon sp. 5 — Camp 1: Lewis 812. Oncostemon sp. 6 — Camp 4: Lewis 1096. Oncostemon sp. 7 — Camp 4: Lewis 1044, 1101. Oncostemon sp. 8 — Camp 4: Lewis 1060. Oncostemon sp. 9 — Camp 3: Lewis 1010. Oncostemon sp. 10 — Camp 3: Lewis 957; Camp 4: Lewis 1106. Oncostemon sp. 11 — Camp 2: Lewis 921. Oncostemon sp. 12 — Camp 1: Lewis 764. Oncostemon sp. 13 — Camp 2: Lewis (s 208); Camp 3: Lewis 986. Oncostemon sp. 14 — Camp 3: Lewis 361; Camp 4: Lewis 1086. Oncostemon sp. 15 — Camp 4: Lewis 1036. Oncostemon sp. 16 — Camp 4: Lewis 1040. Myrtaceae Myrsinaceae Embelia concinna Baker — Camp 3: Lewis 989. Embelia madagascariensis A. DC. — Camp 1: Lewis 757. Maesa lanceolata Forssk. — 50 km S Ambalavao, near abandoned meteo station near Ambala- Syzygium sp. 1 — Camp 2: Lewis 928. Syzygium sp. 2 — Camp 2: Lewis (s 170); Camp 3: Lewis (s 228, 249, 281, 335). Syzygium sp. 3 — Camp 2: Lewis (s 168, 221). Syzygium sp. 4 — Camp 4: Lewis (s 456, 468, 500). Syzygium sp. 5 — Camp 1: Lewis (s 31, 63). Syzygium sp. 6 — Camp 1: Lewis (s 113). 66 FIELDIANA: ZOOLOGY Syzygium sp. 7 — Camp 4: Lewis (s 385). Syzygium sp. 8 — Camp 1: Lewis (s 70). Syzygium sp. 9 — Camp 4: Lewis (s 426). Syzygium sp. 10 — Camp 4: Lewis (s 372). Syzygium sp. 11 — Camp 1: Lewis (s 87); Camp 2: Lewis 899. Syzygium sp. 12 — Camp 3: Lewis (s 303). Syzygium sp. 13 — Camp 1: Lewis 776. Syzygium sp. 14 — Camp 1: Lewis (s 52, 79), 828; Camp 2: Lewis (s 227), 867, 922. Syzygium sp. 15 — Camp 3: Lewis (s 258). Syzygium sp. 16 — Camp 3: Lewis (s 293). Syzygium sp. 17 — Camp 1: Lewis (s 92). Ochnaceae Campylospermum obtusifolium Tiegh. — Camp 1: Lewis 794, 807. Oleaceae Noronhia sp. 1 — Camp 2: Lewis 900. Noronhia sp. 2 — Camp 3: Lewis 1007; Camp 4: Lewis 1107. Noronhia sp. 3 — Camp 3: Lewis (s 294). Noronhia sp. 4 — Camp 3: Lewis (s 299, 326), 947. Noronhia sp. 5 — Camp 3: Lewis 1011. Noronhia sp. 6 — Camp 1: Lewis (s 5). Noronhia sp. 7 — Camp 1: Lewis (s 37). Orchidaceae Angraecum oblongifolium Toill. — Gen. & Bosser (Andringitra). Angraecum sp. 1 — Camp 3: Lewis 948. Angraecum sp. 2 — Camp 4: Lewis 1091. Benthamia exilis Schltr. var. exilis (Massif d' An- dringitra, 2200 m alt.; 2400 m alt.) — Rocailles, de 2000 m a la cime (Perreir de la Bathie, 1927). Benthamia exilis Schltr. var. tenuissima Schltr. (Massif d' Andringitra, 2200 m alt.). Benthamia macro Schltr. (Massif d'Andringitra, 2200 m alt.). Benthamia monophylla Schltr. (Massif d'Andrin- gitra, Buissons ericoides, vers 2600 m) — Ro- cailles vers 2000 m et au-dessus (Perreir de la Bathie, 1927); 50 km S Ambalavao, near abandoned meteo station above Ambalamari- na, high altitude moss/lichen rain forest-eri- caceous bush, Philippia dominant, 1975 m: Nicoll 262. Benthamia oreodorum Schltr. — Foret a mousses, vers 1 800 m, versant Est (Perrier de la Bathie, 1927). Benthamia praecox Schltr. — 50 km S Ambalavao, near abandoned meteo station above Ambala- marina, high altitude moss/lichen rain forest- ericaceous bush, Philippia dominant, 1975 m: Nicoll 288. Bulbophyllum andringitranum Schltr. — Rocailles, au-dessus de 2000 m, versant Ouest (Perrier de la Bathie, 1927). Bulbophyllum sp. 1 — Camp 2: Lewis 857. Calanthe repens Schltr. var. pauliani Ursch & Toill. — Gen. (Andringitra). Cynorkis alborubra Schltr. (Massif d'Andringitra, 2600 m alt.) — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Cynorkis andingitrana Schltr. — Rocailles, au-des- sus de 2000 m (Perrier de la Bathie, 1927). Cynorkis angustipetala Ridl. var. amabilis (Schltr.) H. Perrier — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927); on trail from Antanifotsy to Pic Boby, high altitude meadow by mountain stream, 2000 m: Nicoll 323. Cynorkis baronii Rolfe (Shady places on humus on summit of Andringitra Mts) — On trail from Antanifotsy to Pic Boby, high altitude meadow by mountain stream, 2000 m: Nicoll 324. Cynorkis bathiei Schltr. (Massif d'Andringitra, 2400 m alt.) — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Cynorkis bella Schltr. — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Cynorkis filiformis H. Perrier — Rocailles, au-des- sus de 2000 m (Perrier de la Bathie, 1927). Cynorkis gibbosa Ridl. — On trail from Antanifot- sy to Pic Boby, high altitude meadow by moun- tain stream, 2000 m: Nicoll 327. Cynorkis gigas Schltr. — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Cynorkis inversa Schltr. — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Cynorkis jumelleana Schltr. var. gracillima LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 67 Schltr. — Rocailles, au-dessus de 2000 m (Per- rier de la Bathie, 1927). Cynorkis jumelleana Schltr. var. jumelleana — Ro- cailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Cynorkis lilacina Ridley var. pulchra (Schltr.) H. Perrier — Rocailles, au-dessus de 2000 m (Per- rier de la Bathie, 1927). Cynorkis ochroglossa Schltr. — Rocailles, au-des- sus de 2000 m (Perrier de la Bathie, 1927). Cynorkis quinqueloba H. Perrier — 50 km S Am- balavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 244. Cynorkis saxicola Schltr. — Rocailles, au-dessus de 2000 m (Perrier de la Bathie, 1927). Cynorkis sp. 1 — Camp 1: Lewis 810. Disa andringitrana Schltr. — Broussailles ericoi- des, 2200 m (Perrier de la Bathie, 1927). Habenaria decaryana H. Perrier — Camp 3: Lewis 949. Liparis andringitrana Schltr. — Rocailles, 2000 m, versant Est (Perrier de la Bathie, 1927). Liparis densa Schltr. — Rocailles, 2000 m, versant Est (Perrier de la Bathie, 1927). Liparis henrici Schltr. — Foret a mousses, vers 1800 m, versant Est (Perrier de la Bathie, 1927). Liparis latilabris Schltr. — Foret a mousses, vers 1800 m, versant Est (Perrier de la Bathie, 1927). Liparis listeroides Schltr. — Foret a mousses, vers 1800 m, versant Est (Perrier de la Bathie, 1927). Liparis sp. 1 — Camp 2: Goodman 6381. Polystachya oreocharis Schltr. — 50 km S Amba- lavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 248. Satyrium trinerve Lindl. — On trail from Antani- fotsy to Pic Boby, high altitude meadow by mountain stream, 2000 m: Nicoll 326. Tylostigma foliosum Schltr. (Massif d'Andringitra, 2500 m alt.) — Savoka a Philippia, vers 2000 m, versant Est (Perrier de la Bathie, 1927). Pandanaceae Pandanus sp. 1 — Camp 1: Lewis 768. Pandanus sp. 2 — Camp 4: Lewis 1032. Pandanus sp. 3 — Camp 1: Lewis 787. Pandanus sp. 4 — Camp 2: Lewis 848. Passifloraceae Deidamia bicolor H. Perrier — Camp 1: Lewis 840. Phormiaceae Dianella ensifolia (L.) Redoute — "Foret dense sclerophylle de montagne [above] 1950 m" (Paulian et al., 1971, p. 256). Piperaceae Peperomia sp. 1 — Camp 3: Lewis 946. Peperomia sp. 2 — Camp 3: Lewis 978. Piper sp. 1 — Camp 2: Lewis 872. Pittosporaceae Pittosporum sp. 1 — Camp 3: Lewis 956. Pittosporum sp. 2 — Camp 1: Lewis 770. Pittosporum sp. 3 — Camp 4: Lewis 1039. Pittosporum sp. 4 — Camp 2: Lewis 903. Poaceae Andropogon trichozygus Baker — "Prairies alti- montaines 2000-2300 m" (Paulian et al., 1971, p. 242); "Vegetation liee a la presence d'eau 2000-2300 m" (Paulian et al., 1971, p. 242). Arundinella nepalensis Trin. — "Vegetation liee a la presence d'eau 2000-2300 m" (Paulian et al., 1971, p. 242). Brachiaria dimorpha A. Camus. (Massif d'An- dringitra)— Foret a mousses, vers 2000 m (Per- rier de la Bathie, 1927); "Prairies altimontaines 2000-2300 m" (Paulian et al., 1971, p. 242). Dichanthium andringitrensis A. Camus — Rocail- 68 FIELDIANA: ZOOLOGY les, 2000 m, versant Ouest (Perrier de la Bathie, 1927). Digitaria perrieri A. Camus — Rocailles, 2000 m (Perrier de la Bathie, 1927). Festuca perrieri A. Camus sp. 1 — pres des eaux, vers 2000 m versant Est (Perrier de la Bathie, 1927). Lasiorrachis viguieri (A. Camus) Bosser — "Prai- ries altimontaines 2000-2300 m" (Paulian et al., 1971, p. 242). Lecomtella madagascariensis A. Camus (Massif d'Andringitra, 1600-2400 m); versant Ouest, de 1600 m a 2400 m (Perrier de la Bathie, 1927). Nastus decaryanus A. Camus (Massif d'Andrin- gitra). Nastus elongatus A Camus — Bois, vers 2000 m, versant Est (Perrier de la Bathie, 1927). Panicum aequinerve Nees — 50 km S Ambalavao, near abandoned meteo station above Ambala- marina, high altitude moss/lichen rain forest- ericaceous bush, Philippia dominant, 1975 m: Nicoll 295. Panicum ambositrense A. Camus — "Foret dense scterophylle de montagne [above] 1950 m" (Paulian et al., 1971, p. 256). Panicum cupressiforme A. Camus — "V6g6tation liee a la presence d'eau 2000-2300 m" (Pau- lian et al., 1971, p. 243); "Groupements her- baces 2000-2500 m" (Paulian et al., 1971, p. 247); "La cuvette de Boby 2500 m" (Paulian et al., 1971, p. 247). Panicum neoperrieri A. Camus (Massif d'Andrin- gitra, 1600 m). Panicum spergulifolium A. Camus (Massif d'Andringitra, 2200-2500 m) — Rocailles, vers 2000 m a la cime (Perrier de la Bathie, 1927); "Prairies altimontaines 2000-2300 m" (Pauli- an et al., 1971, p. 241); "V6g6tation H6e a la presence d'eau 2000-2300 m" (Paulian et al., 1971, p. 242). Panicum subhystrix A. Camus sp. 1 — Rocailles, vers 2000 m (Perrier de la Bathie, 1927). Pentaschistis andringitrensis A. Camus (Massif d'Andringitra). Poa perrieri A. Camus sp. 1 — pres des eaux, vers 2000 m versant Est (Perrier de la Bathie, 1927). Proteaceae Faurea forficuliflora H. Perrier (Massif d'Andrin- gitra, 2400 m). Ranunculaceae Clematis mauritiana Lam. var. sulfurea R. Vig. & H. Perrier (Massif d'Andringitra, vers 2200 m). Restionaceae Restio madagascariensis Cherm. — "Groupements herbac^s 2100-2500 m" (Paulian et al., 1971, p. 247); "La cuvette de Boby 2500 m" (Paulian et al., 1971, p. 247). Rhamnaceae sp. 1— Camp 3: Lewis (s 349), 959, 1 109. Rhopalocarpaceae Rhopalocarpus sp. 1 — Camp 2: Lewis (s 147) Rosaceae Alchemilla andringitrensis R. Vig. & de Wild. — Rocailles, de 2000 m a la cime (Perrier de la Bathie, 1927). Rubus rosaefolius Sm. — "Foret dense scleYophyl- le de montagne 2000 m" (Paulian et al., 1971, p. 250). Rubiaceae sp. 1— Camp 2: Lewis 205. sp. 2— Camp 3: Lewis 970. sp. 3 — Camp 2: Lewis 906. sp. 4 — Camp 4: Lewis (s 424). sp. 5 — Camp 2: Lewis 924. sp. 6 — Camp 3: Lewis 1000. sp. 7 — Camp 3: Lewis 968. Alberta humblotii Drake — Camp 2: Lewis 916. Alberta minor Baill. — Open ericaceous vegeta- tion, on E side Ampasipotsy, above Camp 4, 1800-2000 m: Lewis 1070. Anthospermum ibityense Puff — 50 km S Amba- lavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 69 forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 292. Anthospermum madagascariense Homolle ex Puff (Andringitra Massif, brousse ericoide vers 2600 m). Breonia sp. — Camp 1: Lewis 755. Canthium andrigitrense Cavaco (Massif de 1' An- dringitra, Iratsy, vallees de la Riambava et de l'Antsifotra). Canthium parvistipula Arenes & Cavaco — Camp 1: Lewis 780; Camp 4: Lewis 1049. Canthium sp. 1 — Camp 3: Lewis 984. Canthium sp. 2 — Camp 4: Lewis 1087. Canthium sp. 3 — Camp 3: Lewis 997. Canthium sp. A — Camp 3: Lewis (s 331). Canthium sp. 5 — Camp 2: Lewis 171. Chassalia ambodirianensis Bremek. (Massif de l'Andringitra, foret d'Ambodipaiso, 1100 m). Chassalia bojeri Bremek. var. bojeri — 50 km S Ambalavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia domi- nant, 1975 m: Nicoll 301. Chassalia sp. 1 — Camp 3: Goodman 6421. Danais fragrans Gaertn. — Camp 4: Lewis 1098. Danais paludosa Baker — Camp 2: Lewis 920; Camp 3: Lewis 1012. Gaertnera sp. 1 — Camp 1: Lewis 756, 819; Camp 2: Lewis (s 136, 211). Gaertnera sp. 2 — Camp 2: Lewis 880; Camp 3: Lewis 958. Gaertnera sp. 3 — Camp 1: Lewis (s 2). Gaertnera sp. A — Camp 4: Lewis 1092. Gaertnera sp. 5 — Camp 1: Lewis 786; Camp 2: Lewis 869; Camp 3: Lewis (s 322, 330). Gaertnera sp. 6 — Camp 1: Lewis 808. Gaertnera sp. 7 — Camp 4: Lewis (s 440), 1030, 1084. Gaertnera sp. 8 — Camp 4: Lewis (s 489). Galium andringitrense Homolle ex Puff. Mapouria assimilis Bremek. (Andringitra, 1600 m). Mussaenda arcuata Poir. — Camp 3: Lewis 969; Camp 4: Lewis 1078, 1094. Mussaenda sp. 1 — Camp 1: Lewis 791. Otiophora pauciftora Baker — 50 km S Ambala- vao, near abandoned meteo station above Am- balamarina, high altitude moss/lichen rain for- est-ericaceous bush, Philippia dominant, 1975 m: Nicoll 263. Paederia andringitrensis Homolle. Pauridiantha paucinervis (Hiern) Bremek. ssp. lyallii (Baker) Verde. — Camp 2: Goodman 6374, 6378; Lewis 851; Camp 3: Lewis 1001; Camp 4: Goodman 6443, 6454; 50 km S Am- balavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 299. Psychotria macrochlamys (Bremek) combined. — Camp 3: Lewis (s 290, 292). Psychotria sp. 1 — Camp 2: Lewis (s 172). Psychotria sp. 2 — Camp 1: Lewis (s 103); Camp 2: Lewis (s 183). Psychotria sp. 3 — Camp 3: Lewis 934. Psychotria sp. A — Camp 4: Lewis (s 415), 1095. Psychotria sp. 5 — Camp 1: Lewis 751. Psychotria sp. 6 — Camp 2: Lewis (s 142). Saldinia sp. 1 — Camp 4: Lewis 1033, 1046. Schismatoclada sp. 1 — Camp 2: Lewis 852, 887. Schismatoclada sp. 2 — Camp 2: Lewis 932; Camp 3: Lewis (s 231, 285). Tricalysia cf. cryptocalyx Baker — Camp 3: Lewis 991, 936. Tricalysia sp. 1 — Camp 4: Lewis (s 405). Tricalysia sp. 2 — Camp 1: Lewis (s 116); Camp 4: Lewis (s 379). Rutaceae Evodia sp. 1 — Camp 3: Lewis (s 282). Evodia sp. 2 — Camp 4: Lewis (s 454). Evodia sp. 3 — Camp 4: Lewis (s 433). Teclea sp. 1 — Camp 3: Lewis 985, 995. Vepris fitoravina H. Perrier — Camp 1: Lewis 797; Camp 3: Lewis 998. Vepris sp. 1 — Camp 2: Lewis (s 164). Vepris sp. 2 — Camp 3: Lewis 1021. Santalaceae Thesium pseudocystoseiroides Cavaco & Kerau- dren (Massif de l'Andringitra, 2000-2500 m). 70 FIELDIANA: ZOOLOGY Sapindaceae sp. 1 — Camp 1: Lewis (s 49, 78); Camp 2: Lewis (s 166). sp. 2 — Camp 2: Lewis (s 197). sp. 3 — Camp 1: Lewis 832. sp. 4 — Camp 2: Lewis (s 152); Camp 3: Lewis 940. Allophylus sp. 1 — Camp 2: Lewis 846. Doratoxylon sp. 1 — Camp 4: Lewis 1111. Macphersonia sp. 1 — Camp 3: Lewis (s 319). Tina sp. 1 — Camp 2: Lewis 910. Sapotaceae Chrysophyllum sp. 1 — Camp 1: Lewis (s 77). Faucheria sp. 1 — Camp 4: Lewis (s 430, 460). Labramia sp. 1 — Camp 1: Lewis (s 24). Scrophulariaceae Craterostigma perrieri Bonati — pres des eaux, vers 2000 m (Perrier de la Bathie, 1927). Halleria tetragona Baker — Sclerophyllous forest with canopy height 4-5 m, on E side Ampasi- potsy, above Camp 4, 1800-2000 m: Lewis 1053. Hydrotriche mayacoides A. Raynal (Massif de l'Andringitra, mare, 1600 m). Ilysanthes longipes Bonati — pres des eaux, vers 1200 m (Perrier de la Bathie, 1927). Sopubia trifida Buch.-Hamet — Open, level areas above village of Antanifotsy, partially dis- turbed, probably burned areas dominated by grasses, 1400 m: Lowry 4515. Solanaceae Solanum sp. 1 — Camp 4: Lewis 1083. Dombeya cf. amplifolia Arenes — Camp 3: Lewis (s 242, 320). Dombeya cf. spectabilis Bojer — Camp 2: Lewis (s 67, 156). Dombeya ivohibeensis Arenes — Camp 1: Lewis 796. Dombeya leiomacrantha Hochr. ssp. leiomacran- tha var. angustata (Hochr.) Arenes (Massif de l'Andringitra, Iratsy, vallee de la Riambava et de T Antsifotra, 2000-2500 m). Dombeya leiomacrantha Hochr. ssp. leiomacran- tha var. dissecta (Hochr.) Arenes (Massif de l'Andringitra, 2000 m). Dombeya leiomacrantha Hochr. ssp. leiomacran- tha var. leiomacrantha (Massif de l'Andringi- tra, granite, 1600 m). Dombeya leucomacrantha Hochr. var. crassibrac- tea Hochr. (Massif de l'Andringitra, ravins, 2400 m). Dombeya leucomacrantha Hochr. var. leucoma- crantha (Massif de l'Andringitra, 2400 m, brousse ericoide). Dombeya muscosa Hochr. (partie ouest de massif d'Andringitra, granite, 1600 m). Dombeya sp. 1 — Camp 4: Lewis 1042 (s 378). Dombeya sp. 2 — Camp 4: Lewis (s 445, 449, 457, 488). Dombeya sp. 3 — Camp 2: Lewis (s 196). Dombeya sp. 4 — Camp 3: Lewis 324. Strelitziaceae Ravenala madagascariensis Sonn. — Camp 1: Lewis sight record; Camp 2: Lewis sight rec- ord. Theaceae Asteropeia rhopaloides (Baker) Baill. var. angus- tata H. Perrier (W du massif d'Andringitra). Sterculiaceae Byttneria melleri Baker — Camp 1: Lewis 814. Dombeya borraginopsis Hochr. ( W du massif de l'Andringitra, pres des torrents, 1200 m). Thymelaeaceae Peddiea involucrata Baker — Ericaceous vegeta- tion, on E side Ampasipotsy, above Camp 4, 1800-2000 m: Lewis 1054; 50 km S Ambala- vao, near abandoned meteo station above Am- LEVVIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 71 balamarina, high altitude moss/lichen rain for- est-ericaceous bush, Philippia dominant, 1975 m: Nicoll 296. Rinorea urschii H. Perrier — Camp 1: Lewis 827. Viola abyssinica Steud. — "Foret a strate herbacee 1695-1840 m" (Paulian et al., 1971, p. 254). Tiliaceae Grewia sp. 1 — Camp 2: Lewis (s 154). Grewia sp. 2 — Camp 3: Lewis 1020. Urticaceae Boehmeria sp. 1 — Camp 1: Lewis 758. Elatostema subfavoswn Leandri (Massif de l'An- dringitra, vers 1100 m). Elatostema sp. 1 — Camp 2: Lewis 865. Pilea rivularis Wedd. — 50 km S Ambalavao, near abandoned meteo station above Ambalamarina, high altitude moss/lichen rain forest-ericaceous bush, Philippia dominant, 1975 m: Nicoll 252. Urera sp. 1 — Camp 1: Lewis 754. Viscaceae Viscum humbertii Lecomte "Madagascar: Centre et Hautes Montagnes, Andringitra." Viscum sp. 1 — Camp 3: Lewis 943; Camp 4: Lew- is (s 469). Viscum sp. 2 — Camp 3: Lewis 938. Viscum sp. 3 — Camp 3: Lewis 1014. Viscum sp. A — Camp 4: Lewis 1055. Viscum sp. 5 — Camp 3: Lewis 980. Viscum sp. 6 — Ericaceous vegetation above the tree line, on E side Ampasipotsy, above Camp 4, 1800-2000 m: Lewis 1050. Xyridaceae Velloziaceae Xerophyta dasylirioides Baker (van not deter- mined)— Camp 1: Lewis 772; "Vegetation rup- icole 1800-2000 m" (Paulian et al., 1971, p. 237, 257); "Flore xerophile d'altitude 2300- 2650 m" (Paulian et al., 1971, p. 247). Xerophyta dasylirioides Baker var. andringitren- sis H. Perrier (Andringitra Massif). Verbenaceae Clerodendrum sp. 1 — Camp 2: Lewis 915. Clerodendrum sp. 2 — Camp 4: Lewis (s 490). Vitex sp. 1 — Camp 3: Lewis 1016: Camp 4: Lewis 1034 (s 462). Vitex sp. 2 — Camp 4: Lewis 1075. Violaceae Rinorea cf. arborea Baill. — Camp 1 : Lewis (s 36, 58, 88, 93); Camp 2: Lewis (s 177, 187, 215). Rinorea sp. 1 — Camp 3: Lewis (s 263). Rinorea sp. 2 — Camp 4: Lewis (s 394). Xyris madagascariensis Malme — "Prairies alti- montaines 1700 m" (Paulian et al., 1971, p. 236). Appendix 4-2. Vertebrate Food Plants on the Eastern Slopes of the RNI d'Andringitra Steven M. Goodman, Beverley A. Lewis, Peter B. Phillipson, and Jeannine Raharilala During our meanderings around the forest of the RNI d'Andringitra we had the occasion to ob- serve a wide variety of vertebrates feeding on var- ious plants. In most cases, specimens of these food plants were collected. The identifications of this material were confirmed by Phillipson and Raharilala. When collected, the field numbers of voucher specimens from the catalogs of S. M. Goodman (SMG) and B. Lewis (BL) are present- ed. Voucher specimens have been deposited with Missouri Botanical Garden (MO). 72 FIELDIANA: ZOOLOGY Annonaceae Artabotrys mabifolius Diels — At 720 m, fruits on ground eaten by Nesomys rufus (BL 844). Asteraceae Vernonia alleizettei Humbert var. rienanensis Humbert — At about 1300 m, Zosterops mad- eraspatana was observed picking at fruits (SMG 6402). cifolia" at 1650 m elevation — whole fruits on ground gnawed open by Nesomys rufus in order to extract seeds. Ericaceae Vaccinium secundiflorum Hook. — At 1625 m, Hypsipetes madagascariensis was seen eating green-colored fruits (SMG 6450). Euphorbiaceae Burseraceae Canarium madagascariense Engl. ssp. obtusifol- ium (Scott-Elliott) Leenhouts — At 720 m, sev- eral nuts were found below this tree that had been opened by Daubentonia madagascarien- sis. Also, nuts were regularly found on forest floor that had rodent gnaw marks. At 810 m, 22 nuts were excavated from a burrow of Gym- nuromys roberti, 21 of which were gnawed open. Further, Canarium stones, mixed with Cryptocarya fruits, were found in a Nesomys rufus burrow. At this same elevation, nuts were regularly encountered on the forest floor that had been opened and consumed by Daubenton- ia madagascariensis and numerous species of rodents (see Chapter 6). Canellaceae Cinnamosma fragrans Baill. — At 1650 m, Cora- copsis nigra was observed feeding on fruits (BL 1077). They would remove exocarp from seed and consume the latter. Clusiaceae Symphonia sp. — Coracopsis nigra was observed feeding on unopen blossoms at 870 m in un- disturbed forest (no specimen). Macaranga oblongifolia Baill. — At 1650 m, Phi- lepitta castanea was seen feeding on small or- ange-colored buds (SMG 6458). Uapaca sp. — At 720 m, in slightly disturbed for- est, whole fruits were eaten by Eulemur fulvus. Lamiaceae Plectranthus sp. 1 — At 1210 m, Nectarinia soui- manga was seen visiting lavender-colored flow- ers (SMG 6408). Lauraceae Cryptocarya sp. — At 720 m and 810 m, whole ripe and unripe fruits were eaten by Eulemur fulvus. Animals would feed in canopy and on outer branches of tree, often filling their cheeks, and then rest in the same tree or near- by tree consuming fruits. Feces contained in- tact fruits minus exocarp. Fallen fruits were also eaten by Eliurus spp. and Nesomys rufus. Captive Eliurus preferred fruits with red exo- carps. Excavated burrows of Eliurus webbi at 720 m and Nesomys rufus at 810 m contained quantities of cached and partially eaten fruits of this genus. Alectroenas madagascariensis was observed at 810 m feeding on whole fruits (no specimen). Elaeocarpaceae Sloanea rhodantha (Baker) Capuron var. rhodan- tha form "quadriloba" at 810 and form "quer- Loranthaceae Bakerella clavata (Desr.) Balle var. lenticellata (Baker) Balle — At about 1625 m, Neodrepanis hypoxantha, Nectarinia souimanga, and Zoster- LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 73 ops maderaspatana were seen feeding on ver- milion-colored flowers (SMG 6453 and BL 1029). Bakarella clavata (Desr.) Balle var. peralata (Le- comte) Balle — At 720 m, Neodrepanis corus- cans and Zosterops maderaspatana and at 1210 m, Nectarinia souimanga were observed visit- ing the dull reddish-pink flowers of this plant. Bakerella tandrokensis (Lecomte) Balle — At 2050 m, Neodrepanis hypoxantha, Nectarinia souimanga, and N. notata were seen feeding on dull reddish-pink flowers (SMG 6459). Melastomataceae Myrsinaceae Maesa lanceolata Forssk. — At 740 m, fruits were eaten by Alectroenas madagascariensis. Oncostemum ankifiense Mez — At 1625 m, Zos- terops maderaspatana was seen plucking and eating flower buds (SMG 6430). Oncostemum racemiferum Mez — Tree near forest edge at 730 m; whole ripe fruits were con- sumed by Alectroenas madagascariensis and Hypsipetes madagascariensis. Neomixis tenella and Zosterops maderaspatana were also at- tracted to fruits, but it was not clear if they were actually eating fruits or insects attracted to them (SMG 6352). Gravesia calliantha Jum. & H. Perrier — At 1210 m, Philepitta castanea was seen feeding on fruits (SMG 6409). Medinilla ericarum Jum. & H. Perrier — At 1210 m, Hypsipetes madagascariensis and Philepitta castanea were seen feeding on orangish-red fruits (SMG 6407). Medinilla torrentum Jum. & H. Perrier — At about 1650 m, Nectarinia souimanga was seen visit- ing white-colored flowers and apparently pierc- ing unopened flowers with its bill (SMG 6429). Monimiaceae Myrtaceae Syzygium sp. 1 3 — Whole ripe fruits were eaten by Eulemur fulvus, at about 710 m, at ecotone be- tween forest edge and tavy (BL 776). Orchidaceae Liparis sp. 1 — At 810 m, Neodrepanis coruscans and Nectarinia souimanga were seen visiting flowers (SMG 6381). Tambourissa cf. parvifolia Baker — At 1240 m, whole fruits were eaten by Eulemur fulvus (BL 964). Moraceae Pittosporaceae Pittosporum sp. — At 800 m, whole fruits were eaten by Philepitta castanea (no specimen). Ficus brachyclada Barker — Fruits were eaten by birds at 810 m (BL 841). Ficus reflexa Thunb. — Whole fruits were eaten by Alectroenas madagascariensis and Streptopelia picturata at edge of clearing at 710 m (SMG 6353). Ficus rubra Vahl — Fruits were eaten by birds at 810 m (BL 866). Ficus trichopoda Baker — Fruits were eaten by birds at 720 m (BL 769). Streblus dimepate (Bureau) C.C. Berg — At 730 m, whole fruits were eaten by Alectroenas mad- agascariensis in slightly disturbed forest. Poaceae Nastus sp. — At 810, 1250, and 1700 m, bamboos were found that had been gnawed open by Dau- bentonia madagascariensis, presumbly to ex- tract insect larvae (no specimen). Podocarpaceae Podocarpus madagascariensis Baker — At 810 m, Eulemur fulvus was observed eating fruits (no specimen). 74 FIELDIANA: ZOOLOGY Rubiaceae Pauridiantha paucinervis (Hiern) Bremek. ssp. lyallii (Baker) Verde. — At 810 m, small red Chassalia sp. 1 — At 1210 m, Coracopsis nigra, fruits were eaten by Philepitta castanea, Hyp- Philepitta castanea, and Hypsipetes madagas- sipetes madagascariensis, and Zosterops mad- cariensis were eating whole bright red fruits eraspatana (SMG 6374, 6378), and at 1625 m, (SMG 6421). by Philepitta castanea (SMG 6454). LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 75 Chapter 5 The Pteridophytes of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar France Rakotondrainibe and Fidele Raharimalala Abstract An elevational survey of the pteridophytes occurring along the eastern slopes of the Reserve Naturelle Integrale d'Andringitra was conducted. A total of 188 species were recorded. Within the transect zones, 76 species were collected at 720 m, 115 at 800-900 m, 96 at 1200-1300 m, and 74 at 1600-1700 m. On the basis of species turnover of ferns, the transition zone between lowland and montane forest is between 700 and 900 m, and the transition between montane and sclerophyllous forest is between 1550 and 1600 m. Resume Une etude sur la repartition altitudinale de Pteridophytes a ete menee sur le versant Est de la Reserve Naturelle Integrale d'Andringitra. 188 especes au total ont ete recensees. Le long du transect altitudinal, 76 especes ont ete observees a 720 m, 1 1 5 entre 800 et 900 m, 96 entre 1200 et 1300 m et 74 entre 1600 et 1700 m. La composition floristique des communautes de fougeres permet de situer la zone de transition entre la foret de basse altitude et la foret de montagne a 700-900 m et celle entre la foret de montagne et la foret sclerophylle a 1500- 1600 m. Introduction Methods Our current knowledge of the taxonomy of Malagasy pteridophytes and their distribution is largely based on the work of Christensen (1932), Perrier de la Bathie (1932), and Tardieu-Blot (1951-1971). Using this information, Tardieu- Blot (1948), Leroy (1978), and Jacobsen (1991) discussed the origin and the affinities of this group of plants. In recent years, under the direction of biological inventories associated with conserva- tion projects, a considerable amount of new in- formation has been gathered on the island's pte- ridophytes. This new material advances our knowledge of this flora. Between 13 May and 7 June 1995 we con- ducted a survey of the pteridophytes of the Re- serve Naturelle Integrale (RNI) d'Andringitra, us- ing the same transect as the late 1993 mission (see Chapter 1). The 720 m zone was occupied be- tween 13 and 16 May, the 810 m zone between 17 and 22 May, the 1210 m zone between 23 and 30 May, and the 1625 m zone between 31 May and 5 June. Further, general collections were made in secondary grassland and disturbed forest between 700 and 1 150 m. Voucher specimens are deposited in the Museum National d'Histoire Na- turelle, Paris (P), and FOFIFA, Antananarivo (TEF). 76 FIELDIANA: ZOOLOGY The collections made represent the general pte- ridophyte flora of the eastern slopes of the re- serve. The majority of plants were terrestrial and/ or epiphytic (between 0.5 and 8 m above the ground). A few species, approximately 20, were found between 15 and 20 m above the ground. A considerable portion of the information on epi- phytic canopy pteridophytes was obtained from fallen emergent trees. Further, canopy plants were identified with binoculars. The current nomenclature for the Cyatheaceae is not parallel between continents and is in need of major revision. Thus, we have chosen to utilize a somewhat out-of-date system that reflects the standard taxonomy for Madagascar. Results and Discussion Table 5-1 summarizes the pteridophyte inven- tory in an altitudinal range of 700-1700 m on the eastern slopes of the RNI d'Andringitra. Several vegetation types of the Central Domain (for phy- togeographical division of Madagascar and forest nomenclature, see Humbert, 1955, and White, 1983) were studied: (1) disturbed forests and sec- ondary grasslands located outside of the reserve (columns I, II, and III of Table 5-1), and (2) the mid-montane moist forest within the reserve and its two ecotones between 720 and 900 m with lowland forest and between 1500 and 1600 m with the sclerophyllous forest (column IV of Ta- ble 5-1). A total of 188 species of pteridophytes were found, among which 70 (37.2%) are endemic to Madagascar. Prior to the present study, 61 taxa of pteridophytes had been reported from the Andrin- gitra Massif (Tardieu-Blot, 1951-1971). This fig- ure, when compared to those of other reserves on the island, indicates a very high species richness, stemming from the diversity of vegetation in the area. For example, Rahajasoa (1993) identified 155 taxa of pteridophytes in the Pare National (PN) de Ranomafana between 900 and 1200 m, Rakotondrainibe and Quansah (1994) found 169 taxa in the Reserve Sp6ciale de Manongarivo be- tween 200 and 1 876 m, and Rakotondrainibe and Raharimalala (in press) identified 156 taxa along the western slopes of the Masoala Peninsula be- tween 0 and 1 105 m. The disturbed forests and secondary grasslands outside the reserve boundaries still contain a high species diversity (105 taxa) in comparison to that within the protected area (164 taxa). However, the floristic composition of the endemic pteridophytes in the disturbed forests is 35.2%, compared to 39.6% within the reserve. In the undisturbed for- est, the species richness gradually decreases with increasing altitude: 1 15 taxa are found at 800-900 m, 96 taxa at 1200-1300 m, and 74 taxa at 1600- 1700 m. The most characteristic species of the following altitudinal zones in the RNI d'Andringitra are: up to 900 m — Asplenium bipartitum, A. dregeanum, Christella distans, Ctenopteris excaudata, Micro- lepia madagascariensis, Pteris catoptera, Tectar- ia magnified, Trichomanes bipunctatum, and T. rotundifolium; 800-1300 m — Asplenium preusii, Elaphoglossum scolopendriforme, Gymnosphaera boivinii, and Gymnosphaera coursii; 1300-1700 m — Elaphoglossum subsessile, Grammitis holo- phlebia, and Rumohra lokohensis; and 1600-1700 m — Blechnum humbertii, Dryopteris remotipin- nula, Elaphoglossum aubertii, and Hymenophyl- lum peltatum. The ecotone of the lowland and montane forest occurs between 700 and 900 m. This transition is easily defined by a group of low- land ferns whose upper elevational limit is at about 900 m (Nephrolepis biserrata, Phymatoso- rus scolopendria, and Platycerium madagascar- iense) and a group of montane species whose low- er limit appears to be at approximately 900 m (As- plenium anisophyllum, Elaphoglossum forsythii- majoris, Oleandra distenta, and Trichomanes speciosum). The transition between montane and sclerophyllous forest is well demarcated between 1500 and 1600 m by both abundance and distri- bution of several species (e.g., Blechnum humber- tii, Dryopteris remotipinnula, Elaphoglossum au- bertii, and Hymenophyllum peltatum). Indepen- dent of elevation, the distribution of Elaphoglos- sum spathulatum, E. petiolatum var. salicifolium, and Xiphopteris hildebrandtii is closely correlated with the microhabitat of rocks along riverbanks. The presence and/or abundance of the follow- ing taxa indicate some human or natural distur- bance: on the edge of the forest — Lycopodium cernuum, L. clavatum, Blechnum punctulatum, and Mohria cqffrorum; in clearings surrounded by forest — Hypolepis villoso-viscida and Pteris pseu- dolonchitis; in the secondary grasslands succeed- ing forest fires — Pteridium aquilinum, Sticherus flagellaris, and Dicranopteris linearis; and in dis- turbed forests, particularly along trails — Amauro- pelta bergiana, Ctenopteris villosissima, Cyathea melleri, Elaphoglossum deckenii var. rufidulum, and Selaginella lyallii. Several taxa of pterido- RAKOTONDRAINIBE & RAHARIMALALA: PTERIDOPHYTES OF EASTERN SLOPES 77 Table 5-1. The pteridophytes of the Andringitra Massif: elevational distribution and fioristic composition of populations in disturbed (columns I, II, and III) and undisturbed forest (column IV). Taxa Vegetation Types: I and II III Altitude (m): 700-1150 720 rv rv iv 800-900 1200-1300 1600-1700 Amauropelta aff. oppositiformis (C.Chr.) Holttum Amauropelta aff. strigosa (Willd.) Holttum Amauropelta bergiana (Schltr.) Holttum Anthrophyum boryanum (Willd.) Kaulf. Antigramma virchowii (E. Kuhn) Tardieu Arthropteris monocarpa (Cordem.) C.Chr. Arthropteris parallela C.Chr. Asplenium aethiopicum (Burm.f.) Bech. Asplenium anisophyllum Kunze Asplenium auritum Sw. Asplenium bipartitum Bory Asplenium cancellatum Alston Asplenium cuneatum Lam. Asplenium dregeanum Kunze Asplenium rectum var. zeyheri (Pappe & Rawon) Alston & Schelpe Asplenium erectum Willd. var. erectum Asplenium friesiorum C.Chr. Asplenium herpetopteris Baker var. herpetopteris Asplenium herpetopteris var. massoulae (Bonap.) Tardieu Asplenium inaequilaterale Willd. Asplenium lividum E. Kuhn Asplenium mannii Hook. Asplenium nidus L. Asplenium normale D. Don Asplenium obscurum Blume Asplenium pellucidum Lam. Asplenium petiolulatum Mett. Asplenium poolii Baker Asplenium preusii Brause Asplenium protensum Schrad. Asplenium rutifolium (P.J.Bergius.) Kunze Asplenium sandersonii Hook. Asplenium theciferum (Kunth) Mett. Athyrium scandicinum (Willd.) C.Presl. Belvisia spicata (L.f.) Mirb. Blechnum attenuatum (Sw.) Mett. Blechnum attenuatum var. giganteum (Kaulf.) Bonap. Blechnum humbertii Tardieu Blechnum ivohibense C.Chr. Blechnum punctulatum Sw. Blechnum simillimum (Baker) Diels fa binerve (Hook.) Tardieu Blechnum simillimum (Baker) Tardieu var. simillimum Blechnum simillimum var. xiphophyllum (Baker) Tardieu Blotiella madagascariensis (Hook.) Tard. Cheilanthes madagascariensis Baker Christella dentta (Forssk.) Holttum Christella distans (Hook.) Tardieu ** ** o * ** *** ** *** o 0 *** ** o ** **** o ** * * *** *** *** ** * o * *** *** ** * ** *** *** *** ** ** ** 0 ** ** 0 0 o o * **# 0 * *#* *#** **** ** ** 78 FIELDIANA: ZOOLOGV Table 5-1. Continued. Vegetation Types: I and II III IV IV IV Taxa Altitude (m): 700-1150 720 800-900 1200-1300 1600-1700 Coniogramme madagascariensis C.Chr. * *** ** Ctenitis lanuginosa (Kaulf.) Copel. o Ctenitis magma (Baker) Tardieu * *** * Olenitis poolii (C.Chr.) Tardieu * Ctenopteris devoluta (Baker) Tardieu * * * * Ctenopteris elastica (Willd.) Copel. ** ** O Ctenopteris excaudata (Bonap.) Tardieu * ** *** o Ctenopteris rigescens (Bory) J.Sm. ** Ctenopteris villosissima (Hook.) Harley *** * * * Cyathea approximata Bonap. * ** Cyathea bellisqamata Bonap. ** Cyathea boivinii E.Kuhn * * Cyathea borbonica Desv. var. laevigata Bonap. * ** Cyathea decrescens E.Kuhn * *** Cyathea melleri (Baker) Domin *** * Cyathea melleri (Baker) Tardieu var. virescens Tardieu o Cyathea similis C.Chr. ** o Cyathea tsaratananensis Tardieu * *** *** *** ** Deparia bory ana (Willd.) Kato •*• *** Deparia parvisora (C.Chr.) Kato *** Dicranopteris linearis (Burm.f.) Underw. ** Diplazium zakamenense (Tardieu) Rakotondr. * * * Dryopteris mangindranensis Tard. * * * * o Dryopteris remotipinnula Bonap. *** Dryopteris subcrenulata (Baker) C.Chr. o Elaphoglossum achroalepis (Baker) C.Chr. ** Elaphoglossum acrostichoides (Hook.&Grev.) *** Elaphoglossum aff. ovalilimbatum Bonap. * *** * Elaphoglossum scolopendriforme Tardieu *** **** *** Elaphoglossum angulatum (Blume) T.Moore * Elaphoglossum aubertii (Desv.) T.Moore o *** Elaphoglossum coriaceum Bonap. o ** Elaphoglossum decaryanum Tardieu o ** Elaphoglossum deckenii (Kuhn) C.Chr. var. deckenii * Elaphoglossum deckenii var. rufidulum (Willd.) Tardieu *** ** * Elaphoglossum forsythii-majoris H.Christ. o *** Elaphoglossum humbertii C.Chr. ** o Elaphoglossum hybridum (Bory) Brack. ** *** o o Elaphoglossum leucolepis (Baker) Krajina * * o Elaphoglossum petiolatum (Sw.) J.B.Urb. ssp. salicifolium (Kaulf.) Schelpe o ** Elaphoglossum sieberi (Hook.&Grev.) T.Moore * ** o Elaphoglossum spathulatum (Bory) T.Moore *** Elaphoglossum sp. 6 O * ** Elaphoglossum sp. 7 * ** Elaphoglossum sp. 8 * *** Elaphoglossum subsessile (Willd.) Christ. ** *** *** Grammitis barbatula (Baker) Copel. o ** Grammitis cryptophlebia (Baker) Copel. * o Grammitis holophlebia (Baker) Copel. ** *** Gymnosphaera andohahelensis (Tardieu) Tardieu *** RAKOTONDRAINIBE & RAHARIMALALA: PTERIDOPHYTES OF EASTERN SLOPES 79 Table 5-1. Continued. Taxa Vegetation Types: Altitude (m): I and II 700-1150 III 720 rv 800-900 rv iv 1200-1300 1600-1700 Gymnosphaera boivinii (Ettingsh.) Tardieu Gymnosphaera coursii (Tardieu) Tardieu Gymnosphaera sp. 1 (aff. boivinii) Histiopteris incisa (Thunb.) J.Sm. Huperzia megastachya (Baker) Tardieu Huperzia obtusifolia (Sw.) Rothm. Huperzia pecten (Baker) Tardieu Huperzia pichiana Tardieu Huperzia rubrica (Herter) Tardieu Huperzia sqarrosa (Forssk.) Trevis Huperzia verticillata (L.f.) Trevis Hymenophyllum capense Schrad. Hymenophyllum hirsutum (L.) Sw. Hymenophyllum humbertii C.Chr. Hymenophyllum peltatum (Poir.) Desv. Hymenophyllum perrierii Tardieu Hymenophyllum polyanthos (Sw.) Sw. Hymenophyllum polyanthos var. kuhnii (C.Chr.) Schelpe Hymenophyllum poolii Baker Hymenophyllum sibthorpioidews Mett. Hymenophyllum veronicoides C.Chr. Hypolepis villoso-viscida (A. Thouars) Tardieu Lindsaea flabellifolia (Baker) Kuhn Lindsaea madagascariensis Baker Lindsaea odorata Roxb. Lomariopsis pollicina (P. Willemet) Mett. Lonchitis occidentalis Baker Loxogramme lanceolata (Sw.) C.Presl. Lycopodium carolinianum L. var. affine (Bory) Pic.Serm. Lycopodium cernuum L. Lycopodium clavatum L. var. borbonicum Bory Lycopodium zanclophyllum Wilce Lygodium lanceolatum Desv. Marattia fraxinea J.F.Gmel Microlepia madagascariensis C.Presl. Microsorum punctatum (L.) Copel. Mohria caffrorum (L.) Desv. Nephrolepis biserrata (Sw.) Schott Nephrolepis undulata (Sw.) J.Sm. Odontosoria melleri (Hook.) C.Chr. Oleandra distenta Kunze Ophioglossum palmatum L. Osmunda regalis L. Pellaea angulosa (Bory) Baker Phymatosorus scolopendria (Burm.f.) Pic.Serm. Pityrogramma argentea (Willd.) Domin. Platycerium madagascariense Baker Pleopeltis excavata (Willd.) Sledge Pleopeltis macrocarpa (Willd.) Kaulf. Pneumatopteris subpennigera (C.Chr.) Holttum *** ** ** *** ** *** * * * o o 0 o o 0 0 * o *** *** ** * *** *** **** * * ** *** *** 0 * * ** *** **** o * ** ** *** *** o ** * o o ** ** *** 0 *** *** ** o 0 * * *** *** *#* *** * *** *** *** *** o *** o 80 FIELDIANA: ZOOLOGY Table 5-1. Continued. Taxa Vegetation Types: Altitude (m): I and II 700-1150 III 720 rv 800-900 iv rv 1200-1300 1600-1700 Pneumatopteris unita Kunze Polystichum coursii Tardieu Pseudophegopteris cruciata (Willd.) Holttum Pseudophegopteris pulcher (Willd.) Holttum Pteridium aquilinum (L.) E.Kuhn Pteris catoptera Kunze Pteris cretica L. Pteris griseoviridis C.Chr. Pteris humbertii C.Chr. Pteris lane at folia J.Agardh Pteris pseudolonchitis Kuhn Rumohra adiantiformis (G. Forst.) Ching Rumohra lokohensis (Forssk.) Ching Saccoloma henriettae C.Chr. Schizaea dichotoma (L.) Sw. Selaginella fissidentoides Spring Selaginella goudotana Spring Selaginella lyallii Baker Selaginella pectinata (Willd.) Spring Sphaerostephanos arbuscula (Willd.) Holttum Sticherus flagellaris (Bory) St John Tectaria magnified (Bonap.) Tardieu Trichomanes bipunctatum Poir. Trichomanes bonapartei C.Chr. Trichomanes borbonicum Bosch Trichomanes cupressoides Desv. Trichomanes digitatum Sw. Trichomanes lenormandii Bosch Trichomanes longilabiatum Bonap. Trichomanes mannii Hook. Trichomanes meifolium Willd. Trichomanes montanum Hook. Trichomanes rigidum Sw. Trichomanes rotundifolium Bonap. Trichomanes speciosum Willd. Mi ta ha humblotii Hieron. Vittaria isoetifolia Bory Xiphopteris hildebrandtii (Hieron.) Tardieu Xiphopteris oosora (Baker) Alston var. micropecten C.Chr. Xiphopteris serrulata (Sw.) Kaulf. Xiphopteris sp. 2 *** *** ** • * 0 *** * *** o ** *** ** *** **** *** *** ** * *** * o *** *** ** ** *** *** *** o *** *** *** * * * * * * * *** * * o * •* * * *** ** * *** 0 ** *** *** o Total number of taxa: 188 52 76 115 96 74 105 164 Number of endemic taxa: Percentage: 70 37.52% 37 35.2% 65 39.6% I, savannah; II, relict forest; III, disturbed forest; that it is endemic to Madagascar; a taxon printed i southern part of the Central Domain of Malagasy. oRare. * Infrequent. ** Frequent. *** Common. **** Very common. IV, undisturbed forest. A taxon printed in boldface type indicates in boldface type and underlined indicates that it is endemic to the RAKOTONDRAINIBE & RAHARIMALALA: PTERIDOPHYTES OF EASTERN SLOPES 81 phytes occur only in primary or slightly disturbed forest: Anthrophyum boryanum, Asplenium man- nii, A. anisophyllum, Blechnum ivohibense, and Trichomanes rotundifolium. Further survey work and collections are needed to develop our understanding of the relationships and altitudinal distribution of the Malagasy pte- ridophyte flora. Techniques are being developed to quantify measures of abundance along such gradients; these techniques will allow compari- sons between sites. Several students are currently writing theses associated with taxonomic revi- sions of this flora. Acknowledgments We are grateful to the staff of the World Wide Fund for Nature, particularly Peter Schachenmann and Hanta Rabetaliana, and the people of Amba- lamenjana for making our mission possible. The Laboratoire de Phytomorphologie of the Ecole Pratique des Hautes Etudes, under the direction of Dr. A. Le Thomas, provided partial financial sup- port for this project. Literature Cited Christensen, C. 1932. The pteridophyta of Madagas- car. Dansk Botanisk Arkiv, 80: 1-253. Humbert, H. 1955. Les territoires phytogeographiques de Madagascar. In Les divisions Fcologiques du mon- de. Colloques Internationaux du Centre National de Recherche Scientifique LIX. Annee biologique, se>. 3, 31(5-6): 439-448. Jacob-sen, W. B. G. 1991. Phytogeographical analysis of the pteridophytes of some randomly selected areas of the world. Thaiszia, 1: 69-94. Leroy, J.-F. 1978. Composition, origin, and affinities of the Madagascan vascular flora. Annals of the Missouri Botanical Garden, 65: 535-589. Perrier de la Bathie, H. 1932. Distribution dans rile, pp. 207-221. In Christensen, C, ed. The pteridophyta of Madagascar. Dansk Botanisk Arkiv, 80: 1-253. Rahajasoa, G. M. 1993. Analyse de la Diversite des Pteridophytes dans le Pare national de Ranomafana: Inventaire et mode de repartition. Diplome d' Etudes Approfondies de Sciences Biologiques, option Ecol- ogie V6getale, Universite d'Antananarivo. Rakotondrainibe, E, and N. Quansah. 1994. The di- versity and originality of the pteridophytes of the for- est of Manongarivo Special Reserve (north-west Mad- agascar). Fern Gazette, 14: 259-267. Rakotondrainibe, E, and Raharimalala, E In press. Contribution a 1' etude de la flore des aires protegees de Madagascar: Les Pteridophytes de la presqu'ile Masoala. Centre d' Information et de Documentation Scientifique et Technique, Antananarivo, Serie Sci- ences biologiques. Tardieu-Blot, M. L. 1948. Le peuplement ptendolo- gique de Madagascar. Memoire de l'lnstitut Scienti- fique de Madagascar, sen B, 6: 219-243. Tardieu-Blot, M. L. 1951-1971. Les Pteridophytes. In Humbert, H., ed. Flore de Madagascar et des Co- mores, Fam. 1 a 13, Museum National d'Histoire Na- turelle, Paris, France. White, E 1983. The vegetation of Africa. UNESCO/ AETFAT/UNSO, Paris, 356 pp. 82 FIELDIANA: ZOOLOGY Chapter 6 The Utilization of Canarium (Burseraceae) Seeds by Vertebrates in the Reserve Naturelle Integrale d'Andringitra, Madagascar Steven M. Goodman and Eleanor J. Sterling Abstract The aye-aye (Daubentonia madagascariensis) is known to feed extensively on the fat-rich seeds of Canarium, which are encased in an extremely hard endocarp. Endemic Nesomyinae rodents and introduced Rattus rattus captured in the forests of Reserve Naturelle Integrale d'Andringitra were held in captivity and fed Canarium seed capsules to see if they were able to penetrate the endocarp and consume the cotyledon. Five of the eight endemic rodents tested, as well as Rattus, fed on the seeds in captivity. Information is presented on size and pattern of feeding signs on Canarium seed capsules recovered from the forest floor and rodent burrows. Further, we review information on the utilization of Canarium fruits and seeds by other ver- tebrates and on possible competitive interactions by these animals for this food resource. Resume II est connu que 1' Aye-aye {Daubentonia madagascariensis) se nourrit en grande partie de graines de Canarium riches en lipides, qui sont enveloppees par un endocarpe tres dur. Dans les forets de la Reserve Naturelle Integrate d'Andringitra, nous avons capture des Nesomyinae, rongeurs end^miques, et Rattus rattus, espece introduite. Ces especes ont €t€ gardens en cap- tivity et ont €i€ nourries avec des fruits de Canarium afin de tester s'ils pouvaient percer 1' endocarpe et consommer la graine. Sur les huit especes de rongeurs endemiques prenant part au test, cinq especes endemiques et Rattus rattus se sont nourris des graines lorsqu'ils etaient en captivite. On pr6sente ici des informations sur la taille et sur la fa?on de manger ces fruits de Canarium retrouves sur le sol et dans les terriers des rongeurs. De plus, nous passons en revue les informations relatives a l'utilisation des fruits et des graines de Canarium par d'autres especes de vertdbrds et pr6sentons de possibles interactions comp£titives entre ces animaux pour l'acces a cette ressource alimentaire. Introduction aye-aye {Daubentonia madagascariensis) is closely related to that of C. madagascariense Canarium madagascariense (Burseraceae) is a (Iwano & Iwakawa, 1988). Recent fieldwork on common tree at low elevations in the eastern hu- the Reserve Sp6ciale de Nosy Mangabe, an island mid forests of Madagascar (White, 1 983). Its dru- south of Maroantsetra, has shown that aye-ayes paceous fruits provide an important food resource spend from 30% to 89% of their foraging time for numerous vertebrates (Sterling, 1994). In fact, feeding on the seeds of Canarium (Sterling, it has been proposed that the distribution of the 1994). These fruits have a fleshy pericarp and an GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 83 extremely hard three-loculed endocarp, the latter often surpassing 250 kg/mm2 crushing resistance with a spring tester gauge (Sterling, 1994). Their nutritious cotyledon contains up to 60% fat, in addition to numerous minerals and fat-soluble vi- tamins (Sterling et al., 1994). In short, if an ani- mal has the ability to get through the endocarp of the seed, a valuable food resource lies within. In the Reserve Naturelle Intdgrale (RNI) d'Andringitra, where Canarium madagascariense Engl. ssp. obtusifolium (Scott Elliott) Leenhouts has been identified at altitudes of up to 1000 m (Chapter 4), the distinctive seed capsules of this tree (Fig. 6-1) were regularly encountered on the forest floor. Seed capsules were most often con- centrated under Canarium trees, but single ex- amples or clusters were found regularly on the ground a considerable distance from Canarium trees. In order to investigate the utilization of these seeds by the various mammalian vertebrates occurring in this forest, we collected random sam- ples of seed capsules found on the ground. Fur- ther, as part of the small mammal study (Chapter 22), a wide variety of rodents were captured alive in mammal traps, and we presented whole Can- arium seed capsules to them in order to determine if they were able to penetrate the hard endocarp, if they left distinctive gnawing marks on the en- docarps, and if they ate the cotyledons. Finally, numerous rodent burrow systems were excavated, and cached seed capsules and fruits, including Canarium, were recovered. In this chapter we ask three questions: (1) What vertebrates in the RNI d'Andringitra feed on Can- arium fruits and seeds? (2) Do they leave distinc- tive gnawing marks? (3) Is there any evidence of differential utilization of this resource that might indicate competition between the different ani- mals feeding on the seeds? capsules of C. madagascariense and boivini are not distinguishable from one another (Perrier de la Bathie, 1946), for the purposes of this paper it is simpler to follow the taxonomy of Leenhouts (1959). Samples of Canarium seed capsules were picked up under trees using a quadrate method. Three 1-m2 squares were placed at preset distanc- es (1 m east, 4 m west, and 2 m south) from the base of single trees at 720 and 810 m elevation. All seed capsules within the squares were picked up for later sorting and measuring. Also, several collections were made along trails and scattered localities in the forest. Techniques associated with small mammal trap- ping are described in Chapter 22. In each eleva- tional zone, live animals were housed in National traps in secluded areas covered by a tarpaulin at the periphery of the camp. Generally, two or three whole seed capsules of C. madagascariense, sup- plemented by wet cooked rice, were placed in each trap. Individual animals were never kept in captivity for more than 48 hours. Canarium seed capsules were transported to transect zones above the elevational limit of this tree; otherwise all seed capsules fed to captive animals were from the respective altitudinal zone. When seemingly active rodent ground burrows were encountered, traps were placed at the en- trance, and those burrows that yielded animals were subsequently excavated and searched for cached seed capsules or food remains. Opened seed capsules found on the ground or in burrows were divided into three classes according to how they were opened: aye-aye, rodent, and natural dehiscence (see Fig. 6-1). Measurements of Can- arium seed capsules, taken with dial calipers to the nearest 0.1 mm, include maximum length along the main axis and maximum width across the broadest portion of the midsection. Methods In the most recent revision of Malagasy Can- arium, Leenhouts (1959) recognized a single spe- cies, C. madagascariense. In earlier treatments of this group various numbers of species had been considered valid. For example, Perrier de la Ba- thie (1946) recognized two distinct taxa: C. mad- agascariense, a large leaf form, and C. boivini, a smaller leaf form. Recent field-workers recognize at least two species on the island (Andrianisa, 1989; G. Schatz, pers. comm.). Because the seed Results and Discussion Feeding Experiments A total of nine species of captive rodents, in- cluding introduced Rattus rattus and eight species of endemic Nesomyinae, were fed the seed cap- sules of Canarium madagascariense (Table 6-1). Six of these rodents consumed Canarium seeds by gnawing a small hole in the central portion of the seed capsule and then extracting the cotyle- 84 FIELDIANA: ZOOLOGY Table 6-1. Results of feeding experiments of captive rodents in the RNI d'Andringitra. Gnaw marks, longitu- Species Mean mass Fed on seed dinal- Trapped on Trapped off (number of individuals tested) (g)* in captivity central ground ground Rattus rattus (3) 103.4(11) Yes Yes Yes Yes Brachyuromys ramirohitra^ ( 1 ) 117 (1) Yes Yes Yes No Eliurus majorif (2) 96.6 (20) No — Yes Yes Eliurus minor (3) 36.9 (25) No — Yes Yes Eliurus tanala (2) 81.7(23) Yes Yes Yes Yes Eliurus webbi (2) 72.3 (22) Yes Yes Yes Yes Gymnuromys roberti (2) 119.3(3) Yes Yes Yes No Monticolomys koopmani\ ( 1 ) 26.2 (3) No — Yes* Yes Nesomys rufus (4) 158.5(30) Yes Yes Yes No * Mean body mass of adults based on data in Chapter 22; the figure in parentheses is the number of individuals weighed, t Species occurred above known elevational range of Canarium. % The single individual taken on the ground was in a pitfall bucket. don. Three of the rodent species tested occur above the known elevational limit of Canarium in the RNI d'Andringitra. One of these species, Brachyuromys ramirohitra, readily opened the seed capsules, whereas Eliurus majori and Mon- ticolomys koopmani did not consume the seeds in captivity. At other sites in the eastern humid for- est, B. ramirohitra is known to occur at about 900 m (Carleton & Schmidt, 1990). Further, at other sites Canarium is known from elevations up to 1800 m (Leenhouts, 1959). Of the six rodent species occurring within the elevational zone of Canarium, five consumed the seeds in captivity. The exception was Eliurus mi- nor, adults of which weigh on average 36.9 g, the smallest of any of the rodents known below 1000 Fig. 6-1. Comparison of Canarium seeds (from left to right) opened by natural dehiscence, aye-aye, Neso- mys rufus, Rattus rattus, and Eliurus tanala. m in the RNI d'Andringitra. With the exception of Gymnuromys and Nesomys, which were cap- tured only on the ground, all species within this zone appear to be both terrestrial and arboreal (Table 6-1), and in principle are able to exploit fruits and seeds still on the trees as well as those that have fallen to the ground. The possibility ex- ists that arboreal rodents feed on fruits and seeds still attached to the Canarium trees. There has been some confusion in the literature about the gnawing marks made by aye-ayes on Canarium seed capsules. Without exception, aye- ayes on Nosy Mangabe remove the exocarp and pericarp by creating serrated edges all the way around one end of the fruit and removing a "cap" in one piece. They open the seed capsules by set- ting the superior incisors along the mid-endocarp, gnawing a groove with the inferior incisors into the apex, and then using the incisors as a lever to open the seed. Because they pry open the seeds from a gnawed groove at the apex, the point of breakage is always toward the mid-portion. They then extract the cotyledon with their thin middle finger (Sterling, 1993, 1994). Thus, the incisor groove at the apex is unique to seed capsules opened by aye-ayes (Fig. 6-1). These marks are easily distinguished from those of rodents. Ro- dents generally opened the seed capsules by gnawing a small, usually transverse, hole through the mid-endocarp along the central axis (Fig. 6- 1 ). In rare instances, the endocarp is gnawed open toward or at the apex of the seed capsule. In all cases the widths of the incisor gnaw marks are distinctly smaller than those of the aye-aye. GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 85 — e V w - U 0, £ jU 5 "o £ — =i B V - •tori B V u - V ii S 2- e V 8 .23 2 £ 5 13 £ E a « z A "So fi ^ - ■— B V u i- V On V ;►> ? J3 V < £ £ W> B 0) - -*- B V u la V 2. £ ** s 3 V 0 £ P£ £ =Jt S 4> - c ■a _o <-> ■o 1 s © 0 3 C - a e B - c © a s u W ON ^O +i 7 «"> ^ -* +l^ ii SO ON _; on* +i 7 ■<* K2 © ^d +1 J oc nC v© « 3" --• **: co co *' 2} vo vd S +l ' ii +' ' ii 2 co 2 ^ CO CO •£ V) ON On CO o CM + 1 cm 1 NO o — cm ■■*- nC co o 00 — L< 00 cn 00 1 E ^ CM in (■* CM «* co 3 + 1 1 li II C CM © -4; Z 1-* z Z CO m '* U 3 15 oo o > rr ©' E 3 E 1 CM ■ E ON 1 E 3 E oo c II H II a Z10 z — ; o IN + 1 I >n 22 co oo + 1 I §2 CM co oo _; o CM +1 I co © _; r- +i 7 .n<-i O^_oo"^ cn ^ tN r» *28 oooo rtCO * o £ ri a" oo rr cm ^ ^ no gj +1^ ii +i ' +i i ii +i i cj co f^ co O co co I Oh ?s *1 VI C U i 5 2, > 3 (U £ -B St On the basis of the feeding experiments and the extensive experience of E.J.S. with aye-ayes, the seed capsules with central gnawed holes that Duckworth (1993) recorded as possible signs of aye-aye were most likely opened by a rodent. No difference was found in the gnaw marks between rodent species, although this would be worth fur- ther investigation, because the position of the opening along the axis may be species-specific. In summary, on the basis of our field observations and feeding experiments with captive rodents, it was possible to distinguish between seeds opened by the natural dehiscence of the capsules, by aye- ayes, and by rodents. Random Collections of Seed Capsules Information on the various random collections of Canarium seed capsules found within the RNI d'Andringitra at 720 and 810 m is presented in Table 6-2. There was considerable variation be- tween the four samples in the percentage of seed capsules unopened (range, 0-33.3%). In all four samples, rodents were the most common source of opened seed capsules; this varied from a min- imum of 38.1% to a maximum of 87.5%. Opened seed capsules allocated to aye-ayes were found in two of the four collections and never represented more than 12.5% of the total sample. Seed cap- sules that showed natural dehiscence represented between 0 and 23.8% of each sample. Within the four samples, seed capsules that showed no signs of cotyledon predation by mammals (unopened or natural dehiscence) varied from 0 to 57.1%. In Table 6-2 we present the length and width measurements of each sample; samples are also segregated by class of endocarp breakage. No sta- tistically significant difference was found within or between samples in the size of seed capsules eaten by aye-ayes and rodents, nor did the verte- brate-opened endocarps differ from the general size distribution of random samples. Thus, on the basis of the present data, we conclude that the local utilization of Canarium seeds is highly vari- able within this forest, although rodents appear to be the most common consumers of the cotyle- dons. Seed Capsules from Rodent Burrows In Table 6-3 we present information on seed capsules recovered from rodent burrows. At 810 86 FIELDIANA: ZOOLOGY Table 6-3. Measurements (mm) of Canarium mad- agascariense seed capsules collected from three rodent burrows in the RNI d'Andringitra. Measurements'" Species and elevation Length Width Gymnuromys roberti 38.6 ± 5.9 16.1 ± 1.7 (810 m) 23.1-49.2 13.3-19.9 (N = 21) (N = 21) Nesomys rufus 47.4 ± 2.3 19.0 ± 1.3 (810 m) 44.3-49.8 17.6-20.5 (N = 4) (N = 4) Nesomys rufus 48.3 ± 2.8 19.9 ± 2.2 (810 m) 43.1-50.4 16.8-21.0 (N = 22) (N = 22) * Presented as mean ± standard deviation, minimum- maximum measurements, and sample size in parenthe- ses. m elevation a burrow system of Gymnuromys ro- berti contained 22 Canarium seed capsules; 21 of these had been gnawed open and the cotyledons consumed, whereas one was intact (Table 6-2). There was no significant difference in the size of the seed capsules cached by Gymnuromys and those in the random 810 m collections. A burrow of Nesomys rufus at the same elevation and about 120 m from that of the Gymnuromys contained four seed capsules of Canarium, all gnawed open; eight fruits of Cryptocarya (Lauraceae), six whole and two gnawed open; and one whole unidentified seed. The contents of a Nesomys rufus burrow at 820 m elevation and about 500 m ground distance from the other excavated Nesomys burrow con- tained 22 Canarium seed capsules, 15 of which had been gnawed open. The pooled sample of seed capsules recovered from Nesomys burrows (N = 26) was longer than those in the random 810 m sample (/ = 3.21, P = 0.002); there was no difference in width measurements. Thus, it would appear that Nesomys is preferentially se- lecting and caching the larger seed capsules it en- counters on the forest floor. An excavated Eliurus webbi ground burrow at 720 m contained 20 fruits of Cryptocarya, about half of which had been chewed open (Goodman, 1994). General Consumption of Canarium Seed Capsules In general, little information is available about the food habits of endemic Malagasy rodents. Ne- somys rufus, the only diurnal species of rodent within the RNI d'Andringitra, was observed to eat fallen seed capsules of Canarium, Artabotrys ma- bifolius (family Annonaceae), and Sloanea rho- danta (family Elaeocarpaceae). After a major storm passed through Vatoharanana, Pare Nation- al (PN) de Ranomafana, a large hollow tree was found toppled that contained a cavity within the main trunk packed with Canarium seed capsules, many of which had been gnawed open along the central axis; none of the intact seeds was sprout- ing (C. Hemingway, pers. comm.). On the basis of a description of the cavity, it probably belonged to Brachytarsomys albicauda, a species known to occur within this park (Nicoll & Langrand, 1989). In summary, although information in the literature on Nesomyinae food habits is meager, it is clear on the basis of our feeding experiments that nu- merous rodent species are able to open and feed on the seeds of Canarium. In a prior study on Nosy Mangabe, 204 Can- arium seeds that had been fed upon by aye-ayes were opened to determine the average number of locules per seed capsule and whether viable cot- yledons remain after an aye-aye has fed on the seed capsules (Sterling, unpubl. data). Thirty- eight percent of the fruits bore one locule, 52% bore two locules, and 10% bore three locules. With only one exception, aye-ayes had opened all the locules large enough for viable cotyledons and extracted the contents. In one seed a viable cot- yledon remained untouched. Aye-ayes only at- tempted to open aborted locules twice. The results of recent lemur feeding ecology studies have shown that several species consume the soft pericarp of Canarium fruits. Simons Mor- land (1991) found that Varecia variegata varie- gata on Nosy Mangabe consumes whole fruits of Canarium and probably passes whole seed cap- sules. She found considerable temporal variation in the utilization of these fruits. In November 1987, Canarium fruit was the most common food resource, accounting for 34% of the diet by time spent feeding, whereas in December 1988, Can- arium fruit represented only 3% of the diet. On the Masoala Peninsula, Rigamonti (1993) noted that about 1 .5% of the feeding time of V. v. rubra was on Canarium fruits. Overdorff (1991) reported that Canarium fruits form a significant portion of the diet of Eulemur fulvus rufus and E. rubriventer in the PN de Rano- mafana, particularly between the months of Oc- tober and January, although there were consider- able differences between these species. Andrian- isa (1989) also noted that E. f albifrons on Nosy GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 87 Mangabe feeds on the fruits of Canarium. In ad- dition, bush pigs (Potamochoerus larvatus), a spe- cies probably introduced to Madagascar (Grubb, 1993), frequently ingested fallen Canarium seeds on Nosy Mangabe (Sterling, unpubl.). It appears that Eulemur and Varecia are seed dispersers, whereas aye-ayes, rodents, and bush pigs are seed predators. Experimental tests of seed scarring did not increase germination rates of wild-collected Canarium seed capsules (Tsiza, 1989). Unopened seed capsules cached in rodent burrows would also have the possibility of ger- minating, particularly because a significant por- tion of the seeds remain viable for up to 3 years in storage (Tsiza, 1989). There is evidence of competitive interactions within and between lemur species over the con- sumption of Canarium fruits and seeds. Simons Morland (1991) found that dominant female Va- recia displaced males and other females from food resources, including Canarium. Further, ag- gressive interactions between Varecia and Dau- bentonia have been observed on Nosy Mangabe when both species were feeding at night on fruits in the same Canarium tree (Sterling, 1993). Phenology of Canarium Fruiting Several studies have monitored the fruiting and flowering patterns of Canarium in eastern rain forests of Madagascar (Andrianisa, 1989; Over- dorff, 1991; Sterling, 1993). Some trees have mul- tiple crops per year, while others have single crops. Canarium trees at Ranomafana and on Nosy Mangabe were found to fruit asynchronous- ly. On Nosy Mangabe, fruit was found in all months of the year, although fewer fruits were available during the cold, wet season (June to Au- gust) than the rest of the year (Andrianisa, 1989; Sterling, 1993). and at least two species cached them in their bur- rows. On Nosy Mangabe, aye-ayes open seed cap- sules still in trees as well as those that have fallen to the ground. Thus, aye-ayes compete with other species of arboreal lemurs and potentially arboreal rodents when the fruits are still on the tree, and with terrestrial rodents and bush pigs when the fruits and seed capsules are on the ground. On the basis of current information, it is clear that there is overlap in the diet of the aye-aye and rodents on the eastern slopes of the RNI d'An- dringitra. There seems to be no clear difference in the sizes of Canarium seed capsules that they each exploit, however, although seed capsules cached in Nesomys burrows are larger than those found in random samples under trees. During cer- tain periods of the year, this resource is limited (Andrianisa, 1989; Sterling, 1993), and these an- imals may compete for the nutritious seeds. Competition Between Native Animals and Rattus rattus One of the surprising results of the small mam- mal survey of the RNI d'Andringitra was that a rodent introduced to Madagascar, Rattus rattus, has colonized areas of undisturbed forest (Chapter 22). At other sites on the island, Rattus rattus ap- pears to be replacing native rodents (Goodman, 1995). The finding that Rattus rattus readily fed on Canarium seeds, a resource that forms a sig- nificant proportion of the aye-aye's diet (Sterling, 1994) and may also be important for several spe- cies of endemic rodents, suggests that the spread of Rattus rattus across the island may be having an effect on populations of aye-ayes and Neso- myinae rodents. This point is in need of imme- diate further research, because its implications are potentially catastrophic for endemic animals that have been the focus of so much conservation ef- fort on Madagascar. Competition Between Aye-ayes and Native Rodents? At the 720 and 810 m sites, Canarium mada- gascariense was a common emergent tree (Chap- ter 4). We found few Canarium seed capsules eat- en by aye-ayes; this may reflect the relative rare- ness of the primate in this forest, or perhaps we simply did not find a tree heavily visited by this animal. On the other hand, several species of en- demic rodents feed widely on these seed capsules, Acknowledgments We are grateful to Pete Philippson and Beverly Lewis for identifying various seed remains. Jodi Sedlock drew the figure. Literature Cited Andrianisa, J. A. 1989. Observations phenologiques dans la Reserve Sp6ciale de Nosy Mangabe, Maroant- 88 FIELDIANA: ZOOLOGY setra. M6moire de fin d' Etudes, Ecole Sup6rieure des Sciences Agronomiques, University d* Antananarivo. Carleton, M. D., and D. F. Schmidt. 1990. Systematic studies of Madagascar's endemic rodents (Muroidea: Nesomyinae): An annotated gazetteer of collecting lo- calities of known forms. American Museum Novita- tes, no. 2987: 1-36. Duckworth, J. W. 1993. Feeding damage left in bam- boos, probably by aye-ayes (Daubentonia madagas- cariensis). International Journal of Primatology, 14: 927-931. Goodman, S. M. 1994. A description of the ground burrow of Eliurus webbi (Nesomyinae) and case of cohabitation with an endemic bird (Brachypteraciidae, Brachypteracias). Mammalia, 58: 670-672. . 1995. Rutins on Madagascar and the dilemma of protecting the endemic rodent fauna. Conservation Biology, 9: 450-453. Grubb, P. 1993. Order Artiodactyla, pp. 377-414. In Wilson, D. E., and D. M. Reeder, eds.. Mammal spe- cies of the world: A taxonomic and geographic ref- erence, 2nd ed. Smithsonian Institution Press, Wash- ington, D.C. Iwano, T, and C. Iwakawa. 1988. Feeding behaviour of the aye-aye (Daubentonia madagascariensis) on nuts of ramy (Canarium madagascariensis). Folia Pri- matologica, 50: 136-142. Leenhouts, P. W 1959. Revision of the Burseraceae of the Malaysian area in a wider sense. Xa. Canarium (Stickm.). Blumea, 9: 275-475. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la conservation et des aires prot£g6es. World Wide Fund for Nature, Gland, Switzerland, xvii + 374 pp. Overdorff, D. J. 1991. Ecological correlates to social structure in two prosimian primates: Eulemur fulvus rufus and Eulemur rubriventer in Madagascar. Ph.D. thesis, Duke University, Durham, North Carolina. Perrier de la Bathie, H. 1946. Burs£rac6es (Bursera- ceae). Flore de Madagascar, Famille 106, pp. 1-50. Rigamonti, M. M. 1993. Home range and diet in Red Ruffed Lemurs (Varecia variegata rubra) on the Ma- soala Peninsula, Madagascar, pp. 25-39. In Kappeler, P. M., and J. U. Ganzhorn. eds.. Lemur social systems and their ecological basis. Plenum, New York. Simons Morland, H. 1991. Social organization and ecology of Black and White Ruffed Lemurs (Varecia variegata variegata) in lowland rain forest. Nosy Mangabe, Madagascar. Ph.D. thesis, Yale University, New Haven, Connecticut. Sterling, E. J. 1993. Behavioral ecology of the aye- aye (Daubentonia madagascariensis) on Nosy Man- gabe. Ph.D. thesis, Yale University, New Haven, Con- necticut. 1994. Aye-ayes: Specialists on structurally de- fended resources. Folia Primatologica, 62: 142-154. E. S. Dierenfeld, C. J. Ashbourne, and A. T. C. Feistner. 1994. Dietary intake, food consumption and nutrient intake in wild and captive populations of Daubentonia madagascariensis. Folia Primatologica, 62: 115-124. Tsiza, G. 1989. Essai de monographic sylvicole du ramy (Canarium madagascariensis) avec reference sp^ciale a la c6te est. M6moire de Fin d' Etudes, Ecole Sup6rieure des Sciences Agronomiques, University d' Antananarivo. White, F. 1983. The vegetation of Africa. UNESCO, Paris, 356 pp. GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 89 Chapter 7 Millipedes (Diplopoda) from the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar Henrik Enghoff Abstract Eight species of millipedes (Diplopoda) were recorded in the Reserve Naturelle Integrate d'Andringitra. Two other species had been previously reported from the Andringitra Massif: Betshewna andringitrae Mauries, 1994, and B. orbatum Maudes, 1994. Resume Huit especes de myriapodes (Diplopoda) ont ete recensees dans la Reserve Naturelle Inte- grate d'Andringitra. Deux autres especes, Betcheuma andringitrae Mauries, 1994, etB. orbatum Mauries, 1994, venant du Massif d'Andringitra y avaient ete rdpertoriees precedemment. Introduction About 150 species of millipedes (Diplopoda) are known from Madagascar, and almost all are endemic. There are several endemic genera and two endemic higher taxa: Zoosphaeriini of the family Sphaerotheriidae (giant pill millipedes) and Spiromiminae of the family Pachybolidae (a group of spirobolidan millipedes). The giant pill millipedes (genus Globotherium, etc.) and the large cylindrical species of Sechelleptus (Spiros- treptidae) and Aphistogoniulus (Pachybolidae) are among the most conspicuous invertebrates of the island. The foundation of knowledge of Malagasy mil- lipedes was laid by the impressive monograph of Saussure and Zehntner (1902). Since then, little progress has been made. In addition to scattered records and remarks by various authors, there are three papers that have added significantly to our knowledge. Jeekel (1971) reviewed the endemic genus Aphistogoniulus (nine named species). In a sub- sequent paper, Jeekel (1974) studied relationships within the family Sphaerotheriidae and found the closest relative of the Malagasy forms (tribe Zoosphaeriini) to be the Arthrosphaerini of India and Sri Lanka. Recently, Mauries (1994) described the genus Betscheuma, which includes 13 species of tiny leaf litter and soil millipedes. The species, which were all described as new by Mauries (1994), oc- cur in mountain habitats (forests and "prairies"). Two of the new species were described from the Reserve Naturelle Integrate (RNI) d'Andringitra based on collections made in 1970-1971 during the Recherche Cooperative sur Programme (RCP) No. 225, mission to the Andringitra Massif (Pau- lian et al., 1971). Betscheuma andringitrae: Andringitra, Anjavi- dilava, "foret dense humide de montagne, 1800- 1950 m," "foret dense sclerophylle de montagne a Philippia, 2000-2050 m." Andringitra, Maros- itry, "foret dense sclerophylle de montagne, 2000 m." Betscheuma orbatum: Andringitra, "foret dense 90 FIELDIANA: ZOOLOGY Table 7-1. Elevational distribution of millipedes on the eastern slopes of the RNI d'Andringitra. Ia\ a 720 m 810 m 1210 m 1625 m Sphaerotheriidae undetermined Sechelleptus fulgens Sechelleptus metazonalis Sechelleptus sp. (nov.?) Spirostreptidae genus? sp.? Spirostreptidae genus? sp.? Aphis togon iulus aff. corallipes Pachybolidae genus? sp.? 1M, 2F 1M, IF 3 juv. 1M, 1 juv. 5M, 4F 1 juv. 2M, 2F 3M, IF 1 juv.(?) 5M, IF 2M, 2F 1 juv. 1M, 5F IF 4M, 3F M, male(s); F, female(s); juv., juvenile. sclerophylle de montagne a Philippia, 2000 m," "foret dense humide de montagne, 1950 m." An- dringitra, Pic Bory, zone sommitale, "fourre a bambous, 2550 m." Andringitra, Anjavidilava, "foret dense scleYophylle de montagne a Philip- pia, 2075 m." Andringitra, Andrianony, "foret dense humide de montagne, 1650 m." To the best of my knowledge, no other millipede has been recorded from the Andringitra Massif. Methods During the 1993 survey of the RNI d'Andrin- gitra, members of the expedition made general collections of millipedes. The majority of speci- mens were collected during the day on the ground or on vegetation; they are heavily biased toward large and colorful species. Descriptions of the vegetational communities associated with each al- titudinal zone are given in Chapter 2. Specimens were preserved in 70% alcohol. Four of the five unnamed species of Sphaerotheriidae, Spirostrep- tidae, and Pachybolidae are represented by adult males and are probably all undescribed. For three of them, the current state of the taxonomy of these families precludes even generic assignments. The collections are housed in the Field Museum of Natural History, with duplicates in the Zoological Museum, Copenhagen. Results The presently studied material, obtained in late 1993 from the RNI d'Andringitra, comprises eight species (Table 7-1): Family Sphaerotheriidae Genus? sp.? Family Spirostreptidae Sechelleptus fulgens (Saussure & Zehntner, 1901). Previously recorded (as Spirostreptus ful- gens) from Mandraka by Saussure and Zehntner (1901) and from Tolagnaro (Fort Dauphin) by Saussure and Zehntner (1902). Sechelleptus metazonalis (Saussure and Zehnt- ner, 1902). Previously recorded (as Spirostreptus metazonalis) from Col de Sakavalana by Saussure and Zehntner (1902). Sechelleptus sp. (nov.?) Genus? sp.? 1 Genus? sp.? 2 Family Pachybolidae Aphistogoniulus aff. corallipes (Saussure and Zehntner, 1902). Aphistogoniulus corallipes was recorded from Tolagnaro by Saussure and Zehnt- ner ( 1 902, as Spirobolus corallipes) and from Be- voalavo by Demange (1969, as Mystalides cor- allipes). Genus? sp.? ENGHOFF: MILLIPEDES FROM THE EASTERN SLOPES 91 With the negligible background information avail- able, no biogeographical interpretation of these results is possible. Acknowledgments I am grateful to Steve Goodman for letting me examine this material and for help with the manu- script. An anonymous reviewer provided impor- tant comments on an earlier version of this report. Literature Cited Demange, J.-M. 1969. Myriapodes recoltes a Madagas- car par M. L. Bigot. Bulletin Museum National d'Histoire Naturelle, Paris, 2nd ser, 41(2): 484-489. Jeekel, C. A. W. 1971. Notes on the genus Aphisto- goniulus Silvestri (Diplopoda, Spirobolida, Trigoniu- lidae). Bulletin Zoological Museum, University of Amsterdam, 2: 33-42. 1974. The group taxonomy and geography of the Sphaerotheriida (Diplopoda). Symposium Zoolog- ical Society of London, 32: 41-52. Mauries, J. -P. 1994. Decouverte de Diplopodes Cras- pedosomides a Madagascar: Betscheuma n.g. de la famille gondwanienne des Pygmaeosomatidae Carl, 1941 (Myriapoda, Diplopoda). Bulletin Museum Na- tional d'Histoire Naturelle, Paris, 4th seY, sect. A, 16(1): 55-86. Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, and P. Griveaud. 1971. RCP 225. Etudes des eco- systemes montagnards dans le region malgache. I. Le massif de l'Andringitra. 1970-1971. Geomorphologie, climatologie et groupements vegetaux. Bulletin Socie- te d'Ecologie, 11(2-3): 198-226. Saussure, H., and Zehntner, L. 1901. Myriopoden aus Madagaskar und Zanzibar gesammelt von Dr. A. Voeltzkow. Abhandlungen Senckenbergische Natur- forschende Gesellschaft, 26: 425-464. 1902. Myriapodes de Madagas- car. Hist. phys. nat. politique Madagascar publiee par A. Grandidier, 53: 1-349, 15 pi. 92 FIELDIANA: ZOOLOGY Chapter 8 Ant Diversity Patterns Along an Elevational Gradient in the Reserve Naturelle Integrate d'Andringitra, Madagascar Brian L. Fisher Abstract Inventory methods designed to permit rapid, replicable, and quantitative sampling of the leaf litter ant fauna were tested in wet tropical forest in eastern Madagascar in the Reserve Naturelle Integrate d'Andringitra. Methods involved a combination of pitfall and leaf litter sampling along a 250 m transect. Surveys were conducted at four sites, located at 785, 825, 1275, and 1680 m. From pitfall and leaf litter samples, I collected and identified 27,866 ants belonging to 114 species and 28 genera; general collecting yielded an additional 34 species. The first- order jackknife produced an estimate of a total of 132 species for the four sites collectively. Species accumulation curves demonstrate the efficacy of these techniques in sampling the majority of the ants in the leaf litter. The species collected and their relative abundance are presented. The composition of the fauna is compared with that of other tropical forest sites. Species diversity decreased with elevation. Species turnover and faunal similarity measures showed a division in ant communities between the lowest elevation sites (785 and 825 m) and the highest elevation sites (1275 and 1680 m) that corresponded to the cloud forest transition. Resume Dans la foret tropicale humide de la Reserve Naturelle Integrate d'Andringitra a Test de Madagascar, des techniques d'inventaire, concues pour permettre un dchantillonnage quantitatif rapide et r6p6titif des fourmis vivant dans la litiere de feruilles mortes ont 6te testees. Les techniques combinent l'utilisation de pieges "pitfall" et l'6chantillonnage de la litiere de fer- uilles mortes le long d'un transect de 250 m. Des inventaires ont 6te effecUtes dans quatre sites, localises respectivement a 785, 825, 1275 et 1680 m d'altitude. A partir des pieges "pitfall" et des echantillonnages effectu^s a partir de la litiere de feuilles mortes, 27866 fourmis appartenant a 1 14 especes et 28 genres ont 6te recueillis et identifies; une collecte ateatoire a permis de recolter 34 especes supptementaires. Le "first-order jackknife" a produit un total d'environ 132 especes pour quatre sites. Les courbes d'accumulation des especes ont d^montre" l'efficacite de ces techniques qui permettent de r6pertorier la majorite des fourmis dans la litiere de feuilles mortes. Les especes collectees et leur abondance relative sont presentees. La composition de la faune de fourmis est comparee a celle d'autres sites forestiers tropicaux. La diversite des especes diminue a mesure que 1'on monte en altitude. Le remplacement des especes et la similarite de la composition de la faune de fourmis montrent une division des communautes de fourmis entre les deux altitudes les moins eievees (785 et 825 m) et entre les deux altitudes les plus elevees (1275 et 1680 m). Ces derniers correspondent a la transition vers la foret d'altitude caracterisee par la presence de n£bulosite frequente. FISHER: ANT DIVERSITY PATTERNS 93 Introduction Insects constitute 85% of the world's animal biodiversity (Groombridge, 1992) and deserve increased attention in regions of the world, such as Madagascar, where species-rich habitats are under threat. Conservation methods that priori- tize areas based on diversity patterns (species richness and local endemism) of birds and mam- mals and do not include insect diversity overlook most organisms and thus do not guarantee pres- ervation of the greatest diversity. Few studies have investigated the correlation between inver- tebrate and vertebrate diversity. Pearson and Cassola (1992) showed that for gridded squares, 275 km on a side, across North America, the In- dian subcontinent, and Australia, species rich- ness levels of tiger beetles, birds, and butterflies were positively correlated. On a finer geograph- ical scale, however, there is evidence that verte- brate diversity is not an accurate indicator of in- vertebrate diversity. Prendergast et al. (1993) showed low overlap between highly diverse sites of butterflies, dragonflies, liverworts, aquatic plants, and breeding birds in Britain. At 32 sites in southeastern Australia, Yen (1987) found no correlation between the number of species of vertebrates and beetles. Scharff (1992) conclud- ed that sites in tropical rain forests of eastern Africa where linyphiid spiders showed the high- est diversity were not necessarily the same as those documented for birds, mammals, amphib- ians, and reptiles. Thus, arthropod patterns of di- versity cannot be assumed to be correlated with those of vertebrates, and methods are needed to quantify species richness and endemism in insect taxa. In this chapter, I propose and evaluate inven- tory methodologies to survey leaf litter ant di- versity along an elevational gradient. The ele- vational transect specifically addresses the fol- lowing question: To what degree does elevation influence diversity and abundance of ants? I ad- dress this question by comparing species rich- ness estimates, measures of faunal similarity, and species turnover (beta diversity) for ant species across four sampled elevations in the Reserve Naturelle Integrate (RNI) d'Andringi- tra. Future work will compare these patterns with those exhibited by a sympatric assembly of vertebrates. Study Organism and Methods Importance of Ants Madagascar's ant fauna is remarkable for its di- versity, endemism, ancient affinities, and ecolog- ical dominance. Because of its long isolation from other land masses, the level of endemism of ant species on Madagascar is high. In the only re- gional study of ants conducted on the island, D. M. Olson and P. S. Ward (pers. comm.) found 92% of the ants at a dry forest site in western Madagascar to be endemic to the island. More- over, the ant fauna of Madagascar is one of the least understood. For example, two of the better known ant genera in the world still have over 50% of their species undescribed in Madagascar (Te- traponera: 24 undescribed out of 36 known, P. S. Ward, pers. comm.; Tetramorium: 33 undescribed out of 62 known, B. Bolton, pers. comm.). There are no valid species described from the region of RNI d'Andringitra. There is one form, Campono- tus hova becki var. altior Santschi, recorded from near the RNI d'Andringitra (Santschi, 1923; Per- rier de la Bathie, 1927), but this name is unavail- able taxonomically. The taxonomy of Campono- tus is in chaos, and the identity of this specimen cannot be understood until the complex of forms related to Camponotus hova have been thoroughly revised taxonomically. Assessment of Ant Diversity A primary objective of this project was to de- velop standardized methods for assessing terres- trial ant diversity that can be used in other tropical wet forests. The proposed method does not sam- ple the entire ant community (arboreal, subterra- nean, and terrestrial), but only those ant species that nest or forage in the forest leaf litter layer. Subterranean and arboreal ant faunas pose specific technical problems in developing standardized survey methods. Leaf litter lends itself to replic- able sampling with pitfall traps and leaf litter sift- ing. These two techniques are effective for col- lecting many ant species from a wide variety of habitats, and they provide data on abundance and diversity suitable for rigorous statistical analysis (Olson, 1991). Because we do not yet understand the relationship between leaf litter ant diversity patterns and the total ant fauna in any wet tropical forest, results from the following survey tech- nique may not be representative or indicative of 94 FIELDIANA: ZOOLOGY Fig. 8-1. Mini-Winkler sacks used for extracting leaf litter ants. At each elevation there were 26 mini- Winkler sacks, which were suspended under a tarp. See text for explanation of sampling methods. patterns of rarity, endemism, or species richness of arboreal or subterranean ant communities. In the RNI d'Andringitra, intensive ant surveys were conducted at four sites located at 785, 825, 1275, and 1680 m. These fell within 75 m in el- evation of the project transects centered at 720, 810, 1210, and 1625 m. At each elevation the sur- vey method employed 50 pitfalls and 50 leaf litter samples in parallel lines 10 m apart along a 250 m transect. The site for the transect was chosen with the intent of sampling representative micro- habitats at each elevation. Pitfall traps were placed and the leaf litter samples gathered every 5 m. Pitfall traps consisted of test tubes, 18 mm internal diameter by 150 mm long, that were part- ly filled with soapy water and a 5% ethylene gly- col solution, inserted into polyvinyl chloride (PVC) sleeves, and buried with the rim flush with the soil surface. Traps were left in place for 4 days. I extracted arthropods from samples of leaf lit- ter using a modified form of the Winkler extractor (Besuchet et al., 1987; Ward, 1987; Olson, 1991). The leaf litter samples involved establishing two 1 -m2 plots on each side of the transect line, sep- arated by 1 m. The leaf litter inside each 1-m2 plot was collected and sifted through a sieve of 1 cm grid size. Before sifting, the leaf litter material was minced using a machete to disturb ant nests in small twigs and decayed logs. Approximately 2 liters of sifted litter were taken from the two 1-m2 plots. Ants were extracted from the sifted litter during a 48-hour period in modified Winkler sacks ("mini-Winkler"; Fig. 8-1). The standard Winkler sacks can hold up to four 2-liter samples of sifted litter; mini-Winkler sacks are designed FISHER: ANT DIVERSITY PATTERNS 95 to hold only one 2-liter sample. Each sifted litter sample was held in a mesh sack that was sus- pended in a larger cotton enclosure. The ants dropped out of the mesh sack and were collected in plastic bags ("Whirl Packs") containing 70% ethanol at the bottom of the mini-Winkler sack. At each elevational zone, a tarp was used to pro- tect the 25 mini-Winkler sacks from rain. Because insects are less effectively extracted from wet leaf litter, sifting was done at least 24 hours after sig- nificant rainfall. I also surveyed ants through general collecting, defined as any collection that is separate from the mini-Winkler sack or pitfall transects, including searching in rotten logs and stumps, in dead and live branches, in bamboo, on low vegetation, un- der canopy moss and epiphytes, under stones, and leaf litter sifting (four leaf litter samples were taken between 1800 and 2000 m elevation). At each transect site, general collections were con- ducted for an approximately 2-day period. These collections included samples of the arboreal ants found on low vegetation that were not sampled by pitfalls or leaf litter. Ants sampled using general collection methods, therefore, were not used in the analysis of the efficacy of the survey of the leaf litter ants, of faunal similarity, or beta diver- sity. Identification Specimens were identified to morphospecies based on characters previously established to be important at the species level for each genus. When possible, names were attached to these morphospecies by using taxonomic descriptions (Bolton, 1979) and by comparing specimens to material previously collected by Ward, Olson, and Fisher in Madagascar that was compared to type material. Specimens will be deposited at the Mu- seum of Comparative Zoology, Harvard Univer- sity, and a representative set will be returned to Madagascar. Data Analysis Evaluation of Sampling Method — Analyses of ant diversity patterns require an understanding of the extent to which the inventory technique samples the leaf litter ant community in each tran- sect zone. To assess the completeness of the sur- vey for the elevation sampled, I plotted cumula- tive species per sample curves for each elevation. Species accumulation is plotted as a function of the number of leaf litter and pitfall traps samples taken. For the analysis, each leaf litter sample was paired with the adjacent pitfall sample, collective- ly termed a station sample. Species accumulation curves for the 50 stations per transect and first- order jackknife estimates of total number of spe- cies in the local community from which the sam- ples were taken are plotted for each succeeding station. The first-order jackknife method is a non- parametric approach to improving the estimate of species richness and is based on the observed fre- quency of unique species (the jackknife estimator and standard deviation are defined in Heltshe & Forrester, 1983). For both species accumulation curves, sample order was randomized 100 times, and the means and standard deviation of the jack- knife estimates were computed for each succeed- ing station using the program Estimates (R. K. Colwell, unpubl.; see also Colwell & Coddington, 1994). If the species accumulation or jackknife estimate curves appear to level off before 50 sta- tions, then the transect is arguably sufficient. Con- versely, if the curves do not flatten out, a longer sampling transect may be necessary to accurately compare diversity between elevations. Ant Diversity — Data on both species richness and abundance were used to assess the change in species composition along the elevational gradi- ent. Only records of ant workers were used in these calculations. Because alates may travel con- siderable distances during dispersal, their pres- ence does not necessarily signify the establish- ment of a colony of that species within the tran- sect zone. In addition, collections of queens and males dispersing from nearby nests at the time of the survey may not reflect relative abundance of the species. Because ants are colonial, abundance measures were based not on the total number of individual workers collected at each transect site but on species frequency (proportion of stations, out of 50, in which each species was collected at a site). For each elevation, I compared first-order jack- knife estimates of total species richness and 95% confidence limits. Similarity of the ant fauna of the different elevations was assessed using two different measures: (1) the Jaccard Index, based on presence/absence data only: Q = j/(a + b — j), where j = number of species found at both elevations, a = number of species at elevation A, and b = number of species at elevation B (Ma- 96 FIELDIANA: ZOOLOGY gurran, 1988); and (2) the simplified Morisita In- dex, which incorporates abundance data: ('UU — 2 1 (an, X bn,) (da + db)aN X bN where larif Ibn} da = — — — and db = aN2 bN2 where aN = total number of station/species oc- currences at elevation A, bN = total number of station/species occurrences at elevation B, an, = the number of stations occupied by the /th species at elevation A, and bn, = the number of stations occupied by the /th species at elevation B (Horn, 1966; Wolda, 1981). Beta diversity (species turnover between ele- vations) was calculated in two ways. First, the beta diversity measure of Whittaker (1960) was used: beta-1 = (S/a) - 1, where S = the total number of species in the two elevations com- bined, and a = the mean number of species in each elevation. Because this measure does not dis- tinguish between species turnover and the loss of species along a gradient without adding new spe- cies, the measure of beta diversity developed by Harrison et al. (1992) was also calculated: beta-2 = (S/a^) - 1, where S is the same as beta-1 above and a^ is the maximum value of alpha diversity among the elevations compared. Finally, the proportions of species unique to an elevation were also compared. Results I collected and identified 30,708 ants, compris- ing 148 species and 28 genera, from general col- lections, leaf litter, and pitfall methods. These in- clude 705 queens and 441 males that were not used in any of the following analyses. Leaf litter and pitfall methods produced 27,866 worker ants belonging to 1 14 species. A list of ant species in the RNI d'Andringitra, based on all collecting techniques, and separated by elevation and tech- nique, is presented in Table 8-1. General collec- tions within each ±75 m transect zone are pre- sented. The 785 and 825 m general collections were separated by 3 km and did not overlap spa- tially. The total number of species was greatest in the lowest elevations sampled (89 at 785 m and 81 at 825 m). These two zones also contained the great- est number of unique species, but the highest per- centage of species unique to an elevation was found at 1800-2000 m (50%). If the data for 785 and 825 m are combined, 70% of the species (74 of 105 spp.) are unique to the lowest elevations. The number of species and number of individuals collected from pitfall traps was low compared to mini-Winkler sack and general collection meth- ods. For example, at 785 m, 8,582 workers in 76 species were collected from leaf litter samples, whereas only 218 workers in 19 species were col- lected from pitfall traps. Based on the number of ants extracted from the leaf litter samples over a 48-hour period using mini-Winklers sacks the av- erage density of worker ants in the leaf litter layer was 67 per m2. The relative prevalence of the different subfam- ilies in the combined pitfall and leaf litter samples is shown in Figure 8-2. The fauna is dominated by Myrmicinae in both number of species and number of individuals, followed by Ponerinae. The abundance of ant species is presented in Table 8-2. Both the proportion of stations at which each species was collected and the number of in- dividuals collected are presented. Four out of 1 14 species were present at every elevation, but the abundance of individual species often differed considerably from one site to the next. For ex- ample, Hypoponera sp. 1 had a low abundance at 785 and 825 m (0.08 in both), while at 1275 and 1680 m its abundance was much higher (0.88 and 0.46, respectively). Species-accumulation curves and first-order jackknife estimates of species richness with their standard deviations are presented for each transect zone (Fig. 8-3a-d, Table 8-3). Jackknife and ob- served species accumulation curves leveled off, indicating that within the area of the survey, the techniques employed collected the majority of the ants foraging or living in the leaf litter. In a com- bined analysis of all elevations, the pitfall and mini-Winkler sack methods collected 86% of the total number of leaf litter ant species estimated by first-order jackknife technique to occur in the re- gion (Fig. 8-3e). Faunal similarity values (Table 8-4) based on presence/absence data (Jaccard's Index) and abun- dance (simplified Morisita Index) were lowest be- tween 1680 m and the two lowest elevations (785 and 825 m). The highest value of similarity was between the 785 and 825 m sites. The 1275 m transects had a greater faunal similarity with the FISHER: ANT DIVERSITY PATTERNS 97 Table 8-1. Ant species list for the RNI d'Andringitra, including altitude and collection method. 1800-2000 Genus Species 785 m 825 m 1275 m 1680 m m CERAPACHYINAE Cerapachys 1 2 3 W, G W, G G W W, G G 5 W, G W W 8 W, G W, G W, G 9 G Simopone 1 G DOLICHODERINAE Technomyrmex mayri W FORMICINAE CAMPONOTINI Camponotus 1 2 3 4 5 10 11 13 14 15 16 17 G G G P, G G W W, G W, G P G G P, W G G G hildebrandti G G G LAsmsn Paratrechina 1 2 3 W, P, G G G W, G W, P, G 4 W, G W, G W, G W 5 W, P, G W, P, G 6 W PLAGIOLEPIDINI Plagiolepis 1 3 W, G W, P, G G MYRMICINAE CREMATOGASTRINI Crematogaster 1 2 3 lobata W G G G schenki W, G P P, W, G G DACETONINI Glamyromyrmex 1 W, G W Kyidris 1 W, G W Smithistruma 2 3 5 W, G G G G Strumigenys 1 2 3 4 5 6 7 W, G W, G G W W, G W, G W, G W W 8 W, P,G W W, G 9 W W W, G 10 W, G 11 W 12 W W 13 W W 14 W, P,G W, G 98 FIELDIANA: ZOOLOGY Table 8-1. Continued. Genus Species 785 m 825 m 1275 m 1800-2000 1680 m m FORMICOXENINI Leptothorax PHEIDOLOGETONINI Oligomyrmex PHEIDOLINI Pheidole SOLENOPSIDINI Monomorium TETRAMORIINI Tetramorium 15 W 16 w 17 W 18 W, G W, G 1 G G 1 W, G W, G 2 W, P W, G 1 2 W, P, G 3 6 W, G W, P,G 7 W W W 8 W, P, G 9 W, PG W.PG 10 W, P 11 W, P W, P, G 12 w W, P 13 w G 14 w W, P,G 15 w W 16 w W 17 W, P W, PG 18 G 19 20 w 22 w nemoralis W, P, G W, PG veteratrix W, P, G W, PG P longispinosa W, P,G W, PG 1 W, PG W, PG 2 W W 3 W, PG W, G 4 G 5 W W, G 6 W W, G 7 W, G W 8 W W, G 9 W W 10 W W, PG 12 P 13 14 W, G 1 W W, PG 2 G 3 W W, G 4 P W 5 G G W.PG 7 W, G W W, G 9 W 10 W, P W, P,G W, PG W, PG 11 W, P W, P 13 P 14 W, P G W, P, G 15 W 16 W W 17 G FISHER: ANT DIVERSITY PATTERNS 99 Table 8-1. Continued. 1800-2000 Genus Species 785 m 825 m 1275 m 1680 m m andrei W, P W cognatum W W, G W, G dysalum W, P, G W, P,G electrum W, P W, P marginatum W, P W, P PONERINAE AMBLYOPONINI Amblyopone 2 3 W W Mystrium 1 2 W G W, G G Prionopelta 1 3 W, G W, G W, G W ECTATOMMINI Discothyrea 1 W W, P G Proceratium 1 W PLATYTHYREINI Platythyrea bicuspis G PONERINI Anochetus grandidieri W, G W, G Hypoponera 1 W W W, G W, G G 2 W, G W W 3 G G 6 W W 7 W W W, G G 8 W 9 W W, G W, G G 10 W, G W, G G 11 W, G W W, G W, G 12 W W 13 W W 14 W, G G 15 W 17 G sakalava W, G W, G G Leptogenys 1 3 P W Odontomachus coquereli W, G W, G Pachycondyla cambouei sikorae W, P,G G W, P G PSEUDOMYRMEC1 Tetraponera exasciata grandidieri psw-S/ psw-92 G G W, G W, G G G Total species: G 47 42 36 24 20 Total species P 19 19 14 8 Total species: W 76 64 35 23 Total species: All methods 89 81 51 31 20 Number (%) of unique species 21 (24%) 11(14%) 9(18%) 3(10%) 10(50%) Total number of G collections 55 41 24 36 16 Number of workers: G 479 306 382 256 273 Number of workers P 218 255 238 127 Number of workers: W 8,582 6,479 5,467 6,500 Total number of workers 9,279 7,040 6,087 6,883 273 Only collections of worker ants are presented. Only general collections were conducted at 1800-2000 m; these included four leaf litter samples. P, from pitfall transect samples; W, from mini- Winkler sack and leaf litter transect samples; G, from general collection. A total of 148 ant species and 29,562 workers were collected. 100 FIELDIANA: ZOOLOGY CO o r- c CD O k_ CD CL 70-i 60- 50- 40- 30- 20 10- oiO- __Q □ Species ■ Individuals a f / c? 2000 m) in the RNI d'Andringitra, the leaf litter layer in open grassland/heath environment may receive more radiant heat during fog-free periods and thus support a greater ant diversity than at 1680 m. Although only 20 species were found in one af- ternoon of general collection at 1800-2000 m, compared with 31 species found using all meth- ods at 1680 m, I predict that a thorough study of the grassland/heath ant community at 2000 m will show a greater diversity of ant species than at 1680 m. This would be consistent with Brown's (1973) hypothesis about the role of radiant energy in determining ant diversity. Fauna! Similarities and Species Turnover Faunal similarity and beta diversity measures (Tables 8-3 and 8-4) suggest a division of the RNI d'Andringitra ant fauna into two communities, one occurring in lowland forest (785 and 825 m) and the other in montane cloud forest (1275 and 1680 m). Species turnover is greatest and faunal similarity is lowest between 1275 m and the two low-elevation sites. Although there is no detailed study of the temperature gradient along these el- evations, the montane cloud forest community zone corresponds to the area in which clouds ha- bitually develop on a daily basis (see also Chapter FISHER: ANT DIVERSITY PATTERNS 101 Table 8-2. Proportion of stations (out of 50 paired pitfall and leaf litter samples at each altitude) each species was recorded. The number of individual workers collected is given in parentheses. Genus Species 785 m 825 m 1275 m 1680 m CERAPACHYINAE Cerapachys PLAGIOLEPIDINI Plagiolepis MYRMICINAE CPvEMATOGASTRINI Crematogaster DACETONINI Glamyromyrmex Kyidris Smithistruma Strumigenys PHEIDOLOGETONINI Oligomyrmex PHEIDOLINI Pheidole 1 0.04 (90) 2 0.02 (3) 3 0.10(49) 0.06 (23) 5 0.02 (14) 0.02(1) 8 0.10(9) 0.12(46) 0.04 (30) 0.80(146) 0.54 (66) 3 0.32 (53) schenki 0.02(1) 0.02 (2) 1 0.06 (5) 0.02(1) 1 0.16(151) 0.14(130) 3 0.20(31) 1 0.02 (2) 2 3 0.06 (6) 4 5 6 7 8 0.36 (56) 0.08 (4) 9 0.06 (4) 0.02(1) 10 11 12 0.06 (3) 0.22 (21) 13 0.02(1) 0.02(1) 14 0.24(41) 0.16(11) 15 0.02(1) 16 0.02(1) 17 18 0.98 (417) 0.74(183) 1 0.64 (228) 0.50(114) 2 0.08 (69) 0.16(42) 2 6 7 0.02(1) 0.02(1) 8 9 10 11 0.66(188) 0.74 (348) 12 0.08 (40) 0.14(60) 13 0.30 (83) 0.26(18) 0.08 (5) 0.72 (154) 0.46 (78) 0.24 (25) 0.32 (32) 0.06 (5) 0.02(1) 0.08 (30) 0.76 (547) 0.76 (892) 0.44(102) 0.12(21) DOLICHODERINAE Technomyrmex mayri 0.04 (2) FORMICINAE CAMPONOTINI Camponotus 5 11 13 15 0.02 (3) 0.02(1) 0.04 (2) 0.02(1) 17 0.02(1) 0.04 (2) LASIINI Paratrechina 1 0.82 (869) 0.58 (546) 0.38 (232) 4 0.16(32) 0.62 (349) 0.04 (4) 0.02 (4) 5 0.22(132) 0.80 (560) 6 0.02(1) 0.06 (3) 0.18(14) 0.02(1) 0.02 (2) 0.60 (768) 1.00(3,772) 0.02 (7) 0.38(531) 102 FIELDIANA: ZOOLOGY Table 8-2. Continued. Genus Species 785 in 825 m 1275 m 1680 m SOLENOPSIDINI Monomorium TETRAMORIINI Tetramorium PONERINAE AMBLYOPONINI Amblyopone Mystrium Prionopelta ECTATOMMINI Discothyrea Proceratium PONER1NI Anochetus Hypoponera 14 0.26 (325) 0.52 (472) 15 0.02 (3) 0.10(13) 16 0.10(20) 0.16(24) 17 0.42(165) 0.46 (332) 20 0.16(24) 22 0.02(1) nemoralis 0.80 (828) 0.56 (349) veteratrix 0.60(147) 0.54(186) 0.02 (4) longispinosa 0.30(171) 0.66 (482) 1 0.14(29) 0.44 (246) 2 0.08(10) 0.28 (95) 3 0.28(165) 0.04(31) 5 0.32(146) 0.14(11) 6 0.88(1,862) 0.82 (953) 7 0.58 (228) 0.36(186) 8 0.42(191) 0.44(158) 9 0.16(169) 0.04(10) 10 0.12(14) 0.42 (239) 12 0.02(1) 14 0.32 (63) 1 0.02 (2) 0.86 (247) 3 0.02(1) 0.80 (256) 4 0.02(1) 0.04 (2) 5 0.16(13) 7 0.10(11) 0.04 (4) 0.08 (6) 9 0.12(37) 10 0.58(100) 0.34 (46) 0.62(157) 0.50(179) 11 0.20 (20) 0.06 (4) 13 0.02(1) 14 0.18(216) 0.26 (24) 15 0.02(1) 16 0.08 (5) 0.02(1) andrei 0.12(44) 0.08 (5) cognatum 0.58 (99) 0.22(15) 0.22(31) dysalum 0.34(110) 0.30 (45) electrum 0.16(12) 0.06 (6) marginatum 0.10(7) 0.20(16) 2 0.06 (3) 3 0.02(1) 1 0.02(1) 0.02(1) 1 0.56(122) 0.18(12) 3 0.22(12) 0.10(5) 1 0.02(1) 0.04 (3) 1 0.02(1) grandidieri 0.38 (96) 0.30(113) 1 0.08 (45) 0.08 (57) 0.88 (896) 0.46 (339) 2 0.44(154) 0.04 (3) 0.02 (26) 6 0.44 (236) 0.02 (23) 7 0.20(16) 0.24 (24) 0.90 (875) 8 0.04(10) 9 0.02 (4) 0.46 (92) 0.60(119) 10 0.68 (275) 0.02 (3) 11 0.72 (345) 0.36 (48) 0.76(351) 0.46(155) 12 0.18(21) 0.06 (3) 13 0.76 (227) 0.88 (326) 14 0.04(10) 15 0.18(25) FISHER: ANT DIVERSITY PATTERNS 103 Table 8-2. Continued. Genus Species 785 m 825 m 1275 m 1680 m Leptogenys Odontomachus Pachycondyla PSEUDOMYRMECINAE sakalava 1 3 coquereli cambouei 0.20 (20) 0.08 (4) 0.06 (4) 0.50 (66) 0.12(12) 0.02(1) 0.02(1) 0.06 (4) Tetraponera psw-87 psw-92 0.04 (2) 0.08 (9) 3). Further investigations of the high mountain grassland/heath ant community in Madagascar may also show a distinct ant community. Initial collection revealed 50% of the species unique to 1800-2000 m in the RNI d'Andringitra. Additional ant surveys in eastern Madagascar will help to evaluate whether there exists a dis- tinct, homogeneous, and disjunct montane cloud forest plus grassland/heath ant community in Madagascar. Currently we lack distribution rec- ords on those elevation-specific species to deter- mine whether they are part of a cloud forest or grassland/montane community on other mountain ranges in eastern Madagascar. Because the heath community is limited to the highest mountains of Madagascar, this disjunct habitat may have spe- cies or clades that are more closely related to taxa in other heath communities than in adjacent cloud forest ants. Beta-1 and beta-2 values differed from each other in comparisons of sites with high species turnover (Table 8-5). Beta-1 and beta-2 indices differ in their propensity to emphasize either spe- cies replacement or species loss, the two means of obtaining species turnover. Species replace- ment is most prevalent at the lower elevations. For example, the measured species turnover between Table 8-3. The number of species collected and first-order jackknife estimates of total species richness, based on pitfall and leaf litter transects. Statistics are given for each altitude and for all elevations combined. Species richness Altitude (m) Observed estimate 95% C.I. 785 77 91.7 0.363 825 67 83.7 0.395 1,275 39 46.8 0.247 1,680 23 27.9 0.242 All elevations 114 131.9 0.193 95% C.I. indicates 95% confidence intervals. 785 and 825 m is the result of a difference in species, not just the loss of species at 825 m. At higher elevations, however, there is species loss along the gradient without replacement. A nested dropout of species dominates the differences be- tween the 1275 and 1680 m ant communities. Un- der these circumstances, beta-2, which emphasiz- es species replacement over species loss, is the preferred measure of beta diversity (Harrison et al. 1992). Ant Fauna of RNI d'Andringitra In 1891, Forel provided keys and descriptions to 86 ant species known from Madagascar (Forel, 1891). This was the last comprehensive treatment of the ant fauna of Madagascar. In 1922, Wheeler recorded 207 species from Madagascar (Wheeler, 1922). Considering the number of undescribed taxa, the actual number of ant species in Mada- gascar may be on the order of 800. The chaotic state of the taxonomy of Malagasy ants makes it impossible to estimate the number of undescribed species listed in Table 8-1 without comparison to type material. There are a few gen- era, however, where recent revisions and type comparison make accurate estimates possible. Of the 19 species of Tetramorium collected in RNI d'Andringitra, only five are apparently described species. At least 17 of the 18 Strumigenys are un- described because there is only one available name for the entire Malagasy fauna (assuming that none of these specimens are tramp species). Two of the four Tetraponera are undescribed (P. S. Ward, pers. comm.). In addition, because there are no described species of Glamyromyrmex, Kyi- dris, Smithistruma, Leptothorax, Amblyopone, and Discothyrea from Madagascar, all species in these genera in Table 8-1 (eight total) are presumed to be undescribed. All 148 ant species in Table 8-1 are believed to 104 FIELDIANA: ZOOLOGY 100- (a) 785 m 90- 80- 70- 60- x** 50- i •♦* CD 40- • ♦ O 30- ♦ CD Q_ 20- C/) 10- < i i i i i ) 10 20 30 40 50 90- 80- 70^ 60- 50- 40- 30- 20- 10- 0- (b) 825 m ♦ • ♦ ♦ ( I 1 I 1 I ) 10 20 30 40 50 CD n E 50-, c 0) 40H > (c) 1275 m _C0 30 t 20-I i ♦ o 10-1 lMWw ^ i^1!:: :♦♦♦* 30 -. 25- 20- 15 10- 5- (d) 1680 m - 1 1 1 1 1 10 20 30 40 50 J lip .iiSffi — i 1 1 1 1 0 10 20 30 40 50 Number of stations pooled 140 CO © 120 (e) All elevations I co 100 — o 2 80- E 3 C 60 CD > j5 40 3 E 8 *° ? &* ,#^::::s * 20 — I- 40 -I 1 1 1 1 1 100 120 140 160 180 200 60 80 Number of stations pooled Fig. 8-3. Assessment of leaf litter ant sampling technique for each elevation (a-d) and for all elevations combined (e). The lower species accumulation curve in each chart plots the observed number of species as a function of the number of pooled stations sampled (stations = paired leaf litter and pitfall collections). The upper curve displays the non-parametric first-order jackltnife estimated total species richness (error bars are standard deviation) based on succes- sively larger number of samples from the data set (Heltshe & Forrester, 1983). Only every other point is shown in (a- d) and every fourth point is shown in (e). For all curves, each point is the mean of 100 estimates based on 100 randomizations of sample accumulation order, calculated using the program Estimates (R. K. Colwell, unpublished). FISHER: ANT DIVERSITY PATTERNS 105 Table 8-4. Two measurements of faunal similarity between the four elevational zones sampled. Table 8-5. Beta-1 (above the diagonal) and beta-2 (below the diagonal) diversity values of each pair of altitude sites. Elevation 785 m 825 m 1275 m 1680 m 785 m 0.655 0.160 0.075 Elevation 785 m 825 m 1275 m 1680 m 825 m 0.805 — 0.165 0.084 785 m — 0.208 0.724 0.860 1275 m 0.229 0.164 — 0.319 825 m 0.130 — 0.717 0.844 1680 m 0.096 0.068 0.394 — 1275 m 0.300 0.358 — 0.516 1680 m 0.210 0.239 0.205 Above the diaeona 1 is the Jaccard Index of similarity (presence/absence data) and below the diagonal, the sim- plified Morisita Index of similarity (abundance data; Horn, 1966). Higher values represent greater similarity. Values in boldface represent comparisons of altitudinally adjacent transects. be endemic to Madagascar. This is greater than the estimated 92% endemism in ant species in the dry forest of Kirindy, western Madagascar (P. S. Ward, pers. comm.). The lower level of distur- bance and the lack of open habitats may have pre- vented the invasion of introduced species into the RNI d'Andringitra. Comparison with Other Faunas There is currently no other site in eastern Mad- agascar whose leaf litter ant fauna has been sam- pled with comparable intensity and identified to species. One comparison, however, can be made at the generic level. Most of the ant genera col- lected in RNI d'Andringitra (Table 8-1) were also collected in the RNI d'Andohahela in the Anosy Mountains, southeastern Madagascar, in a parallel survey conducted in 1992 (Fisher, unpubl.) The only exceptions are the genera Pilotrochus, Eu- tetramorium, and Aphaenogaster, which were en- countered only in RNI d'Andohahela, and Tech- nomyrmex, Leptothorax, and Odontomachus, which were collected only in RNI d'Andringitra. The relative prevalence of species from differ- ent subfamilies in the RNI d'Andringitra ant fauna (62% Myrmicinae and 22% Ponerinae; Fig. 8-1) shows striking similarities to other tropical forest leaf litter ant communities (Table 8-6). All of the following studies found the Myrmicinae to be the most common subfamily, followed by the Poner- inae, with an average Ponerinae/Myrmicinae spe- cies ratio of 0.336 ± 0.037 (SD) (Table 8-6): Bel- shaw and Bolton (1994), in moist tropical forest in Ghana; Levings (1983), in moist forest on Bar- ro Colorado Island, Panama; Olson (pers. comm.), in tropical wet forest in western Panama; Longino (1986), in tropical wet forest in Costa Rica; An- Higher values represent greater species turnover. Val- ues in boldface represent comparisons of altitudinally adjacent transects. Overall beta-1 diversity was 1.214 and beta-2 diversity was 0.481. dersen and Majer (1991), in seasonally dry forest in Kimberley, northwestern Australia; and D. M. Olson and P. S. Ward (pers. comm.), in the dry Kirindy forest. The 1:3 ratio of Ponerinae to Myr- micinae in these studies suggests that the species richness of the Myrmicinae can be interpolated from the richness of the Ponerinae. In addition, if future studies in Madagascar confirm an approx- imate prevalence of 22% Ponerinae or 62% Myr- micinae, investigations directed at determining the species richness of leaf litter ants in an area may be able to use the number of Ponerinae or Myr- micinae species as an indicator for the entire fam- ily Formicidae. This extrapolation applies only for estimating the ground-dwelling, leaf litter ant fau- na. Potential Effects of Human Disturbance The 785 and 825 m transects differed slightly in ant composition, with low values of beta-1 and 100-, O 80 o 45°). Station 8 (23 November 1993) was along another river of the same order and parallel to the Volontsangana River, but at a slightly lower elevation; it was sampled only for Simuliidae. This river had a strong water flow and ran under an overlapping gallery forest. Its steep- ness was less than that of station 9, and the sub- strate was predominantly pebbles. 1625 m Elevational Zone — Two steep water- courses were sampled. These were small streams cascading onto flagstone. Station 10 (25 Novem- ber 1993) was in a relatively open area, and water discharge was less than 1 liter per second. Station 11 (24 November 1993) was in forest largely composed of bamboo and had a discharge of a few liters per second. 110 FIELDIANA: ZOOLOGY Western Slope Stations Four stations were sampled on the western slope, all of them along the Zomandao River, a confluent of the Mangoky Basin. Three sites were sampled during the November 1993 mission (sta- tions 13, 14, and 15; 28-30 November 1993), whereas station 12 was sampled during the No- vember 1992 field trip. It was retained because it lay on the Zomandao River at an altitude close to that of stations 1, 2, and 3 on the eastern slope. Station 12 was located near Ankaramena, along a sandy stretch of the Zomandao River (stream or- der 4), at 600 m. Here the watercourse was reg- ularly broken by rocks and was situated in a sa- vannah area. Station 13 was at the level of An- tanifotsy (1400 m); a rocky shelf and large boul- ders impeded the medium water flow (stream order 3). This river ran through a zone of savan- nah. Station 14 was on the western part of the high plateau of the reserve, at an altitude of 1900 m. The site was near a small lake formed by the widening of the river (stream order 2), where it passed through a zone of grassy savannah inter- laced with Ericaceae bush. The river course was relatively level and was made up of flagstone, forming shelves, that alternated with deeper water zones that had stony-gravel bottoms. Station 15 was located at the edge of the high plateau, at an altitude of 1950 m and a short distance from the source zone. At this site, the river ran through a zone of grassy savannah. Its steepness was slight (stream order 2), and the river bottom was made up of either flagstone or pebbles. Methods Sampling Techniques Three different methods of sampling were used: Evening Light Traps — This method was used to collect adults of aquatic insects with nocturnal flight activity. A large shallow pan was filled with water that was mixed with a tension-active agent (soap). The light sources were a white light gen- erated by a natural gas lamp (camping stove type), and an ultraviolet (UV) light produced by a bat- tery lamp with a UV tube. The trap was placed along the river bank, illuminated 10 minutes be- fore sunset, and switched off 1 hour later. Ephem- eroptera and other fragile insects were collected and preserved individually as they were caught in the trap. The other insects were filtered and pre- served in 70% ethanol. This method was highly efficient for the sampling of Trichoptera, less so for Ephemeroptera, and generally useless for Si- muliidae. Butterfly Net — This method was used pri- marily to capture adult Ephemeroptera, which swarm during the day and at dusk, and adult Odo- nata. Collection of Aquatic Stages — The different river substrates (pebbles, vegetation, fallen leaves) were sampled using a submerged net laid down- stream. Aquatic stages were also collected directly on the substrates. All substrate types were sam- pled thoroughly. In many cases the larval stages of various aquatic insects are known, and this technique complements and augments species col- lection by light traps. It was the only effective method for capturing Simuliidae. Results The collections made during the survey of the RNI d'Andringitra were sorted and sent to nu- merous specialists. Definitive results for numer- ous groups are not yet available, and the follow- ing results are far from complete for all aquatic insects. Descriptions of numerous new genera and species will be published elsewhere. In general our current knowledge of the species limits and geographical distribution of all aquatic insects is too imprecise for us to present a general review of the endemicity and ecological require- ments of this group. Herein we concentrate on three groups that we studied: Diptera (Simuli- idae), Ephemeroptera, and Trichoptera. Diptera: Simuliidae Simuliidae females in the imaginal stage are he- matophagous, whereas males are floricolous. Lar- vae live in well-oxygenated water. The identifi- cation of the Simulium species was relatively easy with pupae, particularly using characters associ- ated with the form and the number of gill fila- ments. The identification of larvae and adults to species level is more difficult; these life stages are seldom captured. Nine Simulium species were identified from the 15 stations. Among these nine species, four were previously described {Simulium gyas, S. pentacer- GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 111 os, S. iphias, and S. imerinae), although the S. iphias is problematical and may comprise four distinct species (see below). Two species are new to science and are known only from the Andrin- gitra Massif (5. metecontae n. sp. and S. brunhesi n. sp.) (see Chapter 11). Within our material of Simulium iphias from the Andringitra Massif there are several morpho- logical species, each of which is distinguishable from the others by its number of pupal gill fila- ments (herein filaments are designated "f"), and each of which apparently lives in unique ecolog- ical conditions. Four Simulium iphias s.s.l. mor- phospecies are recognized from the reserve: 5. iphias 8f, S. iphias lOf, S. iphias 15f, and S. iphias 19f. Distribution of Simulium in the Andringitra Area Species found on the eastern slopes of the re- serve included S. metecontae n. sp., 5. iphias lOf, S. pentaceros, and S. gyas; those on the western slopes included S. brunhesi n. sp., S. imerinae, S. iphias 8f, S. iphias 15f, and S. iphias 19f. No Simulium species occurring in the forested areas along the eastern slopes was found on the western side of the reserve; species within this genus thus appear to have relatively strict ecological require- ments (Fig. 9-1). Eastern Slope Simulium Simulium iphias lOf — This form was collected at stations 2, 4, and 9, of stream order 3, 4, and 5 (not respective). It was not found at stations 3 and 8. Stations 2, 4, and 9 had mean water flow ranging between 0.7 and 1.5 m/s. In contrast, wa- ter flow was slower at sampling stations 3 and 8. This species occurs all along the eastern coast of Madagascar, in small rivers running through zones of primary or degraded forest. It is absent from rivers and streams in completely degraded zones. Simulium metecontae n.sp. — This species is currently only known from the Andringitra Massif (see Chapter 11). It was first encountered in the 820 m zone at the four stations (4, 5, 9, and 11) on rivers of stream order 1,3, and 4. Altitude and stream order may be critical parameters for the distribution of this species. Indeed, water flow is Wester n slope Eastern slope 2000- 1500- 1000- m Diptera : Simuliidae |B,8f| °) 14« B, 8f, l| 8 t .« uP 13« ST "^^10 M, 10f, G /T 2 |M, I 5] -^^^ V |15f, 19f| M, 10f,Gr4 n\ 500- Stream order 1 ^^^^ Stream order 4 Stream order 2 ■■■■■ Stream order 5 ^_^_ Stream order 3 *j> ■ sampled stations Fig. 9- 1 . Altitudinal distribution of Simulium in the RNI d' Andringitra on the basis of 1 1 sites on the eastern slopes and four sites on the western slopes. Key to spe- cies: I = Simulium imerinae, B = 5. brunhesi, G = S. gyas, M = 5. metecontae, P = S. pentaceros, 8f = S. iphias with eight filaments, lOf = S. iphias with 10 fil- aments, 15f = 5. iphias with 15 filaments, and 19f = S. iphias with 19 filaments. relatively slow for stations 6, 8, and 10, where this species was not collected. Simulium pentaceros — This species was found only at station 5. It is an infrequently collected species of Simulium, although it is also known from two forested rivers near Ranomafana, Ifan- adiana. It is difficult to make any general state- ments on its ecology. However, the two rivers near Ranomafana are of the same stream order as station 5 and at about 850 m. Simulium gyas — This species was found at three stations (4, 8, and 9) of stream order 4 and 3, but not at station 5. S. gyas is broadly distrib- uted along the eastern coast from Antsiranana to Tolagnaro in small to medium-sized rivers run- ning through primary or degraded forest. It has a broad elevational distribution. Western Slope Simulium Simulium iphias 8f — This species was collected at stations 14 and 15, located on the high plateau 112 FIELDIANA: ZOOLOGY of the Andringitra Massif. The only other known record of this form is from the Ankaratra Massif, at the level of the source of the Sisaony River (1800 m). It is a Simulium of high altitude and open zones. Simulium brunhesi n.sp. — This species is cur- rently known only from the ericaceous savannah of the RNI d' Andringitra (see Chapter 11). It pre- sumably will be found at other high mountain sites on the island with similar habitats (e.g., Tsar- atanana and Ankaratra). Simulium imerinae — This species was collected at stations 13 and 14, along the Zomandao River and above 1300 m. It has been previously record- ed in numerous highland and foothill areas in the southern and eastern portions of the island. This species prefers small brooks running across flag- stone, with water flow of 1.2-2 m/s. Simulium iphias 15f — This species was found only at station 12, on the western slope. It has previously been collected in foothill areas in the southern and western parts of Madagascar from the Mandrare River to the Betsiboka River. Simulium iphias 19f — As with S. iphias 15f, S. iphias 19f was collected only at station 12. It in- habits areas with rapids on large highland rivers. Ankaramena is the lowest altitude at which it has been recorded. The cold waters of the RNI d' Andringitra may account for its low altitudinal distribution. Conclusions — Simuliidae The eastern and western slopes of the RNI d' Andringitra are inhabited by different species of Simulium. Water temperature and aquatic vegeta- tion (nutrients) are known to be important vari- ables accounting for these differences. Adult fe- male Simulium are hematophagous, and their feeding regimens may be strict and dependent on specific hosts (e.g., batrachians, reptiles, and mammals), perhaps at the family, genus, or spe- cies levels. Thus, local extirpation of vertebrates associated with deforestation may have a pro- found effect on the distribution of Simulium. The aquatic habitat of larvae and pupae may be only one of the distribution parameters for Simulium species. Species diversity is greatest in medium to large rivers on both the eastern and western slopes of the massif, particularly in waters of stream orders 2, 3, and 4. The extreme stream orders 1 and 5 are inhabited by only one Simulium species. In our sampling of Simulium in the RNI d' Andring- itra, five species were found on the western slopes and four species on the eastern slopes. A single species new to science was found on each slope. On the basis of variables associated with sam- pling stations (Table 9-1), it is difficult to state precisely what effects altitude and stream order have on Simulium distribution, because small and cold water bodies are located at the highest alti- tudes, and there is no large high-mountain river on the Andringitra Massif. Ephemeroptera Little is known about this order of insects in Madagascar (Demoulin, 1970). We are in the pro- cess of conducting inventories around the island. Many species were collected in the RNI d' An- dringitra, the majority of which are new to science and are currently being studied by several spe- cialists. Two families are considered here, Euthy- plociidae and Leptophlebiidae. It is important to remember that the standard emergence period for most mayflies is usually at the end of the rainy season (April and May), and that the Andringitra field trip was during the month of November. Euthyplociidae Only one genus of this family has been record- ed from Madagascar (Proboscidoplocia Demou- lin, 1966). Until recently, this genus was thought to be monospecific, represented by Proboscido- plocia sikorai (Vayssiere, 1895). Extensive studies are being jointly carried out by the Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE), An- tananarivo, and the Musee Cantonnal de Zoologie de Lausanne. This research has shown that at least 10 Proboscidoplocia species occur on Madagas- car (Elouard et al., unpublished data). Among these 10 species, three are known from the An- dringitra Massif: Proboscidoplocia sikorai sensu stricto (s.s.) (stations 1, 4, 5, and 6); Probosci- doplocia vayssieri n. sp. (Elouard & Sartori, in press) (station 9); and a third species, recorded only at station 4, known only from females, and identified by eggs. The P. sikorai s.s. and P. vays- sieri n. sp. are easily distinguished by the shape of male genitalia. Proboscidoplocia pupae live under rocks in permanent water and were cap- tured with deep nets placed under rocks. Imagos GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 113 were caught by evening light traps. Most Ephem- eroidea appear in masses a few days per year. The November Andringitra mission coincided with one of these synchronous emergences. Specimens of males and females were captured at several sta- tions. Distribution of Euthyplociidae in the RNI d'Andringitra Proboscidoplocia were collected only on east- ern rivers (Fig. 9-2). The distribution of P. sikorai s.s. appears to be independent of stream order and forest cover. This species occurs in rivers of stream orders 5 and 2, passing through degraded and primary forest. Altitude appears to be a major factor for the separation of these two species, which are not elevationally sympatric on the mas- sif. Differences in elevation are related to water temperatures. No Proboscidoplocia adults or pu- pae were found in the 1625 m zone (stations 10 and 11), although these sites were thoroughly in- spected. At this time it is impossible to affirm that their absence from the upper forest zone of the mountain is related to stream order or altitude. Proboscidoplocia sikorai s.s.d. has been recorded in the RNI de Marojejy (Fontaine, 1968), al- though no information is available on stream or- der. Leptophlebiidae Only five species of Leptophlebiidae are de- scribed from Madagascar. Two are known from the nymphal stage and three as imagos. On the basis of our recent collections from numerous sites and systematic studies at LRSAE, there are more than 50 species occurring on the island. This family is essentially forest-dwelling. One special- ist, W. L. Peters, is carrying out an extensive sys- tematic revision of this family in Madagascar. Twelve species of Leptophlebiidae were col- lected on the Andringitra Massif, 1 1 of which are new to science. We have classified these 12 taxa as subimaginal and imaginal morphospecies. The only named species is Nesophlebia adusta Peters and Edmunds, 1964, which is referred to here as Lepto-Q. Eastern Slope Leptophlebiidae Eleven species were collected along the eastern slopes of the RNI d'Andringitra. It is difficult to Western slope Eastern slope 2000 -- 1500 -- 1000 500 -- m Ephemeroptera : Euthyplociidae Proboscidoplocia J? ®& ^Vf * fi^y] "" Stream order 1 ■■m Stream order 4 Stream order 2 i^HHi Stream order S mm Stream order 3 A = sampled stations Fig. 9-2. Altitudinal distribution of Proboscidoplo- cia in the RNI d'Andringitra on the basis of 1 1 sites on the eastern slopes and four sites on the western slopes. Key to species: V = Proboscidoplocia vayssieri, S = P. sikorai, and Sp3 = P. sp. 3. give precise details of habitat preferences of the various morphospecies on the basis of variables associated with collection stations. In general, only the following statements can be made: (1) At stations 1 and 5, five species were obtained at each station, and the species assemblages were totally different from one another. (2) No imago Leptophlebiidae was collected in the 1625 m zone (stations 10 and 11), although indeterminate lar- vae belonging to this family were found on these two small tributaries. The low nymphal densities unquestionably account for the absence of adults. (3) Lepto-P, Lepto-N, Lepto-S, and Lepto-AA were found only in the Iantara River. Stream order may be a critical parameter in the distribution oi these species. (4) Nesophlebia adusta (Lepto-Q) occurs in both the Iantara River (station 1) and one of its small tributaries (1210 m, station 9) For this species, stream order is of least impor- tance in the stock range established in the RNI d'Andringitra. Although this species was not col- lected at stations 2, 3, 4, and 5, its absence ai these stations is not proven. Once again, densit) and emergence periods may account for their lo cal absence at these sites. (5) Lepto-Y, Lepto-W and Lepto-Z were found at three stations of sim 114 FIELDIANA: ZOOLOGY Western slope Eastern slope 14 « ► 2000 -- 1500 -- 1000-- 500 -- m . _ ! Ephemeroptera : Leptophlebiidae P,Q,N,S,AA Stream order 1 . Stream order 2 . Stream order 3 h Stream order 4 ai Stream order 5 t tampled •Uliooi Fig. 9-3. Altitudinal distribution of Ephemeroptera: Leptophlebiidae in the RNI d'Andringitra on the basis of 1 1 sites on the eastern slopes and four sites on the western slopes. Key to species: AA = Lepto-AA, N = Lepto-N, O = Lepto-O, etc. ilar stream order. (6) Lepto-U and Lepto-V oc- curred only at station 5. Western Slope Leptophlebiidae Only one species, Lepto-O, was collected along the western slopes of the RNI d'Andringitra, thus supporting the idea that this family is largely for- est-dwelling. At station 1 2, three Leptophlebiidae subimagos were collected. At this development stage there are no clear characters for morpho- species determination. These three subimagos ap- pear to be different from subimagos obtained on the eastern slopes and from station 15. On the basis of current information, Lepto-O is known only from the high plateau ericaceous savannah of the RNI d'Andringitra, where it is presumably endemic. Other Ephemeroptera Species Among the other Ephemeroptera species col- lected during the 1993 mission to the RNI d'An- dringitra and currently under study by specialists are: (1) Several Baetidae (> 25 species). (2) One Ephemerellidae species. This family lives in cold waters with aquatic vegetation, and it is practi- cally unknown in Africa. Only three species are recorded from South Africa (Demoulin, 1970). In Madagascar, only one specimen was previously known, a larva identified as belonging to the ge- nus Ephemerella (Eurylophella); this identifica- tion may be incorrect. The shape of the nymph and the genitalia of the imago belong to Ephem- erellidae and not Telagodinae. In the RNI d'An- dringitra we were able to collect all developmen- tal stages of a mayfly that does not correspond to Ephemerella. The RNI d'Andringitra material un- questionably belongs to an undescribed genus. (3) Larvae of Trichorthidae, but no imago. (4) Larvae and adults of Prosopistomatidae that are probably referable to Prosopistoma variegatum. In view of the poor original description of this species, how- ever, examination of the holotype is needed to confirm this identification. Unfortunately, it ap- pears that the holotype is lost. (5) Heptageniidae (see Chapter 10). Conclusions — Mayflies More than 40 species of mayflies were collect- ed during the Andringitra mission. The great bulk of these species are new to science and have been distributed to specialists for study. These descrip- tions will be published in due course. The RNI d'Andringitra material will be used in the broad context of an island-wide survey of the aquatic insects of Malagasy rivers, with specific reference to speciation and ecological preferences. The order Ephemeroptera is a good indicator of environmental conditions. However, the ima- ginal sampling of this group is seasonal, and im- agos have a short existence, in general less than 1 day. Furthermore, emergence of most mayfly families on Madagascar is synchronous. Never- theless, the presence of pupae in rivers is constant, and their collection can provide species identifi- cations in most cases. Few associations between larvae and adults, however, are known. Rearing of larvae is critical in determining the life history stages of each species, and this fieldwork is time- consuming. Trichoptera The families and genera collected during this study were Hydropsychidae: Macrostemum, Cheu- GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 115 Table 9-2. Distribution of Leptophlebiidae in the RNI d'Andringitra (see Fig. 9-3). Species Eastern stations Western stations 10 11 12 13 14 15 Lepto-P + Lepto-Q + Lepto-N + Lepto-S + Lepto-AA + Lepto-Y Lepto-U Lepto-V Lepto-W Lepto-X Lepto-Z Lepto-O Rain Rain Rain Rain Rain + Rain + Rain + Rain + + Rain + Rain + Rain Rain matopsyche, and Leptonema; Hydroptilidae: Hy- droptila, Orthotrichia, and Oxyethira; Philopo- tamidae: Chimarra, Dolophilodes, and Pauliano- des; Ecnomidae: Psychomyiellodes; Pisuliidae: Western slope 2000 -- "fill ■- 1500 -r- 1000 500 Eastern slope Trichoptera Philopotamidae aEJdAdBpEpC| aEaFK aGpE MNJK 4 3 JaG " HLaG] - Stream order I ^^^_ Stream order 4 _ Stream order 2 ■■■■■■ Stream order 5 . Stream order 3 N° - sampled stations Fig. 9-4. Altitudinal distribution of Trichoptera: Philopotamidae in the RNI d'Andringitra on the basis of 1 1 sites on the eastern slopes and four sites on the west- ern slopes. Key to species: I = Chimarra sp. I; D = Chimarra sp. AH, sp. D, and sp. E; pD = Paulianodes sp. D; pC = Paulianodes sp. C; pE = Paulianodes sp. E; aG = Chimarra sp. AG; L = Chimarra sp. L; aE = Chimarra sp. AE; M = Chimarra sp. M; N = Chimarra sp. N; J = Chimarra sp. J; aF = Chimarra sp. AF; dA = Dolophilodes sp. A; dB = Dolophilodes sp. B; d. = Dolophilodes sp.; and K = Chimarra sp. K. Pisulia; Lepidostomatidae: Goerodes; Polycentro- podidae: Paranyctionphylax and Pseudoneur- eclipsis; and Leptoceridae: Adicella, Oecetis, Tri- anodes, Setodes, Leptocerus, and Athripsodini. On the large rivers, the caddisfly populations are dominated by the genus Macrostemum and the Athripsodini group, the latter remarkable for its high species diversity. We present an analysis of the family Philopotamidae, because this group was captured commonly and its alpha taxonomy is relatively straightforward. Philopotamidae Philopotamidae were caught at almost all sta- tions (Fig. 9-4 and Table 9-3). Three genera were identified: Paulianodes (three species), Dolophil- odes (two species), and Chimarra (eight species from the forest of the eastern slope and one spe- cies on the high plateau at the foot of Pic Boby). Three Chimarra species collected on an earlier mission to the region and outside the reserve along the Zomandao River (Mangoky Basin) were taken into consideration in order to compare the eastern and western slopes of the massif. All of the Paulianodes, Dolophilodes, and Chi- marra species collected are undescribed (Table 9-4). Most species are being studied at the LRSAE, Antananarivo. Some of these taxa are represented in the collections of the Museum Na- tional d'Histoire Naturelle (MNHN), Paris; these species will be described by J. Olah, who kindly communicated his results. This taxonomic infor- mation allowed us to identify the species collected during this study and to summarize the available information on their geographical distributions. 116 FIELDIANA: ZOOLOGY Table 9-3. Distribution of the Philopotamidae in the RNI d'Andringitra. Sampling station: 15 12 11 10 9 6 5 4 2 3 1 (west) (west) (east) (east) (east) (east) (east) (east) (east) (east) (east) Altitudinal zone (m): 2000 580 1625 1625 1210 810 810 810 720 720 720 Chimarra sp. E Chimarra sp. AH Paulianodes sp. D Dolophilodes sp. Dolophilodes sp. A Dolophilodes sp. B Paulianodes sp. C Chimarra sp. AE Paulianodes sp. E Chimarra sp. J Chimarra sp. AF Chimarra sp. K Chimarra sp. M Chimarra sp. N Chimarra sp. AG Chimarra sp. L + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Table 9-3 lists the species found within the re- serve and the collecting stations. Conclusions — Caddisflies The caddisflies show a considerable faunal dis- similarity between the western and eastern slopes of the Andringitra Massif, specifically between the Zomandao River (Mangoky Basin) and the Sahavatoy and Iantara rivers (Manampatrana Ba- sin), respectively. There are some exceptions: at least one Athripsodini species, one Hydroptila species, and three Orthotrichia species occur up- stream in both basins. For the Philopotamidae the faunal dissimilarity is complete. It is interesting to examine this dissimilarity, because there are few available data on the geo- graphical distribution of aquatic insect species in Madagascar. (1) "Open area" species of the western slope — Chimarra sp. I, C. sp. D, and C. sp. E — also occur upstream of the Mandrare Ba- sin (especially on the Mananara River). Chimar- ra sp. AH is one of the most common Trichop- tera species known in Madagascar, especially in the southern and western regions. (2) Forest spe- cies on the eastern slope — Chimarra sp. N, C. sp. K, Paulianodes sp. D, P. sp. C, P. sp. E, Table 9-4. List of Philopotamidae species. Species Taxonomic status Proposed name Material fabienneae LRSAE langrandi LRSAE goodmani LRSAE andringitra LRSAE sahavatoyae LRSAE voromhola LRSAE sahanivorakyae LRSAE michaeli LRSAE erici LRSAE chatugana LRSAE and MNHN nadia LRSAE and MNHN watayana LRSAE and MNHN wigota LRSAE and MNHN atana LRSAE and MNHN elatrasoa LRSAE anadabolava LRSAE anka LRSAE and MNHN Paulianodes sp. C Paulianodes sp. D Paulianodes sp. E Dolophilodes sp. A Dolophilodes sp. B Chimarra sp. K Chimarra sp. N Chimarra sp. J Chimarra sp. L Chimarra sp. AE Chimarra sp. AF Chimarra sp. AG Chimarra sp. M Chimarra sp. I Chimarra sp. D Chimarra sp. E Chimarra sp. AH Gibon (in prep.) Gibon (in prep.) Gibon (in prep.) Gibon (in prep.) Gibon (in prep.) Gibon (in prep.) Gibon (in prep.) Gibon (in prep.) Gibon (in prep.) Olah (in prep.) Olah (in prep.) Olah (in prep.) Olah (in prep.) Olah (in prep.) Gibon (in prep.) Gibon (in prep.) Olah (in prep.) GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 117 Dolophilodes sp. A, and D. sp. B — are known only from the primary forests of the RNI d'An- dringitra. Chimarra sp. AF, C. sp. AE, C. sp. J, and C. sp. L are also known from the Namorona Basin in the primary forests of the Pare National de Ranomafana. Chimarra sp. AG occurs from the Mandraka River (Mangoro Basin) and the upper Mananara River. Chimarra sp. I has been collected from the latter locality. Finally, Chi- marra sp. M occurs on one tributary of the Man- goro River in the forested area south of Mora- manga (Anosibe an'ala). The "open area" or savannah species are known from other open areas (western and southern Madagascar), whereas "forest species" are known from other forests (eastern Madagas- car) or known only from the RNI d'Andringitra. From an ecological point of view, two distinct faunal assemblages are present on the island: a humid forest group and a savannah group. There is probably no endemicity with regard to streams. However, because parameters account- ing for the distribution of Philopotamidae (phys- icochemistry of water, substrates, and nutrition resources) are dependent on the vegetation pre- vailing around the stream, the effects of the bi- ome are determinant factors. These biogeograph- ical results parallel results from studies conduct- ed on west African Cheumatopsyche (Statzner, 1984) and Chimarra (Gibon, 1985), and they confirm that the geographical distributions of these lotic insects depend more on vegetation zones or biomes than on stream orders. A second point associated with the distribution of species on the eastern slopes (see Table 9-4) is that each of the four elevational zones (710, 820, 1210, and 1625 m) appears to have a relatively distinct fauna. The highest sampled zone on the eastern slopes (1625 m) consists only of small streams, where Paulianodes sp. D was collected, together with a species belonging to Dolophilo- des. The torrent at 1210 m is where Paulianodes sp. C, Dolophilodes sp. A, and D. sp. B were found. The zone around 820 m had the highest species diversity, including Chimarra sp. AF, C sp. K, C. sp. M, and C sp. N. Chimarra sp. L was limited to rivers in the 710 m zone. The re- maining species were distributed in two zones: between 820 m and 1210 m {Chimarra sp. AE, C. sp. J, and Paulianodes sp. E), and between 710 m and 820 m (C. sp. AG). It is premature to provide precise details on habitat specificity of these Philopotamid species given the period of study and the number of spec- imens collected. However, our current information shows a clear pattern of altitudinal distribution of the various species that can be divided into "fau- nal zones" (Table 9-3). If no endemicity is demonstrated at the level of the lower rivers, then the faunistic changes related to elevational variables would support the idea of local endemicity in the mountainous streams. This hypothesis is supported by the fact that the species collected in the 1210 and 1625 m zones are known only from the Andringitra Massif, whereas the majority of the species collected at lower el- evations (710 and 820 m zones) are also known from other forests (Ranomafana, Andohahela, or Moramanga). The three genera of Philopotamidae occurring on the Andringitra Massif are distinct clades (Ross, 1956). Paulianodes is endemic to Mada- gascar and is a primitive element of the fauna; it has probably been isolated since the beginning of the Cretaceous. On the other hand, the genus Chi- marra, which is distributed worldwide, has colo- nized Madagascar in more recent geological time (Miocene). Dolophilodes may occupy an inter- mediary position; its status is still poorly known. It is interesting to note that elements that represent the highest level of endemicity (the subfamily Paulianodinae) and that are among the oldest rep- resentatives of the Malagasy fauna are restricted to small, high-altitude tributaries in primary for- ests. Before this study, the genus Paulianodes was only known from one specimen (P. tsaratananae Ross, 1956), reported from identical environments on the Tsaratanana Massif. Other species belong- ing to Paulianodes and Dolophilodes are being studied at the LRSAE. All of these specimens were caught in intact primary forests (Ranoma- fana, Andohahela, and Montagne d'Ambre), and each locality has its own species assemblage. This confirms the hypothesis of endemism localized in high mountain areas. The highest species richness of Philopotamidae (six species) occurs in the intermediary forest streams (station 5 and 9). Small tributaries were probably undersampled because of the method used (light traps). In contrast, the faunas of the larger rivers (Sahavatoy and Iantara) are relativelj depauperate. This is due to the combination ol steep slopes, frequent torrents, and dramatic in- creases in water level that disrupt the benthic en vironment and suppress the diversification of tht bank ecosystems. 118 FIELDIANA: ZOOLOGY Other Aquatic Insects Among the other aquatic insects of relative im- portance collected during the Andringitra mission are Megaloptera and Plecoptera. Megaloptera — Several pupae belonging to two species were collected; one was found in the 710, 820, and 1210 m zones and the second at 1625 m. Three Megaloptera species are currently known from Madagascar — Protosialis afra Navas 1936, P. madegasca Navas 1927, and Madachau- lioda torrentialis (Paulian, 1951). Protosialis spp. are known from the high mountain regions of Tsaratanana and Ankaratra (Paulian, 1951). The two species recorded in Andringitra seem to be- long to the genus Madachaulioda. At least one of the species is known to science. Because the pu- pae of other Malagasy species have not yet been described, we could not identify our specimens. Plecoptera — All known Malagasy Plecoptera belong to the genera Madanemura and Tsarane- mura (Paulian, 1951). Three species had been pre- viously described by Paulian (1959): Madane- mura andringitrinsis, M. perrieri, and M. descar- pentriesi. During the sampling of the RNI d' Andringitra, we collected only pupae of Plecoptera between 750 and 1210m (stations 5 and 9). The specimens collected in 1993 are probably not representative of species described from the massif by Paulian (1959), because all of those taxa were taken either at the base of Pic Boby (within the Sohanihindra- no Plateau) or at the peak. In these areas, ranging from 2000 to 2600 m, the vegetation is sclero- phyllous forest. Malagasy Plecoptera are known from all the main massifs — Andringitra, Ankara- tra, Tsaratanana, and Montagne d'Ambre. In all known cases the species are endemic and repre- sent geographical isolates. General Conclusions The results presented in this chapter illustrate the main problem associated with research on tropical invertebrates. The vast majority of species collected during the mission are new to science. The proper study of these undescribed forms takes a considerable amount of time and needs to be integrated in a broad systematic review of the spe- cific group. Moreover, the importance of these data in regard to species conservation can only be determined after extensive fieldwork and collect- ing in order to gather data on geographical and ecological distribution. This type of information is generally lacking for most new species. On the basis of the few groups of aquatic in- sects that we have studied on Madagascar, the fol- lowing conclusions can be presented: (1) There are clear differences between the faunas found on the eastern slopes, in humid forest, and those on the western slopes, in open savannah. (2) For sev- eral groups there is elevational variation in the distribution of various taxa, although the precise parameters giving rise to this pattern (e.g., river substrates, water temperatures, and the importance of the overlapping canopy) are unclear. (3) Open area species from the western slopes generally have a broader geographical distribution than hu- mid forest species. Upstream species have more restricted ranges than downstream species. (4) In the case of Madagascar, elements presenting the greatest level of endemicity (Paulianodes, repre- senting a subfamily endemic to the island; Me- galoptera; and Plecoptera) are restricted to rivers associated with primary and high-altitude forests. Acknowledgments We are grateful to Mr. Abel Ralaiteferana for his help during the Andringitra mission. We ex- press our gratitude to the World Wide Fund for Nature for inviting the LRSAE group to partici- pate in the Andringitra mission, and to Steven Goodman for translating the text into English. Literature Cited Demoulin, G. 1966. Contribution a l'enide des Euthy- plociidae IV (un nouveau genre de Madagascar). An- nates de la Soctete" Entomologique de France, N.S., 2: 941-949. 1970. Ephemeroptera des faunes ethiopiennes et malgaches. South African Animal Life, 14: 24-170. Elouard, J.-M., and Sartori, M. In press. Probosci- doplocia (Ephemeroptera: Euthyplociidae), a singular plural. Proceedings of the VHIth International Mayfly Conference, 14-20 August 1995, Lausanne, Switzer- land. Fontaine, J. 1968. Contribution a l'&ude des Ephem- eropteres malgaches: La super f ami lie des Epheme- roidea. Bulletin Mensuel de la Soctet6 Linneenne de Lyon, 20(6): 228-242. Fraser, F. C. 1956. Faune de Madagascar. I. Odonates Anisopteres. ORSTOM/CNRS, Paris and Antananari- vo. Gibon, F.-M. 1985. Recherches sur les Trichoperes GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 119 d'Afrique Occidentale. 3 — Philopotamidae de Cote d'lvoire. Revue d'Hydrobiologie Tropicale, 18(1): 23- 30. Kjener, A. 1963. Poissons, peches et pisculture a Mad- agascar. Centre Technique Forestier Tropical, 24: 1- 160. Malicky, H., and Chantaramongkol, P. 1993. The al- titudinal distribution of Trichoptera in Mae Klang catchment on Doi Inthanon, northern Thailand: Stream zonation and cool- and warm-adapted groups. Revue d'Hydrobiologie Tropicale, 26(4): 279-291. Navas, L. 1936. Communicaciones entomologicas. 19 Insectos de Madagascar (3dmc serie). Revista de la Ac- ademia de Ciencias de Zaragoza, third series, 19: 100-110. Paulian, R. 1949. Decouverte de l'ordre des P16cop- teres a Madagascar. M6moires Institut de Recherche Scientifique de Madagascar, seYie E, 4: 359-363. . 1951. Faune des eaux douces de Madagascar. Plecopteres et M6galopteres. M6moires Institut de Re- cherche Scientifique de Madagascar, s6rie A, 6(1): 53- 61. . 1959. Nouveaux Plecopteres malgaches. M6- moires Institut de Recherche Scientifique de Mada- gascar, s6rie E, 23: 9-16. Peters, W. L., and Edmunds, G. F. 1964. A revision of the generic classification of the Ethiopian Leptophle- biidae (Ephemeridae). Transactions of the Royal En- tomological Society of London, 116(10): 225-253. Ross, H. H. 1956. Evolution and classification of the mountain caddisflies. The University of Illinois Press, Urbana, 213 pp. Schmidt, E. 1951. The Odonata of Madagascar (Zy- goptera). Memoires Institut de Recherche Scientifique de Madagascar, 6(1): 115-279. Statzner, B. 1984. Keys to adult and immature Hy- dropsychinae in the Ivory Coast (West Africa) with notes on their taxonomy and distribution. Spixiana, 7(1): 23-50. Vayssiere, A. 1895. Description zoologique de VEu- thyplocia Sikorai, nouvelle espece d' Ephemeridae de Madagascar. Annales de la Societe Entomologique de France, 64(26): 297-306. 120 FIELDIANA: ZOOLOGY Chapter 10 New Heptageniidae (Insecta: Ephemeroptera) from the Reserve Naturelle Integrate d'Andringitra, Madagascar Michel Sartori and Jean-Marc Elouard Abstract Two new species of Heptageniidae (Insecta: Ephemeroptera) are described from the nymphal stage from the Reserve Naturelle Integrate d'Andringitra. One species belongs to the genus Afronurus and the second to the genus Thalerosphyrus. Resume Deux nouvelles especes d' Heptageniidae (Insecta: Ephemeroptera) sont decrites a partir des larves de la Reserve Naturelle Integrate d'Andringitra. Une d'entre elles appartient au genre Afronurus, la seconde au genre Thalerosphyrus. Introduction Malagasy mayflies (Ephemeroptera) belonging to the family Heptageniidae are poorly known. The first report from the island is a larva identified as Notonurus sp. by Demoulin (1973). In the fol- lowing years Edmunds (1975, 1979) mentioned undescribed species from Madagascar belonging to genera Afronurus, Compsoneuriella, and Thal- erosphyrus. In these publications he drew a gen- eral outline of the biogeography of these taxa. Three new taxa from the Reserve Naturelle In- tegrate (RNI) d'Andringitra are discussed here. The nymphs of two forms are described as new species for which the imagos are unknown, and the nymph of a third species is briefly discussed but not formally named. Genus attribution is pro- visional and awaits the availability of adults, es- pecially male imagos. For details on collection techniques and stations see Chapter 9. In order to avoid any confusion, some terms are defined, as follows (Hubbard, 1995). Nymph: a larva that possesses black or extremely dark wing pads and is almost ready to molt to the subima- ginal stage; larva: any other immature stage youn- ger than the nymph; and pecten(s): "any comb- like structure or organ" (Torre-Bueno, 1989); we use this term to name the "comb-shaped struc- tures" that are found on the outer margin of the galea-lacinia of the maxillae (see Fig. 10-25). Descriptions Afronurus matitensis Sartori & Elouard, new species (Figs. 10-1 to 10-16) Nymph — Total length without caudal filaments: male nymph up to 9.5 mm; female nymph up to 11.0 mm. Overall coloration dark brown, pale brown head, dark brown thorax and abdomen; pale brown femora with a dark dorsal spot at mid- line and apex (Fig. 10-1); yellow-brown ventral side with the same spots as on dorsum; yellow- SARTORI & ELOUARD: NEW HEPTAGENIIDAE 121 ^ vT Figs. 10-1 through 10-4. Afronurus matitensis n. sp. 1: dorsal view of nymph; 2: ventral view of nymph; 3: details of head of a female nymph; 4: prothorax (arrows indicate chitinous creases). 122 FTELDIANA: ZOOLOGY brown tibiae, dark brown tarsi; gray-brown ster- nites with paired small pale spots on each sternite; gray-brown meso- and metasternite with a dark brown transverse stripe (Fig. 10-2). Purple and blackish abdominal gills with visible tracheation; yellow-brown caudal filaments with dark brown rings. Head — Very broad head capsule, broadest part at level of the insertion of antennae; female head with posterolateral margins rectilinear to slightly concave (Fig. 10-3); male characterized by a slightly rounded head. Trapezoidal labrum, about four times wider than long (3.75-4.25); internal margin with a well-marked medium notch; round- ed lateral projections slightly bent backward (Fig. 10-5). Mandibles with long sharp-pointed inci- sors, bearing six to eight small teeth in the internal margin. Asymmetrical prostheca, fine on the left mandible, thicker and slightly indented on the right one. Hypopharynx with lateral lobes round- ed and covered with long bristles up to the inner side (Fig. 10-6). Maxillae densely covered with scattered bristles (Fig. 10-8). Apex of galea-lacin- ia with about 14-16 pectens, fifth pecten with 1 1- 12 long sharp-pointed teeth. Maxillary palpi three-segmented: segment I very long, at least the length of galea-lacina, with 18-22 stout bristles on the external margin; segment II narrower and equal in length to segment I; and segment III about 2.5 times shorter than segment II, with the apex gradually narrowing to form a sharp-pointed tip (Fig. 10-8). Apical part of segment II and seg- ment III form a brush made of bristles on their outer margins. Rib-shaped glossae, rounded api- cally. Paraglossae moderately bent laterally (Fig. 10-7). Labial palpi three-segmented. Inner margin of segment HI markedly sinuous, apex of segment II and entire segment III covered with a brush structure similar to that of maxillary palpi (Fig. 10-9). Thorax — Pronotum laterally rounded, with two chitinous creases along symmetrical line, but with no posterolateral projections (Fig. 10-4). Legs thick with dilated femora. Tarsal claws with two small subapical teeth. Outer margin of femora covered with long and fine bristles; dorsal surface of all femora covered with thick short heart- shaped spines (Fig. 10-10), longest not exceeding 1.5 times width. Abdomen — No lateral projections on segment I, small projections on segment II through IV but increasing in length towards segment VIII, where they reach the middle of segment IX. Posterior margin of tergite V, with one to two rows of sharp microdenticles and a posterior row of stout spat- ulate scales with rounded apex (Fig. 10-11). Sev- en pairs of gills, gill VII without tracheal fila- ments (Figs. 10-12 to 10-16). Gills I elongated with subparallel margins and slightly acute apex; gills II-IV more rounded and slightly asymmetri- cal; gills VII elongated with rounded apex. Distal part of each segment of caudal filaments with a whorl of stout spines apically rounded, as well as long and fine bristles. Examined Material Holotype — One female nymph; Madagascar, Fianarantsoa Province, approximately 45 km S Ambalavao, RNI d'Andringitra, the Matitanana Basin, station 1, 717 m, 22°13'S, 47°01'E, 17 No- vember 1993. It is deposited in the Mus6e Can- tonal de Zoologie, Lausanne, Switzerland. The type locality is the original collection sample sta- tion number P0165 of the LRSAE field series, as- sociated with the "Biodiversity and Biotypology of Malagasy Rivers" program. Paratypes — One female nymph, one male nymph, five larvae, same data as for holotype; 14 larvae at station 5 (P0167), 19 November 1993; seven larvae at station 4 (P0168), 20 November 1993, RNI d'Andringitra (see Chapter 9); two lar- vae, Manampanihy River, 24°41'00"S, 46°53'39"E (P0091), 15 April 1992; one male nymph, Na- morona River, Ranomafana, Ifanadiana, 21°15'15"S, 47°27'34"E (P0212), 17 April 1994. All speci- mens collected by J.-M. Elouard, F-M. Gibon, M. Sartori, and others. Some paratypes deposited in the Museum National d'Histoire Naturelle (MNHN), Paris, and Centre National sur l'Envi- ronnement, Antananarivo. Other material not considered as para- types— Nine larvae, stream near Ambatolampy, near confluence with the Namorona River, 7 Au- gust 1958; two larvae, Amborompotsy River, An- tsampandrano (forest station), 25 July 1958; five larvae, Andranomalona River (forest station An- dasibe), along Antananarivo-Toamasina Road, 3 1 July 1958. Collected by F. Starmuhlner and J. Fontaine. Affinities Considering the morphology of its mouthparts, this new species probably belongs to the genus Afronurus. Afronurus matitensis resembles some SARTORI & ELOUARD: NEW HEPTAGENIIDAE 123 Figs. 10-5 through 10-11. Afronurus matitensis n. sp. 5: labrum; 6: hypopharynx; 7: glossae and paraglossae; 8: maxilla; 9: labial palp; 10: spines on dorsum of hind femora; 11: posterior margin of tergite V. 124 FIELDIANA: ZOOLOGY Figs. 10-12 through 10-16. Afronurus matitensis n. sp. 12: gill I; 13: gill III; 14: gill V; 15: gill VI; 16: gill VII. African species, such as A. oliffi Schoonbee and A. harrisoni Barnard, but it is easily distinguished by the shape of the gills (distinctly more asym- metrical in the African species), by the shape of the maxillary palpi (first segment shorter in A. oliffi and A. harrisoni), by the number of pectens of the galea-lacinia (more than 20 for the above two species), and by the shape of the spines on the dorsum of the femora (sharp-pointed for A. harrisoni; see Schoonbee, 1968). Thalerosphyrus josettae Sartori & Elouard, new species (Figs. 10-17 to 10-32) Nymph — Body length without caudal filaments: male nymph up to 9.5 mm, female nymph up to 10.5 mm. Overall coloration chestnut brown to pale brown; evenly colored head and thorax. Ab- dominal tergites with distinctively dark brown on yellowish brown ground (Fig. 10-17). Sternites also distinctive, yellowish beige, with paired elon- gated dark brown spots and paired points all along the median line, the complete structure lamp- shade shaped (Fig. 10-18). Evenly yellowish brown colored legs, apex of femora of nymphs with dark brown spots. Brown tibia and tarsi dark brown. Head — Labrum about four times wider than long, with nearly rectilinear outer margins and a wide median notch. Lateral projections slightly bent backward (Fig. 10-21). Hypopharynx with lateral lobes rounded and covered with long bris- tles that do not extend to the inner side. Mandibles with indented incisors. Asymmetrical prostheca. Galea-lacinia of maxillae covered with scattered long bristles (Fig. 10-24). Apex with 14-16 pec- tens, fifth with 10-11 thick sharp-pointed teeth (Fig. 10-25). Maxillary palpi three-segmented. Segment I shorter than galea-lacina, with 25-28 stout bristles on its outer margin. Segment II about 1 .2 times longer than the first. Segment III short, with a rounded apex (Fig. 10-24). Rib- shaped glossae, paraglossae moderately stretched toward lateral side (Fig. 10-23). Labial palpi three-segmented. Subrectilinear internal margin for third segment. Thorax — Prothorax markedly rounded on both sides with a beginning of rounded projections in latero-posterior part (Fig. 10-19). Prothoracic creases not heavily sclerotized. Well-developed SARTORI & ELOUARD: NEW HEPTAGENIIDAE 125 --• » 17 id Figs. 10-17 through 10-20. Thalerosphyrus josettae n. sp. 17: dorsal view of nymph; 18: ventral view of nymph; 19: prothorax (arrows indicate chitinous creases); 20: details of thoracic margin (arrows indicate supracoxal spurs). supracoxal spurs for meso- and metathorax (Fig. 10-20). Legs with stout tarsal claws generally with three sharp-pointed subapical teeth. Dorsum of hind femora with spines more than two times longer than wide, with diverging rounded to slightly opened apexes (Fig. 10-26). Abdomen — Lateral projections unmarked on segments I-VI, barely visible on segments VII and VIII. Posterior margin of tergite V presenting two to three rows of microdenticles as well as a row with teeth alternating with pyriform scales about 1.5 times longer than the teeth (Fig. 10-27). Seven pairs of gills, last pair without tracheal fil- aments. All gills are markedly asymmetrical and 126 FIELDIANA: ZOOLOGY Figs. 10-21 through 10-27. Thalerosphyrus josettae n. sp. 21: labrum; 22: hypopharynx; 23: labium; 24: maxilla; 25: median pectens of galea-lacinia; 26: spines on dorsum of hind femora; 27: posterior margin of tergite V. SARTORI & ELOUARD: NEW HEPTAGENIIDAE 127 Figs. 10-28 through 10-32. VII. Thalerosphyrus josettae n. sp. 28: gill I; 29: gill III; 30: gill V; 31: gill VI; 32: gill with sharp-pointed apexes, gills V most prominent (Figs. 10-28 to 10-32). Caudal filaments with whorls of stout spines at articulation of each seg- ment. Examined Material Holotype — One female nymph, Madagascar, Amborompotsy stream, Antsampandrano (forest station), 25 July 1958. Collected by F. Starmuhl- ner and J. Fontaine. Deposited in the Musee Can- tonal de Zoologie, Lausanne. Paratypes — Six larvae, Andringitra Massif, station 14 (P0178; see Chapter 9), 29 November 1993; two larvae, Fenoevo, on tributary of Man- ampanihy River, 24°41'00"S, 46°53'39"E (P0091), 15 April 1992. Some paratypes deposited in MNHN and Centre National sur PEnvironnement, Antananarivo. Affinities Two genera are known whose larvae possess well-developed supracoxal spurs: Compsoneuriel- la Ulmer (syn. Notonurus Crass fide Gillies, 1984) and Thalerosphyrus Eaton. On the basis of the shape of the pronotum, the glossae of the labium, and the arrangement of scattered bristles on the ventral side of the galea-lacinia, this new species probably belongs to the genus Thalerosphyrus, according to Jensen (1972). The 10 known species are all from the Oriental Realm, except one from the Afrotropical Realm. Three are known only at the imaginal stage. The remaining species are T. ethiopicus Soldan, 1977 (Sudan); T. determinatus (Walker, 1853), T. sinuosus (Navas, 1933), T. su- matranus (Ulmer, 1939) (all from the Sunda Is- lands); T. vietnamensis (Dang, 1967) (Vietnam); T. bishopi Braasch and Soldan, 1986 (Malaysia); and T flowersi Venkataraman and Sivaramakrish- 128 FIELDIANA: ZOOLOGY nan, 1987 (South India). Thalerosphyrus josettae can be distinguished from all of these species ex- cept T. ethiopicus and T. sinuosus by the lack of well-developed lateral projections on the abdo- men, as well as by the shape of the gills. Gill I is more slender in T. ethiopicus, T. determinants, and T. sumatranus. Gill VII is especially elongat- ed in T. sinuosus (see Ulmer, 1939, p. 668), whereas in T flowersi it does not possess an acu- minate apex. Thalerosphyrus species A Young larvae similar to T josettae were also captured in the RNI d'Andringitra. These larvae differ from T. josettae in the uniform whitish col- oration of abdominal tergites VI and VII. Apart from tergite coloration, the other main differences are in the maxillae. In Thalerosphyrus sp. A the first segment of the palpi has 15-17 bristles on its outer margin, and the galea-lacina crown bears 12-13 pectens, the fifth of which has 12-13 teeth. The remaining characteristics are similar to those of T. josettae, including pronotum shape, spines on the dorsum of femora, tarsal claws, and den- ticulation on posterior margin of tergite V. Because only young larvae of Thalerosphyrus sp. A were available for this study, it is possible that they represent younger stages of T josettae, and we refrain from giving them a specific epithet. that preferentially lives in the metarhitral zone. Thalerosphyrus josettae n.sp. was found at one station located in a zone of savannah on the west- ern slope of the massif at 1900 m. Water temper- ature at the time of sampling was at 19°C. Station physiognomy would imply that this species lives in small streams and would be characteristic of the epirthral zone. There are obvious ecological differences between these two species. A. matiten- sis lives in rivers of eastern humid forest, whereas T josettae appears to be restricted to warmer streams running through eastern coast savannah. In continental Africa, Afronurus is associated with mountain streams and turbulent waters (Schoon- bee, 1968; Gillies, 1984). Little is known about the ecological requirements of Thalerosphyrus larvae. They live in perennial streams or brooks where water flow is moderate, with shallow and warm water (Soldan, 1977, 1991; Braasch & Sol- dan, 1986; Venkataraman & Sivaramakrishnan, 1987). This information fits what we actually know about the ecology of T josettae. Neverthe- less, both species have been collected once at the same place outside Andringitra (Fenoevo, on trib- utary of Manampanihy River), suggesting that more data are needed before a precise typology can be drawn for these two species. Acknowledgments Examined Material Four larvae from station 14 (P0178; see Chap- ter 9), 29 November 1993; two larvae, Amborom- potsy stream, Antsampandrano (forest station), 25 July 1958. Collected by F. Starmuhlner and J. Fontaine. Deposited in the Museum Cantonal de Zoologie, Lausanne. Ecology of Heptageniidae from RNI d'Andringitra Afronurus matitensis n.sp. was found exclu- sively on the eastern slope of the Andringitra Massif, in primary and degraded forest at altitudes ranging between 715 and 750 m. Water tempera- tures at the time of sampling were 16°-18°C. On the basis of the stream order of the colonized riv- ers (medium to large), A. matitensis is a species We are deeply indebted to Prof. W L. Peters (Tallahassee, Florida) for his help in making avail- able to us some hard to obtain literature, as well as for his advice. Thanks are also due to L. Ruf- fieux (Lausanne) for her help and fruitful discus- sion. Literature Cited Braasch, D., and T. Soldan. 1986. Die Heptageniidae des Gombak River in Malaysia (Ephemeroptera). Rei- chenbachia, 24: 41-52. Dang, T. N. 1967. Nouveaux genres, nouvelles especes de la faune des invert£br£s des eaux douces et sua- matres du Nord Vietnam. Tap San Sinh Vat-dia Hoc VI, ser. 3,4: 155-165. Demoulin, G. 1973. Eph6me>opteres de Madagascar III. Bulletin de l'lnstitut Royal des Sciences Naturelles de Belgique, 49(7): 1-20. Edmunds, G. E, Jr. 1975. Phylogenetic Biogeography of Mayflies. Annals of the Missouri Botanical Garden, 62: 251-263. . 1979. Biogeographical relationships of the Ori- SARTORI & ELOUARD: NEW HEPTAGENIIDAE 129 ental and Ethiopian mayflies, pp. 11-14. In Pasternak, K., and R. Sowa, eds. Proceedings of the Second In- ternational Conference on Ephemeroptera. Polish Academy of Sciences, Krakow, 312 pp. Gillies, M. T. 1984. On the synonym of Notonurus Crass with Compsoneuriella Ulmer (Heptageniidae), pp. 21-25. In Landa, V., et al., eds. Proceedings of the IVth International Conference on Ephemeroptera. Czechoslovak Academy of Sciences, 345 pp. Hubbard, M. D. 1995. Towards a standard methodol- ogy for the description of mayflies (Ephemeroptera), pp. 361-369. In Corkum, L., and I. Ciborowski, eds. Current directions in research on Ephemeroptera. Ca- nadian Scholars' Press, Toronto, 478 pp. Jensen, S. L. 1972. A generic revision of the Hepta- geniidae of the world (Ephemeroptera). Ph.D. thesis, University of Utah (unpublished). Navas, L. 1993. Insecta Orientalia. XII. Memoire Pon- tificia Accademia delle Scienze Nouvi Lincei, 17: 75- 108. Schoonbee, H. J. 1968. A revision of the genus Afro- nurus Lestage in South Africa. Memoirs Entomolog- ical Society of Southern Africa, 10: 1-46 + 7 plates. Soldan, T 1977. Three new species of mayflies (Ephemeroptera) from the mist oasis of Erkwit, Su- dan. Acta Entomologica Bohemoslovaca, 74: 289- 294. 1 99 1 . An annotated list of mayflies (Ephem- eroptera) found in the Nam Cat Tien National Park, pp. 4-9. In Spitzer, K., et al., eds. Nam Cat Tien, Czechoslovak Vietnamese Expedition. November 1989, Research Report. Institute of Entomology, CAS, Ceske Budejovice, 45 pp. Torre-Bueno, J. R. de la. 1989. The Torre-Bueno glossary of entomology, revised edition. American Museum of Natural History, New York. Ulmer, G. 1939. Eintagsfliegen (Ephemeroptera) von den Sunda-Inseln. Archiv fur Hydrobiologie, Supple- ment, 16: 443-692. Venkataraman, K., and Sivaramakrishnan, K. G. 1987. A new species of Thalerosphyrus from South India (Ephemeroptera: Heptageniidae). Current Sci- ence, 56(21): 1126-1129. Walker, F. 1853. Catalogue of the specimens of neu- ropteroid insects in the collection of the British Mu- seum. London, 658 pp. 130 FIELDIANA: ZOOLOGY Chapter 11 Two New Species of Simulium (Diptera: Simuliidae) from the Reserve Naturelle Integrate d'Andringitra, Madagascar Jean-Marc Elouard, Theogene Pilaka, and Fabienne Ranaivoharindriaka Abstract Two new species of Simulium (Diptera: Simuliidae), Simulium metecontae and S. brunhesi, are described from the Reserve Naturelle Integrate d'Andringitra based on pupae. Resume Deux nouvelles especes de Simulies (Diptera: Simuliidae), Simulium metecontae et 5. brun- hesi, provenant de la Reserve Naturelle Integrate d'Andringitra sont decrites en se fondant sur les nymphes. Introduction Female Simulium (Diptera: Simuliidae) are he- matophagous and generally dependent on specific vertebrates for their feeding regime. Collection of females in the wild that are not anthropophagous is extremely rare. Male Simulium are floricolous, thus making capture infrequent. Further, adult Si- mulium are generally diurnal; therefore collection at night by light traps is rare and generally con- fined to cases when they are disturbed in their resting places. In contrast, collecting aquatic stages (larva and pupae) is relatively easy. For pupae, varying shapes of gills or respiratory filaments are the eas- iest and most frequently used characteristics for specific determination. Further, the number and shape of dorsal and ventral hooks (thoracic and abdominal), as well as the shape and structure of the cocoon, provide diagnostic characters. Corre- lation between pupae and imagos can generally only be established after pupae have been reared. Morphological characters used to determine the species of larvae are more difficult than those for pupae, and they are largely based on man- dibles, the post-genal bridge, and overall chae- totaxy. Correlation between these two stages is established by uncoiling seventh-stage larval gills, which often parallel the structures in the future pupae. Larvae at their last stages are not always found associated with pupae. For these reasons, Simulium is a genus for which system- atic characters are mainly based on pupal mor- phology. Two new Simulium species were discovered in November 1993 during the inventory of the Re- serve Naturelle Int6grale (RNI) d'Andringitra. One species was found in the humid forest zone on the eastern slope of the massif, whereas the other was discovered in streams running through the high plateau on the western slope. For infor- mation on collection localities, stations, and stream order, see Chapter 9. ELOUARD ET AL.: TWO NEW SPECIES OF SIMULIUM 131 Description Simulium metecontae Elouard & Pilaka, new species (Fig. 11-la-e) Holotype — Currently only known from the pu- pae. The cocoon is simple, without heel or pro- jection (Fig. 11-le). The overall weft of the co- coon is quite loose. Gills of pupae are made up of two filaments originating from a common stem and curving toward one another. Stem and gills together are lyre-shaped (Fig. 11- lc). Four finer filaments originate from the stem and gill com- plex. Two of these filaments make an apical tip, and the other two filaments branch between two- thirds and three-quarters of their length from the apex. The surface of the gill filament creases (Fig. 11-1 d) . Dorsal fangs are arranged in series of four or five (Fig. 11- la). Ventral fangs are thick but not numerous (Fig. 11-lb). Smete-1 (one tube containing the body of the pupa preserved in 70% alcohol and slide Smete- 1.1, with gills treated with Euparal) is designated as the holotype. It was collected in Fianarantsoa Province, approximately 45 km S Ambalavao, RNI d'Andringitra, station 4, on the Sahanivoraky River, at 735 m and 22°13'S, 47°01'E. It is de- posited at Museum National d'Histoire Naturelle (MNHN), Paris. The original collection sample station number is P0168 of the LRSAE collection field series (POxxx refers to sampled stations in the general CNRE/ORSTOM program— Biodiver- sity and Biotypology of Malagasy Waters). Paratypes — Station 1 (P0167), one pupa; sta- tion 9 (P0171), nine pupae; station 11 (P0175), one pupa. All were in the RNI d'Andringitra (see Chapter 9 for details of sites). Paratypes are de- posited in the Centre National de Recherches sur l'Environnement (Antananarivo), ORSTOM col- lection, Montpellier, MNHN, and the Field Mu- seum of Natural History (FMNH). Etymology — This species is dedicated to Mrs. Sophie Metecont, entomologist and friend. Biological notes — This species was found in small freshwater rivers (16°-18°C at time of sam- pling) at altitudes ranging from 500 to 1600 m. Pupae colonize herbaceous substrates in water. Pupae were located as single individuals; this fact may emphasize their relative rarity or their wide spatial distribution. Pupae of S. metecontae were always associated with other species, although this varied between stations. Sympatric species in- cluded Simulium gyas, S. pentaceros, and 5. iphias lOf. See Chapter 9 for further details on the ecology of this genus in the RNI d'Andrin- gitra. Simulium brunhesi Elouard & Ranaivoharindriaka, new species (Fig. ll-2a-f) Holotype — Currently only known from the pu- pae. The cocoon has no heel, but a structure pro- tecting the upper part of the head (Fig. 1 l-2e and 1 l-2f). The weft of the cocoon is quite fine. Pupa has four subequal gill filaments originating from a common stem (Fig. 11 -2c). Each of these branches terminates in a fine flagellated extremity. Cuticle of branchial arm is reticulate (Fig. 1 l-2d). These gill filaments have a discontinuous and ir- regular reticulation when magnified. Backside fangs are set on first abdominal segments (seg- ments 2-4) (Fig. ll-2a). The eighth abdominal segment has a pecten composed of fine fangs. In contrast, ventral fangs are set on median segments (Fig. 11 -2b). Sbrun-1 (one tube containing the body of the pupa preserved in 70% alcohol and slide Sbrun- 1.1, with gills treated with Euparal) is designated as the holotype. It was collected at station no. 14 at 1900 m (see Chapter 9) and is deposited at MNHN. The original collection sample station number is P0179 of the LRSAE field series. Paratypes — Zomandao River, effluent of the Mangoky Basin on the Andringitra Massif, station no. 14 (P0178), seven pupae; station no. 15 (P0179), two pupae, all collected in the RNI d'Andringitra (see Chapter 9 for details on col- lection stations). Paratypes deposited in Centre National de Recherches sur l'Environnement (An- tananarivo), ORSTOM collection, Montpellier, MNHN, and FMNH. Etymology — This species is dedicated to Dr. Jacques Brunhes, a researcher at ORSTOM. Taxonomic Notes — Several African Simulium species have gills consisting of four filaments. In Madagascar, S. tolongoinae has an arrangement similar to 5. brunhesi, with four gill filaments and a reticulated pattern. However, in S. tolongoinae the gill filaments become progressively narrower and are not made up of finer filaments, as in 5. brunhesi. Two African species, S. nigritarsis and S. katangae, have the same number and gill fila- ment arrangement, but without the flagellated tips. Biological Notes — Simulium brunhesi was found above 1800 m on the high plateau of the 132 FIELDIANA: ZOOLOGY 1 mm 1 mm Fig. 11-1. Pupae of Simulium metecontae. a: abdomen in dorsal view; b: abdomen in ventral view; c: gill; d: ornamentation of the gill; e: cocoon in lateral view. RNI d' Andringitra, near the source of the Zoman- dao River, Mangoky Basin. Waters in this area of the Andringitra Massif are warmer (19°-19.5°C at the time of sampling) at 1900 m compared to wa- ters on the eastern slopes (1 1°-16.5°C). At station 14, 5. brunhesi was associated with S. imerinae and with 5. iphias 8f; at station 15 it was asso- ciated only with S. iphias 8f. For more informa- ELOUARD ET AL.: TWO NEW SPECIES OF SIMULIUM 133 1 mm 0,05 mm Fig. 11-2. Pupae of Simulium brunhesi. a: abdomen in dorsal view; b: abdomen in ventral view; c: gill, d: ornamentation of the gill; e: cocoon with lateral view, f: cocoon in dorsal view. 134 FIELDIANA: ZOOLOGY tion on the ecology of RNI d'Andringitra Simu- lium see Chapter 9. Pupae of three Malagasy Simulium (5. mete- contae, S. brunhesi, and S. starmuhlneri) have thin gill filaments supported by thicker filaments. Simulium species belonging to the continental Af- rican Medusiforme group possess parallel fila- ment structure. However, in this group the number of thin filaments is greater and the overall shape of gills is very different from the three Malagasy species. It is still premature to discuss the evolu- tionary history of the three species, but the shape of fine secondary filaments is probably an impor- tant character to discern relationships. At this stage, it is not clear if these species belong to the same group or imago subgenus, or if the filament characters are convergent. Acknowledgments This study is part of the Biodiversity and Bio- typology of Malagasy Continental Waters (PEC 7) Program, jointly run by CNRE and ORSTOM. The program is financed through the Fonds d'Aide et de Cooperation (FAC) francais. We thank Steven Goodman for translating and cor- recting the manuscript. ELOUARD ET AL.: TWO NEW SPECIES OF SIMULIUM 135 Chapter 12 Parasitic and Commensal Arthropods of Some Birds and Mammals of the Reserve Naturelle Integrale d'Andringitra, Madagascar Barry M. OConnor Abstract Examination of 165 specimens representing 21 species of mammals and 25 specimens rep- resenting 16 species of birds collected in the Reserve Naturelle Integrale d'Andringitra yielded numerous collections of parasitic and commensal arthropods. All mammals and 23 of 25 birds harbored arthropod associates. Preliminary identifications of symbionts are presented primarily at the family and occasionally generic level for the Acari. Parasitic insects belonging to the orders Diptera, Siphonaptera, and Phthiraptera are also reported. The following families of mites are reported from native Madagascar vertebrates for the first time: Parasitiformes: Pa- chylaelapidae; Trombidiformes: Demodicidae, Psorergatidae, Ereynetidae, and Syringophilidae; and Sarcoptiformes: Turbinoptidae and Cytoditidae. Resume L'examen de 165 specimens appartenant a 21 especes de mammiferes et 25 specimens ap- partenant a 16 especes d'oiseaux collectes dans la Reserve Naturelle Integrale d'Andringitra a permis la collecte de nombreux arthropodes parasites et commensaux. Tous les mammiferes et 23 des 25 oiseaux collectes ont recele des arthropodes associes. L' identification preliminaire des organismes commensaux est presentee principalement par famille et par genre pour les acariens. Les insectes parasites appartenant a l'ordre des Diptera, Siphonaptera et Phthiraptera sont aussi repertories. Les families suivantes de mites font l'objet d'une premiere mention en qualite d' ectoparasites de vertebres endemiques de Madagascar: Parasitiformes: Pachylaelapidae; Trombidiformes: De- modicidae, Psorergatidae, Ereynetidae, Syringophilidae; Sarcoptiformes: Turbinoptidae, Cyto- ditidae. Introduction Madagascar harbors a unique assemblage of terrestrial vertebrates, which themselves harbor an extensive fauna of parasitic and commensal ar- thropods. Host groups with limited dispersal abil- ity (amphibians, most reptiles, and non-volant mammals) typically exhibit a high degree of en- demism, whereas many volant taxa (birds and bats) are more widely distributed. Study of the arthropod communities associated with these host groups is in its infancy. The present state of knowledge varies considerably among different taxa, with groups having known medical or vet- erinary importance having received the most in- tense study. The basic taxonomy (i.e., naming of species) in certain groups such as the fleas, ticks, and myobiid and atopomelid fur mites of mam- 136 FIELDIANA: ZOOLOGY mals has been worked out to a large extent (de Meillon, 1950; Lumaret, 1952; Hoogstraal, 1953; Fain, 1976, 1978; Uilenberg et al., 1980). Other diverse groups, such as the bloodsucking laelapid mites, chiggers, and most groups parasitizing birds and reptiles are almost completely unstud- ied. Uilenberg (1969) provided a listing of the parasites of domestic and laboratory animals in Madagascar, and Brygoo (1975) compiled a bib- liography of the vertebrate parasites of the region. Contemporary surveys of Madagascar's diverse vertebrate fauna can assist in expanding our knowledge of the associated parasite and com- mensal arthropod communities if host specimens can be preserved in a manner that preserves the associated fauna intact and uncontaminated. The present report provides a preliminary listing of parasitic and commensal arthropods recovered from mammal and bird specimens collected dur- ing the recent survey of the Reserve Naturelle In- tegrate (RNI) d'Andringitra. Methods A total of 165 mammals and 25 birds were ex- amined for arthropod associates. Specimens were collected by S. M. Goodman and other investi- gators during the survey of the RNI d'Andringitra in 1993. Specimens to be examined for parasites were kept separated from others during collecting, individually wrapped in cheesecloth, and fixed in formalin. The wrapped specimens were later transferred to 70% ethanol and loaned to my lab- oratory at the University of Michigan. There, each specimen was unwrapped and examined in detail using a dissecting microscope. Ectoparasites and commensal arthropods were removed with for- ceps and sorted by taxon and location on the host. The nasal passages of the hosts were flushed using a modification of the technique described by Yunker (1961) to remove upper respiratory en- doparasites. The process of preparing specimens on microscope slides for final identification is continuing, with a large number of previously un- described species expected. The following list is a preliminary report of the results of this sampling. Because final identifica- tions and descriptions of new taxa for most col- lections are still pending, I report here the para- site/commensal faunas for each host species, gen- erally at the level of family or occasionally genus or species. For each host species, the number of individuals examined is indicated following the species name, and the number of host individuals harboring a particular arthropod taxon is listed following the taxon name if more than one host was examined. Most of the following represent the first records of these taxa from the listed host species. Results Class Mammalia Order Insectivora Family Tenrecidae Microgale cowani (N = 16) Acari: Parasitiformes Ixodidae (9) Laelapidae (13) Acari: Acariformes Trombiculidae (16) Myobiidae Microgalobia sp. (7) Madamyobia sp. (5) Glycyphagidae (2) Atopomelidae Listrophoroides (Alistrophoroides) (12) Listrophoroides (Madlistrophoroides) (14) Microgale dobsoni (N = 1) Acari: Acariformes Trombiculidae Microgale gracilis (N = 1 ) Acari: Parasitiformes Ixodidae Laelapidae Acari: Acariformes Trombiculidae Glycyphagidae Atopomelidae Listrophoroides {Alistrophoroides) Listrophoroides {Madlistrophoroides) Microgale longicaudata (N = 5) Acari: Parasitiformes Ixodidae (1) Laelapidae (5) Acari: Acariformes Trombiculidae (4) Myobiidae (1) Atopomelidae (4) OCONNOR: PARASITIC AND COMMENSAL ARTHROPODS 137 Insecta: Siphonaptera undetermined family (1) Microgale melanorrhachis (N = 1) Acari: Acariformes Trombiculidae Myobiidae Microgalobia Madamyobia Atopomelidae Listrophoroides (Alistrophoroides) Listrophoroides (Madlistrophoroides) Microgale parvula (N = 7) Acari: Parasitiformes Ixodidae (6) Laelapidae (4) Acari: Acariformes Trombiculidae (7) Glycyphagidae (6) Atopomelidae Listrophoroides (Alistrophoroides) (6) Listrophoroides (Madlistrophoroides) (6) Microgale taiva (N = 18) Acari: Parasitiformes Ixodidae (13) Laelapidae (16) Acari: Acariformes Trombiculidae (18) Myobiidae Microgalobia (9) Madamyobia (2) Glycyphagidae (11) Atopomelidae Listrophoroides (Alistrophoroides) (14) Listrophoroides (Madlistrophoroides) (14) Microgale gymnorhyncha (N = 3) Acari: Parasitiformes Ixodidae (3) Laelapidae (1) Acari: Acariformes Trombiculidae (3) Myobiidae Microgalobia (3) Madamyobia (1) Glycyphagidae (1) Atopomelidae Listrophoroides (Alistrophoroides) (3) Listrophoroides (Madlistrophoroides) (2) Microgale sp. A (N = 2) Acari: Parasitiformes Ixodidae (2) Laelapidae (1) Acari: Acariformes Trombiculidae (2) Setifer setosus (N = 1) Acari: Parasitiformes Laelapidae Acari: Acariformes Trombiculidae Atopomelidae Tenrecobia pauliana (Lawrence, 1955) Order Chiroptera Family Vespertilionidae Miniopterus minor (N = 28) Acari: Parasitiformes Ixodidae (1) Spinturnicidae (24) Macronyssidae (28) Acari: Acariformes Trombiculidae (1) Myobiidae (22) Chirodiscidae (8) Sarcoptidae Nycteridocoptes sp Notoedres sp. (1) Insecta: Diptera Nycteribiidae (20) Streblidae Ascodipteron sp. (1) Myotis goudoti (N = 4) Acari: Parasitiformes Spinturnicidae (4) Macronyssidae (4) Acari: Acariformes Myobiidae (2) Demodicidae (1) Chirodiscidae (2) Sarcoptidae (1) Insecta: Diptera Nycteribiidae (4) (3) Order Carnivora Family Viverridae Galidia elegans (N = 2) Acari: Parasitiformes Ixodidae Ixodes sp. (2) Haemaphysalis sp. (2) Laelapidae (2) 138 FTELDIANA: ZOOLOGY Acari: Acariformes Trombiculidae (1) Atopomelidae (2) Order Rodentia Family Muridae Eliurus majori (N = 12) Acari: Parasitiformes Ixodidae (3) Laelapidae (10) Pachylaelapidae (1) Acari: Acariformes Trombiculidae (12) Psorergatidae Psorergates (2) Sarcoptidae (3) Atopomelidae Listrophoroides (Eulistrophoroides) (7) Listrophoroides (Pallistrophoroides) (7) Insecta: Phthiraptera: Anoplura unidentified family (12) Insecta: Siphonaptera unidentified family (1) Eliurus minor (N = 10) Acari: Parasitiformes Laelapidae (5) Acari: Acariformes Trombiculidae (10) Psorergatidae Psorergates (1) Glycyphagidae (4) Atopomelidae Listrophoroides (Eulistrophoroides) (4) Listrophoroides (Pallistrophoroides) (5) Insecta: Phthiraptera: Anoplura unidentified family (4) Eliurus tanala (N = 16) Acari: Parasitiformes Ixodidae (5) Laelapidae (10) Pachylaelapidae (3) Acari: Acariformes Trombiculidae (16) Demodicidae: Demodex (1) Ereynetidae (1) Glycyphagidae (12) Atopomelidae Listrophoroides (Eulistrophoroides) (8) Listrophoroides (Pallistrophoroides) (16) Insecta: Phthiraptera: Anoplura unidentified family (4) Insecta: Siphonaptera unidentified family (4) Eliurus webbi (N = 12) Acari: Parasitiformes Laelapidae (12) Pachylaelapidae (1) Acari: Acariformes Trombiculidae (12) Demodicidae Demodex (3) Glycyphagidae (1) Atopomelidae Listrophoroides (Eulistrophoroides) (5) Listrophoroides (Pallistrophoroides) (8) Insecta: Phthiraptera: Anoplura unidentified family (2) Insecta: Siphonaptera unidentified family (1) Gymnuromys roberti (N = 2) Acari: Parasitiformes Ixodidae (1) Laelapidae (1) Acari: Acariformes Trombiculidae (2) Glycyphagidae (1) Atopomelidae Listrophoroides (Pallistrophoroides) (2) Insecta: Phthiraptera: Anoplura unidentified family (2) Monticolomys koopmani (N = 2) Acari: Acariformes Trombiculidae (2) Sarcoptidae (1) Nesomys rufus (N = 15) Acari: Parasitiformes Ixodidae (3) Laelapidae (9) Pachylaelapidae (2) Acari: Acariformes Trombiculidae (15) Ereynetidae (2) Atopomelidae Listrophoroides (Eulistrophoroides) (3) Listrophoroides (Pallistrophoroides) (14) Insecta: Phthiraptera: Anoplura unidentified family (13) Insecta: Siphonaptera unidentified family (1) OCONNOR: PARASITIC AND COMMENSAL ARTHROPODS 139 Rattus rattus (N = 9) Acari: Parasitiformes Ixodidae (1) Laelapidae Laelaps nuttalli Hirst, 1915 (2) Echinolaelaps echidninus (Berlese, 1887) (1) Acari: Acariformes Trombiculidae (9) Ereynetidae (2) Myobiidae Radfordia ensifera (Poppe, 1896) (1) Glycyphagidae (3) Listrophoridae: Afrolistrophorus sp. (8) Insecta: Phthiraptera: Anoplura Polyplacidae Polyplax spinulosa (Burmeister, 1839) (3) Analgidae (5) Proctophyllodidae (4) Avenzoariidae (3) Trouessartiidae (5) Insecta: Phthiraptera: Mallophaga family undetermined (1) Phyllastrephus cinereiceps (N = 1) Acari: Acariformes Analgidae Proctophyllodidae Avenzoariidae Trouessartiidae Insecta: Phthiraptera: Mallophaga family undetermined Family Turdidae Class Aves Order Passeriformes Family Eurylaimidae Neodrepanis coruscans (N = 1) Acari: Parasitiformes Rhinonyssidae Acari: Acariformes Proctophyllodidae Trouessartiidae Neodrepanis hypoxantha (N = 2) Acari: Parasitiformes Rhinonyssidae (1) Acari: Acariformes Proctophyllodidae (2) Trouessartiidae (2) Family Pycnonotidae Phyllastrephus madagascariensis (N = 1) Acari: Acariformes Analgidae Proctophyllodidae Trouessartiidae Insecta: Phthiraptera: Mallophaga family undetermined Phyllastrephus zoster ops (N = 5) Acari: Parasitiformes Rhinonyssidae (1) Acari: Acariformes Ereynetidae (1) Copsychus albospecularis (N = 2) Acari: Parasitiformes Rhinonyssidae (1) Acari: Acariformes Trombiculidae (1) Analgidae (2) Proctophyllodidae (2) Avenzoariidae (1) Trouessartiidae (2) Epidermoptidae (1) Family Sylviidae Nesillas typica (N = 1) Acari: Parasitiformes Ixodidae Acari: Acariformes Analgidae Avenzoariidae Trouessartiidae Newtonia amphichroa (N = 1) Acari: Parasitiformes Rhinonyssidae Acari: Acariformes Avenzoariidae Cytoditidae Cryptosylvicola randrianosoloi (N = 1) no parasites observed Family Monarchidae Terpsiphone mutata (N = 1) Acari: Parasitiformes 140 FIELDIANA: ZOOLOGY Rhinonyssidae Acari: Acariformes Analgidae Proctophyllodidae Trouessartiidae Cytoditidae Insecta: Phthiraptera: Mallophaga family undetermined Family Timaliidae = 1) Oxylabes madagascariensis (N Acari: Parasitiformes Ixodidae Acari: Acariformes Trombiculidae Analgidae Proctophyllodidae Avenzoariidae Trouessartiidae Turbinoptidae Insecta: Phthiraptera: Mallophaga family undetermined Family Nectariniidae Nectarinia souimanga (N = 1) Acari: Acariformes Proctophyllodidae Family Zosteropidae Zosterops maderaspatana (N = 1) Acari: Parasitiformes Rhinonyssidae Acari: Acariformes Ereynetidae Syringophilidae Analgidae Proctophyllodidae Avenzoariidae Family Vangidae Hypositta corallirostris (N = 1) no parasites observed Family Ploceidae Ploceus nelicourvi (N = 1 ) Acari: Acariformes Analgidae Proctophyllodidae Trouessartiidae Insecta: Phthiraptera: Mallophaga family undetermined Foudia omissa (N = 4) Acari: Parasitiformes Ixodidae (1) Acari: Acariformes Analgidae (3) Proctophyllodidae (1) Avenzoariidae (3) Trouessartiidae (2) Literature Cited Brygoo, E.-R. 1975. Bibliographic de la Zoologie Par- asitaire des vertebras de Madagascar. Archives de l'lnstitut Pasteur de Madagascar, 44: 245-288. de Meillon, B. 1950. The Madagascar Siphonaptera. M6moires de l'lnstitut des Sciences de Madagascar, Sene A, 4: 67-73. Fain, A. 1976. Arachnides Acariens, Astigmata Listro- phoroidea. Faune de Madagascar, 42: 1-131. . 1978. Les Myobiidae d'Afrique au sud du Sa- hara et de Madagascar Acarina-Prostigmata. Annales du Musee Royal de l'Afrique Centrale, Serie 8°, Sci- ences Zoologiques, 224: 1-186. Hoogstraal, H. 1953. Ticks (Ixodoidea) of the Mala- gasy faunal region excepting the Seychelles; their or- igin and host-relationships; with description of five new Haemaphysalis species. Bulletin of the Museum of Comparative Zoology, Harvard University, 111: 37-113. Lumaret, R. 1952. Insectes Siphonapteres. Faune de Madagascar, 15: 1-107. Uilenberg, G. 1969. Inventaire artropodes, protozoai- res et rickettsiales parasites des animaux domestiques et des animaux de laboratoire a Madagascar. Archives de l'lnstitut Pasteur de Madagascar, 38: 69-105. Uilenberg, G., H. Hoogstraal, and J.-M. Klein. 1980. Les tiques (Ixodoidea) de Madagascar et ler rdle vec- teur. Archives de l'lnstitut Pasteur de Madagascar, nu- mero special, 1-153. Yunker, C. E. 1961. A sampling technique for intra- nasal chiggers (Trombiculidae). Journal of Parasitol- ogy, 47: 720. OCONNOR: PARASITIC AND COMMENSAL ARTHROPODS 141 Chapter 13 Blood Parasites from Birds in the Reserve Naturelle Integrate d'Andringitra, Madagascar Ellis C. Greiner, Michael S. Putnam, and Steven M. Goodman Abstract Six of the 10 blood smears collected from birds in the Reserve Naturelle Integrale d'An- dringitra showed blood parasite infection. Resume Sur dix frottis sanguins realises a partir d'echantillons collectes sur des oiseaux de la Reserve Naturelle Integrale d'Andringitra, six ont montre des infections sanguines parasitaires. Introduction There has been only one previous report on avi- an hematozoa from birds on Madagascar (Bennett & Blancou, 1974). This note reports the results of a recent small sample from the Reserve Naturelle Integrale (RNI) d'Andringitra, southeastern Mad- agascar. Methods During the netting operation conducted in the RNI d'Andringitra (Chapter 18), blood smears were obtained from six birds collected as speci- mens and from clipped claws of four others that were released. Smears were fixed in the field with methanol and prepared in the laboratory with Giemsa stain. Results Below we present the results from the RNI d'Andringitra survey. Field Museum of Natural History (FMNH) and field catalog numbers of S.M.G. (SMG) are also given for each bird spec- imen. Otus rutilus (FMNH 363793, SMG 6373)— 810 m, 6 October 1993, with Haemoproteus. Philepitta castanea (FMNH 363798, SMG 6368)— 810 m, 6 October 1993, with Leuco- cytozoon and microfilariae. Phedina borbonica — 720 m, 24 September 1993, no parasites found. Motacilla madagascariensis (FMNH 363815, SMG 6340)— 720 m, 25 September 1993, no parasites found. Phyllastrephus madagascariensis (FMNH 363819, SMG 6340)— 720 m, 25 September 1993, no parasites found. Terpsiphone mutata (rufous male) — 720 m, 26 September 1993, single schizont of Plasmodi- um. Terpsiphone mutata (rufous male) — 720 m, 26 September 1993, Haemoproteus or Plasmodi- um. Vanga curvirostris (FMNH 363831, SMG 142 FIELDIANA: ZOOLOGY 6397)— 810 m, 10 October 1993, Leucocyto- zoon and microfilariae. Ploceus nelicourvi (male) — 720 m, 26 September 1993, no parasites found. Foudia omissa (FMNH 363869, SMG 6396)— 810 m, 10 October 1993, Leucocytozoon and Plasmodium. Discussion The following blood parasites were found in six of 10 individuals (60%): Haemoproteus, Leuco- cytozoon, Plasmodium, and microfilariae nema- todes. This prevalence is higher than that found by Bennett and Blancou (1974) on Madagascar, where they identified parasites in 14 of 64 (22%) birds sampled. The present study also contrasts with that of Bennett and Blancou (1974) in that Haemoproteus, as well as multiple infections, were found for the first time on Madagascar. Two studies of avian hematozoa on the Mascarene Is- lands found Leucocytozoon, Plasmodium, multi- ple infections, and combined prevalences of 40% on Mauritius, 54% on Reunion, and 29% on Ro- drigues. Haemoproteus was found only in the in- troduced Rock Dove (Columba livid), whereas microfilariae were absent (Peirce et al., 1977; Peirce, 1979). Twenty-two percent of the land birds sampled on Aldabra were infected with Haemoproteus, Plasmodium, or both, with micro- filariae found in wading birds (Lowery, 1971). The results reviewed above therefore suggest that blood parasites are not uncommon in land birds of the western Indian Ocean. Species of the families Culicidae, Simuliidae, and Ceratopogonidae, that elsewhere contain vec- tors of avian hematozoa, are known from Mada- gascar (Paulian, 1961), although their role in avi- an hematozoan transmission is unknown. Nor is the pathogenicity of the parasites found in this small survey known. Literature Cited Bennett, G. E, and J. Blancou. 1974. A note on the blood parasites of some birds from the Republic of Madagascar. Journal of Wildlife Diseases, 10: 239- 240. Lowery, R. S. 1971. Blood parasites of vertebrates on Aldabra. Philosophical Transactions of the Royal So- ciety London B, 260: 577-580. Paulian, R. 1961. Faune de Madagascar. 13. La zo- ogeographie de Madagascar et des iles voisines. Office de la Recherche Scientifique, Paris, 484 pp. Peirce, M. A. 1979. Some additional observations of haematozoa of birds in the Mascarene Islands. Bul- letin of the British Ornithologists' Club, 99: 68-71. Peirce, M. A., A. S. Cheke, and R. A. Cheke. 1977. A survey of blood parasites of birds in the Mascarene Islands, Indian Ocean, (with descriptions of two new species and taxonomic discussion by M. A. Peirce). Ibis, 119: 451-461. GREINER ET AL.: BLOOD PARASITES FROM BIRDS 143 Chapter 14 Elevational Variation in Soil Macroinvertebrates on the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar Steven M. Goodman, Philip P. Parrillo, Sam James, and Petra Sierwald Abstract A series of soil plots in the Reserve Naturelle Integrate d'Andringitra were sampled in three microhabitats (ridge, slope, and valley) within the 720, 810, 1210, and 1625 m elevational zones to assess variation within and between these sites in the soil macroinvertebrates. Of the 166 morphospecies identified from the samples, 126 (76%) were restricted to a single elevation, 21 (13%) occurred at two adjacent elevations, and 19 (11%) were identified from three or four zones. Within each altitudinal zone there was no clear relationship found between the three microhabitats and the density or diversity of soil macroinvertebrates. In general the overall species richness and density were lower than those found at other sites in the tropics, presum- ably due to sampling technique. Two sites at approximately the same elevation but with dif- ferences in the level of human-induced habitat degradation showed no marked differences in the soil macroinvertebrates. Total taxonomic diversity peaked at 1210 m, whereas the total number of individuals per plot increased steadily between the 720 and 1625 m sites. Thus, species richness (as measured by taxonomic diversity) and density (as measured by the number of individuals per plot) do not exhibit precisely the same altitudinal pattern. Resume Une serie de quadrats de sol a ete echantillonnee dans trois microhabitats (crete, versant et vallee) au sein des zones altitudinales suivantes: 720, 810, 1210, et 1625 m, arm d'evaluer les variations de la macrofaune endogee des trois differents sites. En general, la richesse et la densite globale des especes etaient inferieures a celles trouvees au sein d'autres sites tropicaux. Quelques taxons presentent des distributions altitudinales lim- itees, meme si la majorite des especes echantillonnees de facon satisfaisante presente de grandes variations altitudinales dans leur distribution. Au sein de chaque zone d' altitude, la plus haute densite de macrofaune endogee se trouve dans les quadrats localises dans les vallees. La ma- crofaune endogee ne presente pas de difference nette au sein des sites d' altitude equivalente, mais seulement au sein des habitats affectes par des niveaux variables de degradation d'origine anthropique. La diversite taxinomique globale semble atteindre un maximum a 1210 m, tandis que le nombre total d'individus par quadrat augmente graduellement pour les sites localises entre 810 et 1625 m. Ainsi, la richesse specifique mesuree d'apres la diversite taxinomique, et la densite mesuree d'apres le nombre des individus par quadrat, ne montrent pas la meme tendance generate. 144 FIELDIANA: ZOOLOGY Introduction A significant proportion of the animal biomass and species richness in tropical forests is soil and litter invertebrates (Stork, 1988). However, little is known about the density, diversity, and eleva- tional distribution of these invertebrates in the tropics. Olson (1994) studied leaf litter ants, spi- ders, and beetles along an elevational gradient in Panama between 300 and 2020 m and found little evidence of fine-grain elevational zonation in the insect assemblages. The mean elevational range of most invertebrate species in his sample was 500 m, and about one-half of the species found in an elevational zone did not occur at 500 m in elevation above or below the zone. There was a mid-elevational peak in species richness at about 800 m elevation. This is in contrast to the study of Collins et al. (1984) in the Gunung Mulu Na- tional Park, Sarawak, where soil macroinverte- brates declined with increasing altitude, from 2,579 individuals per square meter at 130 m ele- vation to 145 individuals per square meter at 2,376 m elevation. However, total invertebrate biomass does not necessarily parallel species rich- ness. In the Volcan Barva, Costa Rica, there are significant differences in litter invertebrates and soil invertebrates occurring in the first 15 cm of soil (Atkin & Proctor, 1988). There are also dra- matic seasonal differences in the density and spe- cies richness of soil invertebrates (Chiba et al., 1975). In short, although the above comparisons are of different measures of two different micro- habitats in geographically isolated regions of the world, they confirm that few generalities can be made about soil and litter macroinvertebrates in tropical forests, particularly along an elevational transect. Confounding variables such as sampling techniques and seasonal variation further compli- cate interpretation. Moreover, soil and litter in- vertebrate densities and species richness are pre- sumably site or region specific, and they probably hold the key to understanding the distribution of numerous other organisms, such as insects, that feed extensively on macroinvertebrates. We are unaware of any studies of the soil mac- roinvertebrates on Madagascar. In this chapter we examine the distribution and species richness of the invertebrate macrofauna in three microhabitats (ridge, slope, and valley) within and between four elevational transects. Two of the transects were at approximately the same altitude, but one (720 m) was in a forest degraded by human activity, whereas the other (810 m) was in relatively intact forest. The other two transects (1210 and 1625 m) were in pristine forest (see Chapter 2 for a de- scription of the sites). In a subsequent chapter (Chapter 20), the data on soil macroinvertebrates are used to examine ecological correlates with the altitudinal distribution and species richness of In- sectivora. Methods In each elevational zone, pitfall buckets were set up to capture small mammals, reptiles, and amphibians (see Chapters 17 and 20). After the pitfall lines had been in place for between 6 and 8 days, a series of soil plots were set up adjacent to each line. Five plots were sampled per pitfall line during daylight hours. Each plot measured 2 m (length) X 2 m (width) X 100 mm (depth). Plots within a line were 20 m apart. The leaf litter above each plot was discarded, and the underlying soil was placed in rice sacks. Within 1 hour the soil was examined by hand, and all visible inver- tebrates were removed and preserved in alcohol for later sorting and determination. Earthworms were fixed in 12% formalin and subsequently transferred to alcohol. P. P. P. was responsible for all of the invertebrate identifications, with the ex- ception of Annelida and Arachnida; these were done by S.J. and PS., respectively. The level of taxonomic determination varied among the mac- roinvertebrate groups. Even though some organ- isms were identified only to order or family, most could be assigned to "morphospecies." Our des- ignation of species richness is a tabulation of the number of morphospecies. Voucher specimens are deposited at the Field Museum of Natural History (FMNH). Results A general summary of the macroinvertebrates recovered from each of the plots per pitfall line is presented in Table 14-1. Nematomorpha Two individuals of one morphospecies were re- covered from one plot at 1625 m. GOODMAN ET AL.: ELEVATIONAL VARIATION IN SOIL MACROINVERTEBRATES 145 Table 14-1. Summary of invertebrates recovered from soil plots along pitfall lines. Pitfall 720 m 810 m 1210 m 1625 m line 1 2 3 4 5 6 7 8 9 10 11 12 placement* R S V R S V R V S V R S Nematomorpha 1 1/2 Gastropoda Subulinidae 1 1/1 Annelida ~5 1/3 2/3 3/5 3/6 2/5 5/5 2/6 Arachnida Aranea Miturgidae 1 1/1 1/1 1/3 1/1 1/3 Amaurobidae 1 1/1 Dipluridae 1 1/2 Mygalomorpha 1 1/31 1/28 Scorpionidae Buthidae 1 1/1 Chilopoda Geophilomorpha 11 1/1 1/1 1/1 1/1 2/3 1/3 2/9 3/19 4/11 5/22 Scolopendromorpha 5 1/5 1/1 1/2 2/5 2/5 3/6 2/3 1/5 1/1 1/8 3/14 Lithobiomorpha 1 1/2 1/1 1/2 1/1 1/1 1/1 1/1 Scutigeromorpha 1 1/1 1/1 Crustacea Isopoda 6 1/4 1/1 2/4 2/6 1/2 2/4 1/20 1/1 2/8 1/1 Diplopoda 5 1/1 1/1 1/4 1/2 1/2 3/11 4/9 1/4 2/2 3/10 Diplura Diplura Heterojapygidae 1 1/2 1/2 1/1 1/4 1/2 1/2 Japygidae 1 1/2 1/1 1/3 1/4 1/1 1/3 1/6 1/15 Insecta Blattodea Blaberidae 5 1/1 1/1 2/2 2/2 1/1 1/1 1/1 1/1 Coleoptera 10 1/1 1/1 1/1 5/5 1/1 1/1 Anthribidae 1 1/1 Carabidae 4 1/1 1/4 3/4 1/1 Cerambycidae 1 1/1 Dermestidae 1 1/1 Elateridae 21 2/3 6/6 3/3 2/2 2/3 2/5 2/2 2/2 3/4 3/3 Scarabaeidae 38 1/1 1/1 4/8 8/10 7/7 4/12 4/5 8/35 4/19 2/2 2/2 1/2 Staphylinidae 4 2/2 2/4 Tenebrionidae 12 2/12 2/4 1/2 4/4 4/5 3/4 4/18 1/4 2/4 1/1 Dermaptera 1 1/1 Labiidae 2 1/1 2/3 1/1 Diptera 12 2/2 1/1 1/1 1/1 2/2 3/3 1/1 2/2 1/1 1/1 Hemiptera Cicadidae 1 1/2 1/3 1/2 1/1 Cydnidae 1 1/1 Enicocephalidae 1 1/1 1/2 1/1 Hymenoptera 3 1/1 1/1 1/1 Formicidae 1 1/1 Isoptera Termitidae 2 1/8 1/10 1/1 1/1 2/4 Lepidoptera 2 1/1 1/1 Orthoptera Gryllacrididae 1 1/2 1/1 Gryllidae 1 1/2 1/1 1/1 Gryllotalpidae 1 1/1 146 FIELDIANA: ZOOLCXT' Table 14-1. Continued. Pitfall 720 m 810 m 1210 m 1625 m line placement* 1 R 2 S 3 V 4 R 5 S 6 V 7 R 8 V 9 S 10 V 11 R 12 S Phasmatodea 1 1/1 Thysanura Nicoletiidae 1 1/14 1/3 1/3 1/13 1/3 1/11 1/1 1/2 1/5 1/3 1/7 Total taxonomic diversityt per line 8.4 5-12 2.97 4.4 0-10 4.39 5.6 3-11 3.21 10.6 7-14 2.51 8.6 4-12 2.97 2.6 0-7 2.70 8.8 4-13 3.70 7.2 2-15 5.06 8.0 5-12 3.31 7.0 3-10 2.92 6.8 4-10 2.39 8.8 5-16 4.55 per elevational zone 6.1 0-12 3.74 7.3 1-14 4.33 8.7 2-15 3.98 7.5 3-16 3.29 Total individuals per line 13.6 6-22 6.35 5.0 0-12 5.09 8.6 3-16 6.77 14.2 8-24 6.18 9.4 4-13 3.51 4.8 0-14 5.45 11.6 5-20 6.27 16.8 2-35 13.08 16.0 7-34 11.90 16.8 3-51 19.63 10.4 5-14 4.09 24.8 5-43 17.53 per elevational zone 9.1 0-22 6.73 9.5 0-24 6.22 14.9 2-35 10.32 16.7 3-51 14.87 * R = ridge; S = slope; V = valley. Number after name is total number of morphospecies identified in combined samples. Data are presented as total number of morphospecies/total number of individuals per line. t Descriptive statistics are presented as mean, range (minimum-maximum), and standard deviation per line. In each line there are five different plots. Gastropoda The only record from the soil samples is one individual at 720 m. er percentage of the biomass, because A. diffrin- gens is much larger. Arachnida Annelida Earthworms were recovered from all elevation- al zones, but not from each plot. The highest den- sity of earthworms was recorded in the 810 and 1210 m zones. Five was the maximum number of morphospecies recorded along any pitfall line (1210 m, slope). Only one of three lines at 1625 m yielded earthworms. In general, at 810, 1210, and 1625 m the greatest density of earthworms was in valley bottom plots (Table 14-1). In the 810 m zone the majority of collected earthworms belonged to one introduced species, Amynthas diffringens, a native of eastern Asia. This species was not recorded in the other three transect zones. Amynthas diffringens has been widely transported throughout the world and is now present in many temperate and tropical areas (Gates, 1972). Madagascar native species were present in the 810 m transect, but they represented only about 10% of the total individuals and a low- The only spider specimens represented in the soil samples by adults (two females) belonged to the Malagasy genus Uduba Simon, 1980 (Family Miturgidae). Presently, the genus contains two species, U. dahli and U. madagascariensis, and it is currently under revision by C. E. Griswold (pers. comm.). Both adult females belong to the same morphospecies and may be conspecific with Uduba spp. as figured by Griswold (1993, figs. 15 and 16). The subadult Uduba specimens are cer- tainly congeneric; whether they are conspecific cannot be determined at this point. Members of Uduba apparently occur at least over wide por- tions of the eastern humid forest and have been collected from Pare National de Ranomafana, about 120 km northeast of the RNI d'Andringitra (Griswold, 1993:7). Uduba are represented in samples at 725, 810, and 1625 m, and they occur in both pristine forest and regions with human dis- turbance. The low number of spiders recovered from the GOODMAN ET AL.: ELEVATIONAL VARIATION IN SOIL MACROIN VERTEBRATES 147 soil samples appears to be a result of the collect- ing method, because only soil below the leaf litter was examined. According to C. E. Griswold (pers. coram.), Uduba live in subterranean burrows. Chilopoda Eighteen species of Chilopoda were recovered from the soil samples. Of these, the Geophilo- morpha were the most dominant, representing 1 1 species or 61% of the centipede fauna. Diversity was lowest at the two lower elevations, with sin- gle species recorded at each of the 720 and 810 m sites. Within the Geophilomorpha there were positive correlations between elevation and mor- phospecies diversity (df = 3, r2 = 0.99, P - 0.007) and elevation and density (df = 3, r2 = 0.92, P - 0.04). The Scolopendromorpha repre- sented 28% of the recovered chilopods, with in- creasing density as a function of elevation (df = 3, r2 = 0.97, P = 0.01). The Lithobiomorpha and Scutigeromorpha (Scutigeridae) were each repre- sented by a single species, and both groups had broad altitudinal distributions but low densities. Overall, there was a strong positive correlation for the Chilopoda between elevation and diversity (df = 3, r2 = 0.99, P = 0.001) as well as between elevation and density (df = 3, r2 = 0.92, P = 0.04). Seven species (39%) were found at a single site. On the basis of these samples (Table 14-1), Chilopoda showed no significant preference for ridges, slopes, or valleys. Crustacea-Isopoda Six morphospecies were recovered from the plots, and density peaked at 1210 m. Isopoda showed a gradual replacement of species from 720 to 1625 m. Three species were found only along a single line. No correlation was found be- tween diversity or density and microhabitat. Diplopoda Five morphospecies were identified, three of which showed broad altitudinal distributions, oc- cupying at least three elevational zones. Two spe- cies were recorded only at single plots. Diversity and density peaked at 1210 m, with five morpho- species and 22 individuals, respectively. No rela- tionship was found between microhabitat and den- sity. Diplura Two families were identified in the soil sam- ples. The Heterojapygidae were restricted to the lowland forest below 810 m elevation, whereas the Japygidae were distributed across the com- plete elevational range sampled and were distinct- ly more common in the 1625 m zone. Insecta In general, a wide variety of insects, many of which were rare (e.g., Hemiptera: Cicadidae, Cyd- nidae, and Enicocephalidae; Lepidoptera; Orthop- tera; Phasmatodea; and Hymenoptera), were iden- tified from the soil samples. Thysanura — A single morphospecies was col- lected in the soil samples along the complete al- titudinal gradient. Densities were relatively con- stant at each elevation. Ridge microhabitats seemed to support greater populations at 720, 8 1 0, and 1210 m than slopes or valleys. Blattodea — Five morphospecies were recov- ered from the soil samples. Three were found with distributions ranging across three elevational sites, and two were found only at single sites. Roaches occurred in low densities in all transects, reaching a peak in both diversity and density at 810 m. No recognizable association with microhabitat was found. Dermaptera — Three morphospecies were iden- tified from the transect samples. Two were found only at 810 m and the third at 720 m. Coleoptera — Beetles represented the majority of the species diversity from the soil samples, composing 55% of the taxa recovered. Of these, 77% were Elateridae, Scarabaeidae, and Tenebri- onidae. The remaining families consisted of Anthribidae (1 morphospecies), Carabidae (4), Staphylinidae (4), Dermestidae (1), Cerambycidae (1), and 10 specimens that could not be deter- mined to family. Nineteen (90%) of these 21 mor- phospecies were found only along single lines. No clear microhabitat specificity was found for any beetle group. Carabids showed maximum diver- sity and density at 1210 m. Staphylinidae were not present above 810 m. Twenty-one morpho- species of Elateridae were identified from the samples, 17 (81%) were found at single sites, and 148 FIELDIANA: ZOOLOGY species diversity and density were evenly distrib- uted across the transect. The most abundant beetle family was the Scarabaeidae, with 38 morpho- species identified from the soil samples. This group showed elevational specificity, with 32 (84%) of the taxa found within a single line. Scar- abaeid density peaked at 1210 m. Twelve mor- phospecies of Tenebrionidae were identified from the samples. Four of these were collected from one line, and this group reached maximum density at 1210 m. Diptera — Within the soil samples, Diptera were not common. Of the 1 2 morphospecies iden- tified, 1 1 (92%) were found at single sites. Discussion The maximum density of earthworms recorded in the RNI d'Andringitra soil samples was seven individuals per square meter at 810 m, substan- tially less than at many other tropical forest sites. For example, densities (individuals/m2) have been reported as high as 401 in Volcdn Barva, Costa Rica (Atkin & Proctor, 1988), and 121 at Chajul, Mexico (Fragoso & Lavelle, 1987). In general. New World tropical forest sites tend to have a higher density and biomass of earthworms than Old World sites, and islands tend to have a lower density and biomass than continental sites (Fra- goso & Lavelle, 1992). Furthermore, there is con- siderable variation in density and species richness along elevational transects. Atkin and Proctor (1988) sampled earthworms in Costa Rica every 500 m between 100 and 2600 m elevation; they found the highest density at 2000 m and the high- est biomass at 500 m. On Mt. Kosciusko, Austra- lia, there is little difference in the species richness along an elevational transect between 910 and 2100 m elevation, although there is some species turnover (Wood, 1974). In general, the number of insects recovered from the soil plots was distinctly less than in other tropical forest sites. For example, Olson (1994) recorded up to 63 morphospecies of staphylinid beetles within an elevational zone in litter samples in Panama, whereas in the RNI d'Andringitra a maximum of two morphospecies was identified from soil samples in a single elevational zone. These differences certainly reflect collecting tech- niques (Leakey & Proctor, 1987), specifically, that leaf litter was not used in the RNI d'Andringitra samples. On the other hand, certain groups such as Annelida and Chilopoda showed equivalent or higher levels of species richness at Andringitra compared to Olson's study site in Panama (Olson, 1994). Microhabitat Variation Within each elevational zone the soil plots were placed in three microhabitats: ridges, slopes, and valleys. In contrast to vertebrates captured in the pitfall traps adjacent to the soil plots (see Chapter 20), no clear differences were found between mi- crohabitats in the taxonomic diversity or density of soil macroinvertebrates (Table 14-1). At the two highest elevations, the taxonomic diversity and the number of individuals recovered in the valley and slope plots were distinctly higher than in ridge plots. The Effects of Human Disturbance The 720 m elevation site was along a major forest trail between two villages and in close proximity to a small community of swidden ag- riculturalists, whereas the nearby 810 m site was relatively intact forest. Thus, differences in the soil macroinvertebrate communities of these two sites may reflect the affects of forest degradation. On the basis of presence or absence of various taxa, species richness as measured by the number of morphospecies, or the numbers of individuals recovered from the soil plots, no difference, in general, could be ascertained between the soil macroinvertebrate communities in the 720 and 810 m zones. Human activity often results in the movement of soil and plants, and with them, earthworms (e.g., Michaelsen, 1903). Numerous species of earthworms, but only a small fraction of the total species pool, have become globally distributed by human activity. Where cultivation alters natural vegetation or other human disturbance changes the habitat, exotic earthworms are generally suc- cessful colonizers. However, at least in some tem- perate zone forests, habitat alteration is not nec- essary. The presence of exotic species is likely in small remnants of original vegetation, whereas native species are usually found in large areas of unaltered vegetation (Kalis/ & Dotson, 1989). Unfortunately for biodiversity conservation, the invasion of exotic species often means local ex- tinction or severe reduction of native species pop- GOODMAN ET AL.: ELEVATIONAL VARIATION IN SOIL MACROINVERTEBRATES 149 ulations (Smith, 1928; Lee, 1961; Ljungstrom, 1972). Something like this may be in progress in the 810 m zone, where an exotic earthworm is the most abundant species. It was probably intro- duced accidentally by the movement of plants, root stocks, or soil. However, why exotic earth- worms were more common at 810 rather than 720 m (closer to agricultural areas) is unclear. One of us (S.W.J.) has observed trails through undis- turbed forests in the Caribbean Islands to be high- ways of earthworm dispersal. The same unknown process by which earthworms are dispersed along these trails may operate on the trail by the 810 m elevation site. Because earthworm self-dispersal is usually quite slow (maximal rates around 10 m per year), some human assistance is probably in effect. The distance between the nearest village or tavy and the pitfall trap sites is more than a few hundred meters. Another factor contributing to the colonization of many of the widely distributed, or peregrine, earthworms is uniparental reproduction. Though not all peregrins are parthenogenetic or otherwise capable of uniparental reproduction (earthworms are hermaphroditic), some are. The fully devel- oped Amynthas diffringens specimens collected at 810 m showed some signs of lacking the male functions: loss of prostate glands, absence of ma- tured sperm on the male funnels, and absence of sperm received in spermathecae. Collection data on Madagascar earthworms, ei- ther native or introduced, are limited and scat- tered. Michaelsen (1897) listed 15 introduced spe- cies in addition to describing four new native spe- cies and giving new collection locations for six other native species. Gates (1972) is a more recent source of information on the exotics. The native species are generally known only from the origi- nal descriptions, in which localities are seldom given with any degree of precision. In view of the extensive habitat modification since the last of these descriptions was published in the early 20th century, it is unlikely that the native species still exist at type localities. Elevational Gradients Using the 810, 1210, and 1625 m sites as points along an elevational gradient, there are some ap- parent patterns in the soil macroinvertebrates. The taxonomic diversity at 810 m elevation is on av- erage 7.3 morphospecies per plot; at 1210 m, 8.7; and at 1625 m, 7.5. Even though there is a con- siderable amount of variance within each eleva- tional zone, the data suggest that the taxonomic diversity of the soil macroinvertebrates is highest at around 1210 m elevation. The total number of individual specimens per plot rises from 9.1 at 710 m, to 9.5 at 825 m, 14.9 at 1210 m, and 16.7 at 1625 m (df = 3, r2 = 0.95, P = 0.03). Once again there is a large amount of variance within an elevational zone in the number of individuals found in each plot, but the density increases as a function of increasing altitude. These results sug- gest that species richness and species density do not show the same elevational trend. Acknowledgments We are grateful to John Slapcinsky of the Field Museum of Natural History for identifying the gastropod. Brian Fisher and an anonymous re- viewer provided important comments on an ear- lier version of this paper. Literature Cited Atkin, L., and J. Proctor. 1988. Invertebrates in the litter and soil on Volcan Barva, Costa Rica. Journal of Tropical Ecology, 4: 307-310. Chiba, S., T. Abe, J. Aoki, G. Imadate, K. Ishikawa, M. Kondoh, M. Shiba, and H. Watanabe. 1975. Studies on the productivity of soil animals in Pasoh Forest Reserve, West Malaysia. 1 . Seasonal change in the density of soil mesofauna: Acari, Collembola and others. Science Reports of the Hirosaki University, 22: 87-124. Collins, N. M., J. M. Anderson, and H. W. Vallack. 1984. 2. Studies on the soil invertebrates of lowland and montane rain forests in the Gunung Mulu Nation- al Park. Sarawak Museum Journal, 30: 19-33. Fragoso, C, and P. Lavelle. 1987. The earthworm community of a Mexican tropical rain forest (Chajul, Chiapas), pp. 281-295. In Bonvincini Paglai, A. M., and P. Omodeo, eds., On earthworms. Mucchi, Mo- dena, 562 pp. . 1992. Earthworm communities of tropical rain forests. Soil Biology and Biochemistry, 24: 1397- 1408. Gates, G. E. 1972. Burmese earthworms. Transaction? of the American Philosophical Society, new series. 62(7): 1-326. Griswold, C. E. 1993. Investigations into the phylog- eny of the lycosoid spiders and their kin (Arachnida Araneae: Lycosoidea). Smithsonian Contributions tc Zoology, 539: 1-39. Kalisz, P. J., and D. B. Dotson. 1989. Land-use his tory and the occurrence of exotic earthworms in th( 150 FIELDIANA: ZOOLOGY mountains of eastern Kentucky. American Midland Naturalist, 122: 288-297. Leakey, J. G., and J. Proctor. 1987. Invertebrates in the litter and soil at a range of altitudes on Gulum Silam, a small ultrabasic mountain in Sabah. Journal of Tropical Ecology, 3: 1 19-129. Lee, K. E. 1961. Interactions between native and intro- duced earthworms. Proceedings of the New Zealand Ecological Society, 8: 60-62. Ljungstrom, P. O. 1972. Taxonomical and ecological notes on the earthworm genus Udeina and a requiem for the South African acanthodrilines. Pedobiologia, 12: 100-110. Michaelsen, W. 1897. Die Terricolen des Madagas- sischen Inselgebiets. Abhandlungen herausgegeben von der Senckenbergischen Naturforschenden Ge- sellschaft, 21: 217-252. . 1903. Oligochaeten von Peradeniya auf Cey- lon, ein Beitrag zur Kenntnis des Einflusses botan- ischer Garten auf die Einschleppung peregriner Thi- ere. Sitzungsberichte der koniglichen bohmischen Ge- sellschaft der Wissenschaften (Prague), 40: 1-16. Olson, D. M. 1994. The distribution of leaf litter in- vertebrates along a Neotropical altitudinal gradient. Journal of Tropical Ecology, 10: 129-150. Smith, E 1928. An account of changes in the earth- worm fauna of Illinois and a description of one new species. Bulletin of the Illinois Natural History Sur- vey, 17: 347-362. Stork, N. E. 1988. Insect diversity: facts, fiction and speculation. Biological Journal of the Linnaean Soci- ety, 35: 321-337. Wood, T. G. 1974. The distribution of earthworms (Megascolecidae) in relation to soils, vegetation and altitude on the slopes of Mt Kosciusko, Australia. Journal of Animal Ecology, 43: 87-106. GOODMAN ET AL.: ELEVATTONAL VARIATION IN SOIL MACROINVERTEBRATES 151 Chapter 15 Shrimps (Crustacea: Decapoda: Atyidae) of the Reserve Naturelle Integrate d'Andringitra, Madagascar Mahefason Richard Andriamihaja Abstract Four Caridina species (Decapoda: Atyidae: Caridinae) were recorded in the humid forest streams of the Reserve Naturelle Integrale d'Andringitra. No Caridinae were found in the streams on the western slopes of the reserve. Resume Quatre especes de Caridina (Decapoda: Atyidae: Caridinae) ont ete trouves dans les cours d'eau de la zone de foret humide de la Reserve Naturelle Integrale d'Andringitra. En revanche aucun Caridinae n'a ete trouve dans les cours d'eau du versant ouest du massif de l'Andringitra. Introduction Malagasy freshwater shrimps belong to the families Atyidae and Palaemonidae (Crustacea: Decapoda). Both families contain genera that are consumed for food and have considerable eco- nomic importance. During the biological inven- tory of the Reserve Naturelle Integrale (RNI) d'Andringitra, shrimps were collected, all of which belong to the genus Caridina (Atyidae). Methods Shrimps were captured with a hoop net and pre- served in 70% ethanol. Determinations were made using the keys published by Holthuis (1965). Spe- cific determination of Caridina species can often be difficult due to insufficient information on in- dividual variation within a species. Thus, it is crit- ical to compare and define morphological differ- ences in ovigerous females and adult males of the same age. In this case specific determination was often based on adult males using numerous exter- nal morphological characters. Results Species of Caridina Edwards, 1 837 captured in the rivers of the RNI d'Andringitra include Car- idina xiphias, C hova, and C isaloensis. A fourth species was also collected that is not mentioned in Holthuis's (1965) keys and is probably new to science. This must be verified because many Car- idina species are described from Indo-Pacific wa- ters, and it may be known from elsewhere. Table 1 5- 1 lists all captures of shrimps made in the RNI d'Andringitra. No species of Caridina was taken at altitudes higher than 810 m (Fig. 15-1). Increasing altitude is correlated with de- creasing water temperatures (see Table 9- 1 , Chap- ter 9). Stream order may also be important in re- stricting the ranges of this genus, because Cari- 152 FIELDIANA: ZOOLOGY Table 15-1. Species, collection sites, and relative abundance of Caridina shrimps collected in the RNI d'An- dringitra. For information on stations see Chapter 9. Stations Relative abundance Males Species 1 2 3 5 Females Caridina hova + + 13.3 5 3 Caridina xiphias + + + + 16.7 10 0 Caridina isaloensis grandidieri + + + + 55.1 27 6 Caridina nov. sp.? + + + + 15.0 7 2 The plus sign ( + ) indicates the presence of the listed species at the particular station. dina species in Malagasy waters are known only at altitudes lower than 1210 m. At about 1210 m most waterways contain rapids and large water- falls, which presumably form a dispersal barrier for Caridinae. Three species of Caridina (C. xiphias, C. isal- oensis grandidieri, and Caridina sp.) were found in local sympatry at four stations (nos. 1, 2, 3, and 5) in the 720 and 810 m zones (Table 15-1). Caridina hova was found only at stations 1 and 3 along the main portion of the Iantara River in the 720 m zone. This is probably related to this spe- Western slope 2000 -- 1500-- 1000-- SOO -- Eastern slope CRUSTACEA Caridina -- 8 o Stream order 1 . Stream order 2 . Stream order 3 aa Stream order 4 ■ Stream Older 5 i aunrj4cd alalioni Fig. 15-1. Altitudinal distribution of Caridina (Crus- tacea) in the RNI d'Andringitra on the basis of 1 1 sites on the eastern slopes and four sites on the western slopes. Key to species: C. n. sp. — Caridina new spe- cies?; C. h. — C. hova; C. x. — C. xiphias; and C. i. g. — C. isaloensis grandidieri. cies's preference for certain water conditions. Generally Caridina species occurring in rivers with slow-flowing waters are found either in the aquatic vegetation or in clumps of fallen leaves deposited on the river bottom. Caridina xiphias Bouvier, 1925 In the RNI d'Andringitra this endemic species was found at all stations where shrimps were col- lected (Table 15-1). It is apparently relatively common near Didy, south of Lac Alaotra, and it is known to occur south of the Fianarantsoa re- gion (Holthuis, 1965). This species is distin- guished by the absence of up to three dorsal teeth of the rostrum placed behind the orbit, a long ros- trum reaching far beyond antennular peduncle, and the toothless ultimate portion of the upper margin of the rostrum. Ten adult males of C. xiph- ias were collected during the survey; they ranged in size from 7 to 26 mm. Caridina hova Nobili, 1905 In the RNI d'Andringitra, C. hova was collect- ed only at stations 1 and 3 (Table 15-1). This spe- cies was previously known only from southeast- ern Madagascar. The type specimens were cap- tured in a small stream, called Mandromodromo- tra, located 20 km north of Tolagnaro. Caridina hova is distinguished from other members of this genus by the rostrum not usually reaching beyond the base of the third antennular segment and even- ly dentate or with a short distal portion lacking teeth. Furthermore, the dorsal teeth of the rostrum are prominent and strong. The specimens collect- ed were five adult males and three ovigerous fe- males. Adult males vary in total length from 8 to ANDRIAMIHAJA: SHRIMPS 153 , and ovigerous females vary from 17 to 18 mm 18 mm Caridina isaloensis grandidieri Bouvier, 1901 Caridina isaloensis grandidieri was found at all stations where shrimps were collected (Table 15-1). This subspecies is known from western Madagascar. The main area of occurrence of C. i. grandidieri is the Fandiamanana River (Fandra- manona River). This locality may be the Fandra- mana River, 22°25'S, 46°20'E, between Ihosy and the Andringitra Massif. Caridina isaloensis grandidieri is similar to C. hova except the dorsal teeth of the rostrum are distinctly smaller. The collection made in the RNI d' Andringitra is composed of 27 adult males and six ovigerous females. Specimens of the former species range in total length from 7 to 24 mm, and ovigerous females range from 17 to 24 mm. The presence of C. isaloensis grandidieri in the RNI d' Andringitra is unexpected. Previously this form was known only from warm waters of west- ern, southern, and northern Madagascar. Other species, such as C. longirostris, C. gracilirostris, and C. serratirostris, have broad distributions across much of the island. They are found in a variety of river types and biomes from savannah to forest. Caridina sp. A fourth species of Caridina taken in the RNI d'Andringitra cannot be identified using the keys of Holthuis (1965) and may well be new to sci- ence. As noted above, this is in need of further study because numerous Caridina species are de- scribed from Indo-Pacific waters, and the unde- termined Andringitra Massif species may be known from areas away from the island. Literature Cited Holthuis, L. B. 1965. The Atyidae of Madagascar. Me- moire Museum National d'Histoire Naturelle, Nou- velle ser., ser. A, Zool., 23(1): 1-48. 154 FIELDIANA: ZOOLOGY Chapter 16 Crayfish (Parastacidae) and Crabs (Potamonidae) of the Reserve Naturelle Integrate d'Andringitra, Madagascar Bako Rabeharisoa Abstract Of the six described species of Malagasy crayfish, only one, Astacoides granulimanus, was recorded during the survey of the R6serve Naturelle Integrate d'Andringitra. They were col- lected in small brooks of the eastern humid forest of the massif. This species was not recorded on the western slopes. Two subspecies of the crab Hydrothelphusa agilis — H. a. agilis and H. a. madagascarien- sis — were recorded in watercourses ranging from 4 to 5 in stream order. The sympatric occur- rence of these two forms at the same elevations and in the same river system brings into question the subspecific arrangement of this species. Resume Parmi les six especes d'Ecrevisses malgaches, seule 1'espece Astacoides granulimanus a 6l6 recoltee lors de la prospection de la Reserve Naturelle Integrate (RNI) d'Andringitra. Elle a 616 trouvee dans de petits ruisseaux de la foret humide du versant Est. En revanche, aucun specimen n'a 6te recolte dans les cours d'eau du meme type du versant Ouest. Deux sous-especes du crabe Hydrothelphusa agilis: Hydrothelphusa a. agilis et H. a. mad- agascariensis ont 6t6 captures sur les cours d'eau d'ordre 4 et 5 de la RNI d'Andringitra. La presence simultande des deux sous-especes a la meme altitude et dans le meme bassin pose le probleme de la valeur taxinomique des deux sous-especes d' Hydrothelphusa agilis. Crayfish All named Malagasy crayfish (Crustacea: De- capoda) belong to the family Parastacidae and the genus Astacoides Gu£rin, 1839, and all species are endemic to the island. Madagascar is one of the few sites in the intertropical zone where cray- fishes are known. In general, the majority of cray- fishes are found in temperate areas, but they are also known in Honduras, Guatemala, New Guin- ea, Aru and Misol Islands, and northern Australia. No species is known on the African continent or in the Indian subregion. Astacoides has a closer affinity with the Tasmanian genus Astacopis than with any other genus in the world. Although Mad- agascar and Tasmania are geographically distant, Astacoides and Astacopis appear to share a com- mon recent ancestor. Until a few years ago, the genus Astacoides was considered monospecific, with Astacoides madagascariensis comprising several subspecies. Currently six species are recognized: Astacoides crosneri Hobbs, 1987; A. petiti Hobbs, 1987; A. granulimanus (Monod & Petit, 1929), A. mada- gascariensis (Edwards & Audouin, 1839), A. caldwelli (Bate, 1865), and A. betsileoensis (Petit, RABEHARISOA: CRAYFISH AND CRABS 155 1923). Their vernacular Malagasy names vary ac- cording to species and local dialects: orana, or- ambanonga, orambato, and oramboro. Methods Collections of crayfish within the Reserve Na- turelle Integrate (RNI) d'Andringitra were made by hand in small tributaries (stream order 1) of larger streams and rivers. Specimens were caught under rocks and inside many-galleried holes in the shallow streams. Water flow was slow, ranging from 1 to 3 liters per second. See Chapter 9 for further information on collecting sites and infor- mation on stream order. Results and Discussion Among the six crayfish species recorded in Madagascar, only one, Astacoides granulimanus, was collected in the RNI d'Andringitra. Speci- mens were obtained at altitudes ranging from 750 (nine specimens) to 1650 m (two specimens) (Fig. 16-1). The highest previous altitudinal record for crayfishes on the island is 1650 m (Hobbs, 1987). The particular conditions in the RNI d'Andringi- tra region, clear fresh forest streams at high alti- tude, presumably account for its local presence. The water systems of the upper mountain zone of the reserve are generally a series of falls between shelves. Presumably crayfishes find their way around waterfalls by passing on solid ground. As- tacoides granulimanus has the broadest distribu- tion of any Malagasy crayfish, from near Anta- nanarivo south to the Farafangana region (Hobbs, 1987). This species is relatively common in the Ikongo Forest, northeast of RNI d'Andringitra (Hobbs, 1987). Astacoides are abundant in shallow freshwater with pH ranges of 4-6. Astacoides attain a rela- tively large size; adults range from 25 to 30 cm in total length and weigh between 100 and 130 g. They are nocturnal or crepuscular. The geographic range of Astacoides appears to be from Anjozo- robe in the north (northeast of Antananarivo) south to the Isaka River, a tributary of the Efaho River (northeast of Tolagnaro), for a total distance of about 700 km along the eastern coast. In the east, crayfish occur at altitudes from 500 to 1700 m, and in westward drainages they are concen- trated in small high-mountain streams (1400 to 1700 m). Compared to other tropical crayfish, the Western slope Eastern slope ,m CRUSTACEA Astacoides & Hydrothelphusa ,«l 2000- 11-- ^\T™ 1500- 130 9 (£&■) 1000- 12 o 500- \ 7 \ L6\ 5J|r*T 3 H.a.a. | h r^4 ' Lm 1 Hxun. [ i Stream order 1 ^^_ Stream order 4 Stream order 2 mMti Stream order 5 Stream order 3 9 = sampled stations <^> Astacoides 1 1 Hydrothelphusa Fig. 16-1. Altitudinal distribution of Astacoides and Hydrothelphusa in the RNI d'Andringitra on the basis of 11 sites on the eastern slopes and four sites on the western slopes. Key to species: A. gr. — Astacoides gran- ulimanus; H.a.a. — Hydrothelphusa agilis agilis; and H.a.m. — H. a. madagascariensis. Malagasy species appear to live in the lowest al- titudes and highest temperatures. Crabs Methods and Results Crabs were collected either by searching under rocks or using worms at the same collecting sites reported for insects (see Chapter 9). Only one spe- cies, Hydrothelphusa agilis (Decapoda: Potamon- idae: Hydrothelphusinae) was captured in the RNI d'Andringitra, at stations 2 and 3 on water of stream orders 4 and 5. On the basis of keys provided in Bott (1965), the specimens are composed of two different sub- species, H. a. agilis and H. a. madagascariensis (Table 16-1). Hydrothelphusa a. agilis was caught using worms in Stillwater zones of streams and H. a. madagascariensis was collected by hand under rocks in stream riffles. These two subspecies were not sympatric at any station, although they were 156 FIELDIANA: ZOOLOGY Table 16-1. Distribution by station of Hydrothel- present in the same elevational zone. Whether phusa crabs captured in the RNI d'Andringitra. For in- m represent distinct taxa or two stages of me formation on stations see Chapter 9. ...«.-. . , , same species with different morphology and ecol- 7~ ogy must be determined. Station OJ no. Subspecies Males Females 2 Hydrothelphusa agilis agilis 3 3 Literature Cited 3 Hydrothelphusa agilis agilis 3 0 Borr, R. von. 1965. Die Susswasserkraben von Mad- 4 Hydrothelphusa agilis 4 3 agascar (Crustacea, Decapoda). Bulletin Museum Na- madagascariensis tional d'Histoire Naturelle, 2nd ser., 37: 335-350. Hobbs, H. H. 1987. A review of the crayfish genus Astacoides (Decapoda: Parastacidae). Smithsonian Contributions to Zoology, 443: 1-49. RABEHARISOA. CRAYFISH AND CRABS 157 Chapter 17 Amphibians and Reptiles of the Reserve Naturelle Integrale d'Andringitra, Madagascar: A Study of Kiev ational Distribution and Local Pandemicity Christopher J. Raxworthy and Ronald A. Nussbaum Abstract Amphibians and reptiles were surveyed on the eastern side of the Reserve Naturelle Integrale (RNI) d'Andringitra, between 650 and 2300 m elevation. This region of the reserve had not previously been studied. A total of 57 amphibian and 35 reptile taxa were found. Of these, 68% of the amphibian and 88% of the reptile species are new records for the reserve. Pitfall captures suggest that species diversity and abundance are highest in valley bottom forest and lowest in ridge forest. No species occurs throughout the complete elevational range in the reserve, and 45% of the species are restricted to just one of the five elevational transects surveyed (650-800, 750-860, 1100-1350, 1550-1700, and 1850-2300 m). Herpetofaunal species diversity declines with increasing elevation above 800 m. Above 1300 m, species diversity is 21% or less of the maximum diversity recorded at 800 m elevation. Twelve taxa are currently considered to be endemic to the Andringitra Massif, which includes species in both montane and low-elevation habitats. All other species (not locally endemic) are widely distributed on the eastern escarpment or other Central High Plateau massifs. Eight montane specialist species occur both at Andringitra and Ankaratra, and they provide evidence that montane habitats were once widespread on the Central High Plateau, possibly during glacial periods of the Pleistocene. Conservation of the herpetofauna at the RNI d'Andringitra will require protection of all habitats found within the reserve, of which the low-elevation rain forest and montane sclerophyllous forest appear to be the most vulnerable. Resume Un inventaire des amphibiens et des reptiles a ete effectue dans la partie est de la Reserve Naturelle Integrale (RNI) d'Andringitra, entre 650 et 2300 m d' altitude. Cette region de la reserve n'avait jamais ete inventorize auparavant. Un total de 57 amphibiens et de 35 reptiles a ete trouve, parmi lesquels 68% des especes d'amphibiens et 88% des especes de reptiles constituent de premiers recensements pour la reserve. Les captures realisees au moyen de pieges "pitfall" demontrent que la diversite des especes et l'abondance des individus sont maximales dans la foret des fonds de vallee et minimales dans la foret de crete. Aucune espece n'apparait sur l'ensemble des etages d'altitude presents au sein de la reserve et sur les cinq transects altitudinaux inventories (650-800, 750-860, 1100-1350, 1550-1700 et 1850-2300 m), 45% des especes ont une presence limitee a un seul transect. La diversite 158 FIELDIANA: ZOOLOGY specifique de 1'herpetofaune d'ecrott a partir de 800 m d'altitude. La diversite specifique au dessus de 1300 m est de 21% ou moins de la diversite maximale enregistree a 800 m d'altitude. Douze taxons sont maintenant considers comme etant endemiques au massif d'Andringitra, parmi lesquelles des especes de montagne et de basse altitude. Toutes les autres especes sont largement distributes au sein des forets de Test ou dans les massifs forestiers du haut-plateau central. Huit especes de montagne ont 6l6 recensees a la fois dans l'Andringitra et dans l'Ankaratra et ont fourni l'dvidence que les habitats montagnards 6taient autrefois repandus sur le haut-plateau central, probablement au cours des periodes glaciaires du Pleistocene. La conservation de 1'herpetofaune de la RNI d'Andringitra necessite la protection de tous les habitats que Ton trouve dans la reserve, parmi lesquels les forets humides de basse altitude et les forets sclerophylles de montagne qui semblent etre les plus vulneYables. Introduction The higher elevation forests and montane heathlands of Reserve Naturelle Integrate (RNI) d'Andringitra have been subject to opportunistic collecting of amphibians and reptiles for more than 70 years. Some of the earliest collections of frogs were made by Petit before 1924 and Millot in 1949 (Guibe\ 1978; Blommers-Schlosser & Blanc, 1991). In 1937, a small collection of am- phibians was made for the Museum National d'Histoire Naturelle (MNHN), Paris (Millot & Guibe, 1950). The first reptile species described from Andringitra was Lygodactylus montanus, which was catalogued in 1956 at MNHN (Pasteur, 1964). The collector, although not named by Pas- teur, was probably Millot. In 1970-1971 a multidisciplined survey was made of RNI d'Andringitra. The survey included zoologists who (primarily C. P. Blanc) collected amphibians and reptiles (Paulian et al., 1971). Like earlier collections made in the reserve, this survey focused on the higher elevation habitats of the massif, with all specimens taken from 1200 to 2658 m. Currently, a total of 28 amphibian spe- cies are recorded for RNI d'Andringitra (Blom- mers-Schlosser & Blanc, 1993). The only reptile list for the reserve is based on published and un- published observations and includes a total of 13 species (Nicoll & Langrand, 1989). Based on existing collections and literature, it was clear that the herpetological diversity of RNI d'Andringitra was incompletely known. This was especially true for the lower elevation wet forests on the eastern slopes that had apparently never been surveyed for amphibians and reptiles. The elevational distribution of most species recorded from RNI d'Andringitra was also largely un- known. The specific aims of our survey were to determine the full elevational range of habitats and species found on the eastern slopes of the reserve. Study Sites Fieldwork was centered on five elevational transects: (1) 650-800 m; Iantara River (camp 1, 720 m, near Ambarongy Village); 22°13'20"S, 47°01'29"E; 14-21 November and 17 December 1993; primary to degraded rain forest, tavy clear- ings. (2) 750-860 m; Sahavatoy River (camp 2, 810 m, junction with Sahanivoraky River); 22°13'40"S, 47°00'13"E; 22-29 November 1993; primary to degraded rain forest, tavy clearings. (3) 1 100-1350 m; Volotsangana River (camp 3, 1210 m). 22°13'22"S, 46°58'18"E; 30 November-7 De- cember 1993; primary rain forest. (4) 1550-1700 m; Ridge E of Volotsangana River (camp 4, 1625 m); 22°11'39"S, 46°58'16"E; 8-12 and 15-16 De- cember 1993; primary rain forest. (5) 1850-2300 m; Kimora River (camp 5, 2075 m, 4 km NE of Pic Bory); 22°10'10"S, 46°56'40"E; 13-14 De- cember 1993; montane heathland. Methods The three members of the herpetological survey team were Nirhy Rabibisoa (DEA student, Univ- ersite d' Antananarivo), C.J.R., and Angelin Ra- zafimanantsoa (research guide, Montagne d'Am- bre). The survey was timed for the onset of the rainy season, when most species are breeding and ac- tivity is at its highest. Field techniques used to sample animals were (1) pitfall trapping with drift fences; (2) opportunistic day and night searching; RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 159 amphibians reptiles 720 810 1210 1625 2075 Transect Fig. 17-1. Herpetofaunal diversity at each transect. and high plateau of the RNI d'Andringitra. The majority of searching was done close to new trails made during the study, although ridges and river- banks were also used to orient search paths. Night searches were made using headlamps. The following information was recorded at the time of capture for each individual: date, time, longitude, latitude, altitude, microhabitat, and cir- cumstances of capture. Animals not retained for specimens were returned to the site of original capture. Voucher specimens were fixed in 10% buffered formalin and later transferred to alcohol. Liver and muscle were removed from represen- tatives of almost all species and frozen in liquid nitrogen. Color slides were taken of live individ- uals of most species. Frog calls were recorded where possible. Collected material was deposited at two main centers, the Museum of Zoology, University of Michigan, and the Department de Biologie Animale, Universite d' Antananarivo. (3) refuge examination (under and in fallen logs and rotten tree stumps; under bark; under rocks; in leaf litter, root-mat, and soil; and in leaf axils of Pandanus screw palms and Ravenala travel- ler's palm. The pitfall traps were buckets (275 mm deep, 290 mm top internal diameter, 220 mm bottom internal diameter), sunk in the ground at 10 m intervals below a 100-m-long drift fence (Fig. 17-1). The handles of the buckets were removed, and small holes (2 mm diameter) were punched in the bottom to allow water drainage. The fence (0.5 m high) was made from plastic sheeting stapled in a vertical position to thin wooden stakes. The fence bottom was buried 50 mm deep into the ground using leaf litter and po- sitioned to run across the middle of each pitfall trap. A pitfall trap was positioned at both ends of the drift fence, with the other nine traps at 10 m intervals. The trap lines were checked each morn- ing and late afternoon. After heavy rain the buck- ets were sponged dry. At each camp a line was placed in each of the following forest types: ridge (along the crest of a ridge), slope (on a gradient, intermediate between ridge top and valley bot- tom), and valley (within 20 m of a stream in a valley bottom). Three lines were used in each of the lower four transect zones, and sampling was made over 6-8 days. A total of 12 lines were erected during the survey. Opportunistic searching and refuge examina- tion were made throughout the full elevation range of habitats available on the eastern slopes Results A total of 57 amphibian and 35 reptile species were recorded during this survey within the RNI d'Andringitra, with another three reptile species (Furcifer lateralis, Zonosaurus madagascariensis, and Z. ornatus) found at Ambalamanenjana Vil- lage, 12 km north of the reserve limit (Table 17-1). Ten amphibian and seven reptile species have not been identified and may represent un- described taxa. Of the 57 amphibian species re- corded, 39 (68%) are new records for the reserve; of the 34 reptile species, 30 (88%) are new re- cords. Amphibian species diversity was greatest at 810 m and lowest at 2075 m (Table 17-1; Fig. 17-1). Reptile species diversity was greatest at 720 m and lowest at 1625 m (Table 17-1; Fig. 17-1). For both amphibians and reptiles, species diversity was much lower at 1550-2350 m than at lower elevations, 650-1350 m. The characteristics of the pitfall trap lines and captures are given in Table 17-2. A total of 124 vertebrates were captured over 902 pitfall trap (bucket) days, giving an average vertebrate cap- ture rate of 0.14 per trap day (14% trap success per trap day). Pitfall lines captured seven amphib- ian, five reptile (all skinks), and 12 small mammal species. At the four transects where pitfalls were used, species diversity and capture rate were greatest in valley forest and lowest in ridge forest, 160 FIELDIANA: ZOOLOGY with the exception of 810 m, where the lowest species diversity and capture rate were in slope forest. The 1210 m pitfalls had higher capture rates and species diversity than the 720, 810, and 1625 m transects, with the only exception being the 810 m valley forest line, which had a slightly greater species diversity. Discussion Species Not Surveyed Although many of the species found during this survey are new records for RNI d'Andringitra, other species reported for the reserve were not found. Table 17-3 is a list of the missing 21 am- phibian and 14 reptile species. Considering first the amphibians, the record of Plethodontohyla tuberata (Nicoll & Langrand, 1989) apparently is not supported by a voucher specimen, and Blom- mers-Schlosser & Blanc (1993) did not list this species for RNI d'Andringitra. The vouchers of the amphibian species reported by Blommers- Schlosser and Blanc (1991, 1993) and Glaw and Vences (1994) need to be examined to determine if they actually represent species different from those recorded in this study. Some "missing spe- cies" (e.g., Mantidactylus pliciferus, M. cornutus, M. brevipalmatus, M. spiniferus, Boophis lauren- ti, and B. untersteini) are easily confused with other species we recorded in the RNI d'Andrin- gitra. Three species, Scaphiophryne madagascar- iensis, Mantidactylus domerguei, and M. elegans, were missed; these species are probably restricted to the high-elevation forests and heathlands that were less intensively studied during this survey. Nicoll and Langrand (1989) recorded seven reptile species that are seemingly not represented by voucher specimens and probably represent per- sonal observations. Two oplurid species, Opiums cuvieri and O. cyclurus, are not known from lo- calities close to RNI d'Andringitra (Blanc, 1977), and they most likely represent misidentified O. quadrimaculatus. The most recent reviews of Ma- buya and Zonosaurus by Brygoo (1983, 1985) do not include RNI d'Andringitra as a locality for Mabuya aureopunctata and Zonosaurus laticau- datus. Two reptile species for which Andringitra vouchers exist were certainly missed by us. These are Mabuya boettgeri and Oplurus quadrimacu- latus (Blanc, 1977; Brygoo, 1983). Both species occur in the heathland on or near rocky outcrops at high elevation, and they are probably present in the 2075 m transect. They may have been over- looked because of the overcast and rainy condi- tions during the short period this transect was sur- veyed. Other reptiles that we missed in the reserve (Furcifer lateralis, Mabuya elegans, Zonosaurus ornatus, and Mimophis mahfalensis) were seen by us in degraded habitats around villages close to the reserve. Glaw and Vences (1994) also record- ed Lygodactylus pictus (near Sendrisoa) and Ma- buya madagascariensis. These identifications re- quire confirmation, although it is not clear if voucher specimens exist for these records. Sampling Pitfall traps yielded just one species of amphib- ian (.Plethodontohyla alluaudi) that was not ob- tained by opportunistic searching. All reptile spe- cies captured in pitfalls were also sampled by di- rect searching. Although pitfall traps did not sig- nificantly contribute to the herpetofaunal species inventory for the reserve, pitfalls did capture spe- cies otherwise represented by very few specimens (e.g., Plethodontohyla sp. 1, P. serratopalpebro- sa, Amphiglossus anosy, and A. macrocercus). Without pitfalls, several species would not have been recorded at some elevations. Pitfall traps also made a significant contribution to the mam- mal studies, particularly of insectivores (see Chapter 20). Overall trap success for vertebrates was slightly lower in the RNI d'Andringitra (14%, 902 trap days) compared to Pare National (PN) de la Montagne d'Ambre, in extreme northern Madagascar, where identical pitfall trapping tech- niques produced an overall trap success of 21% over the course of 2,244 trap days (Raxworthy & Nussbaum, 1994). Few new species were being discovered by the end of each sampling period (Fig. 17-2). The spe- cies lists for the four lowest transects are consid- ered to be nearly complete. Because sampling time at the 2075 m transect was much shorter, some species are likely to have been missed (see above). Of the 14 species with an elevational range that spanned three or more transects, just three species {Plethodontohyla inguinalis, Boophis sp. 4, and Brookesia nasus) have a gap in their dis- tribution (i.e., they were not sampled in an in- termediate transect). Although elevation gaps in distribution suggest that species were missed at RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 161 Table 17-1. Amphibian and reptile species recorded at RNI d'Andringitra. Taxon Elevation of transect Range (m) 720 m 810 m 1210 m 1625 m 2075 m Min. Max. Amphibia Microhylidae Scaphiophryne marmorata Anodonthyla boulengeri Anodonthyla montana Plethodontohyla bipunctata Plethodontohyla inguinalis Plethodontohyla serratopalpebrosa Plethodontohyla alluaudi Plethodontohyla notosticta Plethodontohyla sp. 1 Platypelis turberifera Platypelis sp. 1 Platypelis sp. 2 Mantellidae Mantella madagascariensis Mantidactylus aerumnalis Mantidactylus aglavei Mantidactylus argenteus Mantidactylus asper Mantidactylus bertini Mantidactylus betsileanus Mantidactylus bicalcaratus Mantidactylus biporus Mantidactylus eiselti Mantidactylus femoralis Mantidactylus flavobrunneus Mantidactylus grandidieri Mantidactylus grandisonae Mantidactylus liber Mantidactylus lugubris Mantidactylus luteus Mantidactylus majori Mantidactylus opiparis Mantidactylus peraccae Mantidactylus pulcher Mantidactylus redimitis Mantidactylus tornieri Mantidactylus ulcerosus Mantidactylus sp. 1 Rhacophoridae Aglyptodactylus madagascariensis Boophis albilabris Boophis albipunctatus Boophis boehmei Boophis goudoti Boophis luteus Boophis madagascariensis Boophis microtympanum Boophis miniatus Boophis rappoides Boophis reticulatus Boophis viridis Boophis sp. 1 Boophis sp. 2 Boophis sp. 3 800 820 1240 1700 1700 2300 800 1280 740 1280 1240 1300 800 900 700 800 800 1240 650 860 1240 1240 1240 1240 650 1280 1850 1850 700 840 700 800 700 1250 1450 1650 650 1220 740 780 650 1400 800 1350 700 1250 700 800 700 820 820 820 800 1150 650 1250 650 1700 650 820 650 1400 1240 1260 780 800 650 1700 720 720 1450 1700 720 820 650 850 700 850 820 820 700 820 1450 1850 700 820 650 850 1850 2300 800 1200 700 720 800 1650 700 720 800 820 830 830 1240 1260 162 FIELDIANA: ZOOLOGY Table 17-1. Continued. Taxon Elevation of transect Range (m) 720 in 810 in 1210 m 1625 m 2075 m Min. Max. Boophis sp. 4 Boophis sp. 5 Boophis sp. 6 Hyperoliidae Heterixalus betsileo Ranidae Ptychadena mascareniensis Totals — Amphibia Reptilia Gekkonidae Lygodactylus montanus Phelsuma barbouri Phelsuma I. lineata Phelsuma q. quadriocellata Uroplatus ebenaui Chamaeleonidae Brookesia n. nasus Brookesia superciliaris Calumma b. brevicornis Calumma b. hilleniusi Calumma fallax Calumma g. gastrotaenia Calumma g. andringitraensis Calumma nasuta Calumma oshaughnessyi Furcifer campani Seine idae Amphiglossus anosy Amphiglossus macrocercus Amphiglossus melanopleura Amphiglossus melanurus Amphiglossus punctatus Amphiglossus sp. 1 Amphiglossus sp. 2 Mabuya gravenhorsti Cordylidae Zonosaurus sp. 1 Colubridae Geodipsas heimi Geodipsas infralineata Geodipsas sp. 1 Geodipsas sp. 2 Geodipsas sp. 3 Liophidium rhodogaster Liophidium sp. 1 Liopholidophis grandidieri Liopholidophis stumpffi Liopholidophis thieli Liopholidophis sp. 1 Totals— Reptilia 1240 1850 1240 1240 1240 1280 720 720 650 720 29 36 26 2000 2250 2000 2300 720 750 720 1350 810 840 720 1630 650 840 1200 1300 1550 1850 1220 1260 1100 1280 1550 1680 720 830 1240 1300 1850 2000 1240 1300 720 900 720 800 720 780 800 1300 750 750 2000 2100 650 800 650 * 800 1260 1260 700 820 720 720 1240 1240 1630 1630 750 750 730 850 1300 1300 720 720 750 840 800 800 17 14 11 RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 163 'C <£ Oq^O\^00 VON i~~ $ o > — r~ r» 8* ON 00 >/-) — CM OV-M^ O VOm a* r»» — ■ -h s m o\r~ en o\ r» "^ cs cm r~ — — i 5£ ^ O ■> NO — NO no en en — * — Cl vD $ O o 1 1 is s ^ >> 2 CS c/3 T3 -Q >^ h <- Z E *>£^ — s- > i ude st ty sarr sam 1 tra] Ed u 5 •S ^ B a 1 a J3 u 1 -5 o -a 5i Jo < e pu j p u S CM CM — E E 2 K - "-a S 3 OS 164 FIELDIANA: ZOOLOGY Table 17-3. Species not found during the 1993 survey, but reported as occurring at RNI d'Andringitra. Species Source Amphibia Platypelis grandis Plethodontohyla tuberata Scaphiophryne madagascariensis Mantidactylus brevipalmatus Mantidactylus cornutus Mantidactylus elegans Mantidactylus boulengeri Mantidactylus decaryi Mantidactylus domerguei Mantidactylus madecassus Mantidactylus pliciferus Mantidactylus spiniferus Mantidactylus curtus Mantidactylus alutus Boophis ankaratra Boophis majori Boophis laurenti Boophis untersteini Boophis sp. b Heterixalus tricolor Heterixalus madagascariensis Reptilia Lygodactylus pictus Furicifer lateralis Opiums cuvieri Opiums cyclurus Opiums quadrimaculatus Mabuya aureopunctata Mabuya boettgeri Mabuya elegans Mabuya madagascariensis Zonosaurus laticaudatus Zonosaurus ornatus Leioheterodon madagascariensis Dromicodryas bernieri Mimophis madagascariensis Glaw & Vences (1994) Nicoll & Langrand (1989) Blommers-Schlosser & Blanc Glaw & Vences (1994) Glaw & Vences (1994) Blommers-Schlosser & Blanc Blommers-Schlosser & Blanc Blommers-Schlosser & Blanc Blommers-Schlosser & Blanc Glaw & Vences (1994) Blommers-Schlosser & Blanc Blommers-Schlosser & Blanc Blommers-Schlosser & Blanc Blommers-Schlosser & Blanc Glaw & Vences (1994) Glaw & Vences (1994) Blommers-Schlosser & Blanc Blommers-Schlosser & Blanc Glaw & Vences (1994) Blommers-Schlosser & Blanc Glaw & Vences (1994) (1993) (1993) (1993) (1993) (1993) (1993) (1993) (1993) (1993) (1993) (1993) (1993) Glaw & Vences (1994) Brygoo (1978) Nicoll & Langrand (1989) Nicoll & Langrand (1989) Blanc (1977), Nicoll & Langrand (1989) Nicoll & Langrand (1989) Brygoo (1983) Brygoo (1983) Glaw & Vences (1994) Nicoll & Langrand (1989) Glaw & Vences (1994) Nicoll & Langrand (1989) Nicoll & Langrand (1989) Nicoll & Langrand (1989), Glaw & Vences (1994) some transects, these gaps may be real due to other factors, such as the absence of specific mi- crohabitats. Influence of Human Disturbance Although the 720 and 810 m transects were in similar types of forest, the forest at the lower tran- sect was more disturbed, through clearing and se- lective logging, and the density of human trails was much higher (see Chapter 4). Two frog spe- cies recorded in the 720 m transect, Ptychadena mascareniensis (not endemic to Madagascar) and Heterixalus betsileo, are characteristic of second- ary habitats or forest edge. Both these species were found in degraded habitats along the Iantara River and were not found within the 810 m tran- sect. Thirteen amphibians and three reptiles found in primary rain forest in the 810 m transect were not observed in the 720 m transect (Table 17-1). Other faunal differences between the 720 and 810 m transects, however, are clearly not due to greater habitat disturbances at the lower elevation. Two species of frog (Mantidactylus tornieri and Boophis viridis) and four species of reptile (Am- phiglossus melanopleura, Geodipsas sp. 1, Lio- pholidophis stumpffi, and Liopholidophis sp. 1) that are restricted to primary rain forest were found only in the 720 m transect and were absent from the higher elevation transects. Their absence in the 810 m transect is not explained by the more RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 165 Table 17-4. Herpetofaunal coefficients of similarity between 100 m elevation bands. 012345678 Time (days) Fig. 17-2. Accumulation curves of herpetofaunal di- versity (all sampling techniques). pristine condition of the primary forest in this transect. The faunal differences between the two transects probably reflect (at least in part) their slight difference in elevation. Elevational Distribution No species was distributed throughout the com- plete elevational range sampled within RNI d'Andringitra. The greatest amphibian elevational range was 650-1700 m (Mantidactylus luteus), and for reptiles, 720-1630 m (Brookesia nasus nasus). The herpetological communities changed sharply between each transect (Table 17-1). A comparison was made of the five most inten- sively surveyed 100 m elevation bands, centered at 700, 800, 1250, 1650, and 1900 m, using the following coefficient of similarity, S: where C is the number of species in common and N1+2 is the total number of species found for both transects combined. The coefficients are given in Table 17-4. The highest similarity (0.53) was found between the 700 and 800 m bands. All other coefficients of similarity were less than 0.50, demonstrating that the majority of species were not widely distrib- uted between the elevation bands. No species was 166 ation Elevation ( m) Elev 1200- 1600- 1850- (m) 650-750750-850 1300 1700 1950 650- -750 1.00 0.53 0.20 0.04 0.00 750- -850 1.00 0.26 0.03 0.00 1200- -1300 1.00 0.09 0.02 1600- -1700 1.00 0.17 1850- -1950 1.00 found at both the 1900 m and either the 700 or 800 m band. Elevational limits are typically proposed to help classify rain forest types in Madagascar (e.g., White, 1983; Jenkins, 1987), with low-altitude rain forest occurring below 800 m, montane forest at 800-1300 m, and sclerophyllous rain forest at 1300-2000 m. The 720 and 810 m transects are in the transition between low- and mid-altitude rain forest, and modest differences between the communities of these two transects may reflect, at least in part, their slight difference in elevation. A sharp transition in the herpetofaunal community at 900 m elevation was reported for rain forest at PN de la Montagne d'Ambre (Raxworthy & Nussbaum, 1994). The 1210m transect is in mon- tane rain forest, the 1625 m transect in a mixture of montane and sclerophyllous rain forest, and the 2075 m transect in montane heathland. The low similarity coefficients (^ 0.26) between the her- petofaunal communities of each habitat suggest that many species of amphibians and reptiles are specialized to a subset of vegetation types along this altitudinal gradient. Figure 17-3 shows the change in species diver- sity of amphibians and reptiles as a function of altitude. Elevation is divided into 100 m bands, and species distributions are inferred to be contin- uous between the minimum and maximum ele- vation recorded at RNI d'Andringitra. Both am- phibian and reptile diversity show an obvious trend toward decreasing with increasing elevation. Above 1 300 m, total herpetological species diver- sity does not exceed 1 1 species, representing at most 21% of the maximum species diversity re- corded between 800 and 900 m. Similar results recently were reported for Broo- kesia chameleons in northern Madagascar, where maximum species diversity occurs at 700 m, and above 1300 m just 20% or less of the maximum species diversity is represented (Raxworthy & FIELDIANA: ZOOLOGY 40 -i amphibians reptiles LfcjaJjJlll* 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Altitude (x 100 m) Fig. 17-3. Herpetofaunal diversity as a function of altitude. Nussbaum, 1995). Brookesia also exhibits a mid- altitude bulge in species diversity, between 500 and 700 m in altitude, with species diversity be- low 500 m representing 50% of the maximum species diversity at 700 m. On the eastern slopes of the RNI d' Andringitra, the increase in amphib- ian diversity between 700 and 800 m may repre- sent the maximum of a similar mid-altitude bulge. Explanations for a decrease in species diversity with altitude have not yet been proposed for hab- itats in Madagascar. Elevational diversity gradi- ents in general are poorly understood, although examples are well known for bird communities in the tropics (Brown, 1988). Possible factors are habitat productivity and stability. Both probably decrease at higher elevation due to environmental conditions such as cooler temperatures, extreme seasonality, and increased vulnerability to cyclone and fire damage. Endemicity The following 1 1 taxa are known only from RNI d' Andringitra and are probably endemic to the massif: amphibians — Anodonthyla montana, Plethodontohyla sp. 1, Platypelis sp. 2, Boophis sp. 1, Boophis sp. 3, and Boophis sp. 6; reptiles — Lygodactylus montanus, Calumma gastrotaenia andringitraensis, Amphiglossus sp. 1 , Zonosaurus sp. 1 , and Liopholidophis sp. 1 . Another endemic chameleon subspecies, Brookesia nasus pauliani, is restricted to the Andringitra Massif rain forest between 1620 and 1650 m elevation and repre- sented by only two males (Brygoo, 1978). It is diagnosed by the lack of paired vertebral tubercles and strong heterogeneity of the skin granules. All Brookesia nasus caught during this survey had characters of B. n. nasus (paired dorsal tubercles and moderate heterogeneity of skin granules). However, we found no males at high elevation. A female caught at 1630 m had characters identical to those at lower elevations. To assess the actual status of B. n. pauliani, it will be necessary to examine the type material. The Andringitra Massif is isolated by the Am- balavao Pass to the north, and by the Mananara Valley (south of Pic dTvohibe) to the south. Both areas do not exceed 1000 m elevation. Lower al- titude rain forest (below 1000 m) at RNI d' An- dringitra, therefore, is not isolated (except by riv- ers) from the same forest type to the north and south. By contrast, the montane forests are com- pletely isolated from other montane areas by cor- ridors of lowland forest. Therefore, based on de- gree of isolation, endemicity should be more pro- nounced for high-altitude specialist species. Seven RNI d' Andringitra endemic species (An- odonthyla montana, Platypelis sp. 2, Boophis sp. 3, Boophis sp. 6, Lygodactylus montanus, Cal- umma gastrotaenia andringitraensis, and Broo- kesia nasus pauliani) are restricted to habitats RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 167 above 1000 m altitude, with minimum known el- evations of between 1240 and 2000 m. By con- trast, there are five locally endemic species re- stricted below 1000 m elevation: Plethodontohyla sp. 1, Boophis sp. 1, Amphiglossus sp. 1, Zono- saurus sp. 1, and Liopholidophis sp. 1. Clearly, local endemicity is not confined to just montane taxa. The presence of low-altitude endemic spe- cies in the RNI d'Andringitra may be due to: (1) species actually being more widespread than is currently known, and therefore not regionally en- demic; (2) other barriers (e.g., major rivers) con- tributing to isolation; and (3) local adaptation in the absence of geographical barriers. Comparisons with Other Sites Table 17-5 is a list of species that were record- ed in this survey and have also been sampled at three other rain forest sites: Ankaratra (Central High Plateau), PN de Ranomafana (eastern es- carpment), and the Anosy Mountains (eastern es- carpment). The Anosy Mountains as here defined include both the Anosy and Vohimena Mountains, and they actually represent six sites between 50 and 1200 m elevation (Marovony, Ampamakie- siny, Manangotry, Marosohy, Nahampoana, and Manantantely) surveyed by Raxworthy and Nuss- baum between 1989 and 1991. The Anosy Moun- tains represent the southern limit of the rain forest on the eastern escarpment, with Marovony, the closest site to Andringitra, 200 km to the south. Pare National de Ranomafana, rain forest 120 km north of the Andringitra Massif, was surveyed by Raxworthy (unpublished) in 1990 between 900 and 1050 m elevation. Ankaratra Massif, 300 km north of RNI d'Andringitra, was surveyed by Raxworthy (unpublished data) in 1993 between 1600 and 2640 m, which included both montane forest and heathland. Despite the fact that forest above 1200 m ele- vation was not surveyed in the Anosy Mountains, a large proportion of the Andringitra herpetofauna was recorded in these forests. Twenty-eight am- phibians (48%) and 19 reptiles (56%) were found in the Anosy Mountains, demonstrating that a sig- nificant component of the Andringitra fauna has a geographical range that extends to the southern limit of the eastern rain forest. The following spe- cies appear to drop out somewhere between the RNI d'Andringitra and the Anosy Mountains: Platypelis tuberifera, Heterixalus betsileo, Boo- phis albipunctatus, Boophis reticularis, Phelsuma quadriocellata quadriocellata, Uroplatus eben- aui, and Geodipsas heimi. Only a narrow elevation range of rain forest was surveyed at the PN de Ranomafana (900- 1050 m); nevertheless, many of the Andringitra species also occur at Ranomafana: 32 amphibians (55%) and 16 reptiles (47%). Undoubtedly, other Andringitra Massif species would have been re- corded if a greater elevational range of forest had been surveyed. Most of the species at RNI d'Andringitra, with the exception of the local en- demic species, are widely distributed along the eastern escarpment. Ankaratra, although 300 km north of Andrin- gitra Massif, has similar montane forests and heathlands at the same altitudes. Five amphibians (Mantidactylus aerumnalis, Boophis microtym- panum, B. goudoti, Boophis sp. 4, and Ptychadena mascareniensis) and four reptiles (Phelsuma bar- bouri, Calumma brevicornis hilleniusi, Furcifer campani, and Amphiglossus sp. 2) occur at both sites, representing 80% of the RNI d'Andringitra montane herpetofauna that occurs above 1800 m elevation. Two other species, Mantidactylus dom- erguei and Mabuya boettgeri, recorded at RNI d'Andringitra (see above), also occur at Ankara- tra. The similarity in montane herpetofauna is striking, especially considering the distance and current isolation between the two massifs. The lowest known elevation for the montane species occurring at both the Andringitra and An- karatra massifs (excluding Boophis goudoti, Boo- phis sp. 4, and Ptychadena mascareniensis) is be- tween 1600 and 1850 m elevation. Dispersal be- tween the two massifs is currently impossible for these eight montane species because of the exten- sive areas of relief between 1000 and 1200 m el- evation. The major low-elevation barriers for montane endemic species are the Fianarantsoa plateau (1000-1300 m) and the Ambalavao Pass (800-1000 m). This vicariant distribution pattern strongly suggests a period in Madagascar's history when the montane habitats of Andringitra and An- karatra formed a continuous belt between the massifs, most likely during glacial periods of the Pleistocene when cooler climatic conditions al- lowed montane habitats to expand into lower el- evations. The two high-altitude RNI d'Andringitra en- demics, Lygodactylus montanus and Anodonthyla montana, may have been more widespread during glacial periods. But, because no other populations are known, it appears that either they were not 168 FIELDIANA: ZOOLOGY Table 17-5. The distribution of Andringitra species (1993 survey) at other sites (see text). Site Table 17-5. Continued. Species Anka- Rano- Anosy ratra mafana Moun- 1600- 900- tains 2640 1050 50- m m 1200 m Amphibia Microhylidae Scaphiophryne marmorata Anodonthyla boulengeri Plethodontohyla bipunctata Plethodontohyla inguinalis Plethodontohyla alluaudi Plethodontohyla notosticta Platypelis turberifera Platypelis sp. 1 Mantellidae Mantella madagascariensis Mantidactylus aerumnalis Mantidactylus aglavei Mantidactylus asper Mantidactylus bicalcaratus Mantidactylus bertini Mantidactylus betsileanus Mantidactylus biporus Mantidactylus eiselti Mantidactylus femoralis Mantidactylus grandidieri Mantidactylus grandisonae Mantidactylus liber Mantidactylus lugubris Mantidactylus luteus Mantidactylus majori Mantidactylus opiparis Mantidactylus peraccae Mantidactylus pulcher Mantidactylus redimitis Mantidactylus ulcerosus Mantidactylus sp. 1 Rhacophoridae Aglyptodactylus madagascariensis Boophis albilabris Boophis albipunctatus Boophis boehmei Boophis goudoti Boophis luteus Boophis madagascariensis Boophis microtympanum Boophis miniatus Boophis reticulatus Boophis sp. 2 Boophis sp. 4 Hyperoliidae Heterixalus betsileo Ranidae Pytchadena mascareniensis Totals — Amphibia Site Species Anka- Rano- Anosy ratra mafana Moun- 1600- 900- tains 2640 1050 50- m m 1200 m Reptilia Gekkonidae Phelsuma barbouri * Phelsuma I. lineata * Phelsuma q. quadriocellata * Uroplatus ebenaui * Chamaeleonidae Brookesia nasus * * Brookesia superciliaris * * Calumma b. brevicornis * * Calumma b. hilleniusi * Calumma fallax * Calumma g. gastrotaenia * * Calumma nasuta * * Calumma oshaughnessyi * * Furcifer campani * Scincidae Amphiglossus anosy * Amphiglossus macrocercus * Amphiglossus melanopleura * * Amphiglossus melanurus * * Amphiglossus punctatus * * Amphiglossus sp. 2 * Mabuya gravenhorsti * * Colubridae Geodipsas heimi * Geodipsas infralineata * Geodipsas sp. 1 * Geodipsas sp. 2 * Liophidium rhodogaster * * Liopholidophis stumpjfi * Liopholidophis thieli * * Totals— Reptilia 4 16 19 33 28 successful at colonizing other massifs, or that oth- er populations have gone extinct or unrecorded. Conservation The most vulnerable elements of the RNI d' Andringitra herpetofauna are the six amphibians and five reptile species, and the two reptile sub- species considered to be endemic to the massif. The distribution of the endemics covers virtually the full elevational range of the reserve from 650 to 2350 m and all vegetation types. Conservation RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 169 of these species, therefore, makes it necessary to protect all habitats within the reserve. The most vulnerable habitat appears to be the low-elevation rain forest at the eastern limits of the reserve, which is threatened by clearing for tavy. By contrast, the high-elevation sclerophyl- lous forest (above 2000 m), now confined to small isolated patches, is vulnerable to damage from fire and possibly grazing from cattle. The high herpetofaunal diversity of RNI d'Andringitra (at least 95 species) and the degree of local endemicity (13% of the herpetofauna sur- veyed) demonstrate the importance of the reserve within the network of protected areas in Mada- gascar. The conservation of the natural habitats found within the Andringitra Massif will be crit- ical to maintaining Madagascar's existing biodi- versity. Acknowledgments We thank Angelin Razafimanantsoa and Nirhy Rabibisoa for their substantial help in the field, Steven Goodman for efficiently organizing the complicated logistics of the survey, and Olivier Langrand and Sheila O'Connor (World Wide Fund for Nature) for their strong support for this project. We are grateful to the officials of Direc- tion des Eaux et Forets, Association Nationale pour la Gestion des Aires Protegees, and the Universite d' Antananarivo for assistance during this survey. This research was supported in part by a grant from the National Science Foundation (DEB-90-24505). Literature Cited Blanc, C. P. 1977. Reptiles Sauriens Iguanidae. Faune de Madagascar, 45: 1-400. Blommers-Schlosser, R. M. A., and C. P. Blanc. 1991. Amphibiens (premiere partie). Faune de Mad- agascar, 75(1): 1-385. . 1993. Amphibiens (deuxieme partie). Faune de Madagascar, 75(2): 385-530. Brown, J. H. 1988. Species diversity, pp. 57-89. In Myers, A. A., and P. S. Giller, eds., Analytical bio- geography: An integrated approach to the study of an- imal and plant distributions. Chapman and Hall, Lon- don, 578 pp. Brygoo, E. R. 1978. Reptiles Sauriens Chamaeleonti- dae. Genre Brookesia et complement pour le genre Chamaeleo. Faune de Madagascar, 47: 1-173. . 1983. Systematique de lezards scincides de la region malgache. XI. Les Mabuya de Madagascar. Bulletin du Museum National d'Histoire Naturelle, Paris, 4th ser., sect. A., 5(4): 1079-1108. . 1985. Les Gerrhosaurinae de Madagascar Sau- ria (Cordylidae). M6moires du Museum National d'Histoire Naturelle. Ser. A, Zoology, 134: 1-65. Glaw, F, and M. Vences. 1994. A fieldguide to the amphibians and reptiles of Madagascar. Second edi- tion. Moss Druck, Leverkusen, Germany, 480 pp. Guibe, J. 1978. Les Batraciens de Madagascar. Bonner Zoologische Monographien, 11: 1-140. Jenkins, M. 1987. Madagascar, an environmental pro- file. International Union for the Conservation of Na- ture and Natural Resources, Cambridge, 374 pp. Millot, J., and J. Guibe. 1950. Les batraciens du nord de l'Andringitra (Madagascar). Memoires de l'lnsti- tute Scientifique de Madagascar. Ser. A, Biologie An- imate, 4(1): 197-206. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la Conservation et des Aires Protegees. World Wide Fund for Nature, Gland, 374 pp. Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, and P. Griveaud. 1971. Etude des ecosystemes mon- tagnards dans la region malgache. I. Le Massif de l'Andringitra. 1970-1971. Geomorphologie, climato- logie et groupement vegetaux. RCP 225. Bulletin de la Societe d'Ecologie, 2(2-3): 198-226. Pasteur, G. 1964. Notes preliminaires sur les Lygo- dactylus (Gekkonides). IV. Diagnoses de quelques for- mes africanes et malgaches. Bulletin du Museum Na- tional d'Histoire Naturelle, Paris, 2nd sen, 36(3): 311- 314. Raxworthy, C. J., and R. A. Nussbaum. 1994. A rain forest survey of amphibians, reptiles and small mam- mals at Montagne d'Ambre, Madagascar. Biological Conservation, 69: 65-73. 1995. Systematics, speciation, and biogeogra- phy of the dwarf chameleons (Brookesia Gray; Rep- tilia; Sauria; Chamaeleontidae) of northern Madagas- car. Journal of Zoology, London, 235: 525-558. White, F 1983. The vegetation of Africa. A descrip- tive memoire to accompany the UNESCO/AETFAT/ UNSO vegetation map of Africa. Natural Resources Research 20. UNESCO, Paris, 356 pp. 170 FIELDIANA: ZOOLOGY Chapter 18 The Birds of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar Steven M. Goodman and Michael S. Putnam Abstract An elevational transect of the birds occurring along the eastern slopes of the Reserve Na- turelle Integrate d'Andringitra was conducted using a variety of techniques: point counts, gen- eral observations, and mist-netting. A total of 99 species were recorded; three of these were boreal winter migrants to Madagascar. The resident birds included 82 species at 720 m, 66 at 810 m, 56 at 1210 m, and 46 at 1625 m. Although the ericoid and heathland zone from 1800 to 2200 m was not systematically surveyed, 20 species of birds were found there. Species accumulation curves showed that the majority of species (85%) recorded in each elevational zone were identified after 4 days of survey work. After 6 days at each elevation at least 93% of species were found. In all elevational zones additional species were added for up to 8 days after the start of each elevational transect. Information is also presented on a com- parison of mist-netting results between the elevational zones, composition of mixed-species foraging flocks, the effects of forest disturbance on species richness, human hunting pressure, and general natural history for selected species. Resume Un transect altitudinal destin6 a r£aliser l'inventaire des oiseaux a €\€ mis en place le long du versant est de la R6serve Naturelle Integrale d'Andringitra en utilisant di verses techniques: points de comptage, observations aldatoires et capture au moyen de filets. Un total de 99 especes a 6t6 recense" dont trois sont des migrateurs boreaux hivernant a Madagascar. Les oiseaux residents comptent 82 especes a une altitude de 720 m, 66 a 810 m, 56 a 1210 m et 46 a 1625 m. Meme si la zone de fourr6 encoi'de trouvee de 1800 m a 2200 m n'a pas €\€ syst6matiquement inventorize, 20 especes d'oiseux y ont 6t6 trouvees. Les courbes d'accumulation des especes montrent que la majority (85%) des especes r^per- toriees dans chaque zone altitudinale a pu etre identified apres 4 jours d'inventaire. Apres 6 jours, a chaque altitude, un minimum de 93% des especes a 6l6 inventoried Des especes sup- ptementaires ont 6t6 recensees dans toutes les zones d'altitude jusqu'a 8 jours apres le d£but de chaque transect altitudinal. Les informations relatives a la comparaison des r£sultats acquis au moyen de filets au sein des diffeientes zones d'altitude, a la composition des groupes plurispecifiques, aux effets de la perturbation de la forSt sur la richesse specifique, a la pression de chasse exercee par l'homme et a l'histoire naturelle de quelques especes selectionnees sont egalement presentees. GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 171 Introduction Madagascar is well known for its remarkable avifauna; over 50% of the 201 breeding birds are endemic (Langrand, 1990). In the past decade the amount of ornithological work on Madagascar has increased considerably, and much new informa- tion is available about the island's birds. Examples of new directions in Malagasy ornithology include systematic and phylogenetic studies (Prum, 1993; Schulenberg, 1993; Schulenberg et al., 1993), ecological work on species (Hawkins, 1993; Wat- son et al., 1993), and studies of community struc- ture (Andrianarimisa, 1993; Langrand & Wilme, 1993; Razafimahaimodison & Andrianantenaina, 1993). Furthermore, numerous inventories of pre- viously unknown or poorly known forest sites have been conducted (e.g., Thompson, 1987; Saf- ford & Duckworth, 1990; Thompson & Evans, 1991). However, little information is currently available on the ecological zonation of birds along an elevational transect. The intent of the present study was to examine the pattern of bird species richness along an ele- vational gradient on the eastern slopes of the Re- serve Naturelle Integrate (RNI) d'Andringitra, as well as to gather general information on the nat- ural history of the local avifauna. Both authors made general observations on the birds of the area. S.M.G. was responsible for bird netting (pre- sented in this chapter), and M.S.P was responsible for point count surveys (M. S. Putnam, unpub- lished data). Previous Work Little has been reported on the birds within the RNI d'Andringitra. The only known detailed ob- servations were made by Langrand, mostly on the west side of the reserve and in the high mountain area (Nicoll & Langrand, 1989). No ornithologist accompanied the La Recherche Cooperative sur Programme (no. 225) expedition to the area (Pau- lian et al., 1971). A Cambridge University student expedition to the area in 1993 compiled some un- published bird records (O'Keeffe & Ashmore, not dated). The majority of our work in the RNI d'Andrin- gitra was on the eastern side of the massif in hu- mid and montane forest. The humid forest site closest to the reserve that is ornithologically rea- sonably well known is the Pare National (PN) de Ranomafana, about 120 km to the northeast. In between these two sites is the forest of Ikongo (Fort Carnot), an important late 19th- and early 20th-century collecting locality, but for which lit- tle recent information is available. Methods Observations The main ornithological studies were conduct- ed by both authors during the first field trip to the reserve (23 September-1 November 1993), al- though general observations were made by S.M.G. during the second trip (14 November-18 December 1993). Daily field notes were kept on the species observed or heard in each transect zone, elevation range, and surrounding habitat(s). Altimeters and known positions on maps were used to calculate elevation. Each study zone in- cluded an elevational band of ±75 m, centered on a transect camp (720, 810, 1210, and 1625 m). The only exception was at our first camp, which was at 720 m, and which formed the lower ele- vational limit of the transect. Thus, there was broad elevational overlap between the zones re- ported herein as 720 and 810 m. See Chapter 1 for details concerning the itinerary of the expe- dition and Chapter 2 for descriptions of vegeta- tional zones. Netting Mist nets were operated in each transect zone, in a variety of microhabitats (ridge, slope, and valley bottoms), and within 1 km ground distance of camp. All nets used were 2.6 m high, 12 m long, and with 36-mm mesh size. A 24-hour pe- riod during which a net was left up continuously was termed a "net-day." Nets were regularly checked during daylight hours, before dawn, and after dusk. In each zone, nine nets were run for 5 days, making a total of 45 net-days per transect, and they were left up throughout the night to cap- ture bats and nocturnal birds. The bottom panel of each net was generally within 10 cm of or touching the ground. Net Captures and Specimens Most netted birds were marked for individual recognition and released. All birds caught on the 172 FIELDIANA: ZOOLOGY same day had the same primary feather marked with indelible ink; for example, birds caught on day 1 had the outermost primary feather on the left wing dyed. All birds were measured and weighed. Birds were released in the immediate vi- cinity of the net in which they were captured. Some birds were collected as voucher speci- mens or for systematic studies. These were pre- pared as either standard museum skins, fluid-pre- served specimens, or full skeletons. This material is housed in the Field Museum of Natural History (FMNH), and a representative portion of the col- lection will be returned to the Departement de Biologie Animale, University d' Antananarivo, Madagascar. Tissue samples of collected speci- mens were saved in liquid nitrogen or a buffered solution. Whole carcasses preserved in formalin were wrapped in fine cheesecloth before immer- sion to prevent mixing of ectoparasites (see Chap- ter 12). Blood smears were made from several released and collected birds for a study of blood parasites (see Chapter 13). Systematic Order, Nomenclature, and Common Names Except as noted, we follow the systematic order and English common and scientific names used by Langrand (1990). Results and Discussion General A total of 99 species were recorded on the east- ern slopes of the RNI d' Andringitra; three of these were boreal winter migrants to Madagascar (Table 18-1). The resident birds included 82 species at 720 m, 66 at 810 m, 56 at 1210 m, and 46 at 1625 m. Although the ericoid and heathland zone from 1800 to 2200 m was not systematically surveyed, 20 species of birds were found there. Further- more, Egretta alba, Egretta dimorpha, and Den- drocygna viduata were observed just outside of the reserve in the area between Col de Vohipia and Ambalamamenjana; these three species have not been included within the list. Combining the list presented by Nicoll and Langrand (1989) with our work, 1 14 species are known to occur in the RNI d' Andringitra. This number does not include Pseudocossyphus bensoni, reported from the re- serve by O'Keeffe & Ashmore (not dated), but for which insufficient details of the record are available. Species Addition Curves In advance of any detailed analysis of the or- nithological results of the 1993 mission, it was critical to determine whether our survey was suf- ficient to reflect the species richness for each ele- vational zone. By plotting the number of new spe- cies of birds added to each elevational zone per day of fieldwork, species addition curves could be calculated. Here we combined information from the point counts, netting, and general observa- tions. Each time we shifted camps, the first day in each zone was only partially devoted to bird surveys. Thus, we combined data for the first par- tial and the first complete day in each elevational zone. An examination of these curves (Fig. 18-1) shows that the majority of species (85%) recorded in each elevational zone were identified after 4 days of survey work. After 6 days at each ele- vation, at least 93% of the species were found (Table 18-2). However, in all cases additional spe- cies were added for up to 8 days after the start of each elevational transect. Two species were re- corded on the second field trip to the same portion of the reserve, when no detailed ornithological studies were carried out. These species were Fal- co zoniventris, in the 810 m elevation zone, and Dromaeocercus brunneus, in the 1625 m zone. Two points are clear: (1) there was a rapid de- crease in the addition of unrecorded species through time within each elevational zone; and (2) after 6 days of fieldwork by two ornithologists experienced with Malagasy birds, over 93% of the birds recorded in each zone had already been de- tected. On the basis of species addition curves for each elevational zone, we conclude that the local avifaunas were well documented. Species Richness There was marked elevational variation in bird species richness on the eastern slopes of the RNI d' Andringitra. A total of 99 species were recorded along an altitudinal gradient from 720 to 2200 m. No introduced species was recorded along the transect, although one, Acridotheres tristis, is known from other parts of the reserve and sur- GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 173 Table 18-1. Elevational distribution of birds on the eastern slopes of RNI d'Andringitra. [1800- Species 720 m 810 m 1210 m 1625 m 2200 m]* Tachybaptus pelzelnii + Phalacrocorax africanus'f + Bubulcus ibis + Ardea purpurea + Lophotibis cristatcr\ + + Aviceda madagascariensis + Milvus migrans + Polyboroides radiatus + + + Accipiter francesii'f + + Accipiter madagascariensis + Accipiter henstii + + + Buteo brachypterus + + + + A Falco newtoni + Falco zoniventris^ + Falco eleonoraeX + Falco concolor\% + Falco peregrinus + Turnix nigricollis + Mesitornis unicolorf + + + + Sarothrura insularis + + + + H Canirallus kioloidest + + Actitis hypoleucosX + Streptopelia picturata + + + + Treron australis^ + Alectroenas madagascariensis + + Coracopsis nigra + + + + Coracopsis vasa + + + Agapornis cana\ + Cuculus audebertif + Cuculus rochii + + + + F, H Coua reynauditf + + + + Coua caerulea + + + + Centropus toulou + + Tyto alba + Otus rutilus + + + + Asio madagascariensis + + Caprimulgus madagascariensis^ + Zoonavena grandidiertf + + + + A Cypsiurus parvus^ + Apus barbatus + + A Apus melba + + + + A Alcedo vintsioides + + Ispidina madagascariensis^ + + + Merops superciliosus + + + Eurystomus glaucurus + + Brachypteracias leptosomusf + + Brachypteracias squamigerf + Atelornis pittoides + + + Atelornis crossleyif + + Leptosomus discolor + + + + A Philepitta castanea~\ + + + + Neodrepanis coruscans^ + + Neodrepanis hypoxantha + + F, H Phedina borbonica + A Motacilla flaviventris + + Coracina cinerea + + + + Phyllastrephus madagascariensis + + + Phyllastrephus zosterops + + + Phyllastrephus cinereiceps^ + + + Hypsipetes madagascariensis + + + + F,H Copsychus albospecularis + + + Saxicola torquata + H 174 FIELDIANA: ZOOLOGY Table 18-1. Continued. [1800-2200 Species 720 m 810 m 1210 m 1625 m m]* Pseudocossyphus sharpei + + + + Nesillas typica + + + + F, H Cisticola cherina + Cryptosylvicola randrianasoloif +§ + + F Dromaeocercus brunneus + Dromaeocercus seebohmif H Randia pseudozosteropsi + + + Newtonia amphichroai + + + + F, H Newtonia brunneicauda + + + + Neomixis tenella + + + Neomixis viridisf + + + + Neomixis striatigulai + + + + Hartertula flavoviridisX + + + + Terpsiphone mutata + + + + Pseudobias wardi + + + Oxyiabes madagascariensisX + + + + Crossleyia xanthophrysX + + + + Mystacornis crossleyif + + + Nectarinia notatai + + + + F, H Nectarinia souimanga + + + + F, H Zosterops maderaspatana + + + + F H Calicalicus madagascariensis + + + + Schetba rufaX + + + Vanga curvirostrisX + + + + Xenopirostris pollenif + Leptopterus viridis + + + + Leptopterus chabert + + + Cyanolanius madagascarinus + + Hypositta corallirostrisX + + Tylas eduardi + + + + Dicrurus forficatus + + + + Corvus albus + + H Hartlaubius auratus + Ploceus nelicourvi + + + + Foudia madagascariensis + H Foudia omissai + + + Lonchura nana + Total number of species 82 66 56 46 20 * The elevational zone in brackets was not part of the standard transect, and records from these zones were not obtained in a systematic manner. The habitats include the following: F = upper montane forest; H = heath; A = aerial. t Species whose names are followed by a dagger (t) were not listed for the reserve by Nicoll and Langrand (1989). X Species whose names are followed by a double dagger (X) are boreal winter migrants to Madagascar. § Recorded at the upper limit of the transect zone. rounding areas. The numbers of species recorded in each elevational zone (Table 18-1) were as fol- lows: 82 in the 720 m zone, 66 in the 810 m zone, 56 in the 1210 m zone, 46 in the 1625 m zone, and 20 in the zone between 1800 and 2200 m (Fig. 1 8-2). Thus, species richness decreased as a function of altitude (linear regression, P = 0.007, r2 = 0.94, N = 5; or, excluding the 720 m datum point, the zone of disturbed forest, P = 0.038, r2 = 0.93, N = 4). This is the same basic pattern found throughout the tropics, whether on conti- nental land masses or islands (e.g., Chapman, 1917; Terborgh, 1977; Diamond & LeCroy, 1979; Goodman & Gonzales, 1990), and in North Amer- ica (Able & Noon, 1976; Finch, 1989). Segregation of Congeners Along an Elevational Gradient Of the 76 genera of resident birds known from the eastern slopes of RNI d'Andringitra, 16 in- GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 175 90- 80- CO (!) 70 *w* o 4> Cu 00 60 <*-H o 1- 1 50 a 3 40 Z 30- 20' 10 11 12 Working Days Fig. 18-1. Species addition curves for four elevational zones. Data from transect counts, general observations, and mist-netting are combined in this analysis. eluded more than one species. Very few of these genera contain species with non-overlapping ele- vational ranges, but in several cases these disjunct distributions may be an artifact of the species be- ing rare and only recorded on a few occasions. However, there are a few examples of apparent altitudinal replacement of relatively common spe- cies. Neodrepanis coruscans was recorded in the 720 and 810 m zones and not above 900 m ele- vation, whereas N. hypoxantha was found in the 1210 and 1625 m elevational zones. Thus, the al- titudinal ranges of these two species almost abut, and these two species appear to replace one an- other. Two species in each of three different genera were found across the complete elevational gra- dient of forested habitat from 720 to 1625 m el- evation (Coua, Newtonia, and Nectarinia). Within these three genera there are behavioral or mor- phological features that separate congeners: Coua Table 18-2. Percentage of new species added to each elevational zone after 4 and 6 days' work in a zone. Percentage of new species added Elevation (m) After 4 days After 6 days 720 810 1210 1625 14 15 5 4 caerulea is arboreal, whereas C. reynaudii is ter- restrial; Newtonia brunneicauda is a mid- to up- per-canopy species, whereas N. amphichroa in- habits the understory; and Nectarinia notata has a distinctly longer bill and is more massive than N. souimanga. Other congeneric species pairs have complete or nearly complete overlapping 100 2500 Elevation (m) Fig. 18-2. Plot of total number of species per ele- vational zone. The straight line shows the relationship y = -0.04 Ix + 106.7 (r2 adjusted = 0.94, P = 0.007), where y = number of species and x = elevation in meters. 176 FIELDIANA: ZOOLOGY Table 18-3. Netting results presented by elevational zone. 720 m 810 in 1210 m 1625 m Total Species (45)* (45)* (45)* (45)* (180)* Otus rutilus 1 1 2 Alcedo vintsioides 4 4 Philepitta castanea 6 7 11 9 33 Neodrepanis coruscans 3 3 N. hypoxantha 2 3 5 Phedina borbonica 7 7 Motacilla flaviventris 1 1 Coracina cinerea 1 1 Phyllastrephus madagascariensis 2 8 10 P. zosterops 12 17 6 35 P. cinereiceps 3 3 6 Hypsipetes madagascariensis 4 5 9 Copsychus albospecularis 13 10 3 26 Pseudocossyphus sharpei 2 2 Nesillas typica 1 1 6 8 Cryptosylvicola randrianasoloi 1 1 Newtonia amphichroa 1 2 2 5 N. brunneicauda 1 1 Terpsiphone mutant 12 3 1 16 Oxylabes madagascariensis 5 1 3 9 Crossleyia xanthophrys 1 1 2 Nectarinia souimanga 2 1 4 3 10 Zosterops maderaspatana 5 3 2 10 Vanga curvirostris 2 1 1 4 Cyanolanius madagascarinus 2 2 Hypositta corallirostris 1 1 Dicrurus forficatus 3 3 6 Ploceus nelicourvi 5 3 2 10 Foudia omissa 2 5 3 10 Total individuals 94 62 42 42 240 Total species 23 13 13 15 30 Average birds/net-day 2.1 1.4 0.9 0.9 1.3 * The number in parentheses is the total number of net-days accrued for that zone. elevational ranges (Coracopsis, Neomixis, and Leptopterus). In general, there were few records of ground- rollers, so it is difficult to compare their eleva- tional ranges. Both species of Brachypteracias and Atelornis pittoides tend to be lowland or mid- montane in distribution, whereas A. crossleyi is more common at higher elevations. The distribu- tion of Foudia seems to be related to forest cover. Foudia madagascariensis was found only in the disturbed areas at 720 m and in the open heath- land above 1950 m, whereas F. omissa occurred in forested areas between 810 and 1625 m ele- vation. One of the more interesting cases of differences in the altitudinal ranges of congeneric species is with Phyllastrephus. Phyllastrephus madagascar- iensis and P. zosterops were found at the 720, 810, and 1210 m elevational zones. On the basis of netting results, both of these species were more common at 810 m than 720 m, with P. zosterops being the more common of the two (Table 18-3). At 1210 m P. madagascariensis was rare and ob- served only a few times; P. zosterops was also decreased in relative abundance. The third spe- cies, P. cinereiceps, was rare at 810 m, where it was only observed, and three individuals were netted in both the 1210 m and 1625 m zones. Thus, in sympatry there are some differences be- tween Phyllastrephus species in their relative abundances. In other zones there appears to be species replacement. Elevational Differences in Species Richness Based on Netting Mist nets were used to assess species richness of lower understory birds within four transect zones (Karr, 1981). In total, 180 net-days were GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 177 Table 18-4. Mass measurements (g) of birds netted during the Andringitra survey. Minimum- Standard Species N maximum* Mean deviation Otus rutilus 2 85.5, 110 Alcedo vintsioides 4 17.5-19.5 18.38 0.85 Atelornis crossleyi 1 77.5 Philepitta castanea 33 27.0-42.5 35.02 3.64 Neodrepanis coruscans 4 6.2-6.6 6.43 0.17 N. hypoxantha 5 6.4-8.1 7.18 0.67 Phedina borbonica 7 20.5-25.0 22.86 1.60 Motacilla flaviventris 1 24.5 Coracina cinerea 1 46.5 Phyllastrephus madagascariensis 11 20.5-37.0 30.50 5.22 P. zosterops 29 14.0-20.0 17.20 1.60 P. cinereiceps 6 16.5-24.5 19.33 2.77 Hypsipetes madagascariensis 8 41.5-50.0 45.81 2.74 Copsychus albospecularis 23 18.5-27.5 23.48 2.26 Pseudocossyphus sharpei 2 26.5, 25.0 Nesillas typica 8 16.5-21.5 18.38 1.53 Cryptosylvicola randrianasoloi 2 8.2, 8.3 Newtonia amphichroa 5 11.5-14.5 12.80 1.20 N. brunneicauda 1 12.5 Terpsiphone mutata 17 12.5-15.0 13.65 0.72 Oxylabes madagascariensis 8 19.0-26.0 22.75 2.56 Crossleyia xanthophrys 2 21.5, 21.5 Nectarinia souimanga 7 5.5-9.5 7.47 1.19 Zosterops maderaspatana 11 9.5-14.5 11.55 1.37 Vanga curvirostris 1 67 Cyanolanius madagascarinus 2 22.0, 24.5 Hypositta corallirostris 1 15.5 Dicrurus forficatus 7 41.5-50.5 47.43 3.17 Ploceus nelicourvi 7 21.0-25.5 23.57 1.69 Foudia omissa 10 18.0-21.4 19.45 1.48 * In cases where three or fewer measurements are available, only the masses themselves are given. accrued. The numbers of birds captured and the capture rate per transect zone (total captured/total number of net-days) were 94 and 2.1 at 720 m, 62 and 1.4 at 810 m, 42 and 0.9 at 1210 m, and 42 and 0.9 at 1625 m. The masses of captured birds are presented in Table 18-4. Within the first three transect zones there was a pattern of decreasing bird capture rate as a func- tion of increasing elevation. However, the results from the 1625 m transect zone were identical to those from 1210 m, and on the basis of the current data, understory bird density, if indeed it could be measured by capture rates, was similar in these two zones. The number of species netted within each zone varied, from 23 at 720 m, to 13 at 810 m, 13 at 1210 m, and 15 at 1625 m. On mountains in the tropics, there is generally a strong negative relationship between bird species richness and el- evation (see p. 175), and the patterns emerging from mist-netting on the Andringitra Massif do not follow this pattern. However, this technique introduces several types of biases into interpreta- tions of avifaunal patterns (Karr, 1981; Graves et al., 1983; Remsen & Parker, 1983), particularly by sampling only a small proportion of the species at a given elevation and specifically those occu- pying a subset of available habitats. However, in- traspecific differences in the capture rates between elevational zones of certain species parallel changes in species richness as measured by point counts and general observations. Mixed-Species Flocks Nineteen mixed-species flocks were recorded in the four elevational zones (Table 18-5). Flock members were classified as being either canopy or understory foragers, or as foraging at both lev- els (Table 18-6). Canopy flocks contained species that forage in the lower to upper canopy. Five flocks contained canopy foragers plus one or two species that foraged at both levels and were clas- sified as canopy flocks. Understory flocks con- 178 FIELDIANA: ZOOLOGY Table 18-5. Elevational distribution of different types of mixed-species flocks in RNI d'Andringitra. Eleva- tion (m) Flock type Canopy Combined Understory Total 720 1 2 0 3 810 1 1 2 4 1210 4 3 3 10 1625 0 1 1 2 Totals 6 7 6 19 tained species that forage in the understory. Five flocks contained one of four understory foragers and a bilevel forager (Phyllastrephus cinereiceps); these were considered understory flocks. Thirty-seven species were recorded in the mixed-species flocks (Table 18-6). This total in- cludes three species (Coracopsis vasa, Pseudo- cossyphus sharpei, and Newtonia amphichroa) not previously reported as members of mixed-spe- cies flocks (Langrand, 1990; Eguchi et al., 1992, 1993; Razaflmahaimodison & Andrianantenaina, 1993; Goodman, pers. obs.). Newtonia amphich- roa was thrice found in combined flocks. The presence of the other two species may have been Table 18-6. Members of mixed-species flocks, the level at which they forage, and their percentage of occurrence. The possible number of flocks in which they might occur is based on the number of flock types (Table 18-5) in the elevational range of each species (Table 18-1). Foraging Actual no. Possible no. Percentage Species level* of flocks of flocks of flocks Coracopsis vasa C 1 12 8 Coua reynaudii U 1 12 8 Coua caerulea C 1 13 8 Philepitta castanea B 1 19 5 Neodrepanis coruscans C 1 5 20 Neodrepanis hypoxantha C 2 8 25 Coracina cinerea C 3 13 23 Phyllastrephus madagascariensis U 2 12 17 Phyllastrephus zosterops U 1 10 10 Phyllastrephus cinereiceps B 7 16 44 Hypsipetes madagascariensis C 5 13 38 Pseudocossyphus sharpei U 1 12 8 Nesillas typica U 1 12 8 Cryptosylvicola randrianasoloi c 2 10 20 Randia pseudozosterops c 4 12 33 Newtonia amphichroa u 3 12 25 Newtonia brunneicauda c 3 13 23 Neomixis tenella c 2 12 17 Neomixis viridis c 2 13 15 Neomixis striatigula c 2 13 15 Hartertula ftavoviridis u 4 12 33 Pseudohias wardi c 4 12 33 Terpsiphone mutata c 3 13 23 Oxylabes madagascariensis u 5 12 42 Nectarinia souimanga c 5 13 38 Nectarinia notata c 1 13 8 Zosterops maderaspatana c 8 13 62 Calicalicus madagascariensis c 6 13 46 Vanga curvirostris B 1 19 5 Leptopterus viridis c 5 13 38 Leptopterus chabert c 1 12 8 Cyanolanius madagascarinus c 4 12 33 Hypositta corallirostris c 1 5 20 Tylas eduardi c 7 13 54 Dicrurus forficatus c 6 13 46 Ploceus nelicourvi B 5 19 26 Foudia omissa u 2 10 20 C = canopy forager, U = understory forager, B = forages at both levels. GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 179 coincidental (e.g., Coua reynaudii with Pseudo- cossyphus sharpei or Coracopsis vasa with a can- opy flock). The number of species in each flock type varied. Canopy flocks ranged from three to 12 species (7.8 ± 3.5, mean ± standard deviation, N = 6), and combined flocks from two to 16 (7.9 ± 4.9, N = 7), whereas understory flocks always had two species (N = 6). Canopy and combined flocks each had on average more species than un- derstory flocks, but they did not differ from each other in this respect (Mann- Whitney test: W = 43, P = 0.94). Individual species occurred in 5% to 62% of the mixed-species flocks (Table 18-6). Razafima- haimodison and Andrianantenaina (1993, Table 3) classified species into four groups based on the frequency with which they occurred in mixed-spe- cies flocks. We made too few observations to sim- ilarly classify all of the species that were ob- served. However, we found two canopy species (Zosterops maderaspatana and Tylas eduardi) that occurred in over 50% of the canopy and com- bined flocks and correspond to their classification of a constant or very frequent member ("especes constantes"). Differences Between Disturbed and Undisturbed Forest The placement of our elevational zones was de- signed in part to investigate the potential effects of habitat degradation in forested areas at approx- imately the same elevation and botanical com- munities, but with differences in the level of hu- man disturbance. The 720 m zone was in an area adjacent to a small settlement of slash-and-burn (tavy) agriculturalists, and close to the Iantara River and a trail running along the eastern side of the RNI d'Andringitra. The 810 m site showed fewer signs of human disturbance (see Chapter 4 for descriptions of these two botanical commu- nities). We documented the presence of 82 resident bird species in the 720 m zone and 66 resident bird species in the 810 m zone (Table 18-1). Twenty-four resident species were found at 720 m and not at 810 m. Of these species, three were associated with relatively large and slow-moving rivers {Tachybaptus pelzelnii, Phalacrocorax af- ricanus, and Ardea purpurea), a habitat that did not occur in the 810 m zone. Fourteen of the 24 species were associated with open areas or hu- man-modified habitats: Bubulcus ibis, Milvus mig- rans, Falco newtoni, Turnix nigricollis, Centropus toulou, Agapornis cana, Tyto alba, Caprimulgus madagascariensis, Cypsiurus parvus, Phedina borbonica, Saxicola torquata, Cisticola cherina, Corvus albus, and Lonchura nana. The balance included aerial foragers that were almost certainly missed at 810 m because of the closed canopy (Apus spp.), relatively rare species found at ele- vations higher than 810 m and so presumably overlooked at that site (Brachypteracias leptoso- mus), and those whose upper elevational limit in the reserve may have been 720 m (Brachyptera- cias squamiger). The absence of three species at 810 m that were found at 720 m (Treron australis, Cyanolanius madagascarinus, and Hartlaubius auratus) may simply have been due to chance. Eight species were found at 810 m and not at 720 m. All of these were forest-dwelling birds, two of which had their lower elevational limits at the upper end of the 810 m zone (Phyllastrephus cinereiceps and Cryptosylvicola randrianasoloi). Four species were very rare along the eastern slopes of the RNI d'Andringitra: Accipiter mad- agascariensis, Falco zoniventris, Cuculus aude- berti, and Xenopirostris polleni, and their appar- ent absence at 720 m may again be chance. The remaining two species, Asio madagascariensis and Foudia omissa, have been recorded below 810 m at other sites on Madagascar, and their ab- sence from the 720 m zone cannot be simply a question of altitudinal distribution. Now we return to the question of whether there were any noticeable additions or deletions to the avifaunas of the 720 and 810 m zones that could be attributed to human habitat modification. The major difference in these zones is the presence of about 14 species in the disturbed areas, particu- larly in the immediate area of the Ambarongy tavy, that were not found at 810 m. These were all species associated with heavily modified en- vironments. The areas bordering the Iantara River below our 720 m camp had been heavily dis- turbed, and presumably this disturbance provided a corridor from the lowlands and coastal area for species to colonize the open areas and forest edge near Ambarongy. With the current data it is dif- ficult to demonstrate the disappearance of forest species found at 810 m from the 720 m zone, even though there is some evidence to suggest it. Bird Hunting The trail between Ambatomboay and Ambala- manenjana is regularly traveled by people carry- 180 FIELDIANA: ZOOLOGY ing supplies between the two villages. Some days up to 20-30 people moved along the trail, many of whom carried slings to hunt birds and, occa- sionally, lemurs. Generally, if a shot bird was still alive, the remiges would be removed and the bird carried to a resting site along the trail, where it would be cooked and eaten. Numerous feather re- mains were found in the vicinity of fire pits at the Korokoto River crossing. As we traveled up and down this trail we re- covered the feathers of recently plucked birds. Thus we were able to identify species and roughly quantify the number of individuals taken. Records have been divided into two sections of the trail: (1) Ambalamanenjana to Col de Vohipia — Coua caerulea, two individuals, and Atelornis pittoides, two individuals; (2) Col de Vohipia to 720 m camp — Accipiter francesii, one individual; Coua caerulea, seven individuals; Coua reynaudii, four individuals; Atelornis pittoides, two individuals; Brachypteracias squamiger, one individual; and Coracina cinerea, one individual. Many of the feather remains were found after the start of our surveys, and this apparent rise in hunting activity was almost certainly related to an increase in traffic along the trail associated with porters supplying and shifting our camps. Feather remains of a plucked Canirallus kioloides and a Coua reynaudii were found within the reserve along the trail leading from the 810 to 1210 m camps. These remains are almost certainly those of porters' meals. Although we found no signs of it, several local people mentioned that Lophotibis cristata is captured for its flesh using snares and a variety of different traps. With the exception of Coracina, all of the spe- cies hunted are relatively large-bodied birds (> 100 g), and the majority tend to be ground-dwell- ing, or at least occur in the lower canopy of the forest. Furthermore, many of these species, when disturbed, fly to a nearby perch and remain mo- tionless; they are then easily shot by a keen-eyed hunter. In conclusion, there is hunting pressure on couas, ground-rollers, and other relatively large- bodied birds along trails that pass through forest- ; ed areas. is already available in the general literature on the Malagasy avifauna, we have concentrated on new or interesting natural history aspects of selected species. Tachybaptus pelzelnii The Madagascar Little Grebe was found at sev- eral sites along the Iantara River from near the junction of the Korokoto River to below Amba- tomboay. These sites were generally in areas where the river was relatively broad and slow- moving. On 13 September 1993, near Ambaron- gy, at 720 m elevation, two birds in breeding plumage were observed performing a courtship display, which included head-tossing behavior. Although this species was observed along the Ian- tara during both periods we occupied the 720 m camp, no definite evidence of local breeding was found. Phalacrocorax africanus The Reed Cormorant was observed on a few occasions along the Iantara River. On the large slow-moving rivers outside of the reserve, such as the Manambolo and the Mananantanana, this spe- cies was relatively common. Lophotibis cristata This species, the Madagascar Crested Ibis, was not commonly detected in the reserve. It was found only in the 720 and 810 m zones, generally along riverbanks or in areas with wet ground. Aviceda madagascariensis The Madagascar Cuckoo-falcon was recorded on one occasion at 1210 m elevation. The bird was observed flying across the Volotsangana Riv- er valley. Selected Species Accounts Polyboroides radiatus Rather than reiterating general information on the breeding season, food habits, and distribution of birds occurring in the RNI d'Andringitra that Between 5 and 13 October 1993, two adult Madagascar Harrier Hawks were observed near to or resting on a nest at 850 m elevation and about GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 181 20 m from the Sahavatoy River. We found no ev- idence that the pair was breeding during our visit. Accipiter henstii On 18 November 1993, at 720 m elevation, an adult Henst's Goshawk was observed flying over the forest behind the Ambarongy tavy. The bird called continuously as it flew in circles over the canopy. This behavior continued for about 10 min. On 22 October 1993, at 1210 m and 10:00 am, near the 1210 m camp, a shadow was observed moving over the trail, and a large raptor was found perched in a tree 5-8 m away and 6 m off the ground. The bird had white extending from the throat to the vent. The white underparts were barred with dark gray from the chin or throat to the vent. The tail was long and alternately barred with light and dark gray bars; the latter bars were narrower. The feet were yellow and the talons dark. The bird was observed for 15 s or less be- fore it flew off. The wings were broad and round- ed. They appeared dark brown with no noticeable patterning. Neither the rump nor the upper side of the tail was observed. There was no visible ruff of nape feathers, although the bird was seen only frontally. The only two species that fit this de- scription are Accipiter henstii and Eutriorchis as- tur (the Madagascar Serpent Eagle). Eutriorchis astur is exceptionally rare and until recently was thought to be on the verge of ex- tinction. In the past few years it has been found at sites in northeastern Madagascar (Sheldon & Duckworth, 1990; Raxworthy & Colston, 1992; Thorstrom et al., 1995). Whether the 22 October 1993 observation was of this species remains in- conclusive. However, on the basis of forest type and size, it is quite possible that E. astur inhabits the RNI d'Andringitra. er Malagasy rain forests, Accipiter spp. were ex- ceptionally uncommon in the RNI d'Andringitra. France's Sparrowhawk was recorded in the 720 and 810 m transect zones. Buteo brachypterus While being watched by an arboreal group of brown lemurs (Eulemur fulvus) giving alarm calls, one of us (M.S.R) whistled imitations of the Mad- agascar Buzzard's call. Before the imitations were given, the lemurs looked only toward the ground. During and following imitations of the hawk's call the lemurs alternated between looking toward the ground and scanning the sky. This observation suggests that brown lemurs recognize the calls of potential avian predators. A recent review (Goodman et al., 1993) re- ported a brown lemur group giving alarm calls upon seeing a large raptor and a case of an adult being killed by an unidentified raptor. Harrington (1975) reported brown lemurs looking at flying Buteo and sometimes giving soft grunts. Two species of lemurs in captivity {Lemur catta and Varecia variegata) vocalized in response to the calls of North American Buteo spp. (Macedonia, 1993). Falco zoniventris The only record we have of the Banded Kestrel in the reserve is one observed on 24 November 1993 at the forest edge near the Sahanivoraky River, just below the 810 m camp. Falco peregrinus Accipiter madagascariensis The only record we have of the Madagascar Sparrowhawk in the reserve is one observed at 810 m elevation on 10 October 1993. Accipiter francesii Accipiter spp. were seldom observed, and none was netted. On the basis of our experience in oth- This species, the Peregrine Falcon, was ob- served only in the 1210 m zone. The numerous sheer cliff faces in the reserve might provide nest- ing sites for this species. Turnix nigricollis During the survey, we only found the Mada- gascar Buttonquail in or near the tavy at Amba- rongy, at 720 m elevation. 182 FIELDIANA: ZOOLOGY Mesitornis unicolor Coracopsis nigra The Brown Mesite was recorded between 720 and 1625 m elevation, although on the basis of general observations it appeared to be less com- mon above 1210 m. The previous highest reported elevation for this species was 900 m (Langrand, 1990). On occasion the Brown Mesite was ob- served on tree limbs or suspended dead branches 1-5 m off the ground. Several times birds were scared up from the forest floor. They would often fly 10 or 20 m, land on a tree perch, and remain motionless. After a period of a few minutes they would walk deftly along branches and then jump to the ground. On several occasions birds were ob- served with the atypical plumage coloration de- scribed by Langrand (1990, plate 15, fig. 77b, p. 150), as well as with gray rather than pale orange or black bills. The Lesser Vasa Parrot was observed feeding on the following plants: Cinnamosma fragrans (Canellaceae) — at 1650 m, they would remove exocarp from seed and consume the latter; Sym- phonia sp. (Clusiaceae) — at 870 m, they fed on unopened blossoms and left behind only the flower base and petiole; and Chassalia sp. (Ru- biaceae) — at 1210 m, they consumed whole fruits. Coracopsis vasa The Greater Vasa Parrot was distinctly less common than C. nigra and was not recorded above 1210 m elevation. Sarothrura insularis On the basis of general observations and calls, this species, the Madagascar Flufftail, was record- ed in all forested zones between 650 and 1 800 m. A few individuals were heard calling in areas of tavy near Ambarongy at 720 m and in the heath- land at about 2200 m. On 28 October 1993, at about 1700 m elevation, one individual was ob- served carrying dried grass in its bill. Streptopelia picturata On 10 October 1993, at 810 m elevation, bro- ken eggshells of the Madagascar Turtle Dove were found below a nest. The clutch had hatched that morning. This species was observed at 710 m elevation feeding on whole fruits of Ficus re- flexa (Moraceae). Alectroenas madagascariensis Identified Madagascar Blue Pigeon food items included Ficus reflexa (Moraceae) — whole fruits, at edge of clearing, 710 m; Cryptocarya sp. (Lau- raceae) — whole fruits, at 810 m; Pachytrophe sp. (Moraceae) — whole fruits, at 730 m; Maesa Ian- ceolata (Myrsinaceae) — whole fruits, at 740 m; and Oncostemum racemiferum (Myrsinaceae) — whole ripe fruits, near forest edge, 730 m (see Chapter 4, Appendix 4-2). Agapornis cana This species, the Gray-headed Lovebird, was only recorded in disturbed areas at the lower el- evations, including the Ambarongy tavy at 720 m, and near Ambatomboay at 650 m elevation. Cuculus audeberti On 10 October 1993, at 820 m and 8:15 am, a Thick-billed Cuckoo was observed perched in the top of a 20-m-tall tree about 500 m from the edge of the Sahavatoy River. The observer had an unobstructed view of the bird and watched it for 2 or 3 minutes. Notes made at the time describe its vocalization as "a single strong note followed by a weaker descending note, kind of Green Sand- piper [Tringa ochropus] like in quality, and given in rapid succession and in couplets of two to four." The bird was in adult plumage, with a dis- tinctly white and unbarred ventrum, a dark char- coal dorsum, and a bicolored bill. It was easily distinguishable from Cuculus rochii or other spe- cies in Madagascar with which it could be con- fused. The only other record of this species in Madagascar after 1922 (Langrand, 1990) is from the Maromiza Forest near Analamazaotra/Pennet, about 400 km to the northeast of the eastern part of the RNI d'Andringitra, in late November 1992 (Langrand & Sinclair, 1994). A tape recording of the Thick-billed Cuckoo made in South Africa was broadcast at 15 of the 30 census points in the 720 and 810 m elevational GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 183 zones. A broadcast was also made within 1 hour of the above encounter at the location of the ob- servation. None of these tape playbacks elicited a vocal response. this migratory species was relatively common along the Sahavatoy River near 810 m and in the vicinity of the 720 m camp. During the second trip to the reserve, it was regularly observed or heard in the 720 and 810 m zones. Cuculus rochii This species, the Madagascar Lesser Cuckoo, was relatively common in all forested zones be- tween 650 and 1800 m. A few individuals were heard calling in the heathlands at about 2200 m. Tyto alba The Barn Owl was recorded only near the 720 m camp. On several nights this bird was heard vocalizing from the direction of the Ambarongy tavy. Asio madagascariensis A roost or hunting perch of this species, the Madagascar Long-eared Owl, was found on a ridge at about 900 m elevation, just above the 810 m camp. The bird was seen in the same tree on several nights during the course of 1 week. Below the perch, two whole and several disintegrated pellets were found that contained the remains of at least two adult and two subadult Rattus rattus. A Madagascar Long-eared Owl was heard calling near the village of Ambatomboay during the night of 13-14 September 1993. Caprimulgus madagascariensis On the basis of vocalization, the Madagascar Nightjar was relatively uncommon along the Ian- tara River in the vicinity of our 720 m camp. This species appeared to be more common near Am- batomboay, where on the nights of 13 and 14 Sep- tember 1993, numerous individuals were heard calling. Eurystomus glaucurus Between 23 September and 13 October 1993, the Broad-billed Roller was not recorded in the 720 and 810 m zones. On 1 November 1993, as the first research group was leaving the reserve, Brachypteracias leptosomus On 21 October 1993, in the 1210 m zone, one Short-legged Ground-roller was observed perched at a height of 10 m and holding in its bill a 10- to 15-cm-long black millipede with yellow legs. The bird shook the millipede and seemed to strike it against a branch before swallowing the animal whole. The bird then flew a short distance away. During the time it was dispatching its prey, the ground-roller was mobbed by two Neodrepanis hypoxantha and several Zosterops maderaspatana and Nectarinia souimanga. These birds moved about, in positions close to the ground-roller, and gave repeated shrill calls. Brachypteracias squamiger On 28 September 1993, two Scaly Ground-roll- ers were observed at 720 m elevation, about 850 m from our camp and 10 m from the trail that led from Ambatomboay to Ambalamanenjana. At least one of two birds dropped from a perch to the ground and then ran away in short dashes, followed by freezing postures. A tape recording of this species was played back to the pair. Soon one bird, and then the second, began to approach. The first bird progressed toward the observer with its head lowered to the same level as its back and with its tail spread. The bird occasionally changed direction as it ran on the ground, keeping the tail fanned throughout the sequence. Minutes later the bird made harsh grating sounds. Until this point the tape recording was being played; soon after the tape recording was stopped, the bird began calling. A long sequence of vocalizations was giv- en, similar to the "coo" calls on the playback tape. The bird also performed a wing-flick dis- play: with the bird standing in a regular, head-up posture and the tail spread, the folded wings were rapidly raised above the back and then lowered to a normal position. While giving a short "coo" call, the bird would I begin from a typical upright posture with the head up and the bill parallel to the ground. The head was lowered to a point where the bill was 45° 184 FIELDIANA: ZOOLOGY below its starting position. At the bottom of the bill's trajectory the "coo" call was given, seem- ingly with the bill closed. The head and bill were then returned to their starting position. This site was in an area of forest along the Ian- tara River with gently sloping ground, an open un- derstory, and nearly complete leaf litter coverage of the ground. On 16 November 1993, within a few meters of where the behavioral observations of this species were made on 28 September 1993, a bur- row was found that contained two Eliurus webbi, a nesomyine rodent (see Chapter 22). On 20 No- vember 1993, the burrow was excavated, and the nest of a Scaly Ground-roller was found within the system (Goodman, 1994). On the basis of the lay- out of the tunnel, it was clear that the ground-roll- ers and rodents were cohabiting the burrow. A sin- gle egg was resting in a shallow depression, about 2 cm deep, and surrounded by a platform of small dried leaves. The placement of this nest was similar to that reported in another nesting record of this species (Benson et al., 1976). Atelornis pittoides The Pitta-like Ground-roller was regularly heard at the edge of the Ambarongy village clear- ing. This species appeared to be the least sensitive to human disturbance of the four ground-rollers recorded in the reserve. Atelornis crossleyi On 19 October 1993, at 1230 m elevation, B. Fisher (Chapter 8) was searching for ants in fallen wood. The principal technique was to open rotten logs resting on the ground. For about 2.5 h, a Rufous-headed Ground-roller followed him. After a site was abandoned, the ground-roller would ap- proach the log, often within a few meters of B. Fisher, and peck at areas of the wood that had been opened. Several times the bird was observed exposing and eating beetle larvae removed from the rotten wood. It deftly climbed and jumped over objects, such as large logs and a backpack, on the forest floor. Two birds were observed, al- though one was more retiring than the other and did not approach closely. Little is known about the diet of this species. Langrand (1990) stated that it chiefly eats insects, particularly small Lepidoptera. The stomach con- tents of one collected individual contained re- mains of Carabidae, Tenebrionidae, and Curcu- lionidae beetles, Formicidae, Diplopoda, and Gas- tropoda. These prey items are consistent with a general terrestrial existence, and this species for- ages on the ground and possibly less frequently on stumps or downed logs. The Rufous-headed Ground-roller was only found in the 1210 and 1625 m elevational zones. It was observed in areas with relatively wet soil and thick forest with large trees. Philepitta castanea This species, the Velvet Asity, was observed feeding on several plants (see also Chapter 4, Ap- pendix 4-2): Macaranga oblongifolia (Euphorbi- aceae) — at 1650 m, on small orange-colored buds; Gravesia caliantha (Melastomataceae) — at 1210 m, fruits; Medinilla ericarum (Melastomata- ceae)— at 1210 m, orangish-red fruits; Pittospo- rum sp. (Pittosporaceae) — at 800 m, whole fruits; Pauridiantha lyallii (Rubiaceae) — at 810 m and 1625 m, small cranberry-colored fruits; and Chas- salia sp. (Rubiaceae) — at 1210 m, bright red fruits. Neodrepanis coruscans This species, the Sunbird-asity, was observed visiting the dull reddish-pink flowers of Bakerella clavata (Loranthaceae) at 720 m and the white flowers of Liparis sp. (Orchidaceae) at 810 m. In late September 1993, three males were col- lected. Two of the three were in breeding plumage and had well-developed wattles, fully ossified skulls, and testes measuring left 4x2 mm and right 3X2 mm. The other male was molting into breeding plumage and had developed but dull wattles, an 85% ossified skull, and testes measur- ing left 3x2 mm and right 2X2 mm. Neodrepanis hypoxantha Little is known about the Yellow-bellied Sun- bird-asity. We found it to be relatively common between 1210 and 2050 m. It was neither sparse nor elusive, as suggested by Benson (1974); nor was it "very sparse," as reported for the RNI de Marojejy (Safford & Duckworth, 1990). In the RNI d'Andringitra, N. hypoxantha was seen dai- ly in the 1625 m elevational band. The mean GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 185 number of detections of N. hypoxantha from high counts at census points between 1100 and 1765 m (x = 0.633, standard deviation [SD] = 0.850, N = 30) did not differ from that of N. coruscans at census points between 720 and 960 m (jf = 0.667, SD = 0.606, N = 30) (Mann- Whitney test, P = 0.54). Although we found N. hypoxantha as numer- ous within its restricted elevational range as TV. coruscans was at lower elevations, the former species should remain a concern for conservation (Collar & Stuart, 1985; Safford & Duckworth, 1990). Neodrepanis hypoxantha is known only from elevations above 1050 m (Langrand, 1990), and so the total area of its range is much smaller than that of N. coruscans. Although deforestation of lowland forests has been severe (Green & Sussman, 1990), high-altitude forests are not im- mune to this process, as witnessed by the near elimination of the native biota of Ankaratra, one of the island's highest peaks (Perrier de la Bath- ie, 1927; Nicoll & Langrand, 1989). Many of the localities where N. hypoxantha was collected in the 19th century are now deforested (Dee, 1986). Neodrepanis hypoxantha was observed visiting the vermilion-colored flowers of Bakerella cla- vata (Loranthaceae) at about 1625 m and the dull reddish-pink flowers of Bakerella tandrokensis at 2050 m. A male N. hypoxantha was observed gleaning a small green caterpillar. He passed the animal back and forth through his mandibles be- fore swallowing it. A nest of this species was found at 1600 m and was watched on 29 and 30 October 1993 by M.S. P. The female made frequent visits to the domed nest, lining its interior with dry leaves of climbing bamboo. The male perched nearby much of the time, displaying when the female was pres- ent. He occasionally hung at the nest entrance but was not observed entering the structure or engag- ing in nest construction. The nest did not have the "entrance-roof" typical of the family Eurylaimi- dae (Amadon, 1951; Prum, 1993). The construc- tion presumably had not been completed. This is further supported by the fact that the nest, when collected on 30 October 1993, did not contain eggs. A few specimens of TV. hypoxantha were col- lected in late October 1993. These included a fe- male, on 20 October 1993, with a slightly en- larged oviduct (largest follicle 2 mm) and a fully ossified skull. Two males were collected, one with testes measuring left 5X3 mm, right 4X3 mm, and the other with testes left 7X5 mm, right 6 186 X 4 mm. Both had fully ossified skulls and well- developed wattles. The coloration of the soft parts of the two males differed. In one the portion of the caruncle around the eye was brilliant aqua or cobalt blue, and the balance was brilliant lime green. In the other male the portion surrounding the eye was almost black and the remaining sec- tion cobalt blue. The coloration of the proximal part of the bill in both males paralleled that found on the caruncle. Specifically, the area around the nares was the same color as the inner portion of the caruncle, and the area at the base of the bill was the same color as the outer portion of the caruncle. In both males the eyelid was bright lime green. Phedina borbonica Flocks of up to 30 Mascarene Martins were oc- casionally noted along the Iantara River, resting on dead limbs, in banana trees, or on the river- banks. Hypsipetes madagascariensis We observed the Madagascar Bulbul feeding on several food plants (see also Chapter 4, Ap- pendix 4-2): Oncostemum racemiferum (Myrsi- naceae) — at 730 m, ripe fruits; Pauridiantha lyal- Hi (Rubiaceae) — at 810 m, small cranberry-col- ored fruits; Bakerella sp. (Loranthaceae) — at 120C m, fruits; Medinilla ericarum (Melastomata- ceae) — at 1210 m, orangish-red fruits; Chassalic sp. (Rubiaceae) — at 1210 m, bright red fruits; anc Vaccinium secundiflorum (Ericaceae) — at 162* m, green-colored fruits. Saxicola torquata This species, the Stonechat, was relatively com mon in disturbed areas, particularly open tav fields, near our 720 m camp. It was also observe< in the open heathland above 1950 m elevation. Cryptosylvicola randrianasoloi The type locality of the Cryptic Warbler, a re cently described genus and species, is the easter slope of the RNI d'Andringitra (Goodman et al 1996), where its lower elevational limit was 90 FIELDIANA: ZOOLOG i m. Several individuals were observed or heard vo- calizing in areas of ericaceous vegetation up to 2100 m. A nest of this species with three eggs was found at 1625 m elevation on 24 October 1993. Dromaeocercus brunneus elevation, feeding three fledglings. The adult en- tered relatively open areas, gave a contact call, and then the fledglings would scurry out of the thick herbaceous understory to be fed. Only one adult was seen at a time feeding the young birds. Nectarinia notata We only have one record of the Brown Emutail in the reserve. On 11 December 1993, two birds were observed at about 1650 m elevation in an open area with thick herbaceous vegetation asso- ciated with a rock slide. Dromaeocercus seebohmi The Grey Emutail was only found in the heath- land zone above 1900 m. This rather skulking species occurred in the low and thick heath veg- etation. On the basis of the number of birds call- ing, it was relatively common. On 28 September 1993, at 810 m elevation, one adult male Long-billed Green Sunbird was observed chasing a female. This male gave a chirping note. At one point he assumed a possi- ble display posture; he perched with his neck ex- tended upward, the bill slightly agape, and with the closed wings lifted slightly away from his body. The wings were raised and lowered several times. At 2050 m elevation, this species, along with Neodrepanis hypoxantha and Nectarinia soui- manga, was observed visiting the dull reddish- pink flowers of Bakerella tandrokensis (Loran- thaceae). Rand ia pseudozosterops The majority of our records of Rand's Warbler within the reserve are from tree perches along riv- er margins between 720 and 1210 m. It was often seen or heard singing from exposed trees. Terpsiphone mutata On 1 November 1993, at about 850 m eleva- tion, eggshells of recently hatched birds were found below a nest of the Madagascar Paradise Flycatcher. Nectarinia souimanga We were able to gather some information on the food habits of this bird, the Souimanga Sun- bird. It was often seen feeding on various species of Bakerella (Loranthaceae). At 720, 810, and 1625 m the local species was Bakerella clavata, and at 2050 m it was B. tandrokensis. The Soui- manga Sunbird was also observed at 810 m vis- iting the white-colored flowers of Liparis sp. (Or- chidaceae); at 1210 m the lavender-colored flow- ers of Plectranthus sp. (Lamiaceae); and at 1650 m the white-colored flowers of Medinilla torren- tum. It apparently pierced flower buds with its bill. Oxylabes madagascariensis Schulenberg (1991) described the duet song of this species, the White-browed Oxylabes. On sev- eral occasions in the RNI d'Andringitra we heard and observed pairs of birds giving similar types of vocalizations. Crossleyia xanthophrys An adult Yellow-browed Oxylabes was ob- served on 3 and 4 December 1993, at 1220 m Zosterops maderaspatana We have a few records on the food habits of the Madagascar White-eye. At 720, 810, and 1625 m it was seen feeding on the vermilion-colored flowers of Bakerella clavata (Loranthaceae). It was attracted to the fruits of Oncostemum racem- iferum (Myrsinaceae) at 730 m, but it was not clear whether it was actually eating them or in- sects. At 810 m it was seen feeding on the small cranberry-colored fruits of Pauridiantha lyallii (Rubiaceae). It was seen picking fruits of Vernon- GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 187 ia alleizettei (Asteraceae) at about 1300 m, and, at 1625 m it was plucking and eating flower buds of Oncostemum ankifiense (Myrsinaceae). Schetba rufa On 28 November 1993, at 810 m elevation, an adult Rufous Vanga was observed feeding a fledg- ling. Xenopirostris polleni The only record we have of Pollen's Vanga in the reserve is one on 8 October 1993 at 810 m elevation. Foudia madagascariensis All of our records of the Madagascar Red Fody are from the tavy of Ambarongy at 720 m and in the heathland between 1950 and 2150 m. We found no evidence of this species in the forest proper. Foudia omissa The Forest Fody was only recorded in forested areas. We found no evidence of this species co- existing with the Madagascar Red Fody, nor was there any clear evidence of hybridization between them. At 1200 m this species was observed feed- ing at the tubular red flowers of Bakerella sp. (Loranthaceae). Hypositta corallirostris On 13 October 1993, at 810 m elevation, a male and female Nuthatch Vanga were observed foraging on the trunks of larger trees. The birds characteristically spiraled upward, very creeper- like (Family Certhiidae), along the bole and larger branches. The birds dropped down to the trunk of a nearby tree and then began to work upward. Once the male sallied after a flying insect that was presumably disturbed from the bark of the tree; he then returned to the trunk. Both the male and female gave a high-pitched "tsee" call just as they flew from one tree to another. Acridotheres tristis This introduced species, the Common Myna, was not recorded in the eastern portion of the RNI d'Andringitra. On the northern and western sides of the reserve, particularly in rice paddies and near villages, it was seen often. Ploceus nelicourvi In September and October 1993, male Neli- courvi Weavers were observed at several sites in the process of weaving nests. In virtually all cases, the nests were suspended from branches or limbs of trees over rivers and streams. This may be an adaptation against snake or lemur predation on the eggs and young. Lonchura nana The Madagascar Mannikin was only found at 720 m in the tavy of Ambarongy. Acknowledgments We are grateful to members of the expedition, particularly Brian Fisher and Eleanor Sterling, for sharing their bird observations with us. Philip Par- rillo identified invertebrate remains found in bird stomachs. Frank Hawkins, Olivier Langrand, Tim Moermond, Tom Schulenberg, Steve Zack, and an anonymous reviewer kindly provided comments on an earlier draft of the manuscript. Literature Cited Able, K. P., and B. R. Noon. 1976. Avian community structure along elevational gradients in the northern United States. Oecologia, 26: 275-294. Amadon, D. 1951. Le pseudo-souimanga de Madagas- car. Oiseau et R. F. O., 21: 59-63. Andrianarimisa, A. 1993. Variation structurale de la communaute avienne d'une foret dense seche semi- caducifoliee dans l'Ouest Malgache. Annales Musee Royal de l'Afrique Centrale (Zoologie), 268: 187— 191. Benson, C. W. 1974. Another specimen of Neodrepanis hypoxantha. Bulletin of the British Ornithologists' Club, 94: 141-143. , J. F. R. COLEBROOK-ROBJENT, AND A. WILLIAMS. 1976. Contribution a l'ornithologie de Madagascar. 188 FIELDIANA: ZOOLOGY L'Oiseau et Revue Franchise d'Ornithologie, 46: 209- 242. Chapman, F. M. 1917. The distribution of bird-life in Colombia; a contribution to a biological survey of South America. Bulletin of the American Museum of Natural History, 36: 1-729. Collar, N. J., and S. N. Stuart. 1985. Threatened birds of Africa and related islands. The ICBP/IUCN Red Data Book, part 1, 3rd edition. International Council for Bird Preservation and International Union for Conservation of Nature and Natural Resources, Cambridge, 761 pp. Dee, T. J. 1986. The endemic birds of Madagascar. In- ternational Council for Bird Preservation, Cambridge, 173 pp. Diamond, J. M., and M. LeCroy. 1979. Birds of Kar- kar and Bagabag islands, New Guinea. Bulletin of the American Museum of Natural History, 164: 467-531. Eguchi, K., S. Yamagishi, H. Nagata, M. Nakura, and V. Randrjanasolo. 1992. The mixed-species flocks of forest-living birds in Madagascar, pp. 28-38. In Ya- magishi, S., ed., Social structure of Madagascar higher vertebrates in relation to their adaptive radiation. Osa- ka City University, Osaka, 92 pp. Eguchi, K., S. Yamagishi, and V. Randrianasolo. 1993. The composition and foraging behaviour of mixed-species flocks of forest-living birds in Mada- gascar. Ibis, 135: 91-96. Finch, D. M. 1989. Habitat use and habitat overlap of riparian birds in three elevational zones. Ecology, 70: 866-880. Goodman, S. M. 1994. A description of the ground burrow of Eliurus webbi (Nesomyinae) and case of cohabitation with an endemic bird (Brachypteraciidae, Brachypteracias). Mammalia, 58: 670-672. , and P. C. Gonzales. 1990. The birds of Mt. Isarog National Park, southern Luzon, Philippines, with particular reference to altitudinal distribution. Fieldiana: Zoology, new series, 60: 1-39. -, O. Langrand, and B. M. Whitney. 1996. A new genus and species of passerine from the eastern rain forest of Madagascar. Ibis, 138: 153-159. -, S. O'Connor, and O. Langrand. 1993. A re- view of predation on lemurs: implications for the evo- lution of social behavior in small, nocturnal primates, pp. 51-66. In Kappeler, P. M., and J. Ganzhorn, eds., Lemur social systems and their ecological basis. Ple- num Press, New York, 274 pp. Graves, G. R., M. B. Robbins, and J. V. Remsen, Jr. 1983. Age and sexual difference in spatial distribution and mobility in manakins (Pipridae): inferences from mist-netting. Journal of Field Ornithology, 54: 407- 412. Green, G. M., and R. W. Sussman. 1990. Deforestation history of the eastern rain forests of Madagascar from satellite images. Science, 248: 212-215. Harrington, J. E. 1975. Field observations of social behavior of Lemur fulvus fulvus E. Geoffroy 1812, pp. 259-279. In Tattersall, I., and R. W. Sussman, eds., Lemur biology. Plenum Press, New York, 365 pp. Hawkins, F 1993. Relationships among vegetation structure and White-breasted Mesite Mesitornis var- iegata presence and foraging site choice. Annales Mu- see Royal de TAfrique Centrale (Zoologie), 268: 155— 164. Karr, J. R. 1981. Surveying birds with mist nets. Stud- ies in Avian Biology, 6: 62-67. Langrand, O. 1990. Guide to the birds of Madagascar. Yale University Press, New Haven, 364 pp. , and J. C. Sinclair. 1994. Additions and sup- plements to the Madagascar avifauna. Ostrich, 65: 302-310. , and L. Wilme. 1993. Protection des zones humides et conservation des especes d'oiseaux end6- miques de Madagascar. Annales Musee Royal de TAfrique Centrale (Zoologie), 268: 201-208. Macedonia, J. M. 1993. Adaptation and phylogenetic constraints in the antipredator behavior of ringtailed and ruffed lemurs, pp. 67-84. In Kappeler, P. M., and J. Ganzhorn, eds.. Lemur social systems and their eco- logical basis. Plenum Press, New York, 274 pp. Nicoix, M. E., and O. Langrand. 1989. Madagascar: revue de la conservation et des aires protegees. Fonds Mondial pour la Nature, Gland, xvii + 374 pp. O'Keeffe, J., and L. Ashmore. Not dated. Madagascar 1993 small mammal project. Unpublished Report, Emmanuel College, Cambridge University [available from S. M. Goodman]. Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, and P. Griveaud. 1971. RCP 225. Etudes des eco- systemes montagnards dans le region malgache. I. Le massif de l'Andringitra. 1970-1971. Geomorphologie, climatologie et groupements vlgltaux. Bulletin de la Soci&e" d'Ecologie, 11(2-3): 198-226. Perrier de la Bathie, H. 1927. Le Tsaratanana, l'An- karatra et l'Andringitra. M6moires de I'Acad^mie Malgache, 3: 1-71. Prum, R. O. 1993. Phylogeny, biogeography, and evo- lution of the broadbills (Eurylaimidae) and asities (Philepittidae) based on morphology. Auk, 110: 304- 324. Raxworthy, C. J., and P. R. Colston. 1992. Conclu- sive evidence for the continuing existence of the Mad- agascar Serpent-eagle Eutriorchis astur. Bulletin of the British Ornithologists' Union, 112: 108-111. Razafimahaimodison, J.-C, and R. Andrianantenaina. 1993. Ecologie et comportement des groupes d'ois- eaux dans une foret dense humide de Madagascar. An- nales Musee Royal de TAfrique Centrale (Zoologie), 268: 197-200. Remsen, J. V., Jr., and T A. Parker. 1983. Contribu- tion of river-created habitats to bird species richness in Amazonia. Biotropica, 15: 223-231. Safford, R., and W Duckworth. 1990. A wildlife sur- vey of Marojejy Nature Reserve, Madagascar. Inter- national Council for Bird Preservation, Study Report no. 40. Schulenberg, T. 1991 . On the songs of Oxylabes mad- agascariensis. Working Group on Birds of the Mad- agascar Region Newsletter, 1:8. . 1993. Phylogeny of the Vangidae: inferences from mitochondrial DNA. Annales Musee Royal de TAfrique Centrale (Zoologie), 268: 23-28. -, S. M. Goodman, and J.-C. Razafimahaimodi- son. 1993. Genetic variation in two subspecies of GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 189 Nesillas typica (Sylviinae) in south-east Madagascar. Annales Musee Royal de l'Afrique Centrale (Zoolo- gie), 268: 173-177. Sheldon, B. C, and J. W. Duckworth. 1990. Redis- covery of the Madagascar Serpent-eagle Eutriorchis astur. Bulletin of the British Ornithologists' Club, 110: 126-130. Terborgh, J. 1977. Bird species diversity on an Andean elevational gradient. Ecology, 58: 1007-1019. Thompson, P. M., ed. 1987. Zahamena Forest (Mada- gascar) expedition report 1985. International Council for Bird Preservation, Study Report no. 20. , and M. I. Evans. 1991. The birds of Amba- tovaky, pp. 4-1 to 4-29. In Thompson, P. M., and M. I. Evans, eds., A survey of Ambatovaky Special Re- serve, Madagascar. Madagascar Environmental Re- search Group, London. Thorstrom, R., R. T. Watson, B. Damary, F. Toto, M. Baba, and V. Baba. 1995. Repeated sightings and first capture of a live Madagascar Serpent-eagle Eu- triorchis astur. Bulletin of the British Ornithologists' Club, 115: 40-44. Watson, R. T, J. Berkelman, R. Lewis, and S. Raza- findramanana. 1993. Conservation of the Madagas- car Fish Eagle Haliaeetus vociferoides. Annales Mu- see Royal de l'Afrique Centrale (Zoologie), 268: 192— 196. 190 FIELDIANA: ZOOLOGY Chapter 19 The Shrew Tenrecs (Microgale) (Insectivora: Tenrecidae) of the Reserve Naturelle Integrate d'Andringitra, Madagascar Paulina D. Jenkins, Steven M. Goodman, and Christopher J. Raxworthy Abstract A collection of Microgale species made in humid forest between 720 and 1625 m elevation on the eastern slopes of the Reserve Naturelle Integrate d'Andringitra in late 1993 is reviewed. The material contains a total of 10 taxa, including two previously unknown species, one of which is described; two species that were known only from the type locality (M. parvula and M. soricoides); and three taxa (M. cowani, M. taiva, and M. melanorrhachis) for which the new material allows redefinition of diagnostic characters. We formally remove M. taiva and M. melanorrhachis from synonymy with M. cowani but provisionally synonymize M. pulla with M. parvula. Information is presented on external and craniodental morphology, measure- ments, variation, and reproduction. Resume La collection d'especes de Microgale effectuee dans la foret humide entre 720 m et 1625 m d' altitude sur le versant est de la Reserve Naturelle Integrate d'Andringitra effectuee a la fin de 1'annee 1993 fait 1'objet d'une discussion. Le materiel echantil tonne" comprend un total de dix taxons, y compris: deux especes encore inconnues auparavant, dont une est decrite ci-dessous et deux especes qui ne sont connues que localement (Af. parvula et M. soricoides) et enfin trois especes (M. cowani, M. taiva, et M. melanorrhachis), pour lesquelles les nouveaux echantillons permettant une redefinition des caracteres utilises pour 1' identification des especes. Nous avons abandon n£ la synonymie de M. taiva et M. melanorrhachis avec M. cowani, mais avons donnd M. pulla et M. parvula comme synonymes. Des informations relatives a la morphologie externe et craniodentaire, aux mesures biomemques, a leurs variations et a la reproduction sont presentees. Introduction monly occurring as a commensal with humans The insectivores of Madagascar are composed and almost certainly introduced (Hutterer & Tra- of two families: (1) the Soricidae, including two nier, 1990), and Suncus madagascariensis (Co- species, Suncus murinus (Linnaeus, 1766), com- querel, 1848); and (2) the Tenrecidae. With the JENKINS ET AL.: SHREW TENRECS 191 exception of the subfamily Potamogalinae, the other three subfamilies of Tenrecidae are endemic to Madagascar. The degree of morphological vari- ation within the Tenrecidae is absolutely remark- able, including the spiny "hedgehog"-like genera Echinops Martin, 1838, Hemicentetes Mivart, 1871, Setifer Froriep, 1806, and Tenrec Lacepede, 1799; the aquatic genus Limnogale Major, 1896a; the semi-fossorial Oryzorictes A. Grandidier, 1870; and the shrew-like Microgale Thomas, 1882, and Geogale Milne Edwards & A. Gran- didier, 1872. Although there is no explicit phylog- eny for the group, it is assumed to be monophy- letic and represents one of the more spectacular adaptive radiations of mammals in the world. The largest genus of Malagasy Tenrecidae is Microgale (shrew tenrecs), which has been re- vised by MacPhee (1987). He listed 22 validly published and available names but drastically re- duced the number of recognized species to 10; furthermore, he grouped these species into phe- netic clusters as follows: cowani cluster: M. cowani, M. thomasi, M. par- vula gracilis cluster: M. gracilis longicaudata cluster: M. longicaudata, M. prin- cipula pusilla cluster: M. pusilla brevicaudata cluster: M. brevicaudata dobsoni cluster: M. dobsoni, M. talazaci More recently, three new species have been de- scribed, one of which, Microgale pulla Jenkins, 1988, is here provisionally synonymized with M. parvula G. Grandidier, 1934. The other two spe- cies present greater problems: M. dryas Jenkins, 1992, groups with both the cowani and gracilis clusters, whereas M. soricoides Jenkins, 1993, ap- pears to warrant a distinct cluster. MacPhee's (1987) major revision was based primarily on dental morphology providing the framework for ontogenetic studies associated with tooth eruption patterns and measures of morpho- logical variation within and between various spe- cies groups. Metric data on the cranium and body, and external features, were accorded somewhat less consideration. In a few cases, the limits of several "species" are poorly defined, and new specimen material is necessary to work out these taxonomic problems. This bias away from exter- nal features has in some cases caused subsequent field-workers, often dealing with live animals, to continue to use the nomenclature of Eisenberg and Gould (1970), employing a combination of ratios of external measurements to divide the ge- nus into four behavioral classes, and that of Ge- nest and Petter (1975), whose key primarily em- ployed external features. In particular, several of the nominal species synonymized under Micro- gale cowani by MacPhee have continued to ap- pear as distinct species in the subsequent litera- ture, including M. taiva Major 1896b and M. me- lanorrhachis Morrison-Scott, 1948 (see Nicoll & Rathbun, 1990; Stephenson, 1995). In the past few years, several reports (Nicoll & Rathbun, 1990; Rax worthy & Nussbaum, 1994; Stephenson, 1995) have been made of specimens that were identified in the field, on the basis of external appearance and variations in general be- havior, as "M cowani" , "M. taiva", and "M. me- lanorrhachis" , all synonyms of M. cowani sensu MacPhee (1987). MacPhee maintained that dental evidence supported the view that the holotypes of M. melanorrhachis and M. taiva are juveniles of M. cowani. Among recent collections made for the University of Michigan Museum of Zoology by C.J.R. at Montagne d'Ambre, Mantady, and Ambatovaky, and for the Field Museum by S.M.G. and C.J.R., all three of the field-identified taxa include adult specimens. On the basis of this new material, some redefinition of M. cowani sen- su MacPhee is now feasible. First, all of these species appear to be closely related and show only slight, if any, dental vari- ation. Second, in most cases, diagnosis is depen- dent on a combination of a few, mainly external, characters. Finally, several of the taxa redefined here occur sympatrically, with no evidence of in- termediate specimens. Additional collections are needed to clarify several points. Also, biochemi- cal analysis of tissues already collected may pro- vide additional insight into the relationships of this complicated genus. Herein we report on one of these recent collec- tions, made during a small mammal survey un- dertaken by two of us (S.M.G. and C.J.R.) on the eastern slopes of the Reserve Naturelle Integrate (RNI) d'Andringitra during November and De- cember 1993. The regions surveyed proved highly speciose for shrew tenrecs of the genus Micro- gale. Some of these are common and their pres- ence is predictable from their widespread occur- rence in other areas of Madagascar; a few others are believed to be rare; and two are apparently undescribed. Many of the shrew tenrecs in the sample were identified from external features, but a considerable proportion were not immediately identifiable and were eventually sent to the Nat- 192 FIELDIANA: ZOOLOGY ural History Museum (BM(NH)) for comparative identification by P.D.J. In this paper we review the species limits of the Microgale obtained in this survey. In Chapter 20 the ecology and elevational distribution of Mi- crogale and other insectivores found within the reserve are discussed. Previous Work The only previous work conducted on the in- sectivores of the RNI d'Andringitra was in late 1970 and early 1971 during an expedition to the area (Paulian et al., 1971; see Chapter I, pp. 4-5 for details). Some information on this collection, made by R. Albignac and A. Peyrieras, was pre- sented by Genest and Petter (1975). MacPhee (1987) made reference to these species in his sys- tematic revision of the genus Microgale, and in a few cases he provided new specific identifications. The expedition participants concentrated their ef- forts in the higher elevation zones, and most if not all of their trapping effort was with rodent traps (R. Albignac, pers. comm.). The collection of Albignac and Peyrieras is in the Museum Na- tional d'Histoire Naturelle (MNHN), Paris. A list of the known mammal fauna of the RNA d'An- dringitra was compiled by Nicoll and Langrand (1989), based on a review of the literature and their own unpublished records. The only other known collections of Microgale made in the re- serve were in 1993 during a Cambridge Univer- sity student expedition (O'Keefe & Ashmore, 1994) for eventual return to Pare Botanique et Zoologique de Tsimbazaza, Madagascar. Materials and Methods Trap Lines The principal means of capturing insectivores was with pitfall buckets and drift fence; a few were obtained with standard mammal live traps. See Chapters 17 and 20 for more details on the trapping techniques. Specimens and Measurements Captured animals were prepared as standard museum skins with associated skulls and skele- tons, fluid-preserved carcasses, or full skeletons. Many tissue samples were frozen in liquid nitro- gen for biochemical studies, and the viscera were preserved in alcohol for endoparasite research. Whole carcasses preserved in formalin were wrapped in fine cheesecloth before immersion to prevent mixing of ecotoparasites (see Chapter 12). A representative collection of the taxa described herein will be returned to the D^partement de Biologie Animale, University d' Antananarivo. Cranial measurements, in millimeters, were taken using dial calipers and a microscope mea- suring stage. The dental nomenclature follows that of Mills (1966), Swindler (1976), Butler and Greenwood (1979), and MacPhee (1987). Dental notations are given in parentheses in the text; pre- maxillary and maxillary teeth are noted by upper case letters and mandibular teeth by lower case letters. The following measurements were made from specimens in the flesh or prepared crania. Abbreviations and definitions for these measure- ments (all in millimeters, with the exception of weight [WT], in grams) follow. BB (breadth of braincase): the greatest distance measured across the squamosals. BL (braincase length): from the superior artic- ular facet to the occipital condyle, parallel to the long axis of the skull. CIL (condyloincisive length): cranial length from first upper incisor to occipital condyle. EL (ear length): measured from the notch at the base of the ear to the distalmost edge of the pinna. HB (head and body length): measured from the tip of the nose to the distalmost point of the body (at base of tail). HF (hind foot length): measured from the back edge of the heel to the tip of the longest toe (not including claw). I1-P3 (length of anterior upper teeth): from an- terior of first upper incisor to anterior of sec- ond upper premolar. TL (tail length): measured from the base of the tail (at right angles to the body) to the end of the distalmost vertebra. Does not include terminal hair tufts. TOTL (total length of body and tail): measured from the tip of the nose to the end of the distalmost tail vertebra (does not include any terminal tail hair tufts). Animal is positioned on its back straight with vertebrae parallel to rule, but not stretched out. UTL (upper toothrow length): from anterior of JENKINS ET AL.: SHREW TENRECS 193 first upper incisor to posterior of third upper molar, parallel to the long axis of the skull. WT (weight): measured with Pesola spring scales. Animals weighing less than 10 g were weighed within 0.2 g; those weighing between 10 and 100 g were weighed within 0.5 g. Reproductive condition was recorded for males as length X width of the testes and degree of con- volution of the epididymis. Females were noted as nonperforate or perforate, nonparous or parous, and the numbers and locations of any embryos and placental scars were recorded. The mammary formula is presented as the number of paired ax- ial, abdominal, or inguinal teats. The following age classes are recognized: Infant: individuals in which the deciduous an- temolar dentition and the molars are not fully erupted; premaxillary, parietal, and basioc- cipital sutures are unfused. Juvenile: individuals in which the molars are fully erupted and the deciduous antemolar dentition is erupted and in the process of re- placement by the permanent teeth; cranial su- tures are in the process of fusing. The erup- tion sequence of the permanent teeth has been subdivided into four stages by MacPhee (1987); these stages have been accepted in this text, unless otherwise stated. Adult: individuals with a fully erupted perma- nent dentition; cranial sutures generally fused, although their position is more or less clearly marked. Other Abbreviations BM(NH): The Natural History Museum, Lon- don (formerly British Museum [Natural His- tory]). FMNH: The Field Museum, Chicago. MCZ: Museum of Comparative Zoology, Har- vard. MNHN: Museum National d'Histoire Naturelle, Paris. USNM: National Museum of Natural History, Washington, D.C. (formerly United States National Museum). C/c: canine. d: deciduous. I/i: incisor. M/m: molar. P/p: premolar. Results Ten species of Microgale were collected in the RNI d'Andringitra; they may be distinguished by the characters given in the key (Appendix 19-1). Unless otherwise indicated the data given below are confined to specimens collected at RNI d'Andringitra. Systematic Section Microgale cowani Thomas, 1882 Holotype— BM(NH) 82.3.1.25: adult female body in alcohol, skull extracted. Collected mid- March to mid-February 1 880 by the Reverend W. Deans Cowan. Type Locality — Ankafana Forest, eastern Bet- si leo (Ankafana = Ankafina, Fianarantsoa, Fian- arantsoa Province, 21°12'S 47°12'E; see Mac- Phee, 1987; Carleton & Schmidt, 1990). Referred Material— FMNH 151758, 151759: 43 km S Ambalavao, junction of Sahanivoraky and Sahavatoy rivers, RNI d'Andringitra, 810 m, 22°13'S 47°00'E; FMNH 151653, 151654, 151655, 151767, 151770, 151772, 151776, 151777, 151779, 151782, 151783, 151784, 151786, 151788, 151789, and 151796: 40 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1210 m, 22°13'S 46°58'E; FMNH 151652, 151656, 151798, 151810, and 151811: 38 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1625 m, 22°11'S 46°58'E. Specimens D, E, and F collected by the Cam- bridge University Expedition from RNI d'Andrin- gitra, 1400-1800 m. Description — The following description is based on the holotype and the RNI d'Andringitra specimens. Medium-sized, tail moderately short, shorter or subequal to HB. General appearance of the pelage is dark brown, especially on the rump where lighter speckling is reduced; color dorsally speckled brown, hairs with dark gray bases and a mixture of buff and red-brown tips; ventrally gray with a buff wash, hair bases gray with buff tips; tail bicolored, dark brown dorsally, sharply de- marcated from the paler venter, which is reddish buff in most specimens, especially proximally; tail well-clothed with long scale hairs that partially obscure the scales, those on the mid-dorsal basal third of the tail overlying 2.5-3 scales; hind feet brown above, dark gray below. Skull (Fig. 19-1) 194 FIELDIANA: ZOOLOGY Fig. 19-1. Crania from left to right of Microgale cowani FMNH 151783 and M. taiva FMNH 151643. Above, lateral view; below, dorsal view left and ventral view right. medium in size, rostrum elongated, nasals extend posteriorly into the interorbital region; frontals large, relative to the parietals, which are reduced in length in the mid-dorsal region; braincase rel- atively broad and short, junction of supraoccipital and parietal subrectangular, high on dorsal surface of the braincase, occipital large; pronounced dia- stemata separate teeth of the upper anterior den- tition from the first upper incisor (II) to the sec- ond upper premolar (P3), and diastemata are also present on either side of the lower canine and p2. All elements of the talonid of the third lower mo- lar (m3) are present, including the hypoconid, en- toconid ridge, talonid basin, and entoconid. For distinctive features of the morphology of the per- manent and deciduous dentition, see Figures 19-2 and 19-3. Measurements — External and cranial measure- ments are presented in Tables 19-1 and 19-2. Variation — There is no obvious indication of sexual dimorphism in size of adults, although the sample is too small for this hypothesis to be test- ed. The ratio of females to males in the sample is 1:1; that of juveniles to adults is 1.8:1. All of the juveniles are classed as Stage 1 (MacPhee, 1987, p. 13) or younger; the anterior dentition is com- pletely deciduous in most specimens but the mo- lars are usually fully erupted, although the third upper molar (M3) is incompletely erupted in two specimens (FMNH 151796 and 151798), whereas the third upper incisor (13) is the only permanent antemolar tooth erupting or erupted in the (Stage 1) individuals. Juveniles were compared with adults in four linear dimensions (HB, CIL, UTL, and BB) and WT. The mean of each measurement is slightly smaller in juveniles, with the exception of HB, in which the mean size of juveniles is greater than that of adults (see Table 19-1). These data on body size confirm the observation of Leche (1907) that juveniles may attain adult body size before replacing any deciduous teeth, and that of JENKINS ET AL.: SHREW TENRECS 195 Fig. 19-2. Above, buccal view of permanent left anterior dentition of M. cowani FMNH 15 1783; below, M. taiva FMNH 151635 (upper toothrow) and FMNH 151638 (lower toothrow). Scale 1 mm. MacPhee (1987) that subadults may exceed mean adult size in some parameters, including head and body length. Reproduction — An adult male (FMNH 151652) had abdominal testes measuring 7X5 mm with a convoluted epididymis. Four adult females for 196 which details on reproductive condition were available had enlarged mammae, perforated va- ginas, and in one case two placental scars on the left oviduct and three on the right. Mammary for- mula: 2-0-4 (N = 5). There is little information on the reproductive age of the juvenile specimens; FIELDIANA: ZOOLOGY one female (FMNH 151655) with completely de- ciduous antemolar dentition was nulliparous, whereas a second female (FMNH 151653) with a deciduous antemolar dentition apart from 13 was perforate but lacked embryos. This limited infor- mation suggests that females in this species may become reproductively mature while still dentally immature. Remarks — Formerly considered to be the most widespread and commonly occurring species throughout northern, eastern, and southeastern Madagascar, this species is here restricted to the holotype from Ankafina and the specimens listed above from the RNI d'Andringitra. It may be much rarer in collections than previous accounts have indicated. Microgale cowani occurs sympat- rically with M. melanorrhachis and M. taiva at RNI d'Andringitra; those taxa were considered to be synonyms of M. cowani by MacPhee (1987). Characters used to discriminate this species from others occurring in the RNI d'Andringitra are giv- en in the key (Appendix 19-1) and Table 19-1. Those that separate M. cowani and M. taiva are given in Table 19-2. Microgale taiva Major, 1896b Microgale cowani Thomas: MacPhee, 1987, in part Holotype— BM(NH) 97.9.1.112: juvenile fe- male, skin and skull; BM(NH) 1975.2233: partial skeleton. Collected January 19, 1895 by C. I. For- syth Major. Type Locality — Ambohimitombo forest, Tana- la Country (Ambohimitombo town, Fianarantsoa, Fianarantsoa Province, 20°43'S 47°26'E; see MacPhee, 1987). Referred Material— FMNH 151633, 151634, 151642, 151643, and 151645: 45 km S Ambala- vao, E bank of Iantara River, along Ambalama- nenjana — Ambatamboay Trail, edge of RNI d'Andringitra, 720 m, 22°13'S 47°01'E; FMNH 151635, 151636, 151637, 151755, 151757, 151760: 43 km S Ambalavao, junction of Sahan- ivoraky and Sahavatoy rivers, RNI d'Andringitra, 810 m, 22°13'S 47°00'E; FMNH 151638, 151639, 151761, 151762, 151763, 151765, 151769, 151771, 151774, 151775, 151780, and 151781: 40 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1210 m, 22°13'S 46°58'E; FMNH 151640, 151641, 151657, 151658, 151724, 151725, 151790, 151791, 151792, 151797, 151802, 151809, and 151813: 38 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1625 m, 22°11'S 46°58'E. Cam- bridge University Expedition specimens B and C, RNI d'Andringitra, 1400-1800 m. Description — Based on the holotype and spec- imens collected in the RNI d'Andringitra. Medi- um-sized, tail moderately long, subequal or longer than head and body length (see Tables 19-1 and 19-2). Dorsal pelage dark brown with buffy brown speckling, ventral coloration gray-brown with buffy brown wash; tail not obviously bicol- ored, dark gray-brown above, slightly paler gray below; tail scale hairs short, overlying 1.5-2 scales on the dorsal proximal third of the tail, so that the scales are visible. Skull medium in size (see Fig. 19-1), with a moderately elongated ros- trum, nasals extend to level of zygomatic plate, barely into interorbital region; frontals posteriorly inflated so that the dorsal surface of the skull ap- pears slightly concave in profile; braincase broad and long, parietals moderately large, supraoccip- ital-parietal suture suboval. Short diastemata sep- arate the anterior upper dentition from II to P3 and are present between the lower canine and p2. Talonid of m3 slightly reduced, entoconid lacking. Distinctive features of the morphology of the per- manent and deciduous dentition are shown in Fig- ures 19-2 and 19-3. Measurements — External and cranial measure- ments are presented in Table 19-1 and 19-2. Variation — Although the mean of HB, WT, and three cranial dimensions in adult female M. taiva is consistently higher than that of males, this difference was found to be statistically insignifi- cant, and it is concluded that sexual dimorphism is not present in the small sample available. The sex ratio of females to males in the adult sample is 1:2.3, and the ratio of juveniles to adults is 1: 1.2. Two specimens (FMNH 151641 and 151813) are dentally immature — the antemolar teeth are completely deciduous and the third upper and lower molars (M3 and m3) are in the process of eruption; five specimens show deciduous ante- molar dentition but fully erupted molars; 13 and/or the third lower incisor (i3) are erupting in three other specimens. In one specimen (FMNH 151636) only i3 has erupted and 13 and the first lower premolar (p2) are beginning to erupt; it is classed as Stage 1-2 (following MacPhee, 1987, p. 13), yet it has fairly large testes (4 X 2) and a convoluted epididymis. Another specimen (FMNH 151642) is difficult to fit into a particular category, because II, 13, i3, p2, and p4 are all JENKINS ET AL.: SHREW TENRECS 197 Fig. 19-3. Top, buccal view of deciduous left anterior upper dentition of M. cowani FMNH 151798; middle, M. taiva FMNH 151725; below, M. melanorrhachis FMNH 151626. Scale 1 mm. Figure continues on page 199. starting to erupt, but in contrast it has small testes (2 X 1 ), and the epididymis is not convoluted. In HB, CIL, UTL, and BB, the mean of these juve- niles is slightly less than that of the adults, al- though the mean WT of juveniles is considerably less (9.0 ± 1.8 g, N = 15) than that of adults (12.4 ± 1.2 g, N = 18; see Table 19-1). The den- tally most mature juvenile and many of the other juveniles are well within the adult range in all of these dimensions, supporting the view that im- mature Microgale achieve large size while decid- uous teeth are still present and that the deciduous 198 FIELDIANA: ZOOLOGY Fig. 19-3 (cont.). Top: above, lingual view of deciduous left anterior lower dentition of M. cowani FMNH 151776; middle, M. taiva FMNH 151725; below, M. melanorrhachis FMNH 151626. Scale 1 mm. Bottom: Lingual view of right m3 from left to right of M. cowani FMNH 151776, M. taiva FMNH 151725, and M. melanorrhachis FMNH 151626. Scale 1 mm. dentition may be maintained into adulthood (Le- che, 1907; MacPhee, 1987; Jenkins 1992). Reproduction — Two of the adult females col- lected were carrying embryos, both with single embryos in the left and right oviducts, measuring 6 and 7 mm crown-rump length. Mammary for- mula: 0-2-4 (N = 1), 0-2-2 (N = 1), 2-0-4 (N = 2). The testes of three adult males in the sample (FMNH 151633, 151643, and 151547) measure 6 X 3, 7 X 3, and 8X4 mm, respectively, all with convoluted epididymides. Information is lacking for females of this species, but it seems likely JENKINS ET AL.: SHREW TENRECS 199 K^ — 3 < a 3C rj CO + 1 I +1 q I +1 oc ; CO O J 5 5 i?id J *i P1^ ci >/"} ^^ ^ ^ -~ ^ ro oo oo 1 © J ^ - ^ ^ iCN^iu-l^rt ^ — H 00 ^ O - I VD ■* m — i-c C4 — ON 00 — cm rn ^c ^V^^^VO^ ON — — — < Tfr +i i +i t +i 7 +i 7 !£ +i 7 +i 7 +i 7 +i 7 to +i 7 " oo c ^ n -* c~> in m • — t"» 00 ir> — O — On — © — On — -/"> — oo r-' ■«* ro -"t ""t «n ^ ^* on oo SO — — (N „ S ^ Jj Tt d ™ d © — NOd" d >n >n m — _ — n oo ■<*■ - -3- — on — ■ — — — 00 — /-v ^ Tfr NO ^-v-^-00lO --v NO Tt ' — ' ' — ' 00 — — — • t — '"3" -.w9^9»piNft1iS,»w«w'*w*Pw ir> vo ir> O •/"> © ~™ no "~ vi in oo in d NO n n >n no no ^ . On I +1 I — +1 1 +1 I +1 I +1 I On oo NO v« r^r^inr^ONr^ONNO^moN O — — Tfr" T^ oo oo oo r- On »n n© ^ S 5 " 8 «• . QN " O nC © oo r^ -•,«>»• o in in S^MNOMvo^^^r^^r- .o^, oo oo — r^oo — TtinTtjOTf-'r-NONO^-^ingo^:--'^^^ +i "7 +i 7 ss +i 7 +i 7 +i 7 +i 7 ~ +i 7 — ~ s - o 3 > 3 3 T3 T3 CO ca s _3 .2,-S .2,.S"S c^ b «* I! 1 §§ .8 § ^' ^ 200 FIELDIANA: ZOOLOGY I ^ — 00 nSmSL^- *-" •* o> ^ oo w — '—'— wr»» — ^^o— <^> — — n «r — ©2d (N CI - O + l4+,^+li+loc+l^+loo+loo+l^+l4 NO^NO^oo^r-^ON^ON^ON^ON'^ON nOnOOOOOOnOOnOnOn »2 2-:d22 = «; •-v ^ ^ ^ pn — oo — o\ po Tt/-^'* ^^ w JinJ 3Lo3L"">v]!ro^irr~woowp>")w(Nw''i-wr,|oow — — — — __rn— X^ +l ^i +l ^. +l J> +l ' +l 4, +l J. +l ' +| J, +| J, +| J. +i i - +i J. ® — (N (S NO — O T}- — PO fi t~: OP on -* ; r- "! oo oo X X k NO -J 00 . O ^ O s0 ^0P^rOTt^-(Np O — O — © — O — O — O — PO — PO rfr — (N >n — < oo — ON ON 2 w ^ T*' ,,*d6*od»dodd--dwd(Ndm6w + — « "T — — (NCNfN(NC4 ~ I +1 ' +i i +ii+ii +1 _L +' I +1 I +1 J. po — o _ o CM _ — ™ +1 1^+1 I ""! *°. f*> « NO On 00 tJ- -NON^od(N O fN ri 3 > 3 3 3 "O g T3 "O ,3 a I I Q C S 5 5 5" 5 3'3 £ JENKINS ET AL.: SHREW TENRECS 201 Table 19-2. Dimensions (in mm) distinguishing hii- crogale cowani from hi. taiva in the RNI d'Andringitra. Parameter hi. cowani hi. taiva Tail length 60.6-71.0 71.0-89.0 65.6 ± 3.25 88.4 ± 4.60 Ratio of tail length to condyloincisive length Length of anterior denti tion I-P3 (10) 2.7-3.1 3.0 ± 0.13 (10) 5.5-6.2 5.8 ± 0.2 (14) 3.5-4.2 3.9 ± 0.19 (14) 5.2-5.8 5.5 ± 0.21 Ratio of anterior denti- (ID 52.8-56.4 (14) 49.1-52.8 tion (I-P3) to upper toothrow length Braincase length 53.9 ± 1.08 (11) 7.3-8.2 7.8 ± 0.24 50.7 ± 0.88 (14) 7.8-8.4 8.1 ± 0.17 (ID (14) from the well-developed testes in the dentally im- mature (Stage 1-2) male (FMNH 151636) that males may become sexually mature before they acquire their complete permanent dentition. Remarks — The holotype of this species is a ju- venile. It was considered a synonym of Microgale cowani by MacPhee (1987) on the basis of its dentition. Recently collected adults confirm that although the permanent dentition is indeed simi- lar, it does differ slightly from that of M. cowani in that the talonid of m3 is slightly reduced and the entoconid is lacking. The deciduous antemolar dentition also differs from that of M. cowani (and M. melanorrhachis). The two taxa do occur sym- patrically in the RNI d'Andringitra, where they are readily distinguished from each other by dif- ferences in their external appearance, including those of absolute and relative tail length, tail col- oration and pilosity, by a combination of cranial characters, and by minor differences in the den- tition (see Tables 19-1 and 19-2, and Figs. 19-1, 19-2, and 19-3). Microgale melanorrhachis Morrison- Scott, 1948 Microgale cowani Thomas: MacPhee, 1987, in part Holotype— BM(NH) 48.88: juvenile female, skin and skull, collected November 22, 1939 by C. S. Webb. Type Locality — Perinet (= Andasibe), near Moramanga, eastern Madagascar, 19°00'S 48°30'E, altitude 3000 feet (= approximately 900 m). Referred Material — FMNH 151626: 45 km S Ambalavao, E bank of Iantara River, along Am- balamanenjana — Ambatamboay Trail, edge of RNI d'Andringitra, 720 m, 22°13'S 47°01'E; FMNH 151627 and 151756: 43 km S Ambalavao, junction of Sahanivoraky and Sahavatoy rivers, RNI d'Andringitra, 810 m, 22°13'S 47°00'E. Description — This description is based solely on juvenile specimens from RNI d'Andringitra. Similar in size to M. cowani and M. taiva juve- niles of equivalent dental age (HB 68-77, mean 73.7 ± 4.03, N = 3); tail length (67-73, mean 69.0 ± 2.83, N = 3) similar to that of M. cowani but much shorter than that of M. taiva (see Table 19-1). Dorsal pelage gray-brown with some yel- lowish speckling and with a distinct dark brown mid-dorsal stripe extending from the head at the level of the ears to the base of the tail. Dorsal hairs with light gray bases, buffy red terminally with short dark brown tips; venter gray with a buff wash, hairs with light gray bases and long white tips. Tail bicolored, dark gray dorsally, buff ven- trally; sparse covering of scale hairs, each over- lapping 1.5-2 scales in the dorsal basal third of the tail. Hind feet buff above, dark gray-brown laterally and below. Skull similar in size to those of juvenile M. cowani and M. taiva (CIL 21.2, 22.5 mm); rostrum moderately elongated and with short diastemata between the anterior upper teeth, as in M. taiva but unlike the longer diastemata of M. cowani; braincase narrower and shallower than that of M. cowani and M. taiva, shorter than that of M. taiva. The deciduous antemolar dentition differs from those of M. cowani and M. taiva (see Fig. 19-3a,b). The talonid of m3 is slightly re- duced, the talonid basin is narrow, and the ento- conid is lacking (see Fig. 19-3c). Measurements — External and cranial measure- ments are presented in Table 19-1. Variation — The pelage of the RNI d'Andrin- gitra specimens differs in coloration from that of the holotype, which is more rufous brown dorsal- ly, and the dorsal surface of the hind feet and ventral surface of the tail are more reddish buff in color than that of the holotype. Reproduction — All of the specimens obtained were juveniles, with deciduous antemolar denti- tions and with fully erupted molars. In one of the two males for which reproductive condition was recorded (FMNH 151626), the immature testes measured 2X2 mm and the epididymis was not convoluted. The single female (FMNH 151627) 202 FIELDIANA: ZOOLOGY in which i3 was beginning to erupt had a possible perforated vagina but showed no other signs of sexual maturity. Remarks — MacPhee (1987) considered M. me~ lanorrhachis to be a synonym of M. cowani on the basis of its dentition, arguing that the mid- dorsal stripe used to characterize the taxon is vari- ably present or absent in M. cowani. Although Grandidier (1934) suggested that striping is a ju- venile characteristic, MacPhee (1987) pointed out that loss or diminution of the stripe, perhaps fol- lowing molting, was not in phase with replace- ment of the deciduous dentition and also occurred in dental adults. The contention that M. mela- norrhachis is a synonym of M. cowani has never gained wide acceptance. Nicoll and Rathbun (1990) states that adults of M. cowani and M. me- lanorrhachis occur with overlapping ranges at Andasibe, whereas Rax worthy and Nussbaum (1994) and Stephenson (1995) also treat M. me- lanorrhachis as a distinct species. All of the spec- imens from the RNI d'Andringitra are juveniles and are clearly distinguishable from juveniles of M. cowani and M. taiva by craniodental features in addition to the distinctive pelage. Microgale longicaudata Thomas, 1882 Microgale majori Thomas, 1918 Holotype— BM(NH) 82.3.1.15: adult female, body in alcohol, skull extracted, collected mid- March to mid-February 1 880 by the Reverend W. Deans Cowan. Type Locality — Ankafana Forest, eastern Bet- sileo (Ankafana = Ankafina, Fianarantsoa, Fian- arantsoa Province, 21°12'S 47°12'E; see Mac- Phee, 1987; Carleton & Schmidt, 1990). Referred Material— FMNH 151628, 151630, and 1 5 1 63 1 : 45 km S Ambalavao, E bank of Ian- tara River, along Ambalamanenjana-Ambatam- boay Trail, edge of RNI d'Andringitra, 720 m, 22°13'S 47°01'E; FMNH 151632, 151795, 151799, 151800, 151803, and 151804: 38 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1625 m, 22° ITS 46°58'E; Cam- bridge University Expedition specimen A: RNI d'Andringitra, 1400-1800 m. Description — Body small in size, with a very long tail, considerably longer than that of the head and body. Dorsal pelage dark brown with a red- dish brown wash, ventrally dark gray with a red- dish buff wash; tail gray-brown above, sharply distinguished from the reddish buff ventral col- oration; hind feet brown but reddish buff laterally. Skull small, dorsal profile slightly concave (Fig. 19-4), rostrum moderately short, nasals extend posteriorly into the interorbital region; braincase long, parietals and occipital large, suboval su- praoccipital suture on dorsal surface of the brain- case. Diastemata are present between II and 12 and on both sides of C and P2; well-developed anterior and posterior accessory cusps are present on 12, C, and P2. Lower p2 caniniform. Talonid of m3 with well-developed hypoconid and hypo- conulid, arrow talonid basin, and reduced ento- conid ridge and entoconid. Measurements — External and cranial measure- ments are presented in Table 19-1. Variation — In the sex ratio of the small sam- ple available, males outnumber females (2.5:1); the juvenile to adult ratio is 1.5:1. The juveniles collected in the RNI d'Andringitra exhibit a com- pletely deciduous antemolar dentition and in most M3 is incompletely erupted. There are only slight differences between the morphology of the decid- uous incisors and first premolar (dp2) and that of the permanent teeth. Reproduction — Only two of the specimens showed signs of reproductive activity — a male (FMNH 151799) with testes measuring 5x3 mm and a convoluted epididymis; and a lactating fe- male (FMNH 151630), collected on November 20, 1993, with single placental scars on the left and right oviducts. Mammary formula: 0-2-2 (N = 1), 2-0-4 (N = 1). Remarks — Microgale longicaudata has been reported in eastern Madagascar from Antsiranana in the extreme north (12°16'S 49°18'E) to Anka- fina (21°12'S 47°13'E) (MacPhee, 1987; Jenkins, 1993). The presence of this species in the RNI d'Andringitra marks a southern extension of the recorded range. Thomas (1882) described a long- tailed species (M. longicaudata) from a series of specimens from Ankafina, smaller specimens of which he later transferred to a new species, M. majori Thomas, 1918. MacPhee (1987) consid- ered the two taxa synonymous. The specimens from the RNI d'Andringitra fall in the lower part of the size range given by MacPhee (1987) for M. longicaudata. This species shows several morpho- logical features that suggest it is adapted to an arboreal lifestyle (see Thomas, 1918). These fea- tures include the extremely long tail, the bare re- gion near the tip with transverse scales, the elon- gated fifth digit on the hind foot, and the elon- gated cheiridia. The elongated tail distinguishes JENKINS ET AL.: SHREW TENRECS 203 Fig. 19-4. Crania from left to right of M. longicaudata FMNH 151799 and M. parvula FMNH 151623. Above, lateral view; below, dorsal view left and ventral view right. this species from all others occurring in the RNI d'Andringitra. Other distinguishing features are given in the key (Appendix 19-1) and Table 19-1. Microgale parvula Grandidier, 1934 IMicrogale pulla Jenkins, 1988 Holotype — MCZ 45465: juvenile male, body in alcohol, skull extracted. Collected by M. Drou- hard. Type Locality — Environs of Diego Suarez (Antsiranana, ca. 12°16'S 49°18'E— see MacPhee, 1987). Referred Material — FMNH 151621: 45 km S Ambalavao, E bank of Iantara River, along Am- balamanenjana-Ambatamboay Trail, edge of RNI d'Andringitra, 720 m, 22°13'S 47°01'E; FMNH 151622: 43 km S Ambalavao, junction of Sahan- ivoraky and Sahavatoy rivers, RNI d'Andringitra, 810 m, 22°13'S 47°00'E; FMNH 151722, 151764, and 151766: 40 km S Ambalavao, along Volot- sangana River, RNI d'Andringitra, 1210 m, 22°13'S 46°58'E; FMNH 151623, 151723, 151793, 151794, 151801, 151805, and 151806: 38 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1625 m, 22°11'S 46°58'E. Description — Very small, tail subequal in length to that of head and body. Dorsal pelage dark gray-brown, ventral pelage dark gray, tail uniform dark gray. Skull very small, delicate, and elongated in appearance (see Fig. 19-4), rostrum slender, moderately short; braincase shallow and long, frontals and occipital large relative to pari- etals, occipital condyles posterodorsally orientat- ed. Diastemata present between II and 12 and on either side of C and P2; anterior and posterior accessory cusps present on 12, 13, and P2. Diaste- ma between c and p2. Talonid of m3 with well- 204 FIELDIANA: ZOOLOGY Fig. 19-5. Left anterior dentition of M. parvula. Above, buccal view of permanent dentition of FMNH 151623; middle, buccal view of deciduous upper den- tition of FMNH 151806; below, lingual view of decid- uous lower dentition of FMNH 151622. Scale 1 mm. developed hypoconulid but reduced hypoconid, entoconid, and entoconid ridge, and narrow, shal- low talonid basin. See Figure 19-5 for illustrations of the permanent and deciduous dentition. Measurements — External and cranial measure- ments are presented in Table 19-1. Variation — The sex ratio of males to females in the small sample is 1.3:1, and that of juveniles to adults is 1.5:1. The antemolar dentition is de- ciduous, and all molars are fully erupted in all four juveniles. Reproduction — Few of the specimens collect- ed showed active reproductive organs. At least two of the males that were adult based on denti- tion had small testes without convoluted epidid- ymides; the testes of FMNH 1 5 1 623 measured 2 X 1 mm, and those of FMNH 151723 measured 3X2 mm. One adult female (FMNH 151621) obtained on November 15, 1993, had two embry- os in each oviduct measuring 4 mm crown-rump length. Mammary formula: 0-0-4 (N = 1), 0-2-4 (N = 1). Information is lacking on the reproduc- tive condition of juvenile males with deciduous dentition, but a female (FMNH 151766) with de- ciduous antemolar dentition was perforate. Remarks — The specimens collected in the RNI d'Andringitra include both adults and juveniles that are clearly attributable to the same species. Microgale parvula is known only from the orig- inal description of the single juvenile specimen, the dentition of which was illustrated by MacPhee (1987), who also corrected the measurements giv- en in the original description. The dentition of the juveniles from RNI d'Andringitra agrees well with the illustrations of the dentition of M. par- vula in MacPhee (1987), and the corrected upper toothrow length given by MacPhee is similar. There are, however, some differences in size; the corrected skull length of the holotype (15.5 mm) is less than that in the three juvenile specimens from RNI d'Andringitra with deciduous antemolar dentitions (i.e., of equivalent dental age), in which CIL is 16.2-16.7 mm. The tail of the holotype is longer than its head and body length (TL:HB 1.21), whereas the tail is subequal to or slightly larger than head and body length in these speci- mens (TL:HB 0.96-1.07). Comparisons have also been made between the adult specimens from RNI d'Andringitra and the adult holotype specimen of M. pulla Jenkins, 1988. They agree in most fea- tures of size, external characters, and craniodental morphology, except for the TL to HB ratio, which is greater than that in the holotype of M. pulla (0.83). When M. pulla was described (Jenkins, 1988), the possibility was mentioned that it might simply represent the adult of M. parvula, and the conclusion from the study of these specimens from RNI d'Andringitra is that this is probably correct, although direct comparison of specimens is required for confirmation. JENKINS ET AL.: SHREW TENRECS 205 Microgale dobsoni Thomas, 1884 Nesogale dobsoni Thomas, 1918 Holoytpe— BM(NH) 84.10.20.1: immature male, in alcohol, skull extracted. Collected Feb- ruary or March 1884 by W. Waters. Type Locality — Nandesen forest, Central Bet- sileo (Nandihizana, 10 miles S Ambusitra — manuscript note in Thomas's private copy of orig- inal description. Nandihizana, ca. 20 miles [30 km] SSW Ambositra, see MacPhee, 1987. Esti- mated as 20°50'S 47°10'E). Referred Material— FMNH 151624, 151625, and 151785: 40 km S Ambalavao, along Volot- sangana River, RNI d'Andringitra, 1210 m, 22°13'S 46°58'E. Description — Head and body large, tail sub- equal to or longer than head and body. Pelage long, dorsally gray-brown with a buff or reddish buff wash, venter gray with a buff or reddish buff wash. Tail gray above, buff below; hind feet buff. Skull and mandible very large and robust, sutures fused and obscure; rostrum broad, interorbital re- gion long, slightly concave; braincase angular, su- perior articular facets very prominent, occipital region reduced, supraoccipital crests well devel- oped. Diastemata between II and 12 and between 13 and C. Lower i2 considerably larger than ca- nine. The teeth of both individuals are too worn for detailed observations. Measurements — External and cranial measure- ments are presented in Table 19-1. Variation — The two adults collected in 1993 differ in external appearance; one specimen, with TL 93% of HB, is more rufous in coloration than the other, in which the tail is relatively much lon- ger (TL 1 12% of HB). A specimen in the MNHN collected from Anjavidilava is slightly lighter in coloration than either of the FMNH specimens, with TL 96% of HB. All three specimens are within the range of color variation exhibited by other specimens of this taxon in the BM(NH) col- lection. The range of TL:HB is 82-111% (mean 94.1 ± 7.67%, N = 11) for M. dobsoni in the BM(NH) collection, so even the long-tailed spec- imen (FMNH 151624) is only just outside the known range on this feature. There is no indica- tion in the RNI d'Andringitra specimens of the incrassation of the tail found commonly in this species. Reproduction — The two female specimens taken in 1993 were adults, on the basis of denti- tion, and both were lactating. One individual (FMNH 151624) had two placental scars on both -i Fig. 19-6. Above, right lateral view of the rhinarium of M. gymnorhyncha FMNH 151808; below, M. gracilis FMNH 151773. Scale 1 mm. the right and left oviducts. Mammary formula: 2-0-4 (N = 1). One infant (FMNH 151785), with unopened eyes, was found at night along a trail. Remarks — The few specimens collected sug- gest that this species is uncommon in the area or that population numbers were low at the time of the trapping program. Four specimens of M. dob- soni in the MNHN were collected during the 1970-1971 expedition to the Andringitra Massif, in the vicinity of Anjavidilava, and at approxi- mately 1995 m. Two of the specimen labels bear the note "foret mousse." Thus, the elevation range of this species within the reserve is pre- sumed to be between about 1200 and 2000 m. Microgale soricoides Jenkins, 1993 Holotype — BM(NH) 91.565: adult male in al- cohol, skull extracted. Collected April 13, 1991, by C.J.R. Type Locality — Mantady National Park, ca. 15 km north of Perinet (Andasibe), 18°51'S 48°27'E, in primary rain forest, altitude 1 100— 1150 m. Referred Material — FMNH 151768 and 151778: 40 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1210 m, 22°13'S 46°58'E. 206 FIELDIANA: ZOOLOGY Fig. 19-7. Crania from left to right of M. gymnorhyncha FMNH 151807 and M. gracilis FMNH 151773. Above, lateral view; below, dorsal view left and ventral view right. Description — Size large, tail subequal to or longer than head and body. Pelage light gray- brown dorsally, gray-brown ventrally with a buff wash. Tail brown above, paler buffy brown below. Skull and mandible moderately large and robust, rostrum and interorbital region broad, braincase short and broad; parietals large, occipital small, supraoccipital ridge present. First upper II mark- edly robust and proodont. First lower il and i2 robust and procumbent, i2 smaller than il but larger than c. The first upper and lower premolars are very small and have a single root. The teeth are too worn for determination of the features of m3. Measurements — External and cranial measure- ments are presented in Table 19-1. Reproduction — One of the specimens (FMNH 151768) taken on December 1, 1993 was an adult female, based on dentition, and was lactating. Mammary formula: 2-0-4 (N = 1). Variation — The two specimens from RNI d'Andringitra are slightly greater in body size than those from Mantady but fall within the cra- nial size range (dimensions of the type series, with the RNI d'Andringitra specimens in square brack- ets: HB 77.0-85.5 mm [86, 100 mm], ratio of TL to HB 0.95-1.13 [1.04, 1.11], CIL 25.1-26.7 mm [24.6 mm], UTL 12.1-13.0 mm [ca. 12.2, 12.3 mm]). Remarks — This is the second locality record for the species. Whereas the type series was col- lected in primary rain forest at an altitude be- tween 1 100 and 1 150 m, the RNI d'Andringitra specimens were collected at a slightly higher el- evation of 1210 m. One of the RNI d'Andrin- gitra specimens was collected 2 m above the ground on a tree limb 3 cm in diameter that led from the ground to a tangle of vines, so this species is obviously capable of climbing in low vegetation. JENKINS ET AL.: SHREW TENRECS 207 Fig. 19-8. Top, buccal view of permanent left anterior dentition of M. gracilis FMNH 151773. Scale 1 mm. Bottom, buccal view of permanent left anterior dentition of M. gymnorhyncha BM(NH) record number 1995.R258. Scale 1 mm. Figure continues on page 209. 208 FIELDIANA: ZOOLOGY Fig. 19-8 (cont.). Above, lingual view of left P4-M3 of M. gymnorhyncha BM(NH) record no. 1995.R258; middle, M. gracilis FMNH 151773; below, lingual view from left to right of right m3 of M. gymnorhyncha FMNH 151651 and M. gracilis FMNH 151649. Scale 1 mm. Microgale gracilis (Major, 1896a) Oryzoryctes [sic] gracilis Major, 1896a Leptogale gracilis Thomas, 1918 Holotype— BM(NH) 97.9.1.78: adult of un- determined sex; skin and skull. Collected Novem- ber 1 894 by C. I. Forsyth Major. Type Locality — Ambohimitombo forest (Am- bohimitombo town, 43 km [by road] SE Ambo- sitra, 10 km into eastern forest; Fianarantsoa; 20°43'S 47°26'E— see MacPhee [1987]). Mac- Phee gives the altitude for this locality variously as 1300 m [1987, p. 6] and 1200 m [1987, table 5] but, as pointed out by Carleton & Schmidt [1990], the altitude recorded for this locality by Major [1897] is 1400-1500 m.). JENKINS ET AL.: SHREW TENRECS 209 Referred Material — FMNH 151648, 151649, and 151773: 40 km S Ambalavao, along Volotsangana River, RNI d'Andringitra, 1210 m, 22°13'S 46°58'E; MNHN 1972-606, 1972-607: Foret Agauria (Marositry), Andringitra (2000 m). Description — Size large, tail shorter than head and body. Pelage dark brown dorsally with buff speckling, ventrally dark gray with a buff wash, tail dark brown above, light brown below; the ju- veniles are less speckled on the rump than the adults. Muzzle very long; rhinarium large, naked region extends posterodorsally for 4-5 mm, an- terior portion reticulated, striae on the posterior region incomplete (Fig. 19-6). Eyes very small; ears small, partially concealed by the pelage, just reaching the posterior corner of the eye if pressed forward. Forefeet broad with enlarged claws, no- ticeably longer than those of the hind feet. Skull very elongated and gracile (Fig. 19-7), with a slender, elongated, markedly attenuated rostrum; the premaxillae meet the nasals above P2; the na- sals extend posteriorly into the elongated inter- orbital region. The braincase is rounded, moder- ately broad and long; the parietals are small rel- ative to the long frontals; the occipital is relatively shallow. The pterygoid foramina are partially or completely bridged by the palatine lip, which forms an incomplete or complete transverse bar. The mandible is elongated, tapers anteriorly, and the mental foramen lies below p2 or anterior to p3. The dentition is reduced (Fig. 19-8a,b); upper incisors subequal in height, incisors and canine very slender; extensive diastemata between all of the anterior teeth from II to P3, diastema partic- ularly long between P2 and P3; lingual cingulum sometimes present on P2, posterolingual acces- sory cusp absent; talon on P4 narrow; talons on molars very reduced, resembling cingula. The lower canine lacks an accessory cusp; the talonid of m3 is slightly reduced and the entoconid is lacking. Measurements — External and cranial measure- ments are presented in Tables 19-1 and 19-3. Variation — The small sample includes an adult male and female and a juvenile male. The dentition of the juvenile (Fig. 19-9) is at Stage 1 of MacPhee (1987, p. 13), at which most of the antemolar teeth are deciduous, except for 13 and i3, whereas II has just started to erupt, and the upper and lower molars are fully erupted, as usu- al in the genus. Features of the deciduous den- tition that are common to Microgale (small size of the anterior teeth and the resemblance of the 210 Table 19-3. Table of comparative measurements (in mm) of Microgale gracilis and M. gymnorhyncha from the RNI d'Andringitra. Measurements of the holotype of M. gracilis are included in square brackets. M. gymno- Parameter rhyncha M. gracilis Head and body 83-95 91-105 [ca.93] length 91.5 ± 4.92 96.3 ± 5.26 (4) (4) Tail length 59-63.3 75-84 [81] 60.8 ± 1.48 79.3 ±3.19 (4) (4) Ratio of tail length 2.3-2.4 2.7-3.0 [2.8] to condyloincisive 2.3 ± 0.05 2.8 ± 0.13 length (3) (4) Condyloincisive 25.6-26.5 27.9-29.6 [29.0] length 26.0 ± 0.39 28.8 ± 0.72 (3) (4) Upper toothrow 13.5-14.1 14.0-14.9 [14.4] length 13.8 ± 0.24 14.4 ± 0.38 (4) (4) Length of anterior 7.0-7.6 8.1-9.0 [8.4] dentition I-P3 7.4 ± 0.23 8.5 ± 0.34 (4) (4) Ratio of anterior 51.9-54.7 57.8-60.4 [58.3] dentition to upper 53.6 ± 1.11 58.8 ± 1.05 toothrow length (4) (4) Rostral breadth at 2.8-3.0 2.4-2.7 [2.5] level of P2 2.9 ± 0.08 2.6 ± 0.11 (4) (4) Braincase length 7.4-7.9 8.4-8.8 [9.0] 7.6 ± 0.22 8.7 ± 0.15 (3) (4) buccal cusps on the deciduous third upper pre- molar [dP4] to those of Ml and M2) also occur in this specimen. This juvenile is, however, un- usual in the presence of an additional cuspid be- tween the principal and posterior cusp on the first lower deciduous premolar (dp2) and between the metaconid and the posterior cusp on the third lower deciduous premolar (dp3); these cuspids are lacking in adult M. gracilis and have not been observed in other juvenile Microgale. The RNI d'Andringitra specimens closely resemble the holotype. In the small sample from the re- serve, the pelage of the Marositry specimens is marginally longer than that of those the Volot- sangana River. Reproduction — An adult male (FMNH 151648), based on dentition, had abdominal testes measuring 7X3 mm, with a convoluted epidid- ymis. There are no data on the reproductive con- dition of the other two individuals. Remarks — This species is considered to be rare in museum collections (MacPhee, 1987) and is confirmed only from the holotype and the speci- FIELDIANA: ZOOLOGY mens recorded above. Unconfirmed records in- clude a specimen from Ankeramadinika (see Ma- jor, 1896a) and an illustration in Leche (1907) of the "milk" dentition of a specimen that is prob- ably correctly attributed to this species. The three specimens from the 1993 mission to RNI d'An- dringitra are therefore important to increase knowledge of the taxon. The two MNHN speci- mens (MNHN 1972-606 and 1972-607) are from the 1970-1971 Andringitra expedition and were taken in "foret Agauria." Based on the date of collection, this would have been at their camp 4 (Marositry), at about 2000 m (Paulian et al., 1971). Thus, the elevational range of this species on the Andringitra Massif appears to be between about 1200 and 2000 m. Microgale gymnorhyncha, new species Microgale gracilis (Major): MacPhee, 1987, in part Holotype— FMNH 151807: adult female in al- cohol, skull extracted (field number SMG 6697). Collected December 13, 1993, by S.M.G. and C.J.R. Type Locality — 38 km S Ambalavao, RNI d' Andringitra, on ridge E of Volotsangana River, Fianarantsoa Province, 22° 1 1 '39"S 46°58' 16"E, al- titude 1625 m. Paratypes— FMNH 151808: old adult female, intact body in alcohol, (field number SMG 6698) collected December 13, 1993; FMNH 151726: adult female, skull, and complete skeleton (field number SMG 6712), collected December 14, 1993; FMNH 151812: infant, in alcohol, skull ex- tracted (field number SMG 6724) collected De- cember 15, 1993, all with the same collection data as the holotype. FMNH 151650: juvenile male, skin and skull (field number SMG 6579) collected December 1, 1993; FMNH 151651: juvenile male, skin and skull (field number SMG 6614) Decem- ber 4, 1993, both collected by S.M.G. and C.J.R., 40 km S Ambalavao, RNI d' Andringitra, along Volotsangana River, Fianarantsoa Province, 22°13'22"S 46°48'18"E, altitude 1210 m. BM(NH) record number 1995. R258: adult, in alcohol, skull extracted, collected 1993 from unspecified local- ity in the RNI d' Andringitra, Fianarantsoa Prov- ince. Referred Material— MNHN 1961-204: 5 km from Fanovana (18°55'S 48°34'E, altitude ca. 600-800 m [see Carleton & Schmidt, 1990]), near to Perinet (Andasibe). Diagnosis — Rhinarium very large, naked pos- terodorsal extension with transverse striae. Ears small, virtually concealed by the pelage. Skull pyriform, rostrum elongated, blunt anteriorly, braincase short. Description — Moderately large (see Tables 19-1 and 19-3), tail shorter than head and body. Pelage soft, lustrous gray-brown dorsally, gray ventrally; dorsal hairs are three-banded, with light gray bases, light buff terminally, with bright brown tips; guard hairs dark brown or black; ven- tral hairs have light gray bases and brown tips. The feet are light-colored and the tail darker dor- sally, grading into the paler ventral surface. Muz- zle very long, forming a proboscis protruding an- teriorly well beyond mouth; rhinarium very large, transversely striated naked region extends postero- dorsally for ca. 6-7 mm (see Fig. 19-6). Eyes very small. Ears small, virtually concealed in the pel- age, anterior border lies far behind eye when pressed forward against head. Forefeet broad, claws enlarged. Skull long, moderately gracile and pyriform in shape; the anterior part of the skull is elongated in appearance but the posterior interorbital and braincase region appear somewhat anteroposteriorly shortened (see Fig. 19-7). The rostrum is slender and elongated but not notice- ably tapered, and the anterior is blunt and the na- sals slightly flared anteriorly; the nasals extend posteriorly into the interorbital region; the pre- maxillae meet the nasals at a level between C and P2. The braincase is short and broad; the superior articular facets are rectangular and visible in dor- sal view; the frontals and parietals are moderately small and the occipital moderately deep; the su- praoccipital suture is dorsally positioned on the braincase. The palatine lip forms the anterior bor- der of the pterygoid foramina. The mandible is sinuous, moderately elongated, and the mental fo- ramen lies below p3. The dentition is moderately reduced, and long diastemata are present between all of the anterior teeth from II to P3 (see Fig. 19-8b). The first upper incisor is slightly pro- odont, with a well-developed distostyle approxi- mately one-third the height of the principal cusp; II is larger than 12 and 13, and subequal to C; 12 is subequal in crown height to 13 but slightly less than C. The upper canine has a prominent disto- style and a small anterior accessory cusp. The first upper premolar (P2) is tricuspid; P3 is premolar- iform and resembles P4, with a small distinct me- sostyle, an anterior ectostyle, distostyle, and a JENKINS ET AL.: SHREW TENRECS 211 posterolingual accessory cusp; the mesostyle is present on P4, the anterior ectostyle is distinct, the distostyle and posterior ectostyle are present, and the talon is well marked and bicuspid. The molars have narrow but well-marked talons (see Fig. 19- 8c). The first lower incisor (il) is procumbent with a distinct hypoconulid and a well-developed lingual cingulum. The lower canine has a promi- nent accessory cusp. The first lower premolar (p2) is tricuspid, taller than the canine, and with a ro- bust principal cusp. Diastemata are present be- tween all teeth from i2 to p3. The second lower premolar (p3) is similar to but larger than p2. The talonid of the third lower molar (m3) is slightly reduced; a talonid basin, hypoconid, hypoconulid, and entoconid ridge are present with a slight trace of an entoconid (see Fig. 19-8c). Deciduous Dentition — Of the three immature specimens belonging to this species, one is an in- fant (FMNH 151812) in which M2 and m2 are partially erupted and M3 and m3 are just begin- ning to erupt; the other two are juveniles (FMNH 151650 and 151651) with fully erupted molars but a completely deciduous antemolar dentition. The deciduous dentition of this species (Fig. 19-9) shows similar traits to that of other species of Mi- crogale (see MacPhee, 1987), except that the de- ciduous incisors and canines are only slightly smaller than the permanent teeth. The lower sec- ond deciduous incisor (di2) has a prominent lin- gual cingulum unlike i2, and the lower deciduous canine (dc) has a small anterior accessory cusp unlike the permanent lower canine. Measurements — External and cranial measure- ments are presented in Tables 19-3 and 19-4. Variation — The pelage of the single specimen from Fanovana is harsher in texture than that of the RNI d'Andringitra specimens. Reproduction — All three of the adult females collected during the first half of December had large mammae; two specimens were lactating and one had three placental scars (one on left and two on right oviducts). Mammae formula: 2-0-4 (N = 2). Etymology — The name of this species is de- rived from the Greek gymno — naked, rhynch — snout or nose, and refers to the prominent, naked rhinarium. Comparison with Other Species — With the ex- ception of M. gracilis, M. gymnorhyncha is readi- ly distinguished from all other species of Micro- gale recorded from the RNI d'Andringitra by fea- tures including the long muzzle and elongated rhi- narium, the small ears partially concealed in the pelage, the broad forefeet with enlarged claws, and the elongated rostrum and short braincase characteristic of the skull. Microgale gymnorhyncha is remarkably similar morphologically to M. gracilis and there are few differences in external appearance between the two species, which presumably share many adap- tive features. Microgale gracilis is slightly larger than M. gymnorhyncha (see table 19-3), with a longer tail relative to the head and body, In both species the muzzle is prolonged into a flexible proboscis; both have a large rhinarium, but that of M. gracilis is smaller ventrally and the pos- terodorsal extension is shorter (4-5 mm long), with a reticulated anterior portion and incomplete striae on the posterior portion. The eyes of M. gracilis are similarly nearly concealed but are slightly larger than those of M. gymnorhyncha, and the ears are also slightly larger, only partially concealed by the pelage, just reaching the poste- rior border of the eye if pressed forward. Both have broad forefeet with enlarged claws, but the hind foot is shorter in M. gymnorhyncha than in M. gracilis. In contrast to M. gymnorhyncha, the skull of M. gracilis is even more gracile and elon- gated, with a slender, elongated, markedly atten- uated rostrum and a longer interorbital region and braincase; the premaxillae meet the nasals above P2; the braincase is rounded and longer; the pter- ygoid foramina are partially or completely bridged by the palatine lip, which forms an in- complete or complete transverse bar. The mandi- ble of M. gracilis is elongated and anteriorly ta- pering, and the mental foramen lies below p2 or anterior to p3. The dentition of M. gracilis is grac- ile and more reduced than that of M. gymnorhyn- cha, with extensive diastemata between all of the anterior teeth, so that the anterior portion is more elongated relative to the whole toothrow than in M. gymnorhyncha (see Table 19-3). In contrast to M. gymnorhyncha, the upper incisors in M. grac- ilis are subequal in height, the incisors and canine are very slender, and the diastema between P2 and P3 is much longer; a lingual cingulum may be present on P3, but there is no posterolingual ac- cessory cusp; the talon on P4 is narrow; the talons on the molars are very reduced and resemble cin- gula; the lower canine lacks an accessory cusp; the talonid of m3 is more reduced and lacks an entoconid. There are few differences in the decid- uous dentitions of the two species, except that a metaconid is present on the first deciduous lower premolar (dp2) and an accessory cuspid is present posterior to the metaconid on the second lower 212 FIELDIANA: ZOOLOGY Fig. 19-9. Top: left anterior deciduous dentition of M. gymnorhyncha FMNH 151651 (above) and M. gracilis FMNH 151649 {below). Bottom: buccal view of upper dentition (above); lingual view of lower dentition (below). Scale 1 mm. deciduous premolar (dp3) of M. gracilis, in con- trast to M. gymnorhyncha, which lacks a meta- conid on dp2 and shows no trace of an accessory cuspid on dp4 (although a metaconid is present on dp3, contra MacPhee [1987, p. 18], who stated that a metaconid on dp3 is usually absent in all species other than M. parvula). Microgale gymnorhyncha is somewhat similar in external appearance to M. thomasi Major, 1896a, a species that has not been recorded from the RNI d'Andringitra but is known to occur in eastern Madagascar from Ivohimanitra (ca. 20°42'S 47°35'E) to Vondrozo (ca. 22°49'S 47°20'E) (see MacPhee, 1987). Microgale tho- JENKINS ET AL.: SHREW TENRECS 213 masi differs from M. gymnorhyncha in the follow- ing features: the muzzle is shorter, and although the rhinarium is moderately large, it does not ex- tend posteriorly; the ears are obvious and the fore- feet are not broadened, nor are the foreclaws en- larged. The skulls of both species are similar in size but differ in their proportions — that of M. thomasi is more robust. In contrast to that of M. gymnorhyncha, the rostrum of M. thomasi is mod- erately broad and short; the braincase is rounded, broader, longer, and deeper; the parietals are larg- er relative to the frontals and occipital; the pter- ygoid foramina open buccally into the posterior part of the palate; the mandible is robust and not sinuous, and the mental foramen lies below the posterior root of p2. The two species differ den- tally: in M. thomasi the upper incisors and canine are not reduced, the diastemata between II and P3 are much smaller, P2 is larger, the lingual shelves on the upper molars are larger; p2 is very large, the talonid of m3 is similar, although the talonid basin is slightly broader. Remarks — Currently M. gymnorhyncha is known only from the localities listed above in the RNI d'Andringitra, Fianarantsoa Province, south- ern Madagascar and from Fanovana, central east- ern Madagascar. All of the specimens from RNI d'Andringitra were trapped on slopes or ridges in montane forests at an altitude of 1210 m or in upper montane forest at 1625 m. No altitude was recorded for the specimen from Fanovana; it is estimated as 600-800 m based on the relief of the area. No observations have been made on the behav- ior of this species. Although it is inadvisable to interpret morphological features in behavioral terms without direct observations, the combina- tion of such features as the long muzzle with a large, naked rhinarium, the small eyes and par- tially concealed ears, and the broad forefeet with enlarged foreclaws suggests that this animal may use its forefeet and snout to rootle in the humus layer. The elongated rostrum and well-spaced teeth also suggest dietary specializations, al- though the identifiable gut contents of a single specimen included only arthropod fragments. Microgale sp. A Referred Material — FMNH 151646 and 151647: 40 km S Ambalavao, RNI d'Andringitra, along Volotsangana River, Fianarantsoa Province, 22°13'22"S 46°58'18"E, altitude 1210 m; BM(NH) 95.257: Basecamp 1, Maitso, RNI d'Andringitra, ca. 22°10'S, 46°50'E. Description — In preparation (RD.J. et al.). Measurements — External and cranial measure- ments of the specimens from RNI d'Andringitra are given in Table 19-1. Variation — This species has been collected at several widely separated localities. Specimens from the RNI d'Andringitra population are larger on average than any of the other populations in external and cranial dimensions. In particular, the braincase of the RNI d'Andringitra specimens is broader and deeper than that of Pare National (PN) de la Montagne d'Ambre specimens. The two specimens from the PN de Ranomafana, al- though geographically closer to the RNI d'An- dringitra specimens in southeastern Madagascar, are, however, more similar to the PN de la Mon- tagne d'Ambre specimens in size, as is the single juvenile specimen from the Reserve Speciale (RS) d'Ambatovaky. Reproduction — Both of the RNI d'Andringitra specimens collected in early December were preg- nant females with permanent dentition. One in- dividual (FMNH 1 5 1 646) had three embryos (two left and one right) measuring 15 mm crown-rump length, and the other (FMNH 151647) had four embryos (two left and two right) measuring 17 mm crown-rump length. Mammary formula: 2-0-4 (N = 2). Discussion Although there is still doubt concerning the number of distinct species of Microgale occurring in Madagascar, the 10 species here recorded for RNI d'Andringitra form a high proportion of the 17 species currently recognized by us. Sympatric association between several species is a common phenomenon in the genus Microgale, and two to five species were recorded from half of the col- lecting localities mapped by MacPhee (1987). In areas that have been studied intensively, even greater numbers of species have been recorded. For example, six species were recorded from PN de la Montagne d'Ambre (Raxworthy & Nuss- baum, 1994) and seven from RS d'Analamazaotra (Nicoll & Langrand, 1989). This high level of sympatry of RNI d'Andringitra is greater than that observed to date at any other locality, and al- though it certainly reveals the high diversity of the area, it may also reflect the thoroughness of 214 FIELDIANA: ZOOLOGY the inventory. The ecological implications of the speciosity of the area are discussed in Chapter 20. Although MacPhee (1987) emphasized that the clusters he used to group species were a purely phenetic device, these clusters continue to be a very helpful method of associating species that share morphological features. The species occur- ring in the RNI d'Andringitra may be placed in these clusters, which are recorded below with a summary of the features shared by the included species: cowani cluster — Microgale cowani, M. taiva, M. melanorrhachis (M. parvula less certainly associated) — craniodental similarities, foot proportions unmodified. gracilis cluster — M. gracilis, M. gymnorhyn- cha — craniodental similarities, forefeet broad, foreclaws enlarged, rhinarium modi- fied. longicaudata cluster — M. longicaudata — cra- niodental characters, elongated hind feet, elongated tail. dobsoni cluster — M. dobsoni — craniodental characters, foot proportions unmodified. soricoides cluster — M. soricoides — cranio-den- tal characters. pusilla cluster — not represented in the collec- tions from RNI d'Andringitra. brevicaudata cluster — not represented in the collections from RNI d'Andringitra. Limited data on the annual reproductive cycle were provided for M. dobsoni and M. talazaci by Eisenberg and Gould (1970), both from trapping programs and from captive animals. More recent- ly, Stephenson and Racey (1993) provided infor- mation on reproductive energetics in M. talazaci and M. dobsoni, as well as limited data on M. cowani and M. melanorrhachis. The data on re- productive condition for all 10 species collected at RNI d'Andringitra are presented in the preced- ing pages, and, although confined to the months of November and December when the trapping program was in operation, they provide a consid- erable increase of information over that previous- ly available for the genus. Some trends are evident within most species of Microgale found at the RNI d'Andringitra. On the basis of dentition, both adults and juveniles were found in most species, with the exception of M. melanorrhachis (juveniles only) and M. dobsoni, M. soricoides, and Microgale sp. A (adults only). On the basis of reproductive condition of the go- nads, sexually adult animals of both sexes were evident in M. cowani, M. taiva, and M. longicau- data only, but sexually adult animals of one sex were found in the other species. Species in which dental juveniles showed some signs of sexual ma- turity included M. cowani, M. parvula, and pos- sibly also M. melanorrhachis, where individual females were perforate but showed no other evi- dence of sexual maturity, and M. taiva, in which an individual male had well-developed testes. Few conclusions may be drawn from these results, al- though there is evidence to suggest that dental ju- veniles may occasionally be recruited into the sexually adult population. Whether this indicates sexual precocity of juveniles or retardation of dental development into adulthood is impossible to decide from these limited data. Dental development among juveniles of all spe- cies appeared to be at a fairly early stage, and there was no evidence of any development be- yond Stage 1 to 2. This observation may coincide with the months of collection, November and De- cember; these months are also at the beginning of the annual rainy season in eastern humid forest. The breeding season for this genus presumably starts earlier in the year, and it is probable that the presence of juveniles of this age is positively correlated with the start of the rainy season when food supplies should begin to increase. Individu- als in which the molars were only partially erupt- ed occurred in M. cowani, M. taiva, M. longicau- data, and M. gymnorhyncha. Individuals with ful- ly erupted molars but completely deciduous an- temolar teeth were found in M. cowani, M. taiva, M. melanorrhachis, M. longicaudata, M. parvula, and M. gymnorhyncha. Individuals showing Stage 1 to 2 states of development were found in M. cowani, M. taiva, and M. gracilis. It is probably significant that those species most commonly cap- tured, M. cowani and M. taiva, also demonstrated the greatest range of variation in dental develop- ment of juveniles. Acknowledgments For the loan of specimens in their care, we thank Laurent Granjon, Michel Tranier, and Jacques Cuisin, Mammiferes et Oiseaux, MNHN, Paris, and Michael Carleton, Marc Frank, and He- len Kafka, Department of Mammalogy, USNM, Washington, D.C. Maria Rutzmoser, MCZ, Har- vard, kindly provided information on the holotype of M. parvula. Members of the 1993 Cambridge JENKINS ET AL.: SHREW TENRECS 215 University expedition, Louise Ashmore, Juliet O'Keefe, Matthew Thomas, and Oliver Tunstall Pedoe, made a small collection of specimens. For the opportunity to examine these species we thank them and, particularly, Nasolo Rakotoarison, Pare Botanique et Zoologique de Tsimbazaza for his cooperation in the loan of material. During this study the salary of C.J.R. was provided by a grant (BSR 90-24505) from the National Science Foun- dation. The photographs were prepared by Phillip Crabbe, Photographic Unit, BM(NH). Literature Cited Butler, P. M., and M. Greenwood. 1979. Soricidae (Mammalia) from the Olduvai Gorge, Tanzania. Zoo- logical Journal of the Linnean Society, 67: 329-379. Carleton, M., and D. F. Schmidt. 1990. Systematic studies of Madagascar's endemic rodents (Muroidea: Nesomyinae); an annotated gazetteer of collecting lo- calities of known forms. American Museum Novita- tes, 2987: 1-36. Coquerel, C. 1848. Note sur une espece nouvelle de musaraigne trouvee a Madagascar. Annales des Sci- ences Naturelles, Zoologie, 9: 193-198. Eisenberg, J. E, and E. Gould. 1970. The tenrecs: A study in mammalian behavior and evolution. Smith- sonian Contributions to Zoology, 27: 1-138. Froriep, L. F. 1806. C. Dumeril's Analytische Zoologie aus dem Franzosischen, mit Zusatzen. Weimar, 345 pp. Genest, H., and F Petter. 1975. Part 1.1. Family Ten- recidae, pp. 1-7. In Meester, J., and H. W. Setzer, eds., The mammals of Africa: An identification manual. Smithsonian Institution Press, Washington, D.C. Grandidier, A. 1870. Description de quelques animaux nouveaux, decouverts a Madagascar. Revue et Maga- sin de Zoologie, 22: 49-54. Grandidier, G. 1934. Deux nouveaux mammiferes in- sectivores de Madagascar. Bulletin Museum National d'Histoire Naturelle, ser. 2, 6: 474-477. Heim de Balsac, H. 1972. Insectivores, pp. 629-660. In Battistini, R., and G. Richard- Vindard, eds., Bio- geography and ecology in Madagascar. W Junk, the Hague, 765 pp. Hutterer, R. 1993. Order Insectivora, pp. 69-130. In Wilson, D. E., and D. M. Reeder, eds., Mammal spe- cies of the world: A taxonomic and geographic ref- erence. Smithsonian Institution Press, Washington, D.C, 1206 pp. , and M. tranier. 1990. The immigration of Asian house shrew (Suncus murinus) into Africa and Madagascar, pp. 309-320. In Peters, G, and R. Hut- terer, eds. Vertebrates in the tropics. Museum Alex- ander Koenig, Bonn, 585 pp. Jenkins, P. D. 1988. A new species of Microgale (In- sectivora: Tenrecidae) from northeastern Madagascar. American Museum Novitates, 2910: 1-7. . 1992. Description of a new species of Micro- gale (Insectivora: Tenrecidae) from eastern Madagas- car. Bulletin of the British Museum Natural History (Zoology), 58: 53-59. 1993. A new species of Microgale (Insectiv- ora: Tenrecidae) from Eastern Madagascar, with an unusual dentition. American Museum Novitates, 3067: 1-11. Lacepede, B. G. E. de 1799. Tableau des divisions, sous-divisions, ordres et genres des mammiferes. Par- is. Leche, W 1907. Zur Entwicklungsgeschichte des Zahn- systems der Saugetiere, zugleich ein Beitrag zur Stam- mesgeschichte dieser Tiergruppe. Zweiter Teil: Phy- logenie. Zweites Heft: Die Familien der Centetidae, Solenodontidae und Chrysochloridae. E. Schweizer- bartsche (E. Nagele), Stuttgart, 157 pp. Linnaeus, C. 1766. Systema naturae per regna tria na- turae secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis. 12th edi- tion. Holmiae. MacPhee, R. D. E. 1987. The shrew tenreces of Mad- agascar: Systematic revision and Holocene distribu- tion of Microgale (Tenrecidae: Insectivora). American Museum Novitates, 2889: 1-45. Major, C. I. Forsyth 1896a. Diagnoses of new mam- mals from Madagascar. Annals and Magazine of Nat- ural History, (6) 18: 318-325. . 1896b. Descriptions of four additional new mammals from Madagascar. Annals and Magazine of Natural History, (6) 18: 461-463. 1897. On the general results of a zoological expedition to Madagascar in 1894-96. Proceedings of the Zoological Society of London, for 1896: 971-981. Martin, W 1838. On a new genus of insectivorous Mammalia. Proceedings of the Zoological Society of London, for 1838: 17-19. Mills, J. R. E. 1966. The functional occlusion of the teeth of Insectivora. Journal of the Linnean Society (Zoology), 47: 1-125. Milne Edwards, A., and A. Grandidier. 1872. De- scription d'un nouveau mammifere insectivore de Madagascar (Geogale aurita). Annales des Sciences Naturelles, Zoologie, 15, Article 19: 1-5. Mivart, St. George. 1871. On Hemicentetes, a new genus of Insectivora, with some additional remarks on the osteology of that Order. Proceedings of the Zoo- logical Society of London, for 1871: 58-79. Morrison-Scott, T. C. S. 1948. The insectivorous gen- era Microgale and Nesogale (Madagascar). Proceed- ings of the Zoological Society of London, 118: 817- 822. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la conservation et des Aires Protegees. World Wide Fund for Nature, Gland, Switzerland, XVII+374 pp. , and G. B. Rathbun. 1990. African Insectivora and Elephant-shrews: An action plan for their conser- vation. IUCN/SSC Insectivore, Tree-shrew and Ele- phant-shrew Specialist Group, Gland, Switzerland, iv+53 pp. O'Keefe, J., and L. Ashmore. [1994]. Madagascar 1993 small mammal project. Unpublished report. Uni- versity of Cambridge. 216 FIELDIANA: ZOOLOGY Paulian, R., J.-M. Betsch, J.-L. Guillaumet, C. Blanc, and P. Griveaud. 1971. RCP 225. Etude des ecosys- temes montagnards dans le region malgache. I. Le massif de l'Andringitra. 1970-1971. G6omorphologie, climatologie et groupements v6g&aux. Bulletin de la Soci6te" d'Ecologie, 11(2-3): 189-266. Raxworthy, C. J., and R. A. Nussbaum. 1994. A rain- forest survey of amphibians, reptiles and small mam- mals at Montagne d'Ambre, Madagascar. Biological Conservation, 69: 65-73. Stephenson, P. J. 1995. Taxonomy of shrew-tenrecs (Microgale spp.) from eastern and central Madagascar. Journal of Zoology, 235: 339-350. , and P. A. Racey. 1993. Reproductive energet- ics of the Tenrecidae (Mammalia: Insectivora). 2. The shrew-tenrecs, Microgale spp. Physiological Zoology, 66: 664-685. Swindler, D. R. 1976. Dentition of Living Primates. Academic Press, London, 308 pp. Thomas, [M. R.] Oldfield. 1882. Description of a new genus and two new species of Insectivora from Mad- agascar. Journal of the Linnean Society (Zoology), 16: 319-322. 1 884. Description of a new species of Micro- gale. Annals and Magazine of Natural History, (5) 14: 337-338. 1918. On the arrangement of the small Ten- recidae hitherto referred to Oryzorictes and Microga- le. Annals and Magazine of Natural History, (9) 1: 302-307. Appendix 19-1. Key to the Species of Microgale Occurring in RNI d'Andringitra 1- Size very small: HB < 64, CIL < 16.6; TL subequal to HB 0.97-1.03; dark brown dorsal and ventral pelage M. parvula Size larger: HB > 63, CIL > 17.3; pelage not dark brown dorsally and ventrally 2 2- Ratio of TL:HB 1.7-1.9; dorsal pelage reddish M. longicaudata Ratio of TL:HB < 1.4 3 3- Digits and tail tip contrastingly paler than body, tail, and feet Microgale sp. A. Tail tip and digits not obviously paler than rest of body 4 4- Dorsal pelage with a distinct, dark mid-dorsal stripe extending from the head to the base of the tail; tail bicolored, dark brown above, lighter below M. melanorrhachis No distinct mid-dorsal stripe 5 5- Size very large: HB > 108, CIL > 30.0; i2 » c M. dobsoni Size smaller: HB < 105, CIL < 30.0; i2 subequal or > c 6 6- Proboscis long, large rhinarium extends posterodorsally onto muzzle; forefeet broad, foreclaws en- larged 7 Small rhinarium confined to anterior of short proboscis; forefeet slender without lengthened foreclaws 8 7- Posterior region of rhinarium with transverse striae; ratio of I1-P3:UTL < 55.0; BL < 8.0 M. gymnorhyncha Posterior region of rhinarium reticulated; ratio of I1-P3:UTL > 57; BL > 8.4 M. gracilis °- II robust, markedly proodont; il » i2 > c; P2 and p2 very small with single roots M. soricoides II neither robust nor markedly proodont; il < or subequal to i2; P2 and p2 with two roots 9 9- Ratio of TL:CIL 2.7-3.1; ratio of I1-P3:UTL 52.8-56.4 M. cowani Ratio of TL:CIL 3.5-4.2; ratio of I1-P3:UTL 49. 1-52.8 M. taiva JENKINS ET AL.: SHREW TENRECS 217 Chapter 20 Insectivore Ecology in the Reserve Naturelle Integrate d'Andringitra, Madagascar Steven M. Goodman, Christopher J. Raxworthy, and Paulina D. Jenkins Abstract Insectivores were surveyed on the eastern side of the Reserve Naturelle Integrate d'Andrin- gitra, at four elevational zones between 720 and 1625 m. Little previous information was available on the small mammals of the reserve. A total of 12 species of insectivores (Setifer, Tenrec, and 10 species of Microgale) were found in the reserve during the 1993 survey, and with earlier records and specimens, a total of 14 species are known from the Andringitra Massif. Microgale taiva and M. principula occurred across the complete transect, M. melanorrhachis was restricted to the lowland forest, and Microgale sp. A, M. dobsoni, M. soricoides, M. gracilis, and M. gymnorhyncha were only captured in montane forest. On the basis of trap results (principally pitfall buckets with drift fence), the highest diversity and density of Insec- tivora was at 1210 m elevation, with eight species. Pitfall buckets placed in valley bottoms had a higher capture rate and species diversity than those on slopes or ridges, although there was no clear pattern of species habitat segregation. Collections of soil invertebrates were made adjacent to the pitfall lines, and the general trends of invertebrate density and Microgale density and abundance paralleled each other. Resume L'inventaire des insectivores du versant est de la Reserve Naturelle Integrate d' Andringitra a ete effectue dans quatre zones d' altitude comprises entre 720 m et 1625 m. Peu d' informations relatives aux petits mammiferes de la reserve etaient jusqu'alors disponibles. Au cours de l'inventaire de 1993, un total de 12 especes d' insectivores {Setifer, Tenrec et 10 especes de Microgale) a ete trouve dans la reserve. Avec les specimens recenses auparavant, un total de 14 especes est connu du massif d' Andringitra. Microgale taiva et M. principula sont presents sur l'ensemble du transect. M. melanorrhachis est limitee a la foret de basse altitude. Microgale sp. A, M. dobsoni, M. soricoides, M. gracilis et M. gymnorhyncha ont ete capturees dans la foret de montagne. D'apres les resultats des pieges (principalement les "pitfall"), la plus grande diversite et densite d' insectivores est constatee a une altitude de 1210 m avec huit especes. Les pieges "pitfall" places dans les vallees recelent un taux de capture et une diversite spe- cifique plus eleves que ceux places sur les pentes et les cretes, bien qu'il n'y ait aucune evidence de segregation d'habitat par les especes. La collecte d'invertebres du sol a ete faite parallele- ment aux lignes "pitfall" et la densite et l'abondance des invertebres et des Microgale montrent une tendance generate parallele. 218 FIELDIANA: ZOOLOGY Introduction Little is known about the natural history of the tenrecs (Order Insectivora, Family Tenrecidae), all of which, with the exception of the African otter- shrews (Subfamily Potamogalinae), are endemic to Madagascar. The tenrecs are speciose, broadly distributed, and occur in a variety of biomes on the island. They represent a group that shows a remarkable degree of morphological and ecolog- ical diversification. The most in-depth study of the behavioral ecol- ogy of the tenrecs to date is by Eisenberg and Gould (1970). It was conducted at sites in the eastern humid forest, particularly in the Reserve Speciale (RS) d'Analamazaotra (Perinet), 400 km NE of the Reserve Naturelle Integrate (RNI) d'Andringitra. The tenrecs have also been the fo- cus of reproductive and physiology studies (Ni- coll, 1982; Stephenson, 1991; Stephenson & Ra- cey, 1993). The purpose of this study was to document in- sectivore distribution along an elevational gradi- ent on the eastern slopes of the RNI d'Andringitra and to present basic natural history information on these animals. Abiotic variables (climate) and biotic variables (macrofauna soil invertebrates) were compared to our trapping data to explore possible correlations and explanations of the al- titudinal distribution, density, and species richness of Microgale along the slopes of the Andringitra Massif. Table 20-1. The known elevational distribution of Insectivora on the eastern slopes of the RNI d'Andrin- gitra. Includes sight records, live traps, and pitfall traps from 1993 survey. Materials and Methods Trap Lines The principal means of capturing insectivores was with pitfall buckets and a drift fence. In the four transect zones (720, 810, 1210, and 1625 m), three separate pitfall lines (one in valley bottom, one on a slope, and one on a ridge crest), each 100 m long and consisting of 11 buckets, 10 m apart, were in operation for a minimum of 6 nights. More details on the technique are given in Chapter 17. A few insectivores were also captured with standard Sherman live traps. The trap types, placement, baits, etc. for these lines are described in Chapter 22. Traps and pitfalls were visited at least twice per day, once at dawn and again in the late afternoon. A "trap-day" and "bucket-day" are defined as a Elevational range (m) Species 720 810 1210 1625 Microgale cowani - + + + Microgale taiva + + + + Microgale melanorrhachis + + - - Microgale longicaudata + - - + Microgale parvula + + + + Microgale dobsoni - - + - Microgale soricoides - - + + Microgale gracilis - - + - Microgale gymnorhyncha - - + + Microgale sp. A - - + - Setifer setosus - + - - Tenrec ecaudatus + + - - Total number of Microgale spp. 4 4 8 6 Total number of Insectivora 5 6 8 6 24-hour period (dawn to dawn) of one of these devices in use. After rains the buckets were sponged dry. Our inventory of the RNI d'Andrin- gitra was conducted during a period (November 14 to December 18, 1993) when all tenrecs were expected to be active (Stephenson, 1994). The Known Insectivoran Community of the RNI d'Andringitra A taxonomic analysis of the Microgale ob- tained during the 1993 survey of the eastern slopes of the RNI d'Andringitra is presented in Chapter 19. A total of 10 species were collected (Table 20-1). Two other genera and species of Tenrecidae were also found in the RNI d'Andrin- gitra in 1993; others have been reported from the reserve or surrounding areas. Fourteen species have been documented in the reserve (Table 20-2). Below we present a summary of Insectiv- ora reported from the reserve, excepting the Mi- crogale discussed in Chapter 19. Most of the ad- ditional specimen records, housed in the Mus6um National d'Histoire Naturelle (MNHN), Paris, are from the 1970-1971 Recherche Cooperative sur Programme no. 225 du Centre national de la Re- cherche Scientifique expedition to the Andringitra Massif (see Paulian et al., 1971; and Chapter 19). Taxa presented in brackets represent those whose GOODMAN ET AL.: INSECTIVORE ECOLOGY 219 Table 20-2. The known Insectivora community of the RNI d'Andringitra, based on museum specimens and photographs. Microgale cowani Microgale taiva Microgale melanorrhachis Microgale longicaudata Microgale parvula Microgale dobsoni Microgale soricoides Microgale gracilis Microgale gymnorhyncha Microgale sp. A Oryzorictes tetradactylus Hemicentetes nigriceps Setifer setosus Tenrec ecaudatus occurrence in the reserve are in need of further documentation. Subfamily Oryzorictinae [Limnogale mergulus Major, 1896a This semiaquatic tenrec is known to dismantle crustacean and invertebrate prey on rocks along river margins. Nicoll and Rathbun (1990) reported possible feeding signs of this species along a small river 15 km N of Antanifotsy. No evidence of this species was found in 1993 in the RNI d'Andringitra. Villagers living in and around the eastern edge of the reserve were clearly unfamil- iar with a precise description of this animal, par- ticularly the webbed feet and laterally compressed tail. Nevertheless, they noted that a water rat, "voalavo rano," occasionally was recovered in their eel traps.] identification as a juvenile M. cowani. The ab- sence of M. talazaci in the RNI d'Andringitra is commented on below. Oryzorictes tetradactylus Milne Edwards & Grandidier, 1882 The only known records of this species on the Andringitra Massif are four specimens taken in 1970 during the RCP mission. These include sin- gle specimens obtained along the Plateau d'An- dohariana (2030 m) and near the Cuvette Boby (2470 m), just below Pic Boby (Paulian et al., 1971). Two other specimens taken on the same expedition are simply labeled "Andringitra." The known elevational range of this species in the RNI d'Andringitra is above the forested zone and be- tween 2000 and 2500 m. The labels of these spec- imens bear no information concerning the repro- ductive condition of the animals. Subfamily Tenrecinae Hemicentetes nigriceps Gunther, 1875 Nicoll and Langrand (1989) reported this spe- cies, under the name H. semispinosus, as occur- ring in the RNI d'Andringitra. A photograph taken by O. Langrand of an animal in Antanifotsy is clearly referable to H. nigriceps. Furthermore, Glaw and Vences (1994) photographed H. nigri- ceps within the reserve. We found no evidence of this species during the 1993 expedition. We fol- low Eisenberg and Gould (1970) in considering H. nigriceps specifically distinct from H. semis- pinosus. Microgale Species Two species of Microgale that had been pre- viously reported from the Andringitra Massif on the basis of specimens obtained during the RCP campaign were not collected during the 1993 ex- pedition: M. talazaci Major, 1896a and M. drou- hardi Grandidier, 1934. However, P.D.J, reidenti- fied MNHN 1972-609 (originally identified as M. talazaci; MacPhee, 1987) as a juvenile M. dob- soni Thomas, 1884. MNHN 1972-611 was pre- viously identified as M. drouhardi and was sub- sequently listed by MacPhee (1987) under the synonymy of M. cowani; P.D.J, confirmed its Setifer setosus (Schreber, 1777) Our only record of this species from the 1993 expedition is a male captured in a pitfall trap at 810 m. Nicoll and Langrand (1989) also reported Setifer setosus from the reserve. Tenrec ecaudatus (Schreber, 1777) This species was observed on two occasions at 720 m and thrice at 810 m during the 1993 ex- pedition. No specimen was collected. Nicoll and 220 FIELDIANA: ZOOLOGY Langrand (1989) also reported this species from the reserve. [Family Soricidae Subfamily Crocidurinae Suncus madagascariensis (Coquerel, 1848) Nicoll and Langrand (1989) noted that this spe- cies occurs in the reserve, although it was not col- lected during the 1993 survey. We are unaware of any documentation of this species in the reserve.] Analysis and Discussion General A total of 902 pitfall bucket-days were accrued during the small mammal survey of the eastern slopes of the RNI d' Andringitra, between Novem- ber 14 and December 17, 1993, comprising 198 bucket-days at 720 m, 231 at 810 m, 220 at 1210 m, and 253 at 1625 m (Table 20-3). Ninety-one small mammals were captured, including 88 Mi- crogale, one Setifer, and two rodents. Further, 2,475 trap-nights using small mammal traps were also accrued (Chapter 22). Within the Insectivora, 11 species (10 species of Microgale and the monotypic Setifer) were captured. Seven of these species had not been previously recorded in the reserve (MacPHee, 1987; Nicoll & Langrand, 1989), and two are new to science (Chapter 19). Only one species, Microgale dobsoni, was cap- tured in the small mammal traps but not in the pitfall buckets, whereas five species of Microgale were trapped only in the pitfall buckets. Before proceeding with the analysis of the trapping re- sults, it is critical to determine if the sampling effort was sufficient to reflect some measure of completeness for the survey and the actual ten- recid species richness within each elevation zone. Species Accumulation Curves By plotting the total number of species known from each pitfall line ( 1 1 pitfall buckets per 24-hour period) or elevational zone (33 pitfall buckets per 24-hour period), species accumulation curves may be drawn. An examination of these curves (Fig. 20-1) shows that no additional spe- cies of Microgale was added in the 720 m zone after 66 pitfall bucket-days (total four species in 198 pitfall bucket-days), in the 180 m zone after 198 pitfall bucket-days (total four species in 231 pitfall bucket-days), in the 1210 m zone after 132 pitfall bucket-days (total seven species in 23 1 pit- fall bucket-days), and in the 1625 m zone after 198 pitfall bucket-days (total six species in 253 pitfall bucket-days). The decrease in the addition of previously unrecorded species with additional trapping effort did not coincide with a general de- cline in overall pitfall trap success (Fig. 20-1). In general, virtually all of the Microgale spe- cies expected to occur within the reserve on the basis of their known distribution in the southern portion of the eastern humid forests were found during the 1993 survey. The major exception was Microgale talazaci, the largest extant species, known from areas north and south of the Andrin- gitra Massif, as well as within the reserve during the 1970-1971 RCP expedition (Paulian et al., 1971). Of the specimens from the reserve, how- ever, MacPhee (1987) listed two, MNHN 1972- 608 and MNHN 1972-612 (a skin only), taken in the upper elevational zone, outside of the area sampled in 1993. Neither of these specimens was examined for this study, but a third specimen (MNHN 1972-609) bearing the same label name was found to be a juvenile M. dobsoni. At other sites in the eastern humid forest, M. talazaci is known from below 1000 m (MacPhee, 1987; Rax- worthy & Nussbaum, 1994). Microgale talazaci shows seasonal variation in activity (Stephenson, 1994); however, this in itself is insufficient to ex- plain the absence of this species during the 1993 survey, because November and December are months in which it has been trapped in consider- able numbers with pitfalls at other sites (Raxwor- thy & Nussbaum, 1994). Further work is neces- sary to determine if M. talazaci occurs within the RNI d' Andringitra. Within the endemic Insectivora there are a few other trapping anomalies. Two other Microgale are known from the general region, yet they were not captured in the RNI d' Andringitra. M. thomasi Major, 1896a, a species rare in collections, was obtained near Vondrozo, a site about 70 km SE of the reserve, and M. pusilla Major, 1 896b, was taken from Vinanitelo, a similar distance to the NE of the reserve (MacPhee, 1987). Although not recorded in such close proximity, M. principula Thomas, 1918, is known from several localities in the eastern forest from the Reserve Speciale (RS) GOODMAN ET AL.: INSECTIVORE ECOLOGY 221 rJ ON NO 00 •* — 00 o > on no oo m — 00 NO O (N r- NO ON NO 1) n e a e 0 tN (S E o > ->' r-^ t-~ tF r- 2a.-<^£ > en On t-» O > NO >-i vO fNl — / NO — NO — •* — rsi no 6 •£ S e 2 £ E E S3 « 5» >> g-o-o >, 3 E « «•§ £, c c & (O cd c c ca •gag I B •3 g e a s •5 .a .a co o 00 Tf O 00 Tt d oo r~- — no — vS fOI~~ NO — — o O oo nn o Tf m >/"> -o m — ro NO NO r- (N O O NO m o) ^ en ,ls-.kk<£o7 u c § §§ § § § j| i§ § 3 £ § Oh U cS O- O to s 5 fe W 8 -J !• E 3 1 a A ■ 720m 810 m 1210 m 1625 m ■ W - ■ i 33 66 99 132 165 trap nights 198 231 264 297 trap nights Fig. 20-1. Species accumulation curves (top) and pitfall trap success (bottom) plotted for each elevational zone against the total number of trap (bucket) days. Information from the three lines at each zone is combined. d'Ambatovaky southward into southern Madagas- car at Andrahomana (MacPhee, 1987; Jenkins, 1992). Nicoll and Langrand (1989) reported Hem- icentetes (semispinosus) nigriceps and Tenrec ecaudatus from the RNI d'Andringitra. The latter species was observed several times at the 720 and 810 m zones but was not captured. During the 1993 survey we did not find evidence of Hemi- centetes in the reserve, although this species has been captured using the same methodology else- where in the eastern humid forest (Raxworthy, un- publ.). We did not search for Hemicentetes bur- rows; this is apparently an effective means of lo- cating this genus (Eisenberg & Gould, 1970). On the basis of the species accumulation curves for each elevational zone, and extrapolation of what should presumably occur in the reserve and what was found, we conclude that the Insectivora fauna, particularly species of Microgale, on the eastern slopes of the RNI d'Andringitra has been thoroughly, although not completely, surveyed. Trapping Success and Abundance In Table 20-3 the results of the pitfall trapping are presented by species, pitfall line, and eleva- tion. There was considerable variation in capture rates of Microgale between lines at the same el- evation and at different zones on the mountain: 6.0% (range 3.0-10.6%) at 720 m, 4.8% (range 3.9-5.2%) at 810 m, 15.2% (range 10.4-18.2%) GOODMAN ET AL.: INSECTIVORE ECOLOGY 223 at 1210 m, and 12.6% (range 5.7-21.6%) at 1625 m. Using standard small mammal traps, insecti- vores were captured only at 1210 (N = 10) and 1625 m (N = 1). Because the pitfall lines were randomly placed in the microhabitats within each elevation zone, the same protocol was used for each line, species accumulation curves reached a plateau, and there is no evidence of preferential capture of any species, we feel that it is reason- able to equate the capture rates of Microgale spe- cies with their relative abundance in each zone during the period of the survey. However, climatic conditions, particularly rainfall patterns, may have affected trap success between the zones. Further- more, animal activity may have changed over the course of the 2-month survey. The only other published study of Microgale using large pitfall buckets in humid forest was by Raxworthy and Nussbaum (1994) in the Pare Na- tional (PN) de la Montagne d'Ambre, in the ex- treme north. The pitfall system they used was identical to that employed in the RNI d'Andrin- gitra, but the lines were in place for between 1 1 and 16 days. In some cases they had remarkable capture rates, with up to 61 Microgale individu- als, representing four species, in a single line dur- ing 121 accrued pitfall bucket-days (50.4%). The use of an identical pitfall system, however, in transitional dry deciduous/sub-arid thorn scrub forest yielded a different result (Goodman & Ganz- horn, 1994; Raxworthy et al., 1994). In 528 ac- crued pitfall bucket-days a total of 1 1 individuals (2.1%), representing three species (Geogale au- rita Milne Edwards & Grandidier, 1872; Setifer setosus; and Suncus madagascariensis), were cap- tured. The species composition in such a forest is not comparable, therefore, to that in humid forest. Stephenson conducted small mammal invento- ries using Sherman traps and randomly placed pit- fall traps (plastic plant propagators), measuring 20 cm in depth and 12 cm in diameter, without a drift fence. In the RS d'Analamazaotra, Stephenson (1993) captured seven species of Insectivora, in- cluding four Microgale spp. Except for single in- dividuals of Oryzorictes hova A. Grandidier, 1870, and Microgale cowani Thomas, 1882, that were trapped during 811 pitfall bucket-days (0.2%), all other individuals and species were ob- tained in Sherman traps. In the RS d'Ambohitan- tely (Stephenson et al., 1994), 345 pitfall bucket- days were accrued and no Microgale sp. was cap- tured, although four species are known to occur in the reserve (Raxworthy, unpubl.). Thus, it would appear that the pitfall technique used by Stephenson (1993) and Stephenson et al. (1994) is less effective than the much larger buckets used with a drift fence. On the basis of our results from the RNI d'Andringitra, large pitfall buckets are much more effective in capturing certain species than are conventional traps. The Relationship of Abiotic Factors and Capture Rates Information on daily minimum and maximum temperatures and precipitation was recorded at each site (Chapter 3). In order to determine if Mi- crogale capture rates were influenced by these cli- matic variables, a series of nonparametric tests were run for each elevational zone using Ken- dall's rank correlation, with the Y variable being the number of animals captured and the X vari- ables as minimum temperature the morning the pitfall buckets were checked, the maximum tem- perature of the previous afternoon, and the amount of precipitation during the previous 24-hour and 48-hour period. The only significant result was at 810 m, between the number of ani- mals captured and the amount of rain in the pre- vious 24-hour period (r = 0.714, N = 7, P = 0.05). In order to allow us to discern general patterns of the relationship between these abiotic variables and the number of animals captured, a series of regression analyses were conducted. In virtually all cases there was a positive relationship between these variables and the number of Microgale cap- tured. For the 720, 1210, and 1625 m zones, how- ever, a few points were outliers of the general pat- tern, and the relationship was not statistically sig- nificant. At 810 m, all of the climatological vari- ables were positively and significantly correlated with the number of animals captured: minimum temperature (r2 = 0.98), maximum temperature (r2 = 0.99), 24-hour rain (r2 = 0.97), and 48-hour rain (r2 = 0.99). If we assume that the general activity rate of Microgale species is correlated with pitfall trap capture rate, then it appears that during periods of higher minimum and maximum temperature, as well as periods of precipitation, there are in- creased movements of these animals. For temper- ature, this relationship is presumably related to metabolic rate, whereas for precipitation it may be a function of increased moisture and flooding of underground burrow systems. 224 FIELDIANA: ZOOLOGY Species Diversity There is elevational variation in species diver- sity of Insectivora on the eastern slopes of the RNI d'Andringitra. Eleven Insectivora, all of which are endemic to the island, were captured in the reserve during the 1993 mission using stan- dard small mammal traps and pitfall buckets. No introduced species (i.e., Suncus murinus Lin- neaus, 1766) was recorded. The number of Insec- tivora known in each elevational level was: 720 m (lines 1-3) — five species, 810 m (lines 4-6) — six species, 1210 m (lines 7-9) — eight species, and 1625 m (lines 10-12) — six species (Table 20-1); for Microgale the species composition was four species at 720 m, four species at 810 m, eight species at 1210 m, and six species at 1625 m. Thus, in general for Insectivora and specifically for Microgale, there appears to have been a mid- elevational peak in species richness at 1210 m. Two species, Microgale taiva Major, 1896b, and the very small M. parvula G. Grandidier, 1934, occur at all elevations, and it is probable that M. longicaudata Thomas, 1882 is also ubiq- uitous but not collected at the intermediate sites. The only species apparently confined to lowland forest in and around the reserve is M. melanor- rhachis Morrison-Scott, 1948. In contrast, five species, M. dobsoni, M. soricoides Jenkins, 1993, M. gracilis Major 1896a, M. gymnorhyncha, new species (Chapter 19), and Microgale sp. A (Chap- ter 19; Jenkins et al., unpubl.), were found only in montane forest. Microgale cowani may also be more common in montane than lowland regions; in the RNI d'Andringitra it was collected at var- ious localities from 1200 to 1800 m, but it was found only rarely in lowland forest at 810 m. Those species that were captured at all alti- tudes, namely M. taiva and M. parvula, deviate from the general pattern of the highest abundance at 1210 m; for both, the greatest number of indi- viduals were collected at 1625 m, slightly higher than the number collected at 1210 m, but mark- edly greater than either of the lowland localities. M. cowani follows the general pattern, with most individuals being trapped at 1210 m, next at 1625 m, fewest at 810 m, and none at 720 m. tive forms. Microgale longicaudata, with its very long tail, which has been hypothesized as being prehensile (Thomas, 1918; MacPhee, 1987), elon- gated hind foot with lengthened fifth digit, and elongated cheiridia, was assumed by Eisenberg and Gould (1970) possibly to climb and ricochet among branches. When handled by S.M.G. and C.J.R., one M. longicaudata captured in a pitfall trap during the 1993 survey wrapped the tip of its tail around a finger and was able to hang sus- pended for about 20 seconds; thus, at least the tip of the tail is prehensile. Although no M. longi- caudata was obtained in Sherman traps placed off the ground, it is clear that this species is at least scansorial, if not occasionally arboreal. Microgale sp. A shows similar but less extreme scansorial adaptations, with a slightly elongated tail, hind feet, fifth hind digit, and cheiridia. Mi- crogale longicaudata and M. sp. A have not been found in the same elevational zones, but they may well occur in direct sympatry; they differ in body size as well as degree of morphological adaptation and may be assumed to occupy slightly different ecological niches. Microgale gracilis and M. gym- norhyncha both show morphological features that suggest a semi-fossorial habit, such as the broad- ened forefeet with enlarged claws, and the re- duced size of the eyes and ears; both are found in direct sympatry in the reserve. The remaining six species have a fairly gener- alized body form, although M. parvula by virtue of its very small size, M. dobsoni because of its large size, and M. soricoides by its unusual den- tition may be assumed to occupy different eco- logical niches. Indeed, the capture of one speci- men of M. soricoides 2 m above the ground implies some climbing ability. The real problems remain with M. cowani, M. taiva, and M. melanorrhachis, all of which show only slight morphological vari- ation, no obvious adaptive specializations, and yet are clearly distinct species (Chapter 19). Micro- gale cowani and M. taiva, and M. taiva and M. melanorrhachis, occur in direct sympatry in the RNI d'Andringitra. Multiple Captures in a Single Pitfall Bucket Ecological Separation Some of the Microgale species occurring in sympatry along the eastern slopes of the RNI d'Andringitra may be divided into distinct adap- Multiple captures include different animals taken in a pitfall bucket during the time that a line was in place and multiple animals found in a pit- fall bucket on a single occasion. In the latter case, referred to below as "simultaneous" captures, the GOODMAN ET AL.: INSECTIVORE ECOLOGY 225 results might be confounded because larger Mi- crogale species will readily eat smaller species when confined to the same bucket. In both cases, these multiple captures reflect animals moving across the same swath of ground 20 m or less in width. We interpret such movement as direct syn- topy in the utilization of areas of the forest. Fur- thermore, this information allows some inference to be made into the social organization of the var- ious species. Multiple Captures During the Tenure of Line Placement — At 720 m there were only two cases of multiple captures, both in line 3 in a val- ley bottom. Two M. taiva were taken in the same bucket, as were a M. taiva and two M. longicau- data in another bucket. Multiple captures at 810 m included two M. taiva; a M. cowani and M. taiva; and a M. parvula and M. cowani in three different buckets. At 1210 and 1625 m the num- ber of multiple captures in a single bucket in- creased substantially. In numerous cases single buckets in a pitfall line collected three to five small mammals during the tenure of the line place- ment. Exceptional cases included five M. longicau- data taken in a single bucket over 3 days; two M. cowani, two M. taiva, and a M. longicaudata in 2 days; and a M. cowani, two M. taiva, and an un- described genus and species of rodent (see Chapter 21) in 2 days. This demonstrates that in some cases species that are morphologically similar (e.g., M. cowani and M. taiva) occur in microsympatry. Oth- ers, such as M. gracilis, M. gymnorhyncha, and M. soricoides, however, were never taken in any pair- wise combination in the same pitfall bucket. Fur- thermore, M. melanorrhachis was not captured in the same pitfall bucket with M. cowani or M. taiva. In several of these cases, the numbers of animals captured for a given species were few, and these patterns may simply be chance. Simultaneous Multiple Captures — A total of 902 bucket-days accrued during the survey. Of these days there were 823 during which no Mi- crogale was captured, 71 bucket-days when single individuals were captured, 6 bucket-days when two individuals were captured, and 2 bucket-days during which three individuals were captured. These observed frequencies do not deviate from the expected frequencies based on a Poisson dis- tribution. The only case at 720 m of two animals in the same pitfall bucket was a pair of M. longicaudata (adult female and juvenile female). No simulta- neous multiple capture was recorded at 810 m. At 1210 m, three M. taiva were found in a bucket the first morning it was installed; these included two juvenile males and a juvenile female. Also at this elevation there was one case of a M. cowani and a M. gymnorhyncha in the same bucket. At 1625 m, a single bucket contained two M. longi- caudata on two occasions: on December 10, 1993, an adult female and juvenile were caught, and on December 12, 1993, two adult males were caught (On December 11, 1993, this bucket yield- ed a juvenile M. longicaudata). The next bucket in the same line yielded an adult female M. cow- ani and an adult male M. taiva on December 9, 1993, and on December 10, 1993, a juvenile male M. cowani, an adult male M. taiva, and an adult male M. longicaudata were captured. Thus, over the course of four consecutive days, 10 individual Microgale of three species were captured in a sec- tion of a pitfall line no more than 30 m long. The last example was two juvenile male M. taiva and an undescribed genus and species of rodent cap- tured on the same night in a single bucket. Little is known about the breeding systems or social organization of Microgale species. Whether the three juvenile M. taiva captured on the same night were moving together or were attracted to the bucket by movements or calls of captured individ- uals is unknown. Female M. taiva have also been reported with two embryos and up to six mammae, and thus it is possible that the young M. taiva were siblings. In the eight cases of "simultaneous" mul- tiple captures in the same bucket, there was not a single example of a conspecific adult male and fe- male, and there was only one case of conspecific adult males being trapped together. Relationship Between Macrofauna Soil Invertebrates and the Elevational Distribution of Microgale Five soil plots, each 2 (length) X 2 (width) X 0.1 (depth) m and separated by 20 m, were sam- pled during daylight hours along each pitfall line to document the variation in density and species richness of macrofauna soil invertebrates (Chapter 14). Because the vast majority of the tenrecs in the RNI d'Andringitra are terrestrial and presum- ably feed extensively on soil invertebrates, the re- sults of this analysis have potential implications for understanding the distribution of insectivores along the eastern slopes of the massif. Both Mi- crogale gracilis and M. gymnorhyncha have well- developed claws, and these species, as well as others, presumably dig into the ground. 226 FIELDIANA: ZOOLOGY Table 20-4. Variation in macrofauna soil invertebrates and Microgale spp. between different pitfall lines. 720 m 810 m 1210 m 1625 m Pitfall line: 1 2 3 4 5 6 7 8 9 10 11 12 Placement: R S V R S V R V S V R S Invertebrate taxonomic diversity Average per 8.4 4.4 5.6 10.6 8.6 2.6 8.8 7.2 8.0 7.0 6.8 8.8 line (5-12) (0-10) (3-11) (7-14) (4-12) (0-7) (4-13) (2-15) (4-12) (3-10) (4-10) (5-16) 2.97 4.39 3.21 2.51 2.97 2.70 3.70 5.06 3.31 2.92 2.39 4.55 Average per 6.1 7.3 8.7 7.5 line per zone (0-12) 3.74 (1-14) 4.33 (2-15) 3.98 (3-16) 3.29 Total per zone 46 68 76 46 Invertebrate density Average individuals 13.6 5.0 8.6 14.2 9.4 4.8 11.6 16.8 16.0 16.8 10.4 24.8 per line (6-22) (0-12) (3-16) (8-24) (4-13) (0-14) (5-20) (2-35) (7-34) (3-51) (5-14) (5-43) 6.35 5.09 6.77 6.18 3.51 5.45 6.27 13.08 11.90 19.63 4.09 17.53 Average per line per 9.1 9.5 14.9 16.7 zone (0-22) 6.73 (0-24) 6.22 (2-35) 10.32 (3-51) 14.87 Microgale spp. % captured per line 3.0 4.5 10.6 3.9 5.2 5.2 10.4 16.9 18.2 21.6 5.7 10.4 % captured per ele- vational zone 6.0 4.8 15.2 12.6 Species per line 2 2 2 1 2 3 5 4 6 5 1 4 Species per eleva- tional zone 4 4 8 6 R = ridge, S = slope, V = valley. Averages are presented as mean, range (minimum-maximum), and standard deviation. In Table 20-4 we present a summary of the tax- onomic diversity and number of individual mac- rofauna soil invertebrates along each line and at each elevational zone. The taxonomic diversity of invertebrates sampled within each plot per pitfall line showed considerable variation in the mini- mum and maximum values and relatively high standard deviations. However, average diversity between elevational zones varied from a low of 6.1 morphospecies on average per plot at 720 m to a high of 8.7 morphospecies per plot at 1210 m. The average total number of individual inver- tebrates found per plot in each elevational zone (N = 15) increased with altitude (r2 = 0.95, P = 0.03), from an average total of 9.1 (range 5.0- 13.6) at 720 m, 9.5 (range 4.8-14.2) at 810 m, 14.9 (range 1 1.6-16.8) at 1210 m, to 16.7 (range 10.4-24.8) at 1625 m. No information is available on invertebrate biomass in the plots. The general trends of invertebrate taxonomic diversity and Microgale species density and abun- dance (as measured by pitfall trap success) par- allel one another relative to elevation, the peak for both groups being 1210 m. The density of in- vertebrates, however, does not peak at 1210 m, but rather shows a linear increase as a function of elevation. Because we know little about the spe- cific food habits of these various species of Mi- crogale, it is impossible to interpret on a fine scale what types of invertebrates or specific taxa they might consume, although it appears that the den- sity and distribution of shrew tenrecs on the slopes of the Andringitra Massif is related more directly to macrofauna soil invertebrate diversity than to density. One species, M. melanorrhachis, is only found at low elevations, and its elevational distribution does not parallel this general pattern. Microhabitat Segregation Between Species and Its Relationship to Macrofauna Soil Invertebrates At each elevational zone, pitfall traps were set in three distinct positions: valley bottom, slope, and ridge crest. The trapping results for each type GOODMAN ET AL.: INSECT! VORE ECOLOGY 227 Table 20-5. Variation in macrofauna soil invertebrates and Microgale spp. in relationship to microhabitat and elevation. Ridge Slope Valley Parameter measured 720 810 1210 1625 720 810 1210 1625 720 810 1210 1625 in m in in m in m m in in in m Invertebrate taxonomic diversity Average number of species per line 8.4 10.6 8.8 6.8 4.4 8.6 8.0 8.8 5.6 2.6 7.2 7.0 Average total number of species 7.5 5.5 per microhabitat 8.7 Invertebrate density Average number per line 13.6 14.2 11 Average number per line per microhabitat 12.5 Microgale spp. % capture per line 3.0 3.9 10 Average % capture per micro- habitat 5.8 Species per line 2 1 5 Species per habitat 8 Total individuals per habitat 18 13.6 14.2 11.6 10.4 5.0 9.4 16.0 24.8 8.6 4.8 16.8 16.? 13.8 11.8 3.0 3.9 10.4 5.7 4.5 5.2 18.2 10.4 10.6 5.2 16.9 21.6 9.6 13.6 1 2264 2345 6 7 27 43 of microhabitat are given in Tables 20-4 and 20-5. At both lowland forest localities (720 and 810 m), few specimens of Insectivora were captured on slopes and on ridges, whereas a considerable por- tion of the individuals in the 720 m zone were collected in valley bottoms; at 810 m, capture rates in valley bottoms were similar to those of slope and ridge. At 720 m, two species of Micro- gale were obtained in each of these microhabitats, whereas at 810 m, the number of species captured varied, from three species in the valley to two on the slope and one on the ridge (Table 20-3). The pattern was slightly different in montane forest at 1210 m, with nearly equal numbers of individual Microgale captured in the valley and slope lines (13 and 12, respectively; Table 20-3) and only eight individuals taken in the ridge line. At this altitude, six species of Microgale were obtained in the slope line, five in the ridge line, and four in the valley line. The capture success rate of Micro- gale in the 1625 m upper montane/sclerophyllous zone showed considerable variation between mi- crohabitats. The capture rate in the valley was nearly four times greater than that in the ridge line and two times greater than that in the slope line. Taxonomic diversity at this elevation in micro- habitats paralleled overall capture success. There is no convincing evidence from this study to suggest that any species occurs prefer- entially in one of the three habitats. Microgale soricoides was found only on a slope and M. sp. A only on a ridge, whereas M. gymnorhyncha and M. dobsoni both occurred on slope and ridge for- est. The sample size for all four species, however, is too small for conclusions to be drawn about habitat preference. There is slight evidence that M. cowani occurs preferentially in slope forest, as may M. taiva, at least at high elevations. The combined data for the three habitat types are given in Table 20-5. The highest capture rate was in valley bottoms (N = 43), followed by slopes (N = 27) and then ridges (N = 18), where- as the highest species richness was on the ridge (eight species), then the valley (seven species) and then slope (six species). When the macrofauna soil invertebrate samples were segregated by microhabitat, the ridge lines had the highest taxonomic diversity per plot (8.7 species, range 6.8-10.6), followed by slope (7.5 species, range 4.4-8.8), and valley (5.7 species, range 2.6-7.2). The average number of individual invertebrates recovered per plot was remarkably constant between microhabitats, although the val- ley and slope sites showed considerable differ- 228 FIELDIANA: ZOOLOGY ences in the range: ridge, 12.5 individuals (range 10.4-14.2); slope, 13.8 individuals (range 5.0- 24.8); and valley, 11.8 individuals (range 8.6- 16.8). In general there seems to be little correla- tion in the average values per habitat type be- tween the taxonomic diversity and density of macrofauna soil invertebrates and density and specificity of Microgale. When examined on a fin- er scale, there appears to be some relationship be- tween these factors. For example, in slope sites the highest invertebrate density was at 1210 (16.0 individuals/plot) and 1625 m (24.8 individu- als/plot). These two lines also had the highest per- centage capture rate (18.2% and 10.4%, respec- tively) and species diversity (six and four, respec- tively) of the four lines in this habitat. An almost parallel situation occurred in the valley lines at 1210 and 1625 m. However, in the ridge sets this pattern did not exist; at 810 m, the elevation with the highest taxonomic diversity and density of soil invertebrates, only one species of Microgale was collected, and the overall trap success was nearly the lowest for this microhabitat. Thus, in short, the measurements of soil invertebrates along the pitfall lines in three different types of microhab- itats were not closely correlated with density or Microgale specificity. Differences Between Disturbed and Undisturbed Lowland Forest The 720 m zone was classified as disturbed for- est and the 810 m zone as relatively pristine for- est. A comparison of the Microgale species rich- ness (five vs. four species, respectively) and den- sity (seven animals in 198 pitfall bucket-days [= 3.5%] vs. 12 animals in 231 pitfall bucket-days [= 5.2%], respectively) indicates that there is no clear difference in the Microgale species com- munities between these localities. In the RS d'Analamazaotra, Stephenson (1993) found that in some cases differences in the level of human activity between sites at the same ele- vation and in approximately the same contiguous forest were related to species diversity. The great- est species diversity of Microgale was in the site with no or at least little human disturbance. How- ever, two of the four species captured at this site were exceptionally rare, and their absence from the other sites may be a question of chance sam- pling rather than actual differences in species di- versity. Furthermore, several forest-dwelling Mi- crogale do not seem particularly sensitive to gen- eral patterns of habitat disturbance (MacPhee, 1987; Goodman et al., in press). Elevational Variation of Species at Other Sites Microgale melanorrhachis may be a lowland forest species in the RNI d'Andringitra, but it seems to be more wide ranging in elevation else- where. Higher elevational sites include the type locality of Andasibe at 3000 ft (= 920 m) (Mor- rison-Scott, 1948) and Didy at 1000 m (Eisenberg & Gould, 1970). Positively identified specimens have also been collected from valley bottom for- est at Mantady (1100-1200 m) (Rax worthy, un- publ.) and Montagne d'Ambre (1100-1150 m) (Raxworthy & Nussbaum, 1994). At the latter lo- cality, M. melanorrhachis was common in valley bottoms, on ridges, and on slopes between 1125 and 1 250 m, but in a comparable number of buck- et trap-days it was not recorded along the same slope between 650 and 670 m. Microgale melan- orrhachis has also been collected at 5000 ft (= 1525 m) at Pic dTvohibe (Morrison-Scott, 1948), just south of the RNI d'Andringitra. Limited in- formation in the literature suggests that M. grac- ilis may be more common in montane regions; it has been collected at Ambohimitombo (1200 m), Ankeramadinika (1400 m), and Anjavidilava, RNI d'Andringitra (1800-2100 m). The supposed low elevational record of this species at Fanovana, Be- forona (500-1000 m) (MacPhee, 1987), actually refers to a specimen of M. gymnorhyncha, which in the RNI d'Andringitra appears to be contrast- ingly confined to montane forest. Microgale cow- ani sensu stricto may also be more common in montane regions. The type locality is Ankafina (summit at 1570 m, currently most forest confined to 1200-1500 m), and at RNI d'Andringitra it was collected at various localities from 1200 to 1800 m, but only rarely in lowland forest at 810 m. Acknowledgments We are grateful to the officials of DEF and AN- GAP for permission to conduct this survey. Crit- ical comments were provided on an earlier ver- sion of this manuscript by B. D. Patterson and an anonymous reviewer. The participation of C.J.R. in this project was partially supported by NSF grant DEB 90-24505. GOODMAN ET AL.: INSECTIVORE ECOLOGY 229 Literature Cited Coquerel, C. 1848. Note sur une espece nouvelle de musaraigne trouv^e a Madagascar. Annales des Sci- ences Naturelles, Zoologie, 9: 193-198. Eisenberg, J. E, and E. Gould. 1970. The tenrecs: A study in mammalian behavior and evolution. Smith- sonian Contributions to Zoology, 27: 1-138. Glaw, E, and M. Vences. 1994. A Fieldguide to the Amphibians and Reptiles of Madagascar, 2nd edition. Zoologisches Forschungsinstitut und Museum Alex- ander Koenig, Bonn, 480 pp. Goodman, S. M., and J. Ganzhorn. 1994. Les petits mammiferes, pp. 58-63. In Goodman, S. M., and O. Langrand, eds., Inventaire biologique foret de Zom- bitse. Recherches pour le Developpement, Serie Sci- ences biologiques, No. Special, Centre dTnformation et de Documentation Scientifique et Technique, An- tananarivo, 106 pp. Goodman, S. M., R D. Jenkins, and O. Langrand. In press. Exceptional records of Microgale species (In- sectivora: Tenrecidae) in vertebrate food remains. Bonner Zoologische Beitrage. Grandidier, A. 1870. Description de quelques animaux nouveaux, decouverts a Madagascar, en novembre 1869. Revue et Magasin de Zoologie, 22: 49-54. Grandidier, G. 1934. Deux nouveaux mammiferes in- sectivores de Madagascar. Bulletin du Museum Na- tional d'Histoire Naturelle, ser. 2, 6: 474-477. Jenkins, P. D. 1992. Description of a new species of Microgale (Insectivora: Tenrecidae) from eastern Madagascar. Bulletin of the British Museum Natural History (Zoology), 58: 53-59. . 1993. A new species of Microgale (Insectiv- ora: Tenrecidae) from Eastern Madagascar with an un- usual dentition. American Museum Novitates, 3067: 1-11. Linnaeus, C. 1766. Systema naturae per regna tria na- turae secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis. 12th edi- tion. Holmiae. MacPhee, R. D. E. 1987. The shrew tenrecs of Mada- gascar: Systematic revision and Holocene distribution of Microgale (Tenrecidae, Insectivora). American Mu- seum Novitates, 2889: 1-45. Major, C. I. Forsyth. 1896a. Diagnoses of new mam- mals from Madagascar. Annals and Magazine of Nat- ural History, (6) 18: 318-325. . 1896b. Descriptions of four additional new mammals from Madagascar. Annals and Magazine of Natural History, (6) 18: 461-463. Milne Edwards, A., and A. Grandidier. 1872. De- scription d'un nouveau mammifere insectivore de Madagascar (Geogale aurita). Annales des Sciences Naturelles, Zoologie, 15, Article 19: 1-5. . 1882. Description d'une nouvelle espece d' insectivore de Madagascar. Le Naturaliste, 2: 55. Morrison-Scott, T. C. S. 1948. The insectivorous gen- era Microgale and Nesogale (Madagascar). Proceed- ings of the Zoological Society of London, 118: 817— 822. Nicoll, M. E. 1982. Reproductive ecology of Tenrec ecaudatus (Insectivora: Tenrecidae) in the Seychelles. Ph.D. thesis, University of Aberdeen. -, and O. Langrand. 1989. Madagascar: Revue de la conservation et des aires protegees. World Wide Fund for Nature, Gland, Switzerland, XVII+374 pp. Nicoll, M. E., and G. B. Rathbun. 1990. African In- sectivora and Elephant- shrews: an action plan for their conservation. IUCN/SSC Insectivore, Tree-shrew and Elephant-shrew Specialist Group, Gland, Switzerland, iv+53 pp. Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, and P. Griveaud. 1971. RCP 225. Etudes des eco- systemes montagnards dans le region malgache. I. Le massif de l'Andringitra. 1970-1971. Geomorphologie, climatologie et groupements vegetaux. Bulletin de la Societe d'Ecologie, 11(2-3): 198-226. Raxworthy, C. J., and R. A. Nussbaum. 1994. A rain- forest survey of amphibians, reptiles and small mam- mals at Montagne d'Ambre, Madagascar. Biological Conservation, 69: 65-73. Raxworthy, C. J., J.-B. Ramanamanjato, and A. Ra- selimanana. 1994. Les reptiles et les amphibians, pp. 41-57. In Goodman, S. M., and O. Langrand, eds., Inventaire biologique foret de Zombitse. Recherches pour le Developpement, Serie Sciences biologiques, No. Special, Centre dTnformation et de Documenta- tion Scientifique et Technique, Antananarivo, 106 pp. Schreber, J. C. D. von. 1777. Die Saiigethiere in Ab- bildungen nach der Natur mit Beschreibungen. Vol- ume 3. Leipzig. Stephenson, P. J. 1991. Reproductive energetics of the Tenrecidae (Mammalia: Insectivora). Ph.D. thesis, University of Aberdeen. . 1993. The small mammal fauna of Reserve Speciale d'Analamazaotra, Madagascar: The effects of human disturbance on endemic species diversity. Bio- diversity and Conservation, 2: 603-615. -. 1994. Seasonality effects on small mammal trap success in Madagascar. Journal of Tropical Ecol- ogy, 10: 439-444. -, and P. A. Racey. 1993. Reproductive energet- ics of the Tenrecidae (Mammalia: Insectivora). II. The shrew-tenrecs, Microgale spp. Physiological Zoology, 66: 664-685. Stephenson, P. J., H. Randriamahazo, N. Rakotoari- son, and P. A. Racey. 1994. Conservation of mam- malian species diversity in Ambohitantely Special Re- serve, Madagascar. Biological Conservation, 69: 213— 218. Thomas, [M. R.] Oldfield. 1882. Description of a new genus and two new species of Insectivora from Mad- agascar. Journal of the Linnean Society (Zoology), 16: 319-322. . 1884. Description of a new species of Micro- gale. Annals and Magazine of Natural History, (5) 14: 337-338. . 1918. On the arrangement of the small Ten- recidae hitherto referred to Oryzorictes and Microga- le. Annals and Magazine of Natural History, (9) 1: 302-307. 230 FIELDIANA: ZOOLOGY Chapter 21 Systematic Studies of Madagascar's Endemic Rodents (Muroidea: Nesomyinae): A New Genus and Species from the Central Highlands Michael D. Carleton and Steven M. Goodman Abstract We report a new genus and species of nesomyine rodent from Madagascar's Central High- lands, type locality Antananarivo Province, Manjakatompo, at about 1800 m elevation on the Ankaratra Massif. The form also occurs in upper montane forest within the Reserve Naturelle Integrate d'Andringitra at 1625 m. The new genus is specifically contrasted to species of Macrotarsomys, a taxon that occurs in western Madagascar and is hypothesized to be its closest relative among the seven previously known genera of Nesomyinae. The many generalized features of the new genus provide a basis for renewed examination of the Miocene cricetodontid Protarsomys (Kenya, Rusinga Faunal Assemblage); we discount its proposed synonymy with Macrotarsomys and question its progenitive significance in regard to the origin of Nesomyinae. The new genus has disjunct populations on the Ankaratra and Andringitra massifs, a distribution that resembles those of numerous other Malagasy organisms restricted to upper montane- sclerophyllous vegetation. This geographical pattern conforms to Humbert's proposed High Mountain Domain. Resume Nous rapportons un nouveau genre et une nouvelle espece de rongeur nesomyine du haut plateau central de Madagascar, locality du specimen type Province d' Antananarivo, Manjaka- tompo, vers 1800 m dans le massif de 1' Ankaratra. Dans la Reserve Naturelle Integrate d'Andringitra, cette forme apparatt aussi dans la foret de montagne a une altitude sup6rieure a 1625 m. Le nouveau genre se d6marque sp6cifiquement des especes de Macrotarsomys, un taxon qui est distribu6 a l'ouest de Madagascar et qui est consider comme le plus proche parmi les sept genres de Nesomyinae connus jusqu'a maintenant. Les nombreuses caracteYis- tiques g6n£rales de ce nouveau genre fournit une base d'examen pour r6examiner Protarsomys (Kenya, Assemblage faunistique de Rusinga) cricetodontid du Miocene; nous ne tenons pas compte de la synonymie proposed avec Macrotarsomys et mettons en doute sa valeur compte tenu de l'origine des Nesomyinae. Le nouveau genre pr£sente des populations isotees dans les massifs de 1'Ankaratra et de 1'Andringitra, une distribution qui ressemble a celle d'autres or- ganismes a la distribution limited a la v6g6tation scterophylle des hautes montagnes. Cette distribution g6ographique est conforme a celle proposee par Humbert pour le Domaine des Hautes Montagnes. CARLETON & GOODMAN: ENDEMIC RODENTS 231 Introduction The Mission Zoologique Franco-Anglo-Amer- icaine (MZFAA) was a cooperative biological sur- vey of Madagascar conducted from April 1929 to May 1931 (Rand, 1932, 1936). The expedition concentrated its collecting activities on the is- land's avifauna and, to a lesser extent, on larger mammals such as lemurs (Archbold, 1932; Rand, 1935), and all specimens were approximately evenly divided and deposited among the three participating museums — the Museum National d'Histoire Naturelle in Paris (MNHN), the British Museum of Natural History in London (BM[NH]) (now called The Natural History Museum), and the American Museum of Natural History in New York City (AMNH). Because of their selective field efforts, the MZFAA teams obtained relative- ly few examples of Madagascar's endemic ro- dents, the Nesomyinae (about 70, compared to some 470 lemurs — see Carleton & Schmidt, 1990), but, for the era, the specimens collected represented important additions to our knowledge of the indigenous small mammals. Regrettably, systematists never attempted a critical taxonomic synthesis of the mammalian collections made by the MZFAA, such as was undertaken for the birds (Delacour, 1932; Rand, 1936), and long over- looked was one nondescript, mouse-like rodent obtained by Austin L. Rand near Manjakatompo, a forestry station on the upper slopes of the Anka- ratra Massif in the Central Highlands (synony- mous with the Central High Plateau). Recognition of the significance of Rand's cap- ture was further hindered by an innocent process- ing error, which resulted in receipt of the skin at the AMNH and accession of the skull into the MNHN. For decades, the separated parts of the enigmatic specimen went unnoticed and unappre- ciated until the early 1970s, when Karl F. Koop- man, then Associate Curator in the Department of Mammalogy, AMNH, began an attempt to resolve his inability to classify the lone museum skin ac- cording to available taxonomic standards on ne- somyine rodents. Subsequent discussions about muroid diagnostic characters with his colleague Guy G. Musser prompted inquiries and led to the eventual rediscovery of the corresponding crani- um and mandibles in the MNHN. Through the gracious and understanding cooperation of Dr. Francis Petter, Laboratoire des Mammiferes et Oiseaux, MNHN, the errant skull was reunited with its skin in the late 1970s, and the now com- plete museum specimen (AMNH 100727) only confirmed Koopman's suspicions of its morpho- logical uniqueness and newness to the rodent fau- na of Madagascar. Other research commitments and priorities intervened to divert the attention of Koopman and Musser from jointly describing the new taxon. The impetus to report Rand's Manjakatompo specimen at this time and in the present context issues from renewed interest in the native rodents of Madagascar and from the resurgence of field- work on the island. Ongoing revisionary work by Carleton (1994) and Carleton and Schmidt (1990), and the 1993 rediscovery by Goodman of Rand's unremarkable mouse on the Andringitra Massif, over 300 km to the south of Ankaratra, make for- mal identification of this form both propitious and necessary. The fresh material from the Reserve Naturelle Integrate (RNI) d' Andringitra encour- ages fuller elaboration of the form's defining traits, which further attest to the distinctiveness of the taxon and convince us that it represents a ge- nus and species new to the fauna of Madagascar. Materials and Methods The material described herein and the compar- ative series of Macrotarsomys that are referenced are housed in the AMNH, New York City; the Field Museum of Natural History (FMNH), Chi- cago; the MNHN, Paris; the National Museum of Kenya (KNM), Nairobi; and the National Muse- um of Natural History (USNM), Smithsonian In- stitution, Washington, D.C. Nesomyine holdings in other museums have been studied by M.D.C., and one or both authors have examined the ho- lotypes of all described forms of Nesomyinae, except for Peters's (1870) Nesomys rufus. The fol- lowing samples, consisting of conventional round skins with skulls, as well as complete skeletons and fluid-preserved whole carcasses, were con- sulted for the various tabular summaries and an- atomical comparisons. Macrotarsomys bastardi bastardi — Fianaran- tsoa Province: 35 mi (= 56 km) N Ihosy, 3000 ft (= 915 m) (AMNH 119710); 5 km E Route Na- tionale 7, along road to Ivohibe, 750 m (USNM 328793-328807). Macrotarsomys bastardi occidentalis — Maha- janga Province: 40 km S Marovoay, Ampijoroa (USNM 341817-24). Toliara Province: 40 km N Morondava, 7 km from sea, Beroboka (AMNH 232 FIELDIANA: ZOOLOGY 119708-9); Petriky Forest, 5-7 km SE Manam- baro, 20 m (USNM 578715-9 and 578822-3). Macrotarsomys ingens — Mahajanga Province: Ankarafantsika Reserve, Ampijoroa (MNHN 1958.636 and 1961.214-5; USNM 328831 and 576753). Protarsomys macinnesi — Lower Miocene, Ru- singa Faunal Assemblage, Kenya (cast of holo- type, KNM 2350, cranium with associated man- dibles); Lower Miocene, Legetet, Kenya (cast of a referred lower jaw with ml -3, KNM 2188). Six measurements, in millimeters (mm) or grams (g), were taken by S.M.G. for each FMNH specimen in the flesh. Their abbreviations and def- initions are given below. TOTL, total length of body and tail: from the tip of the nose to the end of the caudal ver- tebra (not including terminal hair tuft). HBL, head and body length: from the tip of the nose to the distalmost point of the body (at base of tail). TL, tail length: from the base of the tail (held at right angles to the body) to the end of the last caudal vertebra (not including terminal hair tuft). HFL, hind food length: from the heel to the tip of the longest toe (not including claw). EL, ear length: from the basal notch to the dis- talmost rim of the pinna. WT, weight in grams: measured with Pesola spring scales to ± 0.5 g. Sixteen cranial and two dental dimensions were measured by M.D.C. to the nearest 0. 1 mm using handheld digital calipers accurate to 0.03 mm. These measurements, and their abbreviations, fol- low the anatomical landmarks defined and illus- trated in Carleton (1994): BBC, breadth of the braincase; BIF, breadth of incisive foramina; BM 1 s, breadth of the bony palate across the first upper molars; BOC, breadth across the occipital condyles; BR, breadth of rostrum; BZP, breadth of the zygomatic plate; DAB, depth of the audi- tory bulla; IOB, interorbital breadth; LBP, length of bony palate; LD, length of diastema; LIF, length of the incisive foramina; LM1-3, coronal length of maxillary toothrow; LR, length of ros- trum; ONL, occipitonasal length; PPB, posterior breadth of the bony palate; PPL, postpalatal length; WM1, width of the first upper molar; and ZB, zygomatic breadth. Analytical routines for the standard descriptive statistics were carried out using the Systat (ver- sion 5.05, 1994) computer program. Morpholog- ical terms generally follow Carleton (1980) and Voss (1988); names of dental structures follow Reig (1977), as illustrated by Carleton and Musser (1989). Field methods for the three FMNH spec- imens originating from the RNI d'Andringitra are described in Chapter 22. Monticolomys, new genus Type Species — Monticolomys koopmani, de- scribed below as new. Diagnosis — A form of the murid rodent sub- family Nesomyinae (sensu Carleton & Musser, 1984; Musser & Carleton, 1993) characterized by: small size (TOTL = 205-240 mm; HBL = 89- 101 mm) (in other nesomyines except Eliurus mi- nor and Macrotarsomys bastardi, HBL ^ 130 mm); short (15-18 mm), densely furred, and rounded pinnae (pinnae long, ^ 21 mm, and sparsely haired in Macrotarsomys); tail apprecia- bly longer than head and body together, about 135-140% of HBL (TL < HBL in Brachyuromys, Gymnuromys, Hypogeomys, and Nesomys), with- out noticeable elongation of caudal hairs over dis- tal half (distal tip penicillate in Macrotarsomys, a brushy tuft in Eliurus); hind foot relatively long and broad, outer digits I and V comparatively long (hind foot narrow, outer digits relatively short in Macrotarsomys); plantar surface with six fleshy pads, thenar and hypothenar as large as in- terdigital pads (thenar and hypothenar diminutive in Macrotarsomys); ungual tuft well developed, surpassing tip of claw (tuft hairs shorter than claw in Macrotarsomys). Cranium small (ONL = 26-28 mm, LM1-3 = 3.5-3.7 mm) and delicately built with slender, parallel-sided zygomatic arches and narrow, hour- glass-shaped interorbit; vascular groove on squa- mosal-alisphenoid, stapedial and sphenofrontal foramina present (groove and foramina lacking in Brachytarsomys, Brachyuromys, Eliurus, Gym- nuromys, and Hypogeomys); alisphenoid strut ab- sent (strut present in Brachyuromys, most Eliurus, Hypogeomys, and Nesomys); ectotympanic bullae moderately inflated, narrow ventromedial wedge of periotic exposed (bullae smaller, wide expanse of periotic visible in Brachytarsomys, Eliurus, and Gymnuromys; bullae notably larger, periotic most- ly obscured in Brachyuromys, Hypogeomys, Mac- rotarsomys, and Nesomys); tegmen tympani re- duced, not contacting squamosal (articulation with squamosal in Brachytarsomys, Eliurus, and Gym- nuromys). CARLETON & GOODMAN: ENDEMIC RODENTS 233 234 FIELDIANA: ZOOLOGY Upper toothrows divergent anteriorly (conver- gent anteriorly in Brachyuromys, more or less par- allel in Brachytarsomys, Eliurus, Gymnuromys, Hypogeomys, and Nesomys); molars cuspidate and brachyodont (cheek teeth moderately to extremely hypsodont, cuspation lost or indistinct, and occlu- sal surface planar in Brachyuromys, Brachytar- somys, Eliurus, Gymnuromys, and Hypogeomys); mesolophs(ids) absent, posterolophs rudimentary, anteroconids imperfectly developed (meso- lophs[ids] present in Gymnuromys and Nesomys); lower third molar reduced, notably smaller than second molar (size of third molar approximately equal to second in Brachyuromys, Eliurus, Hy- pogeomys, and Nesomys; conspicuously larger than second molar in Gymnuromys); upper molars with three roots and lower molars with two (upper molars four-rooted in Hypogeomys). Morphological Description — As for the sin- gle known species, described below. Monticolomys koopmani, new species (Figs. 21-1 to 21-5; Tables 21-1 and 21-2) Holotype— AMNH 100727: skin and skull of young adult male; original no. 62, collected May 24, 1929, by Austin L. Rand as part of the MZFAA (see Rand, 1932, 1936). Standard measurements (in mm) from the spec- imen tag attached to the skin of the type include: total length, 205; tail length, 116; hind foot length, without claw, 24 (dry hind foot length, as measured by M.D.C. with claw, is 25.5); and ear length, 15. The condition of the skin is good. The skull is damaged, with its right bullae detached and pres- ent in the vial, both zygomatic arches incomplete, and the orbitosphenoid walls on both sides bro- ken; the mandibles are separated, and their an- gular and coronoid processes are incomplete. Type Locality — Madagascar, Antananarivo Province, Manjakatompo, ca. 19°20'S 47°26'E. Rand identified the locality only as "Monjak- atompo" (spelling in the United States Board on Geographic Names, 1989 gazetteer, as Manjaka- tompo) on his original tag and later (1936) sup- plied the collection elevation as 1800 m. This height probably refers to an area of fragmented natural forest intermixed with secondary grass- land in the vicinity of the forestry station of Man- jakatompo. Elevational records mentioned on oth- er specimen tags or in Richard Archbold's journal (AMNH Mammalogy Department) indicate that the team worked the slopes of Ankaratra at least from 1650 m to 1950 m (Rand's own field notes for the MZFAA were apparently lost or de- stroyed). The little natural forest remaining on the slopes of Ankaratra above Manjakatompo is es- timated to be about 650 ha (Nicoll & Langrand, 1989). Referred Specimens — FMNH 151727 (male, skull with postcranial skeleton), 151899 (female, skull with whole carcass in fluid), and 151900 (male, skull with whole carcass in fluid); all from Fianarantsoa Province, 38 km S Ambalavao, RNI d'Andringitra, ridge E of Volotsangana River, 1625 m; 22° ITS 46°58'E; collected by S.M.G., December 12-14, 1993. Distribution — At present known only from two localities in the Central Highlands, about 1800 m on the Ankaratra Massif and 1625 m on the Andringitra Massif. Diagnosis — As for the genus, above. Morphological Description — Fur soft, rela- tively thick and fine (based on the holotype, AMNH 100727). Cover hairs of dorsum usually tricolored — basal two-thirds plumbeous gray, middle band a deep ochraceous, and short tip dark brown to black; guard hairs entirely black, longer, and more heavily concentrated toward the middorsum and on the rump; general appearance of upperparts a somber or muted dark brown. Fur over inguinum, abdomen, and chest with plum- beous bases and pale buffy to whitish tips, most- ly white to bases on chin and throat; combined effect a dark gray ventrum not sharply demar- cated from dorsum. Mystacial vibrissae medium in length, the longest whiskers reaching the top of the pinnae when appressed to the skin. Pinnae short and rounded (Fig. 21-1; Table 21-1), dense- ly clothed internally and externally with slate hairs. Tail very long relative to head-and-body length (TL about 138% of HBL) and appearing Fig. 21-1. Dorsal and ventral views of round museum skins: left pair, the holotype of Monticolomys koopmani (AMNH 100727, TOTL = 205 mm), a young adult male from Antananarivo Province, Manjakatompo; right pair, an original topotype of Macrotarsomys bastardi occidentalis (AMNH 119709, TOTL = 228 mm), an adult male from Toliara Province, Beroboka. CARLETON & GOODMAN: ENDEMIC RODENTS 235 Fig. 21-2. Plantar view of the left hind foot and lateral view of the distal phalanges of the fourth digit: A-A', Monticolomys koopmani (FMNH 151900, HFL = 25 mm), an adult female from Fianarantsoa Province, RNI d'An- dringitra; B-B', Macrotarsomys bastardi (USNM 578822, HFL = 27 mm), an adult male from Toliara Province, Petriky Forest. Abbreviations: 1-4, interdigital pads one through four; hy, hypothenar pad; th, thenar pad. Note the contrasts in size and position of the plantar pads between Monticolomys (A) and Macrotarsomys (B), greater pilosity on the undersurface of the phalanges in Macrotarsomys (B'), and stronger development of the ungual tuft in Mon- ticolomys (A')- monocolored; caudal hairs a pale brown, medium in length over entire tail without notable peni- cillation or development of distal tuft; scutella- tion fine, partially obscured by tail hairs. Mam- mae number six (based on FMNH 151899), dis- tributed as postaxial, abdominal, and inguinal pairs. Tops of metapodials and phalanges covered with fine white hairs; ungual tufts present, the white hairs extending beyond end of claw (Fig. 21-2); palmar and plantar surfaces and underneath of phalanges naked. Forefoot with stubby pollex bearing a nail, other manual digits with claws. Hind foot relatively long (Table 21-1), about 25% of HBL, and wide, with claws on all digits. Cen- tral digits (II-IV) of hind foot almost as long as corresponding metatarsals; digit V relatively long, its claw reaching to end of second phalanx of digit IV; digit I more reduced, its claw extending to middle of first phalanx of digit II (Fig. 21-3). Plantar surface bearing six moderately sized and cushion-like pads (Fig. 21-2); interdigital pads 2- 4 set close together at base of digits, interdigital 1 situated somewhat apart at base of hallux; hy- pothenar pad about as large as interdigitals, po- sitioned near to but slightly posterior of interdig- ital 1 ; thenar ovate, situated near middle of tarsus- metatarsus, and subequal in size to distal pads (Fig. 21-2). Cranium lightly constructed, delicate in ap- pearance (Fig. 21-4). Rostrum narrow and mod- erately long, about 35% of ONL, taper forward from nasolacrimal capsules to end of nasals very gradual; anterior tips of nasals rounded, posterior margins blunt and coextensive with posteriormost limit of rostral processes of premaxillae. Zygo- matic plate narrow, its anterior edge slightly con- vex and set well behind nasolacrimal capsule; posterior border of plate positioned appreciably in front of anterior root of M 1 ; dorsal notch distinct 236 FIELDIANA: ZOOLOGY Fig. 21-3. Dorsal view of the right hind foot skele- ton: A, Monticolomys koopmani (FMNH 151727, HFL = 24 mm), an adult male from Fianarantsoa Province, RNI d'Andringitra; B, Macrotarsomys bastardi (USNM 578719, HFL = 28 mm), an adult male from Toliara Province, Petriky Forest. Abbreviations: I-V, pedal digits one through five. Note the relatively greater lengths of digits I and V in Monticolomys as compared to those of Macrotarsomys, as well as the tighter coal- ignment of the metatarsals in Macrotarsomys. but shallow. Zygomatic arches thin and parallel- sided to slightly bowed over midportion, project- ing laterad just beyond margins of braincase and weakly converging anteriorly; jugal slender but relatively long, forming most of midspan of arch and distinctly separating zygomatic processes of squamosal and maxillary. Interorbital region narrow, exposing floor of or- bits in dorsal view; hourglass-shaped, lacking su- praorbital shelving and ridging; orbital projection of lacrimal inconspicuous. Braincase similarly free of ridges and crests, smooth and rounded in dorsal appearance (Fig. 21-4); frontoparietal su- ture evenly curved, not defining a sharp angle at midsagittal juncture; interparietal moderately large and rhomboidal in outline, its lateral apices not contacting squamosals; dorsal profile of skull gently arched throughout, its highest point formed just anterior to the frontoparietal junction. Incisive foramina medium in length, spanning about 66% of diastema and terminating at level of anterior root of first molars. Palatal bridge rel- atively broad and smooth, devoid of corrugations and excrescences; posterior palatine foramina oc- cur as simple round holes in maxillopalatine su- ture, about level with abutment of Ml-m2; pos- terior border of palate U-shaped, situated even with end of third molars, and lacking posterolat- eral palatal pits. Lateral borders (pterygoid pro- cesses) of mesopterygoid fossa more or less straight sided and enclosing spacious sphenopal- atine vacuities. Parapterygoid fossae broad com- pared to width of mesopterygoid fossa, relatively short and triangular in shape, recessed little rela- tive to the plane of the hard palate; roof of par- apterygoid fossae almost flat and mostly osseus, pierced only by posterior opening to alisphenoid canal near their posterolateral corner and there scored by a groove for passage of the infraorbital branch of the stapedial artery. Ectotympanic bullae moderately sized, with narrow posteromedial wedge of the periotic visi- ble in ventral aspect; anterodorsal rim of the bulla abuts the ventrolateral margin of the squamosal bone but not obscuring the slit-like middle lacer- ate foramen; anterior flange of tegmen tympani inconspicuous, not contacting ventrolateral mar- gin of squamosal; eustacian tube short, overlap- ping but not enveloping tip of pterygoid process; malleus of the parallel type, orbicularis apophysis knob-like without definitive neck, deflected ven- tromedially. Postglenoid foramen large, semioval in outline, circumscribing an area about three to four times that of the smaller subsquamosal fe- nestra; hamular process of squamosal well defined and slender. Mastoid capsule small and bulbous, perforated by posterodorsal fontanelle. Alisphenoid strut absent, masticatory-buccina- tor and accessory oval foramina conjoined as one expansive opening. Sphenofrontal and stapedial foramina present, squamosal-alisphenoid groove well defined, posterior opening of alisphenoid ca- nal large, and vascular groove crossing the pos- terolateral corner of the parapterygoid plate — an- atomical landmarks suggesting a complete ce- phalic circulatory pattern with retention of the su- praorbital and infraorbital branches of the stapedial artery. Coronoid process of dentary falcate, extending dorsad slightly above the condylar process; sig- moid and angular notches clearly incised but un- remarkable in size and shape; alveolus of lower incisor terminates posteriorly at level of coronoid CARLETON & GOODMAN: ENDEMIC RODENTS 237 .## *«*&£/ Fig. 21-4. Dorsal, ventral, and lateral views (X2) of the cranium and mandible: left, Monticolomys koopmani (FMNH 151899, ONL = 28.1 mm), an adult female from Fianarantsoa Province, RNI d'Andringitra; right, Macro- tarsomys bastardi (USNM 578715, ONL = 29.3 mm), an adult female from Toliara Province, Petriky Forest. process, well below ventral rim of sigmoid notch; capsular process evident as low mound. Axial skeleton (per one specimen, FMNH 151727) with 13 thoracic, seven lumbar, four sa- cral (two pseudosacral), and 38 caudal vertebrae; tuberculum of first rib articulates only with trans- verse process of first thoracic vertebra, not touch- ing transverse process of seventh cervical; neural spine of second thoracic vertebra rises conspicu- ously above the spines of other thoracics. Ent- epicondylar foramen present on humerus. Upper incisors asulcate and opisthodont; enam- el medium orange. Upper molar rows divergent anteriorly; upper and lower second molars about three-quarters the size of the first molars but sim- ilar in occlusal design; upper and lower third mo- lars notably smaller, circular to oval, about one- half to two-thirds the length and area of the con- tiguous second molars, and their serial homologies more obscure. Molars decidedly brachyodont, cuspidate (Fig. 21-5); principal cusps positioned nearly opposite in the upper molars, more alter- nate in the lowers. Anterocone of Ml well de- fined, broad and undivided (weakly bilobate in one specimen); anteroconid of ml hardly distin- guishable from metaconid and anterior cingulum; entoconid poorly defined on m3; low median mure(id) interconnects anterior and posterior pairs 238 FIELDIANA: ZOOLOGY Fig. 21-5. Occlusal views (X20) of the upper (left member) and lower (right member) right molar rows: left pair, Monticolomys koopmani (FMNH 151899, LM1-3 = 3.42 mm), an adult female from Fianarantsoa Province, RNI d'Andringitra; right pair, Macrotarsomys bastardi (USNM 578715, LM1-3 = 3.95 mm), an adult female from Toliara Province, Petriky Forest. of cusps on each first and second molar. Principal lingual and labial enamel folds broad and meet near the midline of the tooth, not interpenetrating; posteroloph (postcingulum) of Ml -2 and corre- sponding posteroflexus indistinct, obliterated after slight wear; posterolophid and broad posteroflexid present and persistent with wear on ml -2, indis- tinct or absent on m3; protoflexus of M2 absent, anteroloph and paraflexus short; metaflexid of ml represented by faint indentation or indiscernible, opposite protoflexid well defined; anterolabial cin- gulum present but small on m2, variable on m3; mesolophs(ids), as well as other accessory crests and styles, absent. Three roots on each upper mo- lar and two on each lower. Notes on Natural History — The habitat where Rand collected the type is characterized on the skin specimen tag as "grass slope with scat- tered bushes." The type specimen, even at the time of the MZFAA in 1929, likely originated from despoiled habitat. In his later summary of the itinerary, Rand (1936: pp. 161 and 163) ex- panded upon the vegetation of Ankaratra: The slopes of the greater part of the mountain were grass-covered and supported a growth of health and bracken . . . just above Monjakatom- po, was an area of humid forest, an isolated remnant of the forest that once covered the cen- tral portion of Madagascar. This wooded area was of the humid forest type, with large trees hung with lianas and mosses. Tree ferns were common and in some places there was consid- erable undergrowth. Populations of Monticolomys koopmani may co-occur with a variety of other nesomyine spe- cies. On Ankaratra, Rand also collected three specimens of Brachyuromys betsileoensis (AMNH 100802, BMNH 35.1.8.341, and MNHN 1932.3521) near Manjakatompo, at 1700 m and 1950 m. The spare habitat annotations on their skin specimen tags ("grass and brush at edge of pond"; "grass and brush ridge") leave open the question of whether the Brachyuromys were syn- topic with Monticolomys. The grasslands on the mountain are not natural vegetation formations but have resulted from human-set fires (Perrier de la Bathie, 1927; Nicoll & Langrand, 1989). To judge from the personal field notes of Richard CARLETON & GOODMAN: ENDEMIC RODENTS 239 Table 21-1. Selected external and craniodental measurements (in mm) of the holotype (AMNH 100727) and referred specimens (FMNH 151727, 151899, and 151900) of Monticolomys koopmani. Holotype (Ankaratra Referred specimens Variable Massif) (Andringitra Massif) TOTL 205.0 236.3 (234-240) HBL 89.0 98.0(94-101) TL 116.0 138.0(134-143) HFL 24.0 24.3 (24-25) EL 15.0 18.3(18-19) WT 26.3 (25-27) ONL 26.3 27.8(27.5-28.1) ZB ... 13.6(13.1-14.1) BBC 12.5 12.7(12.4-13.2) IOB 3.9 4.0 (3.9-4.0) LR 8.5 9.9 (9.6-10.2) BR 4.2 4.7 (4.5-4.9) PPL 8.7 9.5 (9.3-9.7) LBP 3.7 4.2 (4.0-4.4) LIF 5.3 5.2(5.1-5.3) BIF 2.0 2.0(1.9-2.0) LD 7.2 7.9 (7.7-8.0) BMls 5.7 5.9 (5.8-6.0) DAB 4.4 4.8 (4.7-4.9) BZP 2.2 2.2(2.1-2.2) BOC 6.4 6.6 (6.5-6.8) LM1-3 3.7 3.5 (3.4-3.7) WM1 1.1 1.1(1.0-1.1) Refer to the Materials and Methods section for an ex- planation of the abbreviations. Archbold, who accompanied Rand to Ankaratra and also collected around Manjakatompo, popu- lations of Rattus rattus were well established in the area at the time. The more thorough survey conducted within the RNI d' Andringitra documents sympatry of Monticolomys koopmani with seven other rodent species — Rattus rattus, Brachyuromys ramirohi- tra, Eliurus majori, E. minor, E. tanala, Gymnu- romys roberti, and Nesomys rufus (voucher num- bers are listed in Chapter 22). Several of these Nesomyinae are themselves restricted to montane and upper montane vegetational zones {Brachy- uromys ramirohitra, Eliurus majori, E. tanala, and Nesomys rufus) or are widely ranging in al- titudinal occurrence (Eliurus minor and Gymnu- romys roberti). The specimens from Andringitra were collected in upper montane forest (ridge-top sclerophyllous forest) that is regularly shrouded in mist and cloud cover. The area (Fig. 21-6) is dominated by dense bamboo stands of the genera Arundinaria and Nastus and by trees of the families Podocarpa- ceae, Cunoniaceae, and Pandanaceae (Chapter 2). For woody plants over 10 cm diameter at breast height (dbh), trees in the 1625 m zone averaged 17.0 cm dbh and 7.9 m tall (Chapter 4). Further- more, at this elevation, there was a marked in- crease in epiphytes, mosses, and lichens in tree crowns and on trunks in comparison to forests at lower elevations. The relatively long tail, the conformation of the hind foot, and the large plantar pads suggest that individuals of Monticolomys are adept climbers. In the RNI d' Andringitra, both live-caught spec- imens (FMNH 151899 and 151900) were cap- tured on successive nights in the same Sherman trap placed on a nearly horizontal segment of a liana that was less than 5 cm in circumference and 2 m above the ground. The vine originated at the ground and wound its way to the canopy. The third specimen (FMNH 151727) was found the next morning in a pitfall bucket along with two Microgale taiva (FMNH 151724 and 151725). With just three records, little can be said about the species' focus of activity, whether predominantly arboreal, scansorial, or terrestrial. The two males from Andringitra, collected in December 1993, possessed partially or fully de- scended testes; the one female had small teats and an imperforate vagina. Remarks — The natural environments on the Ankaratra and Andringitra massifs inhabited by Monticolomys are isolated, their unique montane populations wholly allopatric to one another (see discussion on biogeography below and Fig. 21-11). The meager sample sizes available from the type locality (N = 1) and from the RNI d' Andringitra (N = 3) hinder any meaningful evaluation of the level of differentiation between the two distantly separated localities. Bias in age representation further complicates appreciation of intra- and interlocality differences — the holotype is a young adult male, his molars little worn but in mature pelage, whereas the three paratypes (two males, one female) are fully adult with mod- erately to heavily worn toothrows. Age-related, post-weaning growth may plausi- bly account for the generally smaller size of the holotype as compared to the three Andringitra specimens (Table 21-1). Certain mensural vari- ables—such as HBL, TL, ONL, LR, LD, PPL, all of which are decidedly shortest in the holotype — may exhibit substantial age effects among the samples typically available to the systematist, even with attempts to control for age biases by eliminating juveniles or restricting analyses to 240 FIELDIANA: ZOOLOGY Table 21-2. Comparison of selected external and craniodental measurements (in mm) of Monticolomys koopmani and species of Macrotarsomys. Statistics for the samples include the mean ± standard deviation, and the range. Monticolomys Macrotarsomys Macrotarsomys Macrotarsomys koopmani b. bastardi b. occidentalis ingens Variable (N = 4) (N = 9) (N = 4) (N = 4) TOTL 228.5 ± 15.9 213.0 ± 11.7 225.5 ± 10.8 336.5 ± 36.6 205.0-240.0 195.0-232.0 213.0-235.0 310.0-390.0 HBL 95.7 ± 5.4 91.6 ± 3.2 96.7 ± 4.2 127.0 ± 15.7 89.0-101.0 86.0-95.0 92.0-102.0 115.0-150.0 TL 132.5 ± 1.16 121.5 ± 12.1 128.7 ± 6.9 209.5 ±21.7 116.0-143.0 100.0-142.0 121.0-136.0 193.0-240.0 HFL 24.3 ± 0.5 24.6 ± 0.9 26.7 ± 1.0 35.3 ± 2.5 24.0-25.0 23.0-26.0 26.0-28.0 32.0-38.0 EL 17.5 ± 1.7 21.3 ± 1.2 23.0 ± 1.6 24.3 ± 1.5 15.0-19.0 20.0-23.0 21.0-25.0 23.0-26.0 WT 26.3 ± 1.2 24.5 ± 2.7 31.0 ± 5.6 57.0 25.0-27.0 21.0-28.0 26.0-38.0 54.0, 60.0 ONL 27.4 ± 0.8 28.1 ± 0.7 28.6 ± 0.5 38.9 ± 2.5 26.3-28.1 26.7-28.9 28.2-29.3 35.4-41.3 ZB 13.6 ± 0.5 14.0 ± 0.5 14.5 ± 0.4 18.9 ± 1.5 13.1-14.1 13.0-14.6 14.1-14.8 16.7-20.2 BBC 12.7 ± 0.4 11.3 ± 0.2 11.7 ± 0.3 14.3 ± 0.5 12.4-13.2 10.9-11.5 11.3-12.1 13.5-14.3 IOB 3.9 ± 0.1 4.6 ± 0.1 4.8 ± 0.1 5.6 ± 0.4 3.9-4.0 4.3-4.7 4.6-4.9 5.1-6.0 LR 9.5 ± 0.7 10.2 ± 0.5 10.1 ± 0.3 14.4 ± 1.2 8.5-10.2 9.5-11.1 9.9-10.5 13.0-16.0 BR 4.5 ± 0.3 4.7 ± 0.2 4.9 ± 0.2 6.8 ± 0.3 4.2-4.9 4.4-5.1 4.7-5.2 6.6-7.1 PPL 9.3 ± 0.4 9.5 ± 0.4 9.8 ± 0.6 13.3 ± 1.1 8.7-9.7 9.1-10.0 9.1-10.3 11.9-14.6 LBP 4.1 ± 0.3 4.7 ± 0.2 4.6 ± 0.3 6.3 ± 0.5 3.7-4.4 4.4-4.9 4.2-4.9 5.6-6.9 LIF 5.2 ± 0.1 5.1 ± 0.2 5.0 ± 0.3 7.1 ± 0.4 5.1-5.3 4.9-5.5 4.7-5.4 6.7-7.5 BIF 2.0 ± 0.1 2.0 ± 0.1 2.1 ± 0.2 2.6 ± 0.2 1.9-2.1 1.8-2.1 1.9-2.3 2.4-2.8 LD 7.7 ± 0.3 7.4 ± 0.4 7.7 ± 0.2 11.6 ± 0.9 7.2-8.0 6.8-8.0 7.5-7.9 10.3-12.5 BMls 5.9 ± 0.1 5.9 ± 0.2 5.9 ± 0.1 7.7 ± 0.5 5.7-6.0 5.5-6.2 5.7-5.9 7.0-8.0 DAB 4.7 ± 0.3 5.5 ± 0.2 5.4 ± 0.1 7.0 ± 0.6 4.4-4.9 5.1-5.7 5.2-5.5 6.3-7.5 BZP 2.2 ± 0.05 2.9 ± 0.2 2.8 ± 0.1 3.8 ± 0.2 2.1-2.2 2.6-3.1 2.7-2.9 3.5-4.0 BOC 6.5 ± 0.2 6.1 ± 0.3 6.5 ± 0.2 8.4 ± 0.6 6.4-6.8 5.6-6.3 6.3-6.7 7.7-8.9 LM1-3 3.57 ±0.12 4.08 ±0.17 3.91 ± 0.05 4.85 ±0.16 3.42-3.70 3.79-4.31 3.85-3.95 4.71-5.04 WM1 1.09 ± 0.02 1.28 ± 0.06 1.19 ± 0.02 1.64 ± 0.04 1.07-1.12 1.21-1.39 1.17-1.21 1.59-1.68 specimens assigned to crudely defined age-tooth- wear categories (Voss & Marcus, 1992). We note that several cranial dimensions measured across the brain (BBC, IOB, BOC) or on the molars LM1-3, WM1), regions that change little or not at all with age following weaning, are basically alike in the Ankaratra and Andringitra examples (Table 21-1). In view of such insufficiencies in sample size and imbalanced age representation, and lacking other evidence of substantial divergence, we treat the two allopatric samples as one species but stress the need for additional voucher material. Broader age series are especially critical to verify the typical condition of traits such as the presence of a posteroloph and formation of the anteroconid. CARLETON & GOODMAN: ENDEMIC RODENTS 241 Fig. 21-6. View through the understory of upper montane forest at 1625 m on the eastern slopes of the RNI d'Andringitra. Portions of this forest, where three specimens of Monticolomys koopmani were taken during the 1993 inventory, are dominated by dense stands of bamboo (Arundinaria and Nastus), intermixed with Podocarpus trees. The groundcover is largely covered by Pandanus (Photo by C. J. Raxworthy). These dental characters are obliterated after slight lifelong systematic contributions of Karl F. Koop- trituration and are best observed in juvenile spec- imens. Etymology — The generic name, mountain- dwelling mouse, evokes the montane occurrence of the form. The species is named to honor the man, Curator Emeritus in the Department ol Mammalogy, AMNH, and one-time Assistant Cu- rator of Mammals, FMNH. His curiosity, piquec by the orphaned skin that would not key out prop erly according to the standard classification, lee 242 FIELDIANA: ZOOLOGY him to the perception that it indeed represented a fundamentally different kind of Malagasy rodent. Morphological Comparisons Individuals of Monticolomys koopmani are im- mediately distinguishable from other kinds of Ne- somyinae on the basis of their diminutive size and generalized murine appearance. The mouse-like form offers sharp visual contrast to the distinctive and highly divergent physiognomies of the much larger Brachytarsomys, Brachyuromys, Gymnu- romys, Hypogeomys, and Nesomys (for example, see descriptions, measurements, and/or keys in Ellerman, 1941, 1949; Petter, 1972, 1975; and Chapter 22, this work). Among the tufted-tailed rats, genus Eliurus, only specimens of E. minor approach those of Monticolomys in overall size. Although comparable in average total length, in- dividuals of Monticolomys possess a still lighter build, shorter head-and-body length, relatively longer tail, and longer hind feet (see Carleton, 1994; and Chapter 22, Table 22-2). Furthermore, they lack the conspicuous "bottle-brush" tail that characterizes E. minor and other species of Eli- urus. Crania and dentitions of E. minor and Mon- ticolomys differ even more strikingly than exter- nal form and permit no confusion of the two (see Carleton, 1994). Although discrimination of Monticolomys from most nesomyines is straightforward, two genera, one living and one extinct, present less-pro- nounced contrasts and raise broader questions of phylogenetic relationship. One, Macrotarsomys, whose gerbil-like physique may suggest very dis- tant phyletic association, in fact resembles Mon- ticolomys quite closely in certain cranial and den- tal details that require critical elaboration. The other, the Miocene form Protarsomys, merits at- tention because some authors have viewed it as close to the ancestry of Madagascar's indigenous rodents (Lavocat, 1973, 1978), if not actually syn- onymous with extant Macrotarsomys (Chaline et al.. 1977; Petter, 1990). Comparison with Macrotarsomys Milne Edwards and G. Grandidier The following morphological descriptions and comparative accounts are based on the four ex- amples of Monticolomys koopmani and specimens CARLETON & GOODMAN: ENDEMIC RODENTS representing both Macrotarsomys bastardi (in- cluding M. occidentalis Ellerman, 1949) and M. ingens (see Materials and Methods section). The type species of Macrotarsomys, M. bastardi Milne Edwards and G. Grandidier (1898), differs in subtle aspects from the later named and phys- ically much larger congener M. ingens (Petter, 1959); however, unless noted otherwise, contrasts to Monticolomys apply to both species of Macro- tarsomys. Pelage and External Form — The dark brown dorsum and dull gray ventrum of Monticolomys present a somber impression compared to the light sandy to clay brown upperparts and wholly white underparts of Macrotarsomys. Lateral demarca- tion of the dorsal-ventral pelage is thus conspic- uous in the latter genus. Pelage texture is similarly fine in both genera, but appears longer and denser, for the size of the animal, in the sample of Mon- ticolomys. The generalized murine appearance of Monti- colomys offers several points of contrast to the gerbilline form that characterizes Macrotarsomys (Fig. 21-1). The pinnae of Monticolomys are small, thick, and rounded, protruding only slightly above the body fur, and densely cloaked with black hairs. Those of Macrotarsomys, however, are much longer and wider (Table 21-2), appear- ing almost naked (clothed with fine white hairs internally and pale brown externally) and more pliant. The tails of both Monticolomys and Mac- rotarsomys bastardi are relatively long (TL 133- 138% of HBL), that of M. ingens exceptionally so (TL 160% of HBL). Examples of Monticolo- mys, however, lack the distal elongation of caudal hairs forming a modest pencil; instead the hairs are roughly even in length, spanning about two caudal annuli, over the entire tail. In specimens of Macrotarsomys, hairs covering the proximal two-thirds of the tail are sparse and short, about one annulus in length, but they lengthen some- what abruptly over the distal one-third to form a noticeably penicillate tip (not like the conspicuous terminal brush, however, found in species of Eli- urus). Notwithstanding the wide range in absolute lengths (Table 21-2), the hind feet of Monticolo- mys and Macrotarsomys have similar proportions (HFL about 26-28% of HBL). Hind Foot — The genera differ markedly in hind foot conformation, size and apportionment of the plantar pads, and distribution of pedal hairs. The pes of Macrotarsomys appears more svelte and absolutely narrower in view of the greater length of the proximal tarsal-metatarsal portion relative to the distal phalanges (Fig. 21-2). Not 243 only are the digits of Macrotarsomys generally shorter compared to those of Monticolomys, the outer members (I and V) are more reduced rela- tive to the central three (II-IV), a proportional configuration evident in both the intact hind foot and the foot skeleton (Figs. 21-2 and 21-3). Thus, in Macrotarsomys, the claw of digit V extends to the junction of phalanges 1-2 of digit IV (junction of 2-3 in Monticolomys), and that of digit I reach- es the end of the second metatarsal (middle of the first phalanx of digit II in Monticolomys). In ad- dition, the five metatarsal bones in Macrotarso- mys are closely appressed, although not fused, along their full length. Those of Monticolomys, however, splay noticeably at their distal ends and appear in looser contact along their proximal sec- tions (Fig. 21-3). Individuals of both Monticolomys and Macro- tarsomys possess six plantar pads, but their size, shape, and spatial arrangement contrast apprecia- bly. The six footpads in Monticolomys are fleshy, bulbous mounds, more of less equal to one anoth- er in circumference. In Macrotarsomys, the pads are not only smaller and lower in relief but differ in size; that is, the posterior three (thenar, hypo- thenar, and interdigital 1) are about half the di- ameter and protuberance of the anterior interdi- gitals 2-4 (Fig. 21-2). The hypothenar and inter- digital 1 are coaligned in Macrotarsomys (more alternate in Monticolomys), and the thenar appears to be positioned more distally from the heel. Per- haps the most noticeable difference when exam- ining the plantar surfaces of the two is the devel- opment of the thenar pad — an ovate, pillowy mound in Monticolomys compared to an insub- stantial, circular papilla in species of Macrotar- somys. In contrast to the naked plantar surface of Mon- ticolomys, scattered hyaline hairs are found on the sole in examples of Macrotarsomys bastardi, as well as along the ventral surface of the distal pha- langes (Fig. 21-2). In particular, a small cluster of hairs typically marks the midsole, between inter- digital pads 2-4 and the paired hypothenar and interdigital 1 . The single fluid specimen of Mac- rotarsomys ingens examined (USNM 328831) lacks such isolated hairs on the midsole but pos- sesses them on the undersides of the toes. Unlike the hairs covering the dorsum of the foot, these plantar hairs seem stiffer, but neither they nor hairs on the phalangeal undersurfaces were de- tected in the two fluid specimens of Monticolo- mys. On the other hand, the ungual tufts, tussocks of hairs that emerge at the base of the claws, ap- 244 pear more profuse in individuals of Monticolomys (Fig. 21-2), arching over and beyond the tip of the claw (in both Macrotarsomys, reaching about one-half to three-quarters of claw length but not surpassing the tip). Cranium and Mandible — The crania of Mon- ticolomys and Macrotarsomys share many quali- tative traits that convey a fundamental resem- blance. Notable among these are moderately elon- gate incisive foramina (LIF 62-68% of LD); a flat, short, and wide palate with simple posterior palatine foramina; broad, shallow parapterygoid fossae and narrow mesopterygoid fossa with large sphenopalatine vacuities; absence of alisphenoid struts; size and form of temporal openings and well-defined hamular process; and presence of vascular foramina and osseous grooves indicative of a primitive carotid blood supply. They princi- pally depart from one another in features of the interorbital region, configuration of the maxillary portion of the zygoma, and anatomy of the audi- tory bullae and surrounding area. The interorbital region and braincase of both genera are smooth and gently contoured, devoid of pronounced temporal ridges or crests. Speci- mens of Macrotarsomys bastardi, however, pos- sess an absolutely and relatively wider interorbit such that little of the orbital floor (top of maxil- lary) is visible from above. Also, they typically display a weak supraorbital shelf at the rear of the frontals that imparts a sharper constriction to the interorbit and longer appearance to the braincase (Fig. 21-4). The interorbit in examples of Macro- tarsomys ingens has a relatively narrow, hour- glass-shaped (amphoral) interorbit, more closely resembling that of Monticolomys. The zygomatic plate in examples of Montico- lomys is absolutely and relatively (BZP 8% of ONL) narrower than that observed in both species of Macrotarsomys (BZP 10% of ONL). Not only do specimens of Macrotarsomys have broader zy- gomatic plates, forming deeper zygomatic notch- es, but the zygomatic process of the maxilla wid- ens appreciably at its union with the plate and has a more rugose surface (Fig. 21-7). A bony ledge from the lacrimal interrupts the anterodorsal pe- rimeter of the orbit in samples of Macrotarsomys bastardi but not those of Monticolomys (Figs. 21-4 and 21-7). This tabular projection, reminis- cent of the condition observed in certain Gerbil- linae (Carleton, 1980), is intermediate in size in M. ingens. Bullar inflation is pronounced in both species of Macrotarsomys (DAB about 48% of BBC) FIELDIANA: ZOOLOGY Fig. 21-7. Left lateral-oblique view of the rostrum and anterior zygomatic region: top, Monticolomys koop- mani (FMNH 151899; Fianarantsoa Province, RNI d'Andringitra); bottom, Macrotarsomys bastardi (USNM 578715; Toliara Province, Petriky Forest). Ab- breviations: lop, orbital projection of the lacrimal bone; pm, premaxilla; zn, zygomatic notch; zp, zygomatic plate. Note the wider zygomatic plate and deeper notch in Macrotarsomys, as well as the tabular extension of the lacrimal bone at the anterodorsal rim of the orbit. less so in Monticolomys (DAB 37% of BBC), and perhaps accounts for certain other cranial dissim- ilarities. In Macrotarsomys, hypertrophy of the ectotympanic (Figs. 21-8 and 21-9), particularly noticeable in the expanse of translucent bone en- closing the middle ear cavity, plausibly relates to reduction of the middle lacerate foramen (present as a slit in Monticolomys), almost complete in- vestiture of the periotic in ventral view (visible as posteromedial wedge in Monticolomys), and for- mation of a deep trough on the medial bullar wall for passage of the internal carotid artery (round foramen in Monticolomys). In both genera, the tegmen tympani, an anterior flange of the periotic that roofs the epitympanic recess, lacks articula- tion with the squamosal, presumably an ancestral connection found in many New World muroids (see Voss, 1993) and in some nesomyines (e.g., Eliurus; see Carleton, 1994, Fig. 18). The tegmen tympani is inconspicuous though visible within the postglenoid foramen of Monticolomys, but it is smaller yet in Macrotarsomys and wholly ob- scured by the thickened, rugose dorsal rim of the external auditory meatus (Fig. 21-8). Mandibular conformation of the two genera dif- fers in several details. In Monticolomys, the cor- onoid process is more substantial and oriented more vertically; in contrast, the process in species of Macrotarsomys tapers to a thin spine that is angled more posterodorsally, in parallel to the ori- entation of the condylar process. As a conse- quence, the sigmoid notch of Macrotarsomys ap- pears narrower and shallower (Fig. 21-4); that of Monticolomys forms a broader half-oval typical of many muroids. The angular notch in specimens of Macrotarsomys is deeply excavated, imparting a broader and longer definition to the angular and condylar processes. The concavity of the angular notch is shallower in Monticolomys, and the an- gular and condylar processes are normally pro- portioned. Dentition — The upper incisors of Macrotar- somys, like those of Monticolomys, lack enamel grooves but differ slightly in color; they are yel- low-orange in the former and medium orange in the latter. The molars of Monticolomys and Macrotarso- mys are so remarkably alike that correct segre- gation of the genera is doubtful, except on the basis of size (Fig. 21-5). A trace of a posteroloph and minute posteroflexus is discernible on the M 1 -2s of most Monticolomys and Macrotarsomys ingens, but they seem to be lacking on those of M. bastardi, even for individuals with little tooth wear. The anteroconid is weakly defined in both genera, appearing as little more than an undiffer- entiated procingulum that originates from the an- terolingual shoulder of the metaconid, forms the front rim of the tooth, and merges imperceptibly with the anterolabial cingulum near the proto- conid. In some examples of Macrotarsomys, how- ever, especially M. ingens (USNM 576753), a shallow indentation (vestige of the metaflexid?) on the anterolingual face of ml imparts some sep- aration of the metaconid from the presumptive an- teroconid. Such evident differences must be viewed as qualified until verified by larger series of each species, including juveniles. CARLETON & GOODMAN: ENDEMIC RODENTS 245 Fig. 21-8. Left lateral view of otic region and associated foramina: left, Monticolomys koopmani (FMNH 15 1900; Fianarantsoa Province, RNI d'Andringitra); right, Macrotarsomys bastardi (USNM 578717; Toliara Province, Petriky Forest). Abbreviations: ab, auditory bulla (ectotympanic); ex, exoccipital; hp, hamular process of the squamosal; ms, mastoid capsule of periotic; pgf, postglenoid foramen; sq, squamosal; ssf, subsquamosal fenestra; tt, tegmen tympani. The tegmen tympani fails to contact the squamosal in either genus, but it remains visible within the postglenoid foramen of Monticolomys and is obscured by the rugose dorsal rim of the ectotympanic in Macrotarsomys. The Miocene Fossil Protarsomys Lavocat The time of arrival of the first nesomyine, or nesomyines, into Madagascar remains inferential, given the absence of terrestrial Tertiary records on the island and the rarity of appropriately aged fos- sil muroids from sites in eastern Africa, the pre- sumed area of origin. Lavocat (1978) postulated the entrance of ancestral nesomyines by over-wa- ter dispersal in the lower Miocene, in part because this epoch marks the first appearance of archaic cricetodontids in Africa. A period later than the lower to middle Miocene would prove difficult to reconcile with the lack of murines in Madagascar; they appear in the late Miocene of northern Africa and become commonplace in Pliocene beds of sub-Saharan Africa (see, e.g., Jacobs, 1985; De- nys, 1987). A more pivotal consideration, however, con- cerns the afrocricetodontine Protarsomys macin- nesi (Muroidea: Cricetodontidae). Lavocat (1973) described this fossil rodent from the Rusinga Fau- nal Assemblage, lower Miocene of Kenya, and viewed (1978: p. 80) it as close to the ancestry of the Malagasy Nesomyinae, in particular Macro- tarsomys: "The genus Protarsomys, approximate- ly the size of recent Mus musculus, is so close to the Recent Macrotarsomys [his comparisons in- volved only M. bastardi, not the larger M. ingens] of Madagascar that one must evidently find in this Protarsomys, or in very close relatives of this ge- nus, the ancestors of the Malagasy rodents." Cha- line et al. (1977) had earlier formalized this in- terpretation by synonymizing Protarsomys under Macrotarsomys, an action and affinity reaffirmed by Petter (1990). Description of Monticolomys, which itself ex- hibits traits suggesting relationship to Macrotar- somys, necessarily requires some consideration of Lavocat 's (1973) Protarsomys. The remarkable holotype of Protarsomys (KNM 2350) consists of a mostly intact skull with associated lower jaws embedded in matrix. Partial removal of the en- casing matrix has revealed the general dimensions and morphology of the cranium and mandibles in dorsal and lateral view, but features of the palate, basicranium, and otic capsules, as well as occlusal surfaces of the upper and lower molars, are still obscured on the type specimen. Accordingly, size comparisons of Protarsomys to specimens of Monticolomys and Macrotarsomys are limited to those variables accessible to measurement on the borrowed cast of the holotype. The configuration of other cranial and dental features has been gleaned from Lavocat 's (1973) excellent descrip- tion, from the cast of a referred lower right man- dible with ml -3 (KNM 2188), and from inspec- tion of his Plate 10, which portrays another attrib- uted specimen (KNM 2353) in which the ventral surface of the skull is exposed. As noted by Lavocat (1973, 1978), Protarso- mys has a small skull, about the size of a house mouse, and as such, it is considerably smaller than 246 FIELDIANA: ZOOLOGY Fig. 21-9. Ventral view of left otic region and parapterygoid fossa: left, Monticolomys koopmani (FMNH 151900; Fianarantsoa Province, RNI d'Andringitra); right, Macrotarsomys bastardi (USNM 578717; Toliara Province, Petriky Forest). Abbreviations: ab, auditory bulla (ectotympanic); cc, carotid canal for passage of the internal carotid artery; iag, groove marking the course of the infraorbital branch of the stapedial artery; mf, mesopterygoid fossa; mlf, middle lacerate foramen; pac, posterior opening to the alisphenoid canal; ppf, parapterygoid fossa; sag, squamosal-alisphenoid groove, which indicates the presence of the supraorbital branch of the stapedial artery; sf, stapedial foramen. The middle lacerate foramen remains patent in Monticolomys but is ventrally obscured by the inflated ectotympanic bulla in Macrotarsomys. individuals of either Monticolomys or Macrotar- somys bastardi (Table 21-3). Disparity in size, of course, is not relevant to generic membership, but the condition of other characters does bring into question the closeness of the purported phyletic affinity. Lavocat (1973) stressed similarities in overall skull shape and curve of the cranial vault of his new genus Protarsomys and Macrotarsomys. They both agree, in a general way, in having a somewhat delicate rostrum, narrow zygomatic span, smoothly contoured braincase, and gently arched skull. Although not expanded laterally, the arches of Protarsomys over their midsection are notably stout for the size of the skull, heavier than those observed in the larger Macrotarsomys. The comparatively broad interorbit and posteriorly di- vergent frontals in Protarsomys convey additional resemblance to the cranium of Macrotarsomys bastardi; however, possession of wide frontals that slightly overhang the rear of the orbit does not characterize the other living member of the genus, M. ingens. Temporal fenestration and definition of a ham- ular process appear comparable in the fossil and the Recent genera, but the otic capsules of Mac- rotarsomys are substantially larger than those of Protarsomys. The inflation of the bullae of Pro- tarsomys with respect to Monticolomys is difficult to ascertain because of the distorted preservation in this region, but they seem still smaller, to judge from the figured specimen (Lavocat, 1973: Plate CARLETON & GOODMAN: ENDEMIC RODENTS 247 Table 21-3. Comparison of selected cranial dimen- sions (in mm) of the Miocene fossil Protarsomys (cast of holotype, KNM 2350) with recent Monticolomys (N = 4) and Macrotarsomys (N = 9). For the latter two taxa, the mean and range are given. Protar- somys Vari- mac- Monticolomys Macrotarsomys ables innesi koopmani b. bastardi ONL 21.6 27.4(26.3-28.1) 28.1 (26.7-28.9) ZB 10.8 13.6(13.1-14.1) 14.0(13.0-14.6) BBC 9.3 12.7(12.4-13.2) 11.3(10.9-11.5) IOB 3.7 3.9 (3.9-4.0) 4.6 (4.3-4.7) LR 7.3 9.5 (8.5-10.2) 10.2(9.5-11.1) BR 4.1 4.5 (4.2-4.9) 4.7(4.4-5.1) LD 5.6 7.7 (7.2-8.0) 7.4 (6.8-8.0) BMls 4.4 5.9 (5.7-6.0) 5.9 (5.5-6.2) BZP 2.0 2.2(2.1-2.2) 2.9(2.6-3.1) 10). Lavocat identified a stapedial foramen and carotid canal in Protarsomys; the presence of oth- er cranial foramina and placement of vascular tracings indicative of the carotid circulatory plan are unknown. The incisive foramina are moder- ately long and wide, as in Macrotarsomys and Monticolomys, and they terminate posteriorly near the level of the anterior roots of the Mis. Protarsomys contrasts strikingly with Macro- tarsomys and Monticolomys in the architecture of its infraorbital canal and accompanying zygomat- ic plate. The infraorbital foramen of Protarsomys is broadly elliptical or nearly circular. That ob- served in Monticolomys and Macrotarsomys re- calls the oval "keyhole shape" typical of more derived muroids; that is, the ventral portion is nar- rowed as a neurovascular fissure relative to the spacious dorsal section, which is rounded for pas- sage of the anterior body of the masseter medialis. Presumably correlated with the different design of the infraorbital foramen is the construction of the zygomatic plate. In Protarsomys, the plate is small, about as long as wide, and marginally in- clined above the horizontal; the front edge of the plate extends little beyond the superior ramus of the anterior zygomatic arch and defines only a shallow zygomatic notch. The zygomatic plate in examples of Monticolomys and Macrotarsomys occupies a greater area, clearly longer (taller dor- sad) than wider, and angles noticeably oblique (dorsolaterad) to the midsagittal plane of the skull; the leading margin of the plate is produced farther forward, more so in Macrotarsomys than in Mon- ticolomys, creating a deeper zygomatic notch. Relative to the molar rows, the zygomatic plate of Protarsomys appears displaced posteriad, its rear edge coaligned with the anterior root of M 1 , a condition that Lavocat (1973) interpreted as ple- siomorphic. In contrast, the plates in Monticolo- mys and Macrotarsomys are positioned further forward on the rostrum, well in advance of the front of the Mis. We emphasize that the infraorbital foramen of Monticolomys and Macrotarsomys, compared to many murids and some nesomyines, forms a fairly broad oval that lacks pronounced compression of the ventral space as a narrow slit distinct from a much wider dorsal opening. Nevertheless, the in- fraorbital-zygomatic construction of these two genera more strongly suggests the typical murid condition. The round infraorbital canal and rudi- mentary, almost horizontal zygomatic plate of Protarsomys, on the other hand, resemble the de- velopment found in other contemporaneous Mio- cene Afrocricetodontinae, such as Afrocricetodon and Notocricetodon (Lavocat, 1973). Lavocat (1973) noted two principal differences between the mandibles of Protarsomys and Mac- rotarsomys. No lateral protuberance marks the posterior limit of the alveolus of the lower incisor in Protarsomys, whereas a small but distinct cap- sular process occurs on the ascending ramus of Macrotarsomys (including M. ingens and Monti- colomys). Furthermore, the apex of the masseteric crest in Protarsomys reaches only as far forward as below the middle of ml; in Macrotarsomys and Monticolomys, the crest passes beyond the ante- rior root of ml to end just behind the mental fo- ramen. All three genera possess opisthodont upper in- cisors and very low-crowned, cuspate molars that retain longitudinal enamel connections (mures and murids). Their dentitions contrast substantial- ly in other particulars. The molar tubercles in ex- amples of Macrotarsomys and Monticolomys are relatively thickly enameled, their borders round- ed, and their occlusal topography bunodont. Cor- responding to the robust cusp development, the enamel folds and valleys appear narrower and more crevice-like. In Protarsomys, the principal cusps are less massive, forming tapered conical peaks, the enamel seemingly thinner (Lavocai [1973] characterized their form as pinched ante riorly-posteriorly or somewhat selenodont); tht valleys between major cones are more open anc their bottoms U-shaped. The anterocone of Mi forms a broad, occasionally bilobed cusp in Ma crotarsomys and Monticolomys; that of Protar somys is narrower, without suggestion of an an 248 FIELDIANA: ZOOLOG" teromedian sulcus or bilobation. An anteroconid is weakly expressed in all three taxa, but that of Protarsomys is physically discrete, a distinct me- taflexid, equal in size to the contralateral proto- flexid, incising its posteromedial margin from the nearby metaconid. Posterolophs and companion posterofiexi are eminently discernible on the M 1 -2s of Protarsomys; these enamel features are imperfectly expressed or vestigial in specimens of Macrotarsomys and Monticolomys. The upper and lower third molars of Protarsomys are larger rel- ative to the length of the toothrow. In particular, the m3 of the Miocene fossil exhibits a small but distinct entoconid, posterolophid, and posteroflex- id — features that are indistinct or lost on the re- duced m3 talonids of Macrotarsomys and Monti- colomys. Finally, paramount among the dental structures mentioned by Lavocat (1973) Protar- somys displays low and thin mesolophs(ids) on its first two molars of the upper and lower toothrows. Such transverse enamel ridges do not occur in in- dividuals of the two living genera. Discussion Comments on genealogical relations and his- torical biogeography of Monticolomys are neces- sarily preliminary in nature, a caveat that under- scores the meager number of specimens and field data as yet available for fuller appreciation of its morphology, ecology, and geographical distribu- tion. Small sample sizes, for instance, hinder de- finitive characterization of dental structures and recognition of qualitative differences, if they ex- ist, between Monticolomys and species of Macro- tarsomys. Treatment of their relationship within a rigorous context is hampered by the lack of res- olution surrounding our understanding of higher- level phylogeny within Muroidea and, most im- portant for our topic, the corroboration of Neso- myinae as monophyletic and a working hypothe- sis of generic interrelationships. In view of these uncertainties, the formal diagnosis of Monticolo- mys was broadly stated with reference to all seven other nesomyine genera. Those diagnostic char- acters, as well as others mentioned below, and the rationale for their polarity are elaborated by Hershkovitz (1962), Bugge (1970), Carleton (1980, 1994), Voss (1988, 1993), and Carleton and Musser (1989). Relationships The extant species of Neosomyinae have long been arranged in only seven genera, each one highly distinctive and all remarkable for the ex- treme scope of intergeneric morphological diver- gence (Ellerman, 1941; Petter, 1972; Carleton & Musser, 1984). The striking variety of Madagas- car's native rodents is attested by the various com- parisons drawn, albeit not always fittingly, to their outward appearances — the genera being likened to rabbits (Hypogeomys), gerbils (Macrotarso- mys), voles (Brachyuromys), Old World murines (Gymnuromys), or generalized cricetines (Neso- mys). Just as remarkably, the seven genera were thought, until recently, to contain among them just ten species (Petter, 1975; Honacki et al., 1982; Rakotondravony, 1987; Corbet & Hill, 1991). However, the portrayal of Madagascar's endemic rodents as depauperate is, at least in part, a spe- cious conclusion based on inadequate biological survey and uncritical taxonomic study (see Carle- ton & Schmidt, 1990). The number of species ac- knowledged based on recent taxonomic revisions totals 16 (Musser & Carleton, 1993; Carleton, 1994), and now 17, with the diagnosis of Monti- colomys koopmani. Macrotarsomys, named by Milne Edwards and Grandidier (1898) almost a century ago, was the last nesomyine genus here- tofore described. The existence of a few strongly differentiated genera (seven) embracing nearly as few species (ten), as the Malagasy rodents were so long un- derstood, certainly influenced Ellerman 's (1941, 1949) systematic treatment of the Nesomyinae. Indeed, Ellerman dismissed the evolutionary re- ality of the subfamily and redistributed the genera among five family-group taxa, several of them de- scribed as new within an inclusive Family Muri- dae: Arvicolinae, Brachytarsomyes (Brachytar- somys); Cricetinae (Hypogeomys, Macrotarsomys, and Nesomys); Gymnuromyinae (Gymnuromys); Murinae, Eliuri (Eliurus); and Tachyoryctinae, Brachyuromyes (Brachyuromys). His taxonomy controverted the conventional viewpoint on ne- somyines — that is, a monophyletic group arising from a single immigrant stock and subsequently radiating in isolation (e.g., as classically narrated by Simpson, 1945) — but in doing so, he did cite characters and provide diagnoses for his reclas- sifications. Ellerman 's (1941, 1949) ideas have yet to be seriously challenged as an alternative to the traditional arrangement. The characterization of Monticolomys, howev- CARLETON & GOODMAN: ENDEMIC RODENTS 249 er, departs from the prevalent pattern of extreme morphological differentiation and obscure kinship alliances so far apparent among nesomyine gen- era. That is to say, several cranial and dental char- acters clearly implicate a sister-group relationship between the new genus and a previously de- scribed form, namely Macrotarsomys, suggesting that the two share a more recent common ancestor relative to the cladogenesis of the six other ne- somyine genera. Possible synapomorphies include reduction of the tegmen tympani and subsequent loss of contact with the squamosal, loss of the alisphenoid struts, marginal definition of the pos- terolophs, absence of mesolophs(ids), and a com- parable degree of reduction of the upper and low- er third molars. To be sure, many more resemblances of Mon- ticolomys and Macrotarsomys involve traits that are plausibly considered as primitive or those whose evolutionary polarity is equivocal. Among these are: size and extent of the incisive foramina; ovate infraorbital canal; forward position and dor- solateral orientation of the zygomatic plate; short, simple palate unmarked by corrugations or pos- terolateral palatal pits; wide, shallow paraptery- goid fossae; mesopterygoid fossa comparatively narrow, perforated by spacious sphenopalatine va- cuities; postglenoid and subsquamosal foramina present, defining a slender hamular process; smooth braincase without beading and ridging; a primitive orbitofacial blood supply (presence of vascular grooves across squamosal-alisphenoid and pterygoid; sphenofrontal and stapedial foram- ina patent); and development of a capsular pro- cess on the dentary. Dental similarities between the two genera are equally finely molded: tooth- rows anteriorly divergent; low molar crowns and bunodont form of their cusps; cusp position and retention of interconnecting, low-relief mures(ids); undivided anterocone; indistinct formation of an anteroconid and the virtual absence of a metaflex- id contrasted to the deep protoflexid; and the num- ber of roots on the molars. Considered by themselves, the few derived character states may seem insufficient to posit sis- ter-group stature between Monticolomys and Mac- rotarsomys. Nor, we acknowledge, are most or any of these character-state transformations, as ar- ticulated, unique within Muroidea. At the same time, and notwithstanding the weakness of infer- ring kinship on the basis of shared plesiomor- phies, one cannot help but be impressed by the utter exactitude and number of their presumably ancestral resemblances. In weighing the five syn- apomorphic features against the otherwise fun- damental likeness of the genera in so many as- pects of their craniodental morphology, we be- lieve that a hypothesis of cognate affinity warrants attention at this formative stage of phylogenetic understanding among nesomyines (Fig. 21-10). Sharply countershaded and bright pelage col- ors, silky fur texture, large pinnae, elongate pes, penicillate tail, and moderately inflated bullae are some of the superficial resemblances that have earned Macrotarsomys the sobriquet of the Mad- agascar gerbil (for instance, Webb, 1954). Among nesomyines, both M. bastardi Milne Edwards and Grandidier (1898) and M. ingens Petter (1959) possess these characteristics and other derived features that suggest their close kinship and vin- dicate their union in the genus Macrotarsomys (Fig. 21-10). The muted pelage tones with agouti-banded ventral hairs, broader hind foot and relatively long lateral digits, well-developed ungual tufts, and conformation of the plantar pads in Monitcolomys suggest the physique of a scansorial forest-dwell- er. Such characteristics provide substantial exter- nal contrast to species of Macrotarsomys, in ad- dition to cranial differences in the anterior xygo- matic region and auditory bullae (Fig. 21-10). Ac- cordingly, the diagnostic traits we have advanced for Monticolomys are unique in combination and make its recognition unambiguous. Nevertheless, among the characters surveyed, the identification of wholly autapomorphic features that define Monticolomys is less clear-cut. The apparent lack of uniquely derived character states poses the question of whether the construc- tion of Monticolomys is paraphyletic with regard to Macrotarsomys. Other than acknowledging the issue, a firm answer cannot be immediately or simply mustered without undertaking a detailed phylogenetic review of Nesomyinae. We are per- suaded, nonetheless, that generic isolation of Monticolomys koopmani is both a reasonable tax- onomic hypothesis at this stage of investigation and a defensible one, as based on the number and kinds of morphological differences so far uncov- ered that set it apart from Macrotarsomys (Fig. 21-10). In this regard, the amount of anagenetic change that distinguishes M. koopmani from spe- cies of Macrotarsomys far surpasses that observed within other genera (such as Eliurus and Neso- mys) that contain different species inhabiting dry western versus humid eastern biomes of Mada- gascar (Carleton, 1994, and unpubl. data). Mor- phologically and ecologically, Monticolomys ex- 250 FIELDIANA: ZOOLOGY M°n tic o\° Macrotarsomys i(i9 e"1' 1. Large, pliant pinnae 2. Tail moderately penicillate 3. Dorsal-ventral pelage contrast 4. Ungual tufts reduced 5. Plantar pads papillate 6. Hairs on plantar surface 7. Lateral pedal digits short 8. Metatarsals long, appressed 9. Zygomatic plate broad, notch deep 10. Lacrimal tabular process large 11. Incipient supraorbital shelf 12. Sigmoid notch compressed 13. Ectotympanic bullae inflated Middle lacerate for. obscured Tegmen tympani reduced Loss of alisphenoid strut Mesolophs absent 18. Posteroloph vestigial 19. Reduction of 3rd molars Fig. 21-10. Distribution of presumptive synapomorphies (solid rectangles) and hypothesized cladistic relationship between Monticolomys koopmoni and species of Macrotarsomys. Half-filled rectangles denote intermediate character states. hibits many of the characteristics that one might postulate for an ancestor to Macrotarsomys. The generic validity of Monticolomys and its proposed relationship to Macrotarsomys should be subject- ed to expanded character analyses and cladistic scrutiny, drawing upon representatives of all ne- somyine genera as well as other intelligently se- lected outgroups within Muroidea, and assimilat- ing information from additional anatomical sys- tems, karyology, and genetics. Whatever the outcome of such future phylo- genetic investigation, characteristics of the Mio- cene form Protarsomys, as reviewed herein, sub- vert neither the definition of Monticolomys nor the CARLETON & GOODMAN: ENDEMIC RODENTS 251 recognition of Macrotarsomys as a genus apart. Many resemblances between Protarsomys and Macrotarsomys that so impressed Lavocat (1973) are aspects of shape and form, features that do not simply lend themselves to formulation of char- acter states and argument of polarity. Other gen- eral likenesses — for instance, a brachyodont den- tition, presence of mures(ids), and a rudimentary anteroconid — are plausibly interpreted as shared primitive conditions that do not necessarily pro- vide support for their close relationship. More- over, the strict anatomical correspondence of sim- ilarities so generally stated is less persuasive when examined in detail (such as anteroconid formation and cuspate topography of the brachyodont crown). When comparison is restricted to quali- tative characters, one is struck more by the many and ample dissimilarities: configuration of the in- frarorbital canal; size, position, and orientation of the zygomatic plate; inflation of the auditory bul- lae; formation of the anteroconid and definition of the metafiexid; occurrence of mesolophs and me- solophids, posterolophs and posteroflexi; and de- gree of reduction of the third molars. The union of the fossil species Protarsomys macinnesi La- vocat (1973) with recent Macrotarsomys species cannot be construed on such profound differences and the want of any contravening synapomorphic traits. Lavocat (1973, 1978), it should be underscored, did not himself advocate the synonymy of Pro- tarsomys under Macrotarsomys. That action was promulgated by Chaline et al. (1977) and reiter- ated by Petter (1990), without discussion of char- acter data justifying the new combination. Instead, Lavocat (1973, 1978) drew attention to their like- nesses and proposed the descent of Nesomyinae from a Miocene, eastern African form such as Protarsomys. He retained the latter as a genus of Afrocricetodontinae, Family Cricetodontidae, set apart from a broadly defined Family Nesomyidae, in which he included Nesomyinae and several other archaic African groups (see comments in Carleton & Musser, 1984). In arguing against the generic equivalency of Protarsomys and Macro- tarsomys, we do not necessarily discount Lavo- cat's (1973, 1978) basic thesis with respect to the taxonomic and geographical origin of nesomyi- nes, yet neither can we, at this stage, identify per- suasive character information that would support it. Like other early Miocene muroids, Protarso- mys displays a somewhat generalized morpholo- gy, but from such evidence, one can argue no more strongly for its phyletic association with ne- somyines than with another African group such as the petromyscines. Continued exploration of Ter- tiary fossil sites in eastern Africa may disclose new cricetodontids whose relationship to neso- myines is incontestable. The discovery of relevant Miocene-Pliocene fossils in Madagascar remains a possibility. Fresh examination of the original type series of Protarsomys could feasibly reveal new characters that would bear critically on the question of its relationship to nesomyines. The problem is addressable and amenable to improved resolution. Biogeography In a phytogeographic classification of the floral associations of Madagascar, Humbert (1955) rec- ognized a unique High Mountain Domain that oc- curs patchily on a few isolated peaks between An- dohohela in the south and Tsaratanana in the north. Two of the mountaintop communities fall- ing within this domain are found on the massifs of Ankaratra and Andringitra. Upper montane for- ests (above approximately 1500 m) on these two massifs, particularly in the zone just below the tree line, share many endemic sclerophyllous plants, principally members of the family Erica- ceae (Koechlin, 1972; Koechlin et al., 1974). Fur- thermore, the fragmented distributions of numer- ous animal species — including, for example, ter- restrial insects (Paulian, 1961), aquatic insects (Chapter 9, this work), and several amphibians and reptiles (Chapter 17) — are restricted to the upper reaches of these central and south-central highlands. So far as is currently known, the dis- tributional pattern of the nesomyine rodent Mon- ticolomys koopmani also conforms to Humbert's (1955) High Mountain Domain (Fig. 21-11). The present-day discontinuity in the ranges of all these organisms and the reiterative pattern of vicariance strongly indicate that a corridor of up- per montane vegetation once connected the sum- mits of Ankaratra and Andringitra. On the basis of modern topography, feasible links between these two massifs may have occurred along the Fianarantsoa Plateau, between 1000 and 1300 m, and through the Ambalavao Pass, between 800 and 1000 m (Raxworthy & Nussbaum, in press). Thus, depression of upper montane zones to as low as 1000 m would have been necessary in or- der for animals such as Monticolomys to become 252 FIELDIANA: ZOOLOGY Fig. 21-11. Southeastern sector of Madagascar's Central High Plateau with reference to the two known geographic occurrences of Monticolomys koopmani: 1) Manjakatompo, about 19°20'S 47°26'E; 2) 38 km S Ambalavao, RNI d'Andringitra, 22°1 1'S 46°58'E. Stippled areas indicate uplands above 1500 m; this elevation approximates the pres- ent-day lower boundary of upper montane vegetation. The dot-and-dashed line corresponds to the 1000-m contour, the hypothesized lowest depression of the upper montane-sclerophyllous zone during the Holocene (see the Discussion section). distributed continuously over these mountains (Fig. 21-11). Evidence for cooler periods and consequent lowering of vegetation zones is available for the Quaternary, when Madagascar experienced cycli- cal climatic shifts (Burney, in press). Pollen cores from Lake Tritrivakely, near Antsirabe (80 km S Ankaratra and 250 km N Andringitra) on the Cen- tral Plateau (Fig. 21-11), reveal considerable vari- ation in the natural vegetational communities of the area over the past 36,000 years (Burney, 1987; Gasse et al., 1994). During both the Pleistocene CARLETON & GOODMAN: ENDEMIC RODENTS 253 and early Holocene, ericaceous pollen generally accounted for over 50% of the taxa deposited in the lacustrine sediments, but after the middle Ho- locene this proportion dropped to between 0 and 30%. During portions of the Quaternary, high mountain floral communities likely descended to approximately 1000 m (Burney, in press), an el- evation sufficiently low for upper montane forest and sclerophyllous vegetation to have extended between the regions of Ankaratra and Andringitra. Some time in the middle Holocene, the upper montane and sclerophyllous floras retreated to the uppermost slopes of the mountains and severed the geographic ranges of animals dependent on such habitats. In summary, information derived from paly- nology and from the congruent distributions of other plants and animals convincingly explains the present-day vicariant occurrence of Montico- lomys koopmani on the Andringitra and Ankaratra massifs. Specimen-based documentation of neso- myine distributions, in general poor over much of Madagascar, is almost nonexistent for the island's high mountain zones, particularly those on such huge mountain systems as Tsaratanana and Ma- rojejy in the north and Andohohela in the south. Whether the distributions of other nesomyine ro- dents repeat a similar geographic pattern is there- fore uncertain, but Eliurus majori, as its broken range is currently understood, is a possible can- didate (see Carleton, 1994; and Chapter 22, this work). Note added in proof Monticolomys koopmani still occurs on the An- karatra Massif. During a biological inventory of the Foret de Nosiarivo (2000 m) in February 1996, conducted by S. M. Goodman and D. Rak- otondravony, a specimen of Monticolomys koop- mani was captured (FMNH 156211). The animal was trapped on the ground and in degraded habitat close to the forest edge. Acknowledgments We thank Louis L. Jacobs, Southern Methodist University, for loaning the casts of Protarsomys from his personal research collection. Access to comparative material and/or holotypes of Macro- tarsomys was provided by Guy G. Musser (AMNH) and Francis Petter and Michel Tranier (MNHN). The morphological descriptions and comparisons related in our paper have benefited substantially from the hind foot drawings and cra- niodental photographs carefully prepared by Da- vid F Schmidt. Peter Goldberg photographed the museum skins. We are grateful to Guy G. Musser and Bruce D. Patterson for their critical comments on the penultimate version of the manuscript. Last, the first author acknowledges his deep grat- itude to Karl F. Koopman and Guy G. Musser, AMNH, for their unselfish encouragement to pur- sue the description of Rand's Manjakatompo specimen, particularly because each had already devoted sufficient time and effort to the problem to appreciate its systematic contribution. Literature Cited Archbold, R. 1932. A new lemur from Madagascar. American Museum Novitates, 518: 1. Bugge, J. 1970. The contribution of the stapedial artery to the cephalic arterial supply in muroid rodents. Acta Anatomica, 76: 313-336. Burney, D. A. 1987. Pre-settlement vegetation changes at Lake Tritrivakely, Madagascar. Palaeoecology of Africa, 18: 357-381. -. In press. Theories and facts regarding Holo- cene environmental change before and after human colonization. In Goodman, S. M., and B. D. Patterson, eds., Natural and Human-Induced Change in Mada- gascar. Smithsonian Institution Press, Washington, DC. Carleton, M. D. 1980. Phylogenetic relationships in neotomine-peromyscine rodents (Muroidea) and a re- appraisal of the dichotomy within New World Cricet- inae. Miscellaneous Publications of the Museum of Zoology, University of Michigan, 157: 1-146. . 1994. Systematic studies of Madagascar's en- demic rodents (Muroidea: Nesomyinae): Revision of the genus Eliurus. American Museum Novitates, 3087: 1-55. Carleton, M. D., and G. G. Musser. 1984. Muroid rodents, pp. 289-379. In Anderson, S., and J. K. Jones, Jr., eds., Orders and Families of Recent Mam- mals of the World. John Wiley and Sons, New York, 686 pp. . 1989. Systematic studies of oryzomyine ro- dents (Muridae, Sigmodontinae): A synopsis of Mi- croryzomys. Bulletin of the American Museum of Nat- ural History, 191: 1-83. Carleton, M. D., and D. F. Schmidt. 1990. Systematic studies of Madagascar's endemic rodents (Muroidea: Nesomyinae): An annotated gazetteer of collecting lo- calities of known forms. American Museum Novita- tes, 2987: 1-36. Ch aline, J., P. Mein, and F. Petter. 1977. Les grandes lignes d'une classification evolutive des Muroidea. Mammalia, 41: 245-252. 254 FIELDIANA: ZOOLOGY Corbet, G. B., and J. E. Hill. 1991. A World List of Mammalian Species, 3rd edition. Natural History Mu- seum Publications, London, vii + 243 pp. Delacour, M. J. 1932. Les oiseaux de la Mission Zoo- logique Franco-Anglo-Americaine a Madagascar. L'Oiseau et la Revue de Francaise d'Ornithologie, nouvelle serie, 2: 1-96. Denys, C. 1987. Rodentia and Lagomorpha, pp. 118- 170. In Leakey, M. D., and J. H. Harris, eds.. Laetoli, a Pliocene site in Tanzania. University of Oxford Press, London. Ellerman, J. R. 1941. The Families and Genera of Liv- ing Rodents. Volume 2, Family Muridae. London: British Museum (Natural History), xii+690 pp. . 1949. The Families and Genera of Living Ro- dents. Volume 3, Appendix II [Notes on the rodents from Madagascar in the British Museum, and on a collection from the island obtained by M. C. S. Webb]. London: British Museum (Natural History), v+210 pp. Gasse, E, E. Cortijo, J.-R. Disnar, L. Ferry, E. Gil- bert, C. Kissel, F. Laggoun-Defarge, E. Lallier- Verges, J.-C. Miskovsky, B. Ratsimbazafy, F. Ran- AIVO, L. ROBISON, P. TUCHOLKA, J.-L. SAOS, A. SlFFED- ine, M. Taieb, E. Van Campo, and D. Williamson. 1994. A 36 ka environmental record in the southern tropics: Lake Tritrivakely (Madagascar). Comptes ren- dus de l'Academie des Sciences, seYie II, 318: 15 13— 1519. Hershkovitz, P. 1962. Evolution of Neotropical crice- tine rodents (Muridae) with special reference to the phyllotine group. Fieldiana, Zoology, 46: 1-524. Honacki, J. H., K. E. Kinman, and J. W Koeppl. 1982. Mammal Species of the World. Allen Press and the Association of Systematics Collections, Kansas, ix+694 pp. Humbert, H. 1955. Les temtores phytogeographiques de Madagascar. Leur cartogarphie. Annee Biologique, serie III, 31(5-6): 195-204. Jacobs, L. L. 1985. The beginning of the age of murids in Africa. Acta Zoologica Fennica, 170: 149-151. Koechlin, J. 1972. Flora and vegetation of Madagascar, pp. 145-190. In Battistini, R., and G. Richard- Vin- dard, eds., Biogeography and Ecology in Madagascar. W. Junk, The Hague, xv + 765 pp. Koechlin, J., J.-L. Guillaumet, and P. Morat. 1974. Flore et veg&ation de Madagascar. J. Kramer, Vaduz, Germany. Lavocat, R. 1973. Les rongeurs du Miocene d'Afrique Orientale. I. Miocene infeYieur. M6moires et Travaux de l'Ecole Pratique des Hautes Etudes, Institute de Montpellier, 1: 1-284. . 1978. Rodentia and Lagomorpha, pp. 69-89. In Maglio, J. V, and H. B. S. Cooke, eds.. Evolution of African Mammals. Harvard University Press, Cam- bridge, ix+641 pp. Milne-Edwards, A., and G. Grandidier. 1898. De- scription d'une espece nouvelle de Murid6 provenant de Madagascar. Bulletin de Museum National d'His- torie Naturelle, serie 1, 4: 179-181. Musser, G. G., and M. D. Carleton. 1993. Family Muridae. pp. 501-755. In Wilson, D. E., and D. M. Reeder, eds.. Mammal Species of the World: A Tax- onomic and Geographic Reference, 2nd edition. Smithsonian Institution Press, Washington, D.C., xviii+1207 pp. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la conservation et des Aires Protegees. World Wide Fund for Nature, Gland, Switzerland, xvii + 374 pp. Paulian, R. 1961. La zoogeographie de Madagascar et des iles voisines. Faune de Madagascar, 13: 1-484. Perrier de la Bathie, H. 1927. Le Tsaratanana, l'An- karatra, et l'Andringitra. M6moire de l'Academie Malgache. 3: 1-68. Peters, W 1870. Uber Nesomys rufus, eine neue gat- tung und art madagascarischer nager. Sitzungs-Beri- chte der Gesellschaft naturforsender Freunde, Berlin, 1870: 54-55. Petter, F. 1959. Un nouveau rongeur de Madagascar (Nesomyinae): Macrotarsomys ingens nov. sp. Mam- malia, 23: 139-148. . 1972. The rodents of Madagascar: The seven genera of Malagasy rodents, pp. 661-665. In Battis- tini, R., and G. Richard- Vindard, eds., Biogeography and Ecology in Madagascar. W. Junk, The Hague, xv+765 pp. 1975. Family Cricetidae: Subfamily Nesomyi- nae. Part 6.2, pp. 1-4. In Meester, J., and H. W Setzer, eds.. The Mammals of Africa: An Identification Man- ual. Smithsonian Institution Press, Washington, D.C. 1990. Relations de parente" des rongeurs de Madagascar. Accademie Nazionale dei Lincei, Atti dei Convegni Lincei, 85: 829-837. Rakotondravony, D. A. 1987. Les rongeurs a Mada- gascar, pp. 93-94. In Mittermeier, R. A., L. H. Rak- otovao, V. Randrianasolo, E. J. Sterling, and D. De- vitre, eds., Priorit^s en matiere de conservation des especes a Madagascar. IUCN, Gland, Switzerland, 167 pp. Rand, A. L. 1932. Mission Franco-Anglo- Am^ricaine a Madagascar. Notes de voyage. L'Oiseau et la Revue de Francaise d'Ornithologie, nouvelle serie, 2: 227- 282. 1935. On the habits of some Madagascar mam- mals. Journal of Mammalogy, 16: 89-104. 1936. The distribution and habits of Madagas- car birds. A summary of the field notes of the Mission Zoologique Franco-Anglo-Am6ricaine a Madagascar. Bulletin of the American Museum of Natural History, 72: 143-499. Raxworthy, C. J., and R. A. Nussbaum. In press. Zoo- geographic patterns of reptile endemicity and vicari- ance in eastern Madagascar. In Goodman, S. M., and B. D. Patterson, eds.. Natural and Human-Induced Change in Madagascar. Smithsonian Institution Press, Washington, D.C. Reig, O. A. 1977. A proposed unified nomenclature for the enamelled components of the molar teeth of the Cricetidae (Rodentia). Journal of Zoology, London, 181: 227-241. Simpson, G. G. 1945. The principles of classification and a classification of mammals. Bulletin of the American Museum of Natural History, 85: 1-350. United States Board on Geographic Names. 1989. CARLETON & GOODMAN: ENDEMIC RODENTS 255 Gazetteer of Madagascar, 2nd edition. Defense Map- ine" characters. American Museum Novitates, 3073: ping Agency, Washington, D.C., xvii+826 pp. 1-44. Voss, R. S. 1988. Systematics and ecology of ichthy- Voss',R: S-.AND L'.J Marcus 1992. Morphological omyine rodents (Muroidea): Patterns of morphological ^volutlon in muroid S?*™* "/ Cnmiometnc factor \ .. „ .. ., n „*\. %„. divergence in seven Neotropical genera, with expen- ^^i^s^l^^^!a^-^^QofA^ mental results from Zygodontomys. Evolution, 46: American Museum of Natural History, 188: 259-493. 1918-1934 1993. A revision of the Brazilian muroid ro- Webb, C. S. 1954. The odyssey of an animal collector. dent genus Delomys with remarks on "thomasomy- Longmans, London, xv+368 pp. 256 FIELDIANA: ZOOLOGY Chapter 22 The Rodents of the Reserve Naturelle Integrate d'Andringitra, Madagascar Steven M. Goodman and Michael D. Carleton Abstract Rodents were surveyed on the previously unstudied eastern side of the Reserve Naturelle Intdgrale (RNI) d'Andringitra, between 720 and 1625 m elevation. Ten species were trapped, and one more is known from earlier collections made in the reserve. One introduced rodent, Rattus rattus, has colonized the reserve, and the balance of species belong to the endemic Subfamily Nesomyinae. We present information on identification and distribution, measure- ments, and ecology and reproduction for each species. Trap results showed the greatest rodent species diversity and numbers to be at 1625 m. Two species, Eliurus minor and Gymnuromys roberti, occur across the complete elevational range sampled. Nesomys audeberti and Eliurus webbi occur in forests below 810 and Brachyuromys ramirohitra and Monticolomys koopmani in forests above 1625 m. The balance of species have relatively broad altitudinal ranges. No rodent taxon is known to be endemic to the RNI d'Andringitra, although the recently discovered Monticolomys koopmani is apparently restricted to the upper portions of the Andringitra and Ankaratra massifs. Resume Dans la partie est de la Reserve Naturelle Integrate (RNI) d'Andringitra, ou aucun inventaire n'avait encore 6\€ effectue\ on a r6alis6 l'inventaire des rongeurs entre 720 m et 1625 m d' altitude. Dix especes ont 6t6 capturees et une supptementaire 6tait deja connue de collectes pr6c6dentes effectuees dans la reserve. Rattus rattus, espece introduite, a colonist la reserve et le reste des especes appartient a la sous-famille end^mique des Nesomyinae. Nous pr^sentons des informations sur l'identification, la distribution, les mensurations, 1'ecologie et la repro- duction de chaque espece. La plus grande diversity sp6cifique et le plus grand nombre d'in- dividus ont €t€ documented a 1625 m d'altitude a 1'aide de pieges. Deux especes, Eliurus minor et Gymnuromys roberti, apparaissent sur 1'ensemble de la ligne altitudinale inventorize. On a recensd Nesomys audeberti et Eliurus webbi dans les forets localises en-dessous de 810 m d'altitude et Brachyuromys ramirohitra et Monticolomys koopmani dans les forets localisees au dessus de 1625 m. Le reste des especes a une distribution altitudinale relativement large. Aucun taxon de rongeurs n'est end6mique de la RNI d'Andringitra, mais il faut noter la r^cente d6couverte de Monticolomys koopmani, qui est apparemment limitee aux parties sup^rieures des massifs de 1' Andringitra et 1' Ankaratra. GOODMAN & CARLETON: RODENTS 257 Introduction The native rodents of Madagascar, as currently understood, comprise a group of 15 species that are placed within the endemic Subfamily Neso- myinae of a broadly defined family Muridae (Carleton & Musser, 1984; Musser & Carleton, 1993). Nesomyinae contains a remarkable array of morphological types yet embraces relatively few species, and, if indeed monophyletic (see dis- cussions in Petter, 1972, 1990, and Carleton & Musser, 1984), the subfamily represents one of the more unusual radiations compared to other mu- roid groups that have diversified within an insular setting (see Carleton & Schmidt, 1990). Species diversity, relationships, and geographic limits re- main poorly known (Carleton & Schmidt, 1990), and revisionary systematic studies have recently begun (Carleton, 1994). Petter (1972) and Rako- tondravony (1987) offered general reviews of the rodents of Madagascar, and Petter (1975) provid- ed keys to the genera and species then recognized. Although the endemic carnivores and primates of Madagascar have long inspired attention from biologists, interest in its indigenous rodents has emerged only in the past decade. This interest is manifested in general faunal inventories (Duck- worth & Rakotondraparany, 1990; Barden et al., 1991; Goodman & Ganzhorn, 1994), elevational transects (Ryan et al., unpubl.), review of Holo- cene subfossils (Goodman & Rakotondravony, in press), and ecological or behavioral studies (Ni- coll et al., 1988; Ryan et al., 1993; Stephenson, 1993). Still, this surge of activity has done little to advance basic information about the ecology, reproductive biology, and populations of the is- land's rodents, even for those forms inhabiting more frequently visited areas of the eastern humid forest. The present study documents nesomyine rodent diversity in the Reserve Naturelle Integrate (RNI) d'Andringitra, examines species richness and numbers along an elevational gradient on the east- ern slopes of the reserve, and summarizes basic natural history information on these animals. The broad range of plants and animals studied during this survey affords a unique opportunity to ex- amine the biotic interactions between rodents and other organisms along the slopes of the Andrin- gitra Massif (see Chapter 6). Previous Work Although there were earlier biological surveys, little has been published on the rodents of the RNI d'Andringitra per se. From November 1970 to January 1971, a group of scientists visited the An- dringitra Massif to study the local geomorpholo- gy, climate, and vegetational structure (Paulian et al., 1971). Participants in this campaign, under the name of Recherche Cooperative sur Programme (RCP) no. 225, included the mammalogist Roland Albignac, then of the ORSTOM laboratory in An- tananarivo, and Andre Peyrieras, who assisted in the collection of small mammals (see pp. 4-5 for a listing of the RCP no. 225 itinerary and locali- ties). The expedition members concentrated their efforts on the higher elevational zones, and all small mammals collected were deposited in the Museum National d'Histoire Naturelle (MNHN), Paris. A complete analysis of the collection of Albig- nac and Peyrieras has never been published in its entirety, but selected specimens and related com- munications have been cited in several works (Petter, 1975; MacPhee, 1987; Carleton & Schmidt, 1990; Carleton, 1994). Nicoll and Lan- grand (1989) also compiled the known mammal fauna of the RNI d'Andringitra based on review of the literature and their own unpublished rec- ords. They listed four rodent species, all of them nesomyines: Brachyuromys betsileoensis, B. ra- mirohitra, Gymnuromys roberti, and Eliurus "myoxinus" (probably E. majori, according to Carleton, 1994). In 1993, a group of Cambridge University stu- dents conducted a small mammal survey on the north-central slopes of the reserve in the Foret Imaitso and near Anjavidilava (O'Keeffe & Ash- more, no date). They listed four rodent species from the reserve: two nesomyines (Gymnuromys roberti and Eliurus "myoxinus") and two mu- rines (Rattus rattus and R. norvegicus). Materials and Methods This investigation is based on fieldwork con- ducted between November 14 and December 18, 1993 by the senior author (S.M.G.), who is re- sponsible for general synthesis of the systematic and ecological data. The junior author (M.D.C.) assisted with taxonomic determinations and sys- tematic comparisons. 258 FIELDIANA: ZOOLOGY Table 22-1. Summary of trap lines. Elevation No. of traps Length of line (m) Mean distance between traps (m) Mean height above ground (m) 720 m (November 15-21) Line 1 50 Line 2 50 Total 100 280 340 620 5.5 ± 3.68 (0-19) 7.2 ±5.51 (0-29) 6.4 ± 4.79 (0-29) 1.6 ±0.69 (0.50-3.0), N = 20 1.3 ± 0.69 (0.25-2.0), N = 17 1.5 ± 0.73 (0.25-3.0), N = 37 810 m (November 22-29) Line 3 50 Line 4 50 Total 100 370 410 780 7.6 ± 5.67 (0-24) 8.4 ±7.03 (1-32) 8.0 ± 6.43 (0-32) 1.4 ±0.78 (0.2-3.0), N = 25 1.5 ± 0.63 (0.5-2.5), N = 23 1.4 ±0.72 (0.2-3.0), N = 48 1210 m (December 1-7) Line 5 50 Line 6 50 Total 100 340 375 715 7.0 ± 4.47 (2-19) 7.7 ±5.28 (1-26) 7.4 ±4.94 (1-26) 1.8 ±0.70 (0.75-3.0), N = 21 2.1 ±0.93(1.0-4.0), N = 21 1.9 ±0.85 (0.75-4.0), N = 42 1625 m (December 8-15) Line 7 50 Line 8 50 Total 100 315 275 590 6.4 ±3.80 (1-18) 5.6 ±4.45 (1-22) 6.0 ±4.18 (1-22) 1.7 ±0.62 (0.5-3.0), N = 15 2.6 ±2.99 (1.0-15), N = 20 2.2 ±2.37 (0.5-15), N = 35 Note: Descriptive statistics are presented as mean ± SD (range). Each line consisted of 10 National and 40 Sherman live-traps (see page 259). Field Methods and Trapping Protocol At each of four elevational levels (720, 810, 1210, and 1625 m), two separate trap lines were in operation for a minimum of 5 nights (Table 22-1). Each trap line, numbered sequentially start- ing with the 720 m zone, consisted of 50 stations: 40 Sherman live traps (9 X 3.5 X 3 in.) and 10 National live traps (16 X 5 X 5 in.). Traps were baited daily, generally between 1500 and 1700 hours, with a fresh mixture of finely ground pea- nut butter and oat grain mixed in proportions to make a paste. Pitfall traps were installed in each elevational zone (see Chapter 17); their use re- sulted in the capture of numerous insectivores (see Chapters 19 and 20), but few rodents. Traps were visited at least twice per day, once at dawn and in the late afternoon. A "trap-day" is defined as one trap in use for a 24-hour period (dawn to dawn). Captured animals were prepared as either standard museum skins with associated skulls and skeletons, fluid-preserved carcasses, or full skeletons. Many tissue samples were frozen in liquid nitrogen for biochemical studies and the viscera preserved in alcohol for endoparasite re- search. Whole carcasses were wrapped in a fine cheesecloth, before immersion in formalin, to pre- vent loss or mixing of ectoparasites (see Chapter 12). In order to quantify differences in the spatial distribution of small mammal captures, several trapping variables were systematically recorded: (1) type of trap; (2) total length of trap line; (3) distance between traps; and (4) specific placement of each trap, including its substrate, surrounding forest structure, and position on or height above the ground. The precise location of each trap (variable 4) was categorized by microhabitat site, whether trap placement was on or above the ground (angles are relative to the ground), as follows: On Ground — 1', in leaf litter; 2', in leaf litter by hole; 3', under downed tree (not decomposed); 4', under downed tree (at least partially decom- posed); 5', by tree root; 6', by tree root with cav- ity or hole; 7', by at least partially decomposed tree trunk or root with hole or cavity; 8', on ground (not fitting other categories); 9', stream edge (at or less than 0.5 m from edge). Above Ground — 1", on liana, limb, or trunk less than 5 cm circumference in horizontal to 15° position; 2", on liana, limb, or trunk between 5 and 10 cm circumference in horizontal to 15° po- sition; 3", on liana, limb, or trunk greater than 10 cm circumference in horizontal to 15° position; 4", on liana, limb, or trunk less than 5 cm on circum- ference at 15° to 45° angle; 5", on liana, limb, or trunk between 5 and 10 cm circumference at 15° to 45° angle; 6", on liana, limb, or trunk greater GOODMAN & CARLETON: RODENTS 259 than 10 cm circumference at 15° to 45° angle; 7", on liana, limb, or trunk less than 5 cm circumfer- ence at 45° to 90° angle; 8", on liana, limb, or trunk between 5 and 10 cm circumference at 45° to 90° angle; 9", on liana, limb, or trunk greater than 10 cm circumference at 45° to 90° angle; 10", on downed tree (not decomposed); 11", on downed tree (at least partially decomposed); 12", on exposed rock; 13", in tree hole. Specimens and Measurements A total of 148 rodent specimens were preserved as vouchers during the field census of the RNI d'Andringitra conducted by S.M.G. This material is housed in the Field Museum of Natural History (FMNH), Chicago, and a representative series will be returned to the Departement de Biologie Ani- mate, Universite d' Antananarivo. To amplify specimen-based documentation for the RNI d'Andringitra, the rodents collected under the Re- cherche Cooperative sur Programme (RCP) and now stored in the MNHN, Paris are also listed. To confirm taxonomic identifications, nesomyine holdings in other museums (see Appendix) were also consulted, and one or both authors have ex- amined the holotypes of all described forms of Nesomyinae, except for Peters' (1870) Nesomys rufus. Six measurements, in millimeters (mm) or grams (g), were taken by S.M.G. for each speci- men in the flesh; their abbreviations and defini- tions are given below. TOTL, total length of body and tail: from the tip of the nose to the end of the last caudal vertebra (not including terminal hair tuft). HBL, head and body length: from the tip of the nose to the distalmost point of the body (at base of tail). TL, tail length: from the base of the tail (held at right angles to the body) to the end of the last caudal vertebra (not including terminal hair tuft). HFL, hind foot length: from the heel to the tip of the longest toe (not including claw). EL, ear length: from the basal notch to the dis- tal tip of the pinna. WT, weight in grams: measured with Pesola spring scales, to ± 0.5 g for animals less than 100 g and to ± 1.0 g for those between 101 and 300 g. Sixteen cranial and two dental dimensions were measured by M.D.C. to the nearest 0.1 mm using handheld digital calipers accurate to 0.03 mm. These measurements, and their abbreviations, fol- low the anatomical landmarks defined and illus- trated by Carleton (1994): BBC, breadth of the braincase; BIF, breadth of incisive foramina; BMls, breadth of the bony palate across the first upper molars; BOC, breadth across the occipital condyles; BR, breadth of rostrum; BZP, breadth of the zygomatic plate; DAB, depth of the audi- tory bulla; IOB, interorbital breadth; LBP, length of bony palate; LD, length of diastema; LIF, length of the incisive foramina; LM1-3, coronal length of maxillary toothrow; LR, length of ros- trum; ONL, occipitonasal length; PPB, posterior breadth of the bony palate; PPL, postpalatal length; WM1, width of the first upper molar; and ZB, zygomatic breadth. Standard descriptive statistics (mean, range, standard deviation) were derived for adult speci- mens in each species sample. We define "adult" as the age cohort consisting of animals that lack the finer, juvenile pelage and that possess fully erupted, though sometimes unworn, third molars. Where sample sizes permitted, two-sample t tests and one-way analyses of variance (ANOVA) were applied to the mensural variables, with gender as the categorical variable. For some taxa, patterns of morphometric differentiation were summarized using principal component analysis, extracted from the variance-covariance matrix and based on natural log transformations of the 1 8 craniodental variables. Analytical routines were carried out us- ing Systat (version 5.05, 1994). Mammae formu- lae are presented as the number of paired post- axial, abdominal, or inguinal teats. Accounts of Species Family Muridae: Subfamily Murinae Rattus rattus (Linnaeus, 1758) Identification — In general, all external mea- surements (Table 22-2) fall within the known range of Malagasy populations of Rattus rattus (Rakotondravony, 1992). No sexual dimorphism is discernible in our limited series, although other studies have found males to be larger than females in body measurements (e.g., King, 1990). Ecology and Reproduction — During the 1993 260 FIELDIANA: ZOOLOGY Table 22-2. External measurements and sample statistics of adult rodents from RNI d'Andringitra. Species N TOTL HBL TL HFL EL WT Rattus rattus 10 364.4 167.0 194.8 32.8 24.7 105.7 17.2 8.0 11.2 1.0 0.7 13.1 340-388 151-178 176-212 31-34 24-26 86-134 Brachyuromys betsileoensis 14 222.5 147.1 76.4 30.6 19.3 16.6 11.0 7.5 1.3 1.0 200-252 125-165 70-92 29-34 17-21 B. ramirohitra 3 252.0 155.0 95.3 32.3 22.7 90.5 222-282 140-165 84-110 30-34 21-24 64-117 Eliurus majori 23 353.6 157.9 190.4 28.8 19.6 97.2 15.6 6.6 10.4 1.1 0.7 12.9 315-376 145-171 170-207 27-31 18-21 77-122 E. minor 25 239.6 112.8 128.9 21.4 17.6 35.1 9.5 6.1 5.1 1.0 0.7 5.8 220-262 101-124 119-137 20-23 17-19 21.5-49.5 E. tanala 24 333.1 152.0 178.3 29.9 23.6 81.7 12.0 4.7 10.1 1.3 0.9 7.9 307-350 140-159 152-194 27-33 22-25 66-97.5 E. webbi 19 306.0 138.3 165.7 28.6 22.7 70.1 14.5 13.9 10.1 1.1 0.9 9.7 286-331 105-161 150-183 27-31 21-24 54-90 Gymnuromys roberti 4 335.0 161.5 168.0 34.7 22.3 107.7 310-353 145-168 155-176 34-36 21-23 73-124 Monticolomys koopmani 3 236.3 98.0 138.0 24.3 18.3 26.1 234-240 94-101 134-143 24-25 18-19 25-27 Nesomys audeberti 1 381 185 180 50 30 182 N. rufus 31 343.5 179.4 158.6 43.7 26.1 158.7 9.2 9.0 6.1 1.0 0.9 16.6 323-360 160-196 148-171 42-46 24-28 126-208 Note: Values are means ± SD (range). survey, this murine species, which was introduced to Madagascar, was captured only in forested zones between 810 and 1625 m. Three specimens (MNHN) were earlier obtained by R. Albignac during the RCP survey at Ambalamarovandana (1530 m), Anjavidilava (1995 m), and Ivango- mena (2400-2500 m). Thus, Rattus rattus has an elevational distribution across nearly the complete altitudinal range of the massif. This species was not trapped within the 720 m zone, which was near a small human settlement and partially disturbed forest. The number of Rat- tus rattus trapped increased as a function of alti- tude; there were no captures at 720 m, one at 810 m, four at 1210 m, and six at 1625 m (regression, P = 0.009, r = 0.99, N = 4). This positive cor- relation also reflects increased distances from the nearest cleared forest. Cranial remains of this rat were found in Cryptoprocta ferox scats collected in the heathland zone between 2000 and 2100 m elevation. Individuals of Rattus rattus were captured in a variety of trap placements. Seven of 11 (64%) Rattus were captured on the ground — in leaf litter, by downed dead wood, and by root tangles. An- imals taken from above the ground (36%) had been crossing lianas, limbs, or trunks of various sizes and at a variety of angles. Of the 13 R. rattus taken in 1993, nine were scrotal males, two were adult males with abdom- inal testes, one was a female with enlarged mam- mae, and one was a female with small mammae. Comments — The history of the black rat's col- onization of Madagascar, in particular the eastern humid forest, is unknown. On the basis of RCP specimens collected in 1970-1971, it is clear that Rattus rattus has been on the mountain for over 20 years. The western side of the Andringitra Massif has been settled and cleared for a consid- erable period and contains numerous agricultural areas, particularly rice paddies. The eastern side, however, is relatively intact and lacks dense hu- man population. It is conceivable that R. rattus colonized the eastern slopes from the western GOODMAN & CARLETON: RODENTS 261 side via the central crystalline ridge. This sup- position is supported by the observation that male R. rattus are known to disperse greater dis- tances than females (Spencer & Davis, 1950), and 85% of the animals collected in 1993 in the RNI d'Andringitra were males. This biased trap success is interpreted as an expanding population front. Further, at the time of the 1993 survey, R. rattus was not found in the 720 m zone, which was close to a small village. At other sites on the island R. rattus tends to be common in similar circumstances. The pattern of greater trap suc- cess with increasing elevation, combined with the other points outlined above, would support the idea that this species is invading the eastern portion of the massif from the central ridge. On other tropical islands where R. rattus has been introduced, considerable variation is appar- ent in its ability to colonize montane forest. For example, on some islands in the Philippines and Sulawesi, the species tends to be restricted to dis- turbed lowland forest or agricultural areas (Mus- ser, 1987; Rickart et al., 1993), whereas on others it occurs broadly across an ecological and altitu- dinal gradient from lowland to mossy forest (Hea- ney et al., 1989). What factors lead to such vari- ation in the ability of R. rattus to colonize differ- ent natural forests are not clear, but at least in the Philippines and Sulawesi, they may be related to the presence/absence of sympatric, forest-dwell- ing native congeners. On the basis of feeding trials with captive an- imals in the RNI d'Andringitra, Rattus rattus and nesomyine rodents feed on similar types of forest fruits and nuts (see Chapter 6). Thus, the potential for direct competition between these rodents ex- ists (Goodman, 1995). Whether this competition is sufficient for R. rattus to displace nesomyine rodents in the RNI d'Andringitra is unknown. No evidence of Rattus norvegicus was found within the reserve, but this species was reported to have been captured in 1993 at 1400 m in the Foret Imaitso, in the north-central portion of the reserve (O'Keeffe & Ashmore, no date). Specimens Examined — Anjavidilava (MNHN 1972.521); Ambalamarovandana (MNHN 1972.613); Ivangomena (MNHN 1972.614); 38 km S Ambalavao, ridge E of Volotsangana River, 1625 m (FMNH 151921 through 151924); 40 km S Ambalavao, along Volotsangana River, 1210 m (FMNH 151729 and 151918 through 151920); 43 km S Ambalavao, junction of Sahanivoraky and Sahavatoy rivers, 810 m (FMNH 151728, 151916, and 151917). Family Muridae: Subfamily Nesomyinae Brachyuromys Major, 1896a Recognition — A broad head and compact tor- so, small and rounded pinna, short legs and rela- tively narrow hind feet, and tail vertebrae con- spicuously shorter than the length of head-and- body (TL about 50-60% of HBL) identify mem- bers of this nesomyine genus (Table 22-2). The body fur is thick and fine, its texture soft and lax. The dorsum is colored a rich brown to reddish brown, highlighted by buff or ochraceous tips, and the venter is approximately the same in color and tone, without any lateral border defined be- tween them. The dark monocolored tail is evenly covered with short hairs, which do not lengthen perceptibly toward the tip. Plantar pads number six, with the thenar and hypothenar unreduced and set close to the interdigitals. The mammae count is also six: one pair each postaxial, abdominal, and inguinal. The skull appears stoutly constructed, its length short relative to width, particularly as observed across the braincase and lateral reaches of the zy- goma (Table 22-3). The narrowness of the inter- orbital constriction is accentuated by the heavy zygomatic arches that converge abruptly toward broad zygomatic plates. The rostrum is deep and short, the ectotympanic bullae moderately inflat- ed. The hypsodont molars possess planar occlusal surfaces, oval to cylindriform in shape. The pat- tern of each molar is configured as three broad lamina or enamel loops that are oriented obliquely to the longitudinal axis of the toothrow. Maxillary toothrows are convergent anteriorly, but the man- dibular rows are more or less parallel to one an- other. As remarked by each discoverer of the two know species (Bartlett, 1879; Major, 1896a), these rodents resemble members of the Holarctic sub- family Arvicolinae externally. Both species of Brachyuromys, B. betsileoensis and B. ramirohi- tra, occur in the RNI d'Andringitra. Brachyuromys betsileoensis (Bartlett, 1879) Identification and Distribution — Bartlett's (1879) B. betsileoensis is the noticeably smaller of the two species, as reflected in almost every univariate comparison of the skin and skull (Ta- bles 22-2 and 22-3). Although the cranium of B. betsileoensis is more diminutive than that of B. 262 FIELDIANA: ZOOLOGY Table 22-3. Comparison of selected craniodental measurements of adult Brachyuromys betsileoensis and B. ramirohitra from Andringitra and Ampitambe. B. betsileoensis B. ramirohitra Andringitra Ampitambe Andringitra Ampitambe Variable (N = 16) N = 10 (N = l) (N = 27) ONL 34.6 ± 0.8 33.6 ± 1.5 40.7 39.1 ± 1.3 33.5-36.0 31.5-36.2 36.0-41.5 ZB 20.8 ± 0.6 20.1 ± 1.0 23.0 22.9 ± 0.9 20.1-21.7 18.5-21.4 20.5-24.4 BBC 14.4 ± 0.2 13.9 ± 0.5 15.1 14.8 ± 0.4 14.1-14.6 13.3-14.7 14.1-15.7 IOB 4.9 ± 0.2 5.0 ± 0.2 5.4 5.1 ± 0.2 4.5-5.2 4.7-5.3 4.5-5.5 LR 10.5 ± 0.3 10.2 ± 0.5 13.6 13.0 ± 0.6 9.9-11.1 9.6-11.0 11.6-13.9 PPL 13.4 ± 0.6 13.0 ± 0.8 15.2 14.3 ± 0.8 13.1-14.7 12.0-14.3 12.4-16.0 LIF 7.2 ± 0.3 6.8 ± 0.5 7.9 7.1 ±0.4 6.5-7.7 5.9-7.5 6.1-7.7 LD 9.3 ± 0.7 8.9 ± 0.6 10.2 9.5 ± 0.7 8.3-10.3 8.0-9.7 8.3-10.7 BM1S 8.7 ± 0.4 8.6 ± 0.5 9.5 9.2 ± 0.4 8.3-9.2 8.1-9.7 8.1-10.2 BZP 4.2 ± 0.2 4.1 ±0.3 5.0 5.0 ± 0.3 3.8-4.6 3.6-4.7 3.9-5.4 LM1-3 6.9 ± 0.3 6.9 ± 0.2 9.0 8.7 ± 0.2 6.5-7.4 6.6-7.2 8.1-9.1 WM1 2.2 ±0.13 2.3 ± 0.2 2.9 2.7 ±0.1 1.9-2.5 2.0-2.6 2.4-3.0 Note: Sample parameters are mean ± SD (range). ramirohitra, observed ranges of most variables do overlap, both in the Andringitra samples and in additional series of the two species from Ampi- tambe (Table 22-3). Only the coronal length of their upper molars (LM1-3) does not overlap, a difference that offers a key means of species sep- aration: LM1-3 always < 7.5 mm (range = 6.5- 7.4) in samples of B. betsileoensis and LM 1 -3 al- ways > 8.0 mm (range = 8.1-9.1) in B. ramiro- hitra. Other cranial variables — such as occipito- nasal length (ONL), zygomatic breadth (ZB), and width of the first molar (WM1) — are almost as reliable in providing clear discrimination of the two kinds (Table 22-3; see also Petter, 1975). Other than size, certain qualitative traits may distinguish the species of Brachyuromuys. The pelage of specimens of B. betsileoensis is shorter and less luxuriant than that found on B. ramiro- hitra, and the color is a plainer brown, lacking the deep reddish tones typical of B. ramirohitra, es- pecially on the sides and ventrum. A vestige of the anterior mure connects the two anterior enam- el loops on the first and second upper molars in specimens of B. betsileoensis; the molars of B. ramirohitra generally lack this connection. The voucher specimens preserved by the 1970 RCP expedition constitute the southernmost dis- tributional point for B. betsileoensis, a species that ranges over the central highlands as far north as the vicinity of Didy (1000 m), just southeast of Lac Alaotra (Carleton & Schmidt, 1990). The el- evations for the several Andringitra localities (see below), as later supplied by Paulian et al. (1971), extend from approximately 1 800 to over 2400 m. Based on museum records, Carleton and Schmidt (1990) had reported the altitudinal distribution for this species from about 900 to 2000 m. Ecology and Reproduction — On the Andrin- gitra Massif, the material collected by the RCP team suggests that B. betsileoensis occurs in heathland areas above the tree line down to the upper portion of montane and sclerophyllous for- est (1700-1900 m). In late 1993, cranial remains of B. betsileoensis were recovered from Crypto- procta ferox scats collected in heathland at 2000 and 2100 m. The lower range of the elevational transect performed (720 through 1625 m) presum- ably accounts for the absence of the species in the 1993 survey. No reproductive information is available. GOODMAN & CARLETON: RODENTS 263 Specimens Examined — Foret Aguaria (MNHN 1972.586, 1972.587, 1972.600, and 1972.601); Plateau Andohariana (MNHN 1972.594 through 1972.596); Cuvette de Boby (MNHN 1972.588 and 1972.589); Ivangomena, S Varavarana (MNHN 1972.592, 1972.593, 1972.598, and 1972.599); Gue de la Riembavy (MNHN 1972.590); Varavarana (MNHN 1972.591 and 1972.597). Brachyuromys ramirohitra Major, 1896a Identification and Distribution — Discrimi- nation of this species from its congener B. betsi- leoensis is covered in the previous account (also see Tables 22-2 and 22-3). The meager provenance for B. ramirohitra sug- gests that its geographical and elevational distri- bution is more confined. As with B. betsileoensis, the records of B. ramirohitra on the Andringitra Massif represent its southernmost limit as cur- rently understood. The two specimens obtained in 1993, together with the one collected by Albignac in 1970, document its presence from 1210 to about 2000 m in the RNI d' Andringitra. Carleton and Schmidt ( 1 990) reported B. ramirohitra from two other localities, Ampitambe and Amboasary, at about 900 and 1300 m, respectively. In both places, which lie within the southern sector of the island's Central High Plateau, the two species of Brachyuromys occur in close proximity if not in actual sympatry. Ecology and Reproduction — Brachyuromys ramirohitra apparently inhabits slightly lower vegetational zones and penetrates primary forest. The single RCP specimen was collected in Agua- ria forest, probably near camp 4 at Marositry, at approximately 2000 m, to judge from its date of collection (Paulian et al., 1971). This record and those reported above confirm the syntopic exis- tence of B. betsileoensis and B. ramirohitra at least in the Aguaria zone at the fringe of upper montane (sclerophyllous) forest; in fact, both spe- cies were taken in Aguaria habitat on the same day. The two specimens taken in 1993, at 1210 and 1625 m, were trapped in primary montane and sclerophyllous forest, respectively. The trap at 1625 m was placed at the opening of a natural tunnel system associated with a massive root en- tanglement covered by soil and matted leaf litter. The male from 1210 m has partially erupted third molars and is reproductively immature, its testes small and within the abdomen. The adult female trapped at 1625 m has slightly enlarged mammae and two embryos (3 mm crown-rump length). Specimens Examined — Foret Aguaria (MNHN 1972.585); 38 km S Ambalavao, ridge east of Volotsangana River, 1625 m (FMNH 151660); 40 km S Ambalavao, along Volotsangana River, 1210 m (FMNH 151659). Eliurus Milne Edwards (1885) Recognition — Among Madagascar's endemic rodents, members of this genus are small to me- dium in body size (Table 22-2) and are immedi- ately recognizable by their densely haired tail that bears a conspicuous terminal pencil or tuft. In all species of tufted-tailed rats, the tail length is greater than the head-and-body (TL about 1 15— 120% of HBL), their fur is soft and fine, and the hind foot is relatively short and broad, with six fleshy plantar pads and an elongate fifth digit. The mammae number six, distributed as one postaxial, one abdominal, and one inguinal pair. Diagnostic traits of the skull include the hourglass-shaped in- terorbit devoid of supraorbital shelves or ridges; short incisive foramina, which terminate well an- terior to the molar rows; and the moderately hyp- sodont molars, each with a uniquely trilaminate occlusal configuration (see Carleton, 1994). No regular pattern of sexual dimorphism in cranio- dental size is evident among the RNI d'Andrin- gitra samples of Eliurus, a finding consistent with the analyses by Carleton (1994). Four of the eight known species of Eliurus have been documented in the RNI d' Andringitra. Eliurus majori Thomas, 1895 Identification and Distribution — Prior to this survey, E. majori was known only from five spec- imens from three widely separated localities: two from Montagne d'Ambre, an isolated peak at the northernmost end of Madagascar; two from Am- bohimitombo, the type locality on the Central High Plateau; and one from Anjavidilava in the RNI d' Andringitra (Carleton, 1994). The fine se- ries now available from the Andringitra Massif permits descriptive elaboration and additional tax- onomic comments on the species. Specimens of E. majori rank with those of E. tanala as nearly the largest-bodied in the genus (Table 22-2), a comparison not appreciated by 264 FIELDIANA: ZOOLOGY Carleton (1994) with the limited specimen data then at hand. Relative to its head-and-body size, however, the hind feet of E. majori appear shorter than those observed in examples of E. tanala and E. webbi, and its pinnae are both absolutely and relatively smaller than those of either of those species and scarcely surpass the average ear length recorded for the diminutive E. minor (Table 22-2). The dorsum is consistently somber in tone, colored blackish gray to blackish brown. In con- trast to other Eliurus, chromatic demarcation be- tween the dorsal and ventral pelage is typically indistinct, although the ventrum appears generally lighter; in some individuals, a suffusion of cream- tipped hairs does lend more dorsal-ventral con- trast. The top of the metatarsum is uniformly dusky and contrasts with the dull whitish phalan- ges. Although the tail is well furred and noticeably penicillate in E. majori, the terminal hairs form a less pronounced tuft than in other Eliurus. The caudal hairs gradually become longer toward the tip of the tail, unlike the abrupt elongation into a terminal brush over the distal one-third to one- fourth of the tail, as found in other species, in- cluding the smaller, shorter-tailed E. minor. The caudal hairs are uniformly dark in all 18 speci- mens with an intact tip. The constancy of this trait, as can now be substantiated in a large pop- ulation sample of E. majori, elevates our confi- dence in the diagnostic value of tail color with regard to its specific distinction from E. penicil- latus, a form described from Ampitambe (Thom- as, 1908). Pelage color and cranial morphology of the two species are remarkably alike, except for the white-tipped tail exhibited by all individuals of E. penicillatus from Ampitambe, the one col- lecting site so far known (Carleton, 1994). Aside from the distinctiveness of its pelage and external form, many proportional features of the cranium and dentition set E. majori (and E. pen- icillatus) apart from other species of Eliurus (El- lerman, 1949; Carleton, 1994). These shape con- trasts retain their discriminatory power for the RNI d' Andringitra series of E. majori and are best reflected in their robust molar rows, long and wide incisive foramina relative to the short diastema, and laterally bowed zygomatic arches (compare Tables 22-4 and 22-7). Although the disparate sample sizes discourage meaningful comparisons, substantial morphologi- cal variation is evident among the three widely separated samples of E. majori. In most cranio- dental measurements, the animals from the An- Table 22-4. Comparison of selected craniodental measurements of adult Eliurus majori from Montagne d'Ambre (N = 2), Ambohimitombo (type locality, N = 2), and Andringitra (N = 17). Mt. Ambohimi- Variable d'Ambre* tombof Andringitra ONL 36.5, 37.5 35.2, 35.8 37.6 ± 1.0 (35.5-39.4) ZB 19.3, 19.3 18.3, 18.6 19.1 ±0.5 (18.1-19.8) BBC 14.1, 14.2 13.7, 14.1 14.3 ± 0.4 (13.4-14.8) IOB 5.2, 5.5 5.0, 5.2 5.3 ± 0.2 (4.9-5.7) LR 12.0, 13.1 12.1, 12.2 13.1 ±0.5 (11.6-14.0) PPL 13.0, 13.2 11.9, 12.7 13.4 ± 0.5 (12.4-14.6) LIF 5.5,6.1 5.7, 6.0 6.3 ± 0.3 (5.6-6.7) LD 10.8, 11.1 9.8, 10.2 10.8 ± 0.4 (10.0-11.8) BM1S 7.9, 8.3 7.2, 7.8 8.1 ±0.3 (7.1-8.6) BZP 3.1, 3.2 2.9, 3.1 3.5 ± 0.2 (2.9-3.9) LM1-3 6.1, 6.3 6.3, 6.8 6.6 ± 0.2 (6.1-7.1) WM1 1.7, 1.8 1.5, 1.8 1.8 ±0.1 (1.7-1.9) Note: Values for Andringitra specimens are means SD (range). * AMNH 100854 and 100867. t MCZ 45929 and BMNH 97.9.1.147 (holotype). dringitra Massif average larger and attain greater maxima than recorded for either the two speci- mens from Montagne d'Ambre or the two from Ambohimitombo, the type locality (Table 22-4). However, observed ranges of the RNI d' Andrin- gitra sample generally circumscribe individual values from the latter two sites. Biased age rep- resentation further complicates evaluation of in- terlocality differences and their taxonomic signif- icance. According to evidence from tooth wear, the two specimens from Ambohimitombo are a young animal and a full adult (the type); the 17 from the RNI d' Andringitra represent a mixture of young to old adult animals. The two AMNH records from Montagne d'Ambre are extremely old adults, to judge by their flatly worn molars, yet they appear generally smaller when placed side by side with comparably aged individuals within the Andringitra series. Considering both relative age and sample size, the population from Montagne d'Ambre appears most divergent; this may be anticipated in view of its greater isolation GOODMAN & CARLETON: RODENTS 265 Table 22-5. Microhabitat occurrences of rodent species. Above-ground location Vine, ( .round locatior i limb, or Limbs. Trap position Under By trunk trunks Sus- No. On Above Leaf rotten roots, <10 >10 pended Elevation and species taken ground ground litter wood trunks Misc. cm cm trunks Misc. 720 m Trap distribution 63 37 33 7 18 5 32 2 3 0 Eliurus minor 5 3 2 2 1 0 0 2 0 0 0 Eliurus webbi 10 4 6 4 0 0 0 2 0 2 2 Gymnuromys roberti 1 1 1 0 0 0 810 m Trap distribution 52 48 27 9 16 0 39 1 7 1 Rattus rattus 1 0 1 1 0 0 0 Eliurus minor 13 2 11 1 0 1 0 11 0 0 0 Eliurus tanala 6 0 6 4 0 2 0 Eliurus webbi 13 2 11 1 0 1 0 11 0 0 0 Gymnuromys roberti 1 1 0 1 0 0 0 Nesomys audeberti 1 1 0 1 0 0 0 Nesomys rufus 5 5 0 5 0 0 0 1210 m Trap distribution 58 42 33 6 16 3 22 3 14 3 Rattus rattus 4 3 1 0 0 3 0 0 0 1 0 Eliurus majori 10 0 10 4 1 4 1 Eliurus minor 7 3 4 0 0 3 0 1 0 1 2 Eliurus tanala 9 5 4 3 1 0 1 2 0 2 0 Gymnuromys roberti 3 3 0 0 1 2 0 Nesomys rufus 11 11 0 6 0 4 1 1625 m Trap distribution 65 35 25 8 31 1 23 9 2 1 Rattus rattus 6 4 2 2 2 0 0 1 1 0 0 Brachyuromys ramirohitra 1 1 0 0 0 1 0 Eliurus majori 12 4 8 1 0 2 1 1 6 1 0 Eliurus minor 3 1 2 0 0 1 0 0 2 0 0 Eliurus tanala 9 6 3 0 1 5 0 1 0 2 0 Gymnuromys roberti 1 1 0 0 1 0 0 Nesomys rufus 14 14 0 9 5 0 0 Monticolomys koopmani 2 0 2 2 0 0 0 Totals: 720-1625 m Trap distribution 238 162 118 30 81 9 116 15 26 5 Rattus rattus 11 7 4 2 2 3 0 2 1 1 0 Brachyuromys ramirohitra 1 1 0 0 0 1 0 Eliurus majori 22 4 18 1 0 2 1 5 7 5 1 Eliurus minor 28 9 19 3 1 5 0 14 2 1 2 Eliurus tanala 24 11 13 3 2 5 1 7 0 6 0 Eliurus webbi 23 6 17 5 0 1 0 13 0 2 2 Gymnuromys roberti 6 6 0 2 2 2 0 Nesomys audeberti 1 1 0 1 0 0 0 Nesomys rufus 30 30 0 20 5 4 1 Monticolomys koopmani 2 0 2 2 0 0 0 from the central and southern highlands of Mad- agascar. Additional samples from Montagne d'Ambre and from intermediate localities are re- quired to validate this supposition and to deter- mine whether more than one species is represent- ed among the populations currently classified as E. majori. Specimens taken during the 1993 expedition were restricted to the 1210 and 1625 m zones. In 1971, Albignac collected the sole previous ex- 266 FIELDIANA: ZOOLOGY Table 22-6. Comparison of selected craniodental measurements of adult Eliurus minor from Ampitambe (type locality; N = 2) and Andringitra (N = 17). Sample parameters are mean ± standard deviation, and range. Table 22-7. Comparison of selected craniodental measurements of adult Eliurus tanala (N = 14) and E. webbi (N = 19) from Andringitra. Sample parameters are mean ± standard deviation, and range. Variable Ampitambe* Andringitra Variable /■.". tanala E. webbi ONL 28.9, 30.8 30.1 ± 1.0 ONL 40.5 ± 1.0 37.5 ± 1.1 27.7-31.3 38.8-42.3 35.4-39.0 ZB 15.3, 15.9 15.3 ± 0.6 ZB 19.5 ± 0.6 17.7 ± 0.7 14.1-16.3 18.5-20.7 16.4-19.0 BBC 12.2, 12.4 12.0 ± 0.3 BBC 14.6 ± 0.3 14.0 ± 0.5 11.4-12.5 14.1-15.3 13.1-15.0 IOB 4.8,5.1 5.0 ± 0.2 IOB 5.9 ± 0.3 5.8 ± 0.2 4.8-5.3 5.4-6.4 5.3-6.2 LR 9.8, 10.1 10.3 ± 0.4 LR 14.7 ± 0.6 13.3 ± 0.5 9.4-10.9 13.7-15.5 12.2-14.4 PPL 10.0, 11.7 10.8 ± 0.5 PPL 14.1 ±0.5 13.2 ± 0.5 9.8-11.4 13.6-14.7 12.0-13.9 LIF 4.2, 4.4 4.2 ± 0.3 LIF 5.4 ± 0.3 5.1 ± 0.3 3.7-4.7 4.8-5.9 4.6-5.8 LD 8.3, 8.7 8.8 ± 0.4 LD 12.3 ± 0.5 11.2 ±0.5 8.1-9.3 11.4-13.3 10.2-12.5 BM1S 6.0, 6.3 6.1 ± 0.2 BM1S 7.9 ± 0.2 7.5 ± 0.3 5.7-6.6 7.4-8.0 6.9-7.9 BZP 2.3, 2.5 2.6 ± 0.2 BZP 3.8 ± 0.2 3.3 ± 0.4 2.2-3.0 3.5-4.1 2.9-3.6 LM1-3 4.0, 4.5 4.1 ±0.1 LM1-3 5.7 ± 0.2 5.3 ± 0.2 3.9-4.4 5.3-6.0 4.9-5.6 WM1 1.1, 1.2 1.1 ± 0.05 WM1 1.5 ±0.05 1.4 ± 0.07 1.0-1.2 1.4-1.6 1.2-1.5 * BMNH 97.9.1.153 (holotype) and FMNH 5629. ample of this species on the Andringitra Massif near Anjavidilava, around 2000 m (elevation fide Paulian et al„ 1971). These records collectively document the local altitudinal range of E. majori from 1210 to 2000 m, a zone coinciding with mid- dle to upper montane (sclerophyllous) forest, and they constitute the first vouchered evidence of sympatry with any other species of Eliurus, name- ly E. minor and E. tanala. Elsewhere, the species is known from elevations of approximately 1000 (Montagne d'Ambre) and 1200 m (Ambohimi- tombo). Ecology and Reproduction — Trapping evi- dence portrays E. majori as principally arboreal in habits, with occasional forays to the ground (Table 22-5). All examples taken at 1210 m were in elevated traps placed on lianas, limbs, and trunks of various sizes and different inclinations. At 1625 m, four of 12 animals trapped were on the ground, specifically in leaf litter and by roots and trunks of trees, whereas eight were collected in a variety of trap placements above the ground. The 22 E. majori taken during the 1993 season included nine males with scrotal testes, two males with abdominal testes, nine females with enlarged mammae, and one adult female with small mam- mae. One female had four 11 -cm (crown-rump length) embryos, and three other females each had three discernible placental scars. Specimens Examined — 38 km S Ambalavao, ridge east of Volotsangana River, 1625 m (FMNH 151666, 151667, 151732, 151752, 151851 through 151853, and 151854 through 151858); 40 km S Ambalavao, along Volotsangana River, 1210 m (FMNH 151661 through 151665, 151730, 151731, and 151847 through 151849); Anjavidi- lava (MNHN 1972.602). Eliurus minor Major, 1896b Identification and Distribution — Compared to other Eliurus, individuals of this species are readily recognized by their smaller size in most dimensions of the body and skull (Tables 22-2 and 22-6) and by the well-developed, dark pencil on the distal half of the tail. Within Nesomyinae, only specimens of Monticolomys koopmani ap- proach those of E. minor on the basis of overall size. Individuals of Monticolomys, however, pos- sess a lighter build, shorter head-and-body length, GOODMAN & CARLETON: RODENTS 267 Table 22-8. Comparison of selected craniodental measurements of adult Gymnuromys roberti from An- dringitra (N = 4) and from Ampitambe (N = 20), the type locality. Sample parameters are mean ± standard deviation, and range. Table 22-9. Comparison of selected craniodental measurements of adult Nesomys audeberti (N = 1) and N. rufus (N = 21) from Andringitra. Sample parameters are mean ± standard deviation, and range. Variable Variable N. audeberti Andringitra Ampitambe N. rufus ONL 47.5 44.5 ± 0.9 ONL 39.9 ± 1.9 39.4 ± 1.0 42.3-46.3 37.9-41.7 36.7-41.0 ZB 24.7 23.3 ± 0.6 ZB 19.5 ± 1.1 19.6 ± 0.6 22.5-24.3 18.6-21.0 18.3-20.9 BBC 17.3 16.3 ± 0.3 BBC 14.6 ± 0.03 14.5 ± 0.3 15.7-16.7 14.6-14.7 13.9-15.0 IOB 8.5 8.1 ±0.2 IOB 5.9 ± 0.3 5.9 ± 0.3 7.7-8.4 5.6-6.3 5.4-6.4 LR 18.8 16.9 ± 0.6 LR 14.4 ± 0.7 14.2 ± 0.5 15.6-17.9 13.6-15.0 13.3-15.6 PPL 16.0 14.6 ± 0.4 PPL 13.8 ± 1.1 14.0 ± 0.7 13.6-15.4 12.9-15.0 12.4-15.1 LIF 9.6 8.3 ± 0.4 LIF 4.7 ± 0.4 4.5 ± 0.3 7.4-8.9 4.2-5.1 4.1-5.2 LD 13.0 12.1 ± 0.3 LD 11.6 ±0.5 11.6 ±0.6 11.5-12.6 11.2-12.1 10.5-12.7 BM1S 9.8 9.9 ± 0.4 BM1S 8.2 ± 0.4 8.2 ± 0.2 9.2-10.8 7.8-8.6 7.8-8.7 BZP 4.3 4.5 ± 0.3 BZP 2.9 ± 0.3 2.9 ± 0.2 4.0-5.3 2.5-3.3 2.5-3.3 DAB 7.3 6.7 ± 0.1 LM1-3 5.8 ± 0.1 5.6 ± 0.2 6.4-7.1 5.7-5.9 5.3-6.0 LM1-3 7.1 6.9 ± 0.2 WM1 1.6 ±0.1 1.6 ±0.1 6.6-7.4 1.5-1.8 1.3-1.8 WM1 2.2 2.2 ±0.1 2.1-2.4 relatively longer tail (140% of HBL), and longer hind feet (Table 22-2); furthermore, they lack the conspicuous terminal brush on the tail that is so characteristic of all species of Eliurus. Crania and dentitions of E. minor and Monticolomys differ strikingly and allow no confusion (see Chapter 21). Eliurus minor is found throughout eastern for- est, in lowland and montane settings and from near sea level to over 1500 m. In view of its broad geographic range, the degree of differentiation of certain population samples, particularly those around Antongil Bay, and their taxonomic status, require further clarification. The series from the RNI d' Andringitra, however, agrees closely with the type material from Ampitambe (Table 22-6) and conforms closely in mensural and pelage vari- ation to other E. minor from southern Madagascar, such as specimens collected in Pare National (PN) de Ranomafana and vicinity (see Carleton, 1994: appendix 2). The elevational distribution of E. minor on the eastern slopes of the RNI d' Andringitra (720 to 1625 m) falls within its previously recorded range elsewhere on the island, from near sea level to 1800 m (Carleton & Schmidt, 1990). Ecology and Reproduction — Eliurus minor occurs in all four elevational zones sampled, but relative trapping success suggests that it is most abundant at middle elevations, 720 to 1210 m (Ta- ble 22-10). Data pooled from the four elevations indicate that Eliurus minor frequents a wide variety of ter- restrial and arboreal microhabitats (Table 22-5). Although 50% (14 of 28) of all individual cap- tures occurred on lianas, vines, or trunks less than 10 cm in circumference, the overall proportion of traps placed above ground was just 29% (116 of 400 traps). Eliurus minor appears to move across relatively thin vines and branches. In three of the four elevational zones, E. minor was captured on or off the ground with nearly equal frequency; the exception was at 810 m, where 11 of 13 (85%) animals were in arboreal sets. The reproductive state of the 28 specimens col- lected in 1993 included nine scrotal males, five males with abdominal testes, 10 females with large mammae (four of which were lactating), and 268 FIELDIANA: ZOOLOGY Table 22-10. Number of individuals trapped for each species of small mammal at all four elevations surveyed (total trap-days accrued is shown in parentheses). 720 m 810 m 1210 m 1625 m Species (550) (650) (625) (650) Rutins rutins 1 4 6 Brachyuromys ramirohitra ... 1 Eliurus majori 10 12 Eliurus minor 5 13 7 3 Eliurus tanala 6 9 9 Eliurus webbi 10 13 ... Gymnuromys roberti 1 1 3 1 Nesomys audeberti 1 ... Nesomys rufus 5 11 14 Monticolomys koopmani 2 Microgale spp. 10 1 Total individuals 16 40 54 49 % trap success 2.9 6.2 8.6 7.8 Total rodents 16 40 44 48 % rodent trap success 2.9 6.2 7.0 7.7 Total nesomyines 16 39 40 42 % nesomyine trap success 2.9 6.0 7.0 7.4 three females with small mammae and imperfor- ated vaginas. Two females with large mammae had placental scars, numbering three and four, and no embryos. One female had three embryos mea- suring 10 mm in crown-rump length. Nine fe- males exhibited the mammae apportionment typ- ical of Eliurus, but one appeared to lack the post- axial pair and instead possessed an extra abdom- inal pair (0-4-2). Specimens Examined — 38 km S Ambalavao, ridge E of Volotsangana River, 1625 m (FMNH 151679, 151867, and 151868); 40 km S Amba- lavao, along Volotsangana River, 1210 m (FMNH 151678, 151736 through 151738, and 151864 through 151866); 43 km S Ambalavao, junction of Sahanivoraky and Sahavatoy rivers, 810 m (FMNH 151672 through 151677, 151734 through 151737, and 151859 through 151863); 45 km S Ambalavao, E bank of Iantara River, along Am- balamanenjana-Ambatomboay trail, 720 m (FMNH 151668 through 151671, and 151733). Eliurus tanala Major 1896b Identification and Distribution — Eliurus ta- nala is a distinctive species characterized by mod- erately large size and a long tail with feathery white tip. The white caudal plume provides defin- itive separation from specimens of E. majori and E. webbi, which overlap those of E. tanala in cer- tain external dimensions (such as TOTL, HBL, TL, and WT— Table 22-2). Eliurus tanala and E. webbi are otherwise very similar in craniodental size and shape, but the former can be reliably dis- criminated by its generally longer, broader skull and more robust molar rows (Table 22-7). The morphometric differences between the two spe- cies in the RNI d'Andringitra resemble those demonstrated for other places where they co-oc- cur, for instance the PN de Ranomafana area (Carleton, 1994). Other anatomical traits that al- low for discrimination between the two species are discussed and illustrated by Carleton (1994). Samples of Eliurus tanala are morphologically homogeneous over its range in middle elevation forests along the eastern flanks of the central high- lands. The RNI d'Andringitra records extend the known distribution slightly farther south of its for- mer limit at Vinanitelo, the type locality. Eliurus tanala was found in the 810, 1210, and 1625 m zones in the RNI d'Andringitra; Carleton (1994) had summarized its elevational occurrence as 455 to 1300 m. Ecology and Reproduction — Trap success of this species increased slightly as a function of higher altitude (Table 22-10). At 810 m, the low- est elevational occurrence, E. tanala was captured only in arboreal sets, largely on nearly horizontal small vines and limbs. At higher elevations, ap- proximately half of the additional 1 8 individuals captured were on the ground. In both situations, the species was taken in a variety of sets (Table 22-5). At 1625 m, four individuals were collected over the course of 7 trap-days in the same Na- GOODMAN & CARLETON: RODENTS 269 tional trap placed at the opening of an extensive natural tunnel system under root tangles and mat- ted leaf litter. During November and December 1993, the sample of 23 E. tanala exhibited the following reproductive conditions: four males with scrotal testes, seven males with abdominal testes, six fe- males with large teats, and four females with small teats. One female was pregnant with three embryos, each measuring 15 mm crown-rump length, and another female with small mammae had three placental scars. The one juvenile E. ta- nala (molars partially erupted) was trapped on December 1, 1993 at 1210 m. Specimens Examined — 38 km S Ambalavao, ridge E of Volotsangana River, 1625 m (FMNH 151691, 151692, 151744, and 151878 through 151883); 40 km S Ambalavao, along Volotsan- gana River, 1210 m (FMNH 151690, 151743, 151872 through 151877, and 151897); 43 km S Ambalavao, junction of Sahanivoraky and Sahav- atoy rivers, 810 m (FMNH 151687 through 151689 and 151869 through 151871). Table 22- 1 1 . Comparison of nesomyine rodents documented to date for adequately censused regions in eastern humid forests. Rano- Anala- Andrin- ma- ma- gitra* fanat /aotra 720- 575- 500- Species 2400 m 1225 m 1300 m Brachytarsomys albicauda X X Brachyuromys betsileoensis X X X Brachyuromys ramirohitra X Eliurus majori X Eliurus minor X X X Eliurus petteri X Eliurus tanala X X X Eliurus webbi X X X Gymnuromys roberti X X X Nesomys audeberti X X X Nesomys rufus X X X Monticolomys koopmani X * This study. t Carleton and Schmidt (1990); Ryan et al. (1993). X Compiled from Carleton and Schmidt (1990); Carle- ton (1994). Eliurus webbi Ellerman, 1949 Identification and Distribution — Eliurus webbi suggests a more lightly built version of the larger E. tanala, but with a browner (rather than grayish) cast to its dorsal pelage and with a lush brown (rather than airy white) tail tuft. Other con- trasts are given in the previous account (and see Tables 22-2 and 22-7). As currently known, the species has the greatest range among Eliurus, occurring along the eastern versant of Madagascar, from its northernmost end (Montagne d'Ambre) to the southeastern lowlands (near Vondrozo and south of Farafangana, the type locality). Although much chromatic varia- tion, especially of the ventral pelage, is apparent among locality samples of E. webbi, the close agreement in size and proportions of the skin and skull, and in color and development of the tail pencil, reinforces the notion that only one species is represented. In general, the dimensions of E. webbi specimens taken in the RNI d'Andringitra (Tables 22-2 and 22-7) approximate those of an- imals collected in the PN de Ranomafana region (Carleton, 1994: appendix 2). In the RNI d'Andringitra, E. webbi was con- fined to the 720 and 810 m zones, at which latter elevation it is sympatric with E. tanala. Else- where, specimens of this broadly ranging species have been taken at sites from near sea level to about 1500 m (Carleton & Schmidt, 1990; Carle- ton, 1994). Ecology and Reproduction — Populations of E. webbi appeared approximately equal in abun- dance at 720 and 810 m (Tables 22-10 and 22-1 1). Four of 10 E. webbi obtained at 720 m were on and six off the ground, whereas 810 m the numbers in 13 captures were two and 11, respec- tively (Table 22-5). The only nesting site of Eli- urus found during the 1993 Andringitra survey was a ground burrow of E. webbi, a species for which only six of the 23 individuals (26%) cap- tured were taken with ground sets. Whether this burrow placement is typical of E. webbi is un- known, but this information combined with trap- ping results may indicate that this species nests and forages in different strata of the forest. In forest with open understory near the Iantara River (720 m), a pair of E. webbi were found living in the same underground burrow with nest- ing Scaly-breasted Ground-Rollers (Brachyptera- cias squamiger) (Goodman, 1994). The birds had built a nest in the principal tunnel of the burrow system, anterior to the rodents' food storage cache, the distalmost portion of the tunnel. Chewed seeds of Cryptocarya (Family Lauraceae) were found in the storage cache (see Chapter 4, Appendix 4-2). In contrast to E. tanala, remarkably few of the 270 FIELDIANA: ZOOLOGY 23 E. webbi collected in November and December 1993 showed signs of active reproduction. One of 16 adult males had scrotal testes, and two of seven females had large mammae. No female dissected possessed embryos. Specimens Examined — 43 km S Ambalavao, junction of Sahanivoraky and Sahavatoy rivers, 810 m (FMNH 151684 through 151686, 151741, 151742, 151888 through 151889, and 151891 through 151895); 45 km S Ambalavao, E bank of Iantara River, along Ambalamanenjana-Amba- tomboay trail, 720 m (FMNH 151680 through 151683, 151739, 151740, and 151884 through 151887). Gymnuromys Major, 1896a Recognition — This sleek-furred, dark grey rat is medium-sized but has a stout body, an impres- sion perhaps heightened by its relatively short tail and comparatively robust hind feet (Table 22-2). Cream to white-tipped hairs over slate bases, or sometimes wholly white hairs, give the ventral fur a silvery sheen. The tops of the metapodials and phalanges, both fore and rear, are light colored, covered with fine whitish to hyaline hairs. The hind foot is long but relatively broad, with the outer toes (I and V) somewhat shorter than the central three (II-IV). It bears six plantar pads. The tail is scantily haired, appearing naked to the unaided eye, and only slightly longer than head- and-body length (TL about 104% of HBL); it is countershaded, dusky above and white below, for much of its length, but in most individuals, the distal one-third to one-fourth of the caudal epi- dermis is white over the entire circumference. The six mammae are distributed as one postaxial, one abdominal, and one inguinal pair. The dorsal colors of young Rattus rattus or some Eliurus, such as E. majori, have a superfi- cial resemblance to the grayish-black dorsum of Gymnuromys. Closer inspection discloses that melanistic individuals of Rattus possess noticea- bly longer and monocolored tails (TL about 1 17% HBL). The fur of Eliurus majori is more woolly than sleek, and, like others in the genus, the tail is conspicuously longer than the length of the head-and-body and the hind feet are short and broad compared to those of Gymnuromys (Table 22-2). In Eliurus, the dense caudal hairs, ending in a well-developed distal tuft, provide absolute separation from the naked tail of Gymnuromys. The smoothly contoured braincase, hourglass- shaped interorbital shape, short incisive foramina, and tiny auditory bullae of Gymnuromys recall features of the cranium in members of Eliurus. Specimens of Gymnuromys, however, lack a sub- squamosal foramen, which is typical of most Eli- urus, and their moderately hypsodont molars with convoluted enamel loops and compressed laminae are singularly distinctive (Major, 1 896a; Ellerman, 1941). Size and proportions of the individual mo- lars that constitute each toothrow are unique with- in Nesomyinae, if not within Muroidea. The third molars, both uppers and lowers, are markedly wider and longer than the second molars, which are slightly wider and longer than the first molars, a reversal of the typical muroid pattern, in which the Mis > M2s > M3s. This unusual combination of occlusal configuration and intermolar propor- tions led Ellerman (1940) to create the monotypic subfamily, Gymnuromyinae, within a broadly de- fined Muridae. Gymnuromys roberti Major, 1896a Identification and Distribution — The small series from Andringitra is consistent in average size and range compared to the larger sample of G. roberti from Ampitambe, the type locality (Ta- ble 22-8). Indeed, examples of G. roberti through- out its range appear fairly homogeneous in mor- phology and pelage, although sample sizes, with the exception of Ampitambe, for critically assess- ing geographic variation, are paltry. The specimens from RNI d' Andringitra fall within the known geographic range of Gymnuro- mys, which inhabits eastern forest formations bor- dering the Central High Plateau. Carleton and Schmidt (1990) noted an elevational distribution based on museum records from 500 to 950 m. The elevational expanse of G. roberti is actually great- er, because it occurs at every site sampled, 720 to 1625 m, in the 1993 Andringitra survey, and O'Keeffe and Ashmore (no date) reported it ear- lier, near Anjavidilava, at about 1800 m. Ecology and Reproduction — Although few individuals of G. roberti were captured in 1993, the species is one of the more broadly distributed on the slopes of the RNI d' Andringitra (Table 22-10). Only Eliurus minor and Nesomys rufus are also found in all four elevational zones. Gymnuromys roberti appears to be an exclu- sively terrestrial rat. All six examples were caught on the ground, generally in sparse ground cover and equally split between traps placed in leaf lit- GOODMAN & CARLETON: RODENTS 271 ter, near dead wood, or by roots and trunks (Table 22-5). O'Keeffe and Ashmore (no date) trapped a subadult in a pitfall line set across the ecotone between dry ericaceous habitat and boggy vege- tation at the edge of a small stream. A burrow of the species was discovered on the crest (810 m) of a slight ridge and in an area with relatively open understory. The entrance opened under a fallen log and continued for about 1 m as a relatively straight tunnel at about 15-20° to a small chamber. Twenty-two Canarium (Family Burseraceae) fruits were found in the chamber, 2 1 of which had been gnawed open and the contents eaten (see Chapter 6). A partial cranium of an adult G. roberti was found in a National trap. A carnivore had dragged the trap from its initial placement site and eaten most of the rodent. Among the six Gymnuromys roberti collected, two adult males possessed abdominal testes, and two adult females had enlarged teats, one of which had two placental scars. A subadult male was taken on December 2, 1993 at 1210 m. Specimens Examined — 38 km S Ambalavao, ridge E of Volotsangana River, 1625 m (FMNH 151695); 40 km S Ambalavao, along Volotsan- gana River, 1210 m (FMNH 151896, 151898, and 151927); 43 km S Ambalavao, junction of Sahan- ivoraky and Sahavatoy rivers, 810 m (FMNH 151694); 45 km S Ambalavao, E bank Iantara River, along Ambalamanenjana-Ambatomboay trail, 720 m (FMNH 151693). Monticolomys Carleton & Goodman, 1996 Recognition — Diagnostic characters, differen- tiating comparisons, measurements, and biogeo- graphic comments are presented in Chapter 21. Monticolomys koopmani Carleton & Goodman, 1996 Identification — The small size of this form renders its segregation from most Nesomyinae un- problematic (Table 22-2). Only specimens of Eli- urus minor are similarly diminutive, but they can be distinguished by numerous qualitative features of the skin and skull (see Chapter 21). Ecology and Reproduction — This species was found only in the 1625-m zone (Table 22-5). Both live-trapped individuals were captured in the same set placed on a nearly horizontal liana less than 10 cm in circumference. The third specimen was obtained in a pitfall bucket. With just three records, little can be said about its focus of activ- ity, whether predominantly arboreal or scansorial. The two males obtained possessed partially or fully descended testes; the one female had small teats and an imperforated vagina. Specimens Examined — 38 km S Ambalavao, ridge E of Volotsangana River, 1625 m (FMNH 151727, 151899, and 151900). Nesomys Peters, 1870 Recognition — These are comparatively big, thickset rats with long and narrow hind feet, large pinnae, and a tail not quite as long as the head- and-body (Table 22-2). Their soft fur is generally a reddish brown on the dorsum, often with a dense intermixture of black hairs mid-dorsally and a pronounced suffusion of chestnut to rufous col- oration on the cheeks, flanks, and rump. Through- out the distribution of the genus, the color of the ventrum varies, from a uniform rufous to pure white and all stages of intermediacy between them. The tail pilosity (in eastern Nesomys) is moderately developed, although it is not penicil- late, nor fully obscuring the scutellation; most caudal hairs are blackish, but the terminal seg- ment (about 10-20 mm) may be tipped with white; the length of the caudal vertebrae is either slightly less than or subequal to the head-and- body (TL about 88-97% of HBL). The confor- mation of the foot suggests cursorial locomotion and terrestrial habits: elongate and narrow, the lat- eral digits (I and V) markedly shorter than the central three, and the six plantar pads widely spaced. Mammary glands may usually consist of one pair abdominal and one pair inguinal (N = 4), but in one individual a postaxial pair was ob- served (N = 6). As with external form, the cranium and denti- tion in examples of Nesomys are robust and per- mit ready identification (Table 22-9). The skull has a long rostrum, smoothly contoured braincase, stout and laterally flaring zygomatic arches, and a relatively wide, hourglass-shaped interorbit. The molars are incipiently high-crowned yet retain a cuspidate topography and distinct mesolophs and mesolophids. Carleton and Schmidt (1990) and Musser and Carleton (1993) cautioned that the taxonomy of Nesomys is more complex than had been con- veyed by the recognition of the one species, N. rufus (see, e.g., Petter, 1972, 1975, and Honacki 272 FTELDIANA: ZOOLOGY et al., 1982). At least three species — N audeberti (1879), N. lambertoni Grandidier (1928), and N. rufus ( 1 870) — merit recognition, but the status of other populations remains uncertain pending completion of a generic revision (Carleton, un- publ.). The association of names and morpholo- gies to the two species of Nesomys in the RNI d'Andringitra must therefore by considered ten- tative. Nesomys audeberti (Jentink, 1879) Identification and Distribution — Specimens of N. audeberti average greater than those of N rufus in most cranial and external dimensions, but individual variation and consequent range overlap are substantial. More reliable discrimination is provided by certain length (ONL, LR) and breadth (ZB, BBC, IOB) dimensions of the skull and glo- bosity of the auditory bullae (DAB), less so the size of the molars (Table 22-9). Length of the hind foot is a helpful yardstick (Ryan et al., 1993): typ- ically HFL (with claw) 2: 52 mm in samples of N. audeberti and HFL ^ 50 mm in those of N. rufus. In addition, most specimens of N. audeberti exhibit a large expanse of pure white hairs on their underparts, in contrast to the rufous ventral coloration characteristic of N rufus, but this pel- age distinction is not an absolute criterion of spe- cies assignment. So it is with the one specimen that we provisionally refer to N audeberti — it matches the large size of skin and skull typical of the form (Tables 22-2 and 22-9), yet it has an entirely rufescent ventrum that suggests an indi- vidual of N. rufus. The 810 m record from RNI d'Andringitra falls within the known altitudinal and geographic dis- tribution of N audeberti, which has an extensive north-south range in eastern lowland rain forest (Carleton & Schmidt, 1990). Elsewhere, N au- deberti and N. rufus have been demonstrated to occur in sympatry or actual syntopy in the regions of Analamazaotra and Ranomafana (Carleton & Schmidt, 1990; Ryan et al., 1993). Most localities where A/, audeberti has been documented are un- der 800 m, but potential overlap with the montane N. rufus appears likely within the 800-1000 m elevational zone, or approximately coincident with the floristic transition from lowland rain for- est to montane associations. Ecology and Reproduction — The lone male was captured on a ridge in pristine lowland rain forest; the live-trap was placed on leaf litter and by a tunnel entrance. He appeared reproductively immature or inactive (testes abdominal). Specimens Examined — 43 km S Ambalavao, junction of Sahanivoraky and Sahavatoy rivers, 810 m (FMNH 151696). Nesomys rufus Peters, 1870 Identification and Distribution — Separation of A/, rufus from N. audeberti on the basis of men- sural characters and pelage color is discussed in the preceding account (and see Tables 22-2 and 22-9). The mean weight reported by Ryan et al. (1993) for male and female A/, rufus from the Ranomafana region generally falls within the range of the RNI d'Andringitra series. Naturally, pregnant females are exceptions; the extreme weight (208 g) obtained within the RNI d'An- dringitra sample was that of a female with two near-term embryos (crown-rump length = 47 mm each). Ryan et al. (1993) reported that adult males of N. rufus are significantly heavier than adult fe- males, but this pattern is not manifest in the RNI d'Andringitra sample, even when pregnant fe- males are removed from the comparison. None of the other five external measurements and only one (LR) of the 18 craniodental variables exhibited significant sexual dimorphism in size for the An- dringitra sample (11 males, 10 females). The RNI d'Andringitra marks the southernmost distributional point of N. rufus reported to date; it was previously documented only as far south as the PN de Ranomafana region, about 115 km to the northeast (Carleton & Schmidt, 1990). The presence of this richly colored Nesomys on the upper forested slopes of the Andringitra Massif (810-1625 m) concords with its montane habitus as recorded elsewhere on the island (900-2300 m). Ecology and Reproduction — Nesomys rufus was captured in the 810, 1210, and 1625 m zones (Table 22-10). The number of individuals cap- tured increased with altitude (regression P = 0.07, r = 0.93, N = 4). Other Nesomys, species inde- terminate, were observed during daylight hours at 720 m, but they seemed rare at this elevation and were never trapped. Nesomys rufus is strictly terrestrial and seems to prefer forests with an open understory and rel- atively thick mat of leaf litter (Table 22-5). Of the 30 individuals trapped, two-thirds were in sets GOODMAN & CARLETON: RODENTS 273 placed in leaf litter on the forest floor; others were in a variety of situations under fallen tree trunks or among root entanglements, but always on the ground. In a radio-tracking study of N. audeberti and N. rufus at the PN de Ranomafana, Ryan et al. (1993) found both species to be exclusively diurnal and terrestrial. Several N. rufus were trapped at the openings of burrows that descended below fallen and rotten tree trunks or large limbs. These burrow systems often contained caches of fruits and nuts (e.g., Cryptocarya, Family Lauraceae, and Canarium, Family Burseraceae). Other food plants this di- urnal species was observed eating were Artabo- trys mabifolius (Family Annonaceae) and Sloanea rhodanta (Family Elaeocarpaceae) (see Chapter 4, Appendix 4-2). Signs of active reproduction were evident among the 31 N. rufus collected during the No- vember and December 1993 survey. Twelve of 18 males possessed fully scrotal testes, and nine of 12 females displayed prominently visible mam- mae. Three females with large teats had fresh pla- cental scars (mean = 3; range = 2-4), and four had one or two embryos in utero, varying in crown-rump length from 12 to 47 mm. Specimens Examined — 38 km S Ambalavao, ridge E of Volotsangana River, 1625 m (FMNH 151702, 151703, 151747 through 151751, 151754, and 151908 through 151915); 40 km S Ambala- vao, along Volotsangana River, 1210 m (FMNH 151699 through 151701, 151704, 151745, 151746, and 151904 through 151907); 43 km S Ambalavao, junction of Sahanivoraky and Sahav- atoy rivers, 810 m (FMNH 151697, 151698, 151753, 151901, and 151902). Discussion Trapping Effort and Sampling Confidence The 1993 small mammal survey of the eastern slopes of RNI d'Andringitra collected nine kinds of native rodents, as well as the introduced human commensal Rattus rattus. Another species, Brachy- uromys betsileoensis, had been collected previ- ously by the RCP survey in 1970-1971, and those voucher specimens are also reported above. The 1 1 rodent species documented herein surpass by seven the number that had been previously re- corded in the reserve (Nicoll & Langrand, 1989; Carleton & Schmidt, 1990). During 5 weeks of fieldwork, 2,475 trap-days were approximately evenly divided among the four elevation sites (Tables 22-1 and 22-10) and produced 159 animals, 148 rodents, and 11 insec- tivores (see Chapter 20). No lemurs were live- trapped, although at least two species, Microcebus rufus and Cheirogaleus major, known to occur in the eastern reserve (see Chapter 25), fall within the size range that can enter the traps used (Na- tional and Sherman traps). Overall trap success of small mammals, includ- ing rodents and insectivores, varied from about 3% at 720 m to approximately 8% at the two up- per elevation localities (Table 22-10). Insectivores (Microgale spp.) were captured in live-traps only at 1210 and 1625 m; trap success for rodents only at these two elevations was 7.0% and 7.7%, re- spectively. And when only the native nesomyine rodents are considered, the rate of capture de- clines slightly more, to 6.0% at 810 m, 7.0% at 1210 m, and 7.4% at 1625 m. In general, these rates of trap return agree favorably with the re- sults of other surveys in Madagascar (Stephenson, 1993; Goodman & Ganzhorn, 1994; Stephenson et al., 1994) and on other large tropical islands (Heaney et al., 1989; Rickart et al., 1991, 1993). Moreover, the consistently lower trap success in lowland rain forest at 720 m concords with the lower number of species, and estimates of relative density and relative biomass captured there (Ta- bles 22-10, 22-12, and 22-13). The rodents captured are either terrestrial or scansorial, although probably none is exclusively arboreal (Table 22-5). In several cases, especially for members of Eliurus, there appeared to be al- titudinal variation within species in the proportion taken on and off the ground. At least four, and probably five, rodent species on the Andringitra Massif are strictly ground-dwelling, namely Brachyuromys ramirohitra (and presumably B. betsileoensis), Gymnuromys roberti, Nesomys au- deberti, and N. rufus. Because the vast majority of traps were placed no more than 4 m above the ground (Table 22-1), canopy specialists, if they indeed exist in Malagasy rain forest, would likely have been overlooked. Two lines of evidence, derived from species ac- cumulation curves and analogous results from other small mammal inventories, bolster the no- tion that the trapping effort was suitable to estab- lish basic rodent diversity with the reserve. Plots of the cumulative number of species documented 274 FIELDIANA: ZOOLOGY Table 22-12. Relative numbers of rodents (given as the number of individuals trapped per 100 m of trap line) in the four elevational zones.* The number in parentheses is total length (in m) of trap lines for each zone. 720 m 810 m 1210 m 1625 m Species (620) (780) (715) (590) Grand mean Rattus rattus 0.13 0.56 1.02 0.53 Brachyuromys ramirohitra ... 0.17 0.17 Eliurus majori 1.40 2.03 1.68 Eliurus minor 0.81 1.67 0.98 0.51 1.04 Eliurus tanala 0.77 1.26 1.53 1.15 Eliurus webbi 1.61 1.67 1.64 Gymnuromys roberti 0.16 0.13 0.42 0.17 0.22 Nesomys audeberti 0.13 0.13 Nesomys rufus 0.64 1.54 2.37 1.44 Monticolomys koopmani 0.34 0.34 All rodents 2.58 5.00 5.60 7.12 Averaged only for elevational zones in which a species was trapped. at each elevational zone over the duration of trap- ping disclose distinct plateaus before the end of the survey (Fig. 22- la). That is, no additional ro- dent species were encountered after 200 to 500 trap-days of census, depending upon the elevation and total number of species eventually recorded at each. Nevertheless, the decrease in species add- ed through time within each altitudinal zone does not correspond to any general decline in overall trap success during the course of the survey pe- riod (Fig. 22- lb). The number and composition of rodent species thus far documented for other places in the eastern humid forest at intermediate elevations provide circumstantial evidence about those that might be predicted to occur in the RNI d'Andringitra. Two general localities — PN de Ranomafana and Re- serve Speciale (RS) d'Analamazaotra (Perinet) — are informative in this regard, because consider- able small mammal inventories have been con- ducted in and around each, the former within the past few years and the latter since the early ex- plorations by naturalists. At their closest points, the PN de Ranomafana is 1 1 5 km northeast of the Andringitra Massif and RS d'Analamazaotra some 380 km northeast, but both are similarly sit- uated in moist forest along the eastern flank of the central highlands. Eight and nine species of Ne- somyinae have been reported from Ranomafana and Analamazaotra, respectively, or one less than the number now actually known from the RNI d'Andringitra (Table 22-11). Furthermore, the composition of indigenous rodents is similar among the three areas, because of the extensive geographic ranges of most species — such as Eli- urus minor, E. tanala, E. webbi, Gymnuromys roberti, and Nesomys audeberti — along the mid- dle elevation, forested eastern slopes of the Cen- tral High Plateau and the eastern escarpment. Of the several widely distributed nesomyine Table 22-13. Estimated biomass (g) of rodents trapped along an elevational transect. Summated from average weight of adults captured over the first 550 trap-days within each zone. Species 720 m 810 m 1210 m 1625 m Rattus rattus 103 414 620 Brachyuromys ramirohitra ... 117 Eliurus majori ... 580 869 Eliurus minor 148 443 258 74 Eliurus tanala ... 408 735 654 Eliurus webbi 506 940 ... Gymnuromys roberti 119 119 358 119 Nesomys audeberti ... 181 Nesomys rufus ... 793 1,427 1,427 Monticolomys koopmani ... 52 Number of species 3 7 6 8 Total biomass 773 2,964 3,772 3,932 Average total biomass/species 258 494 629 492 GOODMAN & CARLETON: RODENTS 275 100 200 300 400 500 trap nights 600 700 800 100 200 300 400 500 600 700 800 trap nights Fig. 22- 1 . Species accumulation curves (top) and trap success (bottom) plotted for each elevational zone against the number of trap-nights. species, only one, Brachytarsomys albicauda, is curiously absent, thus far, from the rodent fauna in the RNI d'Andringitra (Table 22-11). This ar- boreal species is nocturnal, dens in tree holes, and, as a result, is relatively difficult to trap using standard techniques. However, it has the agreeable habit of chattering at the entrance of tree holes if disturbed during the day. On numerous occasions, in all elevational zones, S.M.G. located seemingly appropriate cavities for this species, but in no case did a Brachytarsomys pop its head out after the tree was thumped with a hard object, a technique that has worked elsewhere in Madagascar. Nor were local people who lived near the forest fa- miliar with any large arboreal rodent with a white- tipped tail. Other deficiencies in our knowledge of Mala- gasy rodents may affect the completeness of our rodent list. Of immediate relevance here, and as stressed by Carleton and Schmidt (1990), the em- pirical basis underlying comprehension of neso- myine distribution and taxonomic diversity is ex- tremely poor for an island-continent so climati cally, topographically, and floristically diverse 276 FIELDIANA: ZOOLOGY Also relevant is the want of critical systematic study of the limited museum material available. The description of Monticolomys (Chapter 2 1 ) is a telling case in point and underscores the possi- bility that still other undescribed rodents inhabit remote areas of the reserve and certainly of Mad- agascar. Excepting these caveats, which can be ad- dressed only by renewed field and taxonomic studies, extrapolation from the species accumu- lation curves and from the diversity uncovered in comparable survey efforts in eastern forest per- suades us that the rodent fauna on the eastern slopes of RNI d'Andringitra has been largely, though perhaps not exhaustively, documented. Elevation and Rodent Associations Over the limits of the transect (720-1625 m), species diversity becomes greater at higher ele- vations on the eastern slopes of the RNI d'An- dringitra. Of the nine nesomyines collected, the fewest species coexist at the 720 m locality (three) in lowland rain forest, whereas six or seven spe- cies were found at the three higher elevations, from the upper limit of lowland forest to upper montane (sclerophyllous) forest (Table 22-10), where the black rat also occurs. The bulge in spe- cies richness within montane forest associations mirrors the basic pattern found in undisturbed for- ested zones on mountains throughout the Old World tropics, whether on continental land masses or on large islands (see, e.g., Rupp, 1980; Yalden, 1988; Heaney & Rickart, 1990). Differences in Relative Densities and Rela- tive Biomass — Relative density is herein estimat- ed as the number of individuals of a given species per 100 m of trap line, including all eight lines. Because species captures seldom differed notably between the two trap lines within the same ele- vational zone (the exception at 1625 m, where line 7 yielded 4.36 rodents per 100 m and line 8 yielded 9.52 per 100 m), the results are averaged for each elevation of the survey. Overall rodent captures increase as a function of elevation, from an average 2.58 individuals per 100 m at the 720 m site to 7.12 at 1625 m ele- vation (Table 22-12). Particularly marked shifts occur between the 720 and 810 m zones and again between 1210 and 1625 m. These trends parallel results obtained from other rodent studies on trop- ical islands (see, e.g., Heaney et al., 1989; Rickart et al., 1991, 1993). On the basis of total numbers captured, Rattus rattus, Eliurus majori, E. tanala, and Nesomys ru- fus are more common in the 1625 m zone than at lower elevations (Table 22-12). The relative abun- dance of Eliurus minor was lowest at the extremes of its altitudinal distribution (at 720 m, 0.81 in- dividuals per 100 m; at 1625 m, 0.51 individuals per 100 m), but it was greater at the middle ele- vations, particularly 810 m, where the relative density measured 1.67 individuals per 100 m. The captures of Gymnuromys roberti were relatively constant (0.13-0.42 individuals per 100 m) and, on average, the lowest of any rodent species oc- curring in more than one elevational zone. Eliurus webbi showed similar measures of relative density at the two lowest elevations; there were 1.61 in- dividuals per 100 m at 720 m and 1.67 individuals per 100 m at 810 m. Collective altitudinal comparison of numbers of the various rodent species reveals an interesting relationship between relative density and ubiquity of elevational occurrence. Species found in low densities (less than one individual per 100 m) — Rattus rattus, Brachyuromys ramirohitra, Gym- nuromys roberti, and Monticolomys koopmani — are either altitudinally limited or broadly distrib- uted on the eastern slopes of the reserve. Those of intermediate abundance (1-1.5 individuals per 100 m) — Eliurus minor, E. tanala, and Nesomys rufus — are animals found at least across the band of montane-upper montane (sclerophyllous) for- est, from 810 to 1625 m. The last class (relative density > 1.5 individuals per 100 m) consists of two species, Eliurus majori and E. webbi, that re- place one another on the mountain, the latter oc- curring from 720 to 810 m and the former from 1210 to 1625 m. Some tenets of interspecific competition theory hold that when morphologically similar congeners distributed along elevational gradients co-occur, particularly those species that utilize the same as- pects of the environment, their relative abun- dances will shift over the zone of overlap. A pos- sible case involves Eliurus webbi and E. tanala, both of which occur at 810 m elevation. In the former, relative densities were similar at 720 and 810 m, but the latter demonstrated a clinal in- crease from the 810 to 1625 m localities. Al- though no reciprocal change in relative abundance of these two species was evident at the elevation of sympatry (810 m), there appeared to be a change in the portion of the forest strata used by E. tanala (Table 22-5). At 810 m, none of six E. tanala was taken on the ground, at 1210 m five GOODMAN & CARLETON: RODENTS 277 of nine (56%), and at 1625 m six of nine (67%). This increase in arboreality of E. tanala when in contact with E. webbi may be spurious, because the terrestrial incidence of E. webbi does not alter predictably in the presence of E. tanala. There is no persuasive evidence of competitive interaction between these two animals, or between any oth- ers, as measured by changes in relative numbers at zones of sympatry. Because the rodent species trapped in the re- serve range in average body weight from about 25 g (Monticolomys koopmani) to over 180 g (Ne- somys audeberti) (Table 22-2), relative abun- dances of rodents on the eastern slopes of the massif may differ when measured by relative bio- mass compared to relative density. The standing relative biomass represented for each species per elevational zone is based on the total number of individuals captured during the first 550 trap-days and summarized by total and average relative ro- dent biomass as well (Table 22-13). There is a stepwise augmentation in relative biomass of rodents related to elevation, from 773 g at 720 m, through 2,964 g at 810 m and 3,772 g at 1210 m, to 3,932 g at 1625 m (Table 22-13). This relationship parallels the increase in total number of rodents captured (Table 22-10) and em- phasizes the prevalence of more heavy-bodied ro- dents, particularly Nesomys rufus, in higher num- bers at the upper elevations. The increase in rel- ative biomass is not simply related to species di- versity but also reflects the total number of rodents captured per elevational zone. For in- stance, the 1625 m zone, with eight species, sup- ported a greater total rodent relative biomass than did the 1210 m sector, with six species; neverthe- less, the average biomass per species is actually less in the former zone as compared to the latter (Table 22-13). In terms of both rodent species richness and total relative biomass, the forest at the 1625 m zone is the most productive. Multiple Captures in Single Trap Sets — In order to infer aspects of microhabitat utilization and, to a lesser degree, the social organization of rodent species, we tabulated cases when conspe- cific animals were captured in the same trap set. The principal concern was to determine whether there were any clear patterns in the species or age and sex of conspecific animals taken in the same trap, and to use this information to infer aspects of syntopy between species and social systems. A relatively small percentage of the traps used in each elevational zone captured at least two con- specific animals (Table 22-14). No species Table 22-14. Multiple captures of rodents in a sin- gle trap within each elevational zone. No. of traps with Traps with multiple captures Elevation 720 m 810 m 1210 m 1625 m Total with Same Different species captures species species 12 27 37 28 4 11 8 12 showed a clear pattern in the sex ratio of conspe- cific animals taken in the same set, and thus noth- ing can be inferred about their social organization. In a few cases three or more individuals of E. majori, E. minor, and Nesomys rufus were taken in the same trap position. A second aspect of multiple trap captures in- volves cases of interspecific overlap in microhab- itats. When different species were obtained in the same trap set over the course of the survey, a microhabitat overlap between the species is sug- gested. At 720 m all trap sets that captured more than one individual consisted of conspecific ani- mals (Table 22-14). At 810 m, five of the 11 trap sets that caught more than two individuals result- ed in the capture of non-conspecific animals. These included: two traps in leaf litter and by holes in the ground, one of which captured a Ne- somys and Gymnuromys and the other a Nesomys and E. webbi; one set in a dense vine tangle and 1 .5 m off the ground that captured E. minor and E. tanala; one trap placed on a horizontal liana, 3 cm in circumference and 2.5 m off the ground, that trapped one E. minor and one E. webbi; and a trap on a dead 6-cm-circumference limb at 20° with one end on the ground and the other on a liana tangle that captured two E. minor and one E. webbi. At 1210 m elevation all eight traps that cap- tured more than one animal included at least two non-conspecific mammals (Table 22-14). Ground sets included: on a wet talus slope and covered with bracken fern, one Nesomys and Microgale cowani (FMNH 151777); at the base of a large tree with stilted roots, three Nesomys and one Rat- tus; on a downed, rotten, and moss-covered tree, one Nesomys and E. tanala; at the base of massive root tangle, one Gymnuromys and E. minor; and at the base of a tree with a large cavity, one E. minor and two Microgale dobsoni (FMNH 151624, 151625). Elevated sets included traps 278 FIELDIANA: ZOOLOGY placed on a large downed tree spanning the banks across a 10-m-wide river and 3 m above the water, that captured three E. majori and one Rattus; on a horizontal portion of a 3-cm-circumference arched tree trunk that was connected to the ground and a massive vine tangle about 5 m off the ground, that trapped one E. majori and one Mi- crogale soricoides (FMNH 151778); and at the junction of an exposed rock outcrop and the hor- izontal tree roots, about 2-m high, covered with moss and epiphytes, that captured two E. minor and one E. majori. In the 1625 m zone, half of the 14 trap sets that captured at least two mammals included different species (Table 22-14). Traps placed on the ground consisted of two next to downed and rotten hol- low logs, that trapped a Gymnuromys and Rattus, and the other a Nesomys, E. tanala, and Rattus; at the opening of a natural underground passage- way associated with a massive root tangle covered by soil and leaf debris, that captured four E. ta- nala and one Brachyuromys; and under a root tangle, that captured E. tanala and Microgale tai- va (FMNH 151658). Arboreal settings included two on approximately 30-cm-circumference tree trunks at 25-30° and 1.5 m off the ground, one of which linked the ground to an extensive bam- boo thicket and captured two E. tanala and one E. majori, and the other that spanned the distance between the ground and another large tree and trapped two E. minor and one Rattus; and a trap placed on a horizontal tree trunk 6 cm in circum- ference and 1 .5 m off the ground that caught one E. tanala and one E. majori. At the two highest elevational zones, 1210 and 1625 m, the number of trap sets that resulted in the capture of more than one mammal species was higher than at lower elevations (Table 22-14). In these two zones, Rattus, E. majori, E. minor, and E. tanala were often captured in the same arboreal trap sets, and there seems to have been consid- erable overlap between these species in substrate usage. For ground sets, Rattus, Gymnuromys, and Nesomys often occurred in the same microhabitat. Intra- and Interspecific Differences in Re- production— Little is known about the reproduc- tive biology of Malagasay rodents in general, nor has there been any information published on in- traspecific and interspecific variation along an ele- vational gradient to date. Information on breeding condition of the Andringitra rodents is summa- rized here (Table 22-15), but inferences about the seasonality of reproduction are necessarily limited by the brevity and timing of the field survey (5 weeks; November 14, at 720 m, through Decem- ber 18, 1993 at 1625 m). When records for all rodent populations are combined per elevational zone, a clear pattern of increasing reproductive activity as a function of altitude emerges, ranging from only 25% of the rodents sampled at 720 m to over 80% at 1 625 m (Table 22-15). Examples of Eliurus tanala and Nesomys rufus display this elevational trend most sharply, with levels of reproductive activity in- creasing from 17 individuals and 0% at 810 m, 34 animals and 73% at 1210 m, to 66 animals and 86% at 1625 m, respectively. Eliurus minor is the only species found in three or more elevational zones that lacked clear evidence of heightened re- productive activity as a function of altitude. Rat- tus rattus may show a parallel pattern, but the numbers trapped were too few to draw conclu- sions. In this species there was a disproportionate number of males captured. Similarly, the small numbers of Gymnuromys roberti and Nesomys au- deberti captured are insufficient for comment. Among species of Eliurus there appears to be considerable variation in the extent and concur- rence of reproduction, at least during the months of November and December (Table 22-15). Of the 23 E. webbi captured, two were subadults, and only three of 21 adults (13%) exhibited signs of active reproduction. The breeding status of this species' population contrasts with that of E. ma- jori, in which 19 of 22 individuals (86%) gave an indication of ongoing reproduction (males with scrotal testes and convoluted epididymides, and females lactating or carrying embryos). The num- bers of sexually active E. minor and E. tanala fall in between those of E. webbi and E. majori. Among all samples of Eliurus, few subadults were encountered; this may indicate that little breeding activity had transpired in the months preceding the survey. Although the temporal glimpse is brief, the breeding regime of rats of the genus Eliurus, and perhaps for most rodent species in the reserve, appears to be seasonal, and the inception of re- production in populations at higher elevations may precede the onset at lower elevations. The Rodent Faunas in Disturbed and Undisturbed Forest The two lower sites on the eastern slopes of the RNI d' Andringitra were chosen to enable a com- parison of levels of human activity and the pos- GOODMAN & CARLETON: RODENTS 279 4 O — On — ' ■* fS N ^ M O O — -- — — (M--PPQ in O ^ ' » ' M^n NOirim- :0 :m : no ■ £ A m2 Eco, of ^ duction Wide Fund for Nature, Gland, xvn+374 pp. 57_104 Jn Kunz> T H ed Ecology of Bats plenum Pont, S. M., and J. D. Armstrong. 1990. A study of Press, New York, 425 pp. 288 FIELDIANA: ZOOLOGY Chapter 24 The Carnivores of the Reserve Naturelle Integrate d'Andringitra, Madagascar Steven M. Goodman Abstract The eastern slopes of the Reserve Naturelle Integrate d'Andringitra hold five native {Galidia elegans, Galidictis fasciata, Cryptoprocta ferox, Eupleres goudotii, and Fossa fossana) and three introduced carnivores {Canis lupus, Felis silvestris, and Viverricula indica). The greatest diversity was at 820 m, with four native species. The reserve has one of the richest carnivore communities known from any site on the island. Resume On a recense" cinq especes indigenes de cranivores {Galidia elegans, Galidictis fasciata, Cryptoprocta ferox, Eupleres goudotti et Fossa fossana) et trois especes introduites {Canis lupus, Felis silvestris et Viverricula indica) sur le versant est de la Reserve Naturelle Integrate d'Andringitra. A une altitude de 820 m, quatre especes indigenes ont 6l6 inventorizes ce qui reprZsente la plus grande diversite specifique. La reserve recele la plus riche communaute" de carnivore connue parmi tous les sites de l'tle. Introduction During the 1993 survey of the R6serve Natu- relle Integrate (RNI) d'Andringitra no systematic work was conducted on carnivores, but some data were obtained on the locally occurring species and their elevational ranges by incidental obser- vations and a few trap captures. The only known published information on the carnivores of the re- gion was presented by Albignac (1973) and Nicoll and Langrand (1989). Wozencraft (1993) is fol- lowed for the order and taxonomy of the carni- vores. Species Accounts Family Canidae Canis lupus Linnaeus, 1758 Dogs were relatively common in the village of Ambarongy, near our 720 m camp, and they were seen with some regularity in the company of peo- ple traveling along the trail between Ambalama- nenjana and Ambatomboay. A dog was also ob- served in the tavy below our 810 m camp and along the Sahavatoy River. This species was in- troduced to Madagascar by humans. Family Felidae Felis silvestris Schreber, 1775 Feral domestic cats were found near human set- tlements and in lowland forest. Cats were regu- larly noted within 500 m of the village of Am- barongy, on both sides of the Iantara River, and near our 720 m camp. Cats were also heard and observed in the vicinity of our 810 m camp. This species is not native to Madagascar. GOODMAN: CARNIVORES 289 Family Herpestidae Galidictis fasciata (Gmelin, 1788) Subfamily Galidiinae Galidia elegans I. Geoffroy Saint-Hilaire, 1873 This diurnal animal was the most frequently en- countered carnivore on the eastern side of the RNI d' Andringitra. At least four individuals frequented our 810 m camp, where they were remarkably tame, allowing approach to within a few meters. They often raided the camp for any food left out in the open. At 720 m this species was distinctly more retiring and apparently less common. The inhabitants of Ambarongy mentioned their belief that local carnivores, including G. elegans, are a significant predator on chickens. Local people set traps at the ecotone between the forest and agri- cultural fields to trap and kill carnivores. The dif- ference in the commonness of this species be- tween the 720 and 810 m zones may be a result of human persecution near settlements. It was dis- tinctly less common at 1210 m elevation and not recorded in the 1625 m zone or above. Feces of Galidia elegans collected along a trail at 810 m contained beetle keratin and frog bones. At 720 m elevation a G. elegans fled upon my approach from the vicinity of a mist-net placed in the forest, where it had killed captured birds: two Phyllastrephus zosterops and two Terpsiphone mutata. One of the Terpsiphone was still entire; it had been dispatched by a bite to the neck and the vertebral column was broken. On two occa- sions single G. elegans were observed climbing gracefully on vines and limbs 5-10 m above the ground. Such arboreal behavior was previously reported by Rand (1935). On October 8, 1993 at 810 m a pair of G. ele- gans was observed copulating. Uninterrupted co- itus lasted about 15 minutes. Two individuals col- lected in this elevational zone included a male with partially descended testes, measuring 17 X 12 mm, and a slightly convoluted epididymis; and a female with one set of large inguinal mammae. The pelage coloration of these two individuals falls within the range of G. e. elegans, which had been previously reported from the reserve (Albig- nac, 1973) Measurements (6\ 9) — Total length: 673, 638 mm; tail: 291, 287 mm; hind foot (excluding claw): 72, 70 mm; ear: 29, 29 mm; and weight: 965, 715 g. This nocturnal animal was observed and trapped only in the 720 m transect zone. At about 2230 hours, a G fasciata was observed patrolling the area around the camp. The following morning a rice sack that contained provisions was found; it had been torn open and a quantity of food re- moved. An adult female was captured in a Na- tional live trap set on the ground at the base of a large exposed rock in slightly disturbed forest (Fig. 24-1). The subspecies G. f. striata has been reported from just east of the reserve (Albignac, 1973: fig. 5). Measurement — Weight: 520 g ($). Family Viverridae Subfamily Cryptoproctinae Cryptoprocta ferox Bennett, 1833 Inhabitants of the village of Ambarongy, near our 720 m camp, reported that C. ferox is a seri- ous menace to domestic animals, particularly chickens. None of the expedition participants ob- served this species in the RNI d' Andringitra, but within the 810, 1210, and 1625 m transect zones, its tracks were found in soft ground and mud. In the open heathland above 2000 m, feces of this carnivore were collected; they contained the re- mains of rodents Brachyuromys betsileoensis and Rattus rattus, frogs, and large insects. Albignac (1973) noted that C. ferox occurs at 2000 m on the Andringitra Massif. Subfamily Euplerinae Eupleres goudotii Doyere, 1835 This species was observed once during our field trips to the area. At dusk a single adult was ob- served at 810 m moving through an area of rela- tively thick bamboo just above the Sahanivoraky River. This is the first report of this species in the reserve. Fossa fossana (Mtiller, 1776) This nocturnal carnivore was recorded in the 720 and 810 m transect zones. In both cases it 290 FIELDIANA: ZOOLOGY Fig. 24-1. Female Galidictis fasciata captured and released in the 720 m zone. was observed at the edge of the camp or along a trail. Albignac (1973: fig. 5) reported F. fossana south of the RNI d'Andringitra in the Pic d'lvo- hibe area. Subfamily Viverrinae Viverricula indica (Desmarest, 1804) This introduced species was relatively common at lower elevations, particularly at 720 m. The in- habitants of Ambarongy reported that this animal also attacks chickens and that, with some regular- ity, it is captured in carnivore traps set around the village and nearby tavy clearings. In a house with- in the reserve and at the edge of a tavy, skins of two Viverricula, presumably trapped by local peo- ple, were found. Viverricula indica was observed numerous times soon after dusk, moving around the perimeter of our camps or on nearby trails. It was previously reported in the reserve (Nicoll & Langrand, 1989). Summary The eastern slopes of the RNI d'Andringitra hold a total of five native and three introduced carnivores (Table 24-1). Two of the introduced carnivores, Felis and Canis, are associated with human habitation, whereas the third introduced species, Viverricula indica, is not a domestic an- imal and lives away from human settlements. This carnivore diversity is one of the highest known for any rain forest area on the island (Nicoll & Langrand, 1989). The Reserve de Biosphere de Mananara, 225 km north of Toamasina, and be- tween sea level and 570 m in elevation, holds the same suite of native carnivores as the Andringitra area, plus Salanoia concolor. The Masoala Pen- ninsula, in the extreme northeastern portion of the island, has a number of carnivore species equal to that of the RNI d'Andringitra; the only difference between the two sites is that the former lacks Gal- idictis fasciata, and includes Salanoia concolor. The Pare National de Ranomafana, about 1 20 km NE of our 720 m camp, holds basically the same carnivore assemblage as the RNI d'Andringitra. GOODMAN: CARNIVORES 291 Table 24- 1 . Altitudinal distribution of carnivores on the eastern slopes of the RNI d'Andringitra. 720 810 1210 1625 m m in m Family Canidae Canis lupus* + Family Felidae Felis silvestris* + + Family Herpestidae Galidia elegans + + + Galidictis fasciata + Family Viverridae Cryptoprocta ferox + + + Eupleres goudotii + Fossa fossana + + Viverricula indica* + + Total number of native species 3 4 2 1 * Species introduced on Madagascar. The most reasonable explanation for the rela- tively large carnivore lists is related to the amount of time researchers have spent exploring the sites. Perhaps with the exception of Eupleres, all of the carnivores found in the RNI d'Andringitra are rel- atively tolerant of human disturbance. Literature Cited Albignac, R. 1973. Faune de Madagascar. 36. Mam- miferes: Carnivores. Office de la Recherche Scienti- fique et Technique Outre-Mer et Centre National de la Recherche Scientifique, Paris, 210 pp.+plates. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la Conservation et des Aires Protegees. World Wide Fund for Nature, Gland, xvii+374 pp. Rand, A. L. 1935. On the habits of some Madagascar mammals. Journal of Mammalogy, 16: 89-104. Wozencraft, W C. 1993. Order Carnivora, pp. 279- 348. In Wilson, D. E., and Reeder, D. M., eds., Mam- mal Species of the World: A Taxonomic and Geo- graphic Reference, 2nd edition. Smithsonian Institu- tion Press, Washington, D.C., 1,206 pp. 292 FIELDIANA: ZOOLOGY Chapter 25 Rapid Assessment of the Primate Fauna of the Eastern Slopes of the Reserve Natureile Integrate d'Andringitra, Madagascar Eleanor J. Sterling and Marie Gisele Ramaroson Abstract The ongoing deforestation of Madagascar's eastern forests threatens a rich array of primate species. Accurate assessments of the remaining species' geographical distribution, population sizes, and population dynamics directly influence conservation management decisions such as location and size of potential reserves as well as management of existing reserves. Intensive, long-term survey work to decipher the patterns of species distribution and of population abun- dance is ideal but currently impracticable on a large scale. Resource limitations and time constraints dictate that conservation managers use quicker methods, such as rapid biodiversity assessment techniques, to approximate these patterns over the short term. This paper discusses the results of a 6-week rapid survey of the primate fauna in the Reserve Natureile Integrate d'Andringitra. Resume Le d6boisement auquel les forets de Test de Madagascar sont confrontees constitue une menace pour une grande varied de primates. Des inventaires precis sur la distribution g6o- graphique, la tai lie de la population et la dynamique de la population des especes influencent directement les decisions de gestion de la biodiversity, comme par exemple pour ce qui con- cerne la localisation et la taille des reserves futures ainsi que la gestion des reserves existantes. Un inventaire intensif et a long terme destine" a permettre 1' interpretation des caract^ristiques de la distribution des especes et 1'abondance des populations constituerait l'outil id6al, mais cela n'est pas actuellement r6alisable sur une grande 6chelle. La limitation des ressources humaines et financieres, ainsi que les contraintes de temps forcent les responsables de la con- servation a utiliser des techniques moins lourdes, telles que les techniques d' inventaire rapide de la biodiversity, afin d'obtenir a court terme des informations. Ce document pr6sente les r£sultats d'un inventaire rapide de la faune de primates de la Reserve Natureile Integrate d'Andringitra. Introduction slopes (Perrier de la Bathie, 1927; Paulian et al., 1971). These zones support markedly different The Reserve Natureile Integrate (RNI) d'An- faunal complements, according to the few existing dringitra encompasses a variety of ecological survey reports (Perrier de la Bathie, 1927; Nicoll zones due to the major vegetational and climatic & Langrand, 1989). For instance, the list of pri- differences between eastern and western-facing mates found in the reserve includes those widely STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 293 Table 25-1. The 15 primate species and subspecies of the RNI d'Andringitra. Data include lemurs previously reported from the reserve's eastern and western slopes (Perrier de la Bathie, 1927; Nicoll & Langrand, 1989) and those observed during the 1993 inventory on the eastern slopes. Taxon Eastern and western slopes Eastern slopes — 1993 inventory 720 m 810 m 1210 m 1625 m Lemur catta Avahi laniger Varecia variegata variegata Eulemur fulvus fulvus Microcebus rufus Eulemur rubriventer Lepilemur mustelinus Eulemur fulvus albocollaris Eulemur fulvus rufus Hapalemur aureus Hapalemur griseus griseus Hapalemur simus Propithecus diadema edwarsi Cheirogaleus major Daubentonia madagascariensis Total numbers of species and subspecies distributed in eastern rain forests, such as Varecia variegata variegata, as well as those found in dri- er forests of the central plateau and the southwest, such as Lemur catta (Table 25-1). Yet, despite its breadth, we believed the list to be incomplete. Many species, such as Hapalemur griseus, Pro- pithecus diadema, and Cheirogaleus major, were not listed, although they are found in nearby rain forests to the north, (Pare National [PN] de Ran- omafana), or to the south (Reserve Speciale [RS] du Pic d'lvohibe), or to the southeast (RS de Man- ombo) (Table 25-2). Nor were the rare forms, such as Hapalemur aureus, listed in the existing surveys. It is not known to what extent altitude or forest type affects the distribution and density of primate populations, although several researchers have be- gun to address these issues (Skorupa, 1986, 1987, 1988; Johns and Skorupa, 1987; O'Connor et al., 1987; Oates et al., 1990; Thomas, 1991; Ganz- horn, 1994b). In this study, we attempted to com- pare species composition and densities between forests at different elevations and between dis- turbed and undisturbed forests at approximately the same elevation. We were working as part of a team of scientists looking at these issues across taxa in the RNI d'Andringitra. Rapid survey tech- niques for primate populations, however, are in their infancy. We were not sure that they would be robust enough to use for reliable estimates of density and diversity in primates. A variety of techniques have been employed to estimate pri- mate densities over longer periods of time (see National Research Council, 1981). The line tran- sect method is the most frequently used (Edwards, 1992), but we have yet to determine the number of transects necessary for accurate density esti- mates in a region. Results of this current study were therefore used to clarify information on primate distribu- tional boundaries, to provide a basis for evaluat- ing potential ecological correlates in species dis- tributions, to evaluate rapid assessment techniques for primate populations, and to provide recom- mendations for future short-term survey work. Methods Transects were conducted at four elevations (720, 810, 1210, and 1625 m) between September 23 and October 31, 1993. Two of these zones were in some of the lowest remaining forest on the eastern slopes of RNI d'Andringitra, one in a partially degraded forest (720 m) and the other in a relatively intact forest (810 m) located about 2.5 km (by direct line) to the east. The other two tran- sect zones were in undisturbed forest. The line transect method (National Research Council, 1981) was used to estimate density at each site. At the outset, we laid out linear tran- 294 FIELDIANA: ZOOLOGY Table 25-2. List of primate species observed in re- serves within a 100 km radius of RNI d'Andringitra. Site RS RS Pic PN Manombo d'lvohibe Ranomafana (altitudinal (aititudinal (altitudinal range range 775- range Species 1-137 mi 2060 m) 800-1200 m) Efa X Eff ... Efr ... Er ... Ha ... Hgg X Hs ... Lc ... Mr X Pde ... Vvv ... Al X Cm X Dm X Lm ... Total 6 X X X X X X X X X X X 12 Efa = Eulemur fulvus albocollaris; Eff = Eulemur fulvus fulvus; Efr = Eulemur fulvus rufus; Er = Eulemur rubriventer. Ha = Hapalemur aureus; Hgg = Hapale- mur griseus griseus; Hs = Hapalemur simus; Lc = Le- mur catta; Mr = Microcebus rufus; Pde = Propithecus diadema edwarsi; Al = Avahi laniger; Cm = Cheiro- galeus major. Dm = Daubentonia madagascariensis; Lm = Lepilemur mustelinus. Data from: Buettner-Jan- usch and Tattersall (1985), Jenkins (1987), Nicoll and Langrand (1989), Mittermeier et al. (1992), and Good- man and Schulenberg (unpubl.). sects along compass bearings regardless of the terrain, but this was too time consuming and fur- thermore the transects often traversed impassable terrain. We therefore followed existing trails left by bush pigs, cows, and humans, selecting trails to cover equal areas of ridge, slope, and valley habitats. Trails were marked every 10 m with flag- ging for reference during the surveys. Two re- searchers experienced in observing Malagasy pri- mates walked either singly or in tandem along transects of varying lengths (600-3000 m) for ap- proximately 5 hours. Diurnal samples started at 0530, whereas nocturnal surveys started at 1800, 2100, or 0100 hours. We walked approximately 1 km per hour during the day, but we had to slow to 750 m per hour at night due to the difficult terrain. Diurnal and nocturnal animals were de- tected by sound, sight, or movement of branches. In addition to being able to distinguish animals at night by their shape/color differences from the background, one can also locate them by the re- flection of their tapetum lucidum. We paused every 100 m or so to watch and listen for primates, but we never deviated from the path in search of them. Each transect was re- peated a minimum of six times, three times each for diurnal and nocturnal transects. Whenever possible during sequential samples along a tran- sect, we tried to walk in the opposite direction from the previous sample to reduce potential bi- ases (Brockelman & Ali, 1987). Similarly, repeat samples were spaced 48 hours apart when feasi- ble. Due to time constraints, we were forced to undertake sampling under conditions not normally ideal for census work (light rain or fog). We did not conduct samples when our viewing distance was restricted to less than 15 m. For each animal or group seen, we noted spe- cies, mode of detection (sight, vocalization, or sound of movements), number and composition of group, time of contact, position on transect, height from ground, habitat type, behavior, distance from observer to first animal seen, and perpendicular distance from trail to first animal seen. Diurnal animals were defined as members of a "group" if they were observed within 50 m of a conspe- cific animal. Because nocturnal animals are cryp- tic, and therefore more difficult to locate than di- urnal species, we always counted them as individ- uals. We spent no more than 10 minutes with any single cluster of animals trying to determine spe- cies, group composition, and pelage characters. A number of different models exist for calcu- lating density from raw data with the transect method (Charles-Dominique, 1977; Burnham et al., 1980; Whitesides et al., 1988), all of which have their drawbacks (see Whitesides et al., 1988). We used three density estimate methods: two absolute and one relative. The two absolute density models were used in this study because they had been employed previously by primate researchers in Madagascar (Ausilio & Raveloari- noro, 1993; Ganzhorn, 1994a). They involve de- riving a "detection function" from the distribu- tion of observed perpendicular trail-to-animal dis- tances (calculated for each species separately) that is used to estimate the area censused. Absolute Density Estimate Method I (AMI) First, we derived transect width by grouping detection distances for each species into eight fre- STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 295 quency classes between 0 and 40 m from the trail and selecting the distance beyond which obser- vations dropped by at least half (Charles-Domi- nique, 1971; Charles-Dominique & Bearder, 1979; Ausilio & Raveloarinoro, 1993). We mul- tiplied twice this transect width by the average length of a transect to derive total area. The av- erage number of groups (for diurnal species) or individuals (for nocturnal species) observed with- in the derived transect width was divided by total area to calculate density. In principle, as distance from the trail increases, frequency of detection should level off, then drop (Cant, 1978). However, this depends upon the acquisition of an adequate sample of detection distances; it might not be pos- sible to collect such a sample in rapid assessment surveys. these assumptions are violated when working with unhabituated, cryptic primate populations, re- searchers have proposed another index of animal density that is less accurate, but also less exacting (see Cant, 1978). This index, called relative den- sity, is calculated as the number of animals ob- served per unit time or distance, and it is used to compare populations over time or in different ar- eas. In this study, we chose to determine relative density as animals per km surveyed. Results General Absolute Density Estimate Method II (Willi Next, following the methodology of Ganzhorn (1992, 1994a), we took twice the average of the trail-to-animal distances for each species observed on three or more transects and multiplied it by the distance covered to give total surface area sur- veyed per transect. Density estimates were then made by multiplying group size estimates by the number of groups sighted on a transect and divid- ing by the total area for each transect. All obser- vations made on the transect were included, not just those falling within the averaged transect width. It is difficult to estimate distances and number of animals by vocalizations, and there is a greater possibility that animals could be counted twice (through counting both vocal and visual contact). One assumption underlying the transect method is that animals are not counted twice (Burnham et al., 1980; Whitesides et al., 1988; Buckland et al., 1993). Therefore, we did not include in our cal- culations (for either diurnal or nocturnal surveys) animals that were heard but not seen. Relative Density Method The transect method includes other critical as- sumptions: that animals located directly over the transect are always detected; that they are detected at their original location, prior to movement in response to the observer; and that distances are accurately recorded. Due to the fact that some of In total, seven diurnal and five nocturnal spe- cies were observed in the four transect zones (Ta- ble 25-1). Only one diurnal (Eulemur rubriventer) and two nocturnal species (Cheirogaleus major and Microcebus rufus) were seen in all four zones. Overall species and subspecies richness for the RNI d'Andringitra was the highest yet record- ed in any Malagasy forest, if one includes species from the eastern and western slopes (Tables 25-1 and 25-2). Detection Distances and Density Species-specific detection distances were cal- culated by elevation rather than transect because there were far too few observations per transect (Table 25-3). Even so, they did not reach a plateau and drop off as expected. Nevertheless, we cal- culated density on the few species for which we could reasonably estimate transect width (Table 25-4). In comparing density estimates derived from AMI with those from AMII, one can see that the two methods produce similar results for the ele- vational zones, although the relative trends be- tween sites differ (Table 25-4). However, AMI and AMII are correlated significantly only among estimates for nocturnal species (r = 0.83, P = 0.02, N = 7). No other correlation between the different estimates approaches significance, nei- ther for the pooled data set nor when species are subdivided into group-living versus solitary spe- cies. 296 FIELDIANA: ZOOLOGY Table 25-3. Frequency of detection by species for each elevation, by distance from the trail. Distance class (m) Transects 1-5 6-10 11-15 16-20 21-25 26-30 31-35 36-^0 720 m zone Species Efa 2 Er H? Al 4 Cm 5* Lm 1 Mr 9 810 mi zone Species Efa 1 Er 1 Ha 1 Hgg Hs 2 1 H? Cm 13* Mr 14 1210 m zone Species Efa 3 Efr Er Ha 1 Hgg H? 1 Al 1 Cm 8 Lm Mr 12* 1625 m zone Species Er Ha 1 Hgg Al 2 1 Cm 3 Lm 1 Mr 20* 1 1 2* 2 2* 1 3* 1 2* 1 Note: Data for diurnal species represent number of groups observed; for nocturnal species they represent the number of individuals. Species abbreviations are as in Table 25-2. H? = unidentified Hapalemur. * Detection distance used to calculate transect widths in Absolute Method I. Effects of Habitat Disturbance The total number of species observed at the 720 m disturbed and the 810 m undisturbed sites re- mained relatively constant, although the composi- tion of observed species changed (Table 25-1). Group size varied little between these sites for the two species that we observed in both areas (Table 25-5). The average density of Eulemur fulvus al- bocollaris was lower in disturbed than undisturbed habitat using AMI, but AMD estimates showed little difference for this species. AMI and AMD density estimates showed opposite trends for Cheirogaleus major, there was lower density in the disturbed hab- itat according to AMI and lower density in the un- disturbed habitat according to AMII. The density of Microcebus rufus was higher in the undisturbed habitat according to AMD, but it remained relatively similar using AMI. Relative density was higher in the undisturbed that the disturbed sites for all species STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 297 Table 25-4. Average density estimates and standard deviations for primate fauna in the RNI d' Andringitra during September and October 1993, calculated for transects following Absolute Density Estimate Method I (AMI), Absolute Density Estimate Method II (AMII), and Relative Density (RD). Elevations (m) 720 810 N AMI AMII RD N AMI AMII RD Efa 5 140 ± 125 110 ±56 3.0 ± 1.6 5 108 ± 84 123 ± 82 2.4 ± 1.6 Er 3 102 ± 35 0.8 ± 0.2 Hgg Hs 3 96 ± 11 0.5 ±0.1 Ha 3 43 ±21 0.3 ± 0.2 Cm 3 51 ±20 233 ± 118 0.7 ± 0.4 5 115 ±64 156 ± 47 Mr 3 73 ±47 118 ±72 1.5 ±0.9 5 94 ±54 254 ± 146 Note: Species abbreviations are as in Table 25-2. Estimates are given as individuals per km2 for AMI and AMII and as individuals per km for RD. N = number of transects. found in both areas except for Eulemur fulvus al- bocollaris (Table 25-4). Effects of Elevation Changes in elevation did not affect species richness for the two lowest transect zones, but there were fewer species observed at the 1625 m zone. Species composition varied between eleva- tions. Eulemur fulvus albocollaris density de- creased with elevation using AMI but increased using AMII. Cheirogaleus major density de- creased with elevation according to AMI but re- mained relatively constant using AMII. We were unable to calculate the density of this species at the 1625 m elevational zone because it was ob- served on fewer than three transects. Microcebus rufus density increased with elevation using AMI and AMII. Relative density was higher at the 1210 m elevation than at 810 m for all species but Microcebus rufus. Species Accumulation Curves No new diurnal species were accrued after 16 hours of observation in any of the elevation zones (Fig. 25- la). For nocturnal lemurs, no additional species were seen in the 720 and 810 m zones after the first 3 hours of observation (Fig. 25- lb), whereas in the 1210 and 1625 m zones 7 to 8 hours were needed to reach a plateau. Pelage and Morphological Characteristics of Species The following pelage characteristics, which re- mained relatively similar between transect zones, were derived from field observations of lemurs in RNI d' Andringitra. Wherever possible, we used terms for pelage color following Tattersall (1982) in order to standardize observations. In several in- stances we were not able to observe all pelage characteristics for both sexes; only those which were clearly distinguished are noted in this article. Characters associated with sexual dimorphism are restricted to diurnal species. Diurnal species Eulemur fulvus albocollaris — Male: upper parts grey, sometimes a black stripe running down center of back; underparts rust-colored; face grey or black; forehead black; sometimes white above eyes; head black, sometimes white or rust-colored; beard white or orange, thick or flat; tail grey or grey-black, base of tail reddish. Female: upper parts rust-colored; underparts lighter; rust-brown extremities; face grey, black or light rust-colored; sometimes white above eyes; head grey or black; muzzle white at tip; beard grey or orange, less full than males; tail rust-colored with white tip. Eulemur fulvus rufus — Male: upper parts grey; underparts pale grey; scrotum black; face black; black continues up between eyes to crown; head cap rust-colored; whitish patch above eyes; beard and throat grey; dark pygal patch; tail dark- 298 FIELDIANA: ZOOLOGY Table 25-4. Extended. Elevations (m) 1210 1625 N AMI AMII RD N AMI AMII RD 98 ± 34 101 ± 57 82 ± 18 155 ± 65 192 ± 79 130 ±73 90 ±25 174 ±40 299 ± 124 3.3 ± 1.3 1.0 ±0.6 0.4 ± 0.1 71 ± 19 138 ±16 0.8 ±0.1 235 ± 125 397 ± 181 2.5 ± 1.2 ens at tip. Female: upper parts reddish-brown; un- derparts grey; muzzle black; white patches above eyes; crown grey; tail rust-colored, darker at tip. Evlemur rubr/venter — Male: upper parts dark brown; underparts rust-colored; face and muzzle dark, white crescent under both eyes; tail black. Female: upper parts dark brown turning to rust at extremities; underparts white; white throat reach- ing up to base of muzzle. Propithecus diadema edwarsi — Male and fe- male: head and shoulders black; white belt ex- tends from the reddish-brown ventrum to the back, where it is divided in two by a thick black patch; tail black. Hapalemur griseus griseus — Male and female: upper parts dark grey to olivaceous; underparts light grey; tail dark grey. Ears rounded and cropped close to the head. Muzzle short and somewhat pointed. Smallest of known Hapalemur species. Hapalemur aureus — Male and female: medi- um-sized Hapalemur with upper parts olive- brown; underparts yellow; dark, short, broad face with golden-yellow eyebrows, cheeks and throat; ears short, flush with head and grey tipped; head cap olive brown; reddish-gold flanks and thighs; tail gold-brown, with tip darker for the last one- fifth to one-third. Hapalemur simus — Male and female: large- bodied Hapalemur with upper parts charcoal-grey with a reddish tinge; underparts, throat, and end of muzzle grey or white; face dark, muzzle short and broad; large, grey-tufted ears; lower back and flanks lighter grey; pyral patch yellowish; tail grey, turning to black at tip. Hapalemur sp. or color morph of H. simus — A fourth type of Hapalemur was observed at the first three elevation zones: large-bodied (same size as typical H. simus), upper parts, ventrum, head, and throat uniformly deep golden red; dark face, short, broad muzzle; large ears, no tufts; no pygal patch; tail golden red with 10 cm grey seg- ment at tip. Nocturnal Species Avahi laniger laniger — Sexes not distin- guished: dorsum grey to grey-brown; foreparts have rust wash; white underparts, neck, cheeks, and thighs; cap reddish; dark face; white triangle above eyes; pale pygal patch; tail rusty red, darker distally. Cheirogaleus major — Sexes not distinguished: upper parts grey-brown; white underparts; marked dark rings around eyes; grey-brown head; white patch between eyes; medium-sized, pointed naked ears; pointed, dark muzzle with reddish nostrils; grey-brown tail. Daubentonia madagascariensis — Not observed in detail. Lepilemur mustelinus — Sexes not distin- guished: upper parts remarkably variable between individuals, from light brown to chestnut brown; dark dorsal stripe present in some individuals; un- derparts pale grey or brown; ears large and naked; tail darkens distally. The taxonomy of this genus is currently a confusion of species and subspecies. The species L mustelinus is cited here following Ishak et al. (1992). M/crocebus rufus — Sexes not distinguished: upper parts reddish-brown with thin dorsal black stripe on some individuals; underparts off-white; tail reddish-brown. STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 299 diurnal lemurs 15 person hours nocturnal lemurs person hours Fig. 25-1. Species accumulation plots as a function of person-hours for primate censuses in RNI d'Andringitra, Madagascar, for diurnal and nocturnal lemurs. Discussion Results from this study show that 10-day rapid assessment surveys are useful in determining spe- cies richness, but that longer or more intensive sampling efforts are necessary for determining density and abundance. The majority of the pri- mate species observed at each site in the RNI d'Andringitra were detected within 16 hours for diurnal species and within 6 hours for nocturnal species. Only the rarer species, such as Propithe- cus diadema edwarsi or some of the Hapalemur species, may not be detected until much greater effort is given to sampling. For instance, Propi- thecus diadema was observed once in the 720 m zone but did not show up on transects at that el- evation, even after more than 20 diurnal census hours. 300 FIELDIANA: ZOOLOGY Table 25-5. Mean and standard deviations of group sizes for primate species detected in RNI d'Andringitra between September and October 1993. Elevation ( m) Species 720 810 1210 1625 Efa 4.4 ± 1.6(10) 3.7 ± 1.1 (8) 5.3 ± 1.0(9) ND Efr ND ND 4.0 ± 0 (1) ND Er 2 ±0 (1) 2.5 ± 0.5 (5) 3.0 ± 0.8 (5) 3.0 ±0 (1) Ha ND 2.3 ± 0.5 (3) 3.0 ± 0.7 (4) 3.0 ± 0 (2) Hgg ND 1.3 ±0.5(3) 2.0 ± 0 (2) 2.2 ± 0.4 (5) Hs ND 2.0 ± 0 (3) ND ND H? 1.0 ±0 (1) 2.0 ± 0 (2) 2.5 ± 0.5 (2) ND Pd ND 5.0 ±0 (1) ND ND Note: Species abbreviations are as given in Tables 25-2 and 25-3. ND = species not observed on transects. Sample size is given in parentheses. In total, seven diurnal and five nocturnal spe- cies were observed in the four elevational zones, making the eastern slopes of the reserve one of the richest areas of primate diversity in Madagas- car. When one includes the species observed on the western slope as well (Table 25-1), the RNI d'Andringitra has a total of 15 primate species and subspecies. No Varecia were seen or heard during this expedition. Varecia are found in low densities in the PN de Ranomafana (White, 1989) and therefore may be present in the RNI d'An- dringitra. They might have been detected with more intensive sampling efforts. All three described species of Hapalemur were found in the reserve. Hapalemur aureus, a re- cently discovered species (Meier & Rumpler, 1987; Meier et al., 1987), was previously found only in the PN de Ranomafana, where the popu- lation has been estimated at 1,000 individuals (P. Wright in Mittermeier et al., 1992). The individ- uals in the RNI d'Andringitra therefore represent an important repository of genetic diversity for the conservation of this species. Hapalemur simus is similarly currently known only from patchily distributed populations around PN de Ranoma- fana and Kianjavato (Harcourt & Thornback, 1990). However, H simus was previously wide- spread: subfossils have been discovered in north and central Madagascar (Mahe\ 1976; Vuillaume- Randriamanantena et al., 1985; Godfrey & Vuil- laume-Randriamanantena, 1986), and researchers identified H. simus-like calls in the Mananara re- gion, about 600 km NNE of PN de Ranomafana (Nicoll & Langrand, 1989). Pelage characteristics of the Hapalemur species observed in RNI d'Andringitra correspond with known descriptions (Gray, 1870; Forbes, 1894; Hill, 1953; Tattersall, 1982; Meier et al., 1987), with one exception. A large-bodied Hapalemur was observed for several minutes on five occa- sions in different light regimes with a uniform deep golden-red color (as opposed to the typical charcoal-grey of same-sized H. simus) and no py- gal patch. Individuals of this morph were found at the same sites as typical H. simus, suggesting that either there are four species of Hapalemur in the RNI d'Andringitra, one of which is unde- scribed, or that the pelage coloration of H. simus is quite variable. Village elders in forests near Vondrozo, located 70 km W of Farafangana and 70 km SE of our transect in the RNI d'Andrin- gitra, report the existence of a large, red-colored bamboo lemur (Wright et al., 1987). The question of the species limits of this large, golden-red col- ored Hapalemur is in need of further research. A melanistic form of Propithecus diadema ed- warsi, once recognized as a separate subspecies P. d. holomelas (but see Tattersall, 1986), was proposed to exist in the westernmost of the two forest strips that parallel the east coast of Mada- gascar (Milne Edwards & Grandidier, 1875). It appears that all the specimens of the melanistic form come from a forest called Nandihizana, about 150 km due N of RNI d'Andringitra (Tat- tersall, 1986). The few Propithecus diadema in- dividuals seen in the RNI d'Andringitra resem- bled those found in PN de Ranomafana to the north. No melanistic individuals were observed. The RNI d'Andringitra lies at the distributional boundaries of a number of primate species. It is situated near the western limit of the range of Va- recia variegata variegata; the southern limit of the range of Eulemur rubriventer, E. fulvus rufus, and E. f. collaris; and possibly at the northern limit of the range of E. f. albocollaris, whose dis- tribution is not well known. It is unusual that the STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 301 faunal list for RNI d'Andringitra includes E. ful- vusfulvus (Nicoll & Langrand, 1989), because the reserve lies significantly to the south of this sub- species' presumed southern boundary at the Man- goro River (Tattersall, 1982). Six subspecies of E. fulvus are now recognized, yet numerous speci- mens found in museum collections (Meyers & Tattersall, 1988) and individuals seen in the wild (Sterling, pers. obs.) cannot be easily allocated to one of these subspecies. The Eulemur fiilvus morphs found in the reserve proved to be difficult to assign to subspecies. The major distinguishing feature between E. f. albocollaris and E. f. collaris is beard color, which is white in the former and orange in the latter (Tattersall, 1982). On the east- ern slopes of the reserve, we observed some males with white and others with orange beards in the same group. As noted previously (Tattersall, 1982), females also showed variation in beard col- or (from grey to orange in this study), further con- fusing our ability to assign them to subspecies. Finally, we observed one group of four E. fulvus at 1210 m that resembled most closely E. f. rufus in that no adult had a well-defined beard. It is possible that this group migrated from the western portion of the reserve, where E. f. rufus popula- tions are found. Alternatively, it could represent hybrids of two or more E. fulvus subspecies. Eulemur fulvus albocollaris and E. f collaris are reported to have different numbers of chro- mosomes (Buettner-Janusch & Hamilton, 1977; Rumpler et al., 1985; Rumpler, 1990), and in prin- ciple if fertilization indeed takes place, they should not produce fertile hybrid offspring. We tentatively allocate the populations we saw on the eastern slopes to E. f albocollaris because there were more males with white beards than with or- ange ones, but we urge further exploration of the differences between these geographical forms, particularly their distributional boundaries and phenotypic variation, in order better to understand their relationships. Group size in Hapalemur aureus remained at approximately three adults, although we noted a group of five individuals twice in the 1210m zone outside of the transect surveys. This agrees with previous observations on group size in the species (Meier et al., 1987; Wright et al., 1987). Eulemur rubriventer groups mainly consisted of an adult male, an adult female, and a subadult, similar to observations made at other sites (Dague & Petter, 1988; Overdorff, 1993). The group size recorded for Hapalemur griseus griseus in this study is low in comparison with those of most other studies, where group size ranged from one to eight (Petter & Peyrieras, 1970; Pollock, 1986; Wright, 1986). It is not known if the low numbers reflect sam- pling error through underestimation due to miss- ing members of the group or if group sizes in RNI d'Andringitra are truly smaller. Small group sizes could result from several factors, including in- creased hunting pressure in the reserve, increased competition with the other Hapalemur species, or seasonal variations in group size. In a few cases, it was difficult to explain the absence of species from a particular elevational zone. For instance, the absence of Eulemur fulvus individuals at the 1625 m zone was surprising in that there was no sign of decrease between the 720 and 1210 m elevations and because this spe- cies is relatively common in the other elevational zones. Daubentonia was not recorded at the 1210 m site. This may be because Canarium (Bursera- ceae) was absent above 1000 m (see Chapter 4). Seeds from this genus constitute a major part of the diet of Daubentonia elsewhere in Madagascar (Sterling, 1994), and some researchers suggest that the aye-aye's distribution is closely tied with that of Canarium (Iwano et al., 1991). Signs of Daubentonia increased again at 1625 m, where there was a greater prevalence of bamboo. Here there were signs of aye-aye feeding on bamboo, as described by Duckworth (1993). It is not known if these observations represent real changes in aye-aye density at the different ele- vations or if the aye-ayes in this region are feed- ing on foods that do not leave obvious signs. Several sampling artifacts may confound data on species density and composition. For instance, sampling time differed in each site, and climatic conditions became progressively worse towards the end of the expedition. For logistical reasons, trails were freshly cut into the 810, 1210, and 1625 m elevational sites immediately prior to the census work; this may have disturbed some of the primate species. In addition, old slash-and-burn fields were noted less than 1 km from the 810 m "undisturbed" camp. Therefore the differences between the 720 and 810 m sites in terms of forest disturbance (Chapter 4) may have been less mean- ingful for lemurs, who can range over several ha (see Harcourt & Thornback, 1990), than for other animals surveyed. Finally, rain and fog interfered with sampling on at least half of the nights in the 1210 and 1625 m elevations and on more than half of the days at the 1625 m elevation. Conse- quently, the species accumulation curve for diur- 302 FIELDIANA: ZOOLOGY nal primates in the 1625 m elevational zone (Fig. 25-1) may not have reached a plateau. Generalizations about the effects of forest dis- turbance and elevation on primate densities were problematic because we did not have enough ob- servations. Inadequate sample sizes prevented us from determining detection distances by transect, as was undertaken by Ausilio and Raveloarinoro (1993), and even by elevation, in some cases. The absence of a distinct plateau in most cases sug- gests that more data would be necessary to derive meaningful transect widths. Similarly, we were unable to calculate density for several transects because we did not observe species on least three transects. Another problem with density estimates is that group sizes based on limited sightings of unhabituated primates are generally underestimat- ed (see Defler & Pintor, 1985). In addition, the fact that the surveys were not undertaken simul- taneously at the different elevational zones means that seasonal differences may have confounded the data. In particular, Cheirogaleus major under- goes a period of torpor from May or June until mid-September (Starmuhler, 1960; Martin, 1972) and might therefore have been underestimated in the first two transect zones. Trends in species density over elevation, drawn from AMI and AMII (where feasible to calculate), were inconsistent and lacked concurrence. In ad- dition, in this study at least one of the assumptions critical to absolute density estimates derived from transects may have been violated. Consequently, it is difficult to compare absolute density esti- mates between sites. The data from this research, however, seem to support the recent studies show- ing that lemurs are not very good indicators of habitat change (see Ganzhorn, 1994b). In sum, the results from this survey suggest that 10 days of work by two primatologists in a humid forest site was sufficient for determining species richness but not adequate for estimating density and abundance in primate species. In order to ad- equately estimate density, more person-hours are needed. This could be achieved through either a longer stay or a larger team. No study on mini- mum sampling effort exists for density studies in lemur populations, but a few longer term studies exist on other primates. One study suggests that 120 km must be sampled to obtain adequate ac- curacy (Edwards, 1992). This would take two pri- matologists a minimum of 30-40 days to sample each site. Another study specifies that 20-30 rep- etitions are necessary along the same 4 km tran- sect to achieve optimal precision (National Re- search Council, 1981). Again, this would take 20- 30 days for just one data point, because each tran- sect only represents one point. Rapid assessments of primate species' geo- graphical distribution, population sizes, and pop- ulation dynamics are critical to ensuring swift and effective conservation management decisions (Beattie et al., 1991; Thomas, 1991; Oates, 1994). We must concentrate our immediate efforts on es- tablishing minimum requirements for accurate rapid assessments in order to better focus limited time and resources. Acknowledgments Partial support for this work was provided by a North Atlantic Treaty Organization (NATO) post-doctoral fellowship to E.J.S. We thank the World Wide Fund for Nature (WWF) for funding the field portion of this research. Comments by Anna Feistner, Jorg Ganzhorn, Steven Goodman, and Claire Hemingway improved the manuscript. Literature Cited Ausilio, E., and G. Raveloarinoro. 1993. Status et densites des especes de 16muriens de la region d'Ant- salova (ouest de Madagascar). UNESCO Publication 507/INT/40, Antananarivo, Madagascar, 40 pp. Beattie, A. J., J. D. Maier, and I. Oliver. 1991. Rapid biodiversity assessment: A review, pp. 4-14. In Mar- gules, C. R., and M. P. Austin, eds., Nature Conser- vation: Cost Effective Biological Surveys and Data Analysis. CSIRO, Australia, 365 pp. Brockelman, W. Y., and R. All 1987. Methods of surveying and sampling forest primate populations, pp. 23-62. In Marsh, C, and R. A. Mittermeier, eds., Primate Conservation in the Tropical Rain Forest. Alan R. Liss Inc., New York. Buckland, S. T, D. R. Anderson, K. P. Burnham, and J. L. Laake. 1993. Distance Sampling: Estimating Abundance of Biological Populations. Chapman and Hall, London, 446 pp. Buettner-Janusch, J., and A. E. Hamilton. 1977. Chromosomes of Lemuriformes part IV: Karyotype evolution in Lemur fulvus collaris. American Journal of Physical Anthropology, 50: 363-365. Buettner-Janusch, J., and I. Tattersall. 1985. An annotated catalogue of Malagasy primates (Families Lemuridae, Indriidae, Daubentoniidae, Megaladapi- dae, Cheirogaleidae) in the collections of the Ameri- can Museum of Natural History. American Museum Novitates, 2834: 1-45. Burnham, K. P., D. R. Anderson, and J. L. Laake. 1980. Estimation of density from line transect sam- STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 303 pling of biological populations. Wildlife Monographs, 72: 1-202. Cant, J. G. H. 1978. Population survey of the spider monkey, Ateles geoffroyi at Tikal, Guatemala. Pri- mates, 19: 525-535. Charles-Dominique, P. 1971. Eco-6thologie des prosi- miens du Gabon. Biologica Gabonica, 7: 121-228. . 1977. Ecology and Behaviour of Nocturnal Prosimians. Duckworth, London, 277 pp. -, and S. K. Bearder. 1979. Field studies of lo- risid behavior: Methodological aspects, pp. 567-629. In Doyle, G. A., and R. A. Martin, eds., The Study of Prosimian Behavior. Academic Press, New York, 696 pp. Dague, C, and J. J. Petter. 1988. Observations sur le Lemur rubriventer dans son milieu naturel, pp. 78-89. In Rakotovao, L., V. Barre, and J. Sayer, eds., L'equi- libre des Ecosystemes Forestieres a Madagascar. Actes d'un Seminaire International. IUCN, Gland, 338 pp. Defler, T. R., and D. Pintor. 1985. Censusing pri- mates by transect in a forest of known primate density. International Journal of Primatology, 6: 243-259. Duckworth, J. W. 1993. Feeding damage left in bam- boos, probably by aye-ayes (Daubentonia madagas- cariensis). Folia Primatologica, 14: 927-931. Edwards, A. E. 1992. The diurnal primates of Korup National Park, Cameroon: Abundance, productivity and polyspecific associations. Ph.D. diss., University of Florida, Gainesville, 115 pp. Forbes, H. O. 1894. A Hand-book to the Primates, two volumes. W. H. Allen and Co., London, vol. 1-286 pp + vol. 2-296 pp. Ganzhorn, J. U. 1992. Leaf chemistry and the biomass of folivorous primates in tropical forests: Test of a hypothesis. Oecologia, 91: 540-547. . 1994a. Les Lemuriens, pp. 70-72. In Good- man, S. M., and O. Langrand, eds., Inventaire Biolo- gique Foret de Zombitse. Recherches Pour le Devel- oppement. Serie Sciences Biologiques N. Special. Centre d' Information et de Documentation Scienti- fique et Technique, Antananarivo, Madagascar, 106 pp. . 1994b. Lemurs as indicators for habitat change, pp. 51-56. In Thierry, B., J. R. Anderson, J. J. Roeder, and N. Herrenschmidt, eds., Current Pri- matology. Universite Louis Pasteur, Strasbourg. Godfrey, L., and M. Vuillaume-Randriamanantena. 1986. Hapalemur simus: Endangered lemur once widespread. Primate Conservation, 7: 92-96. Gray, J. E. 1870. Notes on Hapalemur simus, a new species lately living in the Gardens of the Society. Proceedings of the Zoological Society of London, [1870]: 828-831. Harcourt, C, and J. Thornback. 1990. Lemurs of Madagascar and the Comoros. IUCN, Gland, 240 pp. Hill, W. C. O. 1953. Primates: Comparative Anatomy and Taxonomy. Volume I: Strepsirhini. New York In- terscience Publishers, Inc., Edinburgh, 798 pp. Ishak, B., S. Warter, B. Dutrillaux, and Y Rumpler. 1992. Chromosomal rearrangements and speciation of sportive lemurs (Lepilemur species). Folia Primatolo- gica, 58: 121-130. Iwano, T, R. Randalana, and G. Rakotoarisoa. 1991. Ecology of the aye-aye (Daubentonia madagascarien- sis), I. Distribution, pp. 41-42. In Ehara, A., and T. Kimura, eds., Primatology Today. Elsevier, Amster- dam. Jenkins, P. D. 1987. Catalogue of primates in the British Museum (Natural History) and elsewhere in the Brit- ish Isles. Part IV: Suborder Strepsirrhini, including the subfossil Madagascan lemurs and Family Tarsiidae. British Museum (Natural History), London, x+189 pp. Johns, C, and J. P. Skorupa. 1987. Responses of rain forest primates to habitat disturbance: A review. In- ternational Journal of Primatology, 8: 157-191. Mahe, J. 1976. Craniometrie des Lemuriens: Analyses Multivariates — Phylogenie. Memoires du Museum National d'Histoire Naturelle, Paris, C, 32: 1-342. Martin, R. D. 1972. A preliminary field study of the Lesser Mouse Lemur (Microcebus murinus J. F Miller 1777). Zeitschrift fur Tierpsychologie, Suppl. 9: 43- 89. Meier, B., and Y Rumpler. 1987. Preliminary survey of Hapalemur simus and of a new species of Hapa- lemur in Eastern Betsileo, Madagascar. Primate Con- servation, 8: 40-43. Meier, B., R. Albignac, A. Peyrieras, Y Rumpler, and P. Wright. 1987. A new species of Hapalemur (Pri- mates) from south east Madagascar. Folia Primatolo- gica, 48: 211-215. Meyers, D. M., and I. Tattersall. 1988. Field re- porting of variation within lemur species. Primate Conservation, 9: 117-118. Milne Edwards, A., and A. Grandidier. 1875. His- toire physique, naturelle et politique de Madagascar. Vol. 1, text. Imprimerie National for A. Grandidier, Paris. MlTTERMEIER, R. A., W R. KONSTANT, M. E. NlCOLL, and O. Langrand. 1992. Lemurs of Madagascar: An Action Plan for their Conservation 1993-1999. IUCN Species Survival Commission, Gland, 58 pp. National Research Council. 1981. Techniques for the Study of Primate Population Ecology. National Acad- emy Press, Washington, D.C., 231 pp. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la Conservation et des Aires Protegees. World Wide Fund for Nature, Gland, xvii+ 374 pp. Oates, J. F 1994. Africa's Primates in 1992: Conser- vation issues and options. American Journal of Pri- matology, 34: 61—71. , G. H. Whitesides, A. G. Davies, P. G. Water- man, S. M. Green, G. L. Dasilva, and S. Mole. 1990. Determinants of variation in tropical forest pri- mate biomass: New evidence from West Africa. Ecol- ogy, 7: 328-343. O'Connor, S. M., M. Pidgeon, and Z. Randria. 1987. Un programme de conservation pour la Reserve d'Andohahela, pp. 31-36. In Mittermeier, R. A., L. H. Rakotovao, V Randrianasolo, E. J. Sterling, and D. Devitre, eds., Priorites en Matiere de Conservation des Especes a Madagascar. Occasional Papers of the IUCN Species Survival Commission, Number 2, Gland, Switzerland, 167 pp. Overdorff, D. J. 1993. Similarities, differences, and 304 FIELDIANA: ZOOLOGY seasonal patterns in the diets of Eulemur rubriventer and Eulemur fulvus rufus in the Ranomafana National Park, Madagascar. International Journal of Primatol- ogy, 14: 721-735. Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, and P. Griveaud. 1971. RCP 255. Etudes des 6co- systemes montagnards dans le region malache. I. Le massif de l'Andringitra. 1970-1971. G6omorphologie, climatologie et groupements v6g6taux. Bulletin de la Soci&e" d'Ecologie, II: 198-226. Perrier de la Bathie, H. 1927. Le Tsaratanana, l'An- karatra et l'Andringitra. M6moires Academie Malga- che, 3: 5-68. Petter, J. J., and A. Peyrieras. 1970. Observations eco-ethologique sur les 16muriens malgaches du genre Hapalemur. La Terre et la Vie, 24: 356-382. Pollock, J. I. 1986. A note on the ecology and behav- ior of Hapalemur griseus. Primate Conservation, 7: 97-100. Rumpler, Y. 1990. Systematiques des L6muriens, pp. 13-22. In Roeder, J. J., and J. R. Anderson, eds., Pri- mates, Recherches Actuels. Masson et Cie, Paris. , S. Warter, Ch. Ratomponirina, M. Hauwy, and B. Ishak. 1985. Cytogenetic study of complex hybrids in the genus Lemur (Primates, Prosimians). Folia Primatologica, 44: 108-116. Skorupa, J. P. 1986. Responses of rainforest primates to selective logging in Kibale Forest, Uganda: A sum- mary report, pp. 57-70. In Benirschke, K., ed., Pri- mates: The Road to Self-sustaining Populations. Springer- Verlag, New York, 1044 pp. . 1987. Do line-transect surveys systematically underestimate primate densities in logged forests? American Journal of Primatology, 13: 1-9. . 1988. The effects of selective timber harvest- ing on rain-forest primates in Kibale Forest, Uganda. Ph.D. Dissertation. University of California, Davis. Starmuhler, F. 1960. Beobachtungen am Mausmaki (Microcebus murinus). Natur u. Volk, 90: 194-204. Sterling, E. J. 1994. Aye-ayes: Specialists on structur- ally defended resources. Folia Primatologica. 62: 142— 154. Tattersall, I. 1982. The Primates of Madagascar. Co- lumbia University Press, New York, 382 pp. . 1986. Notes on the distribution and taxonomic status of some subspecies of Propithecus in Madagas- car. Folia Primatologica, 46: 51-63. Thomas, S. C. 1991. Population densities and patterns of habitat use among anthropoid primates of the Ituri Forest, Zaire. Biotropica, 23: 68-83. Vuillaume-Randriamanantena, M., L. R. Godrey, and M. Sutherland. 1985. Revision of Hapalemur (Prohapalemur) gallieni (Standing, 1905). Folia Pri- matologica, 45: 89-116. White, F 1989. Diet, ranging behavior and social or- ganization of the black and white ruffed lemur, Va- recia variegata variegata, in southeastern Madagas- car. American Journal of Physical Anthropology, 87: 232. Whitesides, G. H., J. F. Oates, S. M. Green, and R. P. Kluberdanz. 1988. Estimating primate densities from transects in a West African rain forest: A com- parison of techniques. Journal of Animal Ecology, 57: 345-367. Wright, P. C. 1986. Diet, ranging behavior and activity pattern of the gentle lemur (Hapalemur griseus) in Madagascar. American Journal of Physical Anthro- pology, 78: 327. , P. S. Daniels, D. M. Meyers, D. J. Overdorff, and J. Rabesoa. 1987. A census and study of Ha- palemur and Propithecus in southeastern Madagascar. Primate Conservation, 8: 84-88. STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 305 Gazetteer of Localities Mentioned in the Text Elevation (m) Latitude S Longitude E Locality O ' 0 ' Ambalamanenjana 22 07 47 02 Ambalamarina 22 08 46 55 Ambalamarovandana 1,530 22 08 46 57 Ambalavao 21 50 46 56 Ambarongy 720 22 13 47 01 Ambatomboay 650 22 15 47 01 Ambatovaky -16 42 -49 15 Ambavahala 1,500 22 04 46 54 Amboasary, Fianarantsoa 1,300 21 51 47 14 Ambohiby 18 48 46 08 Ambohimitombo 1,200-1,500 20 43 47 26 Ambohitantely, RS 1,448-1,662 18 09 47 16 Ambositra 20 31 47 15 Ampamakiesiny 24 31 46 50 Ampanasana River 1,530 22 08 46 55 Ampasipotsy ±2,550 22 12 46 57 Ampasy 22 18 47 01 Ampijoroa 16 15 46 48 Ampitambe ±900 -20 22 -47 46 Analafolaka 22 05 47 03 Analamazaotra, RS 930-1,040 18 56 48 25 Andalamby 23 38 45 49 Andasibe 915 18 56 48 25 Andohahela, RNI -24 40 -46 40 Andohahelo see Andohahela Andohariana 2,030 22 09 46 54 Andrahomana 50 25 50 46 40 Andranomalona River neai • Andasibe Andrianony 22 17 46 52 Andrivola 15 46 49 35 Angavokely 18 55 47 44 Anjavidilava 1,800-2,100 22 09 46 57 Ankafana see Ankafina Ankafina 1,600 21 12 47 12 Ankarafantsika 80-333 16 15 46 55 Ankaratra 2,642 19 25 47 12 Ankarimbelo 22 08 47 20 Ankeramadinika 1,400 20 43 47 26 Antananarivo 1,250-1,400 18 55 47 31 Antambohobe 22 18 46 49 Antanifotsy 1,470 22 07 46 54 Antongil, Baie d' 15 45 49 50 Antsifotra River 2,000 22 10 46 56 Antsiranana 0-40 12 16 49 18 Aody 2,036 22 03 46 53 Beampingaratra 24 30 46 50 Bemera 22 04 46 54 Camp 1 (RCP) 1,650 22 17 46 52 Camp 2 (RCP) 2,030 22 09 46 54 Camp 3 (RCP) 2,470 22 11 46 53 Camp 4 (RCP) 2,000 22 10 46 56 Camp 5 (RCP) 1,995 22 10 46 57 Camp 6 (RCP) 1,530 22 08 46 57 Camp 1 (1993) 720 22 13 47 01 306 FIELDIANA: ZOOLOGY Gazetteer. Continued. Elevation (m) Latitude S Longitude E Locality e ' 0 ' Camp 2 (1993) 810 22 14 47 00 Camp 3 (1993) 1,210 22 13 46 58 Camp 4 (1993) 1,625 22 12 46 58 Camp 5 (1993) 2,075 22 10 46 57 Col de Sakavalana north of Tolagnaro (Viette, 1991) Col de Vohitsatsiva 1,560 22 10 47 02 Col du Tandroka 1,200 22 17 46 54 Cuvette de Boby 2,470 22 11 46 53 Didy 1,000 18 07 48 32 Fanovana 600-800 18 55 48 34 Faraf angaria 0-40 22 49 47 50 Fenoarivo 17 22 49 25 Fenoevo -24 16 -47 29 Fort Carnot see Ikongo Fivahona see Fivanona Fivanona 22 06 46 57 Iatara River see Iantara River Iandroka River see Tandroka River Iantara River 22 13 47 01 Ihovika River 22 18 46 57 Ikongo 21 53 47 28 Imaitso 2,030 22 09 46 57 Itasy 19 04 46 17 Ivangomena 2,100-2,500 22 09 46 53 Ivohimanitra 700 20 42 47 35 Kianjavato 21 22 47 52 Kimora River 22 11 47 01 Korokoro River see Korokoto River Korokoto River 22 11 47 02 Lac Alaotra 17 30 48 30 Lalangina River 700 22 13 47 01 Manampatra River 22 51 47 50 Mananara 16 10 49 46 Mananara River 23 21 47 42 Mananara, RB 0-570 -16 29 -49 42 Manangotry 850 24 45 46 52 Mandraka 18 55 47 56 Mangoky River 21 41 45 10 Mangoro River 20 00 48 45 Manantantely 50-600 24 59 46 56 Manjakatompo 1,550-1,980 19 22 47 26 Manjarivolo 1,650 22 17 46 52 Manombo, RS 0-137 23 02 47 44 Manongarivo, RS 155-1,876 -13 59 -48 23 Mantady 18 51 48 27 Marojejy, RNI 75-2,133 -14 26 -49 15 Marositry 2,000 22 10 46 56 Marovitsika 21 57 47 33 Marovony 50-100 47 20 24 05 Masoala Peninsula 0-1,224 -50 09 -15 38 Matitanana Basin 22 26 47 55 Miarinarivo 22 05 47 04 Montagne d'Ambre, PN 850-1,475 -12 32 -49 10 Nahampoana 75-300 24 58 46 58 Namoly 1,600 22 08 46 54 Nandihizana 1,340 -20 50 -47 10 Nosy Mangabe 0-332 15 30 49 46 Pennet see Andasibe GAZETTEER 307 Gazetteer. Continued. Elevation Latitude S Longitude E Locality (m) o ' 0 ' Pic Boby 2,658 22 11 46 53 Pic Bory 2,630 22 12 46 55 Pic Ivangomena 2,556 22 09 46 53 Pic d'lvohibe, RS 775-2,060 -22 32 -46 59 Pic Soaindra 2,630 22 12 46 54 Pic Vohidray 1,920 22 07 46 56 Pic Vohipia 1,761 22 09 47 02 Ranomafana, PN 800-1,200 -21 16 -47 28 Riembavy River 22 09 46 54 Sahanivoraky River 22 13 47 00 Sahavatoy River 22 13 47 00 Sendrisoa 22 00 46 57 Ste Luce 0-20 24 45 47 11 Tandroka River 22 18 46 53 Toamasina 0-20 13 21 49 45 Tolagnaro 0-40 25 01 46 59 Tsaratanana, RNI 227-2,876 -48 51 -13 59 Vatoharanana 1,000-1,100 21 16 47 26 Varavarana 1,500-1,850 22 08 46 57 Vinanitelo 1,300 21 43 47 16 Vohitsaoka 22 01 46 44 Vohiparara 1,100-1,200 21 16 47 24 Volotsangana River 22 13 46 58 Vondrozo 500 22 49 47 20 Zomandao River 22 07 46 55 Zombitse 485-825 -44 40 -22 47 Note: For geographical localities such as rivers, large reserves, and mountain ranges we have given an intersection of coordinates that allows for easy location on maps. The information in the body of the Gazetteer is based partially on data from the United States Board on Geographic Names (1955), Paulian et al. (1971), and Viette (1991). Literature Cited United States Board on Geographic Names. 1955. Madagascar, Reunion and the Comoro Islands. Gaz- Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, etteer no. 2. U.S. Government Printing Office, Wash- and P. Griveaud. 1971. RCP 225. Etudes des eco- ington, D.C. systemes montagnards dans le region malgache. I. Le massif del'Andringitra. 1970-1971. Geomorphologie, Viette, P. 1991. Principals localites ou des Insectes climatologie et groupements vegetaux. Bulletin de la ont ete recueillis a Madagascar. Faune de Madagascar, Societe d'Ecologie, 11(2-3): 198-226. supplement 2. Private printing. 308 FIELDIANA: ZOOLOGY Index Abrus 62 precatorius 9, 62 Acalypha andringitrensis 61 Acanthaceae 56 Acariformes 137, 138, 139, 140, 141 Accipiter francesii 174, 181, 182 henstii 174, 182 madagascariensis 174, 180, 182 Acridotheres tristis 173, 188 Actitis hypoleucos 174 Adiantaceae 55 Adicella 116 Afrocricetodontinae 246, 248, 252 Afrocricetodon 248 Afrolistrophorus 140 Afronorus 121, 129 harrisoni 125 oliffi 125 matitensis 121-125, 129 Agapornis carta 174, 180, 183 Agauria 9, 17, 18 salicifolia 61 Aglyptodactylus madagascariensis 162, 164, 169 humblotii 69 minor 17, 69 A/fcizia 9, 62 gummifera 41, 46, 62 vintsioides 174, 177, 178 Alchemilla andringitrensis 69 madagascariensis 73, 74, 174, 183 Alistrophoroides 137, 138 Allophylus 71 Aloaceae 17, 56 A/oe 17 andringitrensis 8, 56 capitata 56 decorseii 56 haworthioides 56 macroclada 56 Amaurobidae 146 Amawrope/ra bergiana 11, 78 aff. oppositiformes 78 aff. strigosa 78 Amblyopone 100, 103, 104 Amblyoponini 100, 103 Amphiglossus 163, 167, 168, 169 a/iosy 161, 163, 164, 169 macrocercus 161, 163, 164, 169 melanopleura 163, 164, 165, 169 melanurus 163, 164, 169 punctatus 163, 164, 169 diffringens 147, 150 Amyrai humbertii 61 Anacardiaceae 9, 12, 38, 39, 56 Analgidae 140 Andropogon trichozygus 68 Angraecum 9, 1 1 , 67 oblongifolium 67 Annelida 146, 147, 149 Annonaceae 10, 33, 37, 38, 39, 56, 73, 87 A/ioc/i^/ms grandidieri 100, 103 Anodonthyla boulengeri 162, 164, 169 montana 162, 167, 168 Anoplura 139, 140 Anthericaceae 57 Anthocleista madagascariensis 10, 44, 64 Anthospermum ibityense 69 madagascariense 70 Anthribidae 146, 148 Anthrophyum boryanum 78, 82 A/tfjVfcsma petiolare 39, 41, 46, 61 Awr/gramma virchowii 78 Aphaenogaster 106 Aphistogoniulus 90 aff. corallipes 91 Aphloia 14 theiformis 9, 46, 62 Apiaceae 57 Apocynaceae 37, 43, 46, 57 Apodocephala pauciflora 57 Apodytes 43, 46, 63 Apollonias madagascariensis 8 Apu5 180 melba 174 barbatus 174 Aquifoliaceae 9, 33, 38, 38, 40, 57 Araceae 9, 57 Arachnida 146, 147 Araliaceae 9, 16, 33, 34, 37, 38, 39, 40, 50, 57 Aranea 146 Ardea purpurea 174, 180 Arecaceae 57 Aristea angustifolia 63 cladocarpa 63 humbertii 63 madagascariense 63 mabifolius 56, 73, 87, 274 Arthropodium caesioides 57 Arthropteris monocarpa 78 parallela 78 Arthrosphaerini 90 Arundinaria 13, 17, 240 Arundinella nepalensis 68 Arvicolinae 249 Asclepiadaceae 57 Asc/ep/a.s fruticosa 57 Ascodipteron 138 As/o madagascariensis 174, 180, 184 Aspleniaceae 55 Asplenium 9 aethiopicum 78 anisophyllum 11, 78, 82 auritum 78 bipartitum 11, 78 cancellatum 78 cuneatum 78 dregeanum 11, 78 erectum var. erectum 78 friesiorum 78 herpetopteris var. herpetopteris 78 herpetopteris var. massoulae 78 inaequilaterale 78 lividum 78 mannii 78, 82 marojejyense 55 /;/'(/;/.v 78 normale 78 obscurum 78 pellucidum 78 petiolutatum 78 poo/// 55, 78 preusii 55, 77, 78 protensum 78 rectum var. zeyheri 78 rutifolium 55, 78 sandersonii 55, 78 theciferum 78 Astacoides 155-156 betsileoensis 155 caldwelli 155 crosneri 155 granulimanus 155, 156 madagascariensis 155 perir/ 155 Astacopis 155 Asteraceae 14, 16, 17, 40, 44, 50, 52, 58, 73, 187 Asteroptf/a rhopaloides var. angustata 71 Ate/0/7HS INDEX 309 pittoides 174, 177, 181, 185 crossleyi 174, 177, 178, 185 Athripsodini 115, 117 Athyrium scandicinum 78 Atopomelidae 137, 138, 139 Atyidae 152-154 Avahi laniger 294, 295, 297 laniger laniger 299 Avenzoariidae 140, 141 Aviceda madagascariensis 174, 181 Baetidae 115 Bakerella 11, 186, 187, 188 clavata 185, 186, 187 clavata var. lenticellata 16, 64, 73 clavata var. peralata 9, 64, 74 tandrokensis 64, 74, 186, 187 Balsaminaceae 9, 17, 59 Belvisia spicata 78 Benthamia exilis exilis 67 exilis tenuissima 67 macra 67 monophylla 67 oreodorum 67 praecox 67 Betscheuma 90 andringitrae 90 orbatum 90 Bignoniaceae 12, 59 Blaberidae 146 Blattodea 146, 148 Blechnum 9 attenuatum 78 attenuatum var. attenuatum 78 humbertii 11, 78 ivohibense 78, 82 punctulatum 11, 78 simillimum aff. binerve 78 simillimum var. simillimum 78 simillimum var. xiphophyllum 78 fl/Zg/ua 9 fl/ort'a tanalorum 61 Blotiella madagascariensis 78 Boehmeria 72 floop/»'5 161, 162, 163, 165, 167, 168, 169 albilabris 162, 169 albipunctatus 162, 168, 169 ankaratra 165 boehmei 162, 169 goiKfon 162, 168, 169 laurenti 161, 165 /wtewj 162, 169 madagascariensis 162, 164, 169 majori 165 microtympanum 162, 168, 169 miniatus 162, 169 rappoides 162 reticulatus 162, 168, 169 untersteini 161, 165 viridis 162, 165 Brachiaria dimorpha 68 Brachylaena 14 ramiflora 57 Brachypteracias 111 leptosomus 174, 180, 184 squamiger 174, 180, 181, 184, 270 Brachytarsomys 233, 235, 243, 249 albicauda 87, 270, 275-276 Brachyuromys 233, 235, 243, 249, 262 betsileoensis 239, 258, 261, 262- 264, 270, 274, 283, 290 ramirohitra 85, 164, 222, 240, 258, 261, 262, 263, 264, 266, 269, 270, 274, 275, 277, 279, 283 Brassicaceae 59 Breonia 46, 70 Brooke sia 43, 166 nam* 161, 167, 169 nasus nasus 163, 166, 167 nasMS pauliani 167 superciliaris 163, 169 Bubulcus ibis 174, 180 Buddleja indica 64 Bulbophyllum 9, 11, 67 andringitranum 67 Burasaia madagascariensis 65 Burseraceae 32, 37, 38, 49, 58, 73, 83-88 flwteo 182 brachypterus 174, 182 Buthidae 146 fivtt/jen'a melleri 71 Cactaceae 9, 59 Calanthe repens var. pauliani 67 Calicalicus madagascariensis 175, 179 Calophyllum 9, 33, 37, 44, 59 drouhardi 8, 10, 32, 33, 50, 52, 59 paniculatum 8, 41, 46, 59 Ca/umma brevicornis brevicornis 163, 169 brevicornis hilleniusi 163, 168, 169 /a//ax 163, 169 gastrotaenia andringitraensis 163, 167 gastrotaenia gastrotaenia 163, 169 nasuta 163, 169 oshaughnessyi 163, 169 Campanulaceae 59 Camponotine 98, 102 Camponotus 94, 98, 102 hildebrandti 98 Ziova becfc/ var. a/r/or 94 Campylospermum obtusifolium 67 Canarium 10, 50, 272, 274, 302 boivini 84 madagascariense 32, 42, 46, 83- 88 madagascariense subsp. obtusi- folium 8, 49, 59, 73 Canellaceae 34, 40, 59, 73, 183 Canidae 289, 292 Canirallus kioloides 174, 181 Canw 291 /m/?mj 289, 292 Canthium 10, 37, 43, 44, 46, 70 andrigitrense 70 parvistipula 16, 70 Caprimulgus madagascariensis 174, 180, 184 Carabidae 146, 148, 185 andringitrensis 60 Caridina 152-154 /lova 152, 153 gracilirostris 154 isaloensis 152 isaloensis grandidieri 153, 154 longirostris 154 serratirostris 154 xiphias 152, 153 Casearia 39, 44, 62 elliptica 12, 62 nigrescens 62 nigrescens lucida 62 Cassinopsis 14, 63 tomentosa 10, 44 Cenrropws tow/ow 174, 180 Cerambycidae 146, 148 Cerapachyinae 98, 101, 102 Cerapachys 98, 102 Ceratopogonidae 143 perrieri 57 Certhiidae 188 Chamaeleonidae 163, 169 Chassalia 70, 75, 183, 185, 186 ambodirianensis 70 bojeri var. bojeri 70 Cheilanthes horizontalipinnata 55 madagascariensis 78 Cheirogaleus major 274, 294, 295, 296, 297, 298, 299, 303 Cheumatopsyche 116, 118 Chilopoda 146, 148, 149 Chimarra 116, 117, 118 anadabolava 117 a«A^j 117 atana 117 chatugana 117 elatrasoa 117 310 FIELDIANA: ZOOLOGY erici 117 michaeli 117 nadia 117 sahanivorakyae 117 vorombola 1 1 7 watayana 117 wigota 117 Chirodiscidae 138 Christella dentta 78 distorts 77, 78 Chrysophyllum 42, 46, 71 Cicadidae 146, 148 Ci/i/itf/nay/m* fragrans 34, 44, 59, 73, 183 Cisticola cherina 175, 180 C Union ui 17 Claoxylon 61 Cleistanthus 37 boivinianus 33, 39, 44, 52, 61 Clematis mount mini var. sulfurea 69 Clerodendrum 14, 44, 72 Clusiaceae 8, 9, 10, 13, 32, 33, 37, 38, 39, 40, 50, 59, 73, 183 Coleochloa setifolia 60 Coleoptera 146, 148 Coleotrype goudotii 60 Colubridae 163, 169 Co/mwi/wz //via 143 Commelinaceae 60 Compsoneuriella 121, 128 Co/i/ogra/n/rte madagascariensis 79 Convolvulaceae 60 Cwryza andringitrana 58 Co/wyc/ms albospecularis 140, 174, 177, 178 Coracma c/nerea 174, 177, 178, 179, 181 Coracopsis 177 nigra 73, 75, 174, 183 vasa 174, 179, 180, 183 Cordylidae 163 Corvus albus 175, 180 Costularia cf. purpurea 60 Coma 176 caerulea 174, 176, 179, 181 reynaudii 174, 176, 179, 180, 181 Crassula 60 Crassulaceae 17, 60 Craterostigma perrieri 71 Crematogaster 98, 102 /ofeora 98 scfeniti 98, 102 Crematogastrini 98, 102 Cricetinae 249 Cricetodontidae 246, 252 Crocidurinae 221 Crossleyia xanthophrys 175, 177, 178, 187 Crotalaria andringitrensis 62 Croton 10, 37, 39, 44, 62 myriaster 61 nitidulus 61 Crustacea 146, 148 Cryptocarya 8, 10, 32, 34, 36, 37, 39, 42, 44, 46, 50, 64, 73, 87, 183, 270, 274 crassifolia 14, 33, 34, 46, 64 madagascariensis 32, 42, 46, 64 oppositifolia 64 Cryptoprocta ferox 261, 263, 290, 292 Cryptoproctinae 290 Cryptosylvicola randrianasoloi 140, 175, 177, 178, 179, 180, 186 Ctenitis lanuginosa 79 magma 79 poolii 79 Ctenopteris devoluta 79 elastica 79 excaudata 11, 79 rigescens 79 villosissima 11, 79 Cuculus audeberti 174, 180, 183 rochii 174, 183, 184 Culicidae 143 Cunoniaceae 9, 12, 13, 32, 33, 37, 38, 39, 40, 60, 240 Curculionidae 185 Cyanolanius madagascarinus 175, 177, 178, 179, 180 Cyathea 9, 42, 43, 44, 46 approximata 79 bellisqamata 79 boivinii 79 borbonica var. laevigata 79 decrescens 79 dregei 55 me/teri 77, 79 similis 79 tsaratananensis 79 Cyatheaceae 9, 38, 39, 40, 50, 55 Cyclea 65 Cydnidae 146, 148 Cynanchum andringitrense 58 lineare 58 obovatum 58 papillatum 58 pycnoneuroides 58 Cynorkis 9, 68 alborubra 67 andringitrana 67 angustipetala var. amabilis 67 baronii 67 bathiei 67 *«?//<* 67 filiformis 67 gibbosa 67 g/gas 67 inversa 67 jumelleana 67 lilacina 68 ochroglossa 68 quinqueloba 68 saxicola 68 Cyperaceae 9, 12, 17, 60 calochrous 60 micrantherus 60 nemoralis 60 Cy/w/Mru.y parvus 174, 180 Cytoditidae 140, 141 Dacetononi 98, 102 Dalbergia 8, 9, 32, 39, 46, 47, 52, 62 baroni 9 Danais fragrans 14, 70 paludosa 70 Daubentonia madagascariensis 73, 74, 83, 85, 86, 87, 88, 294, 295, 299, 302 Decaryodendron 49 perrieri 9, 65 Deidamia bicolor 68 Demodex 139 Demodicidae 138, 139 Dendrocygna vidua to 173 Deparia boryana 79 parvisora 79 Dermaptera 146, 148 Dermestidae 146, 148 Desmodium repandum 62 Deuteromallotus 10 Dianella ensifolia 68 Dichaetanthera 65 matitanensis 65 Dichanthium andringitrensis 68 Dichapetalaceae 38, 40, 60 Dichapetalum 42, 46, 60 Dicranopteris linearis 11, 79 Dicrurus forficatus 175, 177, 178, 179 Didymocarpus madagascariensis 12, 63 vestitus 63 Digataria perrieri 68 Diospyros 9, 10, 44, 60, 61 Diplazium zakamenense 79 INDEX 311 Diplopoda 90-91, 146, 148, 185 Diplura 146, 148 Dipluridae 146 Diptera 109, 138, 139, 146, 149 Dipterocarpaceae 50 Disa andringitrana 68 Discothyrea 100, 103, 104 Dolichoderinae 98, 101, 102 Dolichos minutiflorus 62 Dolophilodes 116, 117, 118 andringitra 117 sahavatoyae 1 1 7 Dombeya 9, 32, 34, 43, 44, 46, 71 cf. amplifolia 44, 71 borraginopsis 71 ivohibeensis 71 ivohibeensis subsp. ivohibeensis 9 leucomacrantha var. crassibrac- tea 71 leucomacrantha var. leucoma- crantha 71 leiomacrantha leiomacrantha var. dissecta 71 leiomacrantha leiomacrantha var. angustata 71 leiomacrantha leiomacrantha var. leiomacrantha 71 muscosa 71 cf. spectabilis 46, 71 Doratoxylon 71 Dracaena xiphophylla 9, 43, 46, 60 Dracaenaceae 9, 38, 60 Dromaeocercus brunneus 173, 175, 187 seebohmi 175, 187 Dromicodryas bernieri 165 Dryopteris mangindranensis 79 remotipinnula 11, 79 subcrenulata 79 Drypetes 9, 10 madagascariensis 62 Dypsis 57 linearis 57 Ebenaceae 9, 10, 39, 60 Echinolaelaps echidninus 140 Echinops 192 Ecnomidae 116 Ectatommini 100, 103 Egretta a/fca 173 dimorpha 173 Elaeocarpaceae 10, 16, 32, 36, 37, 38, 39, 40, 50, 61, 73, 87 Elaeocarpus capuronii 61 hildebrandtii 16, 61 subserratus 16, 61 Elaphoglossum achroalepis 56, 79 acrostichoides 79 angulatum 79 aubertii 11, 79 coriaceum 79 decaryanum 79 deckenii var. deckenii 79 deckenii var. rufidulum 11, 79 forsythii-majoris 11, 79 humbertii 79 hybridum 79 leucolepis 79 ovalilimbatum 79 petiolatum var. salicifolium 11, 79 scolopendriforme 11, 79 sieberi 79 spathulatum 11, 79 subsessile 11, 79 Elateridae 146, 148 Elatostema 72 subfavosum 72 £//Mrws 73, 233, 235, 243, 249, 250, 264, 271, 274 mmor 85, 139, 240, 243, 261, 265, 266, 267-269, 270, 271, 272, 275, 277, 278, 279, 280, 283 mo/on 85, 139, 240, 254, 258, 261, 264-267, 269, 270, 271, 275, 277, 278, 279, 280, 283, myoxinus 258 penicillatus 265 tana/a 85, 139, 240, 261, 264, 266, 267, 269-270, 275, 277, 278, 279, 280 wW>/ 73, 85, 87, 139, 185, 261, 265, 266, 269, 270-271, 275, 277, 278, 279, 280 Embelia concinna 66 madagascariensis 66 Enicocephalidae 146, 148 Ephemerella 115 Ephemerellidae 115 Ephemeroidea 114 Ephemeroptera 109, 111, 113-115, 121-129 Ephippiandra 9, 42, 46, 49, 50, 66 microphylla 14, 66 Epidermoptidae 140 Ereynetidae 139, 140, 141 Ericaceae 9, 13, 16, 17, 40, 50, 61, 73, 186, 252 Eriocaulaceae 61 Eriocaulon fenestratum 61 Erythroxylaceae 10, 39, 40, 50, 61 Erythroxylum 61 capitatum 44, 61 pervillei 10 Eulemur fulvus 73, 74, 182, 301, 302 fulvus albifrons 87 fulvus albocollaris 294, 295, 297, 298, 301, 302 fulvus collaris 301, 302 fulvus fulvus 294, 295, 301 fulvus rufus 87, 294, 295, 297, 298, 301, 302 rubriventer 87, 294, 295, 296, 297, 298, 301, 302 Eulistrophoroides 1 39 Euphorbia emirnensis 17, 62 Euphorbiaceae 8, 9, 12, 17, 33, 37, 38, 39, 40, 44, 50, 61, 73, 185 Eupleres goudotii 290, 292 Euplerinae 290 Eurylaimidae 140, 186 Eurylophella 115 Eurystomus glaucurus 174, 184 Euthyplociidae 113-114 Eutriorchis astur 182 Evodia 33, 34, 44, 70 Exacum dipterum 62 emirnense 63 naviculare 63 Fabaceae 8, 9, 32, 37, 38, 50 Falco concolor 174 eleonorae 174 newtoni 174, 180 peregrinus 174, 182 zoniventris 173, 174, 180, 182 Faucheria 34, 44, 71 Faurea forficuliflora 69 Felidae 289, 292 Felis 291 silvestris 289, 292 Festuca perrieri 69 Ficus brachyclada 9, 66, 74 lutea 9, 66 politoria 12, 66 reflexa 66, 74, 183 rubra 66, 74 tiliifolia 9, 66 trichopoda 9, 66, 74 Flacourtiaceae 9, 12, 14, 39, 40 Formicidae 146, 185 Formicoxenni 99 Formincinae 98, 101, 102 Fossa fossana 290, 292 Foudia 111 madagascariensis 175, 177, 188 omissa 141, 143, 175, 177, 178, 179, 180, 188 Furcifer campani 163, 168, 169 lateralis 160, 161, 165 312 FIELDIANA: ZOOLOGY Gaertnera 10, 14, 32, 36, 37, 39, 42, 44, 46, 70 Galidia elegans 138, 290, 292 elegans elegans 290 Galidictis fasciata 290, 291, 292 fasciata striata 290 Galidiinae 290 Galium andringitrense 70 Garcinia 10, 44, 49, 52, 59 Gastropoda 146 Gekkonidae 163, 169 Gentianaceae 62 Geodipsas 163, 165, 169 heimi 163, 168, 169 infralineata 163, 169 Geogale 192 aurita 224 Geophilomorpha 146, 148 Geraniaceae 17, 63 Geranium andringitrense 17, 63 Gesneriaceae 10, 12, 63 Gladiolus andringitrae 63 bojeri 63 dalenii 64 Glamyromyrmex 98, 102, 104 Globotherium 90 Glycyphagidae 137, 138, 139, 140 Goerodes 116 Gramineae 17 Grammitidaceae 55 Grammitis barbatula 79 cryptophlebia 79 holophlebia 11, 79 Gravesia 9, 10, 65 calliantna 65, 74, 185 inappendiculata 65 cf. vestita 65 Grewia 42, 43, 46, 72 Gryllacrididae 146 Gryllidae 146 Gryllotalpidae 146 Gymnosperms 56 Gymnosphaera andohahelensis 79 boivinii 11, 80 coursii 11, 80 Gymnuromyinae 249, 271 Gymnuromys 233, 235, 243, 249, 271, 278 rofc?/?/ 73, 85, 87, 139, 240, 258, 261, 266, 268, 269, 270, 271- 272, 274, 275, 277, 279, 280 Habenaria decaryana 68 Haemaphysalis 138 Haemoproteus 142, 143 Halleria tetragona 16, 71 Hapalemur 299, 300, 301, 302 aureus 294, 295, 297, 298, 299, 301, 302 griseus griseus 294, 295, 297, 298, 299, 301, 302 simus 294, 295, 297, 298, 299, 301 Hartertula flavoviridis 175, 179 Hartlaubius auratus 175, 180 Helichrysum 17 abeitifolium 58 adhaerens 58 attenuatum 58 cf. baronii 58 barorum 58 bracteiferum 58 calocladum 58 cryptomerioides 17, 58 danguyanum 17, 58 hypnoides 58 minutiflorum 58 mirabile 58 retrorsum 58 retrorsum var. empetroides 17 stilpnocephalum 58 tomentosum 58 Hemicentetes 192 nigriceps 220, 223 semispinosus 220, 223 Hemiptera 146, 148 Heptageniidae 115, 121-129 Herpestidae 290, 292 Heterixalus betsileo 163, 165, 168, 169 madagascariensis 165 tricolor 165 Heterojapygidae 146, 148 Hipposideros commersoni 286, 287 commersoni commersoni 285 Histiopteris incisa 80 Hubertia andringitrensis 60 Huperzia megastachya 80 obtusifolia 80 pecten 80 pichiana 80 rubrica 80 sqarrosa 80 verticillata 80 Hyacinthaceae 63 Hydropsychidae 115 Hydroptila 116, 117 Hydroptilidae 116 Hydrothelphusa agilis agilis 156, 157 ag//w madagascariensis 1 56, 1 57 Hydrothelphusinae 156-157 Hydrotriche mayacoides 71 Hymenophyllaceae 12, 55 Hymenophyllum 55 capense 80 compactum 55 hirsutum 80 humbertii 80 peltatum 11, 80 perrierii 80 polyanthus 80 polyanthus var. fcu/in/i 80 poolii 80 sibthorpioidews 80 veronicoides 80 Hymenoptera 146, 148 Hyperoliidae 163, 169 Hypogeomys 233, 235, 243, 249 Hypolepis villoso-viscida 11, 80 Hypoponera 97, 100, 103-104 sakalava 100, 104 Hypositta corallirostris 141, 175, 177, 178, 179, 188 Hypsipetes madagascariensis 73, 74, 75, 174, 177, 178, 179, 186 Icacinaceae 10, 14, 33, 38, 39, 40, 44,63 /fex roifw 9, 50, 52, 57 Ilysanthes longipes 71 Impatiens 9, 17, 33, 59 Indigofera andringitrensis 62 Insectivora 137, 219 Iridaceae 63 Isopoda 146, 148 Isoptera 146 Ispidina madagascariensis 174 Ixodes 138 Ixodidae 137, 138, 139, 140, 141 Japygidae 146, 148 Kalanchoe 17, 60 bergeri 60 gracilipes 60 jongmansi 60 mangini 60 miniata 60 Kosteletzkya macrantha 65 malvocaerulea 65 ATyWm 98, 102, 104 Labiidae 146 Labramia 43, 46, 71 Laelapidae 137, 138, 139, 140 Lae lapis nut lull i 140 Lamiaceae 10, 17, 64, 73, 187 Landolphia myrtifolia 57 Lasiini 98, 102 INDEX 313 Lasiorrachis viguieri 69 Lauraceae 8, 10, 14, 32, 33, 34, 36, 37, 38, 39, 40, 42, 43, 44, 46, 49, 50, 52, 64, 73, 183 Lecomtella madagascariensis 69 Leioheterodon madagascariensis 165 Lemur catta 182, 294 Lentibulariaceae 9, 64 Lepidoptera 146, 148, 185 Lepidostomatidae 116 Lepilemur mustelinus 294, 295, 299 Leptoceridae 116 Leptocerus 1 1 6 Leptogale gracilis 209 Leptogenys 100, 104 Leptonema 116 Leptophlebiidae 109, 114-115 Leptopterus \11 chabert 175, 179 viridis 175, 179 Leptosomus discolor 174 Leptothorax 99, 104, 106 Leucocytozoon 142, 143 Limnanthemum aff. indicum 65 Limnogale 192 mergulus 220 Lindsaea flabellifolia 80 madagascariensis 80 odorata 80 Liophidium 163 rhodogaster 163, 169 Liopholidophis 163, 165, 167, 168 grandidieri 163 s/wm£#i 163, 165, 169 fAiWi 163, 169 Liparis6S, 185, 187 andringitrana 68 densa 68 henrici 68 latilabris 68 listeroides 68 Listophoridae 140 Listrophoroides 137, 138, 139 Lithobiomorpha 146, 148 agrestis 59 Loganiaceae 10, 12, 39, 64 Lomariopsidaceae 56 Lomariopsis pollicina 80 Lonchitis occidentalis 80 nana 175, 180, 188 Lophotibis cristata 174, 181 Loranthaceae 9, 11, 64, 73, 185, 186, 187, 188 Loxogramme lanceolata 56, 80 madagascariensis 14 Lycopodiaceae 56 Lycopodium carolinianum var. aj^we 80 cernuum 56, 77, 80 clavatum 56, 77, 80 zanclophyllum 80 Lygodactylus montanus 159, 167, 168 pictus 161, 165 Lygodium lanceolatum 80 Lythraceae 65 Mabuya 161 aureopunctata 161, 165 boettgeri 161, 165, 168 elegans 161, 165 gravenhorsti 163, 169 madagascariensis 161, 165 Macaranga 62 ankafinensis 62 cuspidata 12, 62 cf. decaryana 62 echinocarpa 14, 44, 52, 62 oblongifolia 62, 73, 185 Macphersonia 44, 71 Macronyssidae 138 Macrostemum 115, 116 Macrotarsomys 232, 233, 243-252 bastardi bastardi 232, 241 bastardi occidentalis 232, 241 mgto a/oa 174, 180, 184 t/apaca 8, 43, 46, 62, 73 feo/'«n 26 t/awoa 147-148 daMi 147 madagascariensis 147 £/rra 44, 46, 70 fitoravina 10, 70 Verbenaceae 14, 34, 40, 50, 72 Vernonia 16, 59 alleizettei 187 alleizettei var. rienanensis 59, 73 rubicundus 59 Vespertilionidae 138, 285, 287 Wo/a abyssinica 72 Violaceae 32, 34, 36, 37, 38, 39, 49, 72 Viscaceae 11, 72 Vwcum 11, 16, 44, 72 humbertii 72 Wtejc 34, 44, 72 V/ttaria humblotii 81 isoetifolia 81 318 FIELDIANA: ZOOLOGY Viverricula indica 291, 292 Viverridae 290, 292 Viverrinae 291 Weinmannia 13, 32, 34, 37, 39, 43, 44,46,47,50,51, 52,60 eriocarpa 60 matnmea 9, 12, 60 Xenopirostris polleni 175, 180, 188 Xerophyta dasylirioides 9, 17, 72 dasylirioides var. andringitrensis 72 Xiphopteris hildebrandtii 55, 77, 81 oosora var. microptera 81 serrulata 81 Xylopia 10, 47, 52, 57 Xyridaceae 72 Xyris madagascariensis 72 Zonosaurus 161, 163, 167 laticaudatus 161, 165 madagascariensis 160 o/7ia/M.s 160, 161, 165 Zoonavena grandidieri 174 Zoosphaeriini 90 Zosteropidae 141 Zosterops maderaspatana 73, 74, 75, 141, 175, 177, 178, 179, 180, 184, 187 NDEX 319