UNIVERSITY OF ILLINOIS LIBRARY AT URBAiMA-CHAMPAIGN BIOLOGY MP g **7n DDDD i i 3 o STANDARD CUSTOM B THESIS DUSTIE n DDDD TABLE CONT. INDEX FRONT COVER BACK COVER ADS 1 m ■D m 3J o o o > -< 33 G *- r« G n »b h- 3 u* o tr h- 3 %i n -n 33 m o cr m z U 3EJ 1 h - • ^ U H Q W ct ri a •« a o h- n a <» m iq h □ 1 MOUNT COVER D POCKET / CLOTH D POCKET / PAPER D n 33 m o m D 00 o o D PERM. CHANGE D NEW TITLE h o a h H- <* 3 << H- 0 n u h K « 0 -x, 3 G 11 CO rf. tn ex, IV3 0 CD CO l .i. 01 • — io w o O u> CD CD CD I I to U.i io cn -rr 8 t- 8 uo |i H ru 0 It til tr H D D co co 5 3 D D L: T7 O H 590.5 FI N.S. (\6.94 BIX PL , T Zoology V SERIES, NO. 94 A Floral and Faunal Inventory of the Reserve Naturelle Integrate d'Andohahela, Madagascar: With Reference to Elevational Variation Steven M. Goodman, Editor June 30, 1999 Publication 1503 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY i VBMAMttK ,.-:':l ■/-,■' , Information for Contributors to Fieldiana i >ns should be submitl I copy pit; nanuscripts will be c< iitor. na, Field M Text: M; must be typewritten double-spaced on standard-weight. I -inch pap. mar. Lf typed on an IBM-compatible computer using MS-EX or 5.0, MultiMate, Displaywrite 2, 3 & 4, Wang PC. Sam ■ WordStar programs or ASCII). r 100 manuscript pages, authors are requested to submit a "Table of ( [Nostra "List of Tables" immediately following title page. In most cases, the text should be preceded ■nd should conclude with "Acknowledgments" (if any) and "Literature Cit in the meti (periods are not used after abbreviated measuremen at and style of headings should follow that of recent issues of Fieldiana. iled style information, see The Chicago Manual of Style (13th ed.), published by The Un of Chicago Press, and also recent issues of Fieldiana. References: In •Literature Cited." book and journal titles should be given in full. Where abbre desiral in citation of synonymies), auth should follow Botanico-Periodicum-Huntianum and TL-2 Taxonomu leu & R. S. Cowan (1976 ct seq.) (botanical papers) or Serial Source 13) published by the BioScienees Information Se imes of botanical authors should follow the "Draft Index of Author Abbreviations, Royal Botam Kew." 1984 ed TL-2. References should be typed in the follov B. 1978. Flora of Barro Colorado Island. Stanl ^ pp. j., J. R. j d T. D. Pennington. 1963. A comparison of montane and lowland rain i lador. I. The forest structure, physiognomy, and fioristics. Journal of Ecology, 51: 567-601. I M. 1979. Ya;: the Siona: Cultural patterns in visions, pp. 63-80. In Browman. D. L. and R. A. Schwarz, eds.. Spirits, Shamans, and Stars. Mouton Publishers, The Hague, Netherlands. MURRA, J. 1946. The historic tribe dor. pp. 785-821. In Steward. J. H., ed.. Handbook of South American Indians. Vol. ndean Civilizations. Bulletin 143, Bureau of American Ethn< Smithsonian Institution, Washington, D.C. R. G. 198; id fern allies of Guatemala. Part II. Potypodiaceae. Fieldiana: Botan; 522. Illustrations: Illustrations are referred to a in the text (not as "plates" ). Figures must be ac ndication of scale, normally a reference bar. Statements in figure captions alone, such as re not: •iable. Captions should be typed double-spaced and consecutively. See recent issues of Fieldiana for del; ations should be marked on the ith author's name, figure number(s), and "top." Id, whene :able, be 8Vi by 1 1 inches (22 X 28 cm) and may not exceed 1 \xh ,e mounted on boards in the arrangement to be obtained in the uginal set ion to the printer as follows: Pen and ink drawings iginals, but within th< n; and! riginal illustrations will be returned t a publication Ul Med. Auth( !Per or color Iei e Prior Page Proofs: ding author will i be made Only on the single set of ; This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Errata — Fieldiana: Zoology, new series, no. 94 A series of errors were introduced into this publication associated with corrections made on the page proofs that were not properly taken over at various production stages. p. viii, line 4 — read "Reserves" as "Reserve" p. 47, column 2, line 19 — read "Bio- logic" as "Biologie" p. 60, under relative density, plot 4 — read " Polysacias" as "Polyscias" p. 67, under Florsitic Parameters, line 11 — read "aprevali" as "aprevalii" p. 73, line 13 — the value of "1.40" should be under column for plot 3, 1150 m p. 73, line 14 — read "13.36" as "13.65" p. 74, line 19 — read "Polyathia" as "Polyalthia" p. 74, lines 30 — 39 — read "Polycias" as "Polyscias" p. 74, line 49 — read "Rohdocolea" as "Rhodocolea" p. 75, line 72 — the value "15.54" should be in bold print p. 76, lines 120 — 121 — read "Albiz- zia" as "Albizia" p. 76, line 163 — read " trichophebia" as "trichophlebia" p. 78, lines 241— 242— read "Sizy- gium" as "Syzygium" p. 78, line 293 — read "Schysmatocla- da" as " ' Schismatoclada" p. 79, line 309— the value "17.85" should be in bold print p. 84, lines 2 — 3 of table — read "Chry- sophyllum boivinianum" as "Chry- sophyllwn boivinianum" p. 84, line 20 — read "Albizzia" as "Albizia" p. 84, line 34 — read "Caesaria" as " 'Case aria" p. 84, line 37 — read "Sizygium" as "Syzygium" p. 85, line 1 1 form bottom — read "Strebulus" as "Streblus" p. 86, line 6 of table — read "trichophe- bia" as "trichophlebia" p. 87, line 4 of table — read "Zanthox- yllum" as "Zanthoxylum" p. 87, line 6 of table — read "Caesaria" as " Casearia" p. 87, line 23 from bottom of table — read "Albizzia" as "Albizia" p. 87, line 21 from bottom of table — read "pervillena" as "pervilleana" p. 87, line 5 from bottom of table — read "Araliceae" as "Araliaceae" p. 89, line 23 of table — read "Polyal- this" as " Polyalthia" p. 91, line 11 of table — read "Drypoe- tes" as "Drypetes" p. 91, line 6 from bottom of table — read " Stronglyodon" as "Strongy- lodon" p. 92, line 8 of table — read "Schys- matoclada" as " Schismatoclada" p. 93, line 8 of table — read "Onocos- temurrC as " Oncostemum,'> p. 93, line 12 of table — read "Sizy- gium" as "Syzygium" p. 93, line 6 from bottom of table — read " Brachilaena" as "Brachylae- na" p. 95, line 3 of table — read "humberti" as "humbertW p. 95, line 12 of table — read "onnco- clada" as " oncoclada" p. 95, line 21 of table — read "Commi- phorra" as "Commiphora" p. 96, line 19 of table — read "micro- cepahala" as "microcephala" p. 96, line 28 of table — read "hilde- brandti" as "hildebrandtii" p. 96, line 10 from bottom of table — read "Cyanchum" as " Cynanchum" p. 96, line 2 from bottom of table — read "Rubiacae" as "Rubiaceae" p. 98, column 2, under Sample Sites, line 10 — read "Potamya" as "Potamyia" p. 210, column 1, line 9 — read "RS d'Anjanaharibe-Sud, or" as "RS d'Anjanaharibe-Sud," p. 246, caption to Table 14 — 11 — read "for five reserves" as "for four re- serves" p. 248, Goodman and Rakotoarisoa ref- erence— read "2 — 4" as "3 — 4" Information for Contributors to Fieldiana Text: or Wor Muse All n Foi Refere desirable (e Data v the "Dr; Ret Illustratio A Floral and Faunal Inventory of the Reserve Naturelle Integrate d'Andohahela, Madagascar ULT-!j The Schlinger m^T> Foundation WWF FIELDIANA Zoology NEW SERIES, NO. 94 A Floral and Faunal Inventory of the Reserve Naturelle Integrate d'Andohahela, Madagascar: With Reference to Elevational Variation Steven M. Goodman, Editor Center for Environmental and Evolutionary Biology Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 U.S.A. World Wide Fund for Nature Aires Protegees B. P. 738 Antananarivo (101) Madagascar Accepted June 30, 1998 Published June 30, 1999 Publication 1503 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 1999 Field Museum of Natural History ISSN 0015-0754 PRINTED IN THE UNITED STATES OF AMERICA c Preface vii 1. Description of the Reserve Naturelle Integrate d'Andohahela, Madagascar, and the 1995 Bi- ological Inventory of the Reserve 1 Steven M. Goodman Appendix 1-1. Participants in the Project (Field and Laboratory) 8 2. An Overview of the Botanical Communities of the Reserve Naturelle Integrate d'Andohahela, Madagascar 11 Nick A. Helme and Pierre Jules Rakotomalaza 3. Pteridophytes of the Eastern Slope of the Reserve Naturelle Integrate d'Andohahela, Mada- gascar: Distribution and Floristic Analysis 25 France Rakotondrainibe 4. Structure and Floristic Composition of the Vegetation in the Reserve Naturelle Integrate d'Andohahela, Madagascar 51 Pierre Jules Rakotomalaza and Nathalie Messmer 5. A Regional Analysis of Species Associations and Distributions of Two Caddisfly Families (Trichoptera: Hydropsychidae and Philopotamidae) in Southeastern Madagascar 97 Francois-Marie Gibon and Patricia Zoe Andriambelo 6. Proboscidoplocia (Ephemeroptera, Polymitarcyidae) from the Reserve Naturelle Integrate d'Andohahela and Surrounding Areas, With a Description of a New Species Ill Jean-Marc Elouard, Michel Sartori, Jean-Luc Gattolliat, and Ranalison Oliarinony 7. Three New Species of Baetidae (Ephemeroptera) from the Reserve Naturelle Integrate d'Andohahela, Madagascar 115 Jean-Luc Gattolliat, Michel Sartori, and Jean-Marc Elouard 8. Aquatic Biodiversity of Madagascar: Simulium (Diptera: Simuliidae) from the Reserve Natu- relle Integrate d'Andohahela and Surrounding Areas 125 Theogene Pilaka and Jean-Marc Elouard 9. Ant Diversity Patterns Along an Elevational Gradient in the Reserve Naturelle Integrate d' Andohahela, Madagascar 1 29 Brian L. Fisher 10. Taxonomic and Ecological Observations on the Scorpions Collected in the Reserve Naturelle Integrate d'Andohahela, Madagascar 149 Wilson R. Lourenco and Steven M. Goodman 11. Amphibians and Reptiles of the Reserve Naturelle Integrate d'Andohahela, Madagascar 155 Ronald A. Nussbaum, Christopher J. Raxworthy, Achille P. Raselimanana, and Jean-Baptiste Ramanamanjato 12. Bird Community Variation with Elevation and Habitat in Parcels 1 and 2 of the Reserve Naturelle Integrate d'Andohahela, Madagascar 1 75 A. F. A. Hawkins and Steven M. Goodman 13. Lipotyphla (Tenrecidae and Soricidae) of the Reserve Naturelle Integrate d'Andohahela, Madagascar 1 87 Steven M. Goodman, Paulina D. Jenkins, and Mark Pidgeon 14. Rodents of the Reserve Naturelle Integrate d'Andohahela, Madagascar 217 Steven M. Goodman, Michael D. Carleton, and Mark Pidgeon 15. Notes on the Bats of the Reserve Naturelle Integrate d'Andohahela and Surrounding Areas of Southeastern Madagascar 25 1 Steven M. Goodman 16. Carnivora of the Reserve Naturelle Integrate d'Andohahela, Madagascar 259 Steven M. Goodman and Mark Pidgeon 17. Lemurs of the Reserve Naturelle Integrate d'Andohahela, Madagascar 269 Anna T. C. Feistner and Jutta Schmid Gazetteer of Localities Mentioned in the Text 285 Index to Scientific Names 289 Preface This volume contains the results of a faunal and floral inventory conducted in extreme southeast- ern Madagascar in the Reserve Naturelle Integrate (RNI) d'Andohahela between 3 October and 15 December 1995. This project was the third of five intensive multidisciplinary surveys, completed to date, conducted along elevational transects of mountainous regions on Madagascar. These sur- veys were coordinated by the World Wide Fund for Nature (WWF), Madagascar, and this volume is the third published in this series in Fieldiana: Zoology. The other two include the inventory of RNI d'Andringitra conducted in 1993 {Fieldiana: Zoology, n.s. 85) and that of the Reserve Speciale (RS) d'Anjanaharibe-Sud in 1994 {Fieldiana: Zo- ology, n.s. 90). The results of subsequent altitu- dinal surveys are currently in preparation and in- clude the RS de Marojejy in 1996 and the RS dTvohibe in 1997. We are indebted to WWF staff in both Anta- nanarivo and Tolagnaro for their help in organiz- ing this mission, in particular Lalaniaina Andria- manarivo, Mark Fenn, Roland Laha, Olivier Lan- grand, Sheila O'Connor, Mamy Ravokatra, Tiana Razafimahatratra, and Malalarisoa Razafimpahan- ana. The assistance and cooperation of inhabitants of the villages of Enosiary and Eminiminy were extremely important for the success of the mis- sion. It is with great pleasure that we acknowledge the work of M. Ledada Rachel Razafindravao, the chef de cuisine, who kept us well-fed under rather difficult conditions. Jean-Aime Rakotoarisoa kind- ly examined and dated pottery remains found within the reserve. We are also grateful to the Di- rection des Eaux et Forets and the Association Nationale pour la Gestion des Aires Protegees, es- pecially Celestine Ravaoarinoromanga and Mia- dona Harisoa Faramalala, for permits to work in the reserve. This volume has benefited greatly from the crit- ical commentaries offered by numerous reviewers who evaluated the chapters presented herein. Those reviewers who did not waive anonymity are mentioned in the acknowledgments of each chapter. John Weinstein, Field Museum of Natural History, helped greatly in preparing the photo- graphs. We continue to be indebted to William Burger, Scientific Editor of Fieldiana, and Mar- jorie Pannell, Managing Editor of the Field Mu- seum Press, for accepting the arduous task of pro- ducing these volumes. Their organizational and editorial skills have graced these monographs. The 1995 biological inventory of the RNI d'Andohahela was made possible by a grant from USAID to WWF as part of an integrated conser- vation and development project aiming to protect the Andohahela region. Publication of this volume was significantly aided by generous grants from USAID and The Schlinger Foundation. Steven M. Goodman April 1998 Antananarivo Note added in proofs: Over the past few months the status of several Malagasy reserves has changed, based on national de- crees. We were unable to modify the text of this volume to reflect these changes. The Reserves Naturelle Integrale d'Andohahela, de Marojejy, and d'Andringitra are now the Pares National d'Andohahela, de Marojejy, and d'Andringitra, respectively. — S.M.G., 21 December 1998. Vlll Chapter 1 Description of the Reserve Naturelle Integrate d'Andohahela, Madagascar, and the 1995 Biological Inventory of the Reserve Steven M. Goodman1 Madagascar is well known for a remarkable va- riety of habitats, from lush rain forests to subdes- ert spiny bush forest. The Reserve Naturelle In- tegrate (RNI) d'Andohahela is located in extreme southeastern Madagascar, in a zone of dramatic ecotones between the humid eastern and the dry southern portions of the island. Much of this hab- itat diversity is represented in the RNI d'Andohahela, largely as a result of the effect of the Anosyenne Mountains, which act as a rain barrier (Battistini, 1964; Goodman et al., 1997). The abrupt shift in habitats is due to this north- south-aligned mountain chain. The eastern slopes of these ancient mountains are a relatively short distance from the Indian Ocean coast, with their summital zone and higher ridges blocking the movement of rain-bearing clouds coming in from the east (Paulian et al., 1973). The biological communities occurring on either side of this di- vide have little in common with one another, and these differences are discussed in this volume. The reserve, comprising 76,020 ha, is divided into three noncontiguous parcels (Nicoll & Lan- grand, 1989; Fig. 1-1): parcel 1 (humid forests on the eastern flank of the Anosyenne Mountains), between 46°37'-46°52'E and 24°30'-24°52'S— 63,100 ha; parcel 2 (dry spiny forest and some degraded gallery forest along river margins to the west of the Anosyenne Mountains), between 46°33'-46°38'E and 24°48'-24°58'S— 12,420 ha; and parcel 3 (transitional humid and dry forest located just to the west of the Anosyenne Moun- tains), between 46°37'-46°39'E and 24°59'- 25°02'S— 500 ha. The initial 30,000 ha of the re- 'Field Museum of Natural History. Roosevelt Road at Lake Shore Drive. Chicago. IL 60605-2496, U.S.A. serve was established in 1939 after the botanical explorations of H. Humbert (1935, 1941). In 1966 an additional portion of the regional forest was placed within this protected area (Nicoll & Lan- grand, 1989). The abrupt ecotone between wet and dry across this zone gives rise to an extraordinary level of species turnover in the biotic communities over the distance of a few kilometers. To emphasize this dramatic effect, I excerpt entries made in my field notebook during a December 1992 hike with Mark Pidgeon and Sheila O'Connor in parcel 1 of the reserve between Eminiminy and Mahama- vo along the Isedro Trail: In the early afternoon after climbing over the Col d'Ambatomaniha along the Isedro Trail we came to the final ridge before the descent down the western slopes of the Anosyenne Mountains . . . The ridge, the western limit of parcel 1, forms the ecotone between humid and dry forest of the reserve. Here we were surrounded by large trees and terrestrial leeches [characteristic of humid forests below 1200 m]. Just below we could see to the west a dramatic shift in vegetation. Large baobabs were within a kilometer or so of us and with binoculars the char- acteristic cactus-like Didiereaceae of the spiny bush forest could be seen in the not so far distance. The calls of humid forest birds were all around us, while those characteristic of spiny bush could be heard just below. The shift is so abrupt that it seemed possible to stand in the humid forest and throw a rock into the spiny bush. Another interesting aspect of the region is that although all of the RNI d'Andohahela falls south of the Tropic of Capricorn, parcel 1 of the reserve retains characteristic aspects of a tropical humid or rain forest flora in both structure and species composition. This parcel is one of the southern- most "tropical" forests in the Old World. FIELDIANA: ZOOLOGY, N.S., NO. 94, JUNE 30, 1999, PP. 1-297 Isaka-tvondro Nahampoana Mandena Berarrty Reserve ,tf Parcel 3 ./ Ranoplso I TOLAGNARO \<* & & QO® ^ Legend A Summit ■ Town, Village /\/ River Boundary of the Reserve A Study Site N 10 Kilometers Source GIS WWF Antananarivo Printed by GIS WWF Antananarivo 04/98 WWF Fig. 1-1. Map of extreme southeastern Madagascar showing the position of the three parcels of the RNI d' Andohahela and the positions of our transect sites during the 1995 inventory of the reserve. The map was designed by the GIS Unit, WWF, Madagascar. A group of field scientists from seven different countries, representing numerous fields of interest in the fields of botany and zoology, studied the biota of the reserve between 19 October and 15 December 1995. Five camps were placed in parcel 1 at different elevational zones (440, 810, 1200, 1500, and 1875 m) along the humid eastern flank and summital zone of the Anosyenne Mountains. One site in the gallery and spiny forest of parcel 2 (at 120 m) was also studied. Each site was oc- cupied for a minimum of 8 days. The results pre- sented in this volume are largely confined to par- cels 1 and 2. All of the groups studied and col- lected during this mission are reported on herein; FIELDIANA: ZOOLOGY the single exception is the report for the terrestrial snails. Many of the mission members and virtu- ally all of the field methods were identical to those on similar transects conducted in the RNI d'Andringitra (Goodman, 1996), the Reserve Spe- ciale (RS) d'Anjanaharibe-Sud (Goodman, 1998), the RNI de Marojejy in 1996 (Goodman, in prep.), and the RS d'lvohibe and the corridor link- ing that reserve to the RNI d'Andringitra in 1997 (Goodman, in prep.). We are currently organizing a mission to the RS de Manongarivo for the up- coming 1998 field season. In this chapter details are presented on the 1995 expedition to the RNI d'Andohahela, along with some data on the climate, geology, and scientific exploration of the area. For more information on the RNI d'Andohahela readers are referred to O'Connor et al. (1985), Nicoll and Langrand (1989), and Goodman et al. (1997). PBZT PN RB RCP RNI RS UMMZ USNM WWF Pare Botanique et Zoologique de Tsimbazaza, Antananarivo Pare National Reserve de Biosphere La Recherche Cooperative sur Pro- gramme No. 225, under the Centre National de la Recherche Scientifique Reserve Naturelle Integrale Reserve Speciale University of Michigan Museum of Zoology, Ann Arbor National Museum of Natural History, Washington, D.C. (formerly United States National Museum) World Wide Fund for Nature Transect Sites Abbreviations Used in the Text ANGAP Association Nationale pour la Gestion des Aires Protegees, Antananarivo BM(NH) The Natural History Museum, Lon- don (formerly British Museum [Nat- ural History]) CNRE Centre National de Recherche sur l'Environnement, Antananarivo CNRS Centre National de la Recherche Scientifique, Antananarivo dbh diameter at breast height DEF Direction des Eaux et Forets FAC Fonds d'Aide et de Cooperation FMNH Field Museum of Natural History, Chicago FTM Foiben-Taosarintanin'i Madagasikara, Antananarivo (Institut National de Geodesie et Cartographie) LRSAE Laboratoire de Recherche sur les Sys- temes Aquatiques et leur Environne- ment, Antananarivo MBG Missouri Botanical Garden, St. Louis MNHN Museum National d'Histoire Natu- relle, Paris MRAD Ministere de la Recherche Appliquee au Developpement, Antananarivo ORSTOM Institut Francais de Recherche pour le Developpement en Cooperation (for- merly Office de la Recherche Scien- tifique et Technique Outre-Mer) During our 1995 inventory of the reserve, co- ordinates for each transect site were determined with the use of a geographical positioning system, and the names for various localities were taken from maps (Institut Geographique National, 1961a,b; FTM, 1979, 1990) and discussions with local people. A gazetteer of the localities men- tioned in the text is presented at the end of this volume (page 285). The first four transect zones in parcel 1 of the reserve ranged generally ±75 m in elevation relative to and centered around our first four camps. Due to problems associated with access to water and the lack of a relatively flat zone in which to install a large number of people, our fifth camp was not the center of the transect but established north of Pic Trafonaomby, at about 1700 m. The fifth and final transect site in parcel 1 was at 1875 m ±75 m. In parcel 2 there was little topographic variation. Elevations were determined with the use of al- timeters. The positions of each camp during the survey are indicated in Figure 1-1. Parcel 1 440 m (camp 1) 19-28 October 1 995— Madagas- car: Province de Toliara, Reserve Naturelle Integrale d'Andohahela, 8 km NW of Emi- niminy, 46°45.9'E, 24°37.6'S 810 m (camp 2) 28 October-7 November 1995— Madagascar: Province de Toliara, Reserve Naturelle Integrale d'Andohahela, 12.5 km NW of Eminiminy, 46°44.3'E, 24°35.6'S GOODMAN: DESCRIPTION OF THE RNI D'ANDOHAHELA 1200 m (camp 3) 7-17 November 1995— Mada- gascar: Province de Toliara, Reserve Natu- relle Integrate d'Andohahela, 13.5 km NW of Eminiminy, 46°44.1'E, 24°35.0'S. 1500 m (camp 4) 17-27 November 1995— Mad- agascar: Province de Toliara, Reserve Natu- relle Integrale d'Andohahela, 15.0 km NW of Eminiminy, 46°43.9'E, 24°34.2'S. 1875 m (camp 5) 27 November-5 December 1995 — Madagascar: Province de Toliara, Re- serve Naturelle Integrale d'Andohahela, 20.0 km SE of Andranondambo, 46°43.3'E, 24°33.7'S. Parcel 2 120 m (camp 6) 7-15 December 1995 — Mada- gascar: Province de Toliara, Reserve Natu- relle Integrale d'Andohahela, 7.5 km ENE of Hazofotsy, 46°36.6'E, 24°49.0'S. Itinerary of the 1995 Expedition Before our inventory of parcel 1 of the RNI d'Andohahela, two trails existed that traversed the northern portion of this forest. The Enakara-An- tseva Trail formed, in part, the northern boundary of the parcel and cut across the Marosohy Forest; the Eminiminy-Mahamavo Trail (also known as the Isedro Trail) passed via the Col d'Ambatomaniha. Neither of these paths ap- proached the summital zone of the Anosyenne Mountains, however, and it was decided that a new trail needed to be created that linked the low- est-lying areas of the parcel with the summit of Pic Trafonaomby (1956 m). Over the course of numerous trips to the region between May and September 1995, P. J. Rakoto- malaza and N. Messmer conducted extensive re- connaissance trips in collaboration with local vil- lagers, and they were able to establish a trail sys- tem that ran from the west of Eminiminy at about 300 m, entered the forest slightly below 400 m, and continued in a northwesterly direction up to about 1500 m. Four camps were set up along this trail at 440, 810, 1200, and 1500 m. The means to reach the summital zone was established from the 1500 m camp. Our fifth transect was at 1875 m, just below Pic Trafonaomby. The fifth camp was northwest of the peak, in a relatively flat area with running water, and about 2-2.5 km from the fifth transect zone. During the field mission (19 October- 15 De- cember 1995), members of the scientific group in- cluded F. Andriatsiferana (botany), J.-M. Elouard (aquatic insects at the 440 and 810 m sites only), A. Feistner (lemurs), S. M. Goodman (small mammals, birds, and bats), A. F A. Hawkins (birds), N. Helme (botany), R. Laha (botany and logistics), N. Messmer (botany), J. M. Rakotoar- ison (terrestrial snails in parcel 1 only), P. J. Rako- tomalaza (botany), F. Rakotondrainibe (ferns), J.- B. Ramanamanjato (reptiles and amphibians), B. Randriamampionona (ferns), A. Raselimanana (reptiles and amphibians), M. Ravokatra (birds), M. Pidgeon (small mammals and carnivores), and J. Schmid (lemurs). Including scientific staff, as- sistants, and visitors, the number of people inhab- iting various camps reached a maximum of 24. Numerous other scientists participated in the de- termination of specimens and analysis of data ob- tained during the 1995 field survey. In addition, based on an earlier field trip to another area of parcel 1, a chapter by B. L. Fisher on the ants of RNI d'Andohahela is included in this volume. The names and addresses of all field and labora- tory researchers involved in this project are given at the end of this chapter (Appendix 1-1). Logistics and Trail Systems Parcel 1 of the RNI d'Andohahela is not ac- cessible by motor vehicle. An old road that ter- minated at Eminiminy has not been passable for several decades. Access to the eastern edge of the parcel was by a foot trail that began just slightly north of Isaka-Ivondro on the road linking the To- lagnaro area to Manantenina via Ranomafana du Sud. This trail, known as the Tananana Trail, starts at 85 m, climbs up to the Col de Tananana (approximately 750 m), then drops into the Man- ampanihy River Valley and to the villages of Em- iniminy and Enosiary at about 300 m. These two villages were used as bases for the transfer of food reserves, supplies, and specimens during our mission, and the majority of porters employed during the displacements between camps (up to 40 individuals) were from these two villages. Food and research supplies were divided by camp (transect zone), packed in rice sacks, and stored in the WWF office in Tolagnaro. Two days before each scheduled camp shift, a WWF em- FIELDIANA: ZOOLOGY ployee in Tolagnaro purchased fresh food in the local market and transported these goods and the stored supplies for the next camp by car to Isaka- Ivondro. From there porters were engaged to car- ry the material to Enosiary (via the Tananana Trail), a 5 -hour trip. The following morning por- ters from Enosiary and Eminiminy carried the supplies to the site we were preparing to leave. There the baggage was rearranged, and the re- search group with porters climbed to the next site. After the second camp the distances were too long for the Enosiary/Eminiminy porters to make the round-trip over the course of 1 day, and makeshift camps had to be installed along the trail with cooking pots and food to accommodate the por- ters for the night. Between Eminiminy and our last camp near Pic Trafonaomby, the round-trip for the porters took 3 or 4 days. As mentioned earlier, in order for the research group to have access to the summital zone of par- cel 1 of the RNI d'Andohahela it was necessary to open and establish a new trail system. A pre- existing trail that entered a short distance into the forest was used to access the reserve to the east of Eminiminy. The site of our first camp at 440 m was along the Andranohela River in an area of forest that showed ancient signs of human distur- bance, including old tombs and certain aspects of the flora that indicated old secondary growth. This area of the reserve is occasionally visited by vil- lagers to exploit various forest products, such as medicinal plants and honey, as well as for hunt- ing. Above our first camp the preexisting trail ended; here P. J. Rakotomalaza and N. Messmer began their efforts to establish a new trail system toward Pic Trafonaomby. The trail linking camp 1 (440 m) to camp 2 (810 m) passed along a series of undulating hills that climbed slowly in altitude. The trail crossed numerous small streams, some perennial and oth- ers seasonal, and areas with large granitic boul- ders. The camp was located along a tributary of the Andranohela River in a relatively deep valley surrounded by intact forest. Although we were initially under the impression that this forest was undisturbed, pottery dating from the 15th or 16th century was found at the site (Goodman & Rak- otoarisoa, in press). Furthermore, there is a rich modern oral history associated with human oc- cupation of the area (Charles, 1985; Razanabahi- ny, 1995; Rakotoarisoa, 1998) that probably dates from the same period as the pottery. Thus, al- though no modern trails exist in this area of the forest, it was formerly frequented by humans. Our third camp, at 1200 m, was on a ridge above the Andranohela River in a zone with rel- atively steep slopes. Here there was a marked change in topography from the relatively flat val- ley basins associated with the lowlands, particu- larly at camp 1 , to the generally sharp topography and often narrow ridges leading up to the sum- mital region of the Anosyenne Mountains. From just above camp 2 to near Pic Trafonaomby, no sign of recent or past human occupancy or utili- zation of the forest was found. Access to the fourth camp, at 1500 m, was via a narrow ridge that climbed abruptly above the 1200 m camp. The 1500 m site was in a sort of hanging valley or steppe below the steep slopes leading to the summital zone. Sufficient flat ground and running water were found in this val- ley to establish our camp. Numerous small rivers drained the basin, coalesced, and then dropped off the valley in a series of waterfalls that formed the headwaters of the Andranohela River. Pic Trafonaomby (1956 m) is a small dome that rises from a narrow and gently sloping plateau at about 1875 m, and the fifth transect was centered on this plateau. Because of lack of running water in this immediate area, it was necessary to estab- lish our camp associated with the 1875 m transect in an area to the north, about 2-2.5 km from the study zone. The campsite was in an isolated forest fragment with a small running stream, and just above a marsh known locally as Ankebotsy. The limit of the continuous forest that we had entered weeks earlier above the village of Enosi- ary was found to the north of the plateau below Pic Trafonaomby. This disturbed region, although still within parcel 1 , had been opened up for cattle pasture and is regularly burned to provide new forage. The lower slopes to the north and west, outside of the reserve, are largely anthropogenic grasslands with a few remnant islands of forest. Herders with their animals from the high moun- tain villages, such as Vohibaka and Antseva, and those on the lower slopes to the west, such as Esomony, regularly pass through this zone. Days before our arrival at the site around the 1875 m transect, a fire had been set that entered into the forest and seriously damaged the understory of an area approximately 3-4 ha in size. After discussion with local cattle herders that frequented Ankebotsy, it became clear that it was possible to exit parcel 1 via the western slopes of the Anosyenne Mountains. This route takes ap- proximately 5 hours to walk and leads to Eso- mony, where we were met by vehicles. After 7 GOODMAN: DESCRIPTION OF THE RNI D'ANDOHAHELA Table 1-1. Summary of minimum and maximum temperatures and precipitation during 1995 expedition to RNI d'Andohahela. Periods of measurement Temperature (°C)* within each transect Minimum Maximum Rainfall (mm)f Parcel 1 440 m 7, 16.5-20.0, 7, 26.0-28.5, 3, 0.2-27.5, 19-28 Oct. 17.0 ± 1.22 27.9 ± 0.87 12.5 ± 11.29 810 m 9, 11.0-14.0, 9, 15.5-29.0, 5, 0.1-4.0, 28 Oct.-7 Nov. 12.0 ± 1.05 23.6 ± 4.89 1.5 ± 1.33 1200 m 10, 11.5-16.5, 9, 21.5-26.5, 1, 0.7 7-17 Nov. 14.4 ± 1.53 24.6 ± 1.78 1500 m 10, 9.0-15.5, 9, 16.0-21.5, 8, 0.5-11.0, 17-27 Nov. 13.2 ± 1.64 18.0 ± 2.29 5.9 ± 6.00 1875 mt 7, 6.0-12.0, 6, 11.0-25.5, 6, 0.5-17.5, 27 Nov-5 Dec. 9.1 ± 2.03 18.8 ± 5.50 5.0 ± 6.0 Parcel 2 120 m 8, 14.0-22.0, 7, 28.0-37.5, 2, 0.5-1.0 7-15 Dec. 18.4 ± 2.38 33.8 ± 2.75 * Data are presented as number of records, range, mean, and SD. t Data are presented as number of days with rain, range, mean, and SD. + Weather station installed within transect zone rather than near camp. weeks of intensive fieldwork, the Esomony op- tion, rather than the 3-day walk back to where we had entered parcel 1, was ideal. Two days before our final displacement from parcel 1, a member of our group descended to Esomony to organize the porters. The final site visited was in parcel 2 of the re- serve in an area of spiny bush forest to the east- northeast of Hazofotsy. The camp, just a few tens of meters outside the northern limit of parcel 2, was accessible by vehicle and was along a tribu- tary of the Mananara River. The forest, most of which was in the reserve, had an extensive trail system. Meteorology The principal factor giving rise to the remark- able habitat variation of the RNI d'Andohahela is the dramatic difference in annual precipitation be- tween the eastern and western sides of the Ano- syenne Mountains (Ratsivalaka-Randriamanga, 1985, 1987). The moisture-laden weather systems that move into the region from the east to the west are trapped by the windward side of the Ano- syenne Mountains (Paulian et al., 1973). The lee- ward side receives considerably less precipitation. "Over the distance of some sixty kilometres as the crow flies, there is a transition from mean rainfalls of less than 600 mm to amounts in excess of 1500 mm" (Donque, 1972, p. 136). This shift in rainfall is much more abrupt than Donque de- scribed, but owing to a lack of meteorological sta- tions in transition areas, it is impossible to be more precise (Paulian et al., 1973; Donque, 1975; Goodman et al., 1997). Mean annual temperatures are distinctly higher to the west of the Anosyenne Mountains than to the east. More details are pre- sented on the regional climate by Paulian et al. (1973). Along the eastern slopes of the Anosyenne Mountains, covering an elevational range from near sea level to over 1900 m, there is consider- able variation in both rainfall and temperature. Weather stations do not exist along this gradient. Data collected on the minimum and maximum daily temperatures (°C) and daily precipitation during the 1995 expedition to the RNI d'Andohahela can be used to partially examine variation along the eastern slope of this range (Ta- ble 1-1). As expected, shifts in daily minimum and max- imum temperatures showed elevational differ- ences, with the warmest temperatures being in the lower-lying areas and the coldest temperatures to- ward the summital zone. Temperatures at the 1200 m site, however, were on average warmer than those at the 810 m site. Whether this difference FIELDIANA: ZOOLOGY is a question of chance related to passing weather systems during our stay in each zone or is an or- ographic effect is unknown. We had more rainy days at the upper two camps; it rained for 8 of 10 days at 1500 m and 6 of 7 days at 1875 m. This portion of the mountain was often shrouded in clouds and mist for a good portion of the day. We strongly suspect that this pattern was not a tem- poral shift related to seasonal changes as we moved up the slopes over the course of the in- ventory, but rather associated with air circulation over the Anosyenne Mountains (Humbert, 1935). The weather in parcel 2 at the 120 m site was generally clear and warm (Table 1-1). Little rain fell during our 9-day stay at the site. The major exception was a very heavy rain that occurred moments after we arrived at the site. This down- pour occurred before the rain gauge had been in- stalled, but an estimated 20-30 mm of rain fell over the course of 40 minutes. Geology Parcel 1 of the RNI d' Andohahela is dominated by the Anosyenne Mountains. This mountain chain runs, at its southern limit, from just west of Ranopiso north to the Isandra Valley at the base of the Midongy-Sud Massif (Battistini, 1964; Pau- lian et al., 1973). The range is formed from Pre- cambrian gneiss and granitic rocks, and their de- posited alluvium is lateritic or ferralitic soils (Bourgeat, 1972). The eastern slopes form the drainage for the Efaho and Manampanihy rivers and the western slopes for the Mananara and Mandrare rivers. To the west of the Anosyenne Mountains, in- cluding parcel 2 of the RNI d'Andohahela, is a large basin with little relief and largely xerophi- lous vegetation. Soils shift abruptly from the la- teritics of the mountains to silicaceous sands. The region is geologically complex, with the juxta- position and infolding of numerous formations (Noizet, 1953). For more detailed information on the geology of the region see Battistini (1964), Besairie (1970), Brenon (1972), Paulian et al., (1973), and Goodman et al. (1997). Literature Cited Battistini, R. 1964. Etude geomorphologique de l'extreme sud de Madagascar. Editions Cujas, Tou- louse, 340 pp. Besairie, H. 1970. Carte Geologique de Madagascar, 1/ 500,000. Feuille Ampanihy No. 8. Service Geolo- gique et Centre de lTnstitut Geographique National a Madagascar, Tananarive. Bourgeat, F. 1972. Sols sur socle ancien a Madagascar. M6moires ORSTOM no. 57, Paris. Brenon, P. 1972. The geology of Madagascar, pp. 27- 86. In Battistini, R., and G. Richard-Vindard, eds. Bio- geography and ecology in Madagascar. W. Junk, The Hague. Charles, C. S. 1985. Les Mahafale de l'Onilahy: des clans au royaume du XV° siecle a la conquete colon- iale. These Pantheon-Sorbonne, Paris. Donque, G. 1972. The climatology of Madagascar, pp. 87-144. In Battistini, R., and G. Richard-Vindard, eds. Biogeography and ecology in Madagascar. W. Junk, The Hague. Donque, G. 1975. Contribution geographique a l'etude du climat de Madagascar. Nouvelle Imprimerie des Arts Graphiques, Antananarivo, vii + 478 pp. FTM. 1979. Ranopiso, 1:100,000, feuille M-62. Foiben Taosarintanin'i Madagasikara, Antananarivo. . 1990. Tolanaro, 1:100,000, feuille no. 62. Foi- ben Taosarintanin'i Madagasikara, Antananarivo. Goodman, S. M., ed. 1996. A floral and faunal inven- tory of the eastern slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar: With reference to elevational variation. Fieldiana: Zoology, n.s., 85: 1-319. , ed. 1998. A floral and faunal inventory of the Reserve Speciale d'Anjanaharibe-Sud, Madagascar: With reference to elevational variation. Fieldiana: Zo- ology, n.s. 90: 1-246. Goodman, S. M., M. Pidgeon, A. F. A. Hawkins, and T S. Schulenberg. 1997. The birds of southeastern Madagascar. Fieldiana: Zoology, n.s. 87: 1-132. Goodman, S. M., and J. -A. Rakotoarisoa. In press. Un regard sur l'utilisation historique et sur la regen- eration des habitats naturels a Madagascar. Akon'ny Ala. Humbert, H. 1935. L'extinction des derniers vestiges de certains types de vegetation autochtone a Mada- gascar. Archives Museum National d'Histoire Natu- relle, Paris, serie 6, 12: 569-587. . 1941. Le massif de l'Andohahela et ses de- pendances (Madagascar, Reserve Naturelle no. XI). Compte rendu sommaire des seances, Societe de Bio- geographie, 18: 31-37. Institut Geographique National. 1961a. Andranon- dambo, 1:100,000, feuille M-61. Institut Geogra- phique National, Paris. . 1961b. Ranomafana du Sud, 1:100,000, feuille N-61. Institut Geographique National, Paris. 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. Noizet, G. 1953. Carte geologique au 1/100,000 Tran- omaro Marohotro, Service Geologique, Antananarivo, Madagascar. O'Connor, S., M. Pidgeon, and Z. Randria. 1985. Un programme de conservation pour la Reserve d'Andohahela, pp. 31-36. In Mittermeier, R. A., L. A. GOODMAN. DESCRIPTION OF THE RNI D'ANDOHAHELA 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, Gland. Paulian, R., C. Blanc, J.-L. Guillaumet, J.-M. Betsch, P. Griveaud, and A. Peyrieras. 1973. Etude des eco- systemes montagnards dans la region malgache. II. Les chaines Anosyennes. Geomorphologie, climato- logie et groupements vegetaux. (Campagne RCP 225, 1971-1972). Bulletin Museum National d'Histoire Naturelle, Paris. Ecologie generate 1, serie 3, 118: 1- 40. Rakotoarisoa, J. -A. 1998. Mille ans d'occupation hu- maine dans le Sud-Est de Madagascar: Anosy, une tie au mileu des terres. Editions l'Harmattan, Paris. Ratsivalaka-Randriamanga, S. 1985. Recherches sur le climat de Tolagnaro (ex Fort-Dauphin) (Extreme Sud de Madagascar). Madagascar Revue de Geogra- phic 46: 47-67. . 1987. Climat et vegetation de la region de Fort-Dauphin. Recherches pour le Developpement, serie Sciences de 1' Homme et de la Societe, 3 (Pre- mier semestre): 51-64. Razanabahiny, V. 1995. Le Dina (Convention entre Membres de Communautes Villageoises) son oppor- tunite ou non dans la conservation de la nature. Cas de la Reserve Naturelle Integrale d'Andohahela — To- lagnaro. Memoire C. A. P.E.N, Ecole Normale Super- ieure, Universite d' Antananarivo. Appendix 1-1 Participants in the Project (Field and Laboratory) A total of 30 scientists and field-workers from seven different countries were involved in this multidisciplinary study. This number included the field participants listed in the previous section as well as researchers responsible for some of the laboratory studies. The addresses of all scientific participants follow: Andriambelo, P. Z., Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE), ORSTOM, B.R 434, Antananarivo, Madagascar. Andriatsiferana, E, Missouri Botanical Garden, B.P. 3391, Antananarivo (101), Madagascar. Carleton, M. D., Division of Mammals, National Museum of Natural History, Smithsonian Insti- tution, Washington, D.C. 20560, U.S.A. Elouard, J.-M., Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE), ORSTOM, B.P. 434, Antananarivo, Madagascar. Feistner, A., Jersey Wildlife Preservation Trust, Les Augres Manor, Trinity, Jersey JE3 5BP, Channel Islands. Fisher, B. L., Department of Entomology, Uni- versity of California, Davis, California 95616, U.S.A. Current address: Life Sciences Divi- sion, South African Museum, P.O. Box 61, 8000 Cape Town, South Africa. Gattolliat, J.-L., Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement, ORSTOM, B.R 434, Antananarivo, Madagas- car. Current address: Musee Cantonal de Zoo- logie, CP 448, CH-1000, Lausanne 17, Swit- zerland. Gibon, F.-M., Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE), ORSTOM, B.P. 434, Antananarivo (101), Madagascar. Goodman, S. M., Field Museum of Natural His- tory, Roosevelt Road at Lake Shore Drive, Chi- cago, Illinois 60605-2496, U.S.A., and World Wide Fund for Nature, Aires Protegees, B.P. 738, Antananarivo (101), Madagascar. Hawkins, A. E A., B.P. 8511, Antananarivo (101), Madagascar. Helme, N. A., 189 Main Road, Kalk Bay, 7975, South Africa. Jenkins, P. D., Mammal Group, The Natural His- tory Museum, Cromwell Road, London SW7 5BD, United Kingdom. Laha, R., World Wide Fund for Nature, B.P. 42, Tolagnaro (614), Madagascar. Lourenco, W R. Laboratoire de Zoologie (Arthro- podes), Museum National d'Histoire Naturelle, 61, rue de Buff on, 75005 Paris, France. Nussbaum, R. A., Museum of Zoology, The Uni- versity of Michigan, Ann Arbor, Michigan 48109-1079, U.S.A. OConnor, B. M., Museum of Zoology, The Uni- versity of Michigan, Ann Arbor, Michigan 48109-1079, U.S.A. Oliarinony, R., Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE), ORSTOM, B.P. 434, Antananarivo, Madagascar. Messmer, N., Conservatoire et Jardin Botaniques de la Ville de Geneve, Case Postale 60, CH- 1292 Chambesy/Geneve, Switzerland. Pidgeon, M., World Wide Fund for Nature, B.P. 738, Antananarivo (101), Madagascar. Current address: Route de St. Cergue, 1270 Trelex, Switzerland. Pilaka, T, Laboratoire de Recherche sur les Sys- temes Aquatiques et leur Environnement FIELDIANA: ZOOLOGY (LRSAE), ORSTOM. B.P. 434, Antananarivo (101), Madagascar. Rakotoarison, J. M., Institute for the Conservation of Tropical Environments. B.R 3715, Antana- narivo (101), Madagascar. Rakotomalaza, R J., Missouri Botanical Garden, B.R 3391, Antananarivo (101), Madagascar. Rakotondrainibe, E, Ecole Pratique des Hautes Etudes, 16, rue de Buffon, 75005 Paris, France. Ramanamanjato, J.-B., Departement de Biologie Animale, Universite d' Antananarivo, B.P. 906, Antananarivo (101), Madagascar. Randriamampionina, B., World Wide Fund for Nature, B.R 42, Tolagnaro (614), Madagascar. Raselimanana, A., Departement de Biologie Ani- male, Universite d' Antananarivo, B.P. 906, An- tananarivo (101), Madagascar. Current ad- dress: World Wide Fund for Nature, B.P. 738, Antananarivo (101), Madagascar. Ravokatra, M.. World Wide Fund for Nature, B.P. 738, Antananarivo (101), Madagascar. Raxworthy, C. J., Center for Environment Re- search and Conservation, Columbia University, 1200 Amsterdam Avenue, New York, New York 10027, U.S.A. Current address: Natural History Museum, University of Kansas, Law- rence, Kansas 66045-2454, U.S.A. Sartori, M., Musee Cantonal de Zoologie, CP 448. CH-1000, Lausanne 17, Switzerland. Schmid. J. Deutsches Primatenzentrum, Kellner- weg 4, Gottingen 37077, Germany, and Abtei- lung fur Verhaltenphysiologie, Beim Kupfer- hammer 8. 72070 Tubingen, Germany. GOODMAN: DESCRIPTION OF THE RNI D'ANDOHAHELA Chapter 2 An Overview of the Botanical Communities of the Reserve Naturelle Integrale d'Andohahela, Madagascar Nick A. Helme1 and Pierre Jules Rakotomalaza2 Abstract An outline of the botanical communities of the Reserve Naturelle Integrale d'Andohahela is presented. The study area was centered on an elevational transect between 440 and 1950 m along the eastern slopes of parcel 1 of the reserve, and also within a limited area of the spiny bush forest of parcel 2 at 120 m. A shift in species composition of pteridophytes and, somewhat less pronouncedly, in angiosperms occurred between the first (420 m) and second (810 m) transect zones, apparently reflecting a shift from humid lowland to mid-altitude forest, but a true mid-elevation montane community was not present below 1000 m. Monocarpic Acantha- ceae and bamboo tended to dominate the understory within the mid-altitude and montane zones (800-1800 m). The exposed ridges above 1200 m supported a distinct, small-leaved, sclero- phyllous forest community characterized by low, gnarled trees that were covered in epiphytes. Large trees, up to 30 m tall, were recorded in the valleys at 1600 m, but they were significantly smaller above this elevation. The upper slopes ( 1 800-2000 m) supported humid montane forest, with an abundance of both epiphytic and terrestrial mosses and ferns. Exposed rock was a feature of the highest elevations, where plant communities included Philippia scrubland and shallow-soil areas dominated by sedges and grasses. Parcel 2, which has low rainfall, had the typical spiny bush forest of the region. It was dominated by species of Euphorbia and Didi- eraceae, with a dense, shrubby understory. Resume Les communautes botaniques de la Reserve Naturelle Integrale d'Andohahela sont presentees en grandes lignes dans ce chapitre. Le site d'etude est axe sur un transect altitudinal compris entre 440 et 1950 m d'altitude sur le versant oriental de la parcelle 1 de la reserve, ainsi que dans le fourre xerophile de la parcelle 2 de la reserve a 120 m d'altitude. Bien que peu marque, un changement de la composition floristique, plus evident toutefois au niveau des pteridophytes que des angiospermes, est observe entre le premier site d'etude (420 m) et le deuxieme (810 m), refletant le passage de la foret dense humide de basse altitude vers la foret de moyenne altitude. Une foret de moyenne altitude typique n'est cependant pas observee en dessous de 1000 m d'altitude. Des especes monocarpiques, des Acanthaceae et des bambous ont tendance a dominer le sous-bois des zones de moyennes altitudes et des forets de montagnes (800-1800 m). Les cretes exposees au-dessus de 1200 m incluent une foret sclerophylle caracterisee par des arbres bas, tortueux, aux feuilles petites, et couverts d'epiphytes. En revanche, des arbres 1 189 Main Road. Kalk Bay, 7975, South Africa. 2 Missouri Botanical Garden, B.P. 3391, Antananarivo (101). Madagascar. HELME & RAKOTOMALAZA: BOTANICAL COMMUNITIES 1 1 atteignant presque 30 m de hauteur sont a nouveau observes dans les vallees a 1600 m. Au- dessus de cette altitude, la hauteur des arbres diminue significativement. Le versant entre 1 800 et 2000 m est domine par une foret humide de montagne, riche en mousses et fougeres epi- phytes et terrestres. Les rochers exposes presentent des fourres ericoi'des caracteristiques des plus hautes altitudes, et les zones a sol peu profond sont dominees par des Cyperaceae et des Poaceae. La parcelle 2 de la reserve quant a elle contient un fourre xerophile typique de cette region, domine par des especes d' Euphorbia et de Didiereaceae, et a sous-bois dense et arbustif. Introduction The earliest botanical exploration of the An- dohahela Massif was carried out by Humbert (1935, 1941), who discovered some interesting species, mainly along the summital ridge. The re- sults of his visits to the region formed the basis for establishing the Reserve Naturelle Integrate (RNI) d'Andohahela. Various other botanists (Ca- puron, Cours Darne, and Saboureau) explored the lower western slopes of the massif but did not publish information regarding their botanical col- lections. No further studies were made in the for- ests of the eastern slope until the 1980s, when Missouri Botanical Garden (MBG) staff started a program in the area, concentrating on the lower slopes and adjoining ridges. Other than the initial exploration by Humbert, however, the mid-alti- tude and upper forests remained poorly known until the current expedition. The RNI d'Andohahela is the most southerly example of Malagasy humid montane forest. The entire reserve, composed of three noncontiguous parcels, is south of the Tropic of Capricorn. Al- though the forests are thus not strictly tropical, they are of the same broad type as is found further north along the eastern escarpment of the island. Due to the topographic variation within parcel 1 (300-2000 m), three major phytogeographic do- mains (sensu Humbert, 1955) are represented: the Eastern Domain, or humid lowland forest (tradi- tionally 0-800 m); the Central Domain, or mid- altitude humid forest (traditionally 800-2000 m); and the High Mountain Domain, or dwarf mon- tane forest and heathland (usually above 1900 m). Du Puy and Moat (1996) slightly modified the classification of evergreen formations of eastern and central Madagascar, delimiting the zone of mid-altitude humid forest to 800-1800 m and re- ferring to lower humid montane forest as that oc- curring from 1800 m to 2000 m. Their classifi- cation is not used in this chapter. A unique and important feature of the area is the phenomenal rainfall gradient from the upper slopes of parcel 1 (e.g., at the base of Pic Trafonaomby) toward the spiny forest in the west encompassing parcels 2 and 3. Over a horizontal distance of no more than 15 km, the average annual rainfall drops from an estimated 3,000 mm to 500 mm (Paulian et al., 1973), producing one of the most striking biotic gradients that can be seen anywhere in Madagascar (Goodman et al., 1997). Although none of our study sites were on these western slopes of parcel 1 , most of which fall outside the boundaries of the reserve, it should be borne in mind that there are nevertheless some steep rain- fall gradients within the relatively small area (63,000 ha) of parcel 1 of the RNI d'Andohahela. This summary focuses on the environmentally more homogeneous eastern slopes of parcel 1. It is by no means a definitive account of the entire reserve, owing to the presence of the abrupt hab- itat gradient on the leeward side of the mountain. Information is also provided on parcel 2, the dry, spiny forest portion of the reserve. For further de- tails on the topography, climate, and geology of the reserve and information on study sites, see Chapter 1. A further important feature of the eastern slopes of the RNI d'Andohahela is that the forests are still largely intact, providing a fine example of the unbroken transition from lowland to mid- elevation to eastern humid montane forest. Our lower two elevational sites, at 440 and 810 m, were located in areas that probably supported a small human population many centuries ago, while the upper three study sites were in appar- ently virgin forest. Although most of the forest at the lower sites was mature humid lowland forest, there were indications that certain areas had been disturbed several hundred years ago. In any case, the forest throughout the eastern slopes is essen- tially undisturbed today, with few preexisting trails, and there is no logging or subsistence ag- riculture (tavy) above about 300 m. A detailed analysis of the structure, floristics, and phytogeography of the area, based largely on 1 ha plots, is presented in Chapter 4. This chapter 12 FIELDIANA: ZOOLOGY reviews the characteristic aspects of the vegeta- tion and flora of the various elevational zones. 440 m After leaving behind the last of the Enosiary rice paddies and open anthropogenic grasslands, at an altitude of about 300 m, we entered an area of disturbed forest, with numerous small lianas and traveler's palm (Ravenala madagascariensis). The forest at this low elevation appeared to be more deciduous than that growing slightly higher up the valley, in the vicinity of our study area, centered at 440 m. Various saxicolous plants such as Kalanchoe were seen growing on the enormous granite boulders that littered the valley. Our initial camp was located along the Andranohela River at an altitude of 440 m. The forest in the valley and along the lower slopes was structurally humid lowland forest, with a canopy of 15-20 m and a large component of buttressed, emergent species 25-30 m tall (Fig. 2-1). There was a high liana density, and although most were small, some enormous Piper lianas were recorded. Epiphytes were present, but by no means abundant; they in- cluded the bird's nest fern {Asplenium nidus), Po- thos scandens (hemi-epiphytic Araceae), and var- ious mosses that covered no more than 20% of the available surface area. Common emergents in- cluded Dilobeia thouarsii, Chrysophyllum boivi- nianum, Sloanea rhodantha var. rhodantha form quadriloba. and Ocotea spp. Canopy species in- cluded Sorindeia madagascariensis, Ilex mitis, Syzygium spp., Oncostemum spp., Tambourissa spp., Diospyros spp., Polyscias sp., Dracaena re- flexa, Dombeya spp., Treculia sp., and various Rubiaceae. The understory was composed of nu- merous seedlings of the canopy species plus dif- ferent Tambourissa spp., Noronhia sp., and vari- ous terrestrial ferns. There was a dense leaf litter in most areas. During our visit to this site, at the end of the dry season, the soil surface was dry, but some moisture occurred a few centimeters be- low the surface. Large palms were not particularly common or diverse, with only one Ravenea and two Dypsis spp. noted. Bamboo clumps were not an important feature of the forest at this elevation. Dalbergia spp. were uncommon, tending to have clumped distributions. Dominant families in the transect (trees with a dbh >10 cm) were Rubi- aceae, Clusiaceae, Lauraceae, Elaeocarpaceae, and Myrsinaceae (Chapter 4, Table 4-5). The rocky, fast-flowing clear water Andrano- hela River supports a well-developed riparian for- est at this elevation, with typical widespread riv- erine plants such as Aphloia theiformis, Ficus sp., Antirrhoea sp., Weinmannia spp., Phyllanthus spp., Albizia sp., Micronychia sp., and Dombeya spp. Ravenala was more common in light gaps along the river than in tall forest. The big black gneissic rocks along the river margins were worn smooth, but a number of grasses and small herbs, such as Lobelia, managed to find a foothold in sandy cracks. Below the waterline were large pop- ulations of the unusual subaquatic plant Hydro- stachys sp., which flowers when the water level drops in the dry season. This genus of some 22 species is restricted to Africa and Madagascar, with most of the species being endemic to Mad- agascar. The riverine plant community is relative- ly constant at elevations below about 1000 m. Above this altitude genera such as Ravenala tend- ed to drop out, being replaced by Cyathea in the narrower streambeds with their lower filtering light values. 810 m The trail between 440 and 810 m was remark- able for its density of lianas. Numerous large granite boulders were present, and in certain areas away from the river the usually loamy soils be- came distinctly sandy, with a high content of large quartzitic crystals derived from the weathered granite. The trail crossed a number of small streams, the majority of them essentially seasonal, although two were perennial, with deep pools and large waterfalls plunging off the steep southeast- facing slopes. Our 810 m camp was situated at the confluence of a perennial tributary and the Andranohela River, and was surrounded on all sides by steep slopes clothed in dense, humid low- land forest (Fig. 2-2). The flowering plant flora in the 810 m eleva- tional zone was not dramatically different in struc- ture or composition from that at 440 m. For pte- ridophytes this difference was much more nota- ble. The fern community showed a distinct spe- cies turnover between the first (420 m) and second (810 m) transect zones (see Chapter 3) that ap- parently reflected a shift from humid lowland to mid-altitude forest. At 810 m in parcel 1 virtually all the emergent and large canopy trees had at least 20-50% epiphyte cover, with Usnea lichens, Asplenium ferns, and Bulbophyllum orchids dom- inating. The hemi-epiphytic Pothos scandens, so HELME & RAKOTOMALAZA: BOTANICAL COMMUNITIES 13 Fig. 2-1. View of humid lowland forest at 440 m in parcel 1 of the RNI d'Andohahela. In this area the canopy height was generally 15-20 m, with emergent species reaching up to 25-30 m. There was a high liana density, and epiphytes were present but not abundant, covering less than 20% of the available surface area. The understory was composed of numerous seedlings of canopy species (Sorindeia madagascariensis, Ilex mitis, Syzygium spp., Oncos- temum, Tambourissa, Diospyros, Polyscias, Dracaena reflexa, Dombeya, Treculia, and Rubiaceae), plus different Tambourissa, Noronhia, and terrestrial ferns. (Photograph by M. Pidgeon.) 14 FIELDIANA: ZOOLOGY Fig. 2-2. Tributary of the Andranohela River at about 820 m and just above our second camp in parcel 1 of the RNI d'Andohahela. The canopy height here was similar to that of the 440 m site, and emergent species consisted of Sloanea rhodantha var. rhodantha form quadriloba, a few scattered Canarium obovatum, Dilobeia thouarsii, Ocotea, Myrtaceae, and Moraceae. The majority of large trees had at least 20-50% epiphyte cover, and moss was common on vertical and horizontal branches. The understory was not dominated by a Tambourissa shrub as at 440 m, but by two species of Acanthaceae that reached 1.5 m in height. (Photograph by N. Helme.) common at 440 m, had virtually disappeared. Moss was common on vertical and horizontal branches. The canopy height at 810 m was similar to that at 440 m. Emergents averaged 20-30 m, while the main canopy was usually between 15 and 20 m. Emergents included the ubiquitous Sloanea rho- dantha var. rhodantha form quadriloba, a few scattered Canarium obovatum, Dilobeia thouarsii, Ocotea spp., Myrtaceae spp., and Moraceae spp. The canopy was dominated by members of the Moraceae (Trilepisium, Treculia, Streblus), Myr- taceae (especially Syzygium spp.), and Monimi- aceae {Tambourissa spp.), and included other gen- era such as Macaranga, Allophylus, Plagioscy- phus, Filicium, Antidesma, Noronhia, Vepris, Po- lyscias, Oncostemum, and various Clusiaceae. The understory was dominated not by a Tam- bourissa shrub, as at 440 m, but by two species of Acanthaceae that reached 1.5 m in height. This domination by a monocarpic giant herb, which be- came even more pronounced at higher elevations, is a feature of many paleotropical montane forest understories (Davis et al., 1994). Bamboo (especially the lianescent Nastus) was fairly common and occasionally formed dense tangles, especially along light gap edges. There seemed to be fewer lianas than at 440 m. Various species of Oncostemum were common understory shrubs at elevations up to 1200 m, and their red berries were regularly seen being eaten by frugiv- orous birds such as Philepitta castanea. Large palms were rare at 810 m, although small under- story species of Dypsis were common. There was a high degree of heterogeneity in this zone, with remarkably different plant com- munities adjacent to each other, suggesting some strong environmental/edaphic gradients that were not immediately evident to us. Just north of our camp, for example, an area at the base of the ridge leading toward the 1 200 m site was characterized by a more open understory with fewer liana tan- HELME & RAKOTOMALAZA: BOTANICAL COMMUNITIES 15 Fig. 2-3. View of regenerating vegetation along natural landslide just above the Andranohela River at about 800 m in parcel 1 of the RNI d'Andohahela. The landslide probably occurred within the preceding 10-15 years, as judged by the size of the small trees, and it appeared to have been caused by water-saturated earth slumping down a relatively steep slope. The area is characterized by pioneer species such as the fern Dicranopteris linearis, the trees Dombeya, Croton, Weinmannia, and Maesa, and on drier, thinner soils by a species of Philippia and a tall Helichrysum. Nastus bamboos were common along the forest/landslide boundary. (Photograph by N. Helme.) gles, species such as Cyathea (indicating moist, deep soils), patches of Ivodea madagascariensis (Rutaceae), and small Dypsis palms. The bamboo Nastus was common, its leaves often making up the bulk of the leaf litter. Large specimens of Dal- bergia were more common than elsewhere (this also seemed to be their altitudinal limit), and they shared the canopy with Dombeya, Grewia, Trile- pisium, Macaranga, Brillantaisia madagascarien- sis, Filicium, and Allophylus. Another unusual habitat was formed by what was clearly regeneration on a landslide track (Fig. 2-3). Similar patches of vegetation could be seen throughout the study area, and although they were not abundant (perhaps one per 200 ha), they un- doubtedly play an important role in the dynamics of the forest, providing regeneration opportunities for light gap species. One such area was visible along the river at 800 m, covering about 1.5 ha on a 50° slope. The landslide probably occurred relatively recently, within the preceding 10-15 years, judging by the size of the small trees, and it appeared to have been caused by water-saturat- ed earth slumping from a particularly steep por- tion of the slope. The seep was still visible and was in fact so wet that Typha bulrushes were found growing at the base of the slide. The area was characterized by pioneer species such as the fern Dicranopteris linearis and trees Dombeya, Croton, Weinmannia, and Maesa. The drier hab- itats, found on the thinner soils with underlying rock, were characterized by a species of Philippia and a tall Helichrysum. Nastus was common along the forest/landslide boundary. 1200 m This altitudinal zone was located on a narrow ridge that was reached by climbing a steep trail up from the Andranohela River. Being on a ridge, both groundwater and large boulders were much rarer than at the lower sites. Much of the ground was covered by a spongy carpet of moss and fine 16 FIELDIANA: ZOOLOGY ¥/ mi O w*^~:' ^ *«fs8 Jm^nl '^ffiffifr T IT « ^X-jj 5?£* ^-_ P£\* j* '( . v \»* . 13fV ^"w*}- ^f ii ' ■ Mr' I IB Sis*- '* :; t *■ f. \r {■■» # ■•■• ► £f ►• V ' . > . • ' -'■■'.»* pi ■■■ ■ % •• • ^ • • . ■ • • ■• 5* • Fig. 2-4. View from edge of steep slope at about 1200 m in parcel 1 of the RNI d'Andohahela. Evening mist was a regular occurrence in this zone and at higher elevations, related to increased epiphyte cover (50-80%), mostly mosses and lichens. Lianas were still present in certain areas, but they were on average less common than at lower elevations. The understory was relatively open and usually consisted of Acanthaceae. young saplings of canopy species, and some Oncostemum, Cyathea, and Marattia fraxinea in the moister areas. (Photograph by N. Helme.) root material that was to occur from this zone on up to the summit. The underlying dry soil was not sandy and turned quickly to dust by the passing of many feet. A major structural change occurred progres- sively between 1000 and 1200 m, highlighting the shift from lowland to moist montane forest. Prob- ably the most obvious change was an increase in tree density, with 50% more stems in our 1 ha plot than in those plots at 440 and 840 m (Chapter 4, Table 4-1). Paralleling this increase in total stem number was a small decrease in mean stem diameter. The plot was dominated by Myrtaceae {Syzygium spp.), with other important families be- ing Lauraceae (Cryptocarya, Ocotea), Sterculi- aceae (Dombeya), Moraceae, and Monimiaceae {Tambour issa). This high density of trees was re- peated in our plot at 1950 m, where Lauraceae was once again dominant. This was the only el- evation at which large palms (Ravenea sp.) formed a significant part of the canopy. Canopy height varied from 12 to 20 m, and there were no clear emergents. Sloanea rhodantha was still present, but perhaps because of its lesser average height at this elevation, it did not have the enor- mous buttress roots so characteristic of this spe- cies at lower elevations. Dilobeia thouarsii dropped out completely by 1 000 m, but Canarium boivini was still present in small numbers. Other large canopy elements included the predominantly lowland Chrysophyllum boivinianum and the widespread Croton monge. Evening mist was a feature of the forest at 1200 m and above; this was reflected in the substantial increase in epiphyte loads (Fig. 2-4). The majority of the epiphyte community consisted of mosses and lichens (50-80% cover), with ferns (especial- ly Hymenophyllum and Asplenium spp.) making up a smaller component. Epiphytic orchids (Bul- bophyllum spp.), melastomes (Medinilla sp.), Rhipsalis baccifera, and Peperomia spp. were present in low numbers. Lianas were still common in certain areas, but they were on average less common than at lower elevations. The understory HELME & RAKOTOMALAZA: BOTANICAL COMMUNITIES 17 was relatively open and usually consisted of Acanthaceae, young saplings of canopy species, some Oncostemum spp., and the ferns Cyathea spp. and Marattia fraxinea in the moister areas. The forest along the ridge leading up to the higher zones of the massif was structurally and floristically very different from the forest on the small plateau above the 1200 m site. The ridge had thin soils and was exposed to strong winds and rapid changes in humidity, with the result that the canopy was seldom above 8 m and trunks were often twisted and densely packed. Prominent trees included Brachylaena, Weinmannia spp., Symphonia, Anthocleista, Syzygium spp., Macar- anga, and various Lauraceae, Rubiaceae, and Ar- aliaceae. The epiphyte load was heavy. The un- derstory was often dominated by a species of Arundinaria bamboo that grew to 2 m and ap- peared to have an inhibiting influence on the re- generation of other small plants. The lianescent bamboo Nastus was also present on the ridge, but it seldom dominated. At various points along the ridge were prominent granite outcrops, and on these grew another, larger species of bamboo {Arundinaria sp.) with irritating hairs on the culm leaves. This species was common within the 1875 m elevational zone in similar rocky habitats. 1500 m The trail between the 1200 m and 1500 m zones followed the ridge leading up to Pic Tra- fonaomby and passed through the stunted ridge forest with its bamboo understory. The 1500 m site was situated in a large bowl (virtually a hang- ing valley) to the southeast of the peak. This bowl was drained by a number of small streams that came together just below the camp to form one of the main tributaries of the Andranohela River. The soils in the area were deep, presumably hav- ing accumulated from centuries of erosion of the surrounding slopes, and large boulders had also come to rest in the valley. The forest in the valley was classified as mid- altitude montane forest (Humbert & Cours Darne, 1965), while the forest on the ridges had distinctly sclerophyllous characteristics typical of more ex- posed situations. The trees in the valley bottom, dominated by Sloanea rhodantha var. rhodantha form quadriloba, were exceptionally large, often attaining a height of 25 m, an extremely unusual situation at this altitude. The deep soils and abun- dant water in the valley may partially explain this unexpected vigor. Strongylodon lianas were abun- dant and often attained a great size in the valley habitat, but they quickly disappeared on the thin- ner ridge soils, as did Sloanea. The slopes im- mediately above the Sloanea zone, at the base of the ridge, were dominated by Ravensara and a large species of Tambourissa. A highly character- istic tree at this elevation was the narrow-leaved Pandanus vandamii, a common canopy element in both valleys and on ridges; this species also occurs in the Western Domain. Large tree ferns {Cyathea spp.) were also a typical feature in this area, indicating the permanently moist nature of the site. Notable by its absence was Podocarpus mad- agascariensis, not recorded anywhere on the mas- sif but previously reported from Beampigaratra, Mandena, and Mahatalaky and known from lit- toral zones up to 1800 m (Koechlin et al., 1974). The apparent absence of this genus in the RNI d'Andohahela is in striking contrast to the RNI d'Andringitra, where it was the second most com- mon tree at 1600 m (Lewis et al., 1996). Other notable absences (all common at lower elevations) included the dwarf Dypsis palms, the 7-m-tall Dypsis scottiana, Diospyros spp., Moraceae, and the large tree Chrysophyllum boivinianum. Large areas in the valley were dominated by the climbing bamboo Arundinaria, which forms dense stands about 5 m above the ground. Scat- tered large trees can be found in these bamboo patches, rising up above the smothering layer, suggesting that the bamboo has only recently col- onized the immediate area. Had the bamboo been there for a long time it is unlikely that the saplings would have survived in the light-starved environ- ment below. This then begs the question as to just what caused the bamboo invasion. One possible scenario is that there was a tree fall, and the bam- boo was able to take advantage of this light gap to colonize the immediate surroundings, outcom- peting other species and gradually coming to dominate the understory. The bamboo will at some stage flower and die, and forest dynamics will then once again determine whether it is bam- boo or woody species that fill the gap. It is diffi- cult to speculate about such vegetation patterns when they are viewed through such a small win- dow of time. Certain areas along the ridge, up to 1750 m, had a dense understory of the same 2-m-tall Nas- tus bamboo. The moister areas in the valley were often dominated by extremely dense populations of a 1.8-m-tall Impatiens, particularly in light 18 FIELDIANA: ZOOLOGY gaps, while the slightly better drained areas sup- ported the monocarpic Ruellia, Gravesia dichaet- antheroides, Vernonia leandrii, Blechnum ferns, and Lapertea meddellii. Common understory shrubs included Pleuridantha liallyii, Filicium, Aphloia, and Oncostemum spp. There were some large clearings (0.5 ha) in the forest, perhaps caused by cyclone-generated windthrow, that were invaded by various leguminous plants, Ur- ticaceae, Acanthaceae, and Balsaminaceae, and by pioneer tree species such as Macaranga and Dom- beya. The reason why bamboos were dominant in some areas and absent in other areas is not known. Epiphytes were predictably abundant, with 80- 100% cover on most horizontal branches. Orchids were no longer dominated by Bulbophyllum, and they included numerous small species. The un- usual and attractive liliaceous epiphyte Rhodoco- don urgineoides, with small white flowers, was common on the trunks of Sloanea and other rough-barked species, and it appeared to be re- stricted to this elevation. The pink-flowered Bak- erella clavata was a common branch parasite, as it was throughout the altitudinal range from 400 to 1800 m, but it was joined by a yellow-flowered form of B. clavata at the higher elevations (>1600 m). The forest on the ridges was composed of small-leaved, twisted trees 8-10 m tall, a growth form molded by the strong winds that are a fea- ture of this habitat. Mosses and lichens covered the trees and created a deep, spongy carpet on the ground. Typical canopy species included Dicory- phe viticoides, Tina isoneura, Elaeocarpus sp., Gaertnera sp., Agauria sp., Anthocleista sp., Dombeya subsquamosae, Ephippiandra sp., and Weinmannia spp. 1875 m Our highest study elevation was located on the northern slopes of Andohahela Massif, due north of Pic Trafonaomby. After leaving the 1 600 m site we passed through humid montane forest and, near the summit, into a zone of moss and bamboo with a high proportion of exposed rock. At 1 875 m the bamboo gave way to a peculiar 8-m-tall montane forest (sensu Humbert & Cours Darne, 1965) with an open understory consisting almost entirely of a single species of sedge, Apodocepha- la pauciflora. As we descended the northern side of the peak the terrain once again became rocky, and the sedge rapidly disappeared. A small pla- teau sloped gently down toward the north before becoming steeper at about 1800 m, and this area of deeper soils was characterized by a dense un- derstory of Acanthaceae, with a few specimens of Gesneriaceae and a humid montane forest canopy of 6-10 m (Fig. 2-5). Secondary grasslands and burnt mid-altitude forest marked the northern boundary of the pristine forest. These short grass- lands and remnant forests form a mosaic that is constantly changing according to the patterns of encroaching man-made fires that may scour the area in the dry season. Some large patches (>100 ha) of remnant forest on these northern slopes be- tween 1400 and 1800 m are still connected by viable forest corridors. Below about 1300 m on the northern slopes the mid-altitude forests end abruptly and give way to secondary grasslands. Still lower, patches of dry forest indicate the be- ginning of the transitional zone between humid and spiny bush forest. Because trees extend right to the top of the mountain, there is no real tree line as there is on certain of the higher peaks in Madagascar. Where trees are not present (mainly on rocky outcrops), having been replaced by shrubs and grasses, this is due to soil depth rather than the effects of al- titude. The most extreme form of the stunted scle- rophyllous forest is, not surprisingly, found on the summit of Pic Trafonaomby (1959 m) itself. The peak is a small granite outcrop, partly covered in sclerophyllous forest on the thin soils, with shrubs and succulents on the rock. The canopy was no more than 3 m tall and consisted of Philippia spp., Alberta, Pittosporum, Agauria, Vaccinium, and Dombeya seyrigii. The ground was covered by a thick mat of moss and Lycopodium, as is common in much of the area above 1 800 m. Exposed rock near the summit supported a large Aloe humbertii, numerous Helichrysum spp., a sedge Kyllinga plurifoliata, and the herbaceous Kniphofia ankar- andrensis. The epiphyte load was heavy, with most tree limbs being densely clothed in moss. Lianas were virtually absent. The 1 ha plot was located at 1950 m on the deep soil plateau with an Acanthaceae understory, just below and to the north of the peak. Although there was a high stem density, species diversity was low, and the area was dominated by the fam- ilies Araliaceae (three species), Lauraceae (seven species), Myrtaceae (seven species), Clusiaceae (one species), and Flacourtiaceae (two species). No palms were recorded within the 1875 m zone, the highest elevational record being scattered Dypsis individuals at 1600 m (both north and south slopes). HELME & RAKOTOMALAZA: BOTANICAL COMMUNITIES 19 Fig. 2-5. View of humid montane forest at 1875 m on plateau below Pic Trafonaomby in parcel 1 of the RNI d'Andohahela. This is the site of the fifth 1 ha plot (see Chapter 4). The forest, with a canopy of 6-10 m, was on a deep soil plateau and characterized by a high stem density with low species diversity, dominated by the families Araliaceae, Lauraceae, Myrtaceae, Clusiaceae, and Flacourtiaceae. The dense understory was made up of Acanthaceae and a few specimens of Gesneriaceae. (Photograph by N. Messmer.) Our camp was located in a valley at about 1700 m, in a patch of mid-altitude montane forest that had been severely degraded by cattle trampling and grazing. The forest patch was separated from the main forest to the south by a band of second- ary grassland approximately 300-900 m wide. Smilax was a common climber in these disturbed forests, and the 6-10-m-tall canopy was com- posed of Tina, Philippia, Cephalostachyum and Nastus bamboos, Cassinopsis madagascariensis, and a Weinmannia sp. The forest margin was made up of a predictable community of shrubs consisting largely of Philippia, together with Vac- cinium and various Asteraceae, such as Brachy- laena merana. The grassland in the vicinity of our camp con- tained a number of species not found elsewhere on the expedition, such as the attractive Gesnier- aceae Tachiadenus longiflorus, a shrubby umbel- lifer Helichrysum spp., a spiny Mimosa dasyphyl- la, and numerous sedges and grasses (Sporobolus centrifngus, Bulbostylis hispidula, and Eleocharis limosa). It is not known whether any of this short grassland is natural (it is short largely because of cattle grazing) or whether it is all secondary, the result of repeated burning of the forest and cattle damage. It is suspected that small areas of natural grassland probably existed around the exposed and highly weathered granite domes in the area, where the soils are too thin to support a forest, although the species present now are not neces- sarily members of the original suite. Supporting evidence comes from a large dome to the east of camp that had apparently never been burned (judging by the surrounding forest and because its steep slopes made it unsuitable for cattle). This large area had few species in common with the grasslands around camp, although the life forms were similar, including various abundant sedges (Costularia baroni), a grass {Danthonia maco- wanii), Myrica phillyreaefolia shrubs, Viscum tieghemii, and scattered Philippia spp. Poor drain- age on the shallow soils is further indicated by Sphagnum moss and the fern Blechnum tabulare. 20 FIELDIANA: ZOOLOGY Fig. 2-6. View of the descent along the western slopes of the Anosyenne Mountains toward Esomony at about 1650 m. The lower limit of the forest was created by human disturbance, largely fire, and the lower grasslands are virtually treeless. The few trees present are scattered Brachylaena and Agauria. At about 900 m the first Kalanchoe beharensis and Pachypodium lamerei were noted, indicating the moist/dry forest ecotone. (Photograph by N. Helme.) Pteridium aquilinum (bracken) was absent, sug- gesting that the area had not been seriously dis- turbed by man and supporting the idea that some sort of grassland may also have been the natural vegetation around the domes near camp. An unusual habitat occurred on the northern slopes below camp, at about 1550 m, in an area known locally as Ankebotsy. The area was per- manently moist, seasonally flooded, and had typ- ical marsh vegetation, characterized by an ab- sence of trees and an abundance of sedges. The 2 ha area had been badly overgrazed by cattle and burned too frequently, but it still supported some interesting species such as the lily Kniphofia an- karandrensis. Sphagnum moss, an Eriocaulaceae, Philippia, and the fern Osmunda regalis. From the 1700 m camp we descended on the footpath to the village of Esomony, located at the northwestern base of the Andohahela Massif. The lowest forest is at about 1300 m and is typical mid-montane forest, as found on the southeastern side of the mountain, although the local rainfall is almost certainly less. The lower limit of the forest was created by fire, and the grassy slopes below are virtually treeless (Fig. 2-6). The few trees present are scattered Brachylaena and Agau- ria, the latter seemingly protected from the flames by its thick bark. In the moist ravines we found small groves dominated by a Weinmannia. At about 900 m we encountered the first Kalanchoe beharensis and Pachypodium lamerei, indicating that we had already crossed the moist/dry forest ecotone. Not much transitional or spiny forest re- mains in the area, and most of what can be seen is restricted to the steep, rocky outcrops. At about 700 m were small (approximately 5 ha) remnant forests dominated by the trees Adansonia za and Alluaudia spp., and the dramatic transition from humid montane forest to dry spiny forest was complete. Spiny Forest at 120 m Our final study site was located in parcel 2 of RNI d' Andohahela, some 50 km southwest of Pic HELME & RAKOTOMALAZA: BOTANICAL COMMUNITIES 21 Fig. 2-7. View of the spiny forest of parcel 2 of the RNI d'Andohahela, some 50 km southwest of Pic Trafona- omby. The spiny forest is characterized by the dominance of the endemic family Didiereaceae, particularly the lanky Alluaudia procera and A. ascendens in the central foreground, and a large number of Euphorbia species. (Photograph by N. Messmer.) Trafonaomby. The area is within the spiny forest, also known as the Southern Domain, and is char- acterized by the dominance of the endemic family Didieraceae and a large number of Euphorbia species (Perrier de la Bathie, 1921; Humbert, 1955). The vegetation is a thicket, at times even a forest, of drought-resistant species, many of them succulent (Fig. 2-7). The camp was located along a tributary of the Mananara River, about 7.5 km east of the village of Hazofotsy. Small granitic hills to the south provided a different, rockier hab- itat than the predominant plains with their sandy soils. These sands are often merely derived from the weathered granite, but there are also patches of deep, red, Pliocene sands. Small outcrops of hard, white calcrete are occasionally visible. Much of the immediate surrounding forest has been selectively logged. The gallery forest along the Mananara has been severely degraded, but there are still examples of deciduous forests, dom- inated by Tamarindus indica, in certain areas. Structurally the spiny forest is fairly homoge- neous, with similar tree densities recorded in a wide range of microhabitats, from sandy valleys to rocky hills. Probably the most characteristic taxa were the lanky Alluaudia procera and A. as- cendens, species that may form a 15-m-tall forest. These trees may be thought of as emergents be- cause they are about 5 m taller than most other species. Burseraceae was well represented by at least five species of Commiphora, while Euphor- bia formed a co-dominant in the canopy, with five species recorded in the area. Other common trees included Gyrocarpus americanus, Cedrelopsis grevei, Boscia longifol- ia, Operculicarya decaryi, Tetrapterocarpon geayi, Grewia spp., Strychnos spp., Fernandoa, Stereospermum, Diospyros sp., Hazunta modesta, Croton spp., and Obetia sp. Baobabs were repre- sented by Adansonia za, some of them reaching a diameter of 5 m, but they were not particularly common. Palms and bamboos were entirely ab- sent. There was an obvious shrub layer about 1-3 m high. This layer was dominated by Dichrostachys decaryana in the disturbed areas, where it was often quite dense. The understory was relatively open in the less disturbed areas and was com- 22 FIELDIANA: ZOOLOGY posed of species such as Rhigozum, various Acan- thaceae, Bauhinia sp., Koehnaria madagascarien- sis, Grewia spp., Aloe vaombe and A. divaricata, Croton spp., Hibiscus spp. and other Malvaceae, and various Fabaceae. Terrestrial grasses were un- common and seldom made up a significant por- tion of the ground cover. By far the most impor- tant component in the ground cover category was a small and extremely drought-tolerant fern — Se- laginella spp. (see Chapter 3). This plant covered large areas, turning green for a few days after a rainfall event but soon reverting to its usual gray, desiccated state. Lianas were present but were usually small and dominated by species of Asclepiadaceae, Apocy- naceae, and Cucurbitaceae. Epiphytes were pre- dictably rare in such a dry environment, the best represented being foliose lichens. A single species of epiphytic orchid (Angraecum) was found. No- ticeably absent were the branch parasites Viscum and Bakerella and the epiphytic Cactaceae Rhip- salis baccifera. The riverine area supported not only large Ta- marindus indica trees, but a number of Ficus spp. (including F. marmorata), large beds of Phrag- mites reed, shrubs such as Phyllanthus, and vari- ous weedy Cyperaceae and Juncaceae. A potential threat to the forest in certain areas were dense infestations of the highly invasive vine Cissus quadrangular e. This plant forms a smothering blanket over all existing vegetation and is capable of severely altering the habitat if not controlled. The rocky hills provided a different habitat, with shallow soils and a change in aspect. The vegetation most characteristic of these hills were the northeast-facing colonies of Pachypodium la- merei, a 3 m, extremely spiny pachycaul shrub. Trees such as Tetrapterocarpon geayi were re- placed by Melia azedarach, and Uncarina sp. be- came more common. The calcrete outcrops had a slightly impover- ished flora dominated by the spiny shrub Euphor- bia stenoclada, a species renowned for forming impenetrable thickets in such habitats further south and along the southern coast. Acknowledgments We are grateful for the extensive comments of Steve Goodman, William Burger, and Nathalie Messmer on an earlier version of this chapter. Literature Cited Davis, S. D., V. H. Heywood, and A. C. Hamilton. 1994. Centres of plant diversity — A guide and strat- egy for their conservation, vol. 1. WWF, IUCN Pub- lications, Cambridge, 354 pp. Du Puy, D. J., and J. Moat. 1996. A refined classifi- cation of the primary vegetation of Madagascar based on underlying geology: Using GIS to map its distri- bution and to assess its conservation status, pp. 205- 218. In Lourenco, W. R., ed., Biogeographie de Mad- agascar. Editions ORSTOM, Paris, 588 pp. Goodman, S. M., M. Pidgeon, A. F. A. Hawkins, and T. S. Schulenberg. 1997. The birds of southeastern Madagascar. Fieldiana: Zoology, n.s. 87: 1-132. Humbert, H. 1935. ^extinction des derniers vestiges de certains types de vegetation autochtone a Mada- gascar. Archives Museum National d'Histoire Natu- relle, Paris, serie 6, 12: 569-587. . 1941. Le massif de l'Andohahela et ses d6- pendances (Madagascar, Reserve Naturelle no. XI). Compte rendu sommaire des seances, Societe de Bio- geographie, 18: 31-37. . 1955. Les territoires phytogeographiques de Madagascar. Centre National de la Recherche Scien- tifique, Paris. Annee Biologique, serie 3, 31: 439- 448. Humbert, H., and G. Cours Darne. 1965. Notice de la carte. Madagascar. Travaux de la Section Scienti- fique et Technique de l'lnstitut Francais du Pondi- chery. Hors serie no. 6. KOECHLIN, J., J.-L. GUILLAUMET, AND P. MORAT. 1974. Flore et vegetation de Madagascar. J. Cramer Verlag, Vaduz. Lewis, B. A., P. B. Phillipson, M. Andrianarisata, G. Rahajasoa, P. J. Rakotomalaza, M. Randriambol- olona, and J. F McDonagh. 1996. A study of the botanical structure, composition, and diversity of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagascar, pp. 24-54. In Goodman, S. M., ed. A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagascar: With reference to eleva- tional variation. Fieldiana: Zoology, n.s. 85: 1-319. Paulian, R., C. Blanc, J.-L. Guillaumet, J.-M. Betsch, P. Griveaud, and A. Peyrieras. 1973. Etude des eco- systemes montagnards dans la region malgache. II. Les Chaines Anosyennes. Geomorphologie, climato- logie et groupements vegetaux. (Campagne RCP 225, 1971-1972). Bulletin du Museum National d'Histoire Naturelle, Paris. Ecologie generate, serie 3, 118: 1- 40. Perrier de la Bathie, H. 1921. La vegetation mal- gache. Annales de l'lnstitut Botanico-geologique Co- lonial de Marseille, serie 3, 9: 1-226. HELME & RAKOTOMALAZA: BOTANICAL COMMUNITIES 23 Chapter 3 Pteridophytes of the Eastern Slope of the Reserve Naturelle Integrate d'Andohahela, Madagascar: Distribution and Floristic Analysis France Rakotondrainibe1 Abstract The pteridophyte flora of the eastern slope of the Reserve Naturelle Integrale (RNI) d'Andohahela (parcel 1) comprises 207 species and varieties in 69 genera. One as yet unde- scribed species of Xiphopteris is currently considered endemic to the massif, and three other species are regional endemics. Data collected from 18 plots, covering a surface area of 13,600 m2 and containing 162 species, were subjected to an ascending hierarchical classification and a principal components analysis. Altitude accounted for 55.5% of the floristic heterogeneity among the plots. Based on floristic composition of the ferns, four altitudinal stages can be recognized between 400 m and 1956 m on the eastern slope. Each of these stages is characterized with regard to the physi- ognomy of its vegetation and various aspects of the pteridophytes present, including the range of growth forms, generic and specific composition, average species density, and the presence of species that are exclusive to a stage or exhibit a strong preference. Comparison of the pteridophyte flora of the RNI d'Andohahela with that of the Reserve Speciale d'Anjanaharibe-Sud, located in extreme northern Madagascar within the same biocli- matic zone but nearly 10° of latitude closer to the equator, shows a striking floristic similarity in the lower altitudinal strata, and equally strong differences above 1500 m. Resume La flore pteridologique du versant est de la Reserve Naturelle Integrale (RNI) d'Andohahela (parcelle 1) comprend 207 especes et varietes reparties en 69 genres. L'une d'elle, non encore decrite, appartenant au genre Xiphopteris, est considered a l'heure actuelle comme endemique du massif et 3 autres sont des endemiques regionales. Les donnees recueillies sur 18 parcelles totalisant une superficie de 13.600 m2 et concernant 162 especes, ont ete soumises a une classification ascendante hierarchique et a une analyse factorielle des correspondances. Le facteur altitudinal rend compte de 55,5% de l'heterogeneite floristique des parcelles. Sur la base de leur composition floristique en Pteridophytes, quatre etages altitudinaux ont ete reconnus sur le versant est, entre 400 et 1957 m. Pour chacun d'eux nous precisons la physionomie de la vegetation et les caracteristiques du peuplement des Pter- idophytes, a savoir: le spectre des types biologiques, la composition generique et specifique, la densite specifique moyenne, les especes caracteristiques exclusives et preferentielles. La comparaison de la flore pteridologique de la RNI d'Andohahela avec celle de la Reserve Speciale d'Anjanaharibe-Sud situe a l'extremite nord de Madagascar, dans la meme zone bio- 1 Laboratoire de Phytomorphologie, Ecole Pratique des Hautes Etudes, 16, rue de Buffon, 75005 Paris, France. RAKOTONDRAINIBE: PTERIDOPHYTES 25 climatique mais a une latitude inferieure de 10°, met en evidence des ressemblances floristiques frappantes dans les strates altitudinales inferieures et au contraire des differences tout aussi remarquables a partir de 1.150 m d' altitude. Introduction Parcel 1 of the Reserve Naturelle Integrate (RNI) d'Andohahela is located in extreme south- eastern Madagascar, in the Anosyenne Mountains (see Chapter 1 ). The information presented in this chapter was gathered in the northern part of the reserve, and specifically in the watershed of the Andranohela River as well as its headwaters on the slopes below the Trafonaomby Massif. This ancient granitic and gneissic massif culminates at 1956 m. Since Humbert (1941) published his overview of the Andohahela area, very little work has been published on the vascular plant flora of the area, despite the increasing number of botanists who have visited the reserve. Careful study of the her- barium material collected in 1928 and 1933-1934 by Humbert in the three massifs of the Anosy Range that are the closest areas to the study site (Pic Varavara, Pic Trafonaomby, and Pic Ando- hahela) resulted in a list of 28 species or varieties of pteridophytes, only five of which came from the eastern slope of Trafonaomby. Based on ex- isting collections and literature it is clear that the fern flora of the RNI d'Andohahela is among the most poorly known in Madagascar. The aims of this chapter are to (1) present an inventory of the pteridophytes occurring on the eastern slope of the RNI d'Andohahela, (2) de- scribe the patterns of distribution of the taxa, and (3) compare the results of the present study con- ducted in the southern part of Madagascar with a similar study conducted in the extreme northern part of the island, in the Reserve Speciale (RS) d'Anjanaharibe-Sud (Rakotondrainibe & Rahari- malala, 1998). Study Area: Climatic Stages and Vegetation The Andranohela Basin is situated in the humid climatic stages, characterized by average annual minimum temperatures between 10° and 13°C and a water deficit of less than 100 mm (Cornet, 1974). The eastern slope of the RNI d'Andohahela, between about 600 and 800 m, falls within the subhumid stage, characterized by a dry season that is attenuated by mist and fog, temperatures between 7° and 10°C, and a water deficit of less than 200 mm. The summit area, for which no climatological data are available, is placed by Cornet (1974) in his montane stage, with temperatures less than 5°C, and specifically within a humid subset with no dry season. The entire study area is covered by dense moist evergreen forest, the physiognomy of which changes progressively with increasing altitude. As elsewhere in Madagascar, the following vegeta- tion types can be recognized (Guillaumet & Koechlin, 1971; Paulian et al., 1973; Cornet & Guillaumet, 1976; F. Rakotondrainibe, pers. ob- serv.): (1) low-elevation dense evergreen forest, with a canopy reaching 20-30 m in height, trunks that are straight and free of epiphytic mosses, al- most no herbaceous ground cover, and large-di- ameter lianas that are abundant within the canopy and the understory; (2) montane dense evergreen forest with a canopy rarely exceeding 20 m in height, an herbaceous ground cover that is open to very dense, abundant mosses on the trunks, and many small-diameter lianas; and (3) dense scle- rophyllous montane forest near the summit, on ridges, and in areas with shallow soil, comprising trees 8-12 m tall, with twisted branches and trunks that are often declining or prostrate, with a thick carpet of mosses on the ground and the bas- es of the trees. Montane thicket vegetation, a low, very dense ericoid formation characteristic of the highest summits, does not occur on Pic Trafonaomby. The transition between low-elevation evergreen forest of the Eastern Domain and the mid-elevation for- est in the Central Domain usually occurs at around 800 m (Perrier de la Bathie, 1921), but in the extreme southern part of Madagascar the tran- sition is found at about 600 m (Humbert, 1935). Methodology The present study was conducted using the same methodology described earlier for inventory work performed in 1994 in the RS d'Anjanaharibe- 26 FIELDIANA: ZOOLOGY Sud (Rakotondrainibe & Raharimalala, 1998) and in 1993 in the RNI d'Andringitra (Rakotondrain- ibe & Raharimalala, 1996). Holistic Survey The floristic inventory was carried out within a 2 km radius around five camps, i.e., between 420 and 550 m (camp 1), 750 and 850 m (camp 2), 1150 and 1280 m (camp 3), 1470 and 1570 m (camp 4), and 1850 and 1956 m (camp 5). The data were supplemented by observations and col- lections made along the main trail connecting the camps that runs from the village of Enosiary to Pic Trafonaomby, following the ridge lines above the left bank of the Andranohela River (see Chap- ter 1). All voucher specimens were deposited in the herbarium of the Departement des Recherches Fo- restieres et Piscicoles, Centre National de la Re- cherche Appliquee au Developpement (DRFP/ FOFIFA), in Antananarivo (TEF), and the Labor- atoire de Phanerogamie, Museum National d'Histoire Naturelle, in Paris (P). Duplicates of most of the species collected will be sent to the Missouri Botanical Garden, St. Louis (MO) and the Royal Botanic Gardens, Kew (K). In most cases, generic delimitations have fol- lowed those proposed by Kramer and Green ( 1 990), although for certain genera whose circum- scription is controversial and which are in need of revision, particularly for the Malagasy flora, more classic concepts have been used: Grammi- tidaceae follows Pichi-Sermolli (1977) and The- lypteridaceae uses the system of Holttum (1974). Megalastrum (Dryopteridaceae) is treated as a separate genus, as suggested by Holttum (1986). The determination of species and varieties was made using Malagasy and African flora treatments (Christensen, 1932; Tardieu-Blot, 1951-1971; Schelpe, 1970; Schelpe & Anthony, 1986) and monographs of Lastreopsis genus (Tindale, 1965), Lindsaeaceae family (Kramer, 1972), and Thelyp- teridaceae family (Holttum, 1974). All of the data presented here concern the pte- ridophytes that occur in moist forest within parcel 1 of the RNI d' Andohahela. Information on a few ferns observed in the spiny forest of parcel 2 is given in Appendix 3-1. Ecological Sampling In order to characterize the pteridophyte asso- ciations in the different biotopes within the study area and to assess the ecological factors) respon- sible for patterns of distribution among the taxa represented, data were collected from plots. As far as possible one plot was established in each bio- tope such that when taken together, they afforded a representative sample of the study area. Thus, 18 ecological plots were established, 14 on slopes (PI -PI 4) and four along watercourses (R1-R4). Tables 3-2 and 3-3 summarize the physical, to- pographic, and biological features of the plots. The minima] area needed for a representative sample of pteridophytes in humid forest in Mad- agascar is between 500 and 1,000 m2, depending on the diversity of the biotope (Rakotondrainibe & Raharimalala, 1998). The area of each plot in this study is 800 m2, except for plot PI 1 (400 m2), established on a narrow ridge. Each plot consisted of eight subunits of 100 m2, the arrangement of which depended on the topography. The number of species present in each plot as a percentage of the total fern flora identified from the general in- ventory in each area provides an indication of how representative they are, with values ranging from 67.9% to 84.0% and averaging 78.3% (which corresponds to 162 taxa present in the plots out of a total of 207 identified from the study area). The lowest value (67.9%) was obtained near the summit, where time constraints made it impossible to sample the vegetation in two areas, a nonforest swamp and a large stone slab. The swamp and stone slab clearly had a low number of highly specialized species that could thus be described adequately without having to use quan- titative methods. Considering that most tropical formations have a rather large number of infre- quently encountered species, the average value of 78.3% is considered sufficient to indicate that the floristic composition of the study plots adequately represented that of the surrounding vegetation. The methods used for the plot analyses (choice of site, size of plots, and procedures for data col- lection) have been described elsewhere (Rakoton- drainibe & Raharimalala, 1998). To facilitate in- terpretation of the results presented here, however, we briefly discuss types of field observations and how they were recorded. The following were re- corded for each plot: A. A complete list of species encountered. B. The number of individuals or colonies of each species, coded by abundance class, according to the following categories: 1 . 1 individual or colony 2. 2-4 individuals or colonies RAKOTONDRAINIBE: PTERIDOPHYTES 27 Table 3-1. Pteridophytes of the eastern slope of the RNI d'Andohahela (parcel type, and altitudinal distribution. I): Floristic composition, habit 420- 810- 1280- 1510- 550 1150 1500 1957 Habit No. Taxa* m m m m type 1 Adiantum madagascariense H. Rodend. van pro- longatum (Bonap.) Tardieu 2 Adiantum phanerophlebium (Baker) C. Chr. 3 Amauropelta bergiana (Schltdl.) Holttum 4 Angiopteris madagascariensis de Vriese 5 Antrophyum bivittatum C. Chr. 6 Antrophyum boryanum (Willd.) Kaulf. 7 Antrophyum malgassicum C. Chr. 8 Arthropteris monocarpa (H. L. Cordem.) C. Chr. 9 Arthropteris orientalis (J. F. Gmel.) Posth. var. sub- biaurita (Hook.) C. Chr. 10 Asplenium aethiopicum (Burm.) Bech. 1 1 Asplenium auritum Sw. 12 Asplenium bipartitum Bory 13 Asplenium blastophorum Hieron. 14 Asplenium cuneatum Lam. 15 Asplenium dregeanum Kunze 16 Asplenium erectum Bory ex Willd. var. e rectum 17 Asplenium erectum Bory ex Willd. var. zeyheri (Pappe & Rawson) Alston & Schelpe 18 Asplenium friesiorum C. Chr. 19 Asplenium herpetopteris Baker var. herpetopteris 20 Asplenium herpetopteris Baker var. massoulae (Bonap.) Tardieu 21 Asplenium inaequilaterale Willd. 22 Asplenium lividum Mett. ex Kuhn 23 Asplenium mannii Hook. 24 Asplenium nidus L. 25 Asplenium normale D. Don 26 Asplenium pellucidum Lam. 27 Asplenium petiolulatum Mett. 28 Asplenium poolii Baker var. poolii 29 Asplenium poolii Baker var. linearipinnatum (Bon- ap.) C. Chr. 30 Asplenium prionites Kunze 3 1 Asplenium protensum Schrad. 32 Asplenium rutifolium (P. J. Bergius) Kunze 33 Asplenium sandersonii Hook. 34 Asplenium theciferum Kunth 35 Asplenium thunbergii Kunze 36 Asplenium unilaterale Lam. 37 Asplenium variabile Hook. var. paucijugum (Bal- lard) Alston 38 Asplenium virchowii Kuhn 39 Asplenium viviparioides Kuhn 40 Athyrium scandicinum (Willd.) C. Presl 41 Belvisia spicata Mirbel 42 Blechnum attenuatum (Sw.) Mett. 43 Blechnum attenuatum (Sw.) Mett. var. giganteum (Kaulf.) Bonap. 44 Blechnum bakeri (Baker) C. Chr. 45 Blechnum ivohibense C. Chr. 46 Blechnum madagascariense Tardieu 47 Blechnum punctulatum Sw. 48 Blechnum simillimum (Baker) Diels 49 Blechnum tabulare (Thunb.) Kuhn 50 Blotiella pubescens (Kaulf.) Tryon 51 Christella dentata (Forssk.) Holttum 52 Christella distans (Hook.) Holttum + — — — T + - — - T - - + + T/r + - - - T - - - + E - - + - E/r + + - - E/r + + + + E/r + + - - T/E - + + + E/r - + - - E/r + - - - E/r + + - - T/E/r + + - - E/r - + - - R - + + + T/r - + + - T/r - + + + T/E/r + + + - E/r - - + - E/r + + — — T/r - + - - E - + + + E + + - - E/r - + + + T/r + + - - E/r - + + + E - + + + T/E/r - + - - T/r — + - — E - - + + E/r - + + + E + + + + T/E/r - + - + E/r + + + T/E/r + - - - T/r + - - - T/E/r — + - - R - - - + T - - + - T - + + - E + + + - T/E/r - - + - T — — + T - + + - T - - + - T - - - + T - + + - T/E - - - + T + + + - T + - - - T + + - - T 28 FIELDIANA: ZOOLOGY Table 3-1. Continued. 420- 810- 1280- 1510- 550 150 1500 1957 Habit No. Taxa* m m m no type 53 Christella multifrons (C. Chr.) Holttum 54 Cochlidium serrulatum (Sw.) L. E. Bishop 55 Coniogramme madagascariensis C. Chr. 56 Ctenitis cirrhosa (Schumach.) Ching 57 Ctenitis crinita (Poir.) Tardieu var.? (FR 2869. 2884) 58 Ctenitis madagascariensis Tardieu 59 Ctenopteris devoluta (Baker) Tardieu 60 Ctenopteris elastica (Bory) Copel 61 Ctenopteris flabelliformis (Poir.) I. Sm. 62 Ctenopteris villosissima (Hook.) Harley 63 Ctenopteris zenkeri (Hieron.) Tardieu 64 Cyathea aff. boivinii Mett. (FR 2880, 2881, 3016) 65 Cyathea aff. dregei Kunze (FR 3136, 3234, 3122) 66 Cyathea andohahelensis (Tardieu) (= Alsophila andohahelensis Tardieu) 67 Cyathea aff. bellisquamata Bonap. (FR 3059) 68 Cyathea borbonica Desv. var. laevigata Bonap. 69 Cyathea borbonica Desv. var. 1 (FR 3150) 70 Cyathea borbonica Desv. var. 2 (FR 2968) 71 Cyathea bullata (Baker) Rakotondr. 72 Cyathea costularis Bonap. 73 Cyathea decrescens Mett. 74 Cyathea melleri (Baker) Domin 75 Cyathea pilosula Tardieu 76 Cyathea sp. 2 (groupe C. borbonica) (FR 3165) 77 Cyathea tsilotsilensis Tardieu 78 Cyclosorus interruptus (Willd.) Kato 79 Cyrtomium caryotideum (Wal. ex Hook. & Grev.) C. Presl var. micropterum (Kunze) C. Chr. 80 Davallia chaerophylloides (Poir.) Steud. 81 Deparia parvisora (C. Chr.) Kato 82 Dicranopteris linearis (Burm.) Underw. 83 Didymochlaena truncatula (Sw.) J. Sm. 84 Diplazium aff. zakamenense (Tardieu) Rakotondr. (FR 3081) 85 Doryopteris kitchingii (Baker) Bonap. 86 Drynaria willdenowii (Bory) Moore 87 Dryopteris mangindranensis Tardieu 88 Dryopteris manniana (Hook.) Tardieu 89 Dryopteris remotipinnula Bonap. 90 Elaphoglossum acrostichoides (Hook. & Grev.) Schelpe 91 Elaphoglossum aff. conforme (Sw.) Schott. (FR 2909,3107,3175) 92 Elaphoglossum aff. sieberi (FR 2962, 2991, 3025, 3188) 93 Elaphoglossum aff. stipitatum (FR 2918) 94 Elaphoglossum aff. subsessile (Baker) C. Chr. (FR 3022, 3066) 95 Elaphoglossum angulatum (Blume) Moore 96 Elaphoglossum aubertii (Desv.) Moore 97 Elaphoglossum coursii Tardieu 98 Elaphoglossum decaryanum Tardieu 99 Elaphoglossum deckenii (Kuhn) C. Chr. var. rufidu- lum (Willd.) Tardieu 100 Elaphoglossum forsthii-majoris H. Christ 101 Elaphoglossum humbertii C. Chr. 102 Elaphoglossum hybridum (Bory) Brack. + - - - T - + - - R - + + - T + - - - T + - - - T - + - - T - + + + E + + - - E - - - + T/E - + + + E + + - - E + + - - T - - + - T - - + + T — - + - T + + - - T - - + - T - + - - T - + + - T + + - - T - + + - T - + + + T - + - - T - - + - T + + - - T + - - - T - - - + T + - - - E - + + - T - + - + T - + - - T - + + - T _ - - + R + - - - E + + + - T/r - - + - T - + + + T - - + + E + - - - T - + + - E + - - - R - + + + E _ _ — + E - - + + T/E - - - + T - - + + E - - + + E _ + — - E - + + - E + + + - E/r RAKOTONDRAINIBE: PTERIDOPHYTES 29 Table 3-1. Continued. No. Taxa* 420- 550 m 810- 1150 m 1280- 1500 m 1510- 1957 Habit m type 103 Elaphoglossum leucolepis (Baker) Krajina var.? (FR 2990) 104 Elaphoglossum petiolatum ssp. salicifolium (Willd. ex Kaulf.) Schelpe 105 Elaphoglossum poolii (Baker) H. Christ 106 Elaphoglossum pseudovillosum Bonap. 107 Elaphoglossum sp. 9 (FR 3127) 108 Elaphoglossum spathulatum (Bory) Moore 109 Elaphoglossum subsessile (Willd.) C. Chr. 1 10 Grammitis barbatula (Baker) Copel. 1 1 1 Grammitis cryptophlebia (Baker) Copel. 1 12 Grammitis gilpinae (Baker) Tardieu 1 13 Grammitis holophlebia (Baker) Copel. 1 14 Grammitis microglossa (C. Chr.) Ching 1 15 Grammitis sp. (FR 3197, 3230) 1 16 Grammitis synsora (Baker) Copel. 1 17 Huperzia cavifolia (C. Chr.) Tardieu 1 1 8 Huperzia gagnepainiana (Herter) Tardieu 1 19 Huperzia humbertii-henrici (Herter) Tardieu 120 Huperzia megastachya (Baker) Tardieu 121 Huperzia obtusifolia (Sw.) Rothm. 122 Huperzia ophioglosoides (Lam.) Rothm. 123 Huperzia pecten (Baker) Tardieu 124 Huperzia squarrosa (G. Forst) Trevis. 125 Huperzia verticillata (L. f.) Trevis. 126 Hymenophyllum aff. fumarioides Willd. (FR 3032, 3084, 3189) 127 Hymenophyllum hirsutum (L.) Swartz 1 28 Hymenophyllum perrieri Tardieu 129 Hymenophyllum polyanthos (Sw.) Sw. 130 Hymenophyllum sibthorpioides (Bory ex Willd.) Mett. ex Kuhn 131 Hymenophyllum tunbrigense (L.) Sm. 132 Hymenophyllum veronicoides C. Chr. 133 Hymenophyllum viguieri Tardieu 134 Hypolepis sparsisora (Schrad.) Kuhn 135 Lastreopsis pseudoperrieriana (Tardieu) Tardieu 136 Lepisorus excavatus (Bory ex Willd.) Moore 137 Lepisorus schraderi (Mett.) Ching 138 Lindsaea goudotiana (Kunze) Mett. ex Kuhn 139 Lindsaea madagascariensis Baker 140 Lomariopsis aff. crassifolia (FR 2859, 3000) 141 Lomariopsis aff. pollicina Willem ex Kuhn (FR 2860) 142 Loxogramme humblotii C. Chr. 143 Loxogramme lanceolata (Sw.) C. Presl 144 Lycopodiella caroliniana (L.) Pic. Serm. 145 Lycopodiella cernua (L.) Pic. Serm. 146 Lycopodium clavatum L. 147 Lygodium lanceolatum Desv. 148 Marattia fraxinea Sm. ex J. F. Gmel. 149 Megalastrum aff. magnum (Baker) Holttum (FR 2874, 2934) 150 Megalastrum lanuginosum (Kaulf.) Holttum 1 5 1 Microlepia madagascariensis (Kunze) C. Presl 152 Microlepia speluncae (L.) Moore 1 53 Microsorum pappei (Mett.) Tardieu 154 Microsorum punctatum (L.) Copel. - - - + E - + - - E - - + + E + + - - R - + + + T - - - + E - - + + R - - + - R - - + + E - - + - R - - - + E - + - - E - + - - E - - + - E - - + + E - + - - E - + + + E - + + + E - + + + E - - - + E - + + + E - + + + E/r + + + + E/r - + + + E - + + + E + + + + E - — + — E - - - + E + - - - R - + - - T + - - - T - + + + E - - - + E + + - - E - + + - T + + + - R + + - - E/r + — — — R + + + + E/r - - - + T - + - - T - - - + T + - - - L + + + - T + + - - T + — + - T + + + - T + - - - T - + - - R + + - - E/r 30 FIELDIANA: ZOOLOGY Table 3-1. Continued. No. Taxa* 420- 810- 550 1150 111 Ml 1280- 1500 m 1510- 1957 Habit m type 155 Mohria caffrorum (L.) Desv. = M. marginata J. P. Roux) 156 Nephrolepis biserrata (Sw.) Schott 157 Nephrolepis tuberosa (Bory) C. Presl 158 Nothoperanema squamisetum (Hook.) Ching 159 Oleandra distenta Kunze 160 Osmunda regalis L. 161 Pellaea angulosa (Bory) Baker 162 Pellaea boivinii Hook. 163 Pellaea viridis (Forssk.) Prantl var. glauca Sim. 164 Phymatosorus scolopendria (Burm.) Pic. Serm. 165 Pityrogramma argentea (Willd.) Domin 166 Pityrogramma calomelanos (L.) Link 167 Pleopeltis macrocarpa (Bory ex Willd.) Kaulf. 168 Pneumatopteris remotipinna (Bonap.) Holttum 169 Pneumatopteris subpennigera (C. Chr.) Holttum 170 Polystichum coursii Tardieu 171 Pseudocyclosorus pulcher (Bory ex Willd.) Holt- tum 172 Pteridium aquilinum (L.) Kuhn 173 Pteris catoptera Kunze 174 Pteris elongatiloba Bonap var. remotivenia Bonap. 175 Pteris griseoviridis C. Chr. 1 76 Pteris pseudolonchitis Bory 177 Rumohra adiantiformis (G. Forst.) Ching 178 Rumohra aff. lokohoensis Tardieu (FR 3033) 179 Rumohra capuronii Tardieu 180 Saccoloma henriettae (Baker) C. Chr. 181 Schizaea dichotoma L. 182 Selaginella marinii Stefanovic & Rakotondr. 183 Selaginella pectinata (Willd.) Spring 184 Sphenomeris chinensis (L.) Maxon 185 Stenochlaena tenuifolia (Desv.) Moore 186 Sticherus flagellaris (Bory) St. John 187 Tectaria madagascarica Tardieu 188 Thelypteris connfluens (Thunb.) C. V. Morton 189 Trichomanes bipunctatum Poir. 190 Trichomanes borbonicum Bosch 191 Trichomanes cupressoides Desv. 192 Trichomanes cuspidatum Willd. 193 Trichomanes digitatum Sw. 194 Trichomanes lenormandii Bosch 195 Trichomanes longilabiatum Bonap. 196 Trichomanes mannii Hook. 197 Trichomanes meifolium Bory ex Willd. 198 Trichomanes melanotrichum Schltdl. 199 Trichomanes montanum Hook. var. montanum 200 Trichomanes montanum Hook, van? (FR 2907, 2931) 201 Trichomanes rigidum Sw. 202 Trichomanes rotundifolium Bonap. 203 Trichomanes speciosum Willd. 204 Vittaria humblotii Hieron. 205 Vittaria isoetifolia Bory ex Fee 206 Vittaria sp. (FR 3035) 207 Xiphopteris sp. nov. (FR 3243) + + + - - T/E/r - + - - T/E - - - + T + + + - E - + - + T + + + + T - - - + R - + + R + + - - T/E/r - - - + T + - - - R - + + + E/r - + - - T - + + - T + + + + T - + - - T - - — + T - - + - T - + - - T - - + - T + + + + T - + + + T/E + + - - T/E - - + - E - - + - T + + + - T - + - - R + - - - T - - - + T + - - - T - + + - T + + - - T - + - - T + + - - E/r + + + - E/r + - + - T/r - - + - R - - + - E/r + + + - E + - - - R + + + + E/r - - + - E + + + + E/r + + -f + E/r + + - - R + + + — T + + + + E/r + - - - E/r + + + + E/r - + + + E + + - - E - - - + E * Species endemic to Madagascar are indicated in boldface type. + = present; - = absent; T = terrestrial; E = epiphyte; R = strictly epilithic; r lianescent. Numerals preceded by the initials FR are the author's collection numbers. occasionally epilithic; L t>V RAKOTONDRAINIBE: PTERIDOPHYTES BIOLOGY U: 101 BURRILL HALL a r innfi 31 Table 3-2. Characteristics of the study plots on the eastern slopes of the RNI d'Andohahela. Plot P2 PI P3 P5 P4 P6 Area (m2) 800 800 800 800 800 800 Altitude (m) 430 450 820 820 840 1100 Topographic po- middle plateau middle middle plateau ravine sition slope slope slope Exposure E SE SW E W Slope (°) 20-30 0 15 15-20 2-7 30-45 Canopy 12-15 20-25 14-16 15-20 16-20 10-16 height (m) Woody plant 35 35-45 40 35 30-35 20-30 cover (%) Herbaceous 7-10 7-12 5-7 15-25 10-35 15-30 plant cover (%) Litter 2 2 5-7 0-2 3 2-4 thickness (cm) Humus 2-5 10-20 15-17 0-10 30 10-30 thickness (cm) Soil yellow ochre yellow ochre yellow ochre rock (granite) rock (granite) brown ochre character- clay clay- •sand clay clay istics 3. 5-9 individuals or colonies 4. 10-19 individuals or colonies 5. 20-49 individuals or colonies 6. >50 individuals or colonies Note: For small epiphytic species that covered a large, continuous surface on a trunk (species of Hymenophyllaceae and Grammitidaceae), each area of 400 cm2 (20 X 20 cm) was arbitrarily con- sidered to represent a single colony. For larger terrestrial or epiphytic gemmiferous species (e.g., Asplenium sandersonii2 and A. poolii), individuals or colonies could be scored easily because all plants connected by a rooted rachis form a colony. 2 Authors of the taxa cited in the text are indicated in Table 3-1. C. The growth form of the species, i.e., whether terrestrial, epilithic, epiphytic, or lianescent. Because many epiphytic pteridophytes can also grow on rocks, as can a more limited number of otherwise terrestrial species, the following growth form categories are recog- nized: T/r, always terrestrial or terrestrial and more rarely epilithic E/r, always epiphytic or epiphitic and more rarely epilithic R, strictly epilithic L, lianescent T/E/r, terrestrial, epiphytic, or more rarely epilithic. Precise definitions of these growth form types Table 3-3. Characteristics of RNI d'Andohahela study plots along watercourses. Width of Area Altitude stream- Nature of (m2) (m) bed (m) streambed Flow Plot Rl 800 (200 X 4) 420-480 1-3 large granitic rocks + stones very gentle Plot R2 800 (200 X 4) 500-530 3-4 large granitic rocks + stones gentle Plot R3 800 (200 X 4) 830-840 2-4 large granitic rocks very gentle Plot R4 800 (200 X 4) 1450-1460 1-4 large granitic rocks + stones gentle 32 FIELDIANA: ZOOLOGY Tablk 3-2. Extended. Plot P7 P8 P9 P10 Pll P14 P12 P13 800 800 800 800 400 800 800 800 1150 1280 1500 1510 1530 1760 1830 1900 ridge middle lower ridge ridge middle middle middle slope slope slope slope slope S W E NE S-SW N W-NW 10-15 35-45 20-30 5 0 5-10 5-10 30-35 12-18 12-16 10-16 12-14 1-12 14-16 8-10 10-14 35-45 35 20-25 40-60 30-40 25-35 20 25-30 5-10 15-30 35-45 3 45-50 7-10 10-15 + moss 5-10 4-8 5-15 7-10 10-15 15 2 0 + moss 5 10-20 20-60 30-50 20-35 10 40 30-60 30-40 brown ochre yellow ochre brown ochre brown ochre yellow ochre yellow ochre stony-sand yellow ochre clay-sand clay-stony clay-stony clay-sand clay clay clay-sand are presented elsewhere (Rakotondrainibe & Raharimalala, 1998). Data Analysis The data collected from the plots were treated using two classic and complementary types of multivariate analysis (STATITCF, Version 4): cor- respondence analysis (CA) (Benzecri & Benzecri, 1984; Legendre & Legendre, 1984a,b) and hier- archical ascending classification (HAC) (Jambu, 1978; Legendre & Legendre, 1984b). The data matrix used for the analyses listed in rows all 162 species recorded in the plot studies against columns representing each of the 18 plots sampled (P1-P14 and R1-R4). Each cell of the matrix contained the corresponding abundance code (1-6) as defined above. For the analyses all of the elements in each column were active. The two numerical approaches, CA and HAC, provide an estimation of the floristic similarity be- tween plots by comparing the characteristics of each plot in terms of species composition and the importance of each species as a component in each plot. These techniques make it possible to visualize and interpret the structures that occur within the body of the data by \: analysis of the relationships between the different elements (plots or species) of the matrix. CA yields a represen- tation, in multidimensional space, of the plot and/ or species points, and reveals gradients among the data. Correlation between clusters of species and clusters of plots is used to determine characteristic species for each biotope type. HAC generates a tree or dendrogram comprising a hierarchical and progressive grouping of plot and/or species units in sets of increasing size. The criterion used for grouping units into a class is "the average of the weighted distances." Comparison with the Flora of the RS d'Anjanaharibe-Sud The pteridophyte flora of the RS d'Anjanaharibe-Sud is compared with that of the RNI d'Andohahela to evaluate the genera and species richness, the number of regional and is- land-wide endemic pteridophyte species, and the characteristic species in each altitudinal stage. S0rensen's similarity coefficient (S0rensen, 1948) is also used to compare the degree of similarity of the two floras. RAKOTONDRAINIBE: PTERIDOPHYTES 33 Fig. 3-1. Percentages of the pteridophyte flora (207 species listed in Table 3-1) by habit type in parcel 1 of the RNI d'Andohahela. See Table 3-1 for explanation of abbreviations. Results Overall Floristic Analysis Table 3-1 lists all the taxa observed on the east- ern slope of the RNI d'Andohahela between 400 and 1956 m, including those recorded from the 18 plots; their growth form (terrestrial, epiphytic, or epilithic); and their altitudinal distribution. The delimitation of altitudinal zones is explained be- low. A total of 207 species or varieties, representing 69 genera, were identified. The most speciose genera were Asplenium (27 spp. and three var.), Elaphoglossum (20 spp.), Trichomanes (14 spp. and one var.), Cyathea (12 spp. and two van), Hu- perzia (nine spp.), Hymenophyllum (eight spp.), Blechnum (seven spp. and one var.), and Gram- mitis (seven spp.). A large number of genera (52 in all) were represented by only one or two spe- cies. Growth Forms The histogram of growth form frequencies shown in Figure 3- 1 was derived from the data in Table 3-1. A majority of the species and varieties (87, representing 42.2% of the total) were always terrestrial or terrestrial and more rarely epilithic (T/r), 82 taxa (39.8%) were always epiphytic or epiphitic and more rarely epilithic (E/r), whereas only 15 (7.3%) were terrestrial, epiphytic, or more rarely epilithic (T/E/r). Twenty-one taxa (10.2%) were observed only on rocks (R), and a single species (Lygodium lanceolatum) was lianescent (L). Endemism Sixty-seven species recorded in parcel 1 are en- demic to Madagascar, representing 32.4% of the total pteridophyte flora of the reserve. These taxa are indicated in boldface type in Table 3-1. Most of the pteridophyte species found on the eastern slope of the RNI d'Andohahela are largely distributed throughout the Eastern and Central do- mains (sensu Humbert, 1955) of Madagascar, al- though variations occur in their abundance, fre- quency, and the biotopes they occupy. Some of them (named below) show more restricted distri- butions, and these are more informative for bio- geographical analysis. Our knowledge of the pte- ridophytes occurring in high mountain areas of Madagascar has progressed substantially in recent years as many new collections have become avail- able, and it is now possible to prepare revised species distribution maps that are probably accu- rate in most cases. A description of a new Xiphopteris species (Rakotondrainibe 3243), collected as part of the present study, will be published shortly; it is the only taxon that can be considered endemic to the Trafonaomby Massif. This new species was found on two occasions, once as a very inconspicuous 34 FIELDIANA: ZOOLOGY $ Blechnum punctu latum % Huperzja humbertii-henrici /^JV (D S) 0 A — 15°S M /^f ~- J/ C 4V Mfc ^f - 20 r~ v'*:,(? fa M Ty-A yw #/ 7 — 25° X^y *y--J •* '(tif \Tp/t(fd) 45° I 0 100 200 km 50°E 1 Fig. 3-2. Distribution of several pteridophyte species from parcel 1 of the RNI d'Andohahela with ranges limited to the South-Central and Mid-Central Domains. plant occurring at 1300 m and a second time at 1780 m, forming a large epiphytic colony on a single tree trunk in a very wet forest located at the base of a cliff. An Elaphoglossum species (Rakotondrainibe 3127) that was found only as a sterile specimen may be new to science and en- demic to the massif, but it would be premature to describe it until fertile material can be collected. Three species endemic to Madagascar are re- stricted in distribution to the southeastern region of the country: (1) Cyathea tsilotsilensis was col- lected for the first time by Humbert at the Col de Tsilotsilo, 6 km northeast of Elakelaka, and then found again recently on the Andohahela Massif by van der Werff. Along with the material gath- ered for the present study, the three known col- lections were made between 400 and 1300 m. (2) Tectaria madagascarica was known only from the type collection made by Humbert at 1200 m in a valley between the upper Mananara and the upper Manampanihy. This species can be easily confused with the more widely distributed T. magnified, from which it differs by having a densely hirsute lower surface of the frond and free pinnules of the lower pinnae (Tardieu-Blot, 1958, p. 349). (3) Ctenitis madagascariensis has been observed in the mountains south of Tanandava, along a tributary of the Manampanihy, and on Ka- lambatritra, south of Betroka. Many species present in the RNI d'Andohahela do not reach much further north than the latitude of Antananarivo (Fig. 3-2) and are characteristic of the Mid-center and South-center subdomains of the Central Domain of Humbert and Cours Darne RAKOTONDRAINIBE: PTERIDOPHYTES 35 4t Blechnum humbertii • Hymenophyllum capillare K Ctenopteris humbertii I Blechnum australe 15°S A(DS) ^ Asplenium variabile var .paucijugum • Huperzia gagnepainiana Q Grammitis microglossa ■ Rumohra capuronii ▲ Selaginella marinii -15°S A(DS) -25 45° _l_ 50°E J — 25 Fig. 3-3. Distribution of several pteridophyte species from parcel 1 of the RNI d'Andohahela with distributions absent from the South-Center Domain (left) and with a disjunct, bipolar pattern (right). (1965). Examples include Cyathea andohahelen- sis and Asplenium virchowii, whose northern dis- tributional limit is the Andringitra Massif. Simi- larly, Blechnum punctulatum, Huperzia humbertii- henrici, Asplenium viviparioides, Polystichum coursii, and Asplenium lividum do not reach fur- ther north than the RS d'Ambohitantely. The floristic originality of the southwestern re- gion can also be seen in the absence of certain taxa that are more or less widely distributed in other parts of the country. For example, several species, including Blechnum humbertii, Ctenop- teris humbertii, Hymenophyllum capillare, and Blechnum australe hardly extend south beyond the Tropic of Capricorn (Fig. 3-3, left). Affinities with Other Mountainous Areas in Madagascar Several high-altitude species that occur in the RNI d'Andohahela (on the Trafonaomby Massif) are characteristic of the flora present on the high- est mountain areas in Madagascar and are com- mon on Tsaratanana, Marojejy, Anjanaharibe-Sud, Ankaratra, and Andringitra. These species include Ctenopteris flabelliformis, Hymenophyllum veron- icoides, and Huperzia ophioglossoides, all of which typically occur on ridges; Lycopodium cla- vatum, usually found at forest edges or in the un- derstory of low, open forests at high elevations; Dryopteris kitchingii, a heliophilous species that 36 FIELDIANA: ZOOLOGY B D pi = P2 . R1 . _R2 . "R3 ■ P5 ■ P3 . P6 ■ P7 ■ _P4 ~P14 P12 P13 P10 _P11 ~P9 R4 P8 20 24 29 I 33 34 35 22 21 23 28 31 25 27 30 19 26 32 i Fig. 3-4. Hierarchical ascending classification (HAC) tree using a matrix of 18 plots (P1-P14, R1-R4) and 162 species of pteridophytes in parcel 1 of the RNI d'Andohahela. The criterion used for grouping units into a class is the average of the weighed distances. usually grows in rocky crevices; and Trichomanes meifolium and Elaphoglossum aubertii, both of which occur in more shady areas such as moist depressions surrounded by mosses. Gleichenia polypodioides, which is rather common on ridges above 1700-1800 m in several massifs, was col- lected by Humbert (13505) in 1933 near Pic Tra- fonaomby but was not seen in 1995. Finally, current information indicates that seven fern taxa have bipolar north-south disjunct distri- butions (Fig. 3-3, right). Asplenium variabile var. paucijugum and Trichomanes montanum var. ? (Rakotondrainibe 2907) are widespread and com- mon species only at low altitude, between 10 and 600 m in the southern and northern part of Mad- agascar, but they are absent from the central part, a fact that could be explained by the almost total loss of low-elevation forests on the central part of the island. On the other hand, it is more difficult to explain why species (albeit less common) such as Blechnum madagascariense, Huperzia gagne- painiana, Grammitis microglossa, Rumohra ca- puronii, and Selaginella marinii, which generally occur between 810 and 1700 m, are found only in the southern and northern parts of the island, whereas at this elevation range the dense ever- green forest still covers some areas of the Central Highlands. Distribution of Taxa Presence of an Altitudinal Gradient The tree obtained by HAC from the data matrix of species against study plots is shown in Figure 3-4. By sectioning the tree between nodes 32 and 33, four groups of plots are defined on the basis of the floristic composition of their pteridophytes: group A (comprising PI, P2, Rl, and R2), group B (P3-P7 and R3), group C (P8, P9, and R4), and group D (P10-P14). Sectioning the tree at the node immediately below (between nodes 31 and 32) separates P8 from group C. The positions of groups A through D are shown on axes I— II (Fig. 3-5), I— III (Fig. 3-6), and I-IV (Fig. 3-7). These four axes respectively account for 21.5%, 15.6%, 9.7%, and 8.7% of the inertia of the data points, or 55.5% of the total inertia. For clarity, only structural species are shown, i.e., those contributing to at least 1.8% of the inertia RAKOTONDRAINIBE: PTERIDOPHYTES 37 II T=15.6% 187 154 18713 122" 172 97 Fig. 3-5. Correspondence analysis (CA) of the matrix of 18 plots and 162 pteridophyte species in parcel 1 of the RNI d'Andohahela: projection on axes I and II of all of the plots (P1-P14, R1-R4) and the species that contribute to at least 1.8% of the inertia of one or both of the factorial axes. Species numbers are those given in Table 3-1. of one or both of the factorial axes shown (this cutoff value is explained below). The numerical code for species used in the figures corresponds to those in Table 3-1. In Figure 3-5, projected on axes I and II, the plots are distributed in a crescent shape (Guttman effect), which indicates the presence of a strong gradient among the data. Axes I and II express the same altitudinal gradient: axis I separates group A, i.e., the plots at low altitude (420-550 m) from group D, i.e., the plots at high altitude (1510-1956 m). Axis II separates groups A and D, located in the negative portion of the scale, from the mid-elevation groups B and C (810- 1500 m) in the positive part of the scale. In Figure 3-6, axis III separates the two mid- elevation groups, with group B (810-1150 m) in the positive portion and group C (1280-1500 m) in the negative part. Plot P8, which is included in group C (Fig. 3-4), in fact occupies an interme- diate position between groups B and C. In Figure 3-7, the cloud of points comprising group D is stretched along axis IV and is divided into two subunits corresponding to two altitudinal and topographic subgroups. Plots P10 and Pll form one subgroup, located on ridges at 1510- 1530 m, whereas PI 2 and PI 3, situated on mid- slopes at 1830-1900 m, form a second subgroup. Plot P14, also located on a mid-slope but at 1760 m, is intermediate, which clearly shows that axis IV expresses an altitudinal factor at least in part. The same holds true for the cloud of points mak- 38 FIELDIANA: ZOOLOGY Ill r=9.7% T=21.5% Fig. 3-6. Correspondence analysis of the matrix of 18 plots and 162 pteridophyte species in parcel 1 of the RNI d'Andohahela: projection on axes I and III of all the plots (P1-P14, R1-R4) and the species that contribute to at least l.S9c of the inertia of one or the other of the factorial axes. Species numbers are those given in Table 3-1. ing up group C that lie along this axis. Thus, P8, located on a mid-slope at 1280 m, is opposite R4 along a stream at 1450-1460 m, and P9, on a lower slope at 1500 m, is intermediate between them. Groups A and B are compact and contribute only weakly to the expression of factor IV. The altitudinal gradient, which is also a cli- matic gradient, thus explains most of the floristic heterogeneity among the plots. Based on the pte- ridophyte floristic composition, four altitudinal stages can be recognized on the eastern slope of the RNI d'Andohahela: stage A, at low elevations around 420-530 m (PI, P2, Rl, and R2); stage B, at lower mid-elevations, between 820 and 1 1 50 m (P3-P7 and R3); stage C, at upper mid-elevations, between 1280 and 1500 m (P8, P9, and R4); and stage D, at high elevations, between 1510 m and the summit at 1956 m (P10-P14). In the absence of data from between 600 and 700 m, it is not possible to assess the upper limit of stage A. Each of the two lower stages, A and B, is floristically homogeneous (plot points re- main clustered on all of the correspondence anal- ysis [CA] projection plans). On the other hand, each of the upper stages is floristically heteroge- neous (plot points are stretched along axis II, III, or IV), a fact that reflects both an altitudinal and a topographic gradient, although it is not possible to determine which factor is predominant. Other factors could also influence pteridophyte floristic composition, such as the structure and composi- tion of the soil, the level of human impact in the area, etc., although their influence on the eastern slope of the RNI d'Andohahela appears to be fair- ly weak and does not alter the general pattern of altitudinal distribution. RAKOTONDRAINIBE: PTERIDOPHYTES 39 IV T=8.7% Fig. 3-7. Correspondence analysis of the matrix of 18 plots and 162 pteridophyte species in parcel 1 of the RNI d'Andohahela: projection on axes I and IV of all the plots (P1-P14, R1-R4) and the species that contribute to at least 1.8% of the inertia of one or the other of the factorial axes. Species numbers are those given in Table 3-1. Floristic Analysis of Each Altitudinal Stage Range of Habit Types (Fig. 3-8; Table 3-4) — The percentage of epiphytic species increases pro- gressively with altitude (36.2% in stage A, 44.7% in stage B, and 48.0% in stage C), and epiphytes become dominant (55.1%) in stage D, i.e., above 1510 m. The percentage of terrestrial species is more or less constant between 420 and 1500 m (41.3%, 36.6%, and 40.0%, respectively, for stages A, B, and C) and decreases in stage D to 29.5%. The percentage of species that are both terrestrial and epiphytic remains nearly constant from one stage to another (11.3%, 10.6%, 7.0%, and 10.3%). Those that are strictly epilithic de- crease regularly with increasing altitude despite the presence of numerous rocky outcrops near the summit. These rocky outcrops are actually smooth, covered with little or no humus, and are exposed to intense insolation in the late morning, making them poorly adapted for the establishment of a diverse pteridophyte flora. Floristic Richness at the Generic Level (Fig. 3-9; Table 3-5) — The richness in number of gen- era represented in the pteridophyte flora varies with altitude. The maximum value (49 genera) oc- curs between 850 and 1150 m. Table 3-6 shows, for each altitudinal stage, the number of species in the eight largest genera of ferns in the RNI d'Andohahela. Asplenium is well diversified in all of the stages and is clearly dominant (24 spp.) between 810 and 1150 m. At low elevations only 40 FIELDIANA: ZOOLOGY 60 50 HT/r ■ E/r DT/E/r E3R ■ L 420-550m 810-1, 150m 1,280- 1,500m 1,510-1 ,957m Fig. 3-8. Percentages of pteridophyte flora (207 species listed in Table 3-1) by habit type for each altitudinal zone in parcel 1 of the RNI d'Andohahela. See Table 3-1 for explanation of abbreviations. two genera are notably speciose, Trichomanes (12 spp.) and Asplenium (11 spp.). Cyathea and Blechnum are well represented in the middle al- titudinal stages, whereas Elaphoglossum, Huper- zia, and Grammitis, all of which are most often epiphytes, are absent or rare at low altitude but become more speciose above 600(?) or 800 m as atmospheric humidity increases due to more fre- quent fog and mist. Floristic Richness at the Specific Level — Table 3-5 shows the total number of species re- corded at each altitudinal stage, as indicated in Table 3-1, and also the number per unit surface area (100 m2), based on the plot samples. Overall species richness is highest (123 spp.) between 810 and 1150 m, where the habitat is very diversified. Higher in elevation, between 1280 and 1500 m, forest structure is simpler, more uniform, and the number of species present (100 spp.) is lower despite their being more numerous per unit area (6.7 per 100 m2 vs. 5.5 per 100 m2 at lower elevations). Altitudinal Indicators — In an attempt to as- sess the ecological status of each species, infor- mation from the general inventory (Table 3-1) was used along with the results from the statistical analyses of the plot samples. The first three axes of the CA of the plot-species matrix are inter- preted as altitudinal factors, whereas the fourth axis appears to be either an altitudinal or a topo- graphic factor, or possibly both. Consequently, only species that contribute substantially to the inertia of at least one of the three first axes (con- tribution [CTR] >1.8%) are regarded as useful al- titudinal indicators. A total of 48 species meet these criteria (Table 3-7). The value of 1.8% was chosen a posteriori as the threshold because it eliminates species that are too widely distributed, those that are infrequent, and those with an erratic distribution. Indicator species that are restricted to a single altitudinal stage are referred to as "exclusively characteristic" of that stage (see Table 3-7). Those that are significantly more abundant statistically in a given stage than elsewhere (i.e., those that have higher coefficients of abundance in the con- stituent plots) and have a CTR >1.8% are called "preferentially characteristic" (see Table 3-7). Exclusively characteristic species are shown in boldface type and preferentially characteristic species are shown in normal type in Table 3-7. Group 1 species are characteristic of the low-el- evation stage A; those in group 2 are character- istic of stage B at mid-elevations; group 3 is char- acteristic of mid-elevation stage C; and group 4 is characteristic of stage D at high altitudes. The species in groups 5 and 6 are characteristic, re- spectively, of stages A + B and C + D. RAKOTONDRAINIBE: PTERIDOPHYTES 41 Table 3-4. Pteridophyte habit type for each altitudinal zone. 420-550 m 810—1150 m 1280-1500 m 1510-1957 m T/r 41.3% 36.6% 40.0% 29.5% E/r 36.2% 44.7% 48.0% 55.1% T/E/r 11.3% 10.6% 7.0% 10.3% R 10.0% 8.1% 5.0% 5.1% L 1.2% 0 % 0 % 0 % See Table 3- 1 for explanation of abbreviations. Conclusions and Discussion The pteridophyte flora of the eastern slope of the RNI d'Andohahela comprises 207 species and varieties that represent 69 genera. One of them, an undescribed species of Xiphopteris (Rakoton- drainibe 3243), is currently the only known taxon endemic to the massif, whereas three other taxa are endemic to the southeastern region of Mada- gascar. The results of the quantitative floristic analysis make it possible to recognize four altitudinal stages within the study area: Stage A, at low elevations between 420 and 550 m, occupied by dense, moist evergreen forest. A total of 40 genera and 80 species or varieties of ferns, five of which are exclusively character- istic species, were found in this formation. The genera Trichomanes (12 spp.) and Asplenium (11 spp.) are dominant in terms of the number of spe- cies present. The average density of pteridophyte species is 3.7 per 100 m2. Terrestrial species are most numerous (41.3%), whereas epiphytes rep- resent 36.2% of the total, and epiliths 10.0%. Stage B, at lower mid-elevations between 810 and 1150 m, with dense, moist montane forest in which 49 genera and 123 species or varieties of ferns were recorded. Asplenium is clearly domi- nant (24 spp.); Elaphoglossum (10 spp.) and Cy- athea (9 spp.) are well diversified. No species is restricted to this stage, and the average species density is 5.5 per 100 m2. Epiphytic species are dominant (44.7%), indicating an increase in rela- tive atmospheric humidity as compared to stage A, at low elevation. Terrestrial species are nev- ertheless well represented (36.6%), and strictly epilithic species are more rare (8.1%). Stage C, at upper mid-elevations between 1 280 and 1500 m, which is in fact a transition between dense, moist montane forest and the sclerophyl- 140 120 100 80 60 40 20 0 I I 420-550m 81 0-1 ,150m I HMHL ■ Genera □ Species 1,280- 1,500m 1,51 0-1, 957m Fig. 3-9. Generic and specific floristic richness of pteridophytes (207 species listed in Table 3- 1 ), at each altitudinal zone, in parcel 1 of the RNI d'Andohahela. 42 FIELDIANA: ZOOLOGY Table 3-5. Generic and specific richness for each altitudinal zone. 420-550 m 810-1150 m 1280-1500 m 1510-1957 m Genera (total number) Species (total number) Number of species/ 100 m2 40 80 3.7 49 123 5.5 37 100 6.7 33 78 3.5 lous forests of the highest stage. An intermingling of the forest types characteristic of the stages above and below results in a particularly high av- erage species density (6.5 spp. per 100 m2), bal- anced percentages of epiphytic and terrestrial spe- cies (48.0% and 40.0%, respectively), and a cer- tain level of heterogeneity in the flora that can be seen along a topographic and altitudinal gradient. One hundred species, eight of which are exclu- sively characteristic, were recorded in this stage. The dominant genera are Asplenium (13 spp.), Trichomanes (11 spp.), and Elaphoglossum (10 spp.). Stage D, at high elevations between 1510 and 1956 m, occupied by montane sclerophyllous for- est that includes 33 genera and 78 species of ferns, of which seven are exclusively character- istic. Epiphytic species are highly dominant (55.1%) as compared to terrestrials (29.5%). The most diversified genera are Asplenium (13 spp.) and Elaphoglossum (10 spp). A comparison of the the pteridophyte flora of the RNI d'Andohahela with that of the RS d'Anjanaharibe-Sud, located in the northern part of the island (Rakotondrainibe & Raharimalala, 1998), makes it possible to develop a better un- derstanding of the factors responsible for its cur- rent composition. Comparative data are presented in Table 3-8. Note that the studies conducted in these two reserves were essentially identical: fieldwork covered a 2-month period at the same time of year in each case and was conducted using identical methods. (1) Despite its position in extreme southern Madagascar, outside of the Tropic of Capricorn, the RNI d'Andohahela has a pteridophyte flora as rich as that of the RS d'Anjanaharibe-Sud, where 211 species and varieties were recorded in 1994 (Rakotondrainibe & Raharimalala, 1998). A large portion of the flora is shared between the two ar- eas (121 species and varieties, representing 50% of the taxa in each of the reserves), giving a S0rensen's similarity coefficient of 57.9 (200 X 121/207 + 211). The two reserves have many oth- er aspects in common. They are both situated on ancient granitic massifs that reach approximately the same elevation (1956 m and 2064 m, respec- tively), and they fall within the same bioclimatic stages: the humid stage at low altitudes, the sub- humid stage above about 600-800 m, and the montane stage above ca. 1800 m (Cornet, 1974). These stages are characterized by high rainfall and/or extended periods of mist and fog. (2) At the same time, the fact that the two re- serves are located some 1450 km apart and at op- posite ends of Madagascar (one in the far south and the other in the extreme north), separated by about 10° of latitude, results in substantial climatic differences between them (Table 3-9). The coastal towns Tolagnaro in the south and Antalaha in the north were chosen for comparing the climates of the two regions in which the reserves are located, primarily because of their proximity to the study zones and the availability of complete meteoro- logical records covering many years. Although Tolagnaro is in the same bioclimatic stage as An- Table 3-6. Species richness of the eight argest pteridophyte genera for each altitudinal zone. Total: Genus 420-550 m 810-1150 m 1280-1500 m 1510-1957 m 420-1957 m Asplenium 11 24 13 13 30 Elaphoglossum 4 10 10 10 20 Trichomanes 12 9 11 4 15 Cvathea 4 9 8 2 14 Huperzia 0 6 6 6 9 Hymenophxilum 3 5 6 6 8 Blechnum 1 3 6 2 8 Grammitis 0 1 4 4 7 RAKOTONDRAINIBE: PTERIDOPHYTES 43 •* - r*> (^ cu 9\ Q o» «s 2 3 & in c« o I WD © - is I M © -. I q -a to s. ft 9 S 5-~ s a w — -3 -2 ~ § &&I s as 1 1 2 y 3 S.-S 0 5 3 9 "S ^«2 a. fc § fe -^ "ST ^ -3 -3 S o, « 3^£ Tf m m •* id m cm - n^t — m m c ci m — i so CS vO iri (N >n vo ro VO ci —i rf tj- i/~> m Niriinvivi\o min in in-^-^tvD'sfin^ ^OVO'OVO^tvD^O'rl-VO U-)\D« (N VO ^O -rf in 2 c-l 2 %> .3 c %J 3 -0 Si Q 5 S-S S F4. 5 ° i §"•*> -S .2 -3 -3 5 9 * §.-3 r^vDvO^tiOWinvOvivO'tr, sO ^t in vo- ^r C i» | - «i 3 BO u 3 3 U "u 'S -3 gg ■5 XI Q s 4! 3 <*-«' u-i -C 2 £ s ,a s 3 <4-i 3 ■ — 3 0 £ as Si. e a «) Sj U 3 a •« •« - c X &i 3 3 CQ c EF «s ^. > >l tt QQ U'U'U "Ss s? s .§ ^ S 3 3 5 & K 5 t; tu ^ ^^ I ON 2 "3 3 2 -2 oc so Gift, /"> -3- m Tf m m it, io it, ^) IT) vO vO T. vO ci IT) C-, «T) r> m /"> vC ^T c~, sC IO vD vD * vivOTf - vO vO 't W h'U S a a ts a a ft. E s <*5 I £ 5 a £ •SJ a i| ■« 2 Id a 3.2 §>rl ■a < g a. a. s -2-2 2 Ljj kj O a -a .s 2 S s 8 a j- -a g KI ** 6c S- E-S a • a a oc-a . c a! a v gi S a « -S ■s as S 5 a- a 2 M 3 •2 2 a. E"| B-S. *■> & -2 a 5: a. t*j a: S 5 a •2 -2 a v> c c g 3 S- S-i (j a "3 a a o II a -a •g S* S .y a v> -a 3 •2 a ■a 2 -2S •'"-' ■a a. II ■as o J2 J2 ^ £»> „«j o u -^ "^ ^ S. co u o o •2 k -a a E"5 E-S a a. is %J "J .2 «J 9 a — 3 •J X. 4> M c :/: *o u :a "7 'C U 3 u - > >..a > n. RAKOTONDRAINIBE: PTERIDOPHYTES 45 Table 3-8. Species richness, endemism, and degree of similarity of the pteridophyta flora of the RNI d'Andohahela and RS d'Anjanaharibe-Sud. Parameter evaluated RNI d' Andohahela RSd' Anjanaharibe-Sud Ps Total number of species and varieties Total number of endemic Malagasy species Total number of species endemic to the massif Number of species between 400 and 600 m Number of species between 800 and 1 100 m Number of species between 1250 and 1500 m Number of species between 1 550 and 2000 m 207 211 57.9 67 (32.4%) 1 82 (38.9%) 5 80 52 45.5 123 113 46.6 100 122 49.5 78 117 32.8 Ps = S0rensen's coefficient of similarity. talaha, its climate is markedly drier and cooler. Average annual precipitation differs by 613 mm, and the average number of days with rainfall is greater by 60.4 days at Antalaha. Average relative humidity is similar (differing only by 5%), but both stations are located close to the sea, and in light of the difference in rainfall it is likely that humidity levels are more divergent farther inland. The only substantial differences in average tem- peratures between the two stations are those of the minima; the average temperature during the coldest month is 2.8°C higher at Antalaha, but that for the warmest month differs by only 0.3°C. The difference in average temperatures between the warmest and coldest months is 14.2°C at Tolag- naro and 1 1 .7°C at Antalaha. The recorded differ- ences in rainfall and temperatures between the two regions, associated with varying degrees of human impact, can thus explain (at least in part) the observed differences in the overall floristic composition and in the altitudinal distribution of floristic richness in the two reserves (Table 3-8). The genera Hymenophyllum and Ctenopteris, most of whose species are epiphytes that prefer moist microhabitats, are abundant in the high-el- evation formations; they are less speciose in the RNI d'Andohahela (with 8 and 5 spp., respective- ly) than in the RS d'Anjanaharibe-Sud (14 and 9 spp.). By contrast, Asplenium, the species of which are much more tolerant of short dry peri- ods, is more diverse in the RNI d'Andohahela (30 vs. 26 spp. in the RS d'Anjanaharibe-Sud), es- pecially between 810 and 1 100 m (24 vs. 17 spp.). Between 400 and 1100 m overall pteridophyte species richness is higher in the RNI d'Andohahela, a trend that is reversed starting at 1280 m. This can be interpreted as follows: on the eastern slope of the RS d'Anjanaharibe-Sud the low-elevation moist evergreen forest is now largely destroyed and covers only a small area. On the other hand, this formation is still extensive and well preserved in the RNI d'Andohahela, although clear signs of earlier human presence, such as funerary monu- ments, are evident (see Chapter 1). Above 1250 m the presence of more pteridophyte species in the RS d'Anjanaharibe-Sud is most likely the con- sequence of a climate more favorable to their growth (smaller difference in average temperature and higher average annual precipitation) and al- most certainly is also due to the proximity of an- other large mountain massif, Marojejy. Marojejy must play a role as a local source of diaspores, Table 3-9. Average meteorological data from Tolagnaro and Antalaha. Stations Annual precipitation (mm) Number of days with rainfall Relative humidity (%) Temperature during the warmest month (M) Temperature during the coldest month (m) M - m Tolagnaro Antalaha 1537.1 2150.3 152.8 213.2 76 81 29.5 29.8 15.3 18.1 14.2 11.7 Rainfall and temperature data are from Morat (1969); relative humidity data are from Chaperon et al. (1993). Morat (1969) used data from 1931-1960 for both cities; Chaperon et al. (1993) did not indicate the source of their data, but they were taken prior to 1970 (Chaperon, pers. comm.). 46 FIELDIANA: ZOOLOGY thereby increasing the total area of high-altitude habitats within the region. (3) The floristic similarities observed between the two study areas, as expressed by the S0rensen's coefficient (Ps), are comparatively strong between 400 and 1500 m (45.5 < Ps < 49.5), but are much weaker starting at 1500 m (Ps = 32.8). However, this coefficient only takes into consideration the presence or absence of species. If abundance values are used, a somewhat differ- ent interpretation results; the floristic similarities become evident between 400 and 1150 m, and the two massifs share eight exclusively or preferen- tially characteristic species within this altitudinal range (Asplenium bipartitwn, A. cuneatum, A. in- aequilaterale, A. nidus, A. variabile var. paucijug- um, Lygodium lanceolatum, Microsorum puncta- tum, and Nephrolepis biserrata). Above 1280 m only one shared species (Asplenium friesiorum) is an altitudinal indicator, and with the exception of this taxon, the upper portions of the two massifs have completely different characteristic species, reflecting much weaker levels of floristic similar- ity. These results demonstrate the importance of comparative studies of pteridophytes that take into consideration not only the presence (or absence) of the species but also their abundance. (4) The data currently available indicate that the level of specific endemism is lower in the RNI d' Andohahela than in the RS d' Anjanaharibe-Sud. A total of 32.4% of the fern taxa recorded in the RNI d' Andohahela are endemic to Madagascar and one species is restricted to the Trafonaomby Massif, whereas 38.9% of the taxa at RS d' Anjanaharibe-Sud are Madagascar endemics and five are known only from the Anjanaharibe- Sud Massif. However, botanical exploration is far from complete in southeastern Madagascar, es- pecially with respect to pteridophytes, for which the specimen base in national and international botanical institutions remains small. Acknowledgments I thank Beberonn Randriamampionona for his technical assistance in the field and Sergio Ro- maniuc Neto for assistance with the statistical analyses. I am also very grateful to Pete Lowry, Steve Goodman, Annick Le Thomas, and anony- mous reviewers for comments and suggestions that improved earlier drafts of the paper. Special thanks go to Pete Lowry for translating the manu- script. This project was supported in part by the Laboratoire de Phytomorphologie of the Ecole Pratique des Hautes Etudes. Literature Cited Benzecri, J. P., and F. Benzecri. 1984. Pratique de 1' analyse des donnees. 1. Analyse des correspond- ances, exposd 61ementaire, 2nd ed. Bordas, Paris, 456 pp. Chaperon, P., J. Danloux, and L. Ferry. 1993. Fleuves et rivieres de Madagascar. Editions OR- STOM, Montpellier, 874 pp. Christensen, C. 1932. Pteridophyta of Madagascar. Dansk Botanisk Arkiv, 80: 1-253. Cornet, A. 1974. Essai de cartographie bioclimatique a Madagascar, carte a 1/2 000 000 et notice, no. 55. ORSTOM, Paris. Cornet, A., and J.-L. Guillaumet. 1976. Divisions floristiques et etages de vegetation a Madagascar. Ca- hiers ORSTOM, serie Biologic, 11(1): 35-40. Guillaumet, J.-L. and J. Koechlin. 1971. Contribution de Madagascar a la definition des types de vegetation dans les regions tropicales. Candollea, 26(2): 263- 277. Holttum, R. E. 1974. Thelypteridaceae of Africa and adjacent islands. Journal of South African Botany, 40(2): 123-128. . 1986. Studies in the fern-genera allied to Tec- taria Cav. VI, A conspectus of genera in the Old World regarded as related to Tectaria, with descrip- tions of two genera. Gardens' Bulletin. Straits settle- ments. Singapore, 39(2): 153-167. Humbert, H. 1935. L'extinction des derniers vestiges de certains types de vegetation autochtone a Mada- gascar. Archives du Museum d'Histoire Naturelle. Paris, serie 6, 12: 569-587. . 1941. Le massif de l'Andohahela et ses 66- pendances. Compte rendu de la Soci£t6 de Biogeogra- phie, 18: 32-37. -. 1955. Les territoires phytogeographiques de Madagascar. In Colloques Internationaux du Centre National de la Recherche Scientifique, LIX: Les di- visions ecologiques du monde, moyen d'expression, nomenclature, cartographie, Paris, 1954. Ann6e Biol- ogique, 3e serie, 31(5-6): 439-448. Humbert, H., and G. Cours Darne. 1965. Notice de la carte de Madagascar. Travaux de la Section Scien- tifique et Technique de lTnstitut Francais de Pondi- chery. Hors Serie no. 6. Jambu, M. 1978. Classification Automatique pour T Analyse des Donnees. 1. Methodes et Algorithmes. Editions Bordas, Paris, 310 pp. Kramer, K. U. 1972. The Lindsaeoid ferns of the Old World, IX. Africa and its islands. Bulletin du Jardin Botanique National de Belgique 42: 305-345. Kramer, K. U., and P. S. Green. 1990. Pteridophytes and gymnosperms. In Kubitzki, K., ed. The Families and Genera of Vascular Plants. Springer- Verlag, Ber- lin, 404 pp. Legendre, L., and P. Legendre. 1984a. Ecologie Nu- RAKOTONDRAINIBE: PTERIDOPHYTES 47 merique. 1. Le Traitement Multiple des Donnees Ecol- ogiques, 2nd ed. Editions Masson, Paris, and les Presses de l'Universite Laval, Quebec, 260 pp. 1984b. Ecologie Numerique. 2. La Structure des Donnees Ecologiques, 2nd ed. Editions Masson, Paris, and Les Presses de l'Universite Laval, Quebec, 335 pp. Morat, P. 1969. Note sur 1' application a Madagascar du quotient pluviothermique d'Emberger. Cahiers OR- STOM, Serie Biologie, 10: 117-130. Paulian, R., C. Blanc, J.-L. Guillaumet, J.-M. Betsch, P. Griveaud, and A. Peyrieras. 1973. Etude des eco- systemes montagnards dans la region malgache. II. Les chatnes Anosyennes. Geomorphologie, climato- logie et groupements vegetaux. (Campagne RCP 225, 1971-1972). Bulletin du Museum National d'Histoire Naturelle, Paris. Ecologie generate 1, serie 3, 118: 1- 40. Perrier de la Bathie, H. 1921. La vegetation malga- che. Annales de l'institut botanico-geologique colo- nial de Marseille, serie 3, 9: 1-268. Pichi-Sermolli, R. E. G. 1977. Tentamen Pteridophy- torum genera in taxonomicum ordinem redigendi. Webbia, 31(2): 313-512. Rakotondrainibe, E, and F. Raharimalala. 1996. The pteridophytes of the eastern slope of the Reserve Na- turelle Integrale d'Andringitra, Madagascar, pp. 76- 82. In Goodman, S. M., ed., A floral and faunal in- ventory of the eastern side of the Reserve Naturelle Integrale d'Andringitra, Madagascar: With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. . 1998. The pteridophytes of the Reserve Spe- ciale d'Anjanaharibe-Sud, Madagascar, pp. 17-38. In Goodman, S. M., ed., A floral and faunal inventory of the Reserve Speciale d'Anjanaharibe-Sud, Mada- gascar: With reference to elevational variation. Field- iana: Zoology, n.s. 90: 1-246. Schelpe, E. A. C. L. E. 1969. Review of tropical Af- rican pteridophyta I. 4. The Lomariopsidaceae of con- tinental Tropical Africa. Contribution from Bolus Her- barium, 1: 25-43. . 1970. Pteridophyta. In Exell, A. W., and E. Launert, eds. Flora Zambesiaca. Crown Agents, Lon- don, 254 pp. Schelpe, E. A. C. L. E., and N. Anthony. 1986. Pter- idophyta. In Leistner, O. A., ed. Flora of Southern Africa. Department of Agriculture and Water Supply, Pretoria, 292 pp. S0rensen, T. 1948. A method of establishing groups of equal amplitude in plant sociology based on similarity of species content. Det Kongelige Danske Videnska- bernes Selskab, Biologiske Skrifter, 5(4): 1-34. Tardieu-Blot, M. L. 1951-1971. Les Pteridophytes. In Humbert, H. ed. Flore de Madagascar et des Comores. Museum National d'Histoire Naturelle, Paris. . 1958. 5e famille, Polypodiaceae (sensu lato). In Humbert H., ed. Flore de Madagascar et des Co- mores. Museum National d'Histoire Naturelle, Paris. Tindale, M. D. 1965. A monography of the genus Las- treopsis Ching. Contributions of the New South Wales National Herbarium, 3(5): 249-339. Appendix 3-1 The Pteridophytes of Parcel 2 of the RNI d'Andohahela An inventory of the pteridophytes of parcel 2 of the RNI d'Andohahela was conducted at low elevation (150-180 m) along the Mananara River, 7.5 km NNE of the village of Hazofotsy. The re- gion is situated within the subarid bioclimatolog- ical stage (Cornet, 1974), characterized by an av- erage annual minimum temperature between 10° and 13°C, a cumulative water deficit of 300-400 mm, and a dry season of 9 months. The vegeta- tion present on sandy-clay soils is a low open xeric forest that is rich in Commiphora, Diospy- ros, Euphorbia, P achy podium, and Alluaudia. Sandy and calcareous soils and rocky outcrops are covered by a xeric thicket that is dominated by Didiereaceae and Euphorbiaceae. Fieldwork was conducted during a 3-day period within a 2 km radius around our base camp. A phytoecological study was done in low xeric forest within a 200 m2 plot (P15) established on a gentle slope (15°) with eastern exposure. The distribution of pteri- dophytes in this dry formation was highly hetero- geneous, with the species density per 100 m2 ranging between 0 and 4. The location of the plot was chosen so that as many species as possible would be included (i.e., eight species in the 200 m2 samples). The 1 1 species of ferns recorded in the dry for- mations within parcel 2 are listed in Table A3-1. Five of them (in boldface type), including three species of Selaginella, are endemic to Madagas- car. None of the species occurs in the Eastern Do- main. Adiantum capillus-veneris and Doryopteris madagascariensis are widely distributed in Mad- agascar, including in the Sambirano, Central, Western, and Southern domains. All of the other species are limited to the Western and Southern domains. 48 FIELDIANA: ZOOLOGY Table A3-1. Pteridophytes of parcel 2 of the RNI d'Andohahela. Taxa* Frequency Habit type Ecology Actiniopteris radiata (Koenig ex Sw.) Link Adiantum capillus-veneris L. Doryopteris concolor (Langsd. & Fisch.) Kuhn Doryopteris madagascariensis Tardieu Doryopteris pilosa (Poir.) Kuhn Notholaena lanceolata Bonap. Pallaea calomelanos (Sw.) Link. Pallaea viridis (Forssk.) Prantl var. glauca Sim. Selaginella digitata Spring Selaginella helicoclada Alston ex Alston Selaginella proximo R. M. Tryon + + terrestrial rocky soil + terrestrial stream bank + terrestrial open understory + + terrestrial open understory + terrestrial open understory + + terrestrial rocky soil + terrestrial clay slope + terrestrial open understory + + + terrestrial open understory + terrestrial open understory + + + terrestrial open understory * Species endemic to Madagascar are indicated in boldface type. + = rare (seen 1 to 5 times); ++ = infrequent (seen 6 to 20 times); + + + = frequent (seen 21 to 50 times). RAKOTONDRAINIBE: PTERIDOPHYTES 49 Chapter 4 Structure and Floristic Composition of the Vegetation in the Reserve Naturelle Integrate d9Andohahela9 Madagascar Pierre Jules Rakotomalaza1 and Nathalie Messmer2 Abstract Studies were undertaken on the vegetational structure and floristic composition of five 1 ha plots in the humid forest of parcel 1 of the Reserve Naturelle Integrale (RNI) d'Andohahela. These five plots were established along an altitudinal gradient between 440 and 1 875 m. A total of 4,875 trees (>10 cm dbh) covering 359 species and 48 plant families were recorded. The number of species per hectare ranged from 52 to 147. The total number of trees per hectare varied between 675 and 1,365 individuals. Total basal area varied from 34.1 to 65.9 m2. No one plant dominated all of the five study plots. Few species dominated each plot; 50% of the trees with dbh >10 cm were represented by between nine and 18 species. Thus, the Shannon-Weaver diversity index showed very low values. Data from these plots were first analyzed individually and then compared between plots in order to document floristic and structural variation along the elevational gradient. The patterns found in parcel 1 of the RNI d'Andohahela are compared to other sites in Madagascar, as well as to other areas in Asia, Africa, and South America. Studies were also conducted in the spiny forest of parcel 2 of the RNI d'Andohahela. Ten linear transects ("Gentry transects"), with a total surface area of 0.1 hectare, were established. Trees with a dbh >2.5 cm were censused. Of the 607 individuals, 540 plants were identified as belonging to 85 species in 28 families. The Shannon- Weaver diversity index value for this site was very low due to the dominance of a few families and species. These results are compared with another spiny forest in southern Madagascar and with sites in the Neotropics and Africa. Resume Un inventaire floristique quantitatif et qualitatif de la foret dense humide de la Reserve Naturelle Integrale (RNI) d'Andohahela (parcelle 1) d'une part, ainsi que de la foret dense seche de la RNI d'Andohahela (parcelle 2) d' autre part, a ete effectue. Dans la foret dense humide de la parcelle 1 de la RNI d'Andohahela, cinq parcelles permanentes d'un hectare chacune ont ete mises en place en un gradient altitudinal s'echelonnant de 440 m a 1875 m d' altitude. Un total de 4,875 arbres de diametre a hauteur de poitrine (dbh) egal ou superieur a 10 cm, comptant 359 especes and 48 families de plantes a ete recence. Le nombre d'especes par hectare s'etend de 52 a 147. Le nombre total d'arbres par hectare varie entre 675 et 1,365 individus. L'aire basale total varie entre 34.1 et 65.9 m2. Aucune plante ne domine dans les cinq parcelles permanentes. Et meme si le nombre d'especes dans chaque plot n'est en rien 1 Missouri Botanical Garden, B.P. 3391, Antananarivo (101), Madagascar. 2 Conservatoire et Jardin Botaniques de la Ville de Geneve, Case Postale 60, CH- 1 292 Chambesy/Geneve, Swit- zerland. RAKOTOMALAZA & MESSMER: VEGETATION 51 extreme, le nombre d' especes dominant chaque parcelle permanente est faible, a savoir entre neuf et dix-huit especes representent les 50% des arbres de dbh >10 cm. En consequence, les valeurs de l'indice de diversite de Shannon- Weaver sont tres basses. Ces parcelles permanentes d'un hectare ont d'abord ete analysees individuellement et ensuite mises en relation les unes par rapport aux autres, afin de mettre en evidence les variations floristiques et structurales en fonction de l'altitude. Ces resultats sont compares a ceux d'autres sites a Madagascar, ainsi qu'en Asie, Afrique et Amerique du Sud. Un inventaire floristique effectue dans la foret dense seche de la parcelle 2 de la RNI d'Andohahela a egalement ete mene. Dix transects lineaires de Gentry, totalisant une surface de 0.1 hectare, ont ete mis en place. 607 individus de dbh >2.5 cm ont ete recenses, dont 540 ont ete determines, appartenant a 85 especes et 28 families. La value de l'indice de diversite de Shannon-Weaver est tres basse, a cause de la dominance de quelques especes et families. Ces resultats ont ete compares a un autre site du sud de Madagascar, ainsi que des sites dans les Neotropiques et en Afrique. Introduction The Reserve Naturelle Integrate (RNI) d'Andohahela, with its three noncontiguous par- cels, contains a remarkable range of vegetation types. These parcels consist of parcel 1, a large zone of humid forest along the eastern slope of the Anosyenne Mountains that contains lowland forest to high mountain forest; parcel 2, with spiny forest and gallery forest; and parcel 3, with a form of transitional habitat between humid and spiny forests (see Chapters 1 and 2). During the 1995 inventory of parcels 1 and 2 our intent was not simply to conduct a qualitative survey of the local plant species, but also to make a quantitative assessment of floral and structural aspects of these forests. Given the differences in vegetation between parcels 1 and 2, combined with time limitations inherent to rapid surveys, we were faced with methodological constraints and had to use two different survey techniques, which are discussed in detail below. Accordingly, this report is divided into two parts. The first part addresses vegetation in the humid forest of parcel 1 . Humid forests are usually divided into vegetation zones associated with elevational bands that are based on quanti- tative characteristics (e.g., diameter at breast height [dbh], height, canopy width) and floristic attributes. The second part of this chapter con- cerns the spiny forest of parcel 2. Few studies of forest composition and structure have been con- ducted on this vegetational type. Our specific ob- jectives for both forest types were (1) to provide a quantitative account of the study sites; (2) to provide a comparison of floristic composition be- tween plots established at different elevations in the humid forest (parcel 1); (3) to document the vegetational changes along the elevation gradient in the humid forest (parcel 1); and (4) to augment available information on the flora and structure of the spiny forest (parcel 2). A qualitative and quantitative survey requires sampling. Plots are quantitative, potentially rig- orous statistically, and provide a good measure of the structure, floristic composition, and vegetation variability. Transect sampling is a rapid method that uses less sophisticated measurements. These qualitative (identification by scientific name) and quantitative (number, dbh, and height) data pro- vide a basis with which to assess physical param- eters (e.g., distribution of dbh and height) between different elevations or formations, density, domi- nance, frequency, and species diversity. On the basis of this information it is possible to quantify diversity at each site. General collections were made in each altitudinal zone that allow species lists for each site to be further supplemented; identifications were generally limited to fertile in- dividuals. Ecologists have devoted considerable effort to developing various indices of diversity that factor in aspects of both the number of species and their relative densities. One of the most frequently used is the Shannon- Weaver diversity index (Shannon & Weaver, 1949). Each plot was analyzed sepa- rately and then compared to the others using the Shannon-Weaver diversity index and the Horn similarity index (Goldsmith et al., 1986). Methods The quandary that has hindered developments in the quantification of tropical plant communities 52 FIELDIANA: ZOOLOGY during rapid assessments is that the sampling area needs to be big enough to contain a sufficiently large number of species that are representative of the local community, yet small enough to be prac- ticably and rapidly studied. Botanists often make numerous counts or collect specimens of redun- dant common and easily accessible species and overlook rarer and less obvious species. For the RNI d'Andohahela inventory, sampling and data collections were conducted in five 1 ha permanent plots established at five different altitudinal zones in parcel 1 and in ten transects ("Gentry tran- sects") in parcel 2. Sampling Procedure and Study Sites Permanent Plots — One-hectare permanent plots have been established in other forests of Madagascar largely on the recommendation of the Missouri Botanical Garden. Three 1 ha permanent plots were installed and sampled in the humid for- est of Pare National (PN) de Ranomafana (Schatz, 1994); two on the Masoala Peninsula; one in the RS d'Ambohitantely (Rabevohitra et al., 1996; C. Birkinshaw, pers. comm.); and ten along the east coast: Tampolo (Fenerive est), Andranomintina (Masoala), Tanambao (Manajary), Reserve Spe- ciale (RS) de Manombo (Farafangana), and Ste. Luce (Tolagnaro). . Permanent plots provide both quantitative and floristic data that, with continued sampling, permit analyses of growth rates and forest dynamics. Di- ameter at breast height was measured 1.44 m off the ground (Malcomer, 1991) and not at the more standard height of 1.3 m off the ground (CFT, 1989). Trees of 5:10 cm dbh were marked with numbered metal tags, their dbh measured, and their height estimated from ground level. Voucher specimens, even sterile, were collected. Material from fertile plants was later compared with ma- terial housed in the herbaria of the Pare Botanique et Zoologique de Tsimbazaza (PBZT) and the Di- rection des Ressources Forestieres et Piscicoles (DRFP), Antananarivo. The nonfertile material was assigned to a "morphospecies" when a bi- nomial name could not be determined (Malcom- ber, 1991), and these morphospecies are num- bered sequentially. With the exception of the plot at 1875 m, all of the sites were located close to our camps in parcel 1. Owing to difficulties in capturing satel- lite signals with a global positioning system in forest with a thick canopy, the coordinates of the 1 ha plots given below are those for the adjacent camp sites (Chapter 1, Fig. 1-1). The coordinates for the 1875 m site come from the nearby forest edge. The following plots were established: Plot 1 — 440 m, established near the Andrano- hela River, 8.0 km NW of the village of Eminiminy, 24°37.6'S, 46°45.9'E. Shape 500 x 20 m. Plot 2 — 840 m, placed near a tributary of the An- dranohela River, 12.5 km NW of the village of Eminiminy, 24°35.6'S, 46°44.3'E. Shape 200 X 50 m. Plot 3—1 150 m, on a plateau 13.5 km NW of the village of Eminiminy, 24°35.0'S, 46°44.1'E. Shape 100 X 100 m. Plot 4 — 1550 m, 15.0 km NW of the village of Eminiminy, 24°34.2'S, 46°43.9'E. Shape 250 X 40 m. Plot 5 — 1875 m, established on the plateau just below Pic Trafonaomby (1959 m), 20 km SE of the village of Andranondambo, 24°33.7'S, 46°43.3'E. Shape 200 x 50 m. In each elevational zone general plant collect- ing was conducted outside the plots for additional floristic information. Linear Transects — The linear transect method has proved to be satisfactory for analyses of struc- tural and floristic composition (Gentry, 1982). For example, using this method Gentry (1988) sam- pled 1 30 different forests at about 70 sites in four high-species-diversity areas of the world. In Mad- agascar, this approach has been used previously in dry deciduous forests (Gentry, 1988; Du Puy et al., 1994). Our quantitative and qualitative floristic inven- tory was conducted in the spiny forest of the RNI d'Andohahela (parcel 2) within the floristic zone known as the Mandrare Valley (Koechlin et al., 1974), located 7.5 km ENE of the village of Ha- zofotsy, 24°49.0'S, 46°36.6'E, at 120 m. Ten lin- ear transects, each measuring 50 X 2 m (=0.01 ha) were installed. Trees, shrubs, and lianas with dbh >2.5 cm and rooted within the transect were censused, and their height was estimated from the ground (Gentry, 1982). Voucher specimens offer- tile plants were collected for later species confir- mation in the herbaria of PBZT and DRFP, An- tananarivo. Two duplicates of nonfertile plants were also collected for comparison. Species of trees with dbh <2.5 cm, understory trees, and her- baceous plants were identified, and if fertile, a herbarium specimen was preserved. General col- RAKOTOMALAZA & MESSMER: VEGETATION 53 lecting was also conducted within the surround- ings of the transects, as well as within the gallery forest along the Mananara River, in order to es- timate the diversity of the local flora and to pro- vide a list of species occurring in the parcel. Data Analysis Although different survey techniques were em- ployed in parcels 1 and 2, in many cases the vari- ables measured for each of these techniques were identical. In cases where differences occurred, these are explained below. The plot and transect studies involved several variables recorded in the field: the number of trees, their dbh and estimated height, and the number of species in each family. The relation between dbh and height was analyzed (for plots only). Furthermore, the biovolume (for plots only), density, basal area, relative dominance at specific and familial level, species diversity, and species frequency were calculated. Species accu- mulation curves as a function of area sampled were also plotted. Distribution of dbh — The distribution of dbh classes was divided into increments of 5 cm for the plots and 2.5 cm for the transects. Height Distribution — Height distribution in intervals of 5 m was analyzed for each of the five plots as well as for the transects. Relation Between dbh and Height — The re- lation of these two parameters was selected be- cause they characterize the physiognomy of trop- ical forest. According to Halle et al. (1978), the linear relationship between these two variables is h = 100 dbh. This model was proposed by Halle et al. (1978) based on data from a Guyanese forest. A natural divergence from the proportion 1:100 is possible, particularly in Malagasy forests with a different flora and structure, and it could be the result of local natural factors rather than human perturba- tion. This relation provides a potentially useful comparison among the five plots along the ele- vational transect in parcel 1. Estimation of Biovolume — Biovolume, ex- pressed in m3, corresponds to an estimation of the wood volume of trees in a given area. In this study, it is calculated for trees with a dbh 5:10 cm in the five 1 ha plots. This value was calcu- lated in two different ways, using the biomass cal- culations of Devineau (1984) and those of Lie- berman et al. (1996). We have presented our re- sults using both formulas, hereafter termed bio- volume a and biovolume b (respectively), in order to allow broader comparisons with other pub- lished works. Density — The density measure, used in both the plots and the transects, estimates the number of individuals of a given taxon found in a fixed area (Curtis & Mcintosh, 1950). Basal Area and Relative Dominance — Using dbh, the basal area can be calculated for each tree for both the plots and transects (CTE 1989). The dominance value for a given taxon provides in- formation on the importance of that taxon by con- sidering the area it occupies over the total area studied (Mori et al., 1983). Frequency — Each plot was divided into a fixed number of sampling units, whereas each of the 10 linear transects measured in parcel 2 represented a single sampling unit. The frequency of a taxon corresponds to the number of units in which it is found. This is usually calculated for species (Cur- tis & Mcintosh, 1950). Relative Species Diversity per Family — The species diversity per family is the number of spe- cies counted of a given family in a delimited area (Curtis & Mcintosh, 1950). Species-Area Curve — A species-area curve represents the cumulative number of species (in- cluding morphospecies) previously unrecorded at the site in relation to increasing sample area. If the curve reaches a plateau, then the studied area is representative of the flora within the local hab- itat. Further, based on the shape of the curve, it is also possible to estimate how homogeneous the site is with regard to species representation (Hill et al., 1994). Family Importance Value — A family impor- tance value (FIV) index establishes the impor- tance of a family with respect to others, consid- ering three factors: the abundance (density), the basal area (dominance), and the species diversity (Mori et al., 1983). Importance Value Index — A species impor- tance value index (IVI) allows an evaluation of diversity at the species level, considering three factors: the abundance (density), the basal area (dominance), and the frequency (Curtis & Mcin- tosh, 1950). This index has been criticized by Spurr (1964) and Silva (1980), who have pointed out that in tropical forests relative frequency and relative density are often nearly equal, and there- fore the index does not give enough weight to tree size. However, Schulz (1960, pp. 160-161) dem- onstrated the problem of placing too much value 54 FIELDIANA: ZOOLOGY Table 4-1. Measured general parameters for 1 ha plots at five different sites in parcel 1 of the RNI d'Andohahela. Number of trees Bio- Bio- having Basal volume volume Number Number Altitude adbh Mean dbh Mean height area a b of of Plot (m) >10 cm (cm) (m) (m2) (m3) (m3) families species 1 440 739 20.5 ± 0.9 14.4 ± 0.6 34.1 270 160 31 121 2 840 880 20.9 ± 1.6 13.5 ± 0.9 43.2 430 260 34 146 3 1150 1,216 18.8 ± 0.6 13.8 ± 0.3 43.8 440 270 38 126 4 1550 675 24.4 ± 1.9 14.3 ± 0.4 63.8 740 450 26 65 5 1875 1,365 21.8 ± 1.0 10.1 ± 0.2 65.9 430 260 23 50 on basal area as an indicator of importance. In this report, we employ the IVI to allow broader com- parisons. A Diversity Index, the Shannon-Weaver In- dex— Species diversity, a characteristic of biolog- ical organization at the community level, provides a measure of community structure. In this survey, we chose to use the Shannon-Weaver diversity in- dex, H' (Shannon & Weaver, 1949). Indeed, if measures of species abundance other than counts are used, only the Shannon-Weaver formula is available (Greig-Smith, 1983). Such a sample will not contain representatives of each species in the entire community. However, the absence of rare species has little effect on the value of H' (Brower et al., 1990). A Similarity Index, the Horn Index — After tabulating the species composition of each of the five 1 ha plots studied, one of several approaches can be used to assess how similar they are to each other. We chose the index of community similarity proposed by Horn, which is derived from the Shannon-Weaver diversity index (Brower et al., 1990). The Horn index, R^, was calculated at the species and family levels for FIV and IVI figures derived from the plots (Goldsmith et al., 1986). Results Part I: The Humid Forest of Parcel 1 Data Analysis Structural Parameters Table 4-1 summarizes the number of trees re- corded in each plot having a dbh ^10 cm, the mean dbh, the mean height, the total basal area, RAKOTOMALAZA & MESSMER: VEGETATION the values for biovolume a and biovolume b, the number of families, and the number of species. Number of Trees and Basal Area — The num- ber of trees increased along the elevation gradient, with the exception at 1550 m, where the lowest density was noted. The lowest total basal area was observed in plot 1 at 440 m. Plot 3 at 1 150 m had about 30% more trees than plot 2 at 840 m, al- though the basal areas in both plots were nearly equivalent. The number of trees was about 100% greater at 1875 m than at 1550 m, but the basal areas in both of these plots were similar. Even though the number of trees was lower at 1550 m than in the adjacent plots, the total basal area was much higher at 1550 m and 1875 m than at 1150 m. Thus, the number of trees and the basal area must be considered together. Distribution of dbh — Table 4-2 presents in- formation on the distribution of dbh for each plot; this is graphically presented in Figure 4-1. The majority of trees in the plots had dbh <20 cm, with percentages of 63.1% at 440 m, 65.4% at 840 m, 69.1% at 1 150 m, 54.9% at 1550 m, and 54.7% at 1875 m. Few individuals had dbh values ^50 cm. The principal families with individuals having large dbh values were Elaeocarpaceae, Lauraceae, Monimiaceae, Moraceae, Myrtaceae, and Sapo- taceae. Within each plot the distribution of dbh had an inverted J-shape curve, which is charac- teristic of primary forest (Mori & Boom, 1987). Height Distribution — Table 4-3 presents data on the distribution of tree height measured in each plot. There were few trees exceeding 25 m in height and almost none taller than 30 m. The only trees having a dbh >10 cm and a height <10 m were found at 1 875 m; in this elevational zone the mean height was distinctly lower than in all the other plots (Table 4-1). By setting the lower limit for dbh measurements at 10 cm, a large propor- tion of the trees <10 m in height were not sam- 55 2 iri CO On — « CM — < iri cm ^^^^^ Tf cm «-* CM m — rt — n — o o m IT) o ri ^ ^^^^ ON ^c r- u-> ri — CM — ri cs -T rs _ r- o CM n en ££ g* 8* S? g* — < oo oo no in rn en CM ci ^f ^ e* e* e^ s* NO On >r> — ■< r-; in cS rn no oo g* g< g< g* g« — > (N (N ro ^t g^ g^ ^g^ ^r m (N CM CM en CM ro NO m — ' — ^ — ' — o> i/-, _ r> » r> o ^r; O o CM g^g^ g« gjgi ir> CM O NO ON in oo r- ^t r*> — , — i 100 DBH (cm) Fig. 4-1. Distribution of dbh measurements taken in the five 1 ha study plots in parcel 1 of the RNI d'Andohahela. study plots, although a few families and a few species represented the first 50% of density in each plot. The percentage of species that repre- sented half of the total density is lowest in the 1550 m plot (9%) and highest in the 1875 m plot (18%). These patterns are correlated with the high values of FIV and I VI in the 1550 m plot (Ap- pendices 4-3 and 4-4). In parcel 1 of the RNI d'Andohahela Lauraceae was well represented in every elevational zone on the basis of abundance, basal area, and species diversity. This family, as well as the Myrtaceae, was dominant at the higher altitudinal levels. Weinmannia (Cunoniaceae), a genus typical of high altitudes, was common in the 1550 and 1875 m plots; a few individuals were sampled as low as 840 m. Elaeocarpaceae, represented by the enormous Sloanea rhodantha, was an important family with regard to basal area. Even if the spe- cies diversity of Elaeocarpaceae was low, this family includes most of the emergents, especially at higher elevations. Chrysophyllum boivinianum was mainly observed at 440 m; individuals were still present at 840 m. This tree is typical of low- land forests, where it is often one of the largest emergents. Tambourissa spp. (Monimiaceae) were very common from 440 to 1 150 m. At 1550 and 1875 m Monimiaceae were uncommon; only six small individuals of Tambourissa were en- countered at 1550 m, and species of Ephippiandra are seen at the higher altitudes. Many Oncoste- mum spp. (Myrsinaceae) were counted at 440 m. This is one of the dominant trees of lowland forest but becomes less frequent with increasing eleva- tion. Many individuals of Dombeya (Sterculi- aceae) and Macaranga (Euphorbiaceae) were Plot Table 4-3. Distribution of height classes for the measured trees in each plot. 0-5 5-10 10-15 Height i in i 15-20 20-25 25-30 >30 440 m 0 (0.0%) 0 (0.0%) 260 (35.2%) 163 (22.1%) 71 (9.6%) 7 (1.0%) 0 (0.0%) 840 m 0 (0.0%) 0 (0.0%) 309 (36.3%) 233 (27.4%) 80 (9.4%) 43 (5.1%) 1 (0.1%) 1150 m 0 (0.0%) 0 (0.0%) 446 (35.9%) 307 (24.7%) 157 (12.6%) 20(1.6%) 0 (0.0%) 1550 m 0 (0.0%) 0 (0.0%) 180 (26.7%) 215 (31.9%) 77 (11.4%) 26 (3.9%) 0 (0.0%) 1875 m 20 (1.5%) 523 (38.3%) 745 (54.7%) 75 (5.5%) 0 (0.0%) 0 (0.0%) 0 (0.0%) Note: The total number of individuals may be lower than that presented in Table 4-1 because of the loss of some values. RAKOTOMALAZA & MESSMER: VEGETATION 57 Table 4-4. Relationship between height and dbh, with the respective r2 value. Plot Altitude Linear regression equation 1 440 m h = 55.8 dbh -0.31 2 840 m h = 50.8 dbh -0.58 3 1150 m h = 60.4 dbh -0.51 4 1550 m h = 33.7 dbh -2.75 5 1875 m h = 40.0 dbh -1.10 present in the 1150 and 1550 m plots. These two genera contain colonizing species of trees, and in each of these plots there was an open area in which these trees were concentrated. Area-Species Curve — The species accumula- tion curves for all five 1 ha plots are presented in Figure 4-4. The curves for the three lowest ele- vations, 440, 840, and 1 150 m, reflect a relatively homogeneous vegetation within each plot. The curves are gradual, although the study areas were not large enough to include the majority of spe- cies in the immediate local habitat. In contrast, the number of species occurring in the 1550 and 1875 m plots is distinctly smaller and a near plateau is reached in the species accumulation curves. This presumably also reflects a reduced botanical di- versity as compared to the three lower elevational zones. In general it would appear that the species sampled in the 1550 and 1875 m 1 ha plots are probably more representative of the local floral communities within these zones than the plots at lower elevation. A single 1 ha plot, or perhaps a plot as small as 0.5 ha, at upper elevations may have been sufficient to characterize the local veg- etational communities. As a final point to this sec- tion, we note that species-area curves may vary depending on the site chosen for the plot, owing to the spatial distribution of species (Fangliang et al., 1996). In some of our plots it was clear that certain species showed a clump distribution. The Shannon-Weaver Diversity Index — Within the 1 ha plots sampled, the highest Shan- non-Weaver diversity index value was at 840 m, closely followed by that at the 1150m plot, and finally by that at the 440 m plot (Table 4-8). The lowest diversity index values were observed at the 1550 and 1875 m plots. There was no clear dif- ference between the various index values for the 440 and 840 m plots. Furthermore, because the ratios of the number of species to the number of individuals were similar between these two ele- vational zones, the number of trees sampled has to be considered. The high diversity index at 1 150 and 1875 m is explained by the high number of sampled trees. Indeed, when considering the ratio Table 4-5. Listing of the five most important families for four parameters in each 1 ha plot. Plot 1, 440 m Plot 2, 840 m Plot 3, 1150 m Plot 4, 1550 m Plot 5, 1875 m Relative density Rubiaceae Clusiaceae Lauraceae Myrsinaceae Monimiaceae Monimiaceae Moraceae Myrtaceae Euphorbiaceae Araliaceae Myrtaceae Lauraceae Sterculiaceae Moraceae Monimiaceae Euphorbiaceae Lauraceae Sterculiaceae Monimiaceae Cunoniaceae Araliaceae Lauraceae Myrtaceae Clusiaceae Flacourtiaceae Relative dominance Elaeocarpaceae Clusiaceae Sapotaceae Rubiaceae Lauraceae Elaeocarpaceae Moraceae Myrtaceae Lauraceae Flacourtiaceae Lauraceae Sterculiaceae Myrtaceae Monimiaceae Moraceae Elaeocarpaceae Monimiaceae Lauraceae Sterculiaceae Euphorbiaceae Araliaceae Lauraceae Cunoniaceae Elaeocarpaceae Flacourtiaceae Relative species diversity Rubiaceae Lauraceae Clusiaceae Myrsinaceae Euphorbiaceae Moraceae Clusiaceae Rubiaceae Lauraceae Monimiaceae Rubiaceae Euphorbiaceae Lauraceae Myrtaceae Moraceae Lauraceae Sterculiaceae Rubiaceae Monimiaceae Clusiaceae Lauraceae Myrtaceae Rubiaceae Araliaceae Cunoniaceae FIV Rubiaceae Clusiaceae Lauraceae Elaeocarpaceae Myrsinaceae Moraceae Elaeocarpaceae Monimiaceae Myrtaceae Rubiaceae Lauraceae Myrtaceae Sterculiaceae Moraceae Rubiaceae Elaeocarpaceae Lauraceae Sterculiaceae Euphorbiaceae Monimiaceae Lauraceae Araliaceae Myrtaceae Cunoniaceae Elaeocarpaceae FIV, family importance value (index). 58 FIELDIANA: ZOOLOGY 440 in ■ Relative specific diversity D Relative dominance I Relative density Family 840 m ■ Relative specific diversity D Relative dominance I Relative density Family 1150 m 1550 m ■ Relative specific diversity D Relative dominance □ Relative density ■ Relative specific diversity □ Relative dominance □ Relative density Family 1875 m Qp ■ Relative specific diversity D Relative dominance □ Relative density P P I E CQ (J 3 d Family Fig. 4-2. Relative density, relative dominance, relative species diversity, and FIV for the five 1 ha study plots in parcel 1 of the RNI d' Andohahela. Each plot is presented separately. Abbrevations refer to families: ARL, Araliaceae; CEL, Celastraceae; CLU, Clusiaceae; CUN, Cunoniaceae; ELC, Elaeocarpaceae; EUR Euphorbiaceae; FLC, Fla- courtiaceae; LAU, Lauraceae; MEL, Meliaceae; MNM, Monimiaceae; MOR, Moraceae; MRS, Myrsinaceae; MRT, Myrtaceae; RUB, Rubiaceae; SAR Sapindaceae; SPT, Sapotaceae; and STR, Sterculiaceae. RAKOTOMALAZA & MESSMER: VEGETATION 59 r- fsi — o\ 9-o.p.P •S| 3 | .2 <-i .5 ^ O R 2, b> H -5 2 3 a C 3 3. 0. O CQ Q, r^ -R -2 H -R BE5 11 e,5 J! X Q 3 si y (NOD. Q.— x :&~ r ea s a s 5 Sf o. _ D. c3 •~ D. U s I " g C r ^ £ 2 9- r* 5-R^. 5 * ^ -R "a 3 fr I § ** ^ a u: 3 S3 S -3 - 3z R >. S3 c^ ~3 3 CI ^~ 3 § p. 0 5 S ° a g ^a: g >s 2 5 2^ R 3 S3 R 2 jfSq ■a 5 •< a; "3 c S _ D. - 3 - o,M Q. 3,_ x o "» Si Q.-T N S3 «, U ^ *> T 5 a a 5 ••= 15 s " G 3 3 i 3 3 i. s O -3 r ~ >3; 53^; -3 g co ' ^ £5 2 at f5 E <5"t2 O ^ 3 3 „ a -a slhg 2,2 ^2 5r c .. 3 o • 2 3 5 C R H R IcST R S3 fS^C •* >: n r a; jo S^ <^~, tj- ir, vc 3 '5 * - 1 .3 1 a. a. R -R ^ * ^ -2 ^2 ^ 3 13 ^ 3 , ,tm r ^ o 9> -I tx C"3 ^ p- S * *s 3 3 _ .J3 0 u ■? 3 C C3 C3 '*■ ■3 >> s u 0 a ^ 3 U U S k. ■S -5 3 3 r ^ R R b e 3 3 Q£S2S5 > 3 — •«■ _R § Q. R * 3 n 3 « -2 > CO g .3 Q „ ^".2 in a.— *. — — tf\ _ a. d. n. ..3 yi s. 5T_ U >..3 2i ^ 3 5 3 3. 2 u a, 6, ""■ S J2 R fi a a. R co (N P x 3 p 2 -R P •£ o >■ 2 o R^3s- -a 2 *> "&, S5 ia & S>- g "5. s- ■" c -3 QQ U 00 ~: (N . ^O D. — ^ U Q. -. SJ y; 2 ° 3 3 S 3 c H 3 d. § § R 3 ^ =- S ^ 3 § S3 a .— ss 3 -R 3 g c >^ .a 3 3 3 ~ - •; 3 5 3 3 J 3 J- j; BS ■sT >-. -3 "^«3 9 ^3 O, C-a|3^2^ •3 3 S3 -3 *i) 2 -3 S - 3 ■3 R 3 "_ _ ;3 a v>^^;33,_s3j3 CO C^Sa.r^C!li.O^ - 3 33' •2 -2 .3 3 R-3 a -3 ?c j- r 3 a a. _ n. CA g j. -A 3 -2 3 2 .§*?.§ 2 S3 3 C -R -3 R 4) ^ 3 3. R > u R 3 3 -~ £ S R-S £,-R 3 3 3 3 S ir ^ ,3 tClJ^sU OS 60 FIELDIANA: ZOOLOGY •2 -2 CU d; (5 =S *?r£$ ' a •" 3 D. > »> > b8 H "* ft. 2 H ?\ i^ %, £. oa =; . u o 'a %> U O -S 5 5 o. § g •* 2 « 'c .a "c ?J _r? ? £? 9 2 .■? — Q. C -r. S=. ~ S < 3 z ■•- r* O 3 2 ft. S O to >■ =3 a 5 £ 5 a 0 >ft -2 -ft o Z 5 o£ 2 5 "2 c 3 tn P" 3 ^> Q..~ ^ ^ '" J! . 5 ^ ^Qjt 8. of the number of species to the number of indi- viduals, the 1150 m plot is clearly less diverse than the plots at 440 and 840 m, but it is very similar to that at 1550 m. The plot below the sum- mit at 1 875 m had a proportionately lower diver- sity than the other plots. The Shannon-Weaver diversity index values for the five plots are relatively very low, falling be- tween 1 .49 and 1 .86. This is due to the dominance of some species in each plot. For example, the 1550 m plot had the lowest Shannon- Weaver di- versity index value, and Macaranga sp2 repre- sented 14.07% of the total tree density. Indeed, the Shannon-Weaver diversity index strongly weighs the number of species and the relative density of each species. The low diversity index is the result of the small number of species that account for over 50% of the number of trees in the various size classes; individual species and family dominance within low-diversity forests is high. The Horn Similarity Index — Dendrograms showing the degree of similarity in family and species composition between each plot on the ba- sis of the FIV and IVI are presented in Figure 4- 5. These dendrograms corroborate our earlier in- terpretations of the fioristic relationships between the various elevational zones. Discussion The eastern Malagasy humid forest contains a number of distinct vegetational communities that vary as a function of altitude. According to Hum- bert and Cours Darne (1965) and Koechlin et al. (1974), lowland rain forest reaches up to 800 m and is characterized by the presence of Sloanea rhodantha, Canarium madagascariense, Myristi- caceae, and Anthostema (Euphorbiaceae). The mid-altitude moist forest, occurring between 800 and 1800 m, is distinguished by Weinmannia (Cu- noniaceae), Tambourissa (Monimiaceae), Sym- phonia (Clusiaceae), Ravensara, and Ocotea (Lauraceae), as well as Canarium madagascar- iense. Finally, they described a stratum between 1800 and 2000 m composed of moist montane forest. Podocarpaceae and Weinmannia are two of the taxa characterizing this latter formation. On the basis of the above classification, plots 1 (440 m) and 2 (840 m) would belong to the low- land rain forest; the upper elevational limit of this zone is subject to some variation and may reach RAKOTOMALAZA & MESSMER: VEGETATION 61 440 m 35 30 4 25 20 15 4- 10 5 + 0 ■ Relative frequency D Relative dominance D Relative density Pn 31 32 33 anna 35 36 Species 37 38 39 40 840 m 35 T 30 „ 25 ■■ 3 1 20 -■ > I '5 ■■ " 10 5 4- 0 ■ Relative frequency D Relative dominance □ Relative density 11 31 32 33 34 35 36 37 38 39 40 Species 1150 m 1550 m 35 T 30 25 4 •3 20 > 315 10 5 ■ Relative frequency □ Relative dominance 13 Relative density 0 + Lid | m 31 32 PI 35 30 25 V 3 1 20 s '315 & 10 ■ Relative frequency P Relative dominance □ Relative density 35 36 Species 40 31 32 33 34 35 36 Species 37 38 39 40 1875 m 35 T 30 25 3 | 20 3 15 4- 10 5 | 0 I ■ Relative frequency D Relative dominance □ Relative density p n _L I 31 32 33 34 35 36 Species 37 38 39 40 Fig. 4-3. Relative density, relative dominance, relative frequency, and IVI for the five 1 ha study plots in parcel 1 of the RNI d'Andohahela. Each plot is presented separately. The numbers correspond to those presented in Table 4-6. 62 FIELDIANA: ZOOLOGY Table 4-7. Number of species (n) with respective percentage (%) representing the indicated percentage (25%, 50%, 75%, 95%) of the total relative density. Altitude (m) Relative density 25% 50% 75% 95% Plot n % n % n % n % 1 440 5 4 12 10 34 28 93 78 2 840 8 5 20 14 46 32 107 73 3 1150 4 3 13 10 30 24 81 64 4 1550 2 3 6 9 16 25 41 63 5 1875 3 6 9 18 19 38 37 74 higher elevations in some regions of the island (Lowry et al., 1997). Plots 3 (1150 m) and 4 (1550 m) fall within the range of mid-altitude moist forest, and plot 5, at 1875 m, is classified as moist montane forest. The taxonomic results obtained in the plots at 440 and 840 m are similar to one another (Table 4-5; Fig. 4-5) and corrob- orate the descriptions made by Humbert and Cours Darne (1965) and Koechlin et al. (1974). At 1550 m in the moist montane forest formation, Weinmannia was abundant, whereas it was com- pletely absent in the 1150 m plot. Tambourissa was common at 1150m and appeared to reach its upper limit at about 1200 m. Few Symphonia were counted in the plots at 1150 and 1550 m, but this genus was often noted during the general collecting in each of these zones. Ocotea and Rav- ensara were dominant at these two elevational zones (Table 4-5). Finally, on the basis of eleva- tion, the plot at 1 875 m would fall into the upper zone of the moist montane forest. This plot had many similarities to those at lower elevations, but the presence of some taxa, particularly Ericaceae, attests that this site should be assigned to moist montane forest. This is particularly true at the spe- cies level, whereas at the family level it is very similar to the plot at 1550 m (Fig. 4-5). One surprising observation was the absence or rarity of two often dominant families in Malagasy humid forest, Podocarpaceae and Burseraceae (Canarium spp.). No Podocarpus was found in any of the transect zones studied. This genus has been reported previously from Beampigaratra (Laubenfels, 1972), another region of the Ano- syenne Mountains. Furthermore, at other sites such as the eastern slopes of the RNI — fN Area(m2) Fig. 4-4. Species accumulation curves for the 1 ha plots in five different elevational zones in parcel 1 of the RNI d'Andohahela. RAKOTOMALAZA & MESSMER: VEGETATION 63 Table 4-8. H' (Shannon-Weaver diversity index value) for each plot, using the number of individuals in each taxon. Shannon-Weaver Species : individual Altitude diversity index Plot (m) Species Families ratios 1 440 1.76 1.32 0.16 2 840 1.86 1.34 0.17 3 1150 1.77 1.29 0.11 4 1550 1.49 1.18 0.10 5 1875 1.52 1.21 0.04 d'Andringitra, Podocarpus was one of the domi- nant genera at 1600 m (Lewis et al., 1996). Sim- ilarly, in parcel 1 of the RNI d'Andohahela only four individuals of Canarium (Burseraceae) were counted in the plots, and even during general col- lecting members of this family were rare. Nor- mally, Canarium is present in most Malagasy hu- mid forests. This genus is cited by Humbert and Cours Darne (1965) as one of the characteristic emergent trees up to 1 800 m. Canarium occurs as a large forest emergent, especially between 700 and 1000 m, in the RNI d'Andringitra (Lewis et al., 1996). In the RS de Manongarivo and the RNI de Marojejy (Rakotomalaza & Messmer, pers. ob- serv.) it is also a conspicuous member of the local plant community. The latitudinal position of the RNI d'Andohahela could be a factor to explain this deviation from the accepted classification of plant zone indica- tors. The transect studied is situated south of the Tropic of Capricorn. The island of Madagascar spans 14 degrees of latitude, and climatic differ- ences between the north and the south along the eastern coast have already been discussed by Donque (1972). According to Humbert and Cours Darne (1965) and Koechlin et al. (1974), their bo- tanical classification with regard to vegetation composition or elevational is relatively constant. However, it appears that they did not consider a latitudinal gradient. One possibility would be that at such high latitudes as parcel 1 , taxa descend to lower elevations. For example, Myristicaceae and Anthostema (Euphorbiaceae) are absent from the transect, but genera belonging to Myristicaceae were recorded at the same latitude at lower ele- vations outside of the reserve (Dumetz, 1993). Furthermore, Canarium might occur at elevations lower than 440 m, the lower limit of our parcel 1 study. This hypothesis is supported by the pres- ence of Chrysophyllum boivinianum, Ilex mitis, o.ooo l- 2- 3- 5- 4- 1.000 0.000 2- 1.000 Fig. 4-5. Dendrograms (SYSTAT, version 2) using FIV data (top) and IVI data (bottom) for five different 1 ha plots in parcel 1 of the RNI d'Andohahela. The numbers along the left margin of the dendrograms represent the different plots. 64 FIELDIANA: ZOOLOGY and Sloanea rhodantha among the dominant spe- cies at 440 m, because these are all typically mid- elevation species and not known from low alti- tudes in forests further north (Schatz, 1994; Ra- bevohitra et al., 1996). In addition, few trees of the Proteaceae family, as represented by Dilobeia thouarsii, were found in the 440 and 840 m plots. This species was absent at higher altitudes, al- though it has been recorded up to 1800 m (Koech- lin et al., 1974). These observations support the hypothesis that the vegetation communities in parcel 1 of the RNI d'Andohahela exhibit a downward elevational shift. A few structural parameters also confirm this supposition. For example, canopy height was lower in plots at 440, 840, and 1150 m than is typical for these elevational zones elsewhere in Madagascar (Koechlin et al., 1974). Nevertheless, the 1550 and 1875 m plots had high values for density, basal area, and height. A decrease in tree height and dbh with altitude is expected (Koechlin et al., 1974). According to these two structural parameters, however, the plot at 1550 m is re- markable in comparison with the other plots; the number of trees having a dbh >50 cm was much higher than in the other plots, and it had the low- est value of trees with a diameter of <20 cm. Even if the total density of the 1550 m plot was only nearly half that of the plot at 1 1 50 m, the value of basal area had increased by nearly 50%. The distinctive aspects of the 1550 m plot were visually apparent during the field study. The values of the structural data for parcel 1 of the RNI d'Andohahela are similar to those for other sites in Madagascar (Table 4-9). A broader comparison of the RNI d'Andohahela to forests studied in South America, Africa, and Asia indi- cates that the basal area values and the number of trees with stems >10 cm dbh are much higher along the elevational gradient in parcel 1 than at those other sites (Table 4-9). Family and species diversity indices are also much lower in the An- dohahela plots than those calculated for forests in other parts of the tropics (Table 4-9). Thus, even if the numbers of families and species are similar between parcel 1 of the RNI d'Andohahela and those other forests, the number of trees for each of these taxonomical units is probably consider- ably different. The Shannon- Weaver diversity in- dex considers the number of species and the num- ber of individuals of each species. If a few species of trees have a much higher density than other sympatric species, the habitat is considered less diverse than one with the same number of species in which none is dominant. Along the transect in the RNI d'Andohahela, the relative densities are unbalanced (Table 4-7). The habitats are thus less diverse in the sense of the Shannon-Weaver di- versity index compared to other tropical forests. Monimiaceae is a distinctive family of the Mal- agasy forests and not to the forests in the Neo- tropics, Asia, or Africa. Apart from the Diptero- carpaceae, which is a widespread and particular aspect of the forests of southeastern Asia, other dominant families occurring in this region and Madagascar are the same. They include, for ex- ample, Lauraceae, Myrtaceae, Elaeocarpaceae, and Clusiaceae (Gentry, 1988; Ohsawa, 1991; Ki- tayama, 1992; Pendry et al., 1997). Lauraceae is one of the dominant families, especially above 800 m, in most of the Malagasy humid forest (Schatz, 1994; Lewis et al., 1996; Rabevohitra et al., 1996), as well as in many other humid forests in other tropical regions (Johnston & Gillman, 1995; Kitayama, 1992; Pendry et al., 1997; Mori et al., 1983; Gentry, 1988; Ohsawa, 1991). Legumes are virtually always the dominant family in Neotropical and African lowland pri- mary forest, and their dominance in these regions is parallel to that of Dipterocarpaceae of south- eastern Asia (Mori et al., 1983; Gentry, 1988; Lie- berman et al., 1996). The major exception to this pattern is that legumes appear to be poorly rep- resented among the woody plants with a dbh 5:10 cm in Malagasy humid forests (Schatz, 1994; Lewis et al., 1996; Rabevohitra et al., 1996). The scarcity of the Leguminosae is noticeable in par- cel 1 of the RNI d'Andohahela; in the entire tran- sect studied, only 35 trees out of a total of 4,875 (0.7%) belonged to this family. Part II: The Spiny Forest of Parcel 2 Although several botanists, including pioneers such as Alluaud and Poisson (Poisson, 1912), have collected in the south of Madagascar, no quantitative data are available on spiny forest structure and floristic diversity; the only known exception is the study of Sussman and Rakotozafy (1994) in the RS de Beza Mahafaly. Data Analysis Structural Parameters Distribution of dbh — Of the 607 individuals with a dbh >2.5 cm, 80% measured between 2.5 RAKOTOMALAZA & MESSMER: VEGETATION 65 8 fc -P - H I £ j -a . = £ 2 3 a a £ > 0 5 cc "^ ^ *■ £ o 3 z ca C8 % S a *§ -a © i. 3 S- W M ^ Si Is E o U ^^■^^M^tt^ o o o —I en in n o -*t On — ■UT*00ir)O*C(NOh OOOflVl — f> OvO(Nt^OOOOOO(Nroooni/^inuoininuoOOinOOOOOOOO~~ — — — ~ — ~ — ~ — — 'OOOOOOOOOOOO JJJJJJJJJJJJ t~~ ON « x; o ^ C- ea "3 S "3 NO ^ ON <^> >^ , '— o <_ n 0> *J CJ ON •c e ' — " c ofi ea , OS ca 2 2 c ca ea £ k 2 '»S ea ca ca «5 ca ca ti rt >> E J2 o c 5 o * dS 2-2 ca ca E > o» o ea IS S "ea o JJ , o E ° o OQ S PQ DQ aa °o 6 o *— 2~ O M Q-s 5- PQ? ea ca 3 .fl .C [l o s ca TO PQ 2 o a* 5 S ffl g -• op =y PQ o o ._, . ^^o > «- s ^o o o O 'O „ o ca 02 ca .S U OS ^ M.2 o ca j3 - ic d < < PQ 66 FIELDIANA: ZOOLOGY Z ~~ - y ^ & > « a ■ J= £ = — s2 X! J* z •- so V o S a z — 1 - S J3 - 09 ™ a BQ £ £ c -= - j: E -h '> "O o u II ■^■■^■•^■Ttmr^rom TTror4 — sOOnoooo l-inwt-iNNvOh OOOOOQOQ § § OO^Oi^OiOO — — o o J J q£ SO 5 5 12 N w < and 10 cm, whereas only 2% measured >30 cm (Table 4-10; Fig. 4-6). These latter trees belonged to five species: Operculicarya decaryi, Alluaudia ascendens, A. procera, Commiphora aprevalii, and Tetrapterocarpon geayi. These species and other members of these genera are emergent trees. The only exception is O. decaryi, which is one of the region's characteristic bottle trees. The mean dbh of the measured trees was 7.42 ± 0.52 cm. Hkight Distribution — Of the 607 individual trees with a dbh >2.5 cm, 72% were between 0 and 5 m in height. Only 3.5% of the trees sampled had a height >10 m (Table 4-11; Fig. 4-7) and are represented by four species: Alluaudia ascen- dens, A. procera, Commiphora marchandii, and Tetrapterocarpon geayi. As noted above, the small trees making up the middle stratum had a height of about 2-4 m, and the canopy trees reached up to 5 or 6 m. The mean height of sam- pled trees was 4.18 ± 0.16 m. Floristic Parameters All of the floristic parameters are presented in Appendices 4-5 and 4-6. Density — A total of 10 linear transects were established in which 540 plants with dbh >2.5 cm were identified, belonging to 28 families and 78 species. Four families accounted for 50% of the individuals: Burseraceae (19.1%), Euphorbiaceae (13.3%), Didiereaceae (8.7%), and Ebenaceae (7.6%). The commonest species were Gyrocarpus americanus (Hernandiaceae); Commiphora hum- bertii, C. aprevali, and C. brevicalyx (Bursera- ceae); Alluaudia procera (Didiereaceae); Euphor- bia intisy (Euphorbiaceae); and Diospyros hum- bertiana (Ebenaceae). Basal Area and Relative Dominance — The total basal area was 4.5 m?; 0.45% of the study area (1,000 m:) contained trees having a dbh ^2.5 cm. Half of the total basal area was occupied by only two families: Didiereaceae (two species of Alluaudia) accounted for 26.5% and Anacardi- aceae {Operculicarya decaryi) accounted for 22.2%. Burseraceae, as represented by Commi- phora humbertii, was the third most common family. Relative Species Diversity per Family — For the 0. 1 ha study area, 78 species of trees, shrubs, and lianas with dbh >2.5 cm were sampled. Fa- baceae was the most diverse family, with 10 spe- cies, followed by Euphorbiaceae. Didiereaceae, RAKOTOMALAZA & MESSMER: VEGETATION 67 Table 4-10. Distribution of dbh for the measured trees in the 10 transects in parcel 2 of the RNI d'Andohahela. dbh (cm) 2.5-5 5-10 10-15 15-20 20-25 25-30 >30 n(%) 237 (39.4%) 239 (39.8%) 72 (12.0%) 13 (2.2%) 18 (3.0%) 10(1.7%) 12 (2.0%) Note: The total number of individuals is lower than that presented in the text because of the loss of some values. Rubiaceae, and Burseraceae were represented by four, five, and six species, respectively. Frequency — Five species of Commiphora, two species of Alluaudia, Operculicarya decaryi, one species of Dichrostachys, Euphorbia oncoclada, and Diospyros humbertiana were the commonest censused trees and shrubs with a dbh 2^2.5. The remaining taxa found in the study area occurred with more or less the same frequencies. Family Importance Value (FIV) and Impor- tance Value Index (IVI) — When the three fac- tors of relative density, relative dominance, and relative species diversity are combined, the five most important families were Burseraceae, Didi- ereaceae, Euphorbiaceae, Anacardiaceae, and Fa- baceae. When the three factors of relative density, relative dominance, and relative frequency are considered, the five most important species also belonged to these same families. The Shannon-Weaver Diversity Index — The value of the Shannon- Weaver diversity index, H', using species density data is 1.17. This is a very low value that is due to the dominance of a few species. In contrast, species richness, with 78 dif- ferent species in 0.1 ha, is high. Discussion A total of 540 trees with dbh >2.5 cm were identified in the 0.1 ha study area. This included 78 species belonging to 28 families. When the lower limit of dbh was increased to >5 cm, the density of trees decreased to 364 per 0. 1 ha and comprised 69 species. The mean value of the Shannon-Weaver diversity index is very low for trees with a dbh >2.5 cm; it can be accounted for by the dominance of a few species and the rarity of most species. Thus, in this case, even though the diversity value is low, in actuality the study area had a diverse flora. The spiny forest of parcel 2 of the RNI d'Andohahela has a higher density and diversity than the spiny forest of Beza Mahafaly (Sussman & Rakotozafy, 1994). Because of differences in methodology only the following figures are com- parable to our data: dbh >2.5 cm in 0.1 ha, 440 individuals; and dbh >2.5 cm in 0.25 ha, 69 spe- cies from 25 families (Sussman & Rakotozafy, 1994). The most common families were the same at both sites. The spiny forest of RNI d'Andohahela has also a higher tree density than 100 j 90 - . 80 I 70 i | 60 - 3 50 - I 40- | 30- 20 - 10 u 0-5 5-10 10-15 15-20 20-25 25-30 D 30 DBH (cm) Fig. 4-6. Distribution of dbh measurements taken in the 10 transects in parcel 2 of the RNI d'Andohahela. 68 FIELDIANA: ZOOLOGY Table 4-11. Distribution of the height for the measured trees in the 10 transects in parcel 2 of the RNI d'Andohahela. Height (m) 0-2.5 2.5-5 5-7.5 7.5-10 10-12.5 12.5-15 >15 n(%) 57 (10.4%) 339 (61.9%) 115(21.0%) 18(3.3%) 15 (2.7%) 3 (0.5%) 1 (0.2%) Note: The total number of individuals is lower than that presented in the text because of the loss of some values. dry forests in continental Africa and the Neotrop- ics. Using the same transect method, Gentry (1993) found an average of 369 individuals >2.5 cm dbh/1,000 m2 in a series of Neotropical low- land dry forests and 361 individuals in dry forests in Africa. In the El Pechiche dry forest of Ecua- dor, 57 trees with dbh ^5 cm belonging to only 13 species were counted in 0.1 ha (Josse & Bal- slev, 1994). In western Ecuador, the total density of a 1 ha plot in a dry forest was 538 trees ^5 cm dbh, and the total species richness was 37, belonging to 22 families. In the Paleotropical dry forest, Acanthaceae, Asclepiadaceae, Euphorbi- aceae, and Tiliaceae were well represented (Tho- masson & Thomasson, 1991). Apart from the Eu- phorbiaceae, this is a different floristic composi- tion from the one observed in parcel 2 of RNI d'Andohahela. The gallery forest structure along the Mananara River is much like continental dry forest in Africa and the Neotropics (see Sussman & Rakotozafy, 1994 for review). Floristically, the Mananara Riv- er gallery forest is dominated by Tamarindus in- dica, Adina microcephala, Ficus marmorata, and Terminalia sp. The flora of parcel 2 belongs to the type Hum- bert and Cours Darne (1965) placed in the Southern Domain and named "Fourre a Didierea- ceae et Euphorbia." The annual rainfall in this region is generally less than 500 mm and the an- nual average temperature is 25°C, with a large daily variation in minimum and maximum tem- peratures. Parcel 2 is situated in the semiarid zone (Humbert & Cours Darne, 1965). In such harsh conditions, plants exhibit many adaptations to re- duce water loss, resulting in various types of con- vergent evolution: swollen roots and trunks (bottle trees); pachycauly; and thickened succulent, mi- crophyllia, thorn, and deciduous leaves in dry sea- son (Koechlin et al., 1974). Alluaudia ascendens (Didiereaceae) is confined to the thicket and spiny vegetation that occurs in this region (Koechlin et al., 1974), and some gen- era such as Adansonia za, Kalanchoe, Aloe, Eu- phorbia, Pachypodium, Ficus, and Uncarina are also well represented (Koechlin et al., 1974; Ap- D 15 12.5-15 <£ 10-12.5 2 7.5-10 5b .8 5-7.5 2.5-5 0-2.5 ■ C 10 20 30 40 50 60 70 80 90 100 % of individuals Fig. 4-7. Distribution of estimated heights in the 10 transects in parcel 2 of the RNI d'Andohahela. RAKOTOMALAZA & MESSMER: VEGETATION 69 pendix 4-6). Lianas belonging to families such as Apocynaceae, Asclepiadaceae, Cucurbitaceae, Passifloraceae, and Urticaceae are numerous in this forest, but they do not reach dbh measure- ments of >2.5 cm. The Southern Domain is certainly less diverse climatically, topographically, and floristically than the eastern humid regions. Perrier de la Bathie (1936) estimated that the Southern Domain con- tained about 336 species of plants. Even though a number of new species have been identified in this domain since his estimate (Phillipson, 1996), if we use that figure, then our 0.1 ha study site included nearly 25% of the local flora. Many plants are endemic either to the region or to Mad- agascar at the genus or species level. Some au- thors estimate that 80-85% of the flora in this region may be endemic (Koechlin et al., 1974). Didiereaceae is the only family endemic to south- ern Madagascar. The main threat to the spiny forests of Mada- gascar is goat and cattle overgrazing, which de- stroys the seedlings and saplings of the larger tree species. Along the Mananara River, in the vicinity of our study site, the forest is more ruderal due to human disturbance, and the gallery forest is de- graded. In the area of parcel 2 that we visited, no evidence was found for the conversion of forest into charcoal. well as providing valuable floristic data (Chate- lain, 1996). This method should be complemented by using small plots in order to obtain measures of basal area and biovolumes. The critical point is that standardization of rapid assessment meth- odologies is necessary in order to allow valid comparisons between study sites. Our quantitative and qualitative survey of the flora in parcels 1 and 2 of the RNI d'Andohahela provides results that do not correspond completely with those expected based on the published liter- ature. More research must be conducted to refine the effects of latitude on the structure and floristic composition of Malagasy humid forests, particu- larly with regard to elevational gradients. In ad- dition to the classical floristic division into do- mains and elevational zones, we strongly suspect that latitudinal divisions are applicable. Our study sites in the RNI d'Andohahela con- tained a rich and fascinating flora, with a high diversity of plants and types of floristic formation in a relatively small region. An extended study of the transition between the different vegetation, from the littoral forests of the eastern coast to the xerophytic thicket of the south, would allow a bet- ter understanding of the complexity of the Mala- gasy vegetation. Both Parcels: Conclusion The inventory methods applied to parcel 1 of RNI d'Andohahela gave a good measure of its structural and floristic composition and provided a large amount of useful data. Nevertheless, 1 ha plots are extremely time-consuming and not very practical for rapid assessment surveys. Further- more, in topographically abrupt zones, such as that at the summit of Pic Trafonaomby, insuffi- cient space exists for placing large plots. In con- trast, the quantitative and qualitative results ob- tained using Gentry-style line transects in parcel 2 of the reserve indicate that this methodology is well-adapted for rapid assessment studies. Two methodological improvements can be pro- posed for the type of fieldwork we conducted in the RNI d'Andohahela, particularly in humid for- ests: (1) use of smaller plots, combined with (2) linear surveys as proposed by Gautier et al. (1994). Linear surveys allow the quantification of plant cover and information on stratification as Acknowledgments We thank Steve Goodman for inviting us to take part in the survey of the RNI d'Andohahela. We are grateful to Felix Andriatsiferana and Nick Helme for their efficient help in the field. We ex- press our gratitude to the WWF project in Tolag- naro for providing logistical support during the months we spent in the forests of the RNI d'Andohahela. We thank George Schatz and Jean- ine Raharilala for their assistance with plant iden- tifications. Many thanks also go to the Pare Bo- tanique et Zoologique de Tsimbazaza herbarium director and staff for allowing access to their col- lection. Lucienne Wilme provided important ad- vice and help with statistical analyses. We thank the Missouri Botanical Garden and the Conser- vatoire et Jardin Botaniques de la Ville de Geneve for agreeing to have us participate to this inven- tory. Finally, we gratefully acknowledge the de- tailed comments and advice given by Steve Good- man and four anonymous reviewers on an earlier version of this chapter. 70 FIELDIANA: ZOOLOGY Literature Cited Balslev, H., J. Luteyn, B. Ollgaard, and L. B. Holm- Nielsen. 1987. Composition and structure of adjacent unflooded and floodplain forest in Amazonian Ecua- dor. Opera Botanica 92: 37-57. Brower, J. E., J. H. Zar, and C. N. Von Ende. 1990. Field and Laboratory Methods for General Ecology, 3rd ed. Brown Publishers. Dubuque, Iowa, 237 pp. Chatelain, C. 1996. Possibilites d'application de rimagerie satellitaire a haute resolution pour l'etude des transformations de la vegetation en Cote d'l voire forestiere. These de doctorat. Universite de Geneve, Departement de botanique et de biologie vegetale. Ge- neva, Switzerland. Centre Technique Forestier Tropical (CFT). 1989. Memento du forestier. Techniques rurales en Afrique. Ministere de la Cooperation et du Developpement: 244-245. Curtis, J. T, and R. P. McIntosh. 1950. The interre- lations of certain analytic and synthetic phytosocio- logical characters. Ecology, 31: 434-455. Devineau, J. L. 1984. Structure et dynamique de qu- elques forets tropophiles de l'ouest africain (Cote dTvoire). These de doctorat d'etat. Universite Pierre et Marie Curie, Paris IV. Paris, France. Donque, G. 1972. The climatology of Madagascar, pp. 87-144. in Battistini, R.. and G. Richard- Vindard, eds. Biogeography and Ecology in Madagascar. W. Junk. The Hague. Dumetz. N. 1993. Mise en evidence de l'heterogeneite floristique et structurale de la foret orientale de basse altitude a Madagascar. These de doctorat. Universite de Paris VI. Paris, France. Du Ply. 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 dTnformation et de Documentation Scientifique et Technique, An- tanarivo. Fangliang, H., P. Legendre. and J. V La Frankie. 1996. Spatial pattern of diversity in tropical rain for- est. Journal of Biogeography, 23: 57-74. Gautier, L., C. Chatelain, and R. Spichiger. 1994. Presentation of a releve method for vegetation studies based on fine-scale satellite imagery. XHIth AETFAT Congress, Malawi, pp. 1339-1350. Gentry. A. H. 1982. Patterns of neotropical plant spe- cies diversity. Evolution 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. In Goldblatt. P. L.. ed. Biological Relationships Be- tween Africa and South America. Yale University Press. New Haven. Goldsmith. F. B.. C. M. Harrison, and A. J. Morton. 1986. Description and analysis of vegetation, pp. 437-524. In Moore. P. D.. and S. B. Chapman, eds. Methods in Plant Ecology, 2nd ed. Blackwell Scien- tific Publications, Oxford. Greig-Smith, P. 1983. Quantitative Plant Ecology, 3rd ed. Blackwell Scientific Publications, Oxford. Halle. F. R. A. A. Oldeman, and P. B. Tomlinson. 1978. Tropical trees and forests: An architectural analysis. Springer- Verlag, Berlin. Hill. J. L.. P. J. Curran, and G. M. Foody. 1994. The effect of sampling on the species-area curve. Global Ecology and Biogeography Letters, 4: 97-106. Humbert, H., and G. Cours Darne. 1965. Notice de la carte. Madagascar. Extrait des travaux de la section scientifique et technique de lTnstitut Francais de Pon- dichery. Hors serie no. 6. Johnston. M., and M. Gillman. 1995. Tree population studies in low-diversity forests. Guyana. I. Floristic composition and stand structure. Biodiversity and Conservation, 4: 339-362. Josse, C, and H. Balslev. 1994. The composition and structure of a dry, semideciduous forest in western Ecuador. Nordic Journal of Botany, 14: 425-434. Kitayama, K. 1992. An altitudinal transect study of the vegetation on Mount Kinabalu. Borneo. Vegetatio, 102: 149-171. KOECHLIN. J.. J.-L. GUILLAUMET, AND P. MORAT. 1974. Flore et Vegetation de Madagascar. Cramer. Vaduz, Liechtenstein, 687 pp. Laubenfels (de). D. J. 1972. Podocarpacees. Flore de Madagascar et des Comores, 18: 1-22. LlEBERMAN, D., M. LlEBERMAN, R. PERALTA, AND G. S. Hartshorn. 1996. Tropical forest structure and com- position on a large-scale altitudinal gradient in Costa Rica. Journal of Ecology. 84: 137-152. Lewis. B. A.. P. Phillipson. M. Andrianarisata. G. Ra- hajasoa, P. J. Rakotomalaza, M. Randriambolo- lona, and J. F McDonagh. 1996. A study of the botanical structure, composition, and diversity of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagascar, pp. 24-75. In Goodman. S. M.. ed. A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra. Madagascar: With reference to eleva- tional variation. Fieldiana: Zoology, n.s.. 85. Lowry. P. P.. G. E. Schatz, AND P. B. Phillipson. 1997. The Classification of natural and anthropogenic veg- etation in Madagascar, pp. 93-123. //; Goodman. S. M., and B. D. Patterson, eds. Natural Change and Hu- man Impact in Madagascar. Smithsonian Institution Press. Washington. D.C. Malcomber. S. 1991. Permanent one hectare plots: Missouri Botanical Garden Methodology. Unpub- lished. Mori. S. A., and B. M. Boom. 1987. The Lecythida- ccae of a lowland Neotropical forest: La Fumee Mountain. French Guiana. Chapter II. The forest. Memoirs of the New York Botanical Garden. 44: 9- 29. Mori. S. A.. B. M. Boom, A. M. De Carvalino. and T S. Dos Santos. 1983. Ecological importance of Myr- taceae in an eastern Brazilian wet forest. Biotropica. 15: 68-70. Ohsawa. M. 1991. Structural comparison of tropical RAKOTOMALAZA & MESSMER: VEGETATION 71 montane rain forests along latitudinal and altitudinal gradients in south and east Asia. Vegetatio, 97: 1-10. Pendry, C. A., and J. Proctor. 1997. Altitudinal zo- nation of rain forest on Bukit Belalong, Brunei: Soils, forest structure and floristics. Journal of Tropical Ecol- ogy, 13: 221-241. Perrier de la Bathie, H. 1936. Biogeographie des plantes de Madagascar. Societe des Editions de Geo- graphic Maritime et Coloniale Paris, 156 pp. Phillipson, P. B. 1996. Endemism and non-endemism in the flora of south-west Madagascar, pp. 125-136. In Lourenco, W. R., ed. Biogeographie de Madagascar. Editions ORSTOM, Paris. Poisson, H. 1912. Recherches sur la Flore Meridionale de Madagascar. Librairie Maritime et Colonial, Paris. Rabevohitra, R., P. P. Lowry II, G. E. Schatz, H. Ran- drianjafy, and N. Razafindrianilana. 1996. As- sessment of plant diversity and conservation impor- tance of east coast low elevation Malagasy rain for- ests. Rapport sur le projet. Centre National de la Re- cherche Appliquee au Developpement Rural, Madagascar. Departement de Recherches Forestieres et Piscicoles, Madagascar. Biodiversity Support Pro- gram. Missouri Botanical Garden, St. Louis. Rollet, B. 1983. La regeneration naturelle dans les trouees. Un processus general de la dynamique des forets tropicales humides. Revue Bois et Forets des Tropiques, 201. Schatz, G. E. 1994. Botanical inventory of Ranoma- fana National Park. In Interim Report to the Liz Clair- borne and Art Ortenberg Foundation, Missouri Botan- ical Garden, St. Louis. Unpublished. Schulz, J. P. 1960. Ecological studies on rain forest in northern Suriname. Mededelingen van het Botanisch Museum en Herbarium van de Rijkuniversiteit te Utrecht, 160: 1-267. Shannon, C. E., and W. Weaver. 1949. The Mathe- matical Theory of Communication. University of Il- linois Press, Urbana, 117 pp. Silva, A. F. da. 1980. Composicao floristica e estrutura de um trecho da mata atlantica de encosta no muni- cipio de Ubatuba, S.R Disertacao apressentada ao In- stituto de Biologia da Universidade Estadual de Cam- pinas para a obtencao do titulo de Mestre em Biologia (Ecologia), Campinas, S.R, Brazil. Spurr, S. H. 1964. Forest Ecology. Ronald Press, New York. Sussman, R., and A. Rakotozafy. 1994. Plant diver- sity and structural analysis of tropical dry forest in southwestern Madagascar. Biotropica, 26: 241-254. Thomasson, M., and G. Thomasson. 1991. Essai sur la flore du Sud-Ouest malgache: Originalite, affinites et origines. Bulletin du Museum National d'Histoire Naturelle, Paris, serie 4, section B, Adansonia nos. 1- 2, 13: 71-89. 72 FIELDIANA: ZOOLOGY Appendix 4-1 List of the Families Present in the Five 1 ha Permanent Plots in Parcel 1 of the RNI d'Andohahela. Plot 1, Plot 2, Plot 3, Plot 4, Plot 5, Family 440 m 840 m 1150 m 1550 m 1875 m 1 Agavaceae 9.07 4.24 0.91 2.79 2 Anacardiaceae 10.14 8.08 6.55 3 Annonaceae 11.61 12.51 5.32 4 Apocynaceae 1.69 2.26 5 Aquifoliaceae 4.21 4.98 7.18 2.04 9.90 6 Araliaceae 4.25 17.94 11.47 13.59 43.24 7 Arecaceae 1.02 0.89 4.13 2.26 8 Asteraceae 1.69 2.29 9 Bignoniaceae 6.23 0.90 10 Burseraceae 32.97 0.83 1.10 11 Celastraceae 10.63 12 Clusiaceae 4.42 17.88 5.01 14.67 14.10 13 Connaraceae 1.40 14 Cunoniaceae 3.13 13.36 2.69 12.10 22.73 15 Cyatheaceae 7.66 4.47 1.85 9.29 2.52 16 Ebenaceae 19.80 7.55 4.15 17 Elaeocarpaceae 2.76 22.01 13.67 57.68 22.34 18 Ericaceae 7.20 19 Erythroxylaceae 12.45 1.70 5.46 20 Euphorbiaceae 6.03 17.76 16.56 28.21 9.38 21 Fabaceae 7.02 4.55 0.90 3.00 22 Flacourtiaceae 1.30 14.72 3.94 2.97 16.88 23 Hamamelidaceae 1.28 4.09 24 Icacinaceae 21.77 2.05 25 Lauraceae 9.15 17.18 38.48 35.08 47.41 26 Loganiaceae 3.38 27 Melastomataceae 1.64 28 Meliaceae 1.02 2.01 3.82 11.69 2.23 29 Monimiaceae 12.32 20.99 17.83 25.70 7.16 30 Moraceae 12.71 28.94 20.57 3.42 31 Myrsinaceae 15.90 1.43 4.97 3.98 5.58 32 Myrtaceae 14.01 20.82 34.96 11.21 25.37 33 Oleaceae 1.31 4.22 3.40 34 Pandanaceae 0.98 1.29 2.81 4.07 35 Physenaceae 1.53 36 Piperaceae 2.44 37 Pittosporaceae 1.29 1.25 4.30 38 Proteaceae 2.51 3.28 39 Rhizophoraceae 2.27 40 Rubiaceae 35.40 18.10 18.07 13.65 18.11 41 Rutaceae 3.76 3.78 3.69 3.34 42 Sapindaceae 7.15 2.35 8.96 43 Sapotaceae 16.55 7.72 44 Sterculiaceae 8.37 6.54 29.05 28.53 10.79 45 Tiliaceae 1.68 2.76 46 Verbenaceae 4.80 5.48 1.91 4.99 47 Violaceae 0.93 Note: FIV is indicated in boldface type when the value is >10.0. RAKOTOMALAZA & MESSMER: VEGETATION 73 Appendix 4-2 List of the Species Present in the Five 1 ha Permanent Plots in Parcel 1 of the KM d'Andohahela. Plot 1, Plot 2, Plot 3, Plot 4, Plot 5, Family Genus species 440 in 840 m 1150 m 1550 m 1875 m 1 Agavaceae Dracaena reflexa ssp2 11.59 3.11 2 Agavaceae Dracaena reflexa sspl 0.45 1.08 0.27 3.02 3 Anacardiaceae Anacardiaceae spl 0.35 4 Anacardiaceae Anacardiaceae sp2 1.24 5 Anacardiaceae Anacardiaceae sp3 0.68 0.84 6 Anacardiaceae Anacardiaceae sp4 4.62 0.27 7 Anacardiaceae Anacardiaceae sp5 0.59 8 Anacardiaceae Micronychia macrophylla 0.45 2.35 9 Anacardiaceae Protorhus spl 0.68 10 Anacardiaceae Protorhus sp2 0.31 11 Anacardiaceae Sorindeia madagascariensis 11.77 12 Annonaceae Annonaceae spl 0.45 6.40 13 Annonaceae Annonaceae sp2 0.93 5.41 0.29 14 Annonaceae Annonaceae sp3 0.53 15 Annonaceae Annonaceae sp4 0.34 16 Annonaceae Annonaceae sp5 0.34 17 Annonaceae Isolona spl 2.28 18 Annonaceae Isolona sp2 1.13 19 Annonaceae Polyathia capuronii 0.54 20 Annonaceae Polyalthia spl 2.80 4.66 21 Annonaceae Xylopia spl 0.34 22 Annonaceae Xylopia sp2 0.26 23 Apocynaceae Apocynaceae spl 0.27 24 Apocynaceae Apocynaceae sp2 1.16 25 Apocynaceae Cabucala spl 0.36 26 Apocynaceae Landolphia spl 0.34 27 Aqu foliaceae Ilex mitis 15.58 6.35 8.95 0.65 10.51 28 Aral aceae Araliaceae spl 0.89 29 Aral aceae Gastonia spl 0.35 30 Aral aceae Polycias spl 5.15 8.78 5.67 31 Aral aceae Polycias splO 1.48 32 Aral laceae Polycias sp2 1.60 1.08 33 Aral laceae Polycias sp3 0.89 34 Aral laceae Polycias sp4 0.58 0.95 35 Aral laceae Polycias sp5 0.92 36 Aral laceae Polycias sp6 1.28 37 Aral laceae Polycias sp7 0.78 32.00 38 Aral laceae Polycias sp8 3.26 39 Aral laceae Polycias sp9 7.16 11.77 40 Aral laceae Schefflera spl 1.61 1.23 41 Aral laceae Schefflera sp2 2.60 42 Aral laceae Schefflera sp3 1.38 43 Araliaceae Schefflera sp4 0.85 44 Arecaceae Dypsis spp 4.94 0.39 4.99 1.61 45 Asteraceae Brachylaena ramiflora 0.72 46 Asteraceae Vernonia spl 0.46 47 Bignoniaceae Colea spl 0.26 48 Bignoniaceae Ophiocolea floribunda 6.87 49 Bignoniaceae Rohdocolea linearis 0.42 50 Burseraceae Canarium boivinii 0.45 0.61 51 Burseraceae Canarium spl 0.34 52 Celastraceae Brexiella spl 11.68 53 Clusiaceae Clusiaceae spl 2.41 0.43 54 Clusiaceae Clusiaceae sp2 1.63 3.66 55 Clusiaceae Clusiaceae sp3 2.36 0.46 56 Clusiaceae Clusiaceae sp4 0.53 1.54 57 Clusiaceae Garcinia spl 0.48 1.05 58 Clusiaceae Garcinia sp2 0.90 0.37 16.45 59 Clusiaceae Harungana spl 0.67 74 FIELDIANA: ZOOLOGY Appendix 4-2 Continued. Plot 1, Plot 2, Plot 3, Plot 4, Plot 5, Family Genus species 440 m 840 m 1150 m 1550 m 1875 m 60 Clusiaceae Mammea spl 0.70 5.13 4.79 61 Clusiaceae Mammea sp2 1.24 0.60 62 Clusiaceae Mammea sp3 19.67 3.31 63 Clusiaceae Ochrocarpus spl 1.87 1.06 2.26 64 Clusiaceae Ochrocarpus sp2 1.70 2.64 65 Clusiaceae Ochrocarpus sp3 0.40 66 Clusiaceae Ochrocarpus sp4 0.26 3.73 67 Clusiaceae Symphonia spl 2.11 68 Connaraceae Ellipanthus spl 1.06 69 Cunoniaceae Weinmannia spl 1.37 2.27 70 Cunoniaceae Weinmannia sp2 0.70 1.54 71 Cunoniaceae Weinmannia sp3 0.87 4.36 72 Cunoniaceae Weinmannia sp4 0.84 15.54 12.20 73 Cunoniaceae Weinmannia sp5 5.56 74 Cunoniaceae Weinmannia sp6 1.33 6.82 75 Cunoniaceae Weinmannia sp7 2.92 76 Cyatheaceae Cyathea spl 3.56 4.62 2.42 12.42 1.39 77 Cyatheaceae Cyathea sp2 0.35 78 Ebenaceae Diospyros spl 0.46 0.85 1.05 79 Ebenaceae Diospyros sp2 0.96 0.71 1.81 80 Ebenaceae Diospyros sp3 3.76 5.55 81 Ebenaceae Diospyros sp4 0.98 0.50 82 Ebenaceae Diospyros sp5 0.91 83 Ebenaceae Diospyros sp6 0.59 84 Elaeocarpaceae Elaeocarpus spl 1.12 85 Elaeocarpaceae Elaeocarpus sp2 0.93 86 Elaeocarpaceae Elaeocarpus sp3 0.52 87 Elaeocarpaceae Elaeocarpus sp4 5.95 88 Elaeocarpaceae Elaeocarpus sp5 3.96 89 Elaeocarpaceae Elaeocarpus sp6 0.35 90 Elaeocarpaceae Sloanea rhodantha 3.74 22.60 11.83 57.93 18.65 91 Ericaceae Agauria spl 7.37 92 Erythroxylaceae Erythroxylum spl 16.93 93 Erythroxylaceae Erythroxylum sp2 0.96 94 Erythroxylaceae Erythroxyium sp3 0.47 95 Erythroxylaceae Erythroxylum sp4 6.50 96 Euphorbiaceae Antidesma petiolare 1.16 0.81 97 Euphorbiaceae Antidesma spl 1.32 98 Euphorbiaceae Bridelia pervilleana 2.81 1.10 99 Euphorbiaceae Bridelia tulasneana 2.00 100 Euphorbiaceae Croton monge 0.76 1.95 3.27 101 Euphorbiaceae Croton spl 1.97 3.72 102 Euphorbiaceae Croton sp2 0.49 0.66 103 Euphorbiaceae Croton sp3 2.60 104 Euphorbiaceae Croton sp4 1.32 105 Euphorbiaceae Croton sp5 0.27 106 Euphorbiaceae Dryptetes madagascariensis 2.79 0.53 107 Euphorbiaceae Drypetes spl 0.26 0.53 108 Euphorbiaceae Drypetes sp2 0.35 109 Euphorbiaceae Drypetes sp3 0.53 110 Euphorbiaceae Euphorbiaceae spl 0.34 111 Euphorbiaceae Euphorbiaceae sp2 1.90 112 Euphorbiaceae Macaranga cuspidata 0.54 113 Euphorbiaceae Macaranga spl 7.41 0.91 114 Euphorbiaceae Macaranga sp2 0.82 0.83 27.30 3.59 115 Euphorbiaceae Macaranga sp3 1.23 116 Euphorbiaceae Phyllanthus spl 0.98 117 Euphorbiaceae Phyllanthus sp2 0.27 118 Euphorbiaceae Suregada spl 2.15 119 Euphorbiaceae Uapaca spl 1.43 RAKOTOMALAZA & MESSMER: VEGETATION 75 Appendix 4-2 Continued. Plot 1, Plot 2, Plot 3, Plot 4, Plot 5, Family Genus species 440 in 840 m 1150 m 1550 m 1875 m 120 Fabaceae Albizzia gumifera 1.12 121 Fabaceae Albizzia spl 1.69 122 Fabaceae Fabaceae spl 4.74 3.92 123 Fabaceae Fabaceae sp2 0.55 124 Fabaceae Fabaceae sp3 0.26 125 Fabaceae Strongylodon spl 3.52 126 Flacourtiaceae Aphloia theaeformis 0.54 0.26 2.62 14.67 127 Flacourtiaceae Casearia spl 1.09 2.47 128 Flacourtiaceae Flacourtiaceae spl 2.15 6.01 4.30 129 Flacourtiaceae Flacourtiaceae sp2 0.92 1.10 130 Flacourtiaceae Flacourtiaceae sp3 0.55 131 Flacourtiaceae Scolopia spl 4.25 1.05 132 Flacourtiaceae Tisonia spl 0.45 3.08 0.26 133 Flacourtiaceae Tisonia sp2 0.27 134 Hamamelidaceae Dicoryphe spl 1.10 4.32 135 Icacinaceae Desmatostachys spl 0.98 136 Icacinaceae Icacinaceae spl 1.57 137 Lauraceae Belschmedia spl 14.97 138 Lauraceae Cinnamosma spl 0.54 139 Lauraceae Cryptocarya spl 2.39 4.41 140 Lauraceae Cryptocarya sp2 1.61 141 Lauraceae Cryptocarya sp3 0.34 10.28 142 Lauraceae Cryptocarya sp4 0.37 143 Lauraceae Lauraceae spl 0.92 1.58 1.01 2.86 144 Lauraceae Lauraceae splO 12.13 20.73 145 Lauraceae Lauraceae sp 1 1 1.07 146 Lauraceae Lauraceae spl 2 5.94 147 Lauraceae Lauraceae spl 3 5.51 148 Lauraceae Lauraceae spl 4 6.62 149 Lauraceae Lauraceae spl 5 9.49 150 Lauraceae Lauraceae spl 6 2.19 151 Lauraceae Lauraceae sp2 1.74 0.37 152 Lauraceae Lauraceae sp3 11.47 4.40 153 Lauraceae Lauraceae sp4 1.75 1.20 154 Lauraceae Lauraceae sp5 0.45 0.72 155 Lauraceae Lauraceae sp6 0.86 8.73 156 Lauraceae Lauraceae sp7 0.63 157 Lauraceae Lauraceae sp8 1.05 5.76 0.52 158 Lauraceae Lauraceae sp9 1.12 159 Lauraceae Ocotea spl 0.45 8.42 0.28 160 Lauraceae Ocotea sp2 1.00 161 Lauraceae Ocotea sp3 1.08 4.76 162 Lauraceae Ocotea sp4 0.79 163 Lauraceae Ocotea trichophebia 0.45 164 Lauraceae Potameia spl 3.56 165 Loganiaceae Anthocleista madagascariensis 2.88 166 Loganiaceae Loganiaceae spl 0.28 167 Melastomataceae Memecylon spl 1.34 2.63 168 Meliaceae Malleastrum gracile 1.06 1.27 0.95 169 Meliaceae Malleastrum spl 12.37 0.66 170 Meliaceae Malleastrum sp2 0.34 171 Meliaceae Malleastrum sp3 0.26 172 Meliaceae Turraea spl 0.45 173 Monimiaceae Decarydendron spl 0.35 174 Monimiaceae Ephippiandra spl 8.20 175 Monimiaceae Monimiaceae spl 18.76 176 Monimiaceae Monimiaceae sp2 0.34 2.07 2.47 177 Monimiaceae Monimiaceae sp3 0.34 0.94 178 Monimiaceae Tambourissa spl 1.25 7.00 6.44 179 Monimiaceae Tambourissa sp2 12.43 6.80 8.65 76 FIELDIANA: ZOOLOGY Appendix 4-2 Continued. Plot 1, Plot 2, Plot 3, Plot 4, Plot 5, Family Genus species 440 in 840 m 1150 m 1550 m 1875 m 180 Monimiaceae Tambourissa sp3 0.46 2.69 181 Monimiaceae Tambourissa sp4 1.86 8.96 182 Monimiaceae Tambourissa sp5 3.18 183 Monimiaceae Tambourissa sp6 1.05 184 Monimiaceae Tambourissa sp7 0.44 185 Moraceae Streblus spl 1.80 186 Moraceae Ficus soroceoides 0.34 0.49 187 Moraceae Ficus spl 3.28 188 Moraceae Ficus sp2 0.71 0.27 0.47 189 Moraceae Ficus sp3 1.81 190 Moraceae Ficus sp4 1.26 191 Moraceae Moraceae spl 7.13 7.47 192 Moraceae Moraceae sp2 0.46 2.36 13.15 193 Moraceae Moraceae sp3 0.52 0.35 194 Moraceae Moraceae sp4 0.45 0.68 195 Moraceae Moraceae sp5 0.50 196 Moraceae Moraceae sp6 0.35 197 Moraceae Moraceae sp7 0.83 198 Moraceae Moraceae sp8 0.70 199 Moraceae Moraceae sp9 0.38 200 Moraceae Streblus sp2 0.36 201 Moraceae Streblus dimepate 0.53 7.53 202 Moraceae Streblus mauritianus 2.02 203 Moraceae Treculia spl 2.14 204 Moraceae Trilepisium madagascariensis 2.07 1.77 205 Moraceae Trophis montana 0.35 206 Myrsinaceae Oncostemum sp 1 1 0.49 207 Myrsinaceae Oncostemum spl 0.54 208 Myrsinaceae Oncostemum splO 1.31 209 Myrsinaceae Oncostemum spl 2 6.19 210 Myrsinaceae Oncostemum sp2 8.33 211 Myrsinaceae Oncostemum sp3 1.10 1.17 212 Myrsinaceae Oncostemum sp4 1.34 0.54 213 Myrsinaceae Oncostemum sp5 0.45 214 Myrsinaceae Oncostemum sp6 1.15 215 Myrsinaceae Oncostemum sp7 0.45 216 Myrsinaceae Oncostemum sp8 0.45 1.92 217 Myrsinaceae Oncostemum sp9 0.46 218 Myrtaceae Eugenia emirnense 2.13 219 Myrtaceae Myrtaceae spl 0.59 5.72 220 Myrtaceae Myrtaceae splO 0.34 221 Myrtaceae Myrtaceae spl 1 8.47 222 Myrtaceae Myrtaceae spl 2 2.78 223 Myrtaceae Myrtaceae spl 3 13.64 224 Myrtaceae Myrtaceae spl 4 0.27 225 Myrtaceae Myrtaceae spl 5 6.61 226 Myrtaceae Myrtaceae spl 6 3.85 227 Myrtaceae Myrtaceae spl 7 1.18 228 Myrtaceae Myrtaceae spl 8 2.23 229 Myrtaceae Myrtaceae spl 9 0.68 230 Myrtaceae Myrtaceae sp2 1.26 231 Myrtaceae Myrtaceae sp20 6.31 232 Myrtaceae Myrtaceae sp21 1.32 233 Myrtaceae Myrtaceae sp22 4.86 234 Myrtaceae Myrtaceae sp3 0.59 0.72 0.57 235 Myrtaceae Myrtaceae sp4 1.07 3.38 236 Myrtaceae Myrtaceae sp5 0.35 17.00 237 Myrtaceae Myrtaceae sp6 0.39 238 Myrtaceae Myrtaceae sp7 7.39 239 Myrtaceae Myrtaceae sp8 0.38 2.36 RAKOTOMALAZA & MESSMER: VEGETATION 77 Appendix 4-2 Continued. Plot 1, Plot 2, Plot 3, Plot 4, Plot 5, Family Genus species 440 m 840 m 1150 m 1550 m 1875 m 240 Myrtaceae Myrtaceae sp9 5.58 241 Myrtaceae Sizygium spl 2.10 242 Myrtaceae Sizygium sp2 5.65 243 Oleaceae Noronhia spl 0.99 0.58 244 Oleaceae Noronhia sp2 4.32 245 Oleaceae Noronhia sp3 0.41 246 Oleaceae Noronhia sp4 0.30 247 Oleaceae Noronhia sp5 1.06 248 Pandanaceae Pandanus spl 2.75 249 Pandanaceae Pandanus sp2 1.10 4.68 250 Pandanaceae Pandanus sp3 0.67 251 Physenaceae Physena madagascariensis 1.21 252 Piperaceae Piper spl 1.29 253 Pittosporaceae Pittosporum spl 0.99 0.91 4.90 254 Proteaceae Dilobeia thouarsii 2.69 3.72 255 Rhizophoraceae Rhizophoraceae spl 1.04 256 Rubiaceae Breonia spl 3.23 257 Rubiaceae Craterispermum spl 0.37 258 Rubiaceae Gaertnera spl 3.01 259 Rubiaceae Gaertnera sp2 0.96 1.08 260 Rubiaceae Gaertnera sp3 1.81 261 Rubiaceae Gaertnera sp4 5.05 262 Rubiaceae Gaertnera sp5 0.96 263 Rubiaceae Hyperacanthus spl 11.36 264 Rubiaceae Hyperacanthus sp2 4.98 265 Rubiaceae Ixora spl 1.06 266 Rubiaceae Pouridiantha spl 0.47 267 Rubiaceae Psychotria spl 0.51 1.17 0.27 268 Rubiaceae Psychotria sp2 0.68 269 Rubiaceae Psychotria sp3 0.56 270 Rubiaceae Psychotria sp4 0.53 271 Rubiaceae Psychotria sp5 0.32 272 Rubiaceae Psychotria sp6 0.93 273 Rubiaceae Psychotria sp7 0.67 274 Rubiaceae Rothmania spl 0.30 275 Rubiaceae Rubiaceae spl 1.23 4.73 6.39 276 Rubiaceae Rubiaceae splO 0.52 277 Rubiaceae Rubiaceae spl 1 1.05 278 Rubiaceae Rubiaceae spl 2 0.53 279 Rubiaceae Rubiaceae spl 3 0.83 280 Rubiaceae Rubiaceae spl 4 0.49 281 Rubiaceae Rubiaceae spl 5 2.41 282 Rubiaceae Rubiaceae spl 6 0.94 283 Rubiaceae Rubiaceae spl 7 5.02 284 Rubiaceae Rubiaceae spl 8 3.77 285 Rubiaceae Rubiaceae sp2 4.77 1.33 0.78 286 Rubiaceae Rubiaceae sp3 0.61 1.58 287 Rubiaceae Rubiaceae sp4 0.55 288 Rubiaceae Rubiaceae sp5 0.94 1.04 289 Rubiaceae Rubiaceae sp6 4.04 0.75 290 Rubiaceae Rubiaceae sp7 0.55 1.34 291 Rubiaceae Rubiaceae sp8 0.59 292 Rubiaceae Rubiaceae sp9 0.47 0.62 293 Rubiaceae Schysmatoclada spl 1.40 294 Rubiaceae Tarenna spl 1.31 295 Rutaceae Vepris spl 2.21 296 Rutaceae Vepris sp2 0.94 297 Rutaceae Vepris sp3 0.27 298 Rutaceae Vepris sp4 1.41 0.29 299 Rutaceae Vepris sp5 0.27 78 FIELDIANA: ZOOLOGY Appendix 4-2 Continued. Family Genus species Plot 1, 440 m Plot 2, Plot 3, Plot 4, Plot 5, 840 m 1150 m 1550 m 1875 m 300 Rutaceae 301 Sapindaceae 302 Sapindaceae 303 Sapindaceae 304 Sapindaceae 305 Sapindaceae 306 Sapindaceae 307 Sapindaceae 308 Sapotaceae 309 Sapotaceae 310 Sapotaceae 311 Sapotaceae 312 Sterculiaceae 313 Sterculiaceae 314 Sterculiaceae 315 Sterculiaceae 316 Sterculiaceae 317 Sterculiaceae 318 Sterculiaceae 319 Sterculiaceae 320 Sterculiaceae 321 Sterculiaceae 322 Sterculiaceae 323 Sterculiaceae 324 Sterculiaceae 325 Sterculiaceae 326 Tiliaceae 327 Tiliaceae 328 Verbenaceae 329 Verbenaceae 330 Verbenaceae 331 Verbenaceae 332 Verbenaceae 333 Verbenaceae 334 Verbenaceae 335 Violaceae Zanthoxylum spl Allophylus cobbe Allophylus spl Allophylus sp2 Macphersonia spl Plagioscyphus spl Plagioscyphus sp2 Sapindaceae spl Capurodendron spl Chrysophyllum boiviniamum Mimusops spl Sapotaceae spl Dombeya spl Dombeya splO Dombeya sp 1 1 Dombeya spl 2 Dombeya sp2 Dombeya sp3 Dombeya sp4 Dombeya sp5 Dombeya sp6 Dombeya sp7 Dombeya sp8 Dombeya sp9 Sterculiaceae spl Sterculiaceae sp2 Grewia spl Grewia sp2 Clerodendrum spl Vitex spl Vitex sp2 Vitex sp3 Vitex sp4 Vitex sp5 Vitex sp6 Rinorea spl 1.46 1.06 3.92 0.58 17.85 0.49 2.67 0.84 0.89 0.70 0.56 8.45 0.54 1.66 6.87 0.29 0.31 0.59 4.83 2.72 3.33 3.67 0.71 9.42 1.21 0.42 9.60 0.47 1.82 1.18 1.17 0.83 2.61 4.66 0.97 18.38 5.37 0.90 0.34 3.77 0.35 1.01 0.26 0.69 0.38 1.92 0.27 0.90 1.28 2.29 0.29 2.21 1.50 3.68 5.16 Note: IVI is indicated in boldface type when the value is >10.0. RAKOTOMALAZA & MESSMER: VEGETATION 79 Appendix 4-3 Relative Density, Relative Dominance, Relative Specific Diversity, and FIV. Plot I at 440 m Relative Number Relative Relative Number specific of density Basal area dominance of diversity Family individuals (%) (cm2) (%) species (%) FIV Rubiaceae 79 13.23 27,338.78 9.77 15 12.40 35.40 Clusiaceae 62 10.39 40,056.69 14.32 10 8.26 32.97 Lauraceae 41 6.87 16,267.57 5.81 11 9.09 21.77 Elaeocarpaceae 18 3.02 42,346.69 15.13 2 1.65 19.80 Sapotaceae 29 4.86 28,087.17 10.04 2 1.65 16.55 Myrsinaceae 39 6.53 5,409.25 1.93 9 7.44 15.90 Myrtaceae 26 4.36 13,153.73 4.70 6 4.96 14.01 Moraceae 22 3.69 9,077.59 3.24 7 5.79 12.71 Euphorbiaceae 19 3.18 9,740.10 3.48 7 5.79 12.45 Monimiaceae 34 5.70 11,586.79 4.14 3 2.48 12.32 Aquifoliaceae 31 5.19 15,641.56 5.59 1 0.83 11.61 Annonaceae 19 3.18 3,287.83 1.18 7 5.79 10.14 Liliaceae 30 5.03 6,910.59 2.47 2 1.65 9.15 Anacardiaceae 34 5.70 4,822.35 1.72 2 1.65 9.07 Sterculiaceae 14 2.35 5,303.65 1.90 5 4.13 8.37 Ebenaceae 15 2.51 2,848.21 1.02 5 4.13 7.66 Sapindaceae 8 1.34 9,330.39 3.33 3 2.48 7.15 Flacourtiaceae 11 1.84 2,911.73 1.04 5 4.13 7.02 Fabaceae 11 1.84 4,788.25 1.71 3 2.48 6.03 Cunoniaceae 8 1.34 1,673.17 0.60 3 2.48 4.42 Arecaceae 7 1.17 6,294.33 2.25 1 0.83 4.25 Araliaceae 9 1.51 5,257.12 1.88 1 0.83 4.21 Cyatheaceae 11 1.84 1,283.29 0.46 1 0.83 3.13 Erythroxylaceae 5 0.84 752.33 0.27 2 1.65 2.76 Proteaceae 4 0.67 2,829.77 1.01 1 0.83 2.51 Piperaceae 3 0.50 788.06 0.28 2 1.65 2.44 Physenaceae 2 0.34 1,040.20 0.37 1 0.83 1.53 Oleaceae 2 0.34 417.40 0.15 1 0.83 1.31 Icacinaceae 2 0.34 395.87 0.14 1 0.83 1.30 Burseraceae 1 0.17 80.12 0.03 1 0.83 1.02 Meliaceae 1 0.17 80.12 0.03 1 0.83 1.02 Total 597 100.0 279,800.69 100.00 121 100.00 300.00 Plot 2 at 840 m Number Relative of Relative Relative Number specific individu- density Basal area dominance of diversity Family als (%) (cm2) (%) species (%) FIV Moraceae 71 9.13 34,574.26 8.93 16 10.88 28.94 Elaeocarpaceae 44 5.66 55,429.15 14.31 3 2.04 22.01 Monimiaceae 72 9.25 21,729.45 5.61 9 6.12 20.99 Myrtaceae 60 7.71 29,691.56 7.66 8 5.44 20.82 Rubiaceae 45 5.78 18,726.10 4.83 11 7.48 18.10 Araliaceae 51 6.56 20,402.74 5.27 9 6.12 17.94 Clusiaceae 45 5.78 12,596.73 3.25 13 8.84 17.88 Euphorbiaceae 55 7.07 17,680.37 4.56 9 6.12 17.76 Lauraceae 36 4.63 24,920.31 6.43 9 6.12 17.18 Flacourtiaceae 43 5.53 22,435.90 5.79 5 3.40 14.72 Cunoniaceae 47 6.04 18,942.32 4.89 4 2.72 13.65 Annonaceae 26 3.34 22,343.31 5.77 5 3.40 12.51 Anacardiaceae 16 2.06 10,168.99 2.63 5 3.40 8.08 Ebenaceae 17 2.19 10,231.58 2.64 4 2.72 7.55 Sterculiaceae 9 1.16 7,681.01 1.98 5 3.40 6.54 Bignoniaceae 21 2.70 8,393.33 2.17 2 1.36 6.23 Aquifoliaceae 18 2.31 7,711.92 1.99 1 0.68 4.98 80 FIELDIANA: ZOOLOGY Appendix 4-3 Continued. Number Relative of Relative Relative Number specific individu- density Basal area dominance of diversity Family als (%) (cm2) (%) species (%) FIV Verbenaceae 9 1.16 3,571.53 0.92 4 2.72 4.80 Fabaceae 15 1.93 4,871.78 1.26 2 1.36 4.55 Cyatheaceae 15 1.93 4,572.73 1.18 2 1.36 4.47 Agavaceae 9 1.16 6,664.90 1.72 2 1.36 4.24 Oleaceae 14 1.80 4,109.24 1.06 2 1.36 4.22 Rutaceae 10 • 1.29 4,330.37 1.12 2 1.36 3.76 Proteaceae 6 0.77 7,079.89 1.83 1 0.68 3.28 Sapindaceae 4 0.51 1,851.72 0.48 2 1.36 2.35 Meliaceae 4 0.51 527.36 0.14 2 1.36 2.01 Apocynaceae 2 0.26 289.65 0.07 2 1.36 1.69 Tiliaceae 2 0.26 249.95 0.06 2 1.36 1.68 Melastomataceae 3 0.39 2,241.13 0.58 0.68 1.64 Myrsinaceae 3 0.39 1,398.92 0.36 0.68 1.43 Pittosporaceae 2 0.26 1,383.79 0.36 0.68 1.29 Pandanaceae 2 0.26 165.13 0.04 0.68 0.98 Arecaceae 1 0.13 298.65 0.08 0.68 0.89 Burseraceae 1 0.13 100.29 0.03 0.68 0.83 Total 778 100.00 387,366.02 100.00 147 100.00 300.00 Plot 3 at 1150 m Relative Number Relative Relative Number specific of density Basal area dominance of diversity Family individuals (%) (cm2) (%) species (%) FIV Lauraceae 156 13.82 73,299.19 16.78 10 7.87 38.48 Myrtaceae 183 16.21 57,833.38 13.24 7 5.51 34.96 Sterculiaceae 95 8.41 72,905.35 16.69 5 3.94 29.05 Moraceae 88 7.79 35,171.15 8.05 6 4.72 20.57 Rubiaceae 64 5.67 9,462.54 2.17 13 10.24 18.07 Monimiaceae 73 6.47 35,854.02 8.21 4 3.15 17.83 Euphorbiaceae 47 4.16 9,440.24 2.16 13 10.24 16.56 Elaeocarpaceae 58 5.14 26,951.31 6.17 3 2.36 13.67 Araliaceae 38 3.37 14,753.81 3.38 6 4.72 11.47 Sapindaceae 41 3.63 12,950.96 2.97 3 2.36 8.96 Sapotaceae 23 2.04 14,501.25 3.32 3 2.36 7.72 Aquifoliaceae 29 2.57 16,693.73 3.82 1 0.79 7.18 Anacardiaceae 19 1.68 4,043.29 0.93 5 3.94 6.55 Verbenaceae 18 1.59 3,212.26 0.74 4 3.15 5.48 Annonaceae 21 1.86 4,805.67 1.10 3 2.36 5.32 Clusiaceae 22 1.95 6,493.99 1.49 2 1.57 5.01 Myrsinaceae 12 1.06 3,311.80 0.76 4 3.15 4.97 Ebenaceae 13 1.15 2,774.52 0.64 3 2.36 4.15 Arecaceae 19 1.68 7,237.66 1.66 1 0.79 4.13 Flacourtiaceae 7 0.62 732.78 0.17 4 3.15 3.94 Meliaceae 18 1.59 2,851.39 0.65 2 1.57 3.82 Rutaceae 5 0.44 832.05 0.19 4 3.15 3.78 Oleaceae 7 0.62 1,811.98 0.41 3 2.36 3.40 Loganiaceae 12 1.06 3,233.12 0.74 2 1.57 3.38 Tiliaceae 14 1.24 3,189.71 0.73 1 0.79 2.76 Cunoniaceae 8 0.71 1,793.82 0.41 2 1.57 2.69 Apocynaceae 5 0.44 1,042.20 0.24 2 1.57 2.26 Icacinaceae 4 0.35 3,981.43 0.91 0.79 2.05 Cyatheaceae 9 0.80 1,161.74 0.27 0.79 1.85 Connaraceae 3 0.27 1.517.17 0.35 0.79 1.40 Pandanaceae 4 0.35 649.50 0.15 0.79 1.29 Hamamelidaceae 4 0.35 615.10 0.14 0.79 1.28 RAKOTOMALAZA & MESSMER: VEGETATION Appendix 4-3 Continued. Relative Number Relative Relative Number specific of density Basal area dominance of diversity Family individuals (%) (cm2) (%) species (%) FIV Pittosporaceae 4 0.35 460.33 0.11 0.79 1.25 Burseraceae 2 0.18 580.69 0.13 0.79 1.10 Violaceae 1 0.09 226.98 0.05 0.79 0.93 Agavaceae 1 0.09 143.14 0.03 0.79 0.91 Fabaceae 1 0.09 93.31 0.02 0.79 0.90 Bignoniaceae 1 0.09 91.61 0.02 0.79 0.90 Total 1,129 100.00 436,704.17 100.00 127 100.00 300.00 Plot 4 at 1550 m Relative Number Relative Relative Number specific of density Basal area dominance of diversity Family individuals (%) (cm2) (%) species (%) FIV Elaeocarpaceae 26 4.17 297,370.02 48.96 3 4.55 57.68 Lauraceae 94 15.09 56,990.94 9.38 7 10.61 35.08 Sterculiaceae 74 11.88 36,701.04 6.04 7 10.61 28.53 Euphorbiaceae 106 17.01 31,169.85 5.13 4 6.06 28.21 Monimiaceae 54 8.67 57,441.91 9.46 5 7.58 25.70 Clusiaceae 32 5.14 11,870.89 1.95 5 7.58 14.67 Rubiaceae 16 2.57 2,908.71 0.48 7 10.61 13.65 Araliaceae 31 4.98 15,531.19 2.56 4 6.06 13.59 Cunoniaceae 42 6.74 23,366.95 3.85 1 1.52 12.10 Meliaceae 29 4.65 24,333.28 4.01 2 3.03 11.69 Myrtaceae 20 3.21 20,956.63 3.45 3 4.55 11.21 Cyatheaceae 41 6.58 7,224.38 1.19 1 1.52 9.29 Hamamelidaceae 9 1.44 6,853.60 1.13 1 1.52 4.09 Myrsinaceae 5 0.80 891.02 0.15 2 3.03 3.98 Rutaceae 3 0.48 1,058.80 0.17 2 3.03 3.69 Moraceae 2 0.32 397.00 0.07 2 3.03 3.42 Fabaceae 8 1.28 1,230.35 0.20 1 1.52 3.00 Flacourtiaceae 7 1.12 2,004.74 0.33 1 1.52 2.97 Pandanaceae 7 1.12 1,062.36 0.17 1 1.52 2.81 Agavaceae 7 1.12 909.27 0.15 1 1.52 2.79 Rhizophoraceae 1 0.16 3,599.71 0.59 1 1.52 2.27 Arecaceae 3 0.48 1,570.40 0.26 1 1.52 2.26 Aquifoliaceae 2 0.32 1,210.34 0.20 1 1.52 2.04 Verbenaceae 2 0.32 467.98 0.08 1 1.52 1.91 Erythroxylaceae 1 0.16 143.14 0.02 1 1.52 1.70 Asteraceae 1 0.16 83.32 0.01 1 1.52 1.69 Total 623 100.00 607,347.82 100.00 66 100.00 300.00 Plot 5 at 1875 m Relative Number Relative Relative Number specific of density Basal area dominance of diversity Family individuals (%) (cm2) (%) species (%) FIV Lauraceae 183 13.62 128,961.25 19.79 7 14.00 47.41 Araliaceae 233 17.35 129,617.75 19.89 3 6.00 43.24 Myrtaceae 105 7.84 23,037.00 3.54 7 14.00 25.37 Cunoniaceae 70 5.22 74,959.50 11.50 3 6.00 22.73 Elaeocarpaceae 75 5.60 70,012.50 10.74 3 6.00 22.34 Rubiaceae 73 5.41 17,598.25 2.70 5 10.00 18.11 Flacourtiaceae 83 6.16 43,814.50 6.72 2 4.00 16.88 Clusiaceae 105 7.84 27,789.75 4.26 1 2.00 14.10 82 FIELDIANA: ZOOLOGY Appendix 4-3 Continued. Relative Number Relative Relative Number specific of density Basal area dominance of diversity Family individuals (%) (cm2) (%) species (%) FIV Sterculiaceae 33 2.43 15,380.25 2.36 3 6.00 10.79 Celastraceae 75 5.60 19,788.25 3.04 2.00 10.63 Aquifoliaceae 53 3.92 25,937.75 3.98 2.00 9.90 Euphorbiaceae 30 2.24 7,427.00 1.14 3 6.00 9.38 Ericaceae 18 1.31 25,367.75 3.89 2.00 7.20 Monimiaceae 45 3.36 11,739.00 1.80 2.00 7.16 Myrsinaceae 38 2.80 5,109.50 0.78 2.00 5.58 Erythroxylaceae 35 2.61 5,508.25 0.85 2.00 5.46 Verbenaceae 30 2.24 4,880.75 0.75 2.00 4.99 Pittosporaceae 18 1.31 6,451.50 0.99 2.00 4.30 Pandanaceae 20 1.49 3,772.00 0.58 2.00 4.07 Rutaceae 13 0.93 2,631.50 0.40 2.00 3.34 Cyatheaceae 5 0.37 956.75 0.15 2.00 2.52 Asteraceae 3 0.19 672.00 0.10 2.00 2.29 Meliaceae 3 0.19 259.75 0.04 2.00 2.23 Total 1,340 100.00 651,672.50 100.00 50 100.00 300.00 Note: Data are organized by decreasing values for each plot, and the unidentified material is omitted. RAKOTOMALAZA & MESSMER: VEGETATION 83 Appendix 4-4 Relative Density, Relative Dominance, Relative Frequency, and INI. Plot 1 at 440 n i Number Relative Basal Relative domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Clusiaceae Mammea sp3 38 6.37 26,631.57 9.52 15 3.79 19.67 Sapotaceae Chrysophyllum boivinianum 28 4.69 27,627.21 9.87 13 3.28 17.85 Elaeocarpaceae Sloanea rhodantha var. rhodantha 12 2.01 38,223.33 13.66 5 1.26 16.93 Aquifoliaceae Ilex mitis 31 5.19 15,641.56 5.59 19 4.80 15.58 Monimiaceae Tambourissa sp2 30 5.03 10,818.00 3.87 14 3.54 12.43 Anacardiaceae Sorindeia madagascariensis 33 5.53 4,735.76 1.69 18 4.55 11.77 Agavaceae Dracaena reflexa sspl 29 4.86 6,827.27 2.44 17 4.29 11.59 Lauraceae Lauraceae sp3 26 4.36 8,612.66 3.08 16 4.04 11.47 Rubiaceae Hyperacanthus spl 26 4.36 7,591.27 2.71 17 4.29 11.36 Myrsinaceae Oncostemum sp2 24 4.02 3,579.59 1.28 12 3.03 8.33 Myrtaceae Myrtaceae sp7 16 2.68 6,112.74 2.18 10 2.53 7.39 Moraceae Moraceae spl 13 2.18 6,087.91 2.18 11 2.78 7.13 Araliaceae Polyscias spl 9 1.51 5,257.12 1.88 7 1.77 5.15 Rubiaceae Hyperacanthus sp2 13 2.18 2,889.40 1.03 7 1.77 4.98 Arecaceae Dypsis spp. 7 1.17 6,294.33 2.25 6 1.52 4.94 Rubiaceae Rubiaceae sp2 9 1.51 7,020.09 2.51 3 0.76 4.77 Fabaceae Albizzia spl 9 1.51 4,088.35 1.46 7 1.77 4.74 Rubiaceae Rubiaceae sp6 10 1.68 3,077.39 1.10 5 1.26 4.04 Sapindaceae Plagioscyphus sp2 3 0.50 8,158.80 2.92 2 0.51 3.92 Ebenaceae Diospyros sp3 8 1.34 1,815.77 0.65 7 1.77 3.76 Elaeocarpaceae Sloanea rhodantha 6 1.01 4,123.36 1.47 5 1.26 3.74 Sterculiaceae Dombeya sp2 8 1.34 2,984.68 1.07 5 1.26 3.67 Cyatheaceae Cyathea spl 11 1.84 1,283.29 0.46 5 1.26 3.56 Rubiaceae Gaertnera spl 5 0.84 2,559.58 0.91 5 1.26 3.01 Euphorbiaceae Antidesma petiolare 5 0.84 1,982.41 0.71 5 1.26 2.81 Annonaceae Polyalthia spl 6 1.01 772.52 0.28 6 1.52 2.80 Proteaceae Dilobeia thouarsii 4 0.67 2,829.77 1.01 4 1.01 2.69 Clusiaceae Clusiaceae spl 4 0.67 2,746.18 0.98 3 0.76 2.41 Clusiaceae Clusiaceae sp3 1 0.17 5,436.72 1.94 1 0.25 2.36 Annonaceae Isolona spl 6 1.01 751.34 0.27 4 1.01 2.28 Euphorbiaceae Macaranga cuspidata 5 0.84 847.26 0.30 4 1.01 2.15 Flacourtiaceae Caesaria spl 5 0.84 1,539.84 0.55 3 0.76 2.15 Moraceae Treculia spl 4 0.67 1,988.33 0.71 3 0.76 2.14 Myrtaceae Eugenia emirnense 3 0.50 2,425.58 0.87 3 0.76 2.13 Myrtaceae Sizygium spl 4 0.67 1,870.65 0.67 3 0.76 2.10 Clusiaceae Ochrocarpus spl 4 0.67 1,934.26 0.69 2 0.51 1.87 Lauraceae Lauraceae sp4 3 0.50 1,373.16 0.49 3 0.76 1.75 Lauraceae Lauraceae sp2 2 0.34 2,504.48 0.90 2 0.51 1.74 Clusiaceae Ochrocarpus sp2 5 0.84 1,000.17 0.36 2 0.51 1.70 Euphorbiaceae Uapaca spl 3 0.50 2,618.57 0.94 1 0.25 1.69 Clusiaceae Clusiaceae sp2 3 0.50 1,027.22 0.37 3 0.76 1.63 Sapindaceae Allophylus cobbe 3 0.50 560.60 0.20 3 0.76 1.46 Euphorbiaceae Suregada spl 2 0.34 1,652.29 0.59 2 0.51 1.43 Cunoniaceae Weinmannia spl 3 0.50 1,025.24 0.37 2 0.51 1.37 Myrsinaceae Oncostemum sp4 4 0.67 460.67 0.16 2 0.51 1.34 Euphorbiaceae Croton monge 2 0.34 1,333.29 0.48 2 0.51 1.32 Rubiaceae Tarenna spl 3 0.50 850.44 0.30 2 0.51 1.31 Piperaceae Piper spl 3 0.50 788.06 0.28 2 0.51 1.29 Monimiaceae Tambourissa spl 3 0.50 668.51 0.24 2 0.51 1.25 Clusiaceae Mammea sp2 3 0.50 639.39 0.23 2 0.51 1.24 Rubiaceae Rubiaceae spl 3 0.50 624.60 0.22 2 0.51 1.23 84 FIELDIANA: ZOOLOGY Appendix 4-4 Continued. Relative Number Relative Basal domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Physenaceae Physena madagascariensis 2 0.34 1.040.20 0.37 2 0.51 1.21 Sterculiaceae Dombeya sp3 2 0.34 1,028.56 0.37 2 0.51 1.21 Sterculiaceae Dombeya sp5 2 0.34 950.83 0.34 2 0.51 1.18 Erythroxyla- Erythroxylum sp2 3 0.50 423.55 0.15 2 0.51 1.16 ccac Myrsinaceae Oncostemum sp6 3 0.50 404.49 0.14 2 0.51 1.15 Annonaceae holona sp2 2 0.34 810.41 0.29 2 0.51 1.13 Myrsinaceae Oncostemum sp3 3 0.50 264.99 0.09 2 0.51 1.10 Flacourtiaceae Aphloia theaeformis 2 0.34 704.26 0.25 2 0.51 1.09 Lauraceae Ocotea sp3 2 0.34 679.44 0.24 2 0.51 1.08 Myrtaceae Myrtaceae sp4 1 0.17 1,809.56 0.65 1 0.25 1.07 Sapindaceae Plagioscyphus spl 2 0.34 610.99 0.22 2 0.51 1.06 Oleaceae Noronhia spl 2 0.34 417.40 0.15 2 0.51 0.99 Ebenaceae Diospyros sp4 2 0.34 401.18 0.14 2 0.51 0.98 Icacinaceae Desmatostachys spl 2 0.34 395.87 0.14 2 0.51 0.98 Ebenaceae Diospyros sp2 2 0.34 330.26 0.12 2 0.51 0.96 Erythroxyla- Erythroxylum spl 2 0.34 328.78 0.12 2 0.51 0.96 ceae Rubiaceae Gaertnera sp2 2 0.34 322.97 0.12 2 0.51 0.96 Rubiaceae Rubiaceae sp5 2 0.34 284.17 0.10 2 0.51 0.94 Annonaceae Annonaceae sp2 2 0.34 243.36 0.09 2 0.51 0.93 Flacourtiaceae Flacourtiaceae spl 2 0.34 234.52 0.08 2 0.51 0.92 Lauraceae Lauraceae spl 2 0.34 221.68 0.08 2 0.51 0.92 Ebenaceae Diospyros sp5 2 0.34 200.72 0.07 2 0.51 0.91 Clusiaceae Garcinia sp2 2 0.34 180.23 0.06 2 0.51 0.90 Cunoniaceae Weinmannia sp3 3 0.50 321.06 0.11 0.25 0.87 Lauraceae Lauraceae sp6 0.17 1,225.42 0.44 0.25 0.86 Lauraceae Ocotea sp4 0.17 1,029.22 0.37 0.25 0.79 Euphorbiaceae Bridelia pervilleana 0.17 956.62 0.34 0.25 0.76 Cunoniaceae Weinmannia sp2 2 0.34 326.88 0.12 0.25 0.70 Rubiaceae Rubiaceae sp3 0.17 518.75 0.19 0.25 0.61 Rubiaceae Rubiaceae sp8 0.17 483.05 0.17 0.25 0.59 Myrtaceae Myrtaceae sp3 0.17 471.44 0.17 0.25 0.59 Myrtaceae Myrtaceae spl 0.17 463.77 0.17 0.25 0.59 Sapotaceae Capurodendron spl 0.17 459.96 0.16 0.25 0.58 Rubiaceae Rubiaceae sp4 0.17 373.25 0.13 0.25 0.55 Fabaceae Fabaceae spl 0.17 369.84 0.13 0.25 0.55 Rubiaceae Rubiaceae sp7 0.17 369.84 0.13 0.25 0.55 Flacourtiaceae Flacourtiaceae sp2 0.17 352.99 0.13 0.25 0.55 Euphorbiaceae Croton sp4 0.17 349.67 0.12 0.25 0.54 Lauraceae Cinnamosma spl 0.17 339.80 0.12 0.25 0.54 Myrsinaceae Oncostemum spl 0.17 333.29 0.12 0.25 0.54 Fabaceae Fabaceae sp2 0.17 330.06 0.12 0.25 0.54 Annonaceae Polyalthia capuronii 0.17 326.85 0.12 0.25 0.54 Annonaceae Annonaceae sp3 0.17 304.81 0.11 0.25 0.53 Clusiaceae Clusiaceae sp4 0.17 304.81 0.11 0.25 0.53 Moraceae Strebulus dimepate 0.17 301.72 0.11 0.25 0.53 Moraceae Moraceae sp5 0.17 274.65 0.10 0.25 0.52 Rubiaceae Psychotria spl 0.17 243.29 0.09 0.25 0.51 Moraceae Moraceae sp5 0.17 229.66 0.08 0.25 0.50 Sterculiaceae Dombeya spl 0.17 198.56 0.07 0.25 0.49 Clusiaceae Garcinia spl 0.17 156.15 0.06 0.25 0.48 Sterculiaceae Dombeya sp4 0.17 141.03 0.05 0.25 0.47 Rubiaceae Rubiaceae sp9 0.17 130.70 0.05 0.25 0.47 Myrsinaceae Oncostemum sp9 0.17 109.36 0.04 0.25 0.46 Ebenaceae Diospyros spl 0.17 100.29 0.04 0.25 0.46 Monimiaceae Tambourissa sp3 0.17 100.29 0.04 0.25 0.46 RAKOTOMALAZA & MESSMER: VEGETATION 85 Appendix 4-4 Continued. Relative Number Relative Basal domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Moraceae Moraceae sp2 0.17 100.29 0.04 0.25 0.46 Lauraceae Lauraceae sp5 0.17 96.77 0.03 0.25 0.45 Lauraceae Ocotea spl 0.17 95.03 0.03 0.25 0.45 Moraceae Moraceae sp4 0.17 95.03 0.03 0.25 0.45 Myrsinaceae Oncostemum sp5 0.17 95.03 0.03 0.25 0.45 Lauraceae Ocotea trichophebia 0.17 89.92 0.03 0.25 0.45 Anacardiaceae Micronychia macrophylla 0.17 86.59 0.03 0.25 0.45 Agavaceae Dracaena reflexa ssp2 0.17 83.32 0.03 0.25 0.45 Myrsinaceae Oncostemum sp8 0.17 81.71 0.03 0.25 0.45 Burseraceae Canarium boivinii 0.17 80.12 0.03 0.25 0.45 Flacourtiaceae Flacourtiaceae sp3 0.17 80.12 0.03 0.25 0.45 Meliaceae Turraea spl 0.17 80.12 0.03 0.25 0.45 Myrsinaceae Oncostemum sp7 0.17 80.12 0.03 0.25 0.45 Annonaceae Annonaceae spl 0.17 78.54 0.03 0.25 0.45 Total 597 100.00 279,800.69 100.00 395 99.75 299.75 Plot 2 at 840 m Relative Number Relative Basal domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Elaeocarpaceae Sloanea rhodantha 40 5.14 54,610.72 14.10 18 3.35 22.60 Araliaceae Polyscias spl 23 2.96 12,474.13 3.22 14 2.61 8.78 Myrtaceae Myrtaceae sp 1 1 24 3.08 10,031.35 2.59 15 2.79 8.47 Lauraceae Ocotea spl 18 2.31 15,719.07 4.06 11 2.05 8.42 Moraceae Streblus dimepate 27 3.47 4,197.67 1.08 16 2.98 7.53 Moraceae Moraceae spl 7 0.90 21,099.64 5.45 6 1.12 7.47 Euphorbiaceae Macaranga spl 28 3.60 6,809.20 1.76 11 2.05 7.41 Monimiaceae Tambourissa spl 22 2.83 6,078.97 1.57 14 2.61 7.00 Bignoniaceae Ophiocolea floribunda 20 2.57 7,985.05 2.06 12 2.23 6.87 Monimiaceae Tambourissa sp2 20 2.57 8,461.82 2.19 11 2.05 6.80 Annonaceae Annonaceae spl 8 1.03 16,491.53 4.26 6 1.12 6.40 Aquifoliaceae Ilex mitis 18 2.31 7,711.92 1.99 11 2.05 6.35 Flacourtiaceae Flacourtiaceae spl 21 2.70 4,150.26 1.07 12 2.23 6.01 Myrtaceae Myrtaceae spl 13 1.67 9,924.81 2.56 8 1.49 5.72 Myrtaceae Myrtaceae sp9 15 1.93 7,659.19 1.98 9 1.68 5.58 Cunoniaceae Weinmannia sp5 15 1.93 11,908.14 3.08 3 0.56 5.56 Ebenaceae Diospyros sp3 11 1.41 8,811.35 2.28 10 1.86 5.55 Annonaceae Annonaceae sp2 15 1.93 5,565.35 1.44 11 2.05 5.41 Rubiaceae Rubiaceae spl 13 1.67 7,526.73 1.94 6 1.12 4.73 Cyatheaceae Cyathea spl 14 1.80 4,450.01 1.15 9 1.68 4.62 Anacardiaceae Anacardiaceae sp4 10 1.29 7,872.42 2.03 7 1.30 4.62 Cunoniaceae Weinmannia sp3 17 2.19 4,799.53 1.24 5 0.93 4.36 Oleaceae Noronhia sp2 13 1.67 3,749.56 0.97 9 1.68 4.32 Flacourtiaceae Scolopia spl 10 1.29 5,706.28 1.47 8 1.49 4.25 Fabaceae Fabaceae spl 12 1.54 4,175.62 1.08 7 1.30 3.92 Proteaceae Dilobeia thouarsii 6 0.77 7,079.89 1.83 6 1.12 3.72 Clusiaceae Clusiaceae sp2 11 1.41 2,193.37 0.57 9 1.68 3.66 Clusiaceae Mammea sp3 9 1.16 2,567.97 0.66 8 1.49 3.31 Rubiaceae Breonia spl 9 1.16 4,422.53 1.14 5 0.93 3.23 Monimiaceae Tambourissa sp5 10 1.29 2,307.87 0.60 7 1.30 3.18 Agavaceae Dracaena reflexa sspl 6 0.77 6,156.32 1.59 4 0.74 3.11 Flacourtiaceae Tisonia spl 4 0.51 8,501.73 2.20 2 0.37 3.08 86 FIELDIANA: ZOOLOGY Appendix 4-4 Continued. Relative Number Relative Basal domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Euphorbiaceae Drypetes madagascariensis 8 1.03 2,481.87 0.64 6 1.12 2.79 Monimiaceae Tambourissa sp3 8 1.03 2,818.28 0.73 5 0.93 2.69 Rutaceae Zanthoxyllum spl 7 0.90 2,543.17 0.66 6 1.12 2.67 Clusiaceae Ochrocarpus sp2 9 1.16 1,408.49 0.36 6 1.12 2.64 Flacourtiaceae Caesaria spl 6 0.77 2,245.23 0.58 6 1.12 2.47 Lauraceae Cryptocarya spl 5 0.64 3,167.84 0.82 5 0.93 2.39 Moraceae Moraceae sp2 6 0.77 2,547.42 0.66 5 0.93 2.36 Cunoniaceae Weinmannia spl 10 1.29 1,643.13 0.42 3 0.56 2.27 Moraceae Trilepisium madagascariensis 7 0.90 918.95 0.24 5 0.93 2.07 Moraceae Streblus mauritianus 6 0.77 1,231.01 0.32 5 0.93 2.02 Euphorbiaceae Bridelia tulasneana 4 0.51 3,575.07 0.92 3 0.56 2.00 Euphorbiaceae Croton monge 4 0.51 2,684.76 0.69 4 0.74 1.95 Monimiaceae Tambourissa sp4 6 0.77 1,349.22 0.35 4 0.74 1.86 Sterculiaceae Dombeya sp4 2 0.26 4,596.42 1.19 2 0.37 1.82 Moraceae Streblus spl 4 0.51 2,079.59 0.54 4 0.74 1.80 Lauraceae Cryptocarya sp2 3 0.39 2,586.70 0.67 3 0.56 1.61 Araliaceae Schefflera spl 5 0.64 1,585.95 0.41 3 0.56 1.61 Araliaceae Polyscias splO 4 0.51 1,331.56 0.34 4 0.74 1.60 Cunoniaceae Weinmannia sp2 5 0.64 591.51 0.15 4 0.74 1.54 Clusiaceae Clusiaceae sp4 4 0.51 1,069.37 0.28 4 0.74 1.54 Araliaceae Polyscias splO 5 0.64 1,078.78 0.28 3 0.56 1.48 Rutaceae Vepris sp4 3 0.39 1,787.20 0.46 3 0.56 1.41 Araliaceae Schefflera sp3 6 0.77 1,648.75 0.43 1 0.19 1.38 Rubiaceae Rubiaceae sp7 4 0.51 1,040.98 0.27 3 0.56 1.34 Melastomataceae Memecylon spl 3 0.39 2,241.13 0.58 2 0.37 1.34 Rubiaceae Rubiaceae sp2 3 0.39 1,502.77 0.39 3 0.56 1.33 Myrsinaceae Oncostemum splO 3 0.39 1,398.92 0.36 3 0.56 1.31 Verbenaceae Vite.x sp5 4 0.51 808.32 0.21 3 0.56 1.28 Myrtaceae Myrtaceae sp2 3 0.39 1,236.10 0.32 3 0.56 1.26 Euphorbiaceae Macaranga sp3 4 0.51 601.72 0.16 3 0.56 1.23 Sterculiaceae Dombeya sp5 2 0.26 2,094.14 0.54 2 0.37 1.17 Rubiaceae Psvchotria spl 3 0.39 869.72 0.22 3 0.56 1.17 Elaeocarpaceae Elaeocarpus spl 3 0.39 697.66 0.18 3 0.56 1.12 Fabaceae Albizzia gumifera 3 0.39 696.15 0.18 3 0.56 1.12 Flacourtiaceae Flacourtiaceae sp2 2 0.26 1,832.40 0.47 2 0.37 1.10 Euphorbiaceae Bridelia pervillena 3 0.39 588.15 0.15 3 0.56 1.10 Rubiaceae Gaertnera sp2 3 0.39 543.09 0.14 3 0.56 1.08 Agavaceae Dracaena reflexa ssp2 3 0.39 508.58 0.13 3 0.56 1.08 Clusiaceae Ochrocarpus spl 3 0.39 454.22 0.12 3 0.56 1.06 Meliaceae Malleastrum gracile 3 0.39 445.64 0.12 3 0.56 1.06 Lauraceae Lauraceae sp8 2 0.26 1,644.54 0.42 2 0.37 1.05 VIonimiaceae Tambourissa sp6 3 0.39 414.88 0.11 3 0.56 1.05 Clusiaceae Garcinia spl 2 0.26 1,629.31 0.42 2 0.37 1.05 Rubiaceae Rubiaceae spl 1 3 0.39 405.99 0.10 3 0.56 1.05 Rubiaceae Rubiaceae sp5 2 0.26 1,579.23 0.41 2 0.37 1.04 Verbenaceae Clerodendrum spl 2 0.26 1,465.97 0.38 2 0.37 1.01 Lauraceae Ocotea sp2 3 0.39 944.87 0.24 2 0.37 1.00 Pittosporaceae Pittosporum spl 2 0.26 1,383.79 0.36 2 0.37 0.99 \raliaceae Polyscias sp5 3 0.39 610.08 0.16 2 0.37 0.92 Verbenaceae Vitex sp3 2 0.26 1,028.44 0.27 2 0.37 0.90 Sapindaceae Sapindaceae spl 2 0.26 1,024.94 0.26 2 0.37 0.89 \raliceae Araliaceae spl 3 0.39 496.01 0.13 2 0.37 0.89 Ebenaceae Diospyros spl 2 0.26 864.76 0.22 2 0.37 0.85 Sapindaceae Allophylus spl 2 0.26 826.77 0.21 2 0.37 0.84 Vloraceae Moraceae sp7 2 0.26 781.08 0.20 2 0.37 0.83 Euphorbiaceae Macaranga sp2 2 0.26 731.94 0.19 2 0.37 0.82 IAKOTOMALAZA & MESSMER: VEGETATION 87 Appendix 4-4 Continued. Relative Number Relative Basal domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Rubiaceae Rubiaceae sp6 2 0.26 448.89 0.12 2 0.37 0.75 Lauraceae Lauraceae sp5 2 0.26 349.61 0.09 2 0.37 0.72 Myrtaceae Myrtaceae sp3 2 0.26 338.77 0.09 2 0.37 0.72 Moraceae Ficus sp2 2 0.26 322.91 0.08 2 0.37 0.71 Ebenaceae Diospyros sp2 2 0.26 321.27 0.08 2 0.37 0.71 Sterculiaceae Dombeya sp2 2 0.26 309.78 0.08 2 0.37 0.71 Clusiaceae Mammea spl 2 0.26 272.85 0.07 2 0.37 0.70 Sterculiaceae Dombeya spl 2 0.26 268.80 0.07 2 0.37 0.70 Moraceae Moraceae sp8 2 0.26 264.80 0.07 2 0.37 0.70 Anacardiaceae Anacardiaceae sp3 2 0.26 202.15 0.05 2 0.37 0.68 Anacardiaceae Protorhus spl 1 0.26 1,418.63 0.37 1 0.19 0.68 Moraceae Moraceae sp4 2 0.13 186.05 0.05 2 0.37 0.68 Rubiaceae Psychotria sp2 2 0.26 185.11 0.05 2 0.37 0.68 Pandanaceae Pandanus sp3 2 0.26 165.13 0.04 2 0.37 0.67 Clusiaceae Harungana spl 0.13 1,372.28 0.35 0.19 0.67 Anacardiaceae Anacardiaceae sp5 2 0.26 555.03 0.14 0.19 0.59 Araliaceae Polyscias sp4 0.13 1,023.54 0.26 0.19 0.58 Ebenaceae Diospyros sp4 2 0.26 234.19 0.06 0.19 0.50 Clusiaceae Clusiaceae sp3 0.13 543.25 0.14 0.19 0.46 Clusiaceae Clusiaceae spl 0.13 433.74 0.11 0.19 0.43 Sterculiaceae Dombeya sp3 0.13 411.87 0.11 0.19 0.42 Bignoniaceae Rhodocolea linearis 0.13 408.28 0.11 0.19 0.42 Oleaceae Noronhia sp3 0.13 359.68 0.09 0.19 0.41 Clusiaceae Ochrocarpus sp3 0.13 311.03 0.08 0.19 0.40 Arecaceae Dypsis spl 0.13 298.65 0.08 0.19 0.39 Myrtaceae Myrtaceae sp8 0.13 268.80 0.07 0.19 0.38 Verbenaceae Vitex spl 0.13 268.80 0.07 0.19 0.38 Moraceae Moraceae sp9 0.13 260.16 0.07 0.19 0.38 Clusiaceae Garcinia sp2 0.13 229.66 0.06 0.19 0.37 Lauraceae Cryptocarya sp4 0.13 221.67 0.06 0.19 0.37 Lauraceae Lauraceae sp2 0.13 201.06 0.05 0.19 0.37 Rubiaceae Craterispermum spl 0.13 201.06 0.05 0.19 0.37 Apocynaceae Cabucala spl 0.13 191.13 0.05 0.19 0.36 Moraceae Streblus sp2 0.13 174.37 0.05 0.19 0.36 Araliaceae Gastonia spl 0.13 153.94 0.04 0.19 0.35 Moraceae Moraceae sp3 0.13 153.94 0.04 0.19 0.35 Moraceae Trophis montana 0.13 138.93 0.04 0.19 0.35 Tiliaceae Grewia sp2 0.13 136.85 0.04 0.19 0.35 Myrtaceae Myrtaceae sp5 0.13 128.68 0.03 0.19 0.35 Monimiaceae Decarydendron spl 0.13 126.68 0.03 0.19 0.35 Cyatheaceae Cyathea sp2 0.13 122.72 0.03 0.19 0.35 Euphorbiaceae Drypetes sp2 0.13 122.72 0.03 0.19 0.35 Moraceae Moraceae sp6 0.13 122.72 0.03 0.19 0.35 Anacardiaceae Anacardiaceae spl 0.13 120.76 0.03 0.19 0.35 Elaeocarpaceae Elaeocarpus sp6 0.13 120.76 0.03 0.19 0.35 Tiliaceae Grewia spl 0.13 113.10 0.03 0.19 0.34 Annonaceae Xylopia spl 0.13 109.36 0.03 0.19 0.34 Myrtaceae Myrtaceae splO 0.13 103.87 0.03 0.19 0.34 Burseraceae Canarium spl 0.13 100.29 0.03 0.19 0.34 Annonaceae Annonaceae sp4 0.13 98.52 0.03 0.19 0.34 Apocynaceae Landolphia spl 0.13 98.52 0.03 0.19 0.34 Moraceae Ficus soroceoides 0.13 95.03 0.02 0.19 0.34 Monimiaceae Monimiaceae sp3 0.13 91.61 0.02 0.19 0.34 Euphorbiaceae Euphorbiaceae spl 0.13 84.95 0.02 0.19 0.34 Lauraceae Cryptocarya sp3 0.13 84.95 0.02 0.19 0.34 Meliaceae Malleastrum sp2 0.13 81.71 0.02 0.19 0.34 Monimiaceae Monimiaceae sp2 0.13 80.12 0.02 0.19 0.34 Annonaceae Annonaceae sp5 0.13 78.54 0.02 0.19 0.34 Total 778 100.00 387,254.80 100.00 537 100.00 300.00 FIELDIANA: ZOOLOGY Appendix 4-4 Continued. Plot 3 at 1150 m Number Relative Basal Relative domi- of indi- densitv area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Sterculiaceae Dombeya sp9 42 3.72 55,482.83 12.70 13 1.95 18.38 Myrtaceae Myrtaceae spl3 92 8.15 23,539.12 5.39 23 3.46 17.00 Myrtaceae Myrtaceae spl4 54 4.78 23,573.35 5.40 23 3.46 13.64 Moraceae Trilepisium madagascariensis 74 6.55 17,628.14 4.04 17 2.56 13.15 Lauraceae Lauraceae splO 52 4.61 19,719.64 4.52 20 3.01 12.13 Elaeocarpaceae Sloanea rhodantha 38 3.37 23,186.55 5.31 21 3.16 11.83 Sterculiaceae Dombeya sp3 37 3.28 13.178.82 3.02 22 3.31 9.60 Monimiaceae Tambourissa sp7 49 4.34 8,348.30 1.91 18 2.71 8.96 Aquifoliaceae Ilex mitis 29 2.57 16,693.73 3.82 17 2.56 8.95 Monimiaceae Tambourissa sp4 15 1 .33 25.410.51 5.82 10 1.50 8.65 Sapindaceae Allophylus sp2 34 3.01 9,972.92 2.28 21 3.16 8.45 Sapotaceae Chrysophyllum boivinianum 21 1.86 13,999.02 3.21 12 1.80 6.87 Lauraceae Lauraceae spl4 23 2.04 12.134.87 2.78 12 1.80 6.62 Lauraceae Lauraceae spl2 17 1.51 11.472.39 2.63 12 1.80 5.94 Lauraceae Lauraceae sp9 19 1.68 7,954.79 1.82 15 2.26 5.76 Araliaceae Polyscias spl 22 1.95 8,355.90 1.91 12 1.80 5.67 Lauraceae Lauraceae spl 3 15 1 .33 11.698.17 2.68 10 1.50 5.51 Clusiaceae Mammea spl 21 1.86 6,400.68 1.47 12 1.80 5.13 Rubiaceae Gaertnera sp4 29 2.57 3,608.46 0.83 11 1.65 5.05 Arecaceae Dypsis spp 19 1.68 7,237.66 1.66 11 1.65 4.99 Annonaceae Polyalthis spl 19 1.68 4,455.78 1.02 13 1.95 4.66 Lauraceae Cryptocarya spl 18 1.59 5,749.63 1.32 10 1.50 4.41 Elaeocarpaceae Elaeocarpus sp5 17 1.51 2,819.92 0.65 12 1.80 3.96 Tiliaceae Grewia spl 14 1.24 3,189.71 0.73 12 1.80 3.77 Myrtaceae Myrtaceae sp8 19 1.68 6,117.29 1.40 2 0.30 3.38 Moraceae Ficus sp2 5 0.44 10,426.27 2.39 3 0.45 3.28 Euphorbiaceae Croton monge 10 0.89 4,480.96 1.03 9 1.35 3.27 Loganiaceae Anthocleista madagascariensis 11 0.97 3,070.26 0.70 8 1.20 2.88 Myrtaceae Myrtaceae spl 2 12 1.06 2,265.36 0.52 8 1.20 2.78 Meliaceae Malleastrum gracile 13 1.15 1,860.26 0.43 7 1.05 2.63 Sterculiaceae Dombeya sp7 9 0.80 3,325.72 0.76 7 1.05 2.61 Cyatheaceae Cyathea spl 9 0.80 1.161.74 0.27 9 1.35 2.42 Myrtaceae Myrtaceae sp5 3 0.27 1,921.71 0.44 11 1.65 2.36 Anacardiaceae Micronychia macrophylla 11 0.97 1,425.50 0.33 7 1.05 2.35 Verbenaceae Vitex sp6 9 0.80 1.915.74 0.44 7 1 .05 2.29 Monimiaceae Tambourissa sp2 8 0.71 2,001.90 0.46 6 0.90 2.07 Euphorbiaceae Croton spl 9 0.80 1.182.39 0.27 6 0.90 1.97 Verbenaceae Vitex spl 7 0.62 1.091.11 0.25 7 1.05 1.92 Euphorbiaceae Euphorbiaceae sp2 2 0.18 318.59 0.07 11 1.65 1.90 Rubiaceae Gaertnera sp3 8 0.71 1 .534.73 0.35 5 0.75 1.81 Ebenaceae Diospyros sp2 7 0.62 1.912.83 0.44 5 0.75 1.81 Moraceae Ficus sp4 3 0.27 4.762.85 1.09 3 0.45 1.81 Myrsinaceae Memecylon spl 5 0.44 2.509.21 0.57 5 0.75 1.77 Sapindaceae Sapindaceae spl 5 0.44 2.699.81 0.62 4 0.60 1.66 Lauraceae Lauraceae spl 5 0.44 2.332.67 0.53 4 0.60 1.58 Rubiaceae Rubiaceae sp3 7 0.62 897.93 0.21 5 0.75 1.58 Icacinaceae Icacinaceae spl 4 0.35 3.981.43 0.91 2 0.30 1.57 Cunoniaceae Weinmannia sp6 5 0.44 594.16 0.14 5 0.75 1.33 Euphorbiaceae Antidesma spl 5 0.44 558.97 0.13 5 0.75 1.32 Meliaceae Malleastrum sp3 5 0.44 991.13 0.23 4 0.60 1.27 Moraceae Moraceae sp2 4 0.35 1,332.43 0.31 4 0.60 1.26 Anacardiaceae Anacardiaceae sp2 4 0.35 1.227.09 0.28 4 0.60 1.24 Araliaceae Schefflera spl 3 0.27 2,897.29 0.66 2 0.30 1.23 Myrsinaceae Oncostemum sp3 5 0.44 536.73 0.12 4 0.60 1.17 Apocynaceae Apocynaceae sp2 4 0.35 886.05 0.20 4 0.60 1.16 Lauraceae Ocotea spl 4 0.35 1,357.52 0.31 3 0.45 1.12 RAKOTOMALAZA & MESSMER: VEGETATION 89 Appendix 4-4 Continued. Plot 3 at 1150 m Number Relative Basal Relative domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Pandanaceae Pandanus sp2 4 0.35 649.50 0.15 4 0.60 1.10 Hamamelida- Dicoryphe spl 4 0.35 615.10 0.14 4 0.60 1.10 ceae Araliaceae Polyscias sp2 4 0.35 1,185.40 0.27 3 0.45 1.08 Connaraceae Ellipanthus spl 3 0.27 1,517.17 0.35 3 0.45 1.06 Oleaceae Noronhia sp5 4 0.35 1,100.30 0.25 3 0.45 1.06 Ebenaceae Diospyros spl 4 0.35 393.71 0.09 4 0.60 1.05 Flacourtiaceae Scolopia spl 4 0.35 392.91 0.09 4 0.60 1.05 Euphorbiaceae Phyllanthus spl 4 0.35 741.18 0.17 3 0.45 0.98 Araliaceae Polyscias sp4 3 0.27 1,008.25 0.23 3 0.45 0.95 Elaeocarpaceae Elaeocarpus sp2 3 0.27 944.84 0.22 3 0.45 0.93 Euphorbiaceae Macaranga spl 4 0.35 474.45 0.11 3 0.45 0.91 Pittosporaceae Pittosporum spl 4 0.35 460.33 0.11 3 0.45 0.91 Sterculiaceae Sterculiaceae spl 4 0.35 428.48 0.10 3 0.45 0.90 Araliaceae Polyscias sp3 3 0.27 743.21 0.17 3 0.45 0.89 Araliaceae Schefflera sp4 3 0.27 563.76 0.13 3 0.45 0.85 Cunoniaceae Weinmannia sp4 3 0.27 1,199.66 0.27 2 0.30 0.84 Anacardiaceae Anacardiaceae sp3 2 0.18 923.79 0.21 3 0.45 0.84 Rubiaceae Rubiaceae spl 3 3 0.27 515.85 0.12 3 0.45 0.83 Sterculiaceae Dombeya sp6 3 0.27 489.50 0.11 3 0.45 0.83 Euphorbiaceae Macaranga sp2 3 0.27 477.66 0.11 3 0.45 0.83 Euphorbiaceae Antidesma petiolare 3 0.27 400.73 0.09 3 0.45 0.81 Rubiaceae Rubiaceae sp2 4 0.35 548.01 0.13 2 0.30 0.78 Lauraceae Lauraceae sp8 2 0.18 678.44 0.16 2 0.30 0.63 Rubiaceae Rubiaceae sp9 2 0.18 600.77 0.14 2 0.30 0.62 Burseraceae Canarium boivinianum 2 0.18 580.69 0.13 2 0.30 0.61 Ebenaceae Diospyros sp6 2 0.18 467.98 0.11 2 0.30 0.59 Oleaceae Noronhia spl 2 0.18 531.13 0.10 2 0.30 0.58 Rubiaceae Psychotria sp3 2 0.18 357.71 0.08 2 0.30 0.56 Rutaceae Zanthoxylum spl 2 0.18 344.08 0.08 2 0.30 0.56 Sapindaceae Macphersonia spl 2 0.18 278.23 0.06 2 0.30 0.54 Myrsinaceae Oncostemum sp4 2 0.18 265.86 0.06 2 0.30 0.54 Euphorbiaceae Drypetes sp3 2 0.18 224.32 0.05 2 0.30 0.53 Rubiaceae Psychotria sp4 2 0.18 223.44 0.05 2 0.30 0.53 Rubiaceae Rubiaceae spl 2 2 0.18 218.56 0.05 2 0.30 0.53 Euphorbiaceae Drypetes madagascariensis 2 0.18 214.61 0.05 2 0.30 0.53 Rubiaceae Rubiaceae splO 2 0.18 188.82 0.04 2 0.30 0.52 Moraceae Ficus spl 1 0.09 886.68 0.20 0.15 0.44 Myrtaceae Myrtaceae sp4 2 0.18 262.61 0.06 0.15 0.39 Rubiaceae Psychotria sp5 0.09 369.84 0.08 0.15 0.32 Anacardiaceae Protorhus sp2 0.09 330.06 0.08 0.15 0.31 Sapotaceae Sapotaceae spl 0.09 298.65 0.07 0.15 0.31 Oleaceae Noronhia sp4 0.09 280.55 0.06 0.15 0.30 Rubiaceae Rothmania spl 0.09 248.85 0.06 0.15 0.30 Annonaceae Annonaceae sp2 0.09 240.53 0.06 0.15 0.29 Violaceae Rinorea spl 0.09 226.98 0.05 0.15 0.29 Rutaceae Vepris sp4 0.09 219.04 0.05 0.15 0.29 Sapotaceae Mimusops spl 0.09 203.58 0.05 0.15 0.29 Lauraceae Lauraceae sp7 0.09 201.06 0.05 0.15 0.28 Loganiaceae Loganiaceae spl 0.09 162.86 0.04 0.15 0.28 Apocynaceae Apocynaceae spl 0.09 156.15 0.04 0.15 0.27 Myrtaceae Myrtaceae sp6 0.09 153.94 0.04 0.15 0.27 Rutaceae Vepris sp5 0.09 153.94 0.04 0.15 0.27 Rubiaceae Psychotria spl 0.09 149.57 0.03 0.15 0.27 90 FIELDIANA: ZOOLOGY Appendix 4-4 Continued. Relative Number Relative Basal domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Euphorbiaceae Phyllanthus sp2 0.09 147.41 0.03 0.15 0.27 Flacourtiaceae Tisonia sp2 0.09 147.41 0.03 0.15 0.27 Agavaceae Dracaena reflexa sspl 0.09 143.14 0.03 0.15 0.27 Anacardiaceae Anacardiaceae sp4 0.09 136.85 0.03 0.15 0.27 Moraceae Ficus sp3 0.09 134.78 0.03 0.15 0.27 Verbenaceae Vitex sp2 0.09 118.82 0.03 0.15 0.27 Euphorbiaceae Croton sp5 0.09 116.90 0.03 0.15 0.27 Rutaceae Vepris sp3 0.09 1 14.99 0.03 0.15 0.27 Annonaceae Xylopia sp2 0.09 109.36 0.03 0.15 0.26 Flacourtiaceae Tisonia spl 0.09 107.51 0.02 0.15 0.26 Euphorbiaceae Drypoetes spl 0.09 102.07 0.02 0.15 0.26 Clusiaceae Ochrocarpus sp4 0.09 93.31 0.02 0.15 0.26 Fabaceae Fabaceae sp3 0.09 93.31 0.02 0.15 0.26 Monimiaceae Monimiaceae sp2 0.09 93.31 0.02 0.15 0.26 Bignoniaceae Colea spl 0.09 91.61 0.02 0.15 0.26 Verbenaceae Clerodendrum spl 0.09 86.59 0.02 0.15 0.26 Flacourtiaceae Aphloia theaeformis 0.09 84.95 0.02 0.15 0.26 Total 1,129 100.00 436,704.16 100.00 665 100.00 300.00 Plot 4 at 1550 m Number Relative Basal Relative domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Elaeocarpaceae Sloanea rhodantha var. quercifolia 3 0.48 204,780.58 33.77 3 0.87 35.13 Euphorbiaceae Macaranga sp2 95 15.25 29,016.73 4.79 25 7.27 27.30 Elaeocarpaceae Sloanea rhodantha var. rhodantha 22 3.53 92,198.87 15.21 14 4.07 22.81 Lauraceae Lauraceae splO 66 10.59 20,897.65 3.45 23 6.69 20.73 Monimiaceae Monimiaceae spl 40 6.42 34,282.75 5.65 23 6.69 18.76 Cunoniaceae Weinmannia sp4 42 6.74 23,366.65 3.85 17 4.94 15.54 Cyatheaceae Cyathea spl 41 6.58 7,224.38 1.19 16 4.65 12.42 Meliaceae Malleastrum spl 27 4.33 24,068.97 3.97 14 4.07 12.37 Lauraceae Lauraceae spl 5 16 2.57 22,567.43 3.72 11 3.20 9.49 Sterculiaceae Dombeya sp2 23 3.69 10,069.28 1.66 14 4.07 9.42 Araliaceae Polyscias sp9 23 3.69 13,967.25 2.30 4 1.16 7.16 Myrtaceae Myrtaceae spl 5 12 1.93 12,552.64 2.07 9 2.62 6.61 Monimiaceae Tambourissa spl 7 1.12 19,905.00 3.28 7 2.03 6.44 Sterculiaceae Dombeya sp9 11 1.77 9,496.62 1.57 7 2.03 5.37 Sterculiaceae Dombeya spl 14 2.25 5,083.07 0.84 6 1.74 4.83 Clusiaceae Mammea spl 13 2.09 4,045.46 0.67 7 2.03 4.79 Sterculiaceae Dombeya sp5 9 1.44 7,144.77 1.18 7 2.03 4.66 Hamamelida - Dicoryphe spl 9 1.44 6,853.60 1.13 6 1.74 4.32 ceae Myrtaceae Myrtaceae spl 6 7 1.12 7,697.13 1.27 5 1.45 3.85 Clusiaceae Ochrocarpus sp4 10 1.61 2,307.57 0.38 6 1.74 3.73 Euphorbiaceae Croton spl 9 1.44 1,437.57 0.24 7 2.03 3.72 Lauraceae Potameia spl 6 0.96 6,905.53 1.14 5 1.45 3.56 Fabaceae Stronglyodon spl 8 1.28 1,230.35 0.20 7 2.03 3.52 Sterculiaceae Dombeya sp 1 1 7 1.12 2,823.89 0.47 6 1.74 3.33 Araliaceae Polyscias sp8 2 0.32 223.71 0.04 10 2.91 3.26 Agavaceae Dracaena reflexa sspl 7 1.12 909.27 0.15 6 1.74 3.02 Pandanaceae Pandanus spl 7 1.12 1,062.36 0.18 5 1.45 2.75 Sterculiaceae Dombeya splO 8 1.28 1,668.59 0.28 4 1.16 2.72 RAKOTOMALAZA & MESSMER: VEGETATION 91 Appendix 4-4 Continued. Relative Number Relative Basal domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Flacourtiaceae Aphloia theaeformis 7 1.12 2,004.74 0.33 4 1.16 2.62 Monimiaceae Monimiaceae sp2 5 0.80 3,038.97 0.50 4 1.16 2.47 Rubiaceae Rubiaceae spl5 5 0.80 936.27 0.15 5 1.45 2.41 Clusiaceae Ochrocarpus spl 4 0.64 2,769.58 0.46 4 1.16 2.26 Clusiaceae Symphonia spl 4 0.64 1,840.36 0.30 4 1.16 2.11 Myrsinaceae Oncostemum sp8 4 0.64 677.19 0.11 4 1.16 1.92 Arecaceae Dypsis spp 3 0.48 1,570.40 0.26 3 0.87 1.61 Rubiaceae Schysmatoclada spl 3 0.48 288.75 0.05 3 0.87 1.40 Araliaceae Polyscias sp6 5 0.80 1,134.11 0.19 1 0.29 1.28 Lauraceae Lauraceae sp4 2 0.32 1,790.37 0.30 2 0.58 1.20 Lauraceae Lauraceae sp 1 1 1 0.16 3,728.45 0.61 1 0.29 1.07 Rubiaceae Ixora spl 2 0.32 950.34 0.16 2 0.58 1.06 Rhizophoraceae Rhizophoraceae spl 1 0.16 3,599.71 0.59 1 0.29 1.04 Lauraceae Lauraceae spl 2 0.32 678.78 0.11 2 0.58 1.01 Sterculiaceae Dombeya sp8 2 0.32 414.82 0.07 2 0.58 0.97 Meliaceae Malleastrum gracile 2 0.32 264.31 0.04 2 0.58 0.95 Rutaceae Vepris sp2 2 0.32 229.22 0.04 2 0.58 0.94 Monimiaceae Monimiaceae sp3 2 0.32 215.39 0.04 2 0.58 0.94 Rubiaceae Rubiaceae spl 6 2 0.32 207.74 0.03 2 0.58 0.94 Rubiaceae Psychotria sp6 2 0.32 197.43 0.03 2 0.58 0.93 Araliaceae Polyscias sp7 1 0.16 206.12 0.03 2 0.58 0.78 Verbenaceae Clerodendrum spl 2 0.32 467.98 0.08 0.29 0.69 Aquifoliaceae Ilex mitis 2 0.32 210.34 0.03 0.29 0.65 Clusiaceae Mammea sp2 0.16 907.92 0.15 0.29 0.60 Rutaceae Zanthoxylum spl 0.16 829.58 0.14 0.29 0.59 Myrtaceae Myrtaceae sp3 0.16 706.86 0.12 0.29 0.57 Euphorbiaceae Dry petes spl 1 V 0.16 480.87 0.08 0.29 0.53 Lauraceae Lauraceae sp8 0.16 422.73 0.07 0.29 0.52 Elaeocarpaceae Elaeocarpus sp3 0.16 390.57 0.06 0.29 0.52 Moraceae Ficus soroceoides 0.16 260.16 0.04 0.29 0.49 Euphorbiaceae Croton sp2 0.16 226.98 0.04 0.29 0.49 Rubiaceae Rubiaceae spl 4 0.16 224.32 0.04 0.29 0.49 Myrsinaceae Oncostemum sp 1 1 0.16 213.82 0.04 0.29 0.49 Erythroxyla- Erythroxylum sp3 0.16 143.14 0.02 0.29 0.47 ceae Moraceae Ficus sp2 0.16 136.83 0.02 0.29 0.47 Rubiaceae Pouridiantha spl 0.16 103.87 0.02 0.29 0.47 Asteraceae Vernonia spl 0.16 83.32 0.01 0.29 0.46 Total 623 100.00 606,339.31 100.00 344 100.00 300.00 Plot 5 at 1875 m Number Relative Basal Relative domi- of indi- density area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Araliaceae Polyscias sp7 153 11.38 106,034.75 16.27 10 4.35 32.00 Clusiaceae Garcinia sp2 105 7.84 27,789.75 4.26 10 4.35 16.45 Lauraceae Belschmedia spl 60 4.48 45,719.00 7.02 8 3.48 14.97 Flacourtiaceae Aphloia theaeformis 68 5.04 37,254.25 5.72 9 3.91 14.67 Elaeocarpaceae Sloanea rhodantha var. quercifolia 40 2.99 44,111.00 6.77 7 3.04 12.80 Cunoniaceae Weinmannia sp4 38 2.80 47,114.50 7.23 5 2.17 12.20 Araliaceae Polyscias sp9 73 5.41 15,948.00 2.45 9 3.91 11.77 Celastraceae Brexiella spl 75 5.60 19,788.25 3.04 7 3.04 11.68 Aquifoliaceae Ilex mitis 53 3.92 25,937.75 3.98 6 2.61 10.51 Lauraceae Cryptocarya sp3 40 2.99 24,867.00 3.82 8 3.48 10.28 Lauraceae Lauraceae spl 6 30 2.24 25,314.50 3.88 6 2.61 8.73 92 FIELDIANA: ZOOLOGY Appendix 4-4 Continued. Relative Number Relative Basal domi- of indi- densitv area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Monimiaceae Ephippiandra spl 45 3.36 11.739.00 1.80 7 3.04 8.20 Ericaceae Agauria spl 18 1.31 25.367.75 3.89 5 2.17 7.37 Cunoniaceae Weinmannia sp6 20 1.49 20.53 1 .00 3.15 5 2.17 6.82 Erythroxyla- Erythroxylum sp4 35 2.61 5.508.25 0.85 7 3.04 6.50 ceae Rubiaceae Rubiaceae spl 28 2.05 8,408.75 1.29 7 3.04 6.39 Myrtaceae Myrtaceae sp20 28 2.05 7.923.00 1.22 7 3.04 6.31 Myrsinaceae Onocostemum sp 1 2 38 2.80 5.109.50 0.78 6 2.61 6.19 Elaeocarpaceae Elaeocarpus sp4 23 1.68 8.021.00 1.23 7 3.04 5.95 Elaeocarpaceae Sloanea rhodantha van rhodantha 13 0.93 17.880.75 2.74 5 2.17 5.85 Myrtaceae Sizygium sp2 38 2.80 4.437.25 0.68 5 2.17 5.65 Verbenaceae Vitex sp4 30 2.24 4.880.75 0.75 5 2.17 5.16 Rubiaceae Rubiaceae spl 7 23 1.68 4.767.75 0.73 6 2.61 5.02 Pittosporaceae Pittosporum spl 18 1.31 6.45 1 .50 0.99 6 2.61 4.90 Myrtaceae Myrtaceae sp22 18 1.31 6.185.00 0.95 6 2.61 4.86 Lauraceae Ocotea sp3 23 1.68 8.750.25 1.34 4 1.74 4.76 Pandanaceae Pandanus sp2 20 1.49 3.772.00 0.58 6 2.61 4.68 Lauraceae Lauraceae spl 15 1.12 10.056.75 1.54 4 1.74 4.40 Flacourtiaceae Flacourtiaceae spl 15 1.12 6,560.25 1.01 5 2.17 4.30 Rubiaceae Rubiaceae spl 8 15 1.12 3,101.00 0.48 5 2.17 3.77 Sterculiaceae Sterculiaceae sp2 13 0.93 6,564.50 1.01 4 1.74 3.68 Euphorbiaceae Macaranga sp2 18 1.31 3.572.75 0.55 4 1.74 3.59 Cunoniaceae Weinmannia sp7 13 0.93 7.314.00 1.12 2 0.87 2.92 Lauraceae Lauraceae sp3 8 0.56 9.306.75 1.43 2 0.87 2.86 Araliaceae Schefflera sp2 8 0.56 7.635.00 1.17 2 0.87 2.60 Euphorbiaceae Croton sp3 10 0.75 3.571.25 0.55 3 1.30 2.60 Myrtaceae Myrtaceae spl 8 10 0.75 1.182.25 0.18 3 1.30 2.23 Sterculiaceae Dombeya spl 2 15 1.12 4,283.25 0.66 1 0.43 2.21 Rutaceae Vepris spl 13 0.93 2,631.50 0.40 2 0.87 2.21 Lauraceae Lauraceae sp6 8 0.56 4,947.00 0.76 2 0.87 2.19 Sterculiaceae Dombeya spl 2 5 0.37 4.532.50 0.70 1 0.43 1.50 Cyatheaceae Cyathea spl 5 0.37 956.75 0.15 2 0.87 1.39 Myrtaceae Myrtaceae sp21 5 0.37 524.00 0.08 2 0.87 1.32 Myrtaceae Myrtaceae spl 7 5 0.37 2.417.00 0.37 1 0.43 1.18 Rubiaceae Gaertnera sp5 5 0.37 1.019.00 0.16 1 0.43 0.96 Asteraceae Brachilaena ramiflora 3 0.19 672.00 0.10 1 0.43 0.72 Myrtaceae Myrtaceae spl 9 3 0.19 368.50 0.06 1 0.43 0.68 Rubiaceae Psychotria sp7 3 0.19 302.00 0.05 1 0.43 0.67 Euphorbiaceae Croton sp2 3 0.19 282.75 0.04 1 0.43 0.66 Meliaceae Malleastrum spl 3 0.19 259.75 0.04 1 0.43 0.66 Total 1,340 100.00 651.672.75 100.00 230 100.00 300.00 Note: The data are ranked by decreasing values of IVI for each plot. The unidentified material is omitted. RAKOTOMALAZA & MESSMER: VEGETATION 93 Appendix 4-5 Relative Density, Relative Dominance, and Relative Specific Diversity from Transect Data Collected in Parcel 2 of the RNI d'Andohahela. Number Relative of Relative Number species individ- Relative Basal area dominance of diversity Family uals density (%) (cm2) (%) species (%) FIV Burseraceae 103 20.16 7,660.77 17.53 5 6.41 44.10 Didiereaceae 47 9.20 11,793.93 26.99 4 5.13 41.32 Euphorbiaceae 72 14.09 2,855.32 6.54 8 10.26 30.88 Anacardiaceae 19 3.72 9,912.51 22.69 2 2.56 28.97 Fabaceae 39 7.63 2,490.89 5.70 10 12.82 26.15 Ebenaceae 41 8.02 882.00 2.02 4 5.13 15.17 Hernandiaceae 38 7.44 2,369.15 5.42 1 1.28 14.14 Rubiaceae 14 2.74 278.62 0.64 6 7.69 11.07 Tiliaceae 21 4.11 267.82 0.61 4 5.13 9.85 Loganiaceae 17 3.33 650.51 1.49 3 3.85 8.66 Combretaceae 8 1.57 735.92 1.68 3 3.85 7.10 Boraginaceae 7 1.37 235.62 0.54 4 5.13 7.04 Apocynaceae 11 2.15 937.18 2.14 2 2.56 6.86 Bignoniaceae 15 2.94 402.71 0.92 2 2.56 6.42 Flacourtiaceae 9 1.76 138.82 0.32 3 3.85 5.93 Lythraceae 16 3.13 371.89 0.85 1 1.28 5.26 Capparidaceae 8 1.57 1,004.52 2.30 1 1.28 5.15 Urticaceae 3 0.59 84.04 0.19 3 3.85 4.63 Verbenaceae 2 0.39 208.13 0.48 2 2.56 3.43 Rutaceae 3 0.59 75.40 0.17 2 2.56 3.32 Ptaeroxylaceae 6 1.17 140.59 0.32 1.28 2.78 Sterculiaceae 5 0.98 77.95 0.18 1.28 2.44 Meliaceae 2 0.39 40.84 0.09 1.28 1.77 Celastraceae 0.20 38.48 0.09 1.28 1.57 Erythroxylaceae 0.20 19.63 0.04 1.28 1.52 Asclepiadaceae 0.20 7.07 0.02 1.28 1.49 Malvaceae 0.20 7.07 0.02 1.28 1.49 Cucurbitaceae 0.20 4.91 0.01 1.28 1.49 Total 511 100.00 43,692.29 100.00 78 100.00 300.00 Note: The unidentified material is omitted. 94 FIELDIANA: ZOOLOGY Appendix 4-6 Relative Density, Relative Dominance, and Relative Frequency from Transect Data Collected in Parcel 2 of the KM d'Andohahela. Relative Number Relative domi- of indi- density Basal area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Anacardiaceae Operculicarya decani 13 2.54 9.815.12 22.46 8 3.94 28.95 Didiereaceae Alluaudia procera 29 5.68 6,415.13 14.68 8 3.94 24.30 Burseraceae Commiphora humberti 37 7.24 3.014.36 6.90 9 4.43 18.57 Didiereaceae Alluaudia ascendens 13 2.54 5,317.15 12.17 5 2.46 17.18 Hernandiaceae Gyrocarpus americanus 38 7.44 2,369.15 5.42 6 2.96 15.81 Burseraceae Commiphora aprevalii 22 4.31 2,286.29 5.23 9 4.43 13.97 Ebenaceae Diospyros humbertiana 34 6.65 727.67 1.67 7 3.45 11.77 Burseraceae Commiphora marchandii 17 3.33 1,465.95 3.36 8 33.94 10.62 Euphorbiaceae Euphorbia onncoclada 18 3.52 696.45 1.59 7 3.45 8.56 Burseraceae Commiphora brevicalyx 18 3.52 541.53 1.24 7 3.45 8.21 Fabaceae Dichrostachys spl 18 3.52 188.69 0.43 6 2.96 6.91 Euphorbiaceae Euphorbia intisy 21 4.11 541.34 1.24 2 0.99 6.33 Euphorbiaceae Euphorbia plagiantha 10 1.96 992.74 2.27 4 1.97 6.20 Lythraceae Lythraceae spl 16 3.13 371.89 0.85 4 1.97 5.95 Fabaceae Tetrapterocarpon geayi 5 0.98 1,807.20 4.14 1 0.49 5.61 Burseraceae Commiphorra simplicifolia 9 1.76 352.64 0.81 6 2.96 5.52 Capparidaceae Boscia longifolia 8 1.57 1,004.52 2.30 3 1.48 5.34 Euphorbiaceae Croton sp2 15 2.94 231.10 0.53 3 1.48 4.94 Tiliaceae Grewia sp2 12 2.35 166.11 0.38 4 1.97 4.70 Loganiaceae Strychnos madagascariensis 13 2.54 140.59 0.32 3 1.48 4.34 Apocynaceae Pachypodium geayi 5 0.98 857.65 1.96 2 0.99 3.93 Tiliaceae Grewia spl 5 0.98 64.60 0.15 5 2.46 3.59 Rubiaceae Rubiaceae sp2 7 1.37 103.87 0.24 4 1.97 3.58 Bignoniaceae Fernandoa madagasca riensis 7 1.37 290.99 0.67 3 1.48 3.51 Bignoniaceae Stereospermum nematocarpus 8 1.57 111.72 0.26 3 1.48 3.30 Ptaeroxylaceae Cedrelopsis grevei 6 1.17 140.59 0.32 3 1.48 2.97 Flacourtiaceae Flacourtiaceae spl 6 1.17 100.73 0.23 3 1.48 2.88 Combretaceae Terminalia monoceros 1 0.20 380.13 0.87 3 1.48 2.54 Fabaceae Fabaceae spl 7 1.37 293.15 0.67 1 0.49 2.53 Anacardiaceae Rhus perrieri 6 1.17 97.39 0.22 2 0.99 2.38 Apocynaceae Apocynaceae spl 6 1.17 79.52 0.18 2 0.99 2.34 Rubiaceae Rubiaceae spl 3 0.59 71.67 0.16 3 1.48 2.23 Fabaceae Cassia spl 3 0.59 65.19 0.15 3 1.48 2.21 Combretaceae Terminalia sp2 4 0.78 115.45 0.26 2 0.99 2.03 Didiereaceae Alluaudia dumosa 4 0.78 56.75 0.13 2 0.99 1.90 Loganiaceae Strychnos sp2 1 0.20 490.87 1.12 1 0.49 1.81 Euphorbiaceae Euphorbia stenoclada 3 0.59 277.25 0.63 1 0.49 1.71 Sterculiaceae Sterculiaceae spl 5 0.98 77.95 0.18 1 0.49 1.65 Tiliaceae Grewia sp3 3 0.59 30.04 0.07 2 0.99 1.64 Combretaceae Terminalia spl 3 0.59 240.33 0.55 1 0.49 1.63 RAKOTOMALAZA & MESSMER: VEGETATION 95 Appendix 4-6 Continued. Relative Number Relative domi- of indi- density Basal area nance Absolute Relative Family Genus species viduals (%) (cm2) (%) frequency frequency IVI Euphorbiaceae Euphorbia lecodendron 2 0.39 91.89 0.21 2 0.99 1.59 Ebenaceae Diospyros spl 2 0.39 91.11 0.21 2 0.99 1.59 Rutaceae Zanthoxylum decaryi 2 0.39 62.83 0.14 2 0.99 1.52 Meliaceae Neobeguea mahafaliensis 2 0.39 40.84 0.09 2 0.99 1.47 Flacourtiaceae Flacourtiaceae sp2 2 0.39 33.18 0.08 2 0.99 1.45 Verbenaceae Verbenaceae sp2 1 0.20 113.10 0.26 2 0.99 1.44 Boraginaceae Boraginaceae spl 3 0.59 145.30 0.33 0.49 1.41 Urticaceae Urticaceae spl 1 0.20 38.48 0.09 2 0.99 1.27 Fabaceae Fabaceae sp6 1 0.20 28.27 0.06 2 0.99 1.25 Ebenaceae Diospyros quartzitarum 3 0.59 45.75 0.10 0.49 1.18 Loganiaceae Strychnos spl 3 0.59 19.05 0.04 0.49 1.12 Boraginaceae Boraginaceae sp2 2 0.39 58.12 0.13 0.49 1.02 Euphorbiaceae Croton spl 2 0.39 19.63 0.04 0.49 0.93 Ebenaceae Diospyros sp2 2 0.39 17.48 0.04 0.49 0.92 Verbenaceae Verbenaceae spl 0.20 95.03 0.22 0.49 0.91 Rubiaceae Adina microcepahala 0.20 50.27 0.12 0.49 0.80 Celastraceae Celastraceae spl 0.20 38.48 0.09 0.49 0.78 Urticaceae Obetia sp2 0.20 38.48 0.09 0.49 0.78 Fabaceae Fabaceae sp2 0.20 28.27 0.06 0.49 0.75 Fabaceae Fabaceae sp5 0.20 28.27 0.06 0.49 0.75 Rubiaceae Rubiaceae sp4 0.20 28.27 0.06 0.49 0.75 Boraginaceae Ehretia spl 0.20 19.63 0.04 0.49 0.73 Erythroxyla- Erythroxylum pervillei 0.20 19.63 0.04 0.49 0.73 ceae Fabaceae Bauhinia hildebrandti 0.20 19.63 0.04 0.49 0.73 Fabaceae Fabaceae sp3 0.20 19.63 0.04 0.49 0.73 Rubiaceae Rubiaceae sp3 0.20 19.63 0.04 0.49 0.73 Boraginaceae Ehretia sp2 0.20 12.57 0.03 0.49 0.72 Fabaceae Fabaceae sp4 0.20 12.57 0.03 0.49 0.72 Rutaceae Rutaceae spl 0.20 12.57 0.03 0.49 0.72 Asclepiadaceae Cyanchum spl 0.20 7.07 0.02 0.49 0.70 Malvaceae Malvaceae spl 0.20 7.07 0.02 0.49 0.70 Tiliaceae Grewia sp4 0.20 7.07 0.02 0.49 0.70 Urticaceae Obetia spl 0.20 7.07 0.02 0.49 0.70 Cucurbitaceae Cucurbitaceae spl 0.20 4.91 0.01 0.49 0.70 Didieraceae Alluaudia humbertii 0.20 4.91 0.01 0.49 0.70 Euphorbiaceae Croton sp3 0.20 4.91 0.01 0.49 0.70 Flacourticaeae Flacourtia spl 0.20 4.91 0.01 0.49 0.70 Rubiacae Rubiaceae sp5 0.20 4.91 0.01 0.49 0.70 Total 511 100.00 43,692.29 100.00 203 100.00 300.00 Note: The unidentified material is omitted. 96 FIELDIANA: ZOOLOGY Chapter 5 A Regional Analysis of Species Associations and Distributions of Two Caddisfly Families (Trichoptera: Hydropsychidae and Philopotamidae) in Southeastern Madagascar Francois-Marie Gibon1 and Patricia Zoe Andriambelo1 Abstract Specimens of the caddisfly (Trichoptera) families Hydropsychidae and Philopotamidae col- lected in the Reserve Naturelle Integrale d'Andohahela and surrounding areas are discussed and identified. An examination of the regional distribution of these caddisflies at the species level using correspondence analysis clearly shows a distinct faunal separation between the humid eastern forests and dry western forest habitats. Resume Des recoltes de Trichopteres Hydropsychidae and Philopotamidae, determinees au niveau de l'espece, ont ete menees au sein de la Reserve Naturelle Integrale d'Andohahela et a sa peri- pheric Le traitement des donnees au moyen de 1' analyse des correspondances met en evidence une nette separation entre les faunes de l'habitat humide de l'Est et celle colonisant la zone aride de 1' Quest. Introduction For the past 6 years a project entitled "Bioty- pologie et biodiversite des eaux continentales malgaches," jointly run by ORSTOM (Institut Francois de Recherche pour le Developpement en Cooperation) and CNRE (Centre National de Re- cherche sur l'Environnement), has been actively studying the freshwater faunas of Madagascar. The aim of this work is to understand certain as- pects of freshwater organisms living in the river and stream systems of the island, including tax- onomy, ecology, distribution, and biotic and abi- otic aspects that are related to their biogeography. Taxa that appear to be good indicators of certain ORSTOM. BP 434. Antananarivo (101), Madagas- environmental conditions and are easy to collect were chosen for detailed studies. Among these are Trichoptera (Annulipalpia) belonging to the two families Hydropsychidae and Philopotamidae. We have now obtained sufficient collections and dis- tributional information from southeastern Mada- gascar, including the Reserve Naturelle Integrale (RNI) d'Andohahela, to present a synthesis of our results. The geographical area is defined by the hydrological basins of the Mandrare, Efaho, Man- ampanihy, and all small rivers between the Man- drare to the west and the Manampanihy to the north. Our analysis emphasizes faunistic associations and linked ecological parameters. One of the main factors that we underline is the role of the eastern primary humid forests of the region, particularly the RNI d'Andohahela and surrounding areas. GIBON & ANDRIAMBELO: CADDISFLIES 97 Methods Results Sites were chosen in order to include the major climatic zones of the region (western and eastern; see Chapters 1 and 2) and the different stream orders. We also tried to work at different altitudes and in different vegetational zones (or soil occu- pations). Logistic and climatic conditions have sometimes considerably influenced our choice of sites. To date 56 sites have been sampled at least once in southeastern Madagascar. Although the database is by no means complete, it is sufficient to obtain a broad perspective and detect important ecological parameters associated with the distri- bution of these trichopterans. The study of Trichoptera at the species level is only possible with adults (especially males, be- cause most females cannot be specifically identi- fied). Generally these insects are captured using a system of light traps of two types — gas and black light. Nets are also used. General trapping meth- ods are described in more detail by Gibon et al. (1994, 1996). Samples are preserved in alcohol (70%). We have been working on an almost unknown fauna (Gibon & Elouard, 1996), and much of our ecological work depends first on working out the alpha-taxonomical aspects of these organisms. There are two main collections of Malagasy cad- disflies. The first is the Institut de Recherche Scientifique de Madagascar (IRSM) collection de- posited at the Museum National d'Histoire Natu- relle (MNHN), Paris. J. Olah (Czarvas, Hungary) is currently working on this first collection, which contains more than 220 species. He has shared his results, allowing us to coordinate taxonomic stud- ies. The second collection is at the Laboratoire de Recherche sur les Systeme Aquatiques et leur En- vironnement (LRSAE), Antananarivo, and con- tains more than 500 species. The morphology of the male genitalia of all species is studied in de- tail. This technique, which is basically a morpho- species approach, allows us to complete ecologi- cal analyses before Latin binomials are available for many of these organisms and in turn to trans- fer data to conservation managers without having to wait for long-term taxonomic publications. Information has been installed into the "Bibi- soa" database, written at the LRSAE with NOE software (Hertu & Elouard, 1997). Cartography is realized by CartoNOE software (Hertu, 1995). Data were treated by correspondence analysis (CA) (Gauch, 1992) with ADE software written by Chessel and Doledec. The following list contains the sites that have been sampled and used in the analysis presented below. For each station (= St) information is giv- en on (1) the code number of the drainage basin (12 = Mandrare, 41 = Manampanihy, 89 = Efa- ho, 108 = Tarantsy, and 109 = small coastal ba- sins of Manantenina); (2) the name of the nearest locality (when available); (3) the elevation (me- ters above sea level [masl]); (4) the longitude; (5) the latitude; and (6) names of captured genera and species or morphospecies (Table 5-1). Sampled Sites St 12-01 — Andratina at Imanombo, 213 m, 45°57'32"E, 24°20'20"S. Cheumatopsyche sp. AH; Macrostemum adpictum, Macrostemum scriptum; Chimarra spp. AH, AI, Chimarra dybowskina. St 12-03 — Mandrare at Anadabolava, 209 m, 46°18'30"E, 24°13'18"S. Cheumatopsyche spp. AH, AI, A, L; Macrostemum adpictum, Macrostemum scriptum, Macrostemum spp. C, K; Potamya sp. F; Chimarra spp. E, AH, AI, A, C, Y. St 12-04 — Unnamed small tributary at Amboa- nemba, 223 m, 46°27'45"E, 24°40'33"S. Cheumatopsyche sp. AH; Macrostemum ad- pictum; Chimarra spp. AH, AI. St 12-06— Mandrare at Ifotaka, 60 m, 46°08'14"E, 24°47'55"S. Cheumatopsyche sp. AH. St 12-07 — Unnamed small tributary at Berenty, 20 m, 46°18'14"E, 24°59'37"S. Chimarra sp. AH. Stl2-08— Sambalaly at Talakifeno, 145 m, 46°40'59"E, 24°49'55"S. Macrostemum ad- pictum, Macrostemum scriptum; Chimarra sp. AH. SH2-09— Imonty at Imonty, 175 m, 46°41'27"E, 24°48'51"S. Cheumatopsyche sp. C; Macros- temum adpictum; Chimarra dybowskina. Stl2-10 — Mananara at Betanimena, 118 m, 46°39'20"E, 24°48'17"S. Macrostemum ad- pictum, Macrostemum scriptum; Chimarra spp. AH, AI, D, O. St 12- 12— Imonty at Imonty, 500 m, 46°43'18"E, 24°49'03"S. Chimarra spp. AH, I, AF, AU, AG; Paulianodes sp. A. SH2-14 — Sahandrojo at Betenina, 325 m, 46°25'25"E, 24°25'12"S. Cheumatopsyche sp. 98 FIELDIANA: ZOOLOGY Table 5-1. Madagascar Taxonomic status of Trichoptera (Hydropsychidae and Philopotamidae) known from southeastern Species Taxonomic status Bibliography Cheumatopsyche sp. A Cheumatopsyche sp. AH Cheumatopsyche sp. AI Cheumatopsyche sp. AK Cheumatopsyche sp. AF Cheumatopsyche sp. C Cheumatopsyche sp. L Hydropsyche sp. A Leptonema conicum Leptonema milae Leptonema madagascariense Leptonema sp. E Leptonema sp. G Macrostemum adpictum Macrostemum placidum Macrostemum scriptum Macrostemum sp. C Macrostemum sp. D Macrostemum sp. K Macrostemum sp. O Polymorphanisus guttatus Potamyia sp. E Potamyia sp. F Chimarra sp. A Chimarra sp. B Chimarra sp. D Chimarra sp. E Chimarra sp. F Chimarra sp. G Chimarra sp. I Chimarra sp. O Chimarra sp. Y Chimarra sp. AE Chimarra sp. AF Chimarra sp. AG Chimarra sp. AH Chimarra sp. AI Chimarra sp. AK Chimarra sp. AN Chimarra sp. AP Chimarra sp. AQ Chimarra sp. AS Chimarra sp. AT Chimarra sp. AU Chimarra sp. AV Chimarra dybowskina Dolophilodes sp. C Paulianodes sp. A Paulianodes sp. F Paulianodes sp. K Wormaldia sp. D LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) Flint et al. (1987) Sykora (1964) Ulmer (1905) LRSAE (in prep.) LRSAE (in prep.) Navas (1935) Navas (1935) Rambur (1842) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) Navas (1935) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) Olah/LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) Olah/LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) Olah/LRSAE (in prep.) LRSAE (in prep.) Olah/LRSAE (in prep.) Olah/LRSAE (in prep.) Navas (1931) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) LRSAE (in prep.) 1,3 1 2 2 2,3 3 2 2 2 1, 2, 3 3 1, 3 1, 3 Key to references: 1, Elouard et al. (1994); 2, (1987); and 5. Barnard (1980). Gibon et al. (1996); 3, Gibon and Elouard (1996); 4, Flint et al. GIBON & ANDRIAMBELO: CADDISFLIES 99 AH; Macrostemum adpictum; Chimarra spp. AH, AI. Stl2-15— Bezavo at Iloty, 525 m, 46°36'32"E, 24°38'10"S. Cheumatopsyche spp. AH, A, C; Macrostemum adpictum; Chimarra spp. AS, AH, AI. St 12- 16 — Marotoko upstream of Hazofotsy, 98 m, 46°35'46"E, 24°48'47"S. Cheumatopsyche sp. AH, C; Macrostemum adpictum; Chimarra sp. AH. St 12- 17 — Mananara at Hazofotsy, 98 m, 46°35'46"E, 24°48'57"S. Cheumatopsyche sp. C; Macrostemum adpictum; Chimarra sp. AH. St 12- 19 — Mananara-Sud near Amboasary, 46 m, 46°26'34"E, 24°51'03"S. Cheumatopsyche sp. AH; Macrostemum adpictum. St 12-20 — Marotoko, 2 km north of Mananara, 275 m, 46°38'50"E, 24°44'02"S. Cheumatop- syche spp. AH, C; Macrostemum adpictum, Macrostemum scriptum; Chimarra spp. AH, AI. St 12-21 — Tributary of Mananara at Amboanem- ba, 223 m, 46°27'45"E, 24°40'40"S. Cheu- matopsyche sp. AH; Macrostemum adpictum; Chimarra spp. AH, AI. St 12-22 — Manambolo at 7 km north of Berohan- ga, 440 m, 46°35'11"E, 24°35'07"S. Cheu- matopsyche sp. AH; Macrostemum adpictum, Macrostemum scriptum; Chimarra spp. AS, AH, AI. St 12-23 — Bezavo at Berohanga near Lotibe, 550 m, 46°36'07"E, 24°38'57"S. Cheumatopsyche spp. AH, A; Macrostemum adpictum, Ma- crostemum scriptum; Chimarra spp. AS, AH, AI, AK. St 12-25 — Small tributary of Mandrare between Tranomaro and Tsivory, 280 m, 46°24'25"E, 24°24'27"S. Cheumatopsyche sp. AH; Ma- crostemum adpictum; Chimarra spp. AH, AI. St 12-26 — Small unnamed tributary at Tsivory, 324 m, 46°00'21"E, 24°06'43"S. Cheumato- psyche sp. AH; Macrostemum adpictum, Ma- crostemum scriptum; Chimarra sp. AH. St 12-27 — Sakamamba at Imanombo, 340 m, 45°45'59"E, 24°28'32"S. Cheumatopsyche sp. AH; Macrostemum adpictum, Macrostemum scriptum; Chimarra spp. AH, AI, Chimarra dybowskina. St 12-29 — Antalimanga at Besomosoy, 272 m, 46°27'59"E, 24°05'45"S. Cheumatopsyche spp. AH, AI, A; Macrostemum adpictum; Chimarra spp. AH, AI. St 12-30 — Small unnamed tributary near Andaza, 315 m, 46°34'05"E, 24°03'16"S. Cheumatop- syche spp. AH, AI, A; Macrostemum adpic- tum, Macrostemum scriptum; Potamyia sp. F; Chimarra spp. AH, AI. St 12-31 — Tributary of Sohitay at Ankazomanga, 430 m, 46°37'23"E, 24°02'37"S. Cheumato- psyche spp. AH, A, C; Macrostemum adpic- tum, Macrostemum scriptum; Potamyia sp. F; Chimarra spp. AH, B, O. St 12-3 3 — Manambolo tributary of Mandrare at Maromby, 345 m, 46°34'39"E, 24°23'36"S. Cheumatopsyche sp. AH; Macrostemum ad- pictum; Potamyia sp. F; Chimarra spp. AH, AI. St 12-34 — Abetolo tributary at Esira, 400 m, 46°41'07"E, 24°18'00"S. Cheumatopsyche sp. AH; Macrostemum adpictum; Chimarra spp. AH, AI. St 12-35 — Anatranatra between Esira and Maroas- ara, 325 m, 46°39'04"E, 24°17'37"S. Cheu- matopsyche sp. AH; Macrostemum adpictum; Potamyia sp. F, Chimarra spp. AH, AI, O. SH2-36 — Betroky at Tranomaro, 260 m, 46°28'30"E, 24°35'47"S. Cheumatopsyche sp. AH; Macrostemum adpictum; Chimarra sp. AH. St 12-37 — Esomony at Esomony, 475 m, 46°37'28"E, 24°30'53"S. Cheumatopsyche sp. AH; Macrostemum adpictum; Chimarra sp. AH, Chimarra dybowskina. Stl2-38 — Sakamalio in the RNI d'Andohahela, 750 m, 46°40'56"E, 24°32'07"S. Cheumato- psyche spp. AH, A, AF, C; Leptonema coni- cum, Leptonema sp. G; Macrostemum scrip- tum; Potamyia sp. E; Chimarra spp. AS, I, AV, Chimarra dybowskina, Chimarra sp. AU. St 12-39— Sakamalio in the RNI d'Andohahela, 725 m, 46°40'49"E, 24°32'13"S. Cheumato- psyche sp. A; Macrostemum adpictum, Ma- crostemum scriptum; Potamyia sp. E; Chi- marra sp. I, Chimarra dybowskina, Chimarra sp. AU. St41-01 — Tributary of Manampanihy at Fenoevo, 72 m, 46°53'39"E, 24°41'00"S. Macrostemum scriptum; Chimarra spp. AS, AV, AK, Chi- marra dybowskina, Chimarra spp. AF, A, F, G; Paulianodes sp. A. St41-05 — Manampanihy at Manantenina (ferry crossing), 2 m, 47°18'57"E, 24°16'08"S. Ma- crostemum scriptum; Chimarra spp. AK, A. St41-06 — Manampanihy at Enosiary, 98 m, 46°49'19"E, 24°40'37"S. Cheumatopsyche sp. A; Macrostemum placidum, Macrostemum 100 FIELDIANA: ZOOLOGY scriptum, Macrostemum sp. D; Potamyia sp. E; Chimarra spp. AH, AN, AQ. St41-07— Andranohela at Bevoay, 98 m, 46°49'25"E, 24°40'00"S. Cheumatopsyche sp. A; Macrostemum placidum, Macrostemum scriptum; Potamyia sp. E; Chimarra dybow- skina, Chimarra spp. AN, AQ. St41-09 — Andranohela at camp 1 in the RNI d'Andohahela, 440 m, 46°45'34"E, 24°36'43"S. Cheumatopsyche sp. L; Hydropsyche sp. A; Leptonema milae; Macrostemum sp. O; Poly- morphanisus guttatus; Chimarra spp. AH, AP, AG; Paulianodes sp. E St4 1-12 — Andranohela at camp 2 in the RNI d'Andohahela, 810 m, 46°44'25"E, 24°35'47"S. Leptonema milae; Macrostemum sp. O; Pauli- anodes sp. A. St41-13 — Tributary of Andranohela at camp 2 in the RNI d'Andohahela, -810 m, 46°44'09"E, 24°35'40"S. Leptonema sp. E; Macrostemum sp. O. St4 1-15 — Andranohela at camp 2 in the RNI d'Andohahela -810 m, 46°44'19"E, 24°35'- 33"S. Leptonema milae; Macrostemum sp. O; Chimarra sp. E St89-01— Efaho at Ifarantsa, 20 m, 46°52'12"E, 24°55'37"S. Cheumatopsyche spp. A, C; Ma- crostemum adpictum, Macrostemum placi- dum, Macrostemum scriptum; Chimarra spp. AH, AT. St89-02— Efaho at Soanierana, 20 m, 46°52'07"E, 24°48'20"S. Cheumatopsyche sp. A; Macro- stemum adpictum, Macrostemum scriptum. St89-03 — Small unnamed tributary at Ranopiso I Ambany, 45 m, 46°40'23"E, 25°02'13"S. Cheumatopsyche sp. AH; Macrostemum scriptum; Chimarra sp. AH, Chimarra dy- bowskina, Chimarra spp. B, O. St89-04 — Ambahibe at Isaka-Ivondro, 50 m, 46°51'46"E, 24°47'03"S. Macrostemum scrip- turn; Chimarra sp. AV, Chimarra dybowski- na, Chimarra sp. A. St89-05 — Ambahibe at Isaka-Ivondro, 70 m, 46°51'53"E, 24°46'47"S. Cheumatopsyche sp. AF; Leptonema conicum; Macrostemum pla- cidum, Macrostemum scriptum; Potamya sp. E; Chimarra spp. AH, A, Chimarra dybow- skina, Chimarra sp. O. St89-06— Ambahibe in the RNI d'Andohahela, 330 m, 46°51'09"E, 24°45'07"S. Cheumato- psyche sp. AK; Dolophilodes sp. C; Pauli- anodes spp. A, F; Wormaldia sp. D. St89-07— Ambahibe in the RNI d'Andohahela, GIBON & ANDRIAMBELO: CADDISFLIES 100 m, 46°51'39"E, 24°46'23"S. Paulianodes sp. F. St89-08 — Ambahibe at Ezoambo, 25 m, 46°51'59"E, 24°49'10"S. Cheumatopsyche sp. A; Macrostemum adpictum, Macrostemum placidum, Macrostemum scriptum. St89-09 — Small unnamed tributary at Manamba- ro, 20 m, 46°49'35"E, 25°01'27"S. Macros- temum adpictum, Macrostemum scriptum; Chimarra sp. AH. St89-10 — Antsanira at Ranopiso II Ambony, 100 m, 46°39'30"E, 25°01'27"S. Macrostemum scriptum; Chimarra dybowskina. St89-ll— Ambahibe in the RNI d'Andohahela, 200 m, 46°51'07"E, 24°46'17"S. Paulianodes spp. F, K. St89-12 — Small unnamed tributary at Soanierana, 20 m, 46°52'28"E, 25°00'10"S. Chimarra sp. AH. St89-13 — Small unnamed tributary, 120 m, 46°52'11"E, 24°46'38"S. Macrostemum ad- pictum, Macrostemum placidum, Macroste- mum scriptum; Potamyia sp. E; Chimarra sp. AH, AV, AK, Chimarra dybowskina, Chi- marra sp. O. St 108-01— Tarantsy at Bevilany, 75 m, 46°35'28"E, 25°00'13"S. Chimarra spp. AH, A. St 108-04 — Tarantsy at Antsovela, 20 m, 46°28'12"E, 25°04'47"S. Chimarra sp. AH. St 109-02 — Antorendrika at Belavenoka, 20 m, 47°05'02"E, 24°50'18"S. Chimarra sp. C. St 1 09-04— Anandrano, 12 m, 46°58'53"E, 24°56'43"S. Chimarra sp. A. Information on the stream ecology of each site is presented in Table 5-2. As usual in studies as- sociated with the distribution of organisms on Madagascar, the first major separation of Trichop- tera groups falls out according to eastern (humid) and western (dry) habitats. For this reason sub- sequent analyses have been conducted separately for each habitat type, and also the two trichopter- an families have been separated. Four factorial di- agrams are presented corresponding to: (1) Hy- dropsychidae species of the Mandrare and Tar- antsy basins plotted according to the 1st and 2nd axes (Fig. 5-1); (2) Philopotamidae species of the Mandrare and Tarantsy basins plotted according to the 1st and 2nd axes (Fig. 5-2); (3) Hydro- psychidae species of the eastern basins plotted ac- cording to the 3rd and 4th axes (the 1st and 2nd axes isolate, respectively, Leptonema madagas- cariense and Cheumatopsyche sp. AK, both of 101 Table 5-2. Various parameters associated with each station sampled. Water Altitude Distance* Width temperature Gallery Station (m) (km) (m) (°C) forestf Habitat around station St 108-01 75 17 60 20 - spiny forest St 108-04 20 36 6 25 - spiny forest St 12-01 213 48 20 21 - wooded savannah St 12-03 209 110 50 30 - wooded savannah St 12-04 223 4 0.2 22 - spiny forest St 12-06 60 210 200 22 - spiny forest St 12-07 20 235 250 23 - spiny forest St 12-08 145 8 3 19 - xerophilous forest St 12-09 175 8.5 10 19 - steppe SU2-10 118 16 12 19 + spiny forest SU2-12 500 1 5 17 + xerophilous forest SH2-14 325 1 1.8 20 - steppe Stl2-15 525 7.5 1.5 20 - spiny forest SU2-16 98 23 4 19 - spiny forest SU2-17 98 24.5 6 18 - spiny forest Stl2-19 46 41.5 8 22 - spiny forest St 12-20 275 11 5 16 - xerophilous forest St 12-21 223 32 4 16 - spiny forest SH2-22 440 19 3 16 + grassland SH2-23 550 6 2 18 + grassland St 12-25 280 13 0.5 19 - steppe St 12-26 324 9.5 1 21 - grassland SU2-27 340 18.5 3 24 - wooded savannah St 12-29 272 11 0.3 27 - steppe St 12-30 315 13 12 25 - steppe SH2-31 430 2 0.1 25 + steppe SH2-33 345 50 7 25 - wooded savannah Stl2-34 400 9 0.1 27 + steppe Stl2-35 325 21.5 0.4 25 + steppe Stl2-36 260 19 15 24 - grassland SH2-37 475 3 1 24 + spiny forest SH2-38 750 3 1 22 + grassland SH2-39 725 8 12 21 + grassland St41-01 72 10.5 5 21 - grassland St41-05 2 90 400 25 - grassland St41-06 98 18 35 19.5 - grassland St41-07 98 17 15 23 - grassland St41-09 525 5 15 21 + primary humid forest St41-12 850 4.5 10 17 + primary humid forest St41-13 925 3.5 4 18 + primary humid forest St4 1-15 900 4 5 18 + primary humid forest St89-01 20 27 20 26 - steppe St89-02 20 38 200 20 + spiny bush St89-03 45 4 2.5 20 + steppe St89-04 50 6.5 7 20 + secondary humid forest St89-05 70 3.5 5 26 + secondary humid forest St89-06 330 3 6.5 16 -j. secondary humid forest St89-07 100 5 4.5 19 + secondary humid forest St89-08 25 11 10 20 - spiny bush St89-09 20 16 50 17 - spiny bush St89-10 100 2 2 20 - grassland St89-ll 200 1 11.5 16.5 + secondary humid forest St89-12 20 33.5 110 17 - spiny bush St89-13 120 3 4 24 + secondary humid forest St 109-02 20 12.5 40 23 - wooded savannah St 109-04 12 3 10 23.5 - grassland See text for explanation of station acronyms. * Distance refers to estimated or measured distance of sampling site from sources. t Gallery forest refers to the presence (+) or absence (-) of forest along the banks of the river or stream in the vicinity of the sampling station. 102 FIELDIANA: ZOOLOGY Ce.C Group 2 : altitude species Ce.AF . I .conic um L.G Pt.E H.scriptua " Ce.A Group 1 : Tolerant species a M.adpic turn ■ Ce.AH ■ Pt.F ■ Ce.AI Group 3 : Anadabolava group Ce.L ■ H.K H.C Fig. 5-1. Hydropsychidae species of the Mandrare and Tarantsy basins plotted according to the first and second axes of correspondence analysis. Generic codes: Ce. = Cheumatopsyche, H. = Hydropsyche, Pt. = Potamyia, M. = Macrostemum, L. = Leptonema, Po. = Polymorphanisus, Ci. = Chimarra, W. = Wormaldia, D. = Dolophilodes, and PI. = Paulianodes. ■ Ci.MB Ci.AS ■ Ci.dybowskina ■ Ci.AI Ci.AV ' PI. A ' ci Ci.AF Ci.AG Group 2 altitude species ■ Ci.MO ■ Ci.D Ci.AK ■ Ci.AI (Group 4 ]) Ci.AH Group 1: Tolerant species Group 3: Anadabolava group ■ Ci.A Ci.Y ■ Ci.E Ci.C Fig. 5-2. Philopotamidae species of the Mandrare and Tarantsy basins plotted according to the first and second axes of correspondence analysis. For a key to the generic codes, see the legend to Figure 5-1 . For species abbreviations see Table 5-1. GIBON & ANDRIAMBELO: CADDISFLIES 103 ■ L.E (Forest study sites J ■ M.O i . Madagascar iense Ce.AK ■ L.milae Ce.L Po . guttatus ' H.A Ce.AH H.N C Deforested or savannah areas ■ M.scriptum ■ M.adpictum Ce.C- ce.A M-D * M.placidum Pt-E ' r aC ■ Ce.AF L. con i cum ■ Ce.G Fig. 5-3. Hydropsychidae species of the eastern basins plotted according to the third and fourth axes of corre- spondence analysis. For a key to the generic codes see the legend to Figure 5-1. For species abbreviations see Table 5-1. which were only captured once at stations where no other species occurred) (Fig. 5-3); and (4) Phil- opotamidae species of eastern basins (from Efaho to Manampanihy) plotted according to the 2nd and 3rd axes (Fig. 5-4) (the 1st axis isolates Chi- marra sp. C, which was captured only once on a small coastal tributary where no other species oc- curred). The geographic distribution of some represen- tative species of Philopotamidae (Fig. 5-5 — Chi- marra spp. and Paulianodes spp.) and Hydro- psychidae (Fig. 5-6 — Cheumatopsyche spp., Ma- crostemum spp., Leptonema, Polymorphanisus, Hydropsyche, and Potamyia) are presented. Analysis and Discussion Hydropsychidae, Western Region Three groups of species related to three differ- ent ecological conditions can be distinguished from the analysis (Fig. 5-1). The first group {Macrostemum scriptum, M. ad- pictum, Cheumatopsyche spp. AH and C) is com- posed of widespread species that are broadly dis- tributed in Madagascar, especially in the western and central regions of the island; they are able to endure significant variation in water level change between the dry and wet seasons and heavy loads of suspended material. One, two, or more consti- tuting species of this group can be found over the complete scale of stream orders, and nearly across the whole basin (Fig. 5-6). Within the western re- gion there are two localized exceptions that are presented as the second and third groups. The second group (Cheumatopsyche sp. AF, Leptonema conicum, and Leptonema sp. G) is composed of species captured on the western slopes of Pic Trafonaomby (Fig. 5-6), where sev- eral variables, including lower temperatures, high- er water velocities, heavier and more regular pre- cipitation, and vegetation along the watercourse, create better ecological conditions for the benthic fauna. The third group (Cheumatopsyche sp. L, Ma- crostemum spp. K and C) was characterized by only one station along a medium-sized stream (Fig. 5-6). This atypical element, called the An- adabolava group (from the name of the nearest locality), occupies a quite peculiar intermediate situation. Upstream, the rivers and their smaller tributaries undergo brutal level variations due to irregularity of rains and associated heavy loads of 104 FIELDIANA: ZOOLOGY PI .A D.C W.D (^Forest study sites j ■ Pl.F Pl.K Ci.AG Ci.AP ■ Ci.F ■ Ci.G Ci.AF Ci.AS ■ Ci.A ■ Ci.AV . Ci.AK [Deforested or Ci.C ^savannah areas Ci.dybowskina ■ Ci.O ■ Ci.B ■ Ci.AH Ci.AT Ci.AN Ci.AQ C clear waters group J Fig. 5-4. Philopotamidae species of the eastern basins (from Efaho to Manampanihy) plotted according to the second and third axes of correspondence analysis. For a key to the generic codes see the legend to Figure 5-1. For species abbreviations see Table 5- 1 . transported sediments. The fauna consists of a group 1 species (Cheumatopsyche sp. AH) that appears to be less particular in its habitat require- ments. Downstream the river broadens, deepens, and rapids become rare; thus the conditions are less favorable for rheophile species. In the area of Anadabolava a large portion of the coarse sus- pended matter has already settled, and the river's output is more regular than on smaller tributaries although conditions are not yet really potamic. It is in this habitat that a narrow ecological niche probably exists. The zone described is poorly rep- resented in this study but is suggested by the pres- ence of three species found nowhere else on the Mandrare River. Philopotamidae, Western Region This analysis is a little more complicated than the previous one, but the general structure is sim- ilar (Fig. 5-2). The first group is composed of two cosmopol- itan and ubiquitous species {Chimarra spp. AH and AI). These two species are very tolerant of turbidity and water level variation, and they have broad distributions, with the exception of the highest altitudes (Fig. 5-5). The second group is constituted of species found on the higher slopes. These organisms were found on the western slope of Trafonaomby (as for Hydropsychidae) and on the high tributaries of the Mananara River. Chimarra sp. AS and C. dybowskina constitute a transition group between the two situations. The third group {Chimarra spp. A, Y, C, and E) characterizes the Anadabolava study site (Fig. 5-5); there is a direct parallel to the Hydropsychi- dae and strong support for the preceding analysis. Three of the Philopotamidae species were found nowhere else during this study on the Mandrare River, and two of them {Chimarra spp. V and E) were scarce and probably characteristic of these clear, warm rivers. The fourth group forms a distinct cluster and is composed of Chimarra spp. B, O, and D. On the basis of current information, this group does not have an equivalent among the Hydropsychidae. This atypical association, which is clearly distin- guished from Chimarra sp. AS and C. dybow- skina, can nevertheless occur in the same geo- graphical and morphological conditions. Our hy- GIBON & ANDRIAMBELO: CADDISFLIES 105 Sampling sites and watercourses -Tolagnaro Amboasary sud Map I Chknarra sp. E Chknarra ap. I • Chknarra sp. AH Map IV Map III PauSanodea ap. A MapV ■ Chknarra ap. AS a) Chrnarm sp. 0 PauSanodea sp. F Map VII Fig. 5-5. Distributions of some representative species of Philopotamidae. Map I shows the watersheds of the region. Key to localities figured on Map II: 1 — Mandrare Basin; 2 — Manampanihy Basin; 3 — Ebakika Basin; 4 — Efaho Basin; 5 — Tarantsy Basin; 6 — small coastal basins; 7 — RNI d'Andohahela (parcel 2); and 8 — RNI d'Andohahela (parcel 1). Maps III-VI show the distributions of some representative species of Philopotamidae. pothesis is that these species occupy sites whose vegetation (not the riparian vegetation, but that of the drainage area) is preserved. This hypothesis is supported by the few available data from other basins. For example, before this work Chimarra sp. B was recorded only from small watercourses in the Zombitse Forest near Sakaraha (Elouard et al., 1994). 106 FIELDIANA: ZOOLOGY • Cheumatopsyche ap. AH • Macrostemum adpictum Map I Map II • Macrostenxjm scnptum • Leptonema ap. G m Cheumatopsyche ap. AK Map III Map IV Macrostemum sp. C Potymorphanisus guttatus • Hydropsyche ap. A U Potamyia ap. F MapV MapVI Fig. 5-6. Distributions of some representative species of Hydropsychidae (Maps I-VI). Hydropsychidae, Eastern Region The first axis of Figure 5-3 distinguishes two groups: The first group corresponds to species living in the primary forest zones and the upper tributaries of the Manampanihy River. These species are roughly stratified according to elevation; the sec- ond axis reflects an altitudinal gradient starting at about 500 m, where the genera Potymorphanisus and Hydropsyche were captured, and ending at 900 m, where Leptonema sp. E was obtained. The data are not extensive, and species richness along these slopes is probably underestimated. Nonethe- less there is an altitudinal effect similar to that found on the Andringitra Massif (Gibon et al., GIBON & ANDRIAMBELO: CADDISFLIES 107 1996). The significance of this result is funda- mental for conservation — the eastern primary for- ests have an aquatic fauna that is unique and in- tolerant to changes in vegetational structure. The second group corresponds to species as- sociated with secondary forest, savannah, or ag- ricultural areas. This fauna includes western spe- cies belonging to the first group, some of which penetrate slightly into the eastern area (e.g., Cheu- matopsyche sp. AH), whereas others are common and broadly distributed (Macrostemum spp.). The species separation is based on the same gradient observed in the western region (from Cheumato- psyche sp. AH to Leptonema conicum and Cheu- matopsyche sp. AF). The interpretation of this gradient, which is clearly linked with the altitude in the west region, is more ambiguous in the east; most of these sites are located between 50 and 120 m. Apparently, species restricted to high al- titudes in the Mandrare Basin can also be found at much lower elevations in the east, where pre- cipitation is more abundant and regular through- out the year. Once again, vegetation seems to play a major role. Cheumatopsyche sp. AH occurs in the agricultural coastal plain and Leptonema con- icum and Cheumatopsyche and sp. AF occur in grassland and secondary forests. Philopotamidae, Eastern Region The results from this group confirm those from the Hydropsychidae, although new aspects of groupings appear that are more difficult to ex- plain. The first group is fauna of the primary forest. It is in the strict sense a world apart. Within this group each station has a unique fauna. As for the Hydropsychidae, this indicates that the data are probably incomplete and that speciosity is under- estimated. The second group is the fauna of secondary for- est, savannah, or agricultural areas. Several of the sites have elements of the western fauna, but data are incomplete because some of the species were collected only once or twice. Furthermore, the en- vironments are very diverse: they include a mo- saic of secondary forests, cultivations, meadows, rice fields, etc. Thus, to properly analyze these different habitats, more samples are needed. Even given these limitations in the data, several points are worth mentioning. Three eastern spe- cies (Chimarra spp. AN, AQ, and T) in this sam- ple do not occur in the western region. Their oc- 108 currence is correlated with the presence of forest or with high elevations, but they are apparently associated with relatively clear waters. Our hy- pothesis is that they are found in savannah with low water turbidity either because they have de- scended from the upstream forest zone or because the vegetation, although for the greater part grass, is sufficiently dense to bind the soil and avoid erosion. These species are clearly distinguished from Chimarra sp. AS or Chimarra dybowskina, both of which are generally associated with waters of high turbidity (e.g., in agricultural areas). Conclusion The longitudinal species distribution within hy- drographical systems constitutes a fundamental ecological problem, but there are few studies of such situations from tropical zones (see Botosa- neanu, 1979, and Malicky & Chantaramungkol, 1993). This near void in information is due to limited systematic studies of invertebrates and a lack of synthetic analyses for the little data that is available. In temperate regions, according to the theory of longitudinal zonation (lilies & Botosa- neanu, 1963) or the river continuum concept (Minshall et al., 1985), faunistic transitions exist from the headwaters of rivers to their estuaries. On the basis of the analysis presented here, the same pattern holds for the hydrological systems of southeastern Madagascar. However, at least two other factors affect the distribution of species or groups of species. The first, which we call the "Anadabolava effect," is a good example of the effect of intermediate human perturbation of the environment. The existence on the Mandrare Riv- er of a rich fauna wedged between irregular mud- dy upstream tributaries and a large homogeneous downstream river is one interesting result of this study. The second is the importance of intact veg- etation and intact soil cover in the drainage area. The primary forests of RNI d'Andohahela have a rheophile fauna of Hydropsychidae and Philopo- tamidae. Local endemism for forest species of these families is very high (Gibon et al., 1996). Deforestation is followed by extreme faunistic changes, with the appearance of species resistant to high levels of suspended matter in the water. Beyond the question of the forest that was the aim of this paper and for which results are well de- fined, we suspect the existence of a species com- munity inhabiting open ecosystems at low eleva- FIELDIANA: ZOOLOGY tions, but requiring clear waters associated with weakly perturbed areas. This is an interesting top- ic for future hydrobiological research in the re- gion. Acknowledgments The lack of a firm taxonomical basis for the organisms discussed herein is a major obstacle in quickly advancing this line of research. The sam- pling process is also tedious and often difficult. We are very grateful to those who have contrib- uted and helped in many ways, in particular Fa- bienne Ranaivoharindriaka, Jean-Marc Elouard, Desire Randriamasimanana, and Theogene Pilaka. We thank Fred Bastian and Steve Goodman for correcting the manuscript. This is contribution 10 of the "Biotypologie et biodiversite des eaux con- tinentales malgaches" project, run jointly by CNRE and ORSTOM. Literature Cited Barnard, P. C. 1980. A revision of the Old World Po- lymorphanisini (Trichoptera: Hydropsychidae). Bul- letin of the British Museum, 41(2): 98-101. Botosaneanu, L. 1979. Quinze annees de recherches sur la zonation des cours d'eau: 1963-1978. Revue commentee de la bibliographic et observations per- sonnelles. Bijdragen tot de Dierkunde, 19(1): 109— 134. Chessel, D., and S. Doledec. ADE Software, Multi- variate Analysis and Graphical Display for Environ- mental Data, version 3.3. Ecologie des Eaux douces et des grands fleuves, Universite Claude Bernard Lyon I, 69622 Villeurbanne Cedex. Elouard, J.-M., F.-M. Gibon, and F. Ranaivoharindri- aka. 1994. Les Insectes aquatiques, pp. 31-40. In Goodman, S. M., and O. Langrand, eds. Inventaire Biologique — Foret de Zombitse. Recherches pour le Developpement, Serie Sciences Biologiques, Centre d' Information et de Documentation Scientifique et Technique. No. Special, Antananarivo. Flint, O., J. F. McAlpine, and H. H. Ross. 1987. A revision of the genus Leptonema Guerin (Trichoptera: Hydropsychidae: Macronematinae). Smithsonian Con- tributions to Zoology, 450: 20-24. Gauch, H. G. 1992. Statistical Analysis of Regional Yield Trials. Elsevier, Amsterdam, 278 pp. Gibon, F-M., and J.-M. Elouard. 1996. Etude preli- minaire de la distribution des insectes lotiques a Mad- agascar (exemples des Trichopteres Philopotamidae et Dipteres Simuliidae), pp. 507-516. In Lourenco, W. R., ed. Biogeographie de Madagascar. Editions OR- STOM, Paris. Gibon, F.-M., J.-M. Elouard, and R. M. Andriamihaja. 1994. Biotypologie des cours d'eau. Aspects theo- riques et developpements actuels. Bulletin de l'Academie Nationale Malgache, Cinquantenaire de TORSTOM, numero special, 2: 17-25. Gibon, F.-M., J.-M. Elouard, and M. Sartori. 1996. Spatial distribution of some aquatic insects in the Re- serve Naturelle Integrate d'Andringitra, Madagascar, pp. 109-120. In Goodman, S. M., ed. A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar: With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. Hertu. O. 1995. Logiciel CartoGODET version 2.1. D^- p6t legal no. IDDN.FR.001.340004.R.P 1995.30200 a l'Agence pour la Protection des Programmes. Hertu, O., and J.-M. Elouard. 1997. NOE, Nomen- clature Oecologique et Environnementale. Notice d'utilisation, 350 pp. Depot le'gal no. IDDN.FR. 001. 420001. R.C. 1994.30200 a l'Agence pour la Protection des Programmes. Illies, J., and L. Botosaneanu. 1963. Probleme et methodes de la classification et de la zonation ecolo- gique des eaux courantes, considerees surtout du point de vue faunistique. Mitteilungen Internationale Ver- einiging ftir Theoretische und Angewandte Limnolo- gie, 12: 1-57. Malicky, M.. and P. Chantaramongkol. 1993. The altitudinal distribution of Trichoptera species in Mae Klang catchment on Doi Inthanon, northern Thailand: Stream zonation and cool- and warm-adapted groups. Revue Hydrobiologie Tropicale, 26: 279-291. Minshall, G. W., K. W. Cummins, R. C. Petersen, C. E. Cushing. D. A. Bruns, J. R. Sedell, and R. L. Vannote. 1985. Developments in stream ecosystem theory. Canadian Journal of Fisheries and Aquatic Sciences, 42: 1045-1055. GIBON & ANDRIAMBELO: CADDISFLIES 109 Chapter 6 Proboscidoplocia (Ephemeroptera, Polymitarcyidae) from the Reserve Naturelle Integrate d'Andohahela and Surrounding Areas, With a Description of a New Species Jean-Marc Elouard,1 Michel Sartori,2 Jean-Luc Gattolliat,2 and Ranalison Oliarinony1 Abstract A new species of Proboscidoplocia is described from the Reserve Naturelle Integrate d'Andohahela, Madagascar. Two other species in the same genus, P. vayssierei and P. rnjfieuxae, were also recorded within the reserve and nearby localities. Resume Une nouvelle espece de Proboscidoplocia est decouverte dans la Reserve Naturelle Integrate d'Andohahela. Les auteurs signalent la presence de P. vayssierei et P. rujfieuxae recoltes a l'interieure et a l'exterieure de la reserve. Introduction Members of the genus Proboscidoplocia are the largest mayflies in the world and are endemic to Madagascar. Six species were recognized by Elouard and Sartori (1997). During the biological inventory of the Reserve Naturelle Integrate (RNI) d'Andohahela in 1995 a previously unde- scribed species was collected. This mayfly is de- scribed below. Proboscidoplocia mccqffertyi Elouard & Sartori, new species (Figs. 6-1 to 6-3) Description of the Male Imago Body — Length without cerci = 24.9 mm. Head — Transverse, completely black. L. = 1 ORSTOM. B.P. 434, Antananarivo (101). Madagas- car. :Musee Cantonal de Zoologie. CP 448. CH-1000. Lausanne 17. Switzerland. 0.38 mm; 1. = 1.63 mm. Black eyes, located on the lateral extremity of the head capsule. Thorax — Prothorax pale brown on the lateral tergite, whitish on the center. Its shape is troncon- ic, the narrower side behind the head. Meso- and metathoracic tergites greenish brown. Prothorax L. = 1.02 mm; 1. ant. margin = 1.06 mm, 1. post, margin = 1.7 mm; L. meso + metathorax = 13.5 mm. Forewings — Large and hyaline (Fig. 6- la). Bordering cells of the posterior margin not very abundant and quite wide. Measurements in mm: L. = 15.9; 1. = 8.4; L./l. = 1.89. Hindwings much smaller than the forewings, with few bordering cells (Fig. 6-lb). L. = 7.7; 1. = 4.4; L./l. = 1.79. L. forewing/L. hindwing = 2.06. Legs — Dark brown in color. Measurements presented in Table 6-1 and illustrations in Figure 6-2a to 6-2c. Abdomen — Light brown. L. = 10 mm. Wide at the last abdominal segment 1. = 1.16 mm. Genitalia (Fig. 6-3) — Two segmented forceps- like structures, basal segment long, L. = 2.48 ELOUARD ET AL.: PROBOSCIDOPLOCIA 111 — rTTTTTTT77 Figs. 6-1 to 6-3. Illustrations of the holotype of Proboscidoplocia mccaffertyi. (la) Right forewing, (lb) right hindwing. (2a) Leg 1, (2b) leg 2, and (2c) leg 3. (3) Genitalia of the male imago, ventral view. Table 6-1. Leg measurements (in mm) of Proboscidoplocia mccaffertyi (see Fig. 6-2). Femora Tibia Total tarsus Tarsus 1 Tarsus 2 Tarsus 3 Tarsus 4 Tarsus 5 PI P2 P3 2.1 1.6 2.1 2.9 1.6 0.96 0.32 1.2 0.96 0.9 0.4 0.5 0.38 112 FIELDIANA: ZOOLOGY Manampanihy Small coastal basins Ebakika NId'Andohahela Sampled stations (84) Tarantsy South-eastern basins Proboscidoplocia mccaffertyi Proboscidoplocia ruffieuxae Proboscidoplocia vayssierei Proboscidoplocia genus Figs. 6-4 to 6-9. Distribution of Proboscidoplocia species in the southeastern Malagasy basins. (4) Map of sampled stations and hydrographical network around the RNI d'Andohahela. (5) Localization of the southeastern basins and the position of the RNI d'Andohahela. (6) Proboscidoplocia mccaffertyi, sp. n.; (7) P. ruffieuxae; (8) P. vayssierei; (9) Proboscidoplocia spp. in general (nymphs, male and female imagos). mm; apical segment short, L. = 0.19 mm. Penis quite wholly black, very long, reaching three- quarters of the forceps length (L. = 1 .6 mm) and joined only in the most basal part. Cerci — Broken and missing. This species differs from other members of the genus Proposidoplocia essentially by the length of the penis and its nearly total separation outside of the body. The most closely related species is P. billi Elouard and Sartori, 1997. Etymology — This species is dedicated to W. P. McCafferty, one the world's specialists on Ephemeridoidea. Holotype — Sample P0538, 23 November 1995, station St41-09 (camp 1, Fig. 6-6; see Chapter 1), Manampanihy Basin, Andranohela River, 46°45'34"E, 24°36'43"S, 525 m. Body in alcohol and wings, legs, part of cerci, and geni- talia on slide preparations. Deposited in the La- boratoire d'Entomologie du Museum National d'Histoire Naturelle (MNHN), Paris. Paratypes — This species is known only from ELOUARD ET AL.: PROBOSCIDOPLOCIA 113 the holotype. Some females and nymphs of this genus were recorded at the same station where the holotype of Proboscidoplocia mccqffertyi was collected, but for the time being we cannot as- sume that they belong to the same species. Other Proboscidoplocia Species Recorded Two other previously described Proboscidoplo- cia species were recorded in the RNI d'Andohahela. Proboscidoplocia vaissyerei Elouard and Sartori, 1997, was found in three streams near camp 1 in the forest zone of the up- per part of the Manampanihy Basin. The elevation was around 525 m (Fig. 6-8). Proboscidoplocia ruffieuxae Elouard and Sartori, 1997, was ob- tained at several collecting stations of the Man- ampanihy Basin, one at 100 m outside of the re- serve and the other at 525 m within a forested region of the reserve; at two sites in the Efaho Basin, at 70 and 120 m; and at one station in the upper streams of the Mandrare Basin, at 725 m (Fig. 6-7). Unidentified nymphs were recorded at numerous stations within and outside of the RNI d'Andohahela (Fig. 6-9). At this point we are un- able to determine to which species these nymphs belong. Discussion Proboscidoplocia spp. live mainly in fresh-run- ning rivers along the eastern coast and in the riv- ers of the upper basins of western Madagascar. Generally, they prefer clear forest streams with a current speed varying between 0.5 and 1.5 m/sec. Some species occur in small streams, while others frequent large rivers. In general, Proboscidoplo- cia species along the west coast occur at higher elevations than their congeners in eastern Mada- gascar. An abundance of rain, suitable air and wa- ter temperatures, and the presence of forest are probably important factors that affect the distri- bution and greater speciosity of this genus in the eastern portion of the island. The presence in the southeast of Proboscido- plocia on the Manampanihy River and close to (but outside of) a forested area is probably due to the influence of the nearby forest, which provides cooler water and a flow of rich organic material in the streams. Apart from collection sites located within forest areas of the reserve or in its imme- diate vicinity, Proboscidoplocia is absent from the lower part of the Mandrare, Manampanihy, and Efaho basins (Fig. 6-5). Thus, these organisms are completely dependent on the intact forests of southeastern Madagascar for their continued ex- istence. Acknowledgments This study represents contribution 11 of the project "Biotypologie et biodiversite des eaux continentales malgaches," run jointly by CNRE and ORSTOM. The program is financed through the French Fonds d'Aide et de Cooperation (FAC). We thank our colleagues at Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE) for their help and assis- tance. We are deeply indebted to the Ministere de la Recherche Appliquee au Developpement (MRAD) for the facilities made available to our research program. Maps were drawn with the CartoNOE software, written by O. Hertu. Literature Cited Elouard, J.-M., and M. Sartori. 1997. Proboscido- plocia, a singular plural (Ephemeroptera, Polymitar- cyidae), pp. 439-448. In Landolt, P., and M. Sartori, eds. Ephemeroptera and Plecoptera: Biology, ecology and systematics. MTL, Fribourg. 114 FIELDIANA: ZOOLOGY Chapter 7 Three New Species of Baetidae (Ephemeroptera) from the Reserve Naturelle Integrate d'Andohahela, Madagascar Jean-Luc Gattolliat,1 Michel Sartori,1 and Jean-Marc Elouard2 Abstract Two new species of Afroptilwn (Ephemeroptera: Baetidae) are described from the imaginal stage and for one species of Dabulamanzia (Ephemeroptera: Baetidae) from the nymphal and imaginal stages from the Reserve Naturelle Integrale d'Andohahela. Affinities and ecology are discussed. Resume Deux nouvelles especes $ Afroptilwn (Ephemeroptera: Baetidae) de la Reserve Naturelle Integrale d'Andohahela sont decrites a partir des imagos et une espece de Dabulamanzia (Ephemeroptera: Baetidae) a partir de la larve et de l'imago. Leurs affinites et leur ecologie sont discutees. Introduction Very few taxonomic works have been carried out on Malagasy Baetidae. Eleven species belong- ing to five genera have been described to date. Over the course of the past 6 years, a program organized by the Office de la Recherche Scienti- fique et Technique Outre-mer (ORSTOM) and Centre National de la Recherche Scientifique (CNRE) entitled "Biotypologie et biodiversite des eaux continentales malgaches" has sought to add information on the organisms inhabiting fresh- water ecosystems in Madagascar. We currently es- timate that 40 baetid species occur on the island. Three new species from the Reserve Naturelle Integrale (RNI) d'Andohahela are discussed in this paper. Two of these species are placed in the 1 Musee Cantonal de Zoologie, CP 448, CH-1000, Lausanne 17, Switzerland. 2 ORSTOM, BP 434, Antananarivo (101), Madagas- genus Afroptilwn Gillies, 1990, based on the fore- wing venational pattern, especially the presence of single intercalary veins and two spurs on the hindwings (Gillies, 1990). These two species dif- fer from species of Centroptilum Eaton, 1869, in the shape of the second and third segments of the forceps, the second segment becoming narrower at the apex and the third elongated instead of globular, as in Centroptilum (Gillies, 1990), and in the lack of a prominent median spine between the gonopod bases (McCafferty & Waltz, 1990). Generic attribution is provisional because knowl- edge of the nymphal stage is limited; such infor- mation might be helpful in understanding the re- lationships of these two species. The third species is placed in the genus Dabulamanzia Lugo-Ortiz and McCafferty, 1996. This generic allocation is justified because the larval stage has a bulbous labial segment 3 and proximal arc setae on the tibia. Furthermore, at the imaginal stage the hindwings have a hooked spur and three longitu- GATTOLLIAT ET AL.: BAETIDAE 115 Figs. 7-1 through 7-6. Afroptilum mathildae, new species; male imago. (1) forewing, (2) hindwing, (3) genitalia, (4) head in frontal view (pr = process), (5) head in lateral view, (6) head in dorsal view. Scales are in millimeters. dinal veins, and the male genitalia gonopods have a basomedial projection on segment 2 (Lugo-Ortiz & McCafferty, 1996). A major systematic revision of African Baetidae is about to be completed (McCafferty, pers. comm.). The holotypes and part of the paratype series are housed in the Musee Cantonal de Zoologie, Lausanne, Switzerland. Other paratypes are de- posited in the Museum National d'Histoire Natu- relle (MNHN), Paris, and CNRE, Antananarivo. Descriptions Afroptilum mathildae Gattolliat & Sartori, new species (Figs. 7-1 to 7-6) Male Imago Body Length (without caudal filaments )- 4.7 mm (4.3-5.2). Head — Width, 1.0 mm. Uniform light brown with a dark brown trapezoid figure between com- pound eyes (Fig. 7-6). Turbinated eyes uniformly dark brown-purple, well separated and becoming narrower posteriorly (Fig. 7-5). Well-marked pro- cess lateral to each antenna (Fig. 7-4). Antennae uniformly pale cream. Thorax — Light brown with only margins of sclerites darker brown. Forewings — Mean length, 4.8 mm (4.4-5.4); mean width, 1.9 mm (1.7-2.2); length/width ratio, 2.5. Membrane opaque except distal third of cos- tal area light gray. Pterostigma with four to six horizontal and vertical cross-veins. One interca- lary vein between longitudinal veins except be- tween subcostal and first radial veins (Fig. 7-1). Hindwings — Mean length, 0.8 mm (0.7-0.9); mean width, 0.2 mm (0.2-0.3); length/width ratio, 3.2. Fore- to hindwing ratio, 6.2. Membrane hy- aline with micropores near margin resembling a black border. Two prominent longitudinal veins 116 FIELDIANA: ZOOLOGY Table 7-1. Measurements (mm) of Afroptilum mathildae, new species. Femur Tibia Tarsus and claw PI P2 P3 0.80 0.73 0.73 1.58 0.95 0.89 1.21 0.42 0.42 joined at base. Costal projection with two spurs, upper clearly thinner than lower (Fig. 7-2). Legs — Light brown without markings. Mea- surements are presented in Table 7-1. Abdomen — Pale cream, each segment with a double black lateral line that widens at two-thirds length. Genitalia — Three-segmented gonopods, seg- mentation between first and second segment not well differentiated. Length of segments 1 and 2, 0.4 mm; that of segment 3, 0.1 mm. First segment larger than others with a brush of setae on internal apical margin. Third segment elongated, slightly grooved (Fig. 7-3). Examined Material Holotype — One male imago (no. 405-11), 23 April 1995, Antorendrika Basin, Antorendrika River, locality Belavenoka, St89-17 (LRSAE/OR- STOM station code; see Chapter 5), Madagascar, 47°05'02"E, 24°50'18"S, 20 m. Paratypes — Three male imagos (nos. 405-9, 405-12, and 405-18), 23 April 1995, Antorendrika Basin, Antorendrika River, locality Belavenoka, St89-17, Madagascar, 47°05'02"E, 24°50'18"S, 20 m. Fifty male imagos in alcohol, 23 April 1995, Antorendrika Basin, Antorendrika River, locality Belavenoka, St89-17, Madagascar, 47°05'02"E, 24°50'18"S, 20 m. Other Examined Material — Two male ima- gos (nos. 87-4 and 87-5), 13 April 1992, Mandra- re Basin, Mandrare River, St 12-03, Madagascar, 46°18'30"E, 24°13'18"S. One male imago (no. 91- 9), 15 April 1992, Manampanihy Basin, tributary of Manampanihy River, locality Fenoevo, St41- 01, Madagascar, 46°53'39"E, 24o41'00"E, altitude 72 m. One male imago (no. 526-6), 21 November 1995, Manampanihy Basin, Manampanihy River, locality Enosiary, St41-06, Madagascar, 46°49'19"E, 24°40'37"S. One male imago (no. 548-8), 29 November 1995, Manampanihy Basin, Manampanihy River, locality Enosiary, St41-06, Madagascar, 46°49'19"E, 24°40'37"S, 100 m. Ecology This species has been found in rivers and streams flowing in woody savannah, steppe, and degraded forest areas. It has been captured with evening light traps, and with morning light traps in stations at low altitude. At one station it was the most abundant mayfly, with more than 200 male imagos caught in one evening. Afroptilum gilberti Gattolliat & Sartori, new species (Figs. 7-7 to 7-12) Male Imago Body Length (without caudal filaments) — 5.0 mm (4.7-5.2). Head — Width, 1.2 mm. Light brown with a dark brown patch between compound eyes (Fig. 7-12) and brown marks on lower margin of carena between antennae (Fig. 7-10). Turbinated eyes light brown except dark brown base. Pale cream antennae with dark brown marks on segments 1, 2, and 3. Well-developed carena prolongated clearly below antennae (Fig. 7-11). Thorax — Light brown, with only margin of sclerites darker brown. Forewings — Mean length, 5.3 mm (5.1-5.9); mean width, 2.1 mm (2.0-2.1); length/width ratio, 2.5. Membrane opaque except apical third of cos- tal region light gray. Pterostigma with three to seven vertical cross-veins. One intercalary vein between longitudinal veins except sometimes be- tween subcostal, first, second, and third radial veins (Fig. 7-7). Hindwings — Mean length, 0.9 mm (0.8-0.9); mean width, 0.3 mm (0.2-0.3); length/width ratio. 3.2. Fore- to hindwing ratio, 6.0. Membrane hy- aline without micropores. Two longitudinal veins well marked and joined at the base. Two spurs clearly distinguishable (Fig. 7-8). Legs — Pale cream with brown marks on fem- ora and tibiae. Measurements are presented in Ta- ble 7-2. Abdomen — Pale cream, with a brown narrow line at distal margin of each segment and a black lateral patch in the middle of each segment. Genitalia — Three-segmented gonopods, seg- mentation between first and second segment bare- GATTOLLIAT ET AL.: BAETIDAE 117 Figs. 7-7 through 7-12. Afroptilum gilberti, new species; male imago. (7) forewing, (8) hindwing, (9) genitalia, (10) head in frontal view, (11) head in lateral view, (12) head in dorsal view. Scales are in millimeters. ly visible. Length of segment 1 and 2, 0.4 mm; segment 3, 0.1 mm. First segment very large be- coming narrow only at apex, with an apophysis at base (Fig. 7-9). Third segment elongated, un- streaked. Examined Material Holotype — One male imago (no. 546-3), 27 November 1995, Manampanihy Basin, Andrano- Table 7-2. Measurements (mm) of Afroptilum gil- berti, new species. Tarsus and Femur Tibia claw PI P2 P3 1.02 0.82 0.79 2.05 1.21 1.16 1.84 0.53 0.49 hela River, locality camp 2 (see Chapter 1), St41- 12, Madagascar, 46°44'25"E, 24°35'47"S, 810 m. Paratypes — One male imago (no. 91-6), 15 April 1992, Manampanihy Basin, tributary of Manampanihy River, locality Fenoevo, St41-01, Madagascar, 46°53'39"E, 24°41'00"S, 72 m. Other Material — One male imago (no. 341- 13), 3 June 1994, Mandrare Basin, Marotoko Riv- er, locality 2 km from Mananara, St 12-20, Mad- agascar, 46°38'50"E, 24°44'02"S, 275 m. One male imago (no. 525-7), 20 November 1995, Efa- ho Basin, Ambahibe River, locality Isaka-Ivondro, St89-05, Madagascar, 46°51'53"E, 24°46'47"S, 70 m. Ecology This species has been found in a variety of eco- logical settings from flowing streams in intact hu- mid forest to degraded forest, and at mid- to low 118 FIELDIANA: ZOOLOGY Figs. 7-13 through 7-18. Dabulamanzia duci new species; male imago. (13) forewing, (14) hindwing, (15) gen- italia, (16) head in frontal view, (17) head in lateral view, (18) head in dorsal view. Scales are in millimeters. elevations. It was seldomly captured, and only in the evening with light traps. The species is ap- parently rare and has been found only in or close to the RNI d'Andohahela (Fig. 7-31). It is almost certainly very sensitive to forest destruction and the ensuing ecological changes within the stream ecosystem. Dabulamanzia duci Gattolliat & Elouard, new species (Figs. 7-13 to 7-28) Male Imago Body Length (without caudal filaments) — 5.0 mm (4.8-5.6). Head — Width, 1.1 mm. Light brown with a dark brown patch between lateral ocelli and tur- binate eyes (Fig. 7-18). Turbinate eyes deep or- ange (becoming honey-brown after preservation and storage in alcohol), subcylindrical. Stout car- ena between antennae (height of carena equals distance between antennae; Fig. 7-17). Flagella of antennae quite long, but not always as long as shown in Figure 7-16. The peculiar shape of the head and the color of the eyes are typical of this species. Thorax — Uniformly pale cream. Forewings — Mean length, 4.6 mm (4.0-5.0); mean width, 1.9 mm (1.6-2.1); length/width ratio, 2.4. Membrane opaque except apical third of cos- tal region light gray. Pterostigma with five to sev- en vertical cross-veins. One intercalary vein be- tween longitudinal veins except between subcos- tal and first radial veins and generally also be- tween the second and third radial veins (Fig. 7-13). Hindwings — Mean length, 0.8 mm (0.7-0.9); mean width, 0.2 mm (0.2-0.3); length/width ratio, 3.7. Fore- to hindwing ratio, 5.7. Membrane hy- GATTOLLIAT ET AL.: BAETIDAE 119 0.5 21 Figs. 7-19 through 7-23. Dabulamanzia duci, new species; male nymph. (19) labrum, (20) right mandible, (21) left mandible, (22) right maxilla, (23) labium. Scale is in millimeters. aline. Two longitudinal veins well-marked and generally joined at base, distinctly more promi- nent than the nonintersecting third vein. Costal projection with only one developed spur. Small teeth covering the whole outer margin of wings (Fig. 7-14). Legs — Pale cream without mark. Measure- ments are presented in Table 7-3. Abdomen — Pale cream, with a brown, narrow, transverse line on posterior margin of each seg- ment. Genitalia — Three-segmented gonopods, seg- mentation between first and second segments barely visible. Length of segments 1 and 2, 0.4 mm; segment 3, 0.05 mm. Second segment long, with a stout apophysis bearing a brush of setae, the inner margin covered with small teeth. Third segment ovoid and elongated (Fig. 7-15). Female Imago No significant differences in size or color com- pared to the male imago. The hindwings with their spur on the costal margin and the three longitu- dinal veins are useful characters by which to rec- ognize females of this species. Nymph Body Length (without caudal filaments) — 5.6 mm. Mouthparts — Dorsal surface of labrum with fine setae along apical margin, three larger setae on each side and reaching the margin (Fig. 7-19). Canines of right mandible not fused, with very small teeth, a large brush of setae between canines and molars (Fig. 7-20). Canines of left mandible not fused (Fig. 7-21). Maxillary palpi two-seg- mented and as long as galea-lacinia, second seg- ment elongated with a few fine setae at apex. Api- cal teeth of maxillae broad, brush of setae with two or three longer ones at inner margin, no sen- sory hair behind apical teeth (Fig. 7-22). Labial palpi stout and composed of three segments par- tially fused; setae covering third segment, scarce on first and second segments, and inner margin of second segment slightly convex distally. Glossae as long as paraglossae (Fig. 7-23). 120 FIELDIANA: ZOOLOGY 26b Figs. 7-24 through 7-28. Dabulamanzia duci, new species; male nymph. (24) abdomen in dorsal view, (25) left gill III, (26a) segment VIII in dorsal view, (26b) posterior margin of segment VIII in dorsal view, (27a) left leg, (27b) claw on the left leg. Scale is in millimeters. Legs — Femora with six large bristles on outer margin and a couple of distal bristles, upper sur- face with scattered spine-like bristles; no femoral patch of setae. Tibiae with sparse scattered spine- like bristles on outer margin, a couple of bristles at apex, without a dorsal line of fine setae (Fig. 7-28c). Tarsi with a line of spine-like setae, claws with one row of about five teeth (Fig. 7-28d). Abdomen — Light brown with dark brown marks on each segment (Fig. 7-24). Each segment Table 7-3. Measurements (mm) of Dabulamanzia duci, new species. Tarsus and Femur Tibia claw PI P2 P3 1.13 0.84 0.79 1.95 1.05 1.00 2.05 0.62 0.52 with scale bases and posterior margin with trian- gular spines longer than wide (Fig. 7-27a, b). Gills present on segments I- VII, asymmetrical and rel- atively elongated, serrated at apex, tracheation with few ramifications (Fig. 7-25). Presence of cerci and median filament (broken at one-quarter length of cerci), cerci with hair only on interior side, median filament hairs on both sides (Fig. 7- 24). Examined Material Holotype — One male imago (no. 525-5), 20 November 1995, Efaho Basin, Ambahibe River, locality Isaka-Ivondro, St89-05, Madagascar, 46°51'53"E, 24°46'47"S, 70 m. Paratypes — One male imago (no. 525-6), 20 April 1995, Efaho Basin, Ambahibe River, local- ity Isaka-Ivondro, St89-05, Madagascar, 46°51'53"E, GATTOLLIAT ET AL.: BAETIDAE 121 24°46'47"S, 70 m. One female imago and four males imagos (nos. 91-10, 91-2, 91-3, 91-4, and 91-5), 15 April 1992, Manampanihy Basin, trib- utary of Manampanihy River, locality Fenoevo, St41-01, Madagascar, 46°53'39"E, 24°41'00"S, 72 m. One nymphal exuvia with the corresponding male subimago (no. 339-31) and one male imago (no. 339-18), 2 June 1994, Mandrare Basin, Man- anara River, locality Hazofotsy, St 12- 17, Mada- gascar, 46°35'46"E, 24°48'57"S, 100 m. Other Examined Material — One male imago (no. 90-8), 14 April 1992, Efaho Basin, Efaho River, locality Ifarantsa, St89-01, Madagascar, 46°52'12"E, 24°55'37"S, 20 m. One male imago (no. 341-12) and seven male individuals, 3 June 1994, Mandrare Basin, Marotoko River, locality 2 km after Mananara, St 12-20, Madagascar, 46°38'50"E, 24°44'02"S, 275 m. Three male ima- gos (nos. 343-5, 343-6, and 343-7), 5 June 1994, Mandrare Basin, Manambolo River, locality 7 km from Berohanga, St 12-22, Madagascar, 46°35'11"E, 24°35'07"S, 440 m. One male imago (no. 514-4), 19 October 1995, Rianila Basin, un- named river, locality road to Lakato, St 17-31, Madagascar, 48°21'48"E, 19°02'40"S, 1050 m. One male imago (no. 525-5), 20 November 1995, Efaho Basin, Ambahibe River, locality Isaka- Ivondro, St85-05, Madagascar, 46°51'53"E, 24°46'47"S, 70 m. One male imago (no. 569-1), 2 February 1996, Mandrare Basin, Sakamalio River, locality RNI d'Andohahela, Stl2-38, Mad- agascar, 46°40'56"E, 24°32'07"S, 750 m. Three male imagos (nos. 607-1, 607-2, and 607-3), 5 October 1996, Lokoho Basin, unnamed river, lo- cality RNI d'Marojejy (Camp 1), St44-04, Mad- agascar, 49°46'18"E, 14°25'50"S, 500 m. Three male imagos (nos. 614-1, 614-2, and 614-3), 11 November 1996, Lokoho Basin, Manantenina River, locality RNI d'Marojejy (Camp 1), St44- 03, Madagascar, 49°46'20"E, 14°26'02"S, 450 m. Two male imagos (nos. 619-2 and 619-3), one subimago male (no. 619-4), 13 October 1996, Lo- koho Basin, Manantenina River, locality RNI d'Marojejy (Camp 1), St44-03, Madagascar, 49°46'20"E, 14°26'02"S, 450 m. Ecology Dabulmanzia duci is one of the most common and widespread species of Baetidae in southeast- ern and eastern Madagascar. It has been found from sea level to above 1000 m, in streams and rivers flowing in all types of vegetation, and in- side and outside of the RNI d'Andohahela (Fig. 7-32). It was captured with evening and morning light traps, as well as with hand nets during the morning. The males fly in a quite peculiar man- ner: they apparently wait for the females in hori- zontal flight a few centimeters above flat stones or rocks. Affinities Only one related species has been described from Madagascar: Afroptilum (Centroptilum) electropterum (Demoulin, 1966). Afroptilum mathildae and A. gilberti differ from A. electrop- terum by the presence of two spurs on the hind- wings (instead of one in A. electropterum), by the shape of the gonopods, and by the size and color of the turbinate eyes. Dabulamanzia duci differs from A. electropterum by the presence of three veins in the hindwings (only two in A. electrop- terum), the shape of the genitalia, and the color of the eyes (Demoulin, 1966). Afroptilum mathildae and A. gilberti are much closer to A. decipiens subgenus Afroptilum, group sudafricanum Gillies, 1990, but the shape of the two spurs on the hindwings and especially the shape of the first and second segments of the gon- opods are quite different. Afroptilum gilberti dif- fers from A. mathildae by the color of the turbi- nate eyes, the presence of a carena between the antennae, and the absence of a brush of setae on the margin of the first segment of the gonopods. Dabulamanzia duci is closer to D. tarsale (Gil- lies, 1990), previously assigned to the tarsale group of Afroptilum. However, D. duci differs from D. tarsale by eye color and by the shape of the third segment of the gonopods, which is short- er and more ovoid in D. duci. Dabulamanzia duci is also relatively similar in morphology to D. ba- baora (Wuillot & Gillies, 1993), except for dif- ferences in the shape and the size of the first seg- ment of the gonopods. The nymph is also very similar to that of D. babaora, but the mandibles and the labial palpi are different. Dabulamanzia duci is easily distinguished from D. fica and D. helenae by the shape and number of teeth of the claws (Lugo-Ortiz & McCafferty, 1996). Dabu- lamanzia duci is easily distinguished from A. mathildae and A. gilberti by the shape of the head, the color of the eyes, the shape of the first seg- ment of the gonopods, and especially by the num- ber of longitudinal veins in the hindwings. 122 FIELDIANA: ZOOLOGY Sampled stations Afroptilum mathildae 31 32 Afroptilum gilberti Dabulamanzia duci Figs. 7-29 through 7-32. Distribution of some Afroptilum and Dabulamanzia in basins of southeastern Madagas- car. The gray area delimits parcels 1 and 2 of the RNI d'Andohahela. (29) sampled stations, (30) distribution of A. mathildae, (31) distribution of A. gilberti, (32) distribution of D. duci. Discussion The three new species of Baetidae described herein have different spatial and ecological distri- butions. Afroptilum mathildae is typically a spe- cies of low altitude, locally abundant in large riv- ers (Fig. 7-30). Its presence in other basins along the eastern coast needs to be confirmed. At the opposite extreme is A. gilberti, a species that lives in small and well-oxygenated streams; its distri- bution is directly related to the presence of forest, and it has been found only in or around the RNI d'Andohahela (Fig. 7-31). Its future existence de- pends completely on these environmental condi- tions and consequently on the preservation of the forest. Dabulamanzia duci is a widespread spe- cies, found in large and small streams along the eastern and southeastern coasts. It is known from areas within and outside of the RNI d'Andohahela (Fig. 7-32). Owing to its capacity to thrive in a variety of environments, this species is not endan- gered by forest degradation. At least 15 new spe- cies of Baetidae have been collected in recent years in southeastern Madagascar. Descriptions of these animals and information on their ecology will be the subject of future articles. Acknowledgments This study represents contribution no. 12 of the "Biotypologie et biodiversite des eaux continen- tales malgaches" project, run jointly by CNRE and ORSTOM. The program is financed through the French Fonds d'Aide et de Cooperation (FAC). We thank our colleagues at Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE) for their help and assis- tance. We are deeply indebted to the Ministere de la Recherche Appliquee au Developpement (MRAD) for the facilities made available for our research program. The text of a previous version of this chapter was substantially improved by comments by Steve Goodman and an anonymous reviewer. GATTOLLIAT ET AL.: BAETIDAE 123 Literature Cited Demoulin, G. 1966. Quelques Ephemeropteres nou- veaux de Madagascar. Annales de la Societe Ento- mologique de France, nouv. series 2(3): 71 1-717. Gillies, M. T. 1990. A revision of the African species of Centroptilum Eaton (Baetidae, Ephemeroptera). Aquatic Insects, 12(2): 97-128. Lugo-Ortiz, C. R., and W. P. McCafferty. 1996. The composition of Dabulamanzia, a new genus of Afro- tropical Baetidae (Ephemeroptera), with descriptions of two new species. Bulletin Societe Histoire Natu- relle, Toulouse, 132: 7-13. McCafferty, W. P., and R. D. Waltz. 1990. Revision- ary synopsis of the Baetidae (Ephemeroptera) of North and Middle America. Transactions of the Amer- ican Entomological Society, 116(4): 769-799. Wuillot, J., and M. T. Gillies. 1993. New species of Afroptilum (Baetidae, Ephemeroptera) from West Af- rica. Revue d'Hydrobiologie Tropicale, 26(4): 269- 277. 124 FIELDIANA: ZOOLOGY Chapter 8 Aquatic Biodiversity of Madagascar: Simulium (Diptera: Simuliidae) from the Reserve Naturelle Integrate d'Andohahela and Surrounding Areas Theogene Pilaka1 and Jean-Marc Elouard1 Abstract The distribution of species of the genus Simulium (Diptera: Simuliidae) in rivers situated inside and outside the Reserve Naturelle Integrale d'Andohahela is reviewed. No new species were discovered during the inventory of the reserve. Resume Les auteurs etudient la distribution du genre Simulium (Diptera: Simuliidae) dans les rivieres situees a l'interieure et a l'exterieure de la Reserve Naturelle Integrale d'Andohahela. Aucune espece nouvelle n'a ete decouverte a l'interieure de la reserve. Introduction The hydrographic network of southeastern Madagascar, including many rivers that have their origins in the hills making up the Reserve Naturelle Integrale (RNI) d'Andohahela, com- prises five main drainage systems: the Mandrare, the Efaho, the Tarantsy, the Ebakika, and the j Manampanihy basins. The eastern portion of this i region has, on the higher slopes, relatively intact primary forest typical of the humid zone, where- as much of the lower slopes has been cleared and contains open grassland or secondary forest. The vegetation to the west of the Anosyenne Moun- tains is distinctly drier and contains a variety of natural habitats, from spiny bush to gallery forest and human-degraded habitats (see Chapters 1 and 2). Generally the slopes along the eastern versant are steeper than those on the western ver- 1 Laboratoire de Recherche sur les Systemes Aqua- tiques et leur Environnement, ORSTOM, B.P. 434, An- tananarivo (101), Madagascar. sant. These different ecological factors have giv- en rise to various types of colonization and life history strategies by different Simulium species. The collections reported on here were carried out within the framework of the project "Bio- typologie et biodiversite des eaux continentales malgaches." Given the number of sites visited and the breadth of the collections, it is now pos- sible to estimate the distribution and species richness of several groups of aquatic insects. Here we focus on the inventory and geographic distribution of Simulium spp. existing in south- eastern Madagascar. All preimaginal forms of 57- mulium are aquatic and restricted to running wa- ter. There is considerable variation in this genus in certain life history parameters (e.g., rate of oxygen flow, pH, turbidity, temperature, habitat), and several species are rather strict in their eco- logical requirements. Furthermore, female Si- mulium are hematophagous and in some cases may be rather dependent on a specific source for blood. PILAKA & ELOUARD: SIMULIUM 125 List of Simulium Known from Basins in Southeastern Madagascar Current information on the species diversity of the Malagasy Simulium fauna indicates that it is relatively poor in comparison to that of Africa. Thirty-eight species are known from Madagascar, 32 of which have been named, with six currently being described. By comparison, 90 species are known from the African continent. The water- sheds of southeastern Madagascar contain 20 spe- cies of Simulium; this number is exceptionally high in comparison with other areas of the island. Thus, slightly over 50% of the known Malagasy species of this genus occur in the southeast. These species are: S. adersi Pomeroy, 1921; S. ambositrae Grenier and Grjebine, 1958; S. brunhesi Elouard and Ranaivoharindriaka, 1996; S. gyas de Meillon, 1951; S. imerinae Roubaud, 1951; S. impukane de Meillon, 1936; S. iphias de Meillon, 1951; S. iphias (10 filaments), undescribed; S. iphias (15 filaments), undescribed; S. iphias (17 filaments), undescribed; S. iphias (18 filaments), undescribed; S. iphias (19 filaments), undescribed; S. metecontae Elouard and Pilaka, 1996; S. neireti Roubaud, 1905; S. philipponi Elouard and Pilaka, 1997; S. ruficorne Macquart, 1838; 5. starmuhlneri Grenier and Grjebine, 1963; S. tolongoinae Grenier, 1972; S. unicornutum Pomeroy, 1920; and Simulium new species, undescribed. Discussion On the basis of an analysis of the distribution of Simulium species known from the basins of southeastern Madagascar, it is possible to distin- guish two species groups that correspond to each of the botanical ecosystems. Group 1 In this group are savannicolous and grassland species that live in tepid, medium- or slow-mov- ing waters with sandy bottoms. Larvae most often hang or cling to the floating substrates. Species in this group include Simulium adersi, S. ruficorne, S. iphias 15 filaments, S. iphias 17 filaments, S. iphias 18 filaments, S. iphias 19 filaments, S. im- erinae, S. philipponi, S. impukane, Simulium new species, and unicornutum (Fig. 8-1, maps II-V). Group 2 In this group are forest species that live in fresh waters with relatively rapid currents and occa- sional waterfalls, and often in rivers that circulate under forest galleries that cover all or nearly all of the river's width. This group includes Simulium neireti, S. ambositrae, S. starmuhlneri, S. tolon- goinae, S. iphias 10 filaments, and 5. gyas (Fig. 8-1, maps V and VI). The hydrographic system of the RNI d'Andohahela includes four main basins: the Mandrare (western part), the Tarantsy and the Efaho (southern part), and the Manampanihy (eastern part). Inside the reserve, Simulium cap- tured in the upper streams of the Mandrare are different from those recorded in watercourses of the upper part of the Manampanihy. Thus, S. ad- ersi and S. unicornutum are characteristic of the upper part of the Mandrare Basin, whereas species such as S. gyas, S. starmuhlneri, S. tolongoinae, S. neireti, and S. ambositrae occur in the upper streams of the Manampanihy Basin. Simulium starmuhlneri is a strictly forest-dwell- ing species that has only been found in small trib- utaries under a continuous covering of forest. Oth- ers, such as S. gyas, S. ambositrae, S. iphias 10 filaments, and S. tolongoinae, are found in waters that course through a discontinuous forest cover. Few species occur in microsympatry, and gen- erally one to four species are collected at each site. There is more similarity among the faunas of Simulium spp. found in open savannah or grass- land habitats than among those from forest habi- tats. This may be explained by the fact that sa- vannah rivers or those passing through open coun- try are often larger and contain a greater variety of microhabitats within the same system, whereas mountain streams are often smaller and more ho- mogeneous. Simulium starmuhlneri, S. tolongoinae, and S. neireti were collected in the forest of the RNI d'Andohahela. None of these Simulium are en- demic to the reserve, and they are known to occur at other forested sites of eastern Madagascar. This 126 FIELDIANA: ZOOLOGY w„ _ j • Sampled stations • Simuhum adersi (17) Map II B Simuhum ruficorne (9) ^P Simulium unicornutum (8) ■ Simulium iphias (15fil) ( 10) • Simulium iphias (I7fil)(5) * Simulium iphias ( I8fil) { I ) w Simulium iphias ( 1 9fil) ( 10) • Simulium brunhesi (4) ■ Simulium imerinae (4) ^ Simulium metecontae (I) ■ Simulium philipponi (5) ^ Simulium ambositrae (I) Map V . Simulium gyas (4) "V Simulium iphias (lOfil) ( I) ^ Simulium quilleverei (2) Map VI Simulium impukane (2) Simulium neireti (I) Simulium slarmuhlneri (2) Simulium tolongoinae (1) Fig. 8-1. Maps of the spatial distribution of Simulium spp. occurring in southeastern Madagascar. Areas in gray are the parcels of the RNI d'Andohahela. Map I, hydrographic network of southeastern Madagascar; Maps II to VI. distribution of Simulium spp. Key to localities figured on Map II: 1, Mandrare Basin; 2. Manampanihy Basin; 3, Ebakika Basin; 4, Efaho Basin; 5, Tarantsy Basin: 6, small coastal basins; 7 and 8. RNI d"Andohahela. PILAKA & ELOUARD: SIMULIUM 127 is in contrast to the tributaries of the Mandrare and the Manampanihy basins, which flow through open zones and from which several new species have been discovered (S. philipponi, Simulium new species, and S. brunhesi). Acknowledgments This study represents contribution 13 of the project "Biotypologie et biodiversite des eaux continentales malgaches," run jointly by CNRE and ORSTOM. The program is financed through the French Fonds d'Aide et de Cooperation (FAC). We thank our colleagues at Laboratoire de Recherche sur les Systemes Aquatiques et leur Environnement (LRSAE) for their help and assis- tance. We are deeply indebted to the Ministere de la Recherche Appliquee au Developpement (MRAD) for the facilities made available for our research program. Maps were drawn with the CartoNOE software, written by O. Hertu. S. M. Goodman and two reviewers provided useful comments on an earlier versions of this chapter. Literature Cited Elouard, J.-M., T. Pilaka, and F. Ranaivoharindriaka. 1996. Two new species of Simulium (Diptera: Simu- liidae) from the Reserve Naturelle Integrate d'Andringitra, pp. 131-135. In Goodman, S. M., ed. A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrate d'Andringitra, Mada- gascar: With reference to elevation variation. Fieldi- ana: Zoology, n.s. 85: 1-319. Grenier, P., and J. Bruhnes. 1972. Simulies (Diptera, Simuliidae) de Madagascar: Simulium tolongoinae n. sp., Simulium milloti Grenier et Doucet, Simulium pentaceros n. sp. Cahiers de l'ORSTOM, serie Ento- mologie Medicate et Parasitologic 10(1): 69-75. Grenier, P., and J. Doucet. 1949a. Simulies de Mad- agascar. I, Bulletin de la Societe de Pathologie exo- tique, 42: 127-128. . 1949b. Simulies de Madagascar. II, Bulletin de la Societe de Pathologie exotique, 42: 587-597. 1949c. Simulies de Madagascar. Memoire de lTnstitut de Recherche Scientifique de Madagascar, 3: 301-323. Grenier, P., and A. Grjebine. 1958. Remarques mor- phologiques et biologiques concernant les "Mouka- Fouhi" {Simulium neireti Roubaud, 1905, S. imerinae Roubaud, 1951) de Madagascar et S. ambositrae nom. nov. Bulletin de la Societe de Pathologie exotique, 51: 981-991. . 1963. Une Simulie nouvelle de Madagascar. Bulletin de la Societe de Pathologie exotique, 56: 1055-1062. Grenier, P., and R. Paullan. 1957. Diptera Simuliidae. Memoire de lTnstitut de Recherche Scientifique de Madagascar, 8: 20. Grenier, P., and M. C. le Roy-Moret. 1970. Simulium (Xenosimulium) neireti Roubaud, 1905 (Diptera, Si- muliidae): Description des stades preimaginaux et du male; description complementaire de la femelle. Ca- hiers de T ORSTOM, serie Entomologie Medicate et Parasitologic 8(1): 123-129. Pilaka, T, and J. M. Elouard. 1997. Biodiversite aquatique de Madagascar. 4. Description des Simulium philipponi et S. alidae (Diptera, Simuliidae). Bulletin de la Societe Entomologique de France, 102(1): 89- 92. Roubaud, M. E. 1905. Les "Mouka-fouhi" simulies nouvelles de Madagascar. Bulletin du Museum nation- al d'Histoire naturelle, Paris, 6: 424. 128 FIELDIANA: ZOOLOGY Chapter 9 Ant Diversity Patterns Along an Elevational Gradient in the Reserve Naturelle Integrale d'Andohahela, Madagascar Brian L. Fisher1 Abstract Leaf litter ant faunas were inventoried in Madagascar at 430, 800, and 1250 m in parcel 1 of the Reserve Naturelle Integrale (RNI) d'Andohahela. Within each elevational zone, survey methods involved a combination of pitfall and leaf litter sampling along a 250 m transect. From pitfall and leaf litter samples, I collected and identified 12,285 ants belonging to 25 genera and 111 species; general collecting yielded an additional 28 species. For each elevation, two species richness estimators — incidence-based coverage estimator (ICE) and first-order jackknife — gave comparable results. Species accumulation curves showed decreased rates of species detection and demonstrated the efficacy of these inventory tech- niques. Species collected and their relative abundances are presented. Species richness peaked at mid-elevation. Species turnover, complementarity, and faunal similarity measures demon- strated a division in ant communities between lowland forest at <800 m and montane forest at 1250 m. A mid-elevation peak in species richness is probably the result of the mixing of two distinct, lower and montane forest, ant assemblages. In addition, I compare complementarity and species turnover values, the number of species restricted to a locality, and the number of species shared between the RNI d'Andohahela, the RNI d'Andringitra, the Reserve Speciale d'Anjanaharibe-Sud, and the western Masoala Pen- insula. Resume Les fourmis de litieres ont ete inventories sur les elevations de 430, 480 et 1250 m dans parcelle 1 de la Reserve Naturelle Integrale (RNI) d'Andohahela a Madagascar. Pour chaque zone d' elevation, les methodes d'inventaires utilisees combinent les trous pieges et les echan- tillonnages de litieres le long d'un transect de 250 m. Avec les trous pieges et les echantillonnages de litieres. j'ai pu collecter et identifier 12.285 fourmis appartenant a 1 1 1 especes et 25 genres. Les collections generates ont acquis 28 especes en plus. Pour chaque elevation, les deux estimateurs de richesse d'especes telles que 1'ICE (Incidence-based coverage estimator) et le premier ordre jac knife ont donnes des resultats comparables. Les courbes d'accumulation d'especes ont approchees une asymptote et ont mon- trees l'efficacite de la technique d'inventaires utilisees. Les especes collectees et leur abondance relatives sont presentees. La richesse d'espece est au pic a la mid-elevation. Les mesures de la succession continue d'especes "species turnover," de la complementarite et de la similarite ont demontrees une division sur la communaute des fourmis entre forets basse de <800 m et les 1 Life Sciences Division, South African Museum, P.O. Box 61, 8000 Cape Town, South Africa. FISHER: ANT DIVERSITY 129 forets de montagnes de 1250 m. Le pic en richesse d'especes, a la mi-elevation est probable- ment le resultat de la mixture des fourmis des deux milieux distinctes: fourmis de la foret basse et ceux de la foret de montagne. En plus, j'ai fait une comparaison entre la RNI d'Andohahela, la RNI d'Andringitra, la Reserve Speciale d'Anjanaharibe-Sud, et l'Ouest du peninsule de Masoala. La comparaison est basee sur les valeurs de la complementarite et la succession continue d'especes "species turn- over," le nombre d'especes unique d'un localite, et le nombre d'especes communs pour les sites. Introduction Geographical patterns of species richness and areas of endemism are two criteria for conserva- tion assessment that require baseline information on species distributions (McNeely et al., 1990). For most invertebrates, we lack this necessary in- formation and even the practical approaches and methods to obtain it. To assign priority to areas with high species richness and endemism in Mad- agascar, we need methods to inventory the most diverse taxa. Sampling and estimation procedures for a di- verse and ecologically important group of terres- trial insects, ants, were used to assess diversity along elevational gradients in the Reserve Natu- relle Integrate (RNI) d'Andohahela. Similar meth- ods were used to inventory ants in the RNI d'Andringitra (Fisher, 1996a), and in the Reserve Speciale (RS) d'Anjanaharibe-Sud and on the western Masoala Peninsula (Fisher, 1998). I eval- uate the efficacy of the inventory methods and the effect of elevation on species richness in the RNI d'Andohahela. I compare measures of faunal sim- ilarity and complementarity for ant species across elevations sampled from 430 to 1250 m. In ad- dition, I compare the ant fauna in the RNI d'Andohahela with those of three other localities and discuss the nature and causes of geographical variation in ant diversity in eastern Madagascar. A complementary aspect of this study is the sub- stantial increase in taxonomic and ecological knowledge of the ant fauna in one of the most threatened regions of the world. Methods Study Sites Surveys were conducted between 12 November and 10 December 1992 near the northern bound- ary of the RNI d'Andohahela, parcel 1. RNI d'Andohahela comprises 63,100 ha of humid for- est within the elevational range of 350-1972 m (Nicoll & Langrand, 1989). Collection sites dif- fered from those during the 1995 inventory dis- cussed elsewhere in this volume. The ant inven- tory transects were located at: (1) 10 km NW of Enakara, 24°34'S, 46°49'E, transect at 430 m and general collecting from 400 to 450 m; (2) 1 1 km NW of Enakara, 24°34'S, 46°49'E, transect at 800 m and general collecting from 750 to 850 and 900 to 1000 m; and (3) 13 km NW of Enakara, 24°33'S, 46°48'E, transect at 1,250 m and general collecting from 1180 to 1300 m. Estimated can- opy height was 20-30 m at 430 m, 15 m at 800 m, and 6-8 m at 1250 m. There was no evidence of recent exploitation of the forest at the three transect sites. The transect at 430 m, however, was approximately 200 m from the trail along the northern boundary of the park. The trail was frequently used by local in- habitants for transporting agricultural products and cattle and subsequently showed signs of dis- turbance. Survey Methods In the RNI d'Andohahela, intensive ant surveys were conducted at three sites located at 430, 800, and 1250 m. At each elevation I used 50 pitfall traps and 50 leaf litter samples (mini-Winkler) in parallel lines 10 m apart along a 250 m transect. The site for each transect was chosen with the intent of sampling representative microhabitats found at each elevation (Palmer, 1995). Pitfall traps were placed and leaf litter samples gathered every 5 m along the transect. Pitfall traps con- sisted of test tubes with an 18 mm internal di- ameter and 150 mm long, partly filled to a depth of about 50 mm with soapy water and a 5% eth- ylene glycol solution, inserted into PVC sleeves and buried with the rim flush with the soil surface. 130 FIELDIANA: ZOOLOGY To prevent rainfall from filling the traps, an opaque, ridged piece of plastic was suspended ap- proximately 5 cm above the trap by means of a metal wire support. Traps were left in place for 4 days. I extracted invertebrates from samples of leaf litter (leaf mold, rotten wood) using a modified form of the Winkler extractor (see Fisher, 1996a, Fig. 8-1, and Fisher, 1998, Fig. 4-1). The leaf litter samples involved establishing 50 1 m2 plots sep- arated by 5 m along the transect line. The leaf litter inside each plot was collected and sifted through a wire 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 lit- ter was taken from each 1 m2 plot. At the low elevations (<800 m) litter was occasionally sparse, and sometimes less than 2 liters was taken. If the subsample plot contained a large rotten log or thick litter, 2 liters of litter was the maximum amount taken at each subsample site. This 2 liter limit was imposed because of the size of the Winkler extractor. In those sites where 1 m2 pro- vided an excess of leaf litter, the plot was subsam- pled until 2 liters of litter was obtained. Ants and other invertebrates were extracted from the sifted litter during a 48 hr period in mini-Winkler sacks (for a detailed discussion of the mini-Winkler method, see Fisher, 1996a, 1998). I also surveyed ants through general collecting, defined as any collection that was separate from the mini-Winkler 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. At each transect site, general collections were conducted for an approximately 2-day period. General collections were made within 500 m ground distance and within 75 m in elevation of each transect site. In addition, general collecting was conducted between 900 and 1000 m. General collections included samples of the arboreal ants found on low vegetation that were not sampled by pitfalls or leaf litter. Ants sampled with general collection methods, therefore, were not used in the analysis of the efficacy of the sur- vey of the leaf litter ants, of faunal similarity, or complementarity. Sample Processing For every 50-station transect, which took from 5 to 7 field days to conduct, an average of 1 month was spent in the laboratory sorting, iden- tifying, and curating specimens. After I returned from the field, ant specimens from the pitfall and leaf litter samples were sorted. The saturated salt water extraction procedure described below was very effective in removing organic matter from inorganic matter in the leaf litter samples. Each sample was emptied into a 40 mm diameter, 250 ml graduated cylinder. A near boiling saturated salt water solution was then added until the cyl- inder was half filled. After 2 minutes the solution was stirred. After settling for about 2 minutes the organic matter was decanted off the top into a strainer and rinsed with water, then with 95% eth- anol. This process was repeated two to three times for each sample. The residue at the bottom of the cylinder was checked for large arthropods that were too heavy to float to the top of the cylinder. Next, each sample was sorted by genus. Trained student assistants (parataxonomists sensu Whee- ler, 1995) sorted and identified all material down to the genus level. All ant specimens from a single genus were then sorted to species by me, by ex- amining specimens sequentially from each eleva- tional site. This method allowed the greatest num- ber of specimens within an elevational site to be identified while in alcohol and thus limited the cost in time of mounting specimens. Data for specimens were managed using Biota (Colwell, 1996). Identification Specimens were identified to morphospecies by me, based on characters previously established to be important at the species level for each genus. When possible, species names were attached to these morphospecies by using taxonomic descrip- tions (see Fisher, 1997, for a list of references) and by comparing specimens with those previ- ously collected by P. S. Ward and me in Mada- gascar that had been compared to type material. Species codes used in this paper correspond to species codes used in Fisher (1996a, 1998). A rep- resentative set of specimens will be deposited at the Museum of Comparative Zoology at Harvard University and in Madagascar. Data Analysis Evaluation of Sampling Method — To assess survey completeness for the elevations sampled, FISHER: ANT DIVERSITY 131 Table 9-1. Ant species list for the RNI d'Andohahela, including elevation and collection method. Genus Species 430 m 800 m 900-1000 m 1250 m CERAPACHYINAE Cerapachys Simopone FORMICINAE CAMPONOTINI Camponotus LASIINI Paratrechina PLAGIOLEPIDINI Plagiolepis MYRMICINAE CREMATOGASTRINI Crematogaster DACETONINI Kyidris Smithistruma Strumigenys 2 W 3 W, G 4 5 W 6 P 7 8 W 2 G 2 5 G 6 W, P, G 7 8 9 10 G 12 15 W, G 23 24 28 hildebrandti G 1 W, P, G W, P, G 4 5 W, P G W, P, G 6 W W 2 3 W W 3 W w 4 11 w w schenki 1 w w 2 3 G 1 W, P 2 3 G 13 W P 14 W W 16 w W 18 W, G W 20 W W 21 51 W grandidieh G W W W, G G P G G G G W, G G P W W, P G W W, G W G W G W, G W, P G W, G W W W W, G W W. G 132 FIELDIANA: ZOOLOGY Table 9 - 1 . Continued Genus Species 430 m 800 m 900-1000 m 1250 m PHALACROMYRMECINI Pilotrochus besmerus W, G PHEIDOLINI Aphaenogaster 1 G Pheidole 6 W, P W, G 7 W, P, G w 8 W, R G 10 W, P W, P W, P 11 P W, G 13 w 14 W, P W, R G 17 G 23 R G W, P W 24 W 25 W, P P 26 W 27 W W, P 28 W, R G P 29 W, P 31 G W, P 32 W 33 W 34 G 35 W W 38 W longispinosa W, R G W, R G W nemoralis W, P W, P veteratrix W, R G W, R G G W, P PHEIDOLOGETONINI Oligomyrmex 3 W 6 W W SOLENOPSIDINI Monomorium 5 W, G W W 7 W W 14 W 17 P 18 W W 19 W 20 W, P 21 W 22 W 25 W W, R G 43 G TETRAMORIINI Tetramorium 6 G W, P G W, R G 13 P P 14 W, P 15 P 16 W, P 18 W W, P W, R G 19 G 20 W, G 22 W W, P 23 W, P P 27 W, G W 30 G 31 W, G W 32 G 33 W, P W, G dysalum W, P electrum W, P W, P FISHER: ANT DIVERSITY 133 Table 9-1. Continued Genus Species 430 m 800 m 900-1000 m 1250 m 1NCERTAE SEDIS Undescribed genus PONERINAE AMBLYOPONINI Amblyopone Mystrium Prionopelta ECTATOMMINI Discothyrea Proceratium PLATYTHYREINI Platythyrea PONERINI Anochetus Hypoponera Leptogenys Pachycondyla PSEUDOMYRMECINAE Tetraponera bicuspis grandidieri 1 4 5 6 7 8 9 11 12 13 14 16 18 sakalava 1 2 4 cambouei sikorae grandidieri hysterica psw-70 psw-81 psw-92 W G W W G W G W G P W, R G G W, G G W G W W W W W W, R G W W W W W W W W W P W, P, G W, P G W, P G W, G W W, G w, w R G W. G \v. G w, G W, G R G W, R G W W, R G Total species: G 27 15 12 39 Total species: P 18 30 20 Total species: W 46 63 53 Total species: W and P 49 71 59 Total species: all methods 64 74 71 Number (%) of unique species: all methods 18 (28%) 20 (27%) 1 (8%) 35 (49%) Number (%) of unique species: W and P 9 (18%) 20 (28%) 29 (49%) Total number of G collections 40 12 12 23 Number of workers: G 814 172 98 348 Number of workers: P 852 1 .038 430 Number of workers: W 4,032 3,679 2,254 Total number of workers 5,698 4,889 3,032 Abundance: total number stations collected 519 641 528 Notes: Only collections of workers are presented (G = from general collections; P = from pitfall transect samples; W = from mini-Winkler, leaf litter transect samples). A total of 137 ant species and 13,717 workers were collected. In addition, Cardiocondyla emeryi and Eutetramorium sp. 1 from 1250 m were recorded from queens only. Abundance refers to the total number of stations where each species was collected. 134 FIELDIANA: ZOOLOGY I plotted species accumulation curves for each el- evation. Species accumulation was plotted as a function of the number of leaf litter and pitfall trap samples taken. For the analysis, each leaf lit- ter sample was paired with the adjacent pitfall sample, collectively termed a station sample. Spe- cies accumulation curves for the 50 stations per transect, as well as incidence-based coverage es- timator (ICE) and first-order jackknife estimates of the total number of species in the local com- munity from which the samples were taken, were plotted for each succeeding station sample. ICE and the first-order jackknife methods are nonpara- metric approaches to improving the estimate of species richness. ICE is based on species found in 10 or fewer sampling units (Lee & Chao, 1994; Chazdon et al., 1998). Standard deviations of ICE are based on bootstrap estimates (Colwell, 1997). The first-order jackknife is based on the observed frequency of unique species at a 50-station tran- sect. The jackknife estimator and its standard de- viation are defined in Heltshe and Forrester (1983). For species accumulation curves, sample order was randomized 100 times, and the means and standard deviations of ICE and the jackknife estimates were computed for each succeeding sta- tion using the program Estimates (Colwell, 1997; see also Colwell & Coddington, 1994; Chazdon et al., 1998). 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 bias the relative abundance of the species. Because ants are colonial, abundance measures were not based on the total number of individual workers collected at each transect site, but rather on species frequency defined as the pro- portion of stations, out of 50, in which each spe- cies was collected at a site. For each elevation, I compared ICE and first- order jackknife estimates of total species richness and their 95% confidence limits. Overlap and complementarily (distinctness or dissimilarity sensu Colwell & Coddington, 1994) of the ant assemblages at different elevations were assessed using distance, faunal similarity, and beta-diver- sity indices. The proportion of all species in two sites that occurred in only one or the other was calculated using the Marczewski-Steinhaus (M-S) distance index based on presence/absence data: CMS = (a + b - 2j)/(a + b - j), where j = num- ber of species found at both elevations, a = num- ber of species at elevation A, and b = number of species at elevation B (Pielou, 1984; Colwell & Coddington, 1994). Similarity of the ant fauna was assessed using the simplified Morisita Index, which incorporates abundance data: 2 2 (an, X bn,) (da + db)aN X bN where da 2 aN2 and db - bN2 ' aN = total number of station/species occurrences in elevation A, bN = total number of station/spe- cies occurrences in elevation B, an, = the number of stations occupied by the /th species in elevation A, and bn, = the number of stations occupied by the /th species in elevation B (Horn, 1966; Wolda, 1981). Indices based on presence/absence data have been shown to be strongly influenced by spe- cies richness and sample size (Wolda, 1981). The Morisita Index is nearly independent of species richness and sample size (Wolda, 1981) and may therefore be more appropriate for comparisons of ant assemblages between sites along an elevation- al gradient that differ greatly in species richness. 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 5 = the total number of species in the two elevations combined and a = the mean number of species in each el- evation. Because this measure does not distin- guish 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/amM) - 1, where S is the same as beta-1 above and amsa = the maximum value of alpha- diversity (i.e., number of species) among the el- evations compared. The number of species unique to an elevation and the number of species shared between elevations were also compared. In addition, I also calculated complementarily, species turnover, number of unique species, and number of species shared among the RNI d'Andohahela and three other localities surveyed using the same methods: the RNI d'Andringitra, FISHER: ANT DIVERSITY 135 Table 9-2. Abundance measured as frequency of occurrence (proportion of stations out of 50 paired pitfall and leaf litter samples at which each species was recorded) for each elevation in the RNI d'Andohahela. Genus Species 430 m 800 m 1250 m CERAPACHYINAE Cerapachys 2 3 0.04 (5) 0.06(10) 5 0.02 (1) 0.06 (9) 6 0.02 (1) 7 0.08 (4) 8 0.04 (5) 0.08 (5) FORMICINAE CAMPONOTINI Camponotus 5 6 12 15 24 28 0.04 (2) 0.02 (1) 0.02 (1) 0.06 (126) 0.02 (1) 0.08 (4) LASIINI Paratrechina 1 4 1.00 (2,090) 0.32 (103) 0.36 (112) 0.10(42) 5 0.48 (254) 0.44 (754) 0.04 (29) 6 0.02 (2) 0.06 (6) 0.02 (1) PLAGIOLEPIDINI Plagiolepis 3 0.02 (1) 0.08 (8) 0.42 (112) MYRMICINAE CREMATOGASTRINI Crematogaster 3 0.02 (1) 0.02 (1) 11 0.08 (6) 0.04 (4) 0.28 (17) schenki 0.72 (166) DACETONINI Kyidris 1 0.04 (6) 0.02 (1) Smithistruma 2 0.06 (4) Strumigenys 1 2 0.90 (989) 0.02 (1) 0.04 (13) 13 0.04 (4) 0.02 (1) 14 0.14 (8) 0.32 (38) 16 0.04 (4) 0.02 (1) 0.04 (6) 18 0.42 (39) 0.20 (41) 0.62 (113) 20 0.04 (3) 0.10 (8) 21 0.08 (6) 51 0.02 (1) grandidieri 0.02 (2) PHALACROMYRMECINI Pilotrochus besmerus 0.08 (4) PHEIDOLINI Pheidole 6 0.24 (75) 0.02 (3) 7 0.38 (133) 0.28 (82) 8 0.74 (332) 10 0.04 (2) 0.12 (28) 0.48 (76) 11 0.04 (12) 0.04 (2) 13 0.02 (1) 14 0.14 (16) 0.44 (178) 23 0.12 (95) 0.12 (26) 0.02 (1) 24 0.02 (1) 25 0.54 (110) 0.16(48) 136 FIELDIANA: ZOOLOGY Table 9-2. Continued Genus Species 430 m 8(M) m 1250 m PHEIDOLOGETONINI Oligomyrmex SOLENOPSIDINI Monomorium TETRAMORIINI Tetramorium INCERTAE SEDIS Undescribed genus PONERINAE AMBLYOPONINI Amblyopone Mystrium Prionopelta ECTATOMMINI Discothyrea Proceratium 26 0.36(110) 27 0.36 (72) 0.32 (120) 28 0.90 (224) 0.02 (1) 29 0.12 (72) 31 0.06(11) 32 0.02 (1) 33 0.02 (1) 35 0.02 (1) 0.04 (8) 38 0.02 (3) longispinosu 0.98 (1,010) 0.68 (510) 0.04 (2) nemoralis 0.28 (78) 0.32 (76) veteratrix 0.24 (80) 0.52 (218) 0.38 (107) 3 0.02 (1) 6 0.10 (6) 0.14 (37) 5 0.50 (67) 0.40 (57) 0.34 (142) 7 0.02 (3) 0.10(69) 14 0.16(28) 17 0.02 (1) 18 0.30(108) 0.48 (81) 19 0.02 (2) 20 0.18 (48) 21 0.02 (3) 22 0.10 (98) 25 0.10(7) 0.04 (3) 6 0.18 (21) 0.10 (8) 13 0.02 (1) 0.02 (5) 14 0.20 (39) 15 0.02 (1) 16 0.24 (37) 18 0.02 (1) 0.80 (253) 0.52 (109) 20 0.18 (18) 22 0.02 (4) 0.12 (17) 23 0.14 (19) 0.02 (1) 27 0.02 ( 1 ) 0.02 (1) 31 0.06 (3) 0.06 (5) 33 0.14 (9) 0.30 (43) dysalum 0.30(151) electrum 0.26 (76) 0.18 (28) 0.10 (5) 0.14 (9) 0.02 (I) 0.06 (8) 0.02 ( 1 ) 0.16 (18) 0.02 (1) 0.12 (10) 0.08 (5) 0.10(6) FISHER: ANT DIVERSITY 137 Table 9-2. Continued Genus Species 430 m 800 m 1250 m PLATYTHYREINI Platythyrea PONERINI bicuspis The number of individual workers collected is given in parentheses. 0.02 (1) Anochetus grandidieri 0.40 (35) 0.20 (27) Hypoponera 1 4 5 0.06(14) 0.02 (1) 0.82 (389) 0.02 (8) 6 0.50 (71) 0.24 (86) 7 0.18 (11) 0.38 (42) 8 0.16 (18) 9 0.10(6) 0.08 (14) 11 0.46 (56) 0.66 (232) 12 0.02 (1) 13 0.36 (33) 16 0.02 (2) 18 0.06 (5) sakalava 0.34 (42) Leptogenys 1 0.02 (2) 2 0.08 (4) 0.16 (10) 0.34 (33) 4 0.02 (1) Pachycondyla cambouei 0.68 (72) 0.90 (221) 0.36 (56) PSEUDOMYRMECINAE Tetraponera grandidieri 0.04 (4) 0.04 (2) psw-81 0.06 (3) 0.06 (3) the RS d'Anjanaharibe-Sud, and the western Ma- soala Peninsula. Results In the RNI d'Andohahela I collected and iden- tified 13,717 ants comprising 29 genera and 139 species from general collections, leaf litter, and pitfall methods. These included 155 queens and 86 males. Leaf litter and pitfall methods yielded 12,285 worker ants belonging to 25 genera and 111 species. A list of ant species from this study in the RNI d'Andohahela based on all collecting methods and separated by elevation and technique is presented (Table 9-1). General collections from 900 to 1000 m are also presented. Absent from Table 9-1 are records of species known from queens only: Cardiocondyla emeryi and Eutetra- morium sp. 1, both from 1250 m. Within the RNI d'Andohahela, the 800 m zone had the greatest total number of species recorded (74 species total from all methods; 71 species to- tal from litter and pitfall samples; Table 9-1). The same relative ranking in observed species richness between sites was reached and maintained after three station samples (Table 9-5). The numbers of species and individuals collected from pitfall traps were low compared to those collected by mini- Winkler methods. Only six species that were col- lected by pitfall traps were not also collected by the mini- Winkler method. In a study using com- parable methods in dry forest in southwestern Madagascar, however, pitfall traps collected a greater proportion of individuals and species (Fisher & Razafimandimby, 1997). The abundance of ant species is presented in Table 9-2. Both the proportion of stations at which each species was collected and the number of in- dividuals collected are given. General collections are not included. Only 15 species out of 111 (14%) were found at all three elevations. The rel- ative frequencies of occurrence of these species, however, often differed considerably from one site to the next. For example, Paratrechina sp. 5 had relative frequencies of 0.48 at 430 m, 0.44 at 800 m, and 0.04 at 1250 m (Table 9-2). Thirty-eight species (34%) were collected at two of the three elevational sites. The number of ant species and their abundance, measured as the total number of stations where 138 FIELDIANA: ZOOLOGY 80 CO CD 8 60 Q. CO O40 CD -Q £20 3 (a) 0 200 400 600 800 100012001400 700 -, 600 CD 500 §400 "O §300 .O <200 100 (b) 0 200 400 600 800 100012001400 Elevation (m) Fig. 9- 1 . The number of ant species (a), and the total abundance (b) as a function of elevation. Abundance is measured as the total number of stations where each species was collected (see text for details). Data are from pitfall and mini-Winkler samples. each species was collected, peaked at 800 m (Fig. 9-1). The relative prevalence of each subfamily for the combined pitfall and leaf litter samples for each elevation and for all elevations is shown in Table 9-3. The fauna was dominated by Myrmi- cinae in both numbers of species and individuals, followed by Ponerinae. The ratio of Ponerinae to Myrmicinae was similar at 800 and 1250 m (0.41 and 0.42, respectively), but was lower at 430 m (0.19). Observed number of species, ICE, and first-or- der jackknife estimates of species richness, their standard deviation, and 95% confidence intervals are presented for the RNI d'Andohahela (Table 9- 4). Observed species richness for each elevation surveyed in the RNI d'Andohahela, evaluated at different sample sizes, is presented in Table 9-5. Species accumulation curves for observed, ICE, and jackknife estimates showed a decrease in the rate of species accumulation, but were still in- creasing slowly (Fig. 9-2). In a combined analysis of the three elevations in the RNI d'Andohahela, the pitfall and mini-Winkler methods collected 86% of the total number of leaf litter ant species estimated by ICE that could be collected using these methods in the transect areas (Fig. 9-2d). The greatest dissimilarity (M-S Index) and low- est similarity (simplified Morisita Index) values between adjacent elevations occurred between 800 and 1250 m (Table 9-6). Similarly, the great- est species turnover (beta-diversity) occurred be- tween 800 and 1250 m (Table 9-7). The overall beta-1 and beta-2 values of species turnover be- tween all elevations were 0.860 and 0.563, re- spectively. The 1250 m site had the greatest number and the highest percentage of species restricted to one specific elevation (Table 9-1). The 800 m site had the highest number of species shared with other sites (Table 9-6). The 800 m site shared more spe- cies with the 1250 m site than with the 430 m site (Table 9-6). The ant fauna at the RNI d'Andohahela was compared to the fauna at three other localities where similar inventories have been conducted. The elevations surveyed at all four localities are presented in Table 9-8, and their distances apart (km) are presented in Table 9-9. The matrix of complementarity values shows a high level of dis- tinctness between localities (Table 9-10). The ant faunas of the RNI d'Andohahela and the RNI d'Andringitra (65% distinct; 275 km apart), and between the RS d'Anjanaharibe-Sud and the west- ern Masoala Peninsula (72% distinct; 110 km apart) show the lowest level of distinctness. The RNI d'Andringitra and the RS d'Anjanaharibe- Sud (87% distinct; 900 km apart) and the RNI Table 9-3. Total number and percentage (%) of species of each subfamily for pitfall and leaf litter collections on the RNI d'Andohahela (general collections are excluded). P/M refers to the taxonomic ratio of species in the Ponerinae and Myrmicinae. Subfamily names are abbreviated (see Table 9-1). Elevation (m) Cerap Form Pon Myrm Pseudo P/M 430 0 4 (8%) 7 (14%) 36 (74%) 2 (4%) 0.19 800 5 (7%) 6 (87c) 17 (24%) 41 (58%) 2 (3%) 0.41 1250 3 (5%) 9 (15%) 14 (24%) 33 (56%) 0 0.42 All elevations 6 (5%) 11 (10%) 27 (24%) 65 (59%) 2 (2%) 0.41 FISHER: ANT DIVERSITY 139 Table 9-4. The number of species collected, inci- dence-based coverage estimator (ICE), and first-order jackknife estimates of total species richness (with 95% confidence intervals, CI), based on pitfall and leaf litter transects in the RNI d'Andohahela. Statistics are given for each altitude and for all elevations combined. Elevation (m) Ob- served ICE 95% CI Jack- 95% knife CI 430 800 1250 AH elevations 49 71 59 111 61.7 90.3 72.0 129.4 0.32 0.28 0.31 0.11 61.7 0.61 90.6 0.79 73.7 0.70 135.8 0.20 d'Andohahela and the western Masoala Peninsula (87% distinct; 845 km apart) had the greatest complementarity. Species turnover values show the same pattern (Table 9-10). In a comparison of all 800 m zone sites (Table 9-11) and all 1200 m Table 9-5. Observed species richness for each ele- vation surveyed in the RNI d'Andohahela, evaluated at different sample sizes. Richness values are the means of 1 00 randomizations of sample accumulation order. Stan- dard deviations are given in parentheses. Stations sampled 1 3 5 10 15 20 25 30 35 40 45 50 430 m 800 m 10.7 (2.90) 18.8 (3.19) 23.5 (3.01) 30.6 (2.49) 34.7 (1.95) 37.7 (1.96) 40.4 (1.84) 42.5 (1.77) 44.4 (1.51) 46.2 (1.28) 47.7 (0.98) 49 12.7 (4.01) 25.5 (4.09) 33.6 (3.81) 44.4 (3.09) 50.6 (3.06) 55.2 (2.70) 68.7 (2.52) 61.7 (2.35) 64.4 (2.25) 66.8 (1.78) 69.0(1.18) 71 1250 m 10.8 (2.53) 20.8 (2.69) 26.7 (2.72) 35.3 (2.91) 41.2 (2.94) 45.5 (2.84) 48.8 (2.73) 51.3 (2.30) 53.8 (2.07) 55.8 (1.57) 57.5 (1.19) 59 (0 CD O JS E o 150 Number of stations sampled Fig. 9-2. Assessment of leaf litter ant sampling technique for each elevation (a-c) and for all elevations combined (d) in the RNI d'Andohahela. The lower species accumulation curve (thick line) in each chart plots the observed number of species as a function of the number of stations sampled. The upper curves display the nonparametric first- order jackknife (dashed line) and incidence-based coverage estimator (ICE) (solid line) estimated total species richness based on successively larger numbers of samples from the data set (Heltshe & Forrester, 1983; Lee & Chao, 1994). Curves are plotted from the means of 100 randomizations of sample accumulation order. 140 FIELDIANA: ZOOLOGY Table 9-6. Complementarity and faunal similarity between the three elevational zones sampled in the RNI d'Andohahela. Above the diagonal is the Marczewski- Steinhaus (M-S) distance index (presence/absence data; Pielou, 1984) and below the diagonal, simplified Mor- isita index of similarity (abundance data; Horn. 1966). Higher values represent greater distinctness (M-S) or similarity (Morisita). Bold values represent comparisons of altitudinally adjacent transects. The number of species shared between elevations is presented in parentheses above the diagonal. Elevation 430 m 800 m 1250 m 430 m 800 m 1250 m 0.584 0.257 0.537 (38) 0.411 0.813 (17) 0.725 (28) zone sites (Table 9- 1 2), the same pattern of com- plementarity and species turnover values were found. In Table 9-11, data from the 785 and 825 m sites in the RNI d'Andringitra (see Table 9-8) were combined. The same pattern was obtained when the 785 or 825 m sites were analyzed sep- arately with the balance of the 800 m zone sites from the other localities. Discussion The RNI d'Andohahela Ant Fauna We are far from the saturation point in our ef- forts to discover undescribed ant species in Mad- agascar. In terms of numbers of new species col- lected per site inventoried, we are still on the steep part of the curve. There may be 1,000 spe- cies on the island, with two-thirds of the species undescribed (Fisher, 1996b, 1997). No previous records exist for ants collected in the RNI d'Andohahela. Subsequent collections of ants in other parts of the RNI d'Andohahela region have been made by P. S. Ward, G. D. Alpert, and K. C. Emberton, and they reveal additional species diversity within the region not recorded during this 1992 survey. For the island of Madagascar, 90% of the valid specific and subspecific ant taxa are endemic (Fisher, 1996b, 1997). In the RNI d'Andohahela, nearly 100% of the ants collected are thought to be endemic to Madagascar, except for Cardiocon- dyla emeryi, which is a pantropical tramp species and is known throughout the Malagasy region (Bolton, 1982; Fisher, 1997). Table 9-7. Beta-1 (above the diagonal) and beta-2 (below the diagonal) diversity values of each pair of altitude sites in the RNI d'Andohahela. Higher values represent greater species turnover. Bold values represent comparisons of altitudinally adjacent transects. Overall beta-1 diversity was 0.860 and that of beta-2 was 0.563. Elevation 430 m 800 m 1250 m 430 m 800 m 1250 m 0.155 0.542 0.366 0.437 0.685 0.569 Many interesting and rare taxa were collected at the 1250 m site. For example, a single worker of Aphaenogaster sp. 1 was collected in a general collection from the leaf litter. This is the first rec- ord from the southern half of the island of a mon- tane forest Aphaenogaster. At the 1250 m site, seven workers and two queens of Pilotrochus bes- merus were collected; this is the second record of this endemic Malagasy monotypic genus. Pilotro- chus was previously known only from a single worker collected in a Berlese sample along the road to Anosibe An' Ala, 33 km south of Mora- manga, in east central Madagascar in 1975 (Brown, 1978). The generic name is derived from the Greek pilos (hair) + trochos (wheel), in ref- erence to its amazing mesopleural "hair-organ." In addition, other rare species and genera col- lected at the 1250 m site include two species of Amblyopone, one species each of Smithi struma, Discothyrea, Eutetramorium, and an undescribed myrmicine genus. Previously, Eutetramorium was thought to be endemic to the dry and eastern hu- mid forests of the northern half of Madagascar. At the 1 250 m site, a queen of Eutetramorium sp. 1 was collected in a leaf litter sample. In 1993, P. S. Ward collected a single worker-queen inter- mediate in a Winkler sample in humid forest at 1050 m in the RNI d'Andohahela, 3 km east of Mahamavo. This suggests that although queens and worker-queen intermediates may be found in the leaf litter, the nest is located elsewhere, in the canopy or in fallen hard wood that is not sampled by the leaf litter technique. Fallen trees with hard wood are a microhabitat that is often overlooked by ant collectors. The undescribed myrmicine genus was thought to be endemic to the humid forest of northeastern Madagascar (Fisher, 1998). In the RNI d'Andohahela, a different species of the unde- scribed genus of myrmicine was collected in a large rotten log at 950 m and in a leaf litter sample FISHER: ANT DIVERSITY 141 Table 9-8. Elevations surveyed within each elevational zone in the RNI d'Andohahela, the RNI d'Andringitra, the RS d'Anjanaharibe-Sud, and on the western Masoala Peninsula. Location 0 m 400 m 800 m 1200 m 1600 m 2000 m Andohahela Andringitra Anjanaharibe-Sud Masoala 25 430 800 1250 785, 825 1275 1680 875 1200 1565 425 825 1985 at 800 m. These records of Pilotrochus, Eutetra- morium and the undescribed genus suggest that the geographical ranges of these genera may ex- tend patchily across the entire length of the east- ern humid forest. Elevational Gradient and Complementarity Faunal similarity, distinctness, and species turn- over measures (Tables 9-5 and 9-7) support a di- vision of the ant fauna into two assemblages, one occurring in lowland forests <800 m and the oth- er in montane forests at 1250 m. Between adja- cent sites, species turnover was greater between 800 and 1250 m than between 430 and 800 m. In previous studies in the RNI d'Andringitra (Fisher, 1996a), and in the RS d'Anjanaharibe-Sud and on the western Masoala Peninsula (Fisher, 1998), mid-elevation sites (ca. 800 m) had the highest rate of species turnover. Species richness did not decrease monotonical- ly as a function of elevation (Fig. 9-1). A mid- elevation peak has been documented for ants in Madagascar (Fisher, 1996a, 1998), in Panama (Ol- son, 1994), and for other taxa (Rahbek, 1995). The mid-elevation peak observed in ant species richness in Madagascar may be the result of a mixing of two distinct ant assemblages along an ecotone. As suggested in Fisher (1998) for the ant fauna in the RS d'Anjanaharibe-Sud and on the western Table 9-9. Distance (km) between the RNI d'Andohahela, the RNI d'Andringitra, the RS d'Anjanaharibe-Sud, and the western Masoala Peninsu- Andringitra Anjanaharibe- Sud Masoala Ando- hahela Andringitra Anjanaha- ribe-Sud 275 1,170 1,200 900 845 110 Masoala Peninsula, species richness may increase from low elevation (430 m) to mid-elevation (800 m) because the 800 m site is adjacent to the source-pool of the distinct montane ant fauna as well as those from lower elevations. The prox- imity of elevational zones encourages the estab- lishment of marginal populations from adjacent elevations (Pulliam, 1988; Stevens, 1989, 1992; Rahbek, 1997). A mixing of the lowland and montane ant assemblages results in the peak in species richness. In the RNI d'Andohahela, the mid-elevation site (800 m) has the highest number of species shared with other sites: 38 species with 430 m and 28 species with 1250 m (Table 9-6). The number of species shared by the lowest ele- vation site decreases with increasing change in el- evation. An alternative hypothesis to that of the mixing of lowland and montane ant assemblages is that mid-elevations provide the most "suitable" en- vironment for ants (Rosenzweig & Abramsky, 1993). This assumes that the suitable environment favors an increase in species richness rather than an increase in the population numbers of species. There is currently no accepted explanation of why the most suitable habitats would occur at mid-el- evations or how these habitats would increase species richness (Rosenzweig & Abramsky, 1993; Rahbek, 1997). Efficacy of Inventory Methods The efficacy of the inventory methods can be evaluated by using species accumulation curves (Col well & Coddington, 1994). The criterion I use to evaluate efficacy is the number of species col- lected per unit effort. For every 50-station tran- sect, which takes an average of 7 field days to conduct, 1 month must be spent in the laboratory sorting, identifying, and curating specimens. If in- creased sampling efforts always collect additional species, how many subsamples should be taken? An accumulation curve is specific to the area 142 FIELDIANA: ZOOLOGY Table 9-10. Complementarity (M-S, above the diagonal) and beta-1 (below the diagonal) diversity values between the RNI d*Andohahela. the RNI d'Andringitra, the RS d'Anjanaharibe-Sud, and the western Masoala Peninsula. The number (percentage) of species specific to the locality are presented along the diagonal. Total number of species for all localities is 381. Andohahela Andringitra Anjanaharibe-Sud Masoala Andohahela Andringitra Anjanaharibe-Sud Masoala 37 (33) 0.478 0.739 0.734 0.647 45 (39) 0.776 0.773 0.850 0.874 79(44) 0.562 0.846 0.872 0.720 78 (47) of the survey, the season or year, and the collect- ing techniques employed. Additional collecting methods, or a survey in a different area or season at the same elevation, would most likely collect additional species. If an observed or estimated species accumulation curve demonstrates a suffi- cient decrease in the rate of species accumulation, then the number of subsamples is arguably ade- quate for collecting the species in the area sur- veyed for the particular methods employed. Con- versely, if the curves are rising rapidly, more in- tensive sampling may be necessary to accurately compare diversities between elevations. For hy- perdiverse groups with large numbers of rare spe- cies, more intensive sampling (i.e., larger numbers of subsamples) typically never generate curves that completely flatten out and reach an asymp- tote. For these taxa, rates of species accumulation are expected to slowly decrease with more sam- pling. The entire area may need to be exhaustively surveyed before one can be sure that every spe- cies has been collected, but in most cases com- plete sampling is not possible and is often not the objective. Sufficient sampling for a high level of com- pleteness is therefore the point at which the ac- cumulation curves show an adequate decrease in species detection. The problem is the lack of ex- istence of an asymptote for diverse taxa and the difficulty in quantifying "an adequate decrease in species detection." One possibility is to sample until a certain percentage — say, 80% — of the es- timated species are sampled. In this study, be- tween 78% and 82% of the species had been sam- pled from 50 stations based on ICE and jackknife estimates of species richness (Table 9-4). For all elevations combined, 86% of the ICE and 82% of the jackknife-estimated species richnesses were sampled. The problem with this approach is that the ICE and jackknife-estimated values are sen- sitive to sample size (Fig. 9-2). For example, after 10 stations, comparable percentages of ICE and jackknife estimates were obtained, but species ac- cumulation was still rising rapidly. Therefore, the percentage sampled of the ICE and jackknife-es- timated species richness was not a reliable indi- cator of completeness. For example, at 1250 m (Fig. 9-2a), the ICE, jackknife, and observed curves between 10 and 50 stations are parallel. ICE and jackknife estimates predict comparable levels of completeness within this range of sam- ples (10-50) even though species accumulation was still rising rapidly after 10 stations (Fig. 9- 2a). Sensitivity to sample size prevents using this method for assessing the level of completeness of these inventories. An alternative approach is to sample until ad- ditional sampling efforts achieve a defined per- centage increase in the number of species sam- pled. Species accumulation curves can be extrap- Table 9-11. Complementarity (M-S, above the diagonal) and beta-1 (below the diagonal) diversity values for the 800 m zone sites between the RNI d*Andohahela, the RNI d'Andringitra, the RS d'Anjanaharibe-Sud, and the western Masoala Peninsula. Data from the 785 and 825 m transects from the RNI d'Andringitra were combined. The numbers (percentage) of species specific to the locality within the 800 m zone are presented along the diagonal. Total number of species for all 800 m sites is 242. Andohahela Andringitra Anjanaharibe-Sud Masoala Andohahela Andringitra Anjanaharibe-Sud Masoala 25 (35) 0.487 0.786 0.756 0.664 37 (42) 0.795 0.763 0.880 0.886 45 (46) 0.563 0.861 0.848 0.714 50 (46) FISHER: ANT DIVERSITY 143 Table 9-12. Complementarity (M-S, above the di- agonal) and beta-1 (below the diagonal) diversity values for the 1200 m zone sites between the RNI d'Andohahela, the RNI d'Andringitra, and the RS d'Anjanaharibe-Sud. The numbers (percentage) of spe- cies specific to the locality within the 1200 m zone are presented along the diagonal. Total number of species for all 1200 m sites is 151. Andohahela Andringitra Anjanaharibe- Sud Ando- hahela Andrin- gitra Anjana- haribe-Sud 32 (54) 0.660 0.766 0.795 23 (56) 0.874 0.867 0.933 68 (79) olated to project the increase in species richness expected for a increase in sampling effort (Sob- eron & Llorente, 1993; Colwell & Coddington, 1994). In Figure 9-3, I fitted the observed species accumulation curves using the Soberon and Llo- rente (1993) logarithmic model: S(7) = ln(l + zat)/z, where t is the measure of sampling effort (samples or individuals), and z and a are curve- fitting parameters. Log models do not have an as- ymptote and are considered appropriate for spe- cies-rich taxa (Soberon & Llorente, 1993). That is, I use a nonasymptotic model because I assume the curves will never completely flatten, even with complete sampling. Using a nonasymptotic model may therefore result in a conservative es- timate of the number of species predicted with increasing effort. I fitted the log model using the nonlinear least squares method of regression in JMP (SAS Institute, 1994). Based on the extrap- olation of these curves, a doubling of sampling effort (an additional 50 stations) would achieve only a 13% gain in species richness at the 430 m site, 14% at 800 m, and 15% at 1250 m (Fig. 9- 3). If all samples are combined, only 17 more spe- cies (13%) are predicted from a doubling of sam- pling effort (an additional 150 stations). The relative between-site pattern of species richness would change very little if collection had been made at an additional 50 stations at each elevation. The relative ranking of between-site pattern of species richness stabilized after a few stations (Table 9-5) and is not predicted to change with the addition of 50 more stations at each tran- sect. Additional species from 100 station transects, however, could affect the relative between-site pattern of complementarity. If increased sampling collects rare species that are restricted to a partic- ular elevation, then complementarity can be un- 60-, 50 40 30H 20 10-1 (a) 430 m 0 10 20 30 40 50 60 70 80 90 100 90- (b) 800 m 80- _-_-— ■ CO 70- ^~- — ' 1 500 m in parcel 1 include Mantidactylus elegans, M. guibei, Am- phiglossus sp., Calumma capuroni, and Lygodac- tylus montanus. The restricted distributions of M. guibei and C. capuroni to high elevations in the southern rain forests and the lack of information NUSSBAUM ET AL.: AMPHIBIANS AND REPTILES 167 ■ft 3 e /, < «« ■j >> Xi 5 >> o X 1/1 ta g a u F= u R •— a 5 63 * £ cS < ■S^ ^ .2 boon ■si" ■S- i «zz Za^ 2.3x? 5 0 ■£3 55 ^ o £ o S - u* nj cS 03 ^ >-■ CX, c« . 3 C O c O X .3 X o - 3 "O tfl ^i 00 •212 °S r P-13 u _u 3 c m on 3 3 ' a VI "C _0) ■a "3 3 01 a < GO ■O S CO 1-H 01 0 Im 93 Q.T3 2 GO u 7 J= 0) Iri u « s xs X as GS U9 it 3 a •—. "3 e o> a < GO "O e a — o> u 0> a X >> E x> «( 01 "3 s. xs to 0) 3 SI 3 VI -• 3 '3 C 01 a " GO 6 SI s * On 3?„ BO 3 k. C a a -3 •2 C _2 S B<) -5 _2 IS -a Si 3 "a a J= 5 g yi ,*j y. .< ,fc5 a -5 -5 5 -5 -S t* R. a. R, a. a, Q c 5 0 a c C Q e Q a a •< 03 cq oq qq 03 «5 S c 5 3 s -S 'i: '3 « ^ a ■« -Cl -C5 3 3 ... ■S>.2 -a 5 3s 5 a (j '5 -~ -2 -5 -S: -C -3 -3 -3 -3 -3 ■ -S ^ d 3 §.5 "3 ^ ^ "1 >1 tfl <^ «5 «3 3 3 3 3 3 3 3 ^cS^I cccccckcccs: Is ^ « 5 §§ C 3 TS 6 S -6 < l << =o to to 3 1- ^Q /^ 3 3 3 3 3 3 3 3 ~5 3 5 i 66666666 .^ ,&5 ,«n .*5 /-. a a 3 3 3 3 3 3 0 -S -5 -5 -S -5 rs^^^^^s^s^s c O O Cl 0 0 ^ Q3 oq cq qq 03 CCC3333C 3 4J *> -3 3; ft, 3 ^ i3 J3 ^^ ft, a a 5 c c 6 S s % ^ BO ft g >> to to g -3 3 3 K *s 6 6 g § u I* Bo O ^C 03 — X - fi-SS" 3 .3 to <3 3 3 3 -3 -3 S S 6 ■«-s: -a S" S" 3 ft O 3 ft ft ^ 03 03 ^ J 3 8 ^^ § 3 C *= 6 °° § u ^ s s 6-5 sg^ 6 § ft, Q, -S c 'S 3 to 3 ^) to 3 C S 3 § ^3 -2 3 Bo 3 5 a 3 %> -3 .to c S? 3 ~ I 3 1ft ft -S 3 3 |! s 3 3 s ^. s ft 3 -O s ft to -3, .ft 3 "^ ft. 3 .*» "C 3 (j to * BO g s S 3 sl •a § %j .«j s "5 5 CU S ft r3 r3 i r3 j5 ,* c ^ 5 BO ^ " _3 5 „ 3 "3 3 :§■§£ -3 -3 s 3. a. ft ft" ft" SJ 1^ 168 FIELDIANA: ZOOLOGY ■v e eg rm ii u u a a >> x> •a ■c "3 = 0> < Cfl -O c 99 — 01 u u es O.T3 = r. l> "O Xl a> u ■- 9) n X J3 a rs C •— : '3 C a < c« ■o c n »h a> u u a a >> £ x < 01 ■b so V] a» c 2 09 .a a °u 0 1) a " W3 .a *- '~ "B b~ P ■5 S3 1 I -2 <*> ft S £3 3 s ft^ J ^ft C C %i ft 3 Ssi « a ft 5 5 b -2 Si 5 S 5 ft - -e -2? § ^ a •= "u 5 a ^ ^ ft -r- 5 '2 < ft ft ^N -> "^* j .2?. 2? « -2 -5 IT S> S> ft. 5 5 3 ft .2 a .ft -2 -2 -^ ^ "< ft) a S 2 S 5 k c S S 2 £ -5 ft I A* a 3-2 Z P. .5 55 £ t B-.? c .ft ft C .s .3 _s ,3 § .3 ft. ft.2* ?3 ^ £■ s ^ ^ -S =5 ft. ft. c "? 3?" J!f Oc ^ c f1 - ft ft ft < a*. a c a a *> J2 .2 .2 >-"-s: -2 -S a 1?^2 | C ft ^ ■^ ^J ft. ft. c ft 5 4) -ft -«s 'n vi 3 *) S a a c ^ a rv o ?J r> !*♦ •a .5 "s1^ 11 "^ -2 -2 - j .Sp.SP 2 On 5 S ft S rQ rQ rQ %j ft -^ •« sv ~ 'C ft ft. ft ft- ft- ^ — -2 c ^ ft .ft ■« c ^ ft *) ft ft 00 <3 ft s a ^. 5 3 >5' < S; «- ^> ft * u, ft-ft s 3 k ft ft =C i) 3 S; * s 2 «, ft a ?-ft ft. ft 2 -± NUSSBAUM ET AL.: AMPHIBIANS AND REPTILES 169 on their likely sister-species currently preclude any biogeographical discussion of these forms. The three remaining species, however, also occur in montane rain forests north of the southern An- osyenne Mountains and are of biogeographical in- terest. Mantidactylus elegans is the most widespread of the three. In addition to parcel 1, it has been recorded at high elevations (from south to north) in the RS dTvohibe, RNI d'Andringitra, PN de Ranomafana, RS d' Anjanaharibe-Sud, and RNI de Tsaratanana. We found specimens identical to Am- phiglossus sp. at similarly high elevations on the Andringitra Massif, indicating a historical link be- tween these two sites. The discovery of Lygodac- tylus montanus above 1500 m in parcel 1 also establishes a biogeographical link between the montane forests of the southern Anosyenne Mountains and Andringitra, where we also found this species. It was originally described from the summit (2060 m) of Ivohibe, south of Andringitra and part of the same mountain complex. Parcel 2 (Spiny Forest) The relatively great similarities between the herpetofauna of parcel 1 and other rain forest her- petofaunas as far away as 1,300 km (Montagne d'Ambre) stand in stark contrast to the almost to- tal lack of species in common between parcel 1 and parcel 2, the borders of which are just 5 km apart at the nearest point. The differences result from the rain shadow effect of the Anosyenne Mountains. Westward-flowing air masses leave their moisture burden on the eastern slopes of par- cel 1, with little moisture remaining for the parched western slopes and lowlands. The result is a dry environment with a mosaic of spiny and deciduous forests and more mesic gallery forests along the water courses. The species accumulation curve for site 6 (Fig. 11-2) indicates that our survey of parcel 2 was too brief and that some species may have been missed. In fact, there are two amphibian and 24 reptile species (Table 11-8) that one might expect in similar habitats in this region. We recorded some of these additional species in 1990 at Ber- aketa, only 22 km from site 6 of parcel 2. These include the amphibian Boophis tephraeomystax and the following reptiles: Pyxis arachnoides, Geochelone radiata, Pelomedusa subrufa, Lygo- dactylus tuberosus, Phelsuma modesta, Androngo trivittatus, Voeltzkowia lineata, Heteroliodon oc- cipitalis, Langaha madagascariensis, Leiohetero- don madagascariensis, Leioheterodon modestus, Liophidium apperti, Liophidium torquatus, Lio- pholidophis lateralis, and Lycodryas guentheri. There are at least three possible reasons for the greater diversity recorded at Beraketa. First, as mentioned above, the survey period at parcel 2 may have been too brief. Second, there was a larger survey team (four individuals rather than two) at Beraketa. Third, there may have been real differences in the diversity of the two sites at the times of the surveys because of subtle microhab- itat differences interacting with random and un- known factors that influence local extinction and colonization. Additional species were found in the gallery forests of Berenty Reserve, to the west of parcel 2 along the Mandrare River. These include the typhlopid snake Ramphotyphlops braminus, the colubrid snake Pseudoxyrhopus kely, and the cor- dylid lizard Zonosaurus trilineatus. Other species typical of habitats represented in parcel 2 but not recorded there during the current survey are listed in Table 11-8. As currently documented, the herpetofauna of parcel 2 contains no endemic species and is gen- erally representative of the much broader herpe- tofauna of the Malagasy southwestern deserts. Conservation Prior to our survey of parcel 1 of the RNI d'Andohahela, several species of amphibians and reptiles known from southeastern Madagascar were of concern because they were seemingly rare or of commercial (animal trade) interest. These species are threatened with extinction mainly be- cause of habitat destruction, but also because of their potential value in the animal trade. None of these species is currently protected by law, and prior to our survey none was known to occur within reserves. Foremost among these species of concern were Mantella haraldmeieri, Calumma capuroni, Phel- suma antanosy, Uroplatus malahelo, and U. mal- alama. These species were known from few spec- imens and had been recorded at only a few sites in areas subjected to ongoing habitat destruction, and they are all of commercial interest. Other spe- cies of concern because of their limited distribu- tion, but of less commercial interest, are Para- gehyra gabriellae, Pseudoxyrhopus kely, and P. sokosoko. With the exceptions of Phelsuma an- il 0 FIELDIANA: ZOOLOGY Table 1 1-8. Species not recorded in parcel 2 of RNI d'Andohahela but that possibly occur there, based on our unpublished data for similar habitats in southern Madagascar. Records for Manambaro and Tolagnaro are from within the village and city, and not peripheral areas. Species Non-RNI d'Andohahela records AMPHIBIA HYPEROLIIDAE Heterixalis boettgeri RHACOPHORIDAE Boophis tephraeomystax REPTILIA PELOMEDUSIDAE Pelomedusa subrufa TESTUDINIDAE Geochelone radiata Pyxis arachnoides CORDYLIDAE Zonosaurus trilineatus GEKKONIDAE Lygodactylus tuberosus Phelsuma modesta SCINCIDAE Amphiglossus splendidus Androngo trivittatus Voeltzkowia lineata COLUBRIDAE Heteroliodon occipitalis Ithycyphus oursi Langaha madagascariensis Langaha pseudoalluaudi Leioheterodon modestus Leioheterodon madagascariensis Liophidium apperti Liophidium torquatus Liophidium vaillanti Liopholidophis lateralis Lycodras gaimardi Lycodras guentheri Pseudoxyrhopus quinquelineatus Pseudoxyrhopus kely TYPHLOPIDAE Ramphotyphlops braminus Tolagnaro, Petriky, Cap Ste. Marie; gallery forests, temporary swamps, and streams Beraketa, Tolagnaro, Petriky; gallery forests, temporary swamps, and streams Beraketa, many sites in spiny forest; temporary pools and streams Bereketa; rare in this part of its range Beraketa, Berenty; rare in this part of its range Berenty, spiny forest near Amboasary-Sud; dense brush Beraketa, Berenty; open, dry deciduous forests Beraketa, Berenty; open forests, relatively dry aspect, disturbed sites Bekinana, open slopes of Ambatorongorongo; dry deciduous forests Beraketa, Berenty, Petriky, Tolagnaro; spiny and dry deciduous for- ests, within villages Beraketa, Petriky; spiny and dry deciduous forests Beraketa, Berenty, Amboanemba (south of Tranomaro), Cap Ste. Ma- rie; spiny and dry deciduous forest Petriky, Berenty; dry deciduous forest Beraketa, Petriky; dry, often disturbed, spiny and dry deciduous for- est Near Amboasary-Sud; spiny forest? Rare, perhaps extinct Beraketa, many sites in SE Madagascar; common in spiny forest Beraketa, many sites in SE; open forests Beraketa; gallery and spiny forest Beraketa, Berenty, many sites in Madagascar, including SE Madagas- car; variable habitats Many sites in Madagascar, including SE Madagascar; variable habi- tats Beraketa, Petriky, Manambaro, many sites near Tolagnaro; temporary swamps and open areas Petriky, Berenty, many surrounding sites; open areas and lowland for- ests Beraketa; gallery and dry deciduous forests Amboanemba (south of Tranomaro); sandy areas in spiny and dry de- ciduous forests Berenty; littoral and gallery forests, mainly sandy areas Berenty, Manambaro, Petriky; Tolagnaro; sandy areas in a wide varie- ty of habitats including villages, agricultural plots, and gallery for- ests NUSSBAUM ET AL.: AMPHIBIANS AND REPTILES 171 tanosy and Pseudoxyrhopus kely, all of these spe- cies are now known from within parcel 1 of RNI d'Andohahela and are therefore afforded some protection. In addition, we now know that Uroplatus ma- lahelo has a much wider distribution than was previously known. We recorded it at Analavelona near Sakaraha and in RS de Kalambatritra. Uro- platus malama is currently recorded from the type locality in the Anosyenne Mountains north of An- dohahela, Kalambatritra (unpubl. data), and at one site within Andohahela. Recently discovered specimens of Uroplatus in rain forests of northern Madagascar are similar to U. malama, but their taxonomic status is uncertain. Although Parage- hyra gabriellae has not been found outside the Tolagnaro area, it has been recorded at many sites in the vicinity of Andohahela, including Amba- torongorongo, Grand Lavasoa, Manantantely, Vohisandria, and the Anosyenne Mountains, both south and north of parcel 1 of RNI d'Andohahela. Pseudoxyrhopus sokosoko is also limited locally but is found at many sites within parcel 1 and in nearby forests. Pseudoxyrhopus kely has recently been found in Berenty Reserve and a few other lowland sites near Tolagnaro. Phelsuma antanosy is perhaps of greatest con- cern. This species is known from only four sites, three of which are within the proposed mining zone for ilmenite sands along the coastal plains on either side of Tolagnaro. One of these three sites is the type locality, which was cleared for crop and charcoal production 1 year after the spe- cies was discovered. The other two localities within the mining zone are small blocks of littoral forest. The fourth site, Foret de Malahelo, on the west slope of Ambatorongorongo, is outside the mining zone just a few kilometers west of the type locality. The west slope of Ambatorongorongo, also the type locality of Uroplatus malahelo, had a small patch of degraded, low-elevation rain for- est as of 1995, but the forest was being selectively cut for wood, and the annual burning of the sur- rounding grasslands was destroying 3-10 m of the peripheral forest of Ambatorongorongo each year. Unless action is taken, the forest will be com- pletely eliminated within a few years. Phelsuma antanosy was not found in parcel 1 of the RNI d'Andohahela. The species is not easily detected, however, and a concerted effort is needed to de- termine whether it occurs in some of the lower elevation forests of parcel 1 or in surrounding for- ests. Acknowledgments Our field research in Madagascar was made possible through the cooperation of the Malagasy Ministere de l'Enseignement Superieur, the Min- istere de la Production Animale et des Eaux et Forets, and the Ministere de la Recherche Scien- tifique et Technologie pour le Developpement. Our fieldwork was sponsored by the Madagascar Minerals Project, QIT FER et Titane, Inc.; the U.S. National Science Foundation, US AID, the National Geographic Society, and Earthwatch. We received logistical support from the World Wide Fund for Nature — Madagascar, Conservation In- ternational, and CARE. Literature Cited Andreone, E, and H. Randriamahazo. 1997. Ecolog- ical and taxonomic observations on the amphibians and reptiles of the Anodohahela low altitude rainfor- est, S. Madagascar. Revue Francaise Aquariologie, 24(3-4): 95-129. Goodman, S. M., M. Pidgeon, A. F. A. Hawkins, and T. S. Schulenberg. 1997. The birds of southeastern Madagascar. Fieldiana: Zoology, n.s. 87: 1-132. Jenkins, P. D., C. J. Raxworthy, and R. A. Nussbaum. 1997. A new species of Microgale (Insectivora, Ten- recidae), with comments on the status of four other taxa of shrew tenrecs. Bulletin of the Natural History Museum, London (Zoology), 63(1): 1-12. Morris, D. S. 1994. A herpetological survey of Berenty Reserve, Madagascar. Undergraduate thesis, Depart- ment of Ecology and Evolutionary Biology, Princeton University. Nicoll, R. E., and O. Langrand. 1989. Madagascar: Revue de la conservation et des Aires Protegees. World Wide Fund for Nature, Gland, 374 pp. Nussbaum, R. A., and C. J. Raxworthy. 1994a. The genus Paragehyra (Reptilia: Sauria: Gekkonidae) in southern Madagascar. Journal of Zoology, London, 232: 37-59. . 1994b. A new species of Mabuya Fitzinger (Reptilia: Squamata: Scincidae) from southern Mad- agascar, Herpetologica, 50(3): 309-319. 1994c. A new species of Uroplatus Dumeril (Reptilia: Squamata: Gekkonidae) from southern Madagascar. Herpetologica, 50(3): 319-325. -. 1994d. A new rainforest gecko of the genus Paroedura Giinther from Madagascar. Herpetological Natural History, 2(1): 43-49. 1995a. New Uroplatus Dumeril (Reptilia: Squamata: Gekkonidae) of the ebenaui-group from the Anosy Mountains of southern Madagascar. Co- peia, 1995(1): 118-124. . 1995b. Review of the scincine genus Pseudoa- contias Barboza du Bocage (Reptilia: Squamata: Scin- cidae) of Madagascar. Herpetologica, 51(1): 91-99. . 1995c. A new Mabuya (Reptilia: Squamata: 172 FIELDIANA: ZOOLOGY Scincidae) of the aureopunctata-group from southern Madagascar. Journal of Herpetology, 29(1): 28-38. . 1998a. New long-tailed Mabuya Fitzinger from Lokobe Reserve. Nosy Be, Madagascar (Reptilia: Squamata: Scincidae). Copeia, 1998(1): 114-119. 1998b. Revision of the genus Ebenavia Boett- ger (Reptilia: Squamata: Gekkonidae). Herpetologica, 54(1): 18-34. Nussbaum, R. A., F. Andreone, and C. J. Raxworthy. 1998a. New rain-forest species of Pseudoxyrhopus Giinther (Squamata: Colubridae) from northern Mad- agascar. Copeia, 1998(1): 128-132. Nussbaum, R. A., C. J. Raxworthy, and O. Pronk. 1998b. The ghost geckos of Madagascar: A further revision of the Malagasy leaf-toed geckos (Reptilia, Squamata, Gekkonidae). Miscellaneous Publications, Museum of Zoology, University of Michigan, no. 186: 1-26. Paulian, R., C. Blanc, J.-L. Guillaumet, J.-M. Betsch, P. Griveaud, and A. Peyrieras. 1973. Etude des eco- systemes montagnards dans la region malgache. II. Les chaines Anosyennes. Geomorphologie, climato- logie et groupements vegetaux. (Campagne RCP 225, 1971-1972). Bulletin du Museum Histoire Naturelle, Paris, serie 3, no. 118, Ecologie Generate 1:1-40. Ramanamanjato, J. B. 1993. Contribution a 1' etude des reptiles et amphibiens de la foret ombrophile du sudest de Madagascar. Memoire de D.E.A. des Sciences Biologiques Appliquees, Universite d'Antananarivo, Antananarivo. Raselimanana, A. P. 1993. Contribution a l'etude de la batrachofaune et de l'herpetaufaune des zones semi- arides et des zones littorales de Fextreme sud-est de Madagascar. Memoire de D.E.A. des Sciences Biolo- giques Appliquees, Universite d'Antananarivo, Anta- nanarivo. Raxworthy, C. J., and R. A. Nussbaum. 1993a. Four new species of Amphiglossus from Madagascar (Squa- mata: Scincidae). Herpetologica, 49(3): 326-341. . 1993b. A new Madagascan Phelsuma with a review of Phelsuma trilineata and comments on Phel- suma cepediana in Madagascar. Herpetologica. 49(3): 342-349. . 1994a. A review of the Madagascan snake gen- era Pseudoxyrhopus, Pararhadinaea, and Heterolio- don (Squamata: Colubridae). Miscellaneous Publica- tions, Museum of Zoology, University of Michigan, no. 182: 1-37. . 1994b. A partial systematic revision of the day geckos, Phelsuma Gray, of Madagascar (Reptilia; Squamata: Gekkonidae). Zoological Journal of the Linnean Society, 112: 321-335. . 1994c. A rainforest survey of amphibians, rep- tiles and small mammals at Montagne d'Ambre. Mad- agascar. Conservation Biology, 69: 65-73. 1995. Systematics, speciation and biogeogra- phy of the dwarf chameleons (Brookesia; Reptilia, Squamata, Chameleontidae) of northern Madagascar. Journal of Zoology, London. 235: 525-558. 1996a. Amphibians and reptiles of the Reserve Nataturelle Integrate d'Andringitra. Madagascar: A study of elevational distribution and local endemicity. pp. 158-170. In Goodman, S. M.. ed. A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrate d'Andringitra, Madagascar: With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. 1996b. Montane amphibian and reptile com- munities in Madagascar. Conservation Biology, 10(3): 750-756. . 1996c. Patterns of endemism for terrestrial ver- tebrates in eastern Madagascar, pp. 369-383. In Lour- enco, W. R., ed. Biogeographie de Madagascar, Edi- tions ORSTOM, Paris. 1997. Biogeographie patterns of reptiles in eastern Madagascar, pp. 124-141. In Goodman, S. M., and B. D. Patterson, eds. Natural Change and Human Impact in Madagascar. Smithsonian Institution Press, Washington, D.C. 432 pp. Raxworthy, C. J., F. Andreone, R. A. Nussbaum, N. Rabibisoa, and H. Randriamahazo. 1998. Amphib- ians and reptiles of the Anjanaharibe-Sud Massif, Madagascar: Elevational distributions and regional en- demicity, pp. 79-92. In Goodman, S. M., ed. A floral and faunal inventory of the Reserve Speciale d' Anjanaharibe-Sud, Madagascar: With reference to elevational variation, Fieldiana:Zoology, n.s. 90: 1- 245. NUSSBAUM ET AL.: AMPHIBIANS AND REPTILES 173 Chapter 12 Bird Community Variation with Elevation and Habitat in Parcels 1 and 2 of the Reserve Naturelle Integrate d'Andohahela, Madagascar A. F. A. Hawkins' and Steven M. Goodman2 Abstract A study conducted between 19 October and 14 December 1995 in the Reserve Naturelle Integrale d'Andohahela, southeastern Madagascar, found 123 species of birds, a high total for a small (<700 km:) area. This high species richness was due to the variety of habitats sampled: humid forest from 400 to 1875 m, spiny forest, and open areas. Of these the lowland humid forest is considered to be the highest priority for bird conservation management. The bird community in the humid forest sample was relatively homogeneous, with a low turnover rate corresponding to changes in altitude; only 14 of 50 species, however, were found evenly over the elevational gradient. Species richness was slightly greater at mid-elevation and lower at high elevation. Species richness estimates within elevational zones were found to be distorted unrealistically by compensation for sampling area. This was because most sampling sites within an elevational sample shared the same species. The bird community in the spiny forest was very different from that in the humid forest; of 86 forest species found in all habitats, only 26 occurred in both humid and spiny forest. Raptors and frugivores made up a disproportionately large part of these 26 species, whereas small insectivores were underrepresented. Resume Une etude conduite entre 19 octobre et 14 decembre 1995 dans la Reserve Naturelle Integrale d'Andohahela a permis de deceler 123 especes d'oiseaux, une proportion elevee pour une petite etendue de Madagascar. Ceci peut s'expliquer par la variete d'habitats selectionnee comme echantillons: foret humide comprise entre 400 m et 1875 m d'altitude, foret scche, et des zones ouvert. La foret humide de basse altitude est considered comme la priorite la plus pressante en matiere de gestion de conservation. La communaute d'oiseaux rencontree a 1'interieur de l'echantillon de foret humide est relativement homogene, avec un taux d'apparition de plus en plus faible en montant en altitude; toutefois 14 especes seulement sur 50 sont rencontrees regulierement sur une pente de meme altitude. La richesse en espece est legerement plus eleve en moyenne altitude, moins eleve a haute altitude. Les estimations en richesse en cette espece selon les niveaux d'altitude s'averent erronees si on se base aux zones d'echantillonnage, etant donne que la plupart des sites d'echantillonnage sur une altitude determinee renferment les memes especes. La communaute d'oiseaux dans la foret seche est tres differente; sur toutes les 86 especes identifiers, 26 seulement apparaissent dans la foret humide ainsi que dans la foret 1 BirdLife International. B.P. 1074. Antananarivo (101). Madagascar. : Field Museum of Natural History, Roosevelt Road at Lake Shore Drive, Chicago, IL 60605-2496, U.S.A. HAWKINS & GOODMAN: BIRD COMMUNITIES 175 seche. Les rapaces et frugivores composent une partie disproportionnee de ces 26 especes. Les petits insectivores sont sous-representes. Introduction The Reserve Naturelle Integrate (RNI) d'Andohahela, with its three disjunct parcels (see Chapter 1), is unusual in comparison to most pro- tected areas of Madagascar in that it encompasses a remarkable variety of habitats within a single reserve. These habitats include the humid forest and high mountain zones of parcel 1, the spiny forest and remnant gallery forest of parcel 2, and the transitional zone between humid and spiny forest of parcel 3. The abrupt ecotones between the parcels are reflected in dramatic changes in the bird species inhabiting them. A recently pub- lished monograph on the birds of southeastern Madagascar included ornithological information from the 1995 survey of the RNI d'Andohahela and earlier research conducted in the reserve as well as surrounding areas (Goodman et al., 1997). One of the analyses presented in that study shows high levels of species turnover between the par- cels. The bird fauna of the eastern humid forest of Madagascar, a region extending nearly 1,200 km between parcel 1 of the RNI d'Andohahela and the northernmost site it was compared to, is more homogeneous than the bird population found in parcels 1 and 2 of the RNI d'Andohahela, which are separated by a few kilometers. In this chapter we concentrate on the Andohahela region, review information on the elevational distribution of birds in the reserve, examine species turnover between five different elevational zones in the hu- mid forest of parcel 1 and the single site studied in the spiny forest of parcel 2, and provide esti- mates of densities for certain species. Data on the bird species in parcel 3 are not extensive, and we limit our comparisons to parcels 1 and 2. For ad- ditional details on the natural history of bird spe- cies that occur within the reserve, see Goodman et al. (1997). Materials and Methods Data were collected within a survey area of ±100 m in elevation and less than 3 km in hori- zontal distance from survey area centers (= camps) established at 440, 810, 1200, 1500, and 1875 m in the humid forest of parcel 1, and at 120 m in the spiny forest of parcel 2. Species lists were compiled by direct observation while ob- servers walked along forest trails, by call-play- back of bird vocalizations using a tape recorder, from static observations made from broken-can- opy watch points, and by mist-netting. Specific details on procedures associated with static ob- servation and mist-netting and their results are given in Goodman et al. (1997). Point counts were made at 150 m intervals (in humid forest) or 200 m intervals (in spiny forest) along marked and measured preexisting and new- ly cut forest trails. A minimum of 12 point count sites were sampled within each elevational zone. Where possible a minimum of five point count sites were established within each zone in habitat in each of the following locations: ridge, slope, and valley bottom. Each point count site was sampled twice, on different days, once between 0430 and 0600 hr and once between 0630 and 0900 hr. During each sample count, which lasted for 10 minutes, the following data were noted on each bird contact: species, estimated distance from observer (to nearest 10 m), nature of initial contact (song, call, wing-noise, or visual), and time of contact. This methodology differs slightly from that used in other inventories (Goodman & Putnam, 1996; Hawkins et al., 1998); findings of previous sur- veys (Hawkins et al., 1998) indicated that where many (up to five) repeat samples of fewer point count sites were made, much of the data obtained was unusable for density calculations because a large percentage of contacts could not be regarded as statistically independent. In the present survey an effort was made to obtain the largest number of independent samples possible within an ele- vational zone. Sampling thoroughness for each elevational zone was assessed by examination of species accumulation curves. These curves were calculated based on the day that a species was first detected. It would be possible to evaluate changes in bird species richness with elevation in the humid forest sites by simply using the number of species within each zone. However, this approach does not take into account varying sampling effort, particularly with regard to the surface area sampled. A poten- tial solution is to compare the logarithm of the 176 FIELDIANA: ZOOLOGY surface area sampled (the number of point counts made at an elevational sample is used here as an index of area sampled) with the logarithm of spe- cies richness (Rahbek, 1995). An alternative method is to standardize and compare the number of species recorded within an elevational zone over a similar number of samples or days. Species recorded in parcel 1 were scored as oc- curring in one of five elevational categories ac- cording to their presence and relative abundance at sample sites. Species occurring only at 440 and 810 m were classified as lowland species, those occurring only between 440 and 1200 m as low- to mid-elevation species, those occurring only be- tween 810 and 1500 m as mid-elevation species, those only between 810 and 1875 m as mid- to high-elevation species, and those occurring only between 1200 and 1875 m as high-elevation spe- cies. Species occurring more or less equally at all sites were considered generalists. Species abun- dance within an elevational zone was derived from contact frequency on point counts. For some species that rarely vocalized or vocalized quietly (e.g., Philepitta castanea and Pseudobias wardi), casual contacts revealed that the species was clearly more abundant than contacts on point counts indicated. In these cases the elevational category was adjusted accordingly. Species noted only once within an elevational zone were not considered to occur regularly at that elevation. Only species falling clearly into the elevational categories described above were used in the sub- sequent analysis. To assess the relationships between bird com- munities in sample sites within the reserve, we calculated the Jaccard Index of similarity for the distribution of breeding bird species: r Jaccard Index = Nl + N2 - C where Nl = the number of species at site 1, N2 = the number of species at site 2, and C = the number of species common to both sites. The co- efficients from these indices were used in a cluster algorithm ("Phylip," written by J. Felsenstein us- ing the Fitch-Margoliash method). The computer program DISTANCE (Laake et al., 1993) was used to calculate density estimates for species where sufficient independent contacts (18 per elevational zone) were made. For each species, the highest of the two counts made at each point count site was used. Several species for which the number of contacts was relatively high were excluded from the analysis because most contacts with these species (Cuculus rochii, Coracopsis spp., and Leptosomus discolor) oc- curred at an estimated distance of more than 200 m. Such data produce very unreliable density es- timates (Hawkins et al., 1998) owing to the dif- ficulty of estimating comparatively long distances correctly (Buckland et al., 1993). Results One hundred twenty-three bird species were re- corded during the survey (Table 12-1). This is an exceptionally high figure for a small geographical area of Madagascar and reflects the diversity of habitats sampled — low-, mid-, and high-elevation humid forest; spiny forest; gallery forest; savanna; and riverine habitats. In parcel 1 a total of 59 species were recorded in the 440 m zone, 64 in the 810 m zone, 63 in the 1200 m zone, 47 in the 1500 m zone, and 38 in the 1875 m zone; there were 81 species recorded at 120 m in parcel 2. The ratios of species within these zones utilizing forest and mixed forest/open habitats to those oc- curring in open and aquatic habitats were 10.8, 9.8, 1 1.4, 46.0, 8.3, and 1.3, respectively. It is thus clear that one of the main differences between the bird faunas of parcel 1 and parcel 2 is the much higher percentage of species using open and aquatic habitats in parcel 2. The site surveyed in parcel 2 was adjacent to a tributary of the Man- drare River and areas of aquatic habitat, including a relatively slow-moving river and marshes. These biotopes were largely not present in the various zones surveyed in parcel 1 . The percentage of for- est and mixed-habitat birds was similar at all hu- mid forest sites below 1875 m (Table 12-1). The percentage of forest species was lower at 1875 m. This reflects the presence of a more or less con- stant number of aerial species (those that forage in the air outside the forest body) over the whole elevational gradient. Aerial species, as well as nu- merous open habitat species, make up a higher proportion of the total in the species-poor summit area. Sampling effort varied in parcel 1 from 12 point counts at 1875 m to 21 at 810 m (Table 12- 1). The steepness of slopes, particularly at the 1200 m and 1875 m sites, reduced the potential for increasing sample size. On the basis of the species accumulation curves (Fig. 12-1), more than 90% of the species likely to be present at a site were recorded within the first 3-5 days. HAWKINS & GOODMAN: BIRD COMMUNITIES 177 Table 12-1. Numbers of species, forest species, and species restricted to a sampling site for all sampling sites in this study of the RNI d'Andohahela. Humid forest sites (parcel 1) Spiny Descriptive parameter 440 m 810 m 1200 m 1500 m 1875 m torest (parcel 2) Total 10 10 10 9 7 9 55 17 21 15 20 12 28 113 59 64 63 47 38 81 123 57 61 60 47 31 50 86 (97) (95) (95) (100) (82) (62) (69) 64 70 78 78 79 5 1 1 1 1 40 — Days of survey effort Number of point count sites Number of bird species Number of forest bird species (% of total) Percentage of species recorded on point counts Number of bird species restricted to that site The humid forest sites can be divided into two groups (species-poor and species-rich) based on the number of species present (Fig. 12-2A). De- spite the disparity in the number of point count sites sampled in the 440, 810, and 1200 m sam- ples, the number of species present within each is similar. The two high-elevation sites have distinct- ly fewer species. Of the humid forest sites sam- pled, the 440 m zone had by far the largest num- ber of unique species (five; Table 12-2). All other elevational zones shared almost all species with at least one other zone. Biogeography Given the differences in the number of species in open and forested habitats between the two par- cels, the Jaccard Index has been calculated in two ways: for the overall avifauna and for species that occur only in forested and mixed habitats. Anal- ysis of the complete avifauna in the five surveyed zones in parcel 1 and the single study site in par- cel 2 (Table 12-2; Fig. 12-3 A) indicates that the sites within parcel 1 form a distinct cluster. Within these sites the 810 m and 1200 m zones have a very similar avifauna forming the upper lowland and mid-montane community; these zones are then next closest in faunal similarity to 440 m, followed by 1500 m, and then finally 1875 m. Parcel 2 has a distinctly different avifauna than the humid forest sites. On the basis of this anal- ysis the spiny forest parcel has an avifauna closest to that of the high mountain site in humid forest, at 1875 m. When the analysis is repeated without the 90 ♦ ♦--♦ fi^te^F?^ 4- -A- -A -O' .,4c tA- A A A— " '-"-'- -m H- - - - -a eb- - - ~m — D— 440 m o 810m — -o — - 1200 m A 1500 m - - -B8- - - 1875 m — •♦— • Parcel 2 4 5 6 7 Days sampled 10 1 1 Fig. 12-1. Species accumulation curves for the five elevational zones in humid forest of parcel 1 and the spiny forest of parcel 2 in the RNI d'Andohahela. 178 FIELDIANA: ZOOLOGY M» 1000 1200 1400 Elevation (m) ill 600 MO 1000 1200 1400 1600 IKO 2000 Elevation (m) 200 400 600 MO 1000 1200 1400 1600 1100 2000 Elevation (m) WO 1000 1200 1400 Elevation (m) Fig. 12-2. The number of species recorded within each elevational zone in humid forest in the RNI d'Andohahela as measured by all techniques combined (A), after 12 point counts (B), after 7 days of sampling using all techniques combined (C), as adjusted for surface area sampled (D; Rahbek, 1995), and total number of species recorded by all methods adjusted for the number of survey days (E). aquatic, open habitat, and aerial species, the same basic pattern exists (Fig. 12-3B). The only subtle difference with the latter analysis is that the dis- tances between nodes within the 440, 810, 1200, and 1500 m cluster are smaller, whereas the 1875 m site is further from the lower humid forest sites. Methodological Considerations In this section we consider the relationships be- tween elevation and species richness and we com- pare methods of adjusting for sampling effort. Ex- cept where stated, the total number of species re- corded at a sample site is used. The relationship between elevational zones and unadjusted species richness is illustrated in Figure 12-2 A. A similar relationship is shown if sampling time is kept con- stant (7 days), because of the low number of pre- viously unrecorded species occurring within each elevational zone after 7 days (Fig. 12-2C). If an adjustment is made for area sampled following the method of Rahbek (1995; Fig. 12-2D), however, sites with fewer samples (1200 m and 1875 m) come out as much more species-rich (Fig. 12-2B). When sample size is controlled for by dividing the number of species recorded in each zone by the number of days of sampling (as an index of sampling effort; Fig. 12-2E), the results are sim- ilar to those shown in Figure 12-2A. Again, the sampling zones with lower effort (particularly 1875 m) come out as more species-rich than in the unadjusted analysis. Controlling for sample size by taking the num- ber of species discovered using a fixed number of point count sites (12, the minimum at any site; Fig. 12-2B) produces a result similar to that of the unadjusted analysis (Fig. 12-2A), with the excep- tion that the lower elevational zones come out as rather species-poor. This is due to the lower per- centages of the total number of species present at 440 m and 810 m that were recorded on point counts (Table 12-1). The species accumulation curves for each elevational zone (Fig. 12-4) show that the general pattern occurs because species HAWKINS & GOODMAN: BIRD COMMUNITIES 179 Table 12-2. Presence of bird species within elevational zones in the RNI d'Andohahela. Species Parcel 1 Else- where 440 810 1200 1500 1875 in Elevational Parcel Status m m m m m parcel 1 category 2 Phalacrocorax africanus N Nycticorax nycticorax N Ardeola ralloid.es N Ardeola idae * Bubulcus ibis N Butorides striatus N Egretta ardesiaca N Egretta dimorpha (*) Egretta alba N Ardea purpurea N Ardea cinerea N Scopus umbretta N Lophotibis cristata# * + + Dendrocygna bicolor N Dendrocygna viduata N Sarkidiornis melanotos N Anas erythrorhyncha N Aviceda madagascariensis# * + Milvus migrans N Polyboroides radiatus# * Accipiter henstii# * + + Accipiter francesii# (*) + + + Buteo brachypterus# * 1 3 1 2 + Falco newtoni (*) Falco zoniventris * Falco eleonorae M Falco concolor M Margaroperdix madagasca riensis * Numida meleagris I Turnix nigricollis# (*) Dryolimnas cuvieri (*) + Canirallus kioloidestt * + + 1 2 Sarothrura insularis# * + + + Gallinula chloropus N Tringa nebularia M Actitis hypoleuca M Pterocles personatus * Streptopelia picturatatt (*) 1 6 3 16 4 Oena capensis N Treron australisft (*) + Alectroenas madagascariensis# * + 1 + + + Coracopsis vasa# (*) 4 1 2 7 1 Coracopsis nigra# (*) 7 17 16 22 15 Agapornis cana# * Cuculus rochiiU (*) 25 30 23 15 + Coua gigas# * Coua reynaudiW * 1 2 5 10 6 Coua cursor# * Coua ruficeps# * Coua caerulea# * 6 12 6 4 2 Coua cristataU * Centropus toulou# (*) 3 0 5 2 Otus rutilustt (*) + + + + L-M L-M ?G G G ?M G 14 + L + G ?G 9 G 8 6 L-M 28 16 ?M-H 9 20 G 18 L-M 29 L-M + 180 FIELDIANA: ZOOLOGY Table 12-2. Continued. Parcel 1 Else- where 440 810 1200 1500 1875 in Elevational Parcel Species Status m m m m m parcel 1 category 2 Ninox superciliarisM * + Asio madagascariensis# * + + + L-M Caprimulgus madagasca riensis# (*) + Caprimulgus enarratusti * + Zoonavena grandidieriU (*) 1 1 + + + Cypsiurus parvus N + + + + Apus melba N + + + + + Apus barbatus N + + + + + + Alcedo vintsioides (*) + + + Ispidina madagascariensisU * + + + L-M Merops superciliosus N + + Eurystomus glaucurus N + 5 Brachypteracias leptosomusU * 3 1 5 + L-M Brachypteracias squamiger# * 1 L Atelornis pittoidestt * 6 M Atelornis cross ley i# * 10 17 2 M-H Leptosomus discolor# * 3 9 5 4 4 G 5 Upupa epops# N 1 Philepitta castaneaU * + 1 1 2 + G Neodrepanis coruscanstf * 9 4 M Neodrepanis hypoxantha# * + 8 9 H Mirafra hova * + 14 Riparia paludicola * + Phedina borbonica * + + + + Motacilla flaviventris * + + + + Coracina cinereatt * 2 6 3 3 2 G + Phyllastrephus madagasca riensis# * 22 19 3 L-M Phyllastrephus zosteropstt * 15 9 4 L-M Phyllastrephus cinereicepsU * + 4 3 5 M-H Hypsipetes madagascariensisU * 17 22 5 3 + L-M 23 Copsychus albospecularis# * 7 12 3 L-M 24 Sa.\icola torquatatt * + Pseudocossyphus sharpeitt * + 10 13 8 M-H Acrocephalus newtoni * + Nesillas typicatf * 7 12 22 23 M-H Thamnornis chloropetoidesti * 9 Cisticola cherina (*) + + Dromaeocercus brunneus# * 3 3 5 M-H Randia pseudozosteropstf * 14 10 1 2 L-M Newtonia amphichroatt * 5 14 8 5 M-H New ton ia archboldi# * 7 Newtonia brunneicaudaU * 11 16 8 12 10 G 38 Newtonia fanovanaeU * 7 L Neomixis tenel\a# * 23 12 4 L-M 23 Neomixis viridistt * 6 16 11 10 8 G Neomixis striatigula# * 17 22 8 3 L-M 18 HAWKINS & GOODMAN: BIRD COMMUNITIES 181 Table 12-2. Continued. Parcel 1 Else- where 440 810 1200 1500 1875 in Elevational Parcel Species Status m Ml m m m parcel 1 category 2 Cryptosylvicola randrianasoloi# * 2 12 22 11 M-H Hartertula flavoviridistt * + 2 1 + ?M Pseudobias wardi# * + + 2 + L-M Terpsiphone mutata# * 13 14 2 3 L-M 23 Oxylabes madagascariensis# * + 5 3 + 1 ?G Crossleyia xanthophrys# * 2 1 6 1 M-H Mystacornis crossleyW * 2 3 2 3 L-M Nectarinia souimangatt (*) 10 22 16 27 26 G 18 Nectarinia notataU * 1 1 1 L-M 4 Zosterops maderaspatanaU (*) 20 25 11 15 11 G 1 Calicalicus madagascariensistt * 16 10 4 5 L-M Schetba rufa# * 6 2 L Vanga curvirostris# * 3 1 ?L-M 16 Xenopirostris xenopirostris^ * 5 Xenopirostris pollenitt * 5 8 4 3 M-H Falculea palliata# * 12 Leptopterus viridistt * 3 7 9 4 L-M 11 Leptopterus chabert# * + 3 + ?L-M 2 Cyanolanius madagascarinustt (*) 9 3 5 2 L-M Hypositta corallirostris# * + L Tylas eduardi# * 18 17 11 5 1 L-M Dicrurus forficatustt * 11 7 6 + + L-M 23 Corvus albus# N + Hartlaubius auratustt * 1 3 + L-M Acridotheres tristis I + Ploceus nelicourvW * 1 5 + 2 + G Ploceus sakalava# * 7 Foudia omissatt * 1 7 9 17 5 M-H Foudia madagascariensisU (*) + ?H 14 Lonchura nana * + Species are listed under "Elsewhere in parcel 1 " only if they were seen outside the elevational zones. The numerals indicate number of independent contacts with a species from point counts. A plus symbol ( + ) indicates that the species was recorded within an elevational sample but not on point counts. The pound (#) symbol indicates a species that feeds or breeds regularly in native forest. Under the heading "Status": N = nonendemic resident breeder in Madagascar; * = Malagasy resident breeder; (*) = Malagasy regional endemic breeder; M = migrant nonbreeding visitor; I = introduced. Under the heading "Elevational category," only humid forest (parcel 1) species are analyzed: L = lowland species; L-M = low- to mid-elevation species; M = mid-elevation species; M-H = mid- and high- elevation species; H = high-elevation species; G = elevational generalist. were recorded at a faster rate over a series of point count sites in species-rich elevational zones. Species Densities Density estimates for certain bird species in parcel 1 of the RNI d'Andohahela are presented in Table 12-3. The calculated densities do not show any marked variation with elevation except in the cases of Nesillas typica and Nectarinia souimanga, both of which appear to be more com- mon at 1875 m than elsewhere. The species listed in Table 12-3 are less abundant at other altitudes, where frequency of contact was too low for den- sity calculation. 182 FIELDIANA: ZOOLOGY 0.05 0.04 0.15 0.16 0.18 0.24 0.39 B 0.08 0.06 0.02 0.14 0.12 0.16 0.24 0.36 0.10 1200 m 810 m 440 m 1500 m 1875 m 120 m 1200 m 810 m 440 m 1500 m 1875 m 120 m Fig. 12-3. Cluster analysis of faunal similarity of the resident avifauna of the sites surveyed in the RNI d'Andohahela. The coefficients were derived from the Jaccard Index for the general resident bird community (A) and restricted to forest-dwelling species (B). See text (p. 177) for definition of the Jaccard Index. Discussion Biogeography The bird community in the humid forest (parcel 1) of RNI d'Andohahela is relatively homoge- neous, and there are no clear abrupt shifts in the species that occur between 440 and 1500 m. At a finer level of analysis the species in the 810 and 1200 m zones are more similar to one another than they are to those in the 440 m zone. This is the result of a few lowland species (e.g., Treron australis, Brachypteracias squamiger, Newtonia fanovanae, and Hypositta corallirostris) being confined to the 440 m zone. As one moves up the slopes into the 1500 and 1875 m zones, there is a decrease in similarity, but the Jaccard Index co- efficients between these two zones in both anal- yses are close to 0.60. The dramatic shift in hab- itats between parcel 1 and parcel 2 is mirrored by 5 10 15 20 25 Number of point count sites Fig. 12-4. Species accumulation curves based on point counts for each elevational zone. a change in the avifaunal community. The bird community of parcel 2 is more similar to that in the 1 875 m zone of parcel 1 than to any other site along the slopes of the humid forest of the re- serve. Insofar as the humid forest in parcel 1 of the reserve ends abruptly along the western flank of the Anosyenne Mountains and the spiny forest habitat is geographically closer to the 1 875 m site than most of the other sites surveyed in parcel 1 , the faunistic relationship between the 1875 m zone and parcel 2 might simply reflect the ability of certain species to move between these habitats. Species Distributions Most humid forest species for which sufficient data are available are either elevational generalists or became more scarce above 1500 m. Five spe- cies were limited to lowland, 20 to lowland or mid-elevation, three to mid-elevation, seven to mid- to high elevation, three to high elevation, and 14 were elevational generalists. Thus, few species were completely restricted to the lowland or montane habitats. Of the 50 species for which there were enough data to permit analysis, 72% (36 of 50) showed clear elevational limits. Over- all, 23 species (63% of elevation-restricted spe- cies) were not recorded at higher elevations, whereas 10 species (27%) were not found in low- land samples. A striking feature of this study is the abundance of frugivorous species at 1500 m (Table 12-2) — there were considerably higher numbers of two pigeons (Streptopelia picturata and Alectroenas madagascariensis) and both Coracopsis parrots in HAWKINS & GOODMAN: BIRD COMMUNITIES 183 Table 12-3. Densities of species in the RNI d'Andohahela with more than 18 contacts per elevational zone, with certain exceptions (see text, p. 177). Species (Elevation) Atelornis crossleyi (1500 m) Neomixis tenella (440 m) Cryptosylvicola randrianansoloi (1500 m) Phyllastrephus madagascariensis (440 m) Phyllastrephus madagascariensis (810 m) Nesillas typica (1500 m) Nesillas typica (1875 m) Zosterops maderaspatana (440 m) Zosterops maderaspatana (810 m) Nectarinia souimanga (810 m) Nectarinia souimanga (1200 m) Nectarinia souimanga (1500 m) Nectarinia souimanga (1875 m) Tylas eduardi (440 m) Mean Lower 95% Upper 95% Percentage density confidence confidence coeffi- estimate interval interval cient of Degrees of (birds/km2) (birds/km2) (birds/km2) variation freedom 67.6 34.8 131.4 33 30 436.9 337.2 566.0 12.7 27 119.5 68.5 208.5 27.9 34 214.3 132.9 345.5 24.1 48 156.9 94.5 260.5 24.1 48 792.0 503.8 1,245.2 22.8 51 1,265.1 855.5 1,870.9 19.7 48 688.2 416.7 1,136.7 25.2 38 607.8 408.2 905.0 20.1 63 305.5 222.4 419.9 16.1 80 291.7 213.7 398.2 15.7 67 393.8 283.6 546.7 16.6 71 631.9 485.5 822.5 13.4 97 124.1 77.1 199.7 24.1 55 Densities were calculated with the computer program DISTANCE (Laake et al., 1993) from data collected during point counts. this zone than elsewhere in parcel 1 (the abun- dance of Alectroenas madagascariensis was esti- mated from casual encounters). This may have been due to the large numbers of fruiting trees at this altitude, but there is no botanical evidence to support such a speculation (see Chapter 4). On casual inspection there appear to be no other ob- vious or significant relationships between eleva- tional distribution and dietary preference (Table 12-2). The forest species present in the spiny forest in parcel 2 are markedly different from those in the humid forest. Of a total of 86 forest species for the combined samples, only 26 occurred in both dry and humid forest. These included four raptors (Aviceda madagascariensis, Polyboroides radia- tus, Accipiter francesii, Buteo brachypterus), three larger frugivores (the two species of Cora- copsis and Streptopelia picturata), one owl (Otus rutilus), six large sally-gleaning or gleaning in- sectivores (Cuculus rochii, Leptosomus discolor, Coracina cinerea, Vanga curvirostris, Leptopte- rus viridis, and Dicrurus forficatus), four small sally-gleaning or sallying insectivores (Neomixis tenella, N. striatigula, Newtonia brunneicauda and Terpsiphone mutata), two insectivore/nectar- ivores (two species of Nectarinia), one terrestrial insectivore (Copsychus albospecularis), and three small frugivore/granivores (Hypsipetes madagas- cariensis, Zosterops maderaspatana, and Foudia madagascariensis). Absent or underrepresented on this list are the bark-feeders and terrestrial in- 184 sectivores. Frugivores and raptors appear to be overrepresented. Of the 26 forest species common to the two habitats, at least nine (Coracopsis vasa, C. nigra, Coracina cinerea, Copsychus albospecularis, Neomixis tenella, N. striatigula, Nectarinia soui- manga, Vanga curvirostris, and Leptopterus viri- dis) occur in the different parcels as distinct sub- species (Goodman et al., 1997). In addition, Otus rutilus populations in the two forest types have different songs (Goodman et al., 1997). All of these species except Coracopsis nigra and Nec- tarinia souimanga are restricted to lower eleva- tion humid forest (<1500 m). Neomixis tenella and N. striatigula were both absent from humid forest between 1500 and 1950 m on the eastern side of the RNI d'Andohahela, but they were present in transitional forest at about 1500 m on the western slope of the massif. Coracopsis nigra and Nectarinia souimanga were also present in these transitional forests. For these four species, it is not known which subspecies occur in the transitional forest. The fact that the nine species listed above occur as separate forms in the two forests suggests that populations of these species in the two habitats are separate and do not experience significant gene interchange. Of the other 17 forest birds that occur in both habitats, at least the four raptors, Leptosomus discolor, Cuculus rochii, Zoonavena grandidieri, and Hypsipetes madagascariensis, are capable of relatively long-distance movement FIELDIANA: ZOOLOGY and so might be expected to show dispersal across the mountain chain. Thus, only eight forest bird species (9% of the total), Streptopelia picturata, Centropus toulou, Newtonia brunneicauda, Terp- siphone mutata, Nectarinia notata, Zosterops maderaspatana, Dicrurus forficatus, and Foudia madagascahensis, appear to be phenotypically identical in both forest types. Methodological Considerations In parcel 1 of the RNI d'Andohahela, species richness only starts to drop off at about 1,200 m (Fig. 12-2). Correction for surface area (as sug- gested by Rahbek, 1995) or controlling for sam- pling effort by using the number of days of sam- pling associated with measures of species richness inflates species richness unrealistically in sites with few samples. The reason for this seems to be that most point count stations within an elevation- al zone generally contained the same species. Point count sites at species-rich elevations simply hold more species than those at species-poor el- evations; increasing the point count sample size beyond 1 2 or 15 does not increase the number of species recorded at an elevational sample. One should thus evaluate the variation in species rich- ness as a function of elevation for humid forest birds in Madagascar without correcting for sur- face area sampled or controlling for overall sam- pling effort. The only form of controlling for sam- pling effort should be the use of a similar sam- pling effort (whether surface area, point counts, or days) at each site. It appears that a relatively smaller proportion of the total species is recorded on point counts at lower elevations (Table 12-1), so the number of days spent sampling (including sampling by active searching) would probably be a more appropriate measure to keep constant were a comparison between different elevational zones required. Species Densities Species densities are broadly comparable to those calculated in the Reserve Speciale (RS) d'Anjanaharibe-Sud (Hawkins et al., 1998), and the RNI de Zahamena (Hawkins et al., in press). For instance, the calculated density of Atelornis crossleyi in the higher altitude forest varied from 43 to about 66 singing individuals per km2 in the RS d'Anjanaharibe-Sud, 98 individuals per km2 in the RNI de Zahamena, and 67 individuals per km2 in parcel 1 of the RNI d'Andohahela (Table 12- 3). Cryptosylvicola randrianasoloi occurred at be- tween 21 and 135 individuals per km2 in the RS d'Anjanaharibe-Sud, compared to about 119 in- dividuals km2 in the present study (Table 12-3). The density of Phyllastrephus madagascahensis was higher in parcel 1 of the RNI d'Andohahela than in other sites; it varied from 67 individuals per km2 in the RS d'Anjanaharibe-Sud to 53 per km2 in the RNI de Zahamena, and it was between 157 and 214 per km2 in the present study. Den- sities of Nesillas typica are often extremely high, especially in montane forests. In the RS d'Anjanaharibe-Sud, about 267 individuals per km2 were estimated at 1260 m, rising to 1,807 per km2 at 1650 m and 1,547 per km2 at 1950 m. At 1500 m in the RNI de Zahamena, a density of about 2,600 individuals per km2 was estimated. In the RNI d'Andohahela, estimates ranged from around 790 per km2 at 1500 m to 1,265 per km2 at 1875 m. Conservation and Management Implications With regard to conservation management, the most important ornithological observations re- corded during the inventory of the RNI d'Andohahela included the presence of good pop- ulations of Newtonia fanovanae at 440 m and the occurrence of Xenopirostris polleni at much high- er elevations than previously known (Langrand, 1990). In addition, large populations of Atelornis crossleyi and Neodrepanis hypoxantha were found; these species have been recorded recently in similar montane forests at other sites (Goodman & Putnam, 1996; Hawkins et al., 1998, in press). The presence of five bird species limited to low- land forest (a higher figure than any other eleva- tional zones), coupled with the threat of forest clearance for agriculture within this altitudinal range across Madagascar (Green & Sussman, 1990). means that in the RNI d'Andohahela, as in other eastern humid forests, the lowland sector should be given high priority for immediate con- servation activity. Acknowledgments We are grateful to the Association Nationale pour la Gestion des Aires Protegees (ANGAP) HAWKINS & GOODMAN: BIRD COMMUNITIES 185 and the Ministere des Eaux et Forets for permis- sion to conduct this study. The staff of World Wide Fund for Nature, Andohahela, helped us considerably with logistical and technical prob- lems. Literature Cited Buckland, S. T., D. R. Anderson, K. P. Burnham, and J. L. Laake. 1993. Distance sampling: Estimation of abundance of biological populations. Chapman and Hall, London, 446 pp. Goodman, S. M., M. Pidgeon, A. F. A. Hawkins, and T. S. Schulenberg. 1997. The birds of southeastern Madagascar. Fieldiana: Zoology, n.s., 87: 1-132. Goodman, S. M., and M. S. Putnam. 1996. The birds of the eastern slopes of the Reserve Naturelle Integra- le d'Andringitra, Madagascar, pp. 171-190. In Good- man, S. M., ed., A floral and faunal inventory of the eastern slopes of the Reserve Naturelle d'Andringitra, Madagascar: With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. Green, G. M., and Sussman, R. W. 1990. Deforestation history of the eastern rain forests of Madagascar from satellite images. Science, 248: 213-215. Hawkins, A. F A., A. Andrianarimisa, O. M. Rako- tonomenjanahary, and V. Raminoarisoa. In press. Inventaire des oiseaux dans la Reserve Naturelle In- tegrate de Zahamena. Ostrich. Hawkins, A. F. A., J.-M. Thiollay, and S. M. Good- man. 1998. The birds of the Reserve Speciale d'Anjanaharibe-Sud, Madagascar, pp. 93-127. In Goodman, S. M., ed., A floral and faunal inventory of the Reserve Speciale d'Anjanaharibe-Sud, Mada- gascar: With reference to elevational variation. Field- iana: Zoology, n.s. 90: 1-246. Laake, J. L., S. T Buckland, D. R. Anderson, and K. P. Burnham. 1993. DISTANCE user's guide, version 2.0. Colorado Cooperative Fish and Wildlife Research Unit, Colorado State University, Fort Collins, CO, U.S.A. Langrand, O. 1 990. Guide to the birds of Madagascar. New Haven: Yale University Press. Rahbek, C. 1995. The elevational pattern of species richness: A uniform pattern? Ecography, 18: 200-205. 186 FIELDIANA: ZOOLOGY Chapter 13 Lipotyphla (Tenrecidae and Soricidae) of the Reserve Naturelle Integrate d'Andohahela, Madagascar Steven M. Goodman,1 Paulina D. Jenkins,2 and Mark Pidgeon3 Abstract Lipotyphla (Soricidae and Tenrecidae) were studied in extreme southeastern Madagascar in two different parcels of the Reserve Naturelle Integrale d'Andohahela: on the eastern slopes of humid forest (parcel 1) in five elevational zones between 440 and 1875 m, and in the spiny bush (parcel 2) with a single site at 120 m. Thirteen species of lipotyphlans were recorded in parcel 1. including 10 species of Microgale, whereas only three species of lipotyphlans, and no Microgale, were documented in parcel 2. Although these two parcels are separated by a distance of only 20 km, no species of lipotyphlan was found to occur in both parcels. Within parcel 1 three species of Tenrecidae occurred across the complete elevational range: Microgale longicaudata, M. parvula, and M. dobsoni. The only species restricted to a single elevational zone was Oryzorictes hova, which was recorded at 1875 m. The greatest species richness of Tenrecidae was nine (including eight species of Microgale), at 1200 nr. the other elevational zones had eight species of Tenrecidae, of which six to eight were members of the genus Microgale. Densities of lipotyphlans in the 1875 m zone were exceptionally high, with up to 11.9% capture rates in the pitfall lines. Evidence was found for two distinct communities within the humid forest (parcel 1): a lowland to mid-montane (440-1200 m) fauna consisting of Microgale principula, M. thomasi, and Tenrec ecaudatus, and a montane to sclerophyllous forest fauna that included M. gracilis, M. gymnorhyncha, and Oryzorictes hova. Microgale fotsifotsy, M. cowani, and M. soricoides had elevational distributions that crossed over these two habitat communities. New information is presented here on morphology, reproduction, development, ecology, and systematics, including a partial taxonomic revision of the genus Oryzorictes. The three species of lipotyphlans recorded in parcel 2, Geogale aurita, Echinops telfairi, and Siincus madagascariensis, have distributions across much of the arid southern and western portions of the island. When all sources of information are combined. 16 species of Lipotyphla have been recorded within the reserve, and 10 of those belong to the genus Microgale. Resume Les etudes sur les Lipotyphla (Tenrecidae et Soricidae) ont ete effectuees dans l'extreme Sud-Est de Madagascar dans deux Parcelles differentes de la Reserve Naturelle Integrale d'Andohahela: sur les versants Est de la foret humide (Parcelle 1) etales sur cinq zones d'altitudes comprises entre 440 m et 1875 m. et dans une zone de broussailles epineuses 1 Field Museum of Natural History. Roosevelt Road at Lake Shore Drive. Chicago. IL 60605-2496. U.S.A. : The Natural History Museum. Cromwell Road. London SW7 5BD. United Kingdom. ; Route de St. Cergue. 1270 Trelex. Switzerland. GOODMAN ET AL.: LIPOTYPHLA 187 (Parcelle 2) comprenant un seul site a 120 m d' altitude. Treize especes d'insectivores sont inventoriees a l'interieur de la Parcelle 1, y compris 10 especes de Microgale, contre seulement trois especes d'insectivores et aucun Microgale enregistre dans la Parcelle 2. Bien que ces deux Parcelles soient a une distance de 20 km seulement l'une de 1' autre, aucune espece commune d'insectivore a ete decelee dans les deux Parcelles. A l'interieur de la Parcelle 1, trois especes de Tenrecidae apparaissaient a travers l'ensemble des zones d' altitudes differentes: Microgale longicaudata, M. parvula, et M. dobsoni. La seule espece limitee a une zone d'altitude unique est V Oryzorictes hova, inventorize a une altitude de 1875 m. La communaute la plus importante de Tenrecidae est composee de 9 especes (comprenant 8 especes de Microgale) se trouvant a une altitude de 1200 m; au niveau des autres zones d'altitudes differentes, on rencontre 8 especes de Tenrecidae dont 6 a 8 especes appartiennent au genre Microgale. Les densites d'insectivores dans la zone d'altitude 1875 m sont exceptionnellement elevees, avec une proportion de capture s'elevant jusqu'a 77.9% au niveau des pieges. Une decouverte a ete faite sur deux communautes distinctes a l'interieur de la foret humide (Parcelle 1): une faune typique des zones variant d'une altitude basse a une altitude a mi-flanc d'une montagne (de 440 m a 1200 m) composee de Microgale principula, M. thomasi, et Tenrec ecaudatus, et une faune typique des zones de montagnes a des zones de forets sclerophyles comprenant M. gracilis, M. gymnorhyncha, et Oryzorictes hova. Plusieurs especes sont reparties selon les altitudes dont le cas des habitats de ces deux populations: M. fotsifotsy, M. cowani, et M. soricoides. De nouvelles informations sont presentees sur la morphologie, la reproduction, le developpement, l'ecologie et les systematiques, comprenant une revision partielle taxonomique du genre Oryzorictes. Les trois especes d'insectivores inventoriees dans la Parcelle 2 possedent des colonies a travers la plupart des parties arides du Sud et de l'Ouest de 1'ile. Ces especes comprennent: Geogale aurita, Echinops telfairi, et Suncus madagascariensis. En considerant toutes les sources d' information, 16 especes de Lipotyphla ont ete inventoriees a l'interieur de la reserve, dont 10 appartiennent au genre Microgale. Introduction The southeastern corner of Madagascar con- tains a remarkable range of landscapes, from hu- mid forest to some of the driest areas on the is- land. Reflected in this array of biotopes are vary- ing natural habitats that are often separated by abrupt ecotones. The Reserve Naturelle Integrate (RNI) d'Andohahela contains most of these eco- logical zones. It is divided into three parcels (see Chapter 1). Parcel 1 is a large block of humid forest (63,100 ha) at the southern end of the north-south-aligned Anosyenne Mountain chain that ranges in altitude from about 300 to slightly less than 2000 m. Local rainfall ranges from 1,500 to at least 2,000 mm annually. Within parcel 1 is a wide range of forest types, including areas clas- sified as High Mountain Domain (one of the phy- togeographic zones recognized by Humbert [1955]). Parcel 1 is of particular biogeographical interest because it contains physically diverse ar- eas with correspondingly distinct flora and fauna. To the west of the Anosyennes and within the rain shadow of the mountain range is parcel 2, a large- ly intact block (12,400 ha) of semiarid spiny bush and gallery forest ranging in altitude from 1 10 to 1005 m (Nicoll & Langrand, 1989; Goodman et al., 1997). The yearly rainfall in this area is on the order of 600-700 mm, with a pronounced dry season of 5-6 months. The ecotone between par- cels 1 and 2 is dramatic and abrupt: in just a few kilometers there is a shift from humid to dry for- est. At the very southern end and on the west- facing slopes of the Anosyenne Mountain Range is parcel 3 of the RNI d'Andohahela. This parcel contains a small area (500 ha) of transitional hab- itat with elements of both dry and wet forest. The Lipotyphla (the order currently recognized as including the families Erinaceidae, Solenodon- tidae, Chrysochloridae, Tenrecidae, Soricidae, Talpidae, and their fossil relatives, all of which were formerly placed in the Order Insectivora; see Butler, 1988) is represented in Madagascar by the Tenrecidae and Soricidae. Previous work on the lipotyphlan fauna of the region includes a 1989— 1 990 study of the small mammal fauna of south- eastern Madagascar conducted by G. Ken Creigh- ton, D. Rakotondravony, and colleagues. This FIELDIANA: ZOOLOGY study was part of an environmental impact study associated with the QIT-FER mining project; nothing has yet been published on this material. Creighton et al. visited a wide variety of sites and habitats in the region. Material collected during this study is deposited in the Departement de Biologie Animale, Universite d'Antananarivo (UA) and the National Museum of Natural His- tory (USNM), Washington, D.C. O'Connor et al. (1987) and Nicoll and Langrand (1989) discuss lipotyphlans from the reserve; in most cases their records are based on direct observations or infor- mation from people living near or regularly pass- ing through the reserve. During a small mammal survey undertaken by two of us (S.M.G. and M.P.) in 1995, a transect was conducted up the eastern side of parcel 1 of the reserve and five elevational zones were sam- pled (440, 810, 1200, 1500, and 1875 m). A sixth site was sampled in the spiny bush of parcel 2. Not surprisingly, considerable differences were evident between the small mammal faunas of these two parcels, and altitudinal variation was demonstrated within the fauna of parcel 1 . The purpose of this chapter is to provide infor- mation on the distribution, ecology, and species richness of lipotyphlans of this region, document their elevational distribution within the humid for- ests of parcel 1, and summarize the remarkable level of species turnover between parcel 1 and 2. The humid forests of parcel 1 represent the south- ern limit of this habitat on Madagascar, and thus the distributional data from the site help to define the geographical ranges of the locally occurring small mammals. The Lipotyphla occurring in Madagascar in- clude four subfamilies of the family Tenrecidae (three of which are endemic) and two species of Suncus belonging to the family Soricidae. Three of the four families of Tenrecidae (Tenrecinae, Oryzorictinae, and Geogalinae) are represented in the RNI d'Andohahela, as is one soricid, Suncus madagascariensis (Coquerel, 1848). Information on external and craniodental morphology, mea- surements, variation, population structure, and re- production is given under Systematics, below. Morphological data are presented as brief key fea- tures for species that have been recorded in pre- vious reports (Jenkins et al., 1996, 1997; Good- man & Jenkins, 1998). More complete data are provided for those taxa not so recorded, namely Echinops telfairi Martin, 1838, Oryzorictes hova A. Grandidier, Geogale aurita Milne Edwards & A. Grandidier, 1872, and S. madagascariensis. Detailed descriptions of these species were pre- viously sparse or lacking. The chapter also in- cludes a partial revision of the genus Oryzorictes. Materials and Methods Trap Lines Pitfall traps with associated drift fences were the principal means of capturing lipotyphlans. In all of the elevational zones visited within humid forest (440, 810, 1200, 1500, and 1875 m), as well as the spiny bush site ( 1 20 m), three separate pit- fall lines (generally one in a valley bottom, one on a slope, and one on a ridge crest) were in- stalled. Each line was 100 m long and consisted of 1 1 buckets, 1 0 m apart, in operation for a min- imum of 6 nights. More details on the technique are given in Chapter 1 1 . A few Lipotyphla were also captured with standard Sherman live traps. The trap types, placement, baits, etc. for these lines are described in Chapter 14. 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 24-hr period (dawn to dawn) of use for one of these devices. After rains the buckets were sponged dry. The inventory of the RNI d'Andohahela was conducted between 19 October and 15 December 1995, a seasonal period when all Tenrecidae, in- cluding those that hibernate or aestivate, are ex- pected to be active (Stephenson, 1994a). Captured animals were either released or prepared as standard museum skins with associated skulls and skeletons, as fluid preserved carcasses, or as full skeletons. Voucher specimens are deposited in the Field Museum of Natural History (FMNH), Chica- go, and a representative series was returned to UA. Specimens deposited in the latter institution imme- diately after the survey have not yet been catalogued and are individually referenced by the collector's field numbers (UA-SMG or UA-MP). Measurements Cranial measurements were taken using digital calipers to within 0.1 mm or using a microscope measuring stage; external measurements were made using a standard rule to within 1 mm. The dental nomenclature follows that of Mills (1966), Swindler (1976), Butler and Greenwood (1979), GOODMAN ET AL.: LIPOTYPHLA 189 and MacPhee (1987). Dental notations are given in parentheses in the text. Premaxillary and max- illary teeth are denoted by uppercase letters and mandibular teeth by lowercase letters, as follows: incisor (I/i), canine (C/c), premolar (P/p), molar (M/m). The prefix 'd' indicates deciduous teeth; thus (dI3) refers to the deciduous third upper in- cisor. The following measurements were made of specimens in the flesh or from 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): greatest distance mea- sured across the squamosals BL (braincase length): from the superior articular facet of the occipital condyle, parallel to the long axis of the skull CIL (condyloincisive length): cranial length from anteriormost portion of first upper incisor to oc- cipital condyle Ear (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) IOB (interorbital breadth): shortest distance across the frontals between the orbital fossae LMR (maxillary toothrow length): distance of the occlusal surface from first incisor to last molar ML (mandible length): measured from the con- dylar process across the length of the mandible to the alveolus of first incisor NL (nasal length): greatest length of nasal bone 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 ter- minal hair tufts UTL (upper toothrow length): from anterior of first upper incisor to posterior of third upper molar, parallel to the long axis of the skull WT (weight): measured in grams (g) with Pesola spring scales. Animals weighing <10 g were weighed to within 0.2 g; those weighing be- tween 10 and 100 g were weighed to within 0.5 g ZB (zygomatic breadth): greatest breadth across the zygomatic processes 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 number and location of any embryos and placental scars were recorded. The mammary for- mula is presented as the number of paired axial, abdominal, or inguinal mammae. The following age classes are recognized: "In- fant" refers to individuals in which the deciduous antemolar dentition and the molars are not fully erupted; premaxillary, parietal, and basioccipital sutures are unfused. "Juvenile" includes individ- uals in which the molars are fully erupted, the deciduous antemolar dentition is erupted and in the process of replacement by the permanent teeth and cranial sutures are in the process of fusing. The eruption sequence of the permanent teeth was subdivided into four stages by MacPhee (1987); these stages have been accepted in this text unless otherwise stated. "Adult" individuals have a fully erupted permanent dentition and the cranial su- tures are generally fused, although their position is more or less clearly marked. Other Abbreviations Besides the sites designated RNI (Reserve Na- turelle Integrate), there are two other designations for protected sites in Madagascar: PN (Pare Na- tional) and RS (Reserve Speciale). BM (NH) The Natural History Museum, London (formerly British Museum [Natural History]) FMNH Field Museum of Natural History, Chicago MCZ Museum of Comparative Zoology, Harvard MNHN Museum National d'Histoire Natur- elle, Paris MP Field catalog of Mark Pidgeon SMG Field catalog of S. M. Goodman UA Departement de Biologie Animale, Universite d'Antananarivo, Antanana- rivo UMMZ University of Michigan Museum of Zoology, Ann Arbor. Systematics Family Tenrecidae Subfamily Tenrecinae Echinops telfairi Martin, 1838 Holotype— BM (NH) 1855.12.24.86: skin, skull, and skeleton collected by William Telfair. 190 FIELDIANA: ZOOLOGY Table 13-1. External measurements (mm) and weight (g) of Malagasy Lipotyphla recorded in the RNI d'Andohahela, excluding Microgale. Species HB* TL HF Ear WT Onzorictes hova 107.14 ± 4.30 52.86 ± 1.86 17.71 ± 0.95 12.00 ± 0.82 34.00 ± 3.71 101-112 (7) 51-55 (7) 16-19 (7) 11-13 (7) 28-40 (7) Echinops telfairi — 13 20 25 — Suncus madagascariensis 44.71 ± 3.77 33.00 ± 1.73 7.86 ± 0.69 6.57 ± 0.53 2.46 ± 0.53 39-51 (7) 31-35 (7) 7-9 (7) 6-7(7) 1.8-3.2 Geogale aurita 67.80 ± 5.00 38.67 ± 2.42 11.67 ± 0.52 17.00 ± 1.55 7.03 ± 1.34 61-76 (6) 34-41 (6) 11-12 (6) 15-18 (6) 5.5-8.5 (6) The mean, standard deviation, and range are given, with sample size in parentheses. * HB = head and body length: TL = total length; HF = hind foot length; Ear = ear length; and WT = weight. See text (p. 190) for further explanation of abbreviations. Type Locality — Madagascar(?). Referred Material — FMNH 156489: 7.5 km ENE of Hazofotsy, 24°49'S, 46°36'E, 120 m. Key Features — Dorsum covered with spines; external tail absent. Dorsal profile of skull only slightly curved; rostrum short, deep, and broad; interorbital region elongated, frontals not inflated; braincase deeper than interorbital region; brain- case short, lambdoid crest well developed. Dental formula 2/2 1/1 3/3 2/2 = 32; II taller than C; no diastemata. Measurements — External measurements are presented in Table 13-1. Remarks — This specimen was collected in slightly disturbed spiny forest. Echinops is an an- imal of dry habitat, and the record from parcel 2 of the RNI d'Andohahela is apparently the east- ernmost occurrence of this species in southern Madagascar. On the basis of discussions with for- est guards and local people living near the re- serve, this species has been reported to occur in parcel 1; we strongly suspect that records of hedgehog-like spiny tenrecids from this area are of Setifer and not Echinops. Echinops telfairi is known to live in a variety of ecotypes from spiny bush to dry deciduous forest, and it seems some- what resilient to selective logging (Nicoll & Lan- grand, 1989; Goodman & Ganzhorn, 1994; Gan- zhorn et al„ 1996). Setifer setosus (Schreber, 1777) Holotype — Unknown. Type Locality — Unknown. Referred Material— FMNH 156490: 8 km NW of Eminiminy, 24°37'S, 46°45'E, 440 m. Key Features — Dorsum covered with spines; very short spinous tail present. Skull moderately GOODMAN ET AL.: LIPOTYPHLA robust, dorsal profile curved in lateral view; ros- trum deep and broad; interorbital region broad and elongated, frontals posterodorsally inflated; interorbital region deeper than braincase; brain- case short, lambdoid crest present. Dental formula 2/2 1/1 3/3 3/3 = 36; II well developed, slightly shorter than C; short diastemata on either side of C and first lower premolar (p2). Remarks — The single specimen was found in undisturbed lowland forest with relatively open understory. It has been previously reported to oc- cur in parcel 1 and inferred to occur in parcel 3 on the basis of published maps of its distribution (O'Connor et al., 1987). This species is broadly distributed across the eastern humid forests, gen- erally at lower-lying elevations, from the RNI d'Andohahela north to the PN de la Montagne d'Ambre in the Antsiranana region (Raxworthy & Nussbaum, 1994; Goodman et al., 1996a). It is also known from a variety of sites in western de- ciduous forest (Ganzhorn et al., 1996) and spiny forest (Nicoll & Langrand, 1989). Tenrec ecaudatus (Schreber, 1777) Holotype — Unknown. Type Locality — Unknown. Referred Material— FMNH 156491: 12.5 km NW of Eminiminy, 24°35'S, 46°44'E, 810 m; FMNH 156492, UA-SMG 7547: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m. Key Features — The largest of the Tenrecinae. Dorsal pelage of coarse bristly hair intermixed with spines; short tail present. Skull elongated; rostrum with deep sockets in ventrolateral region of the premaxillae, which accommodate the lower canines when the jaw is closed; interorbital region narrow, elongated; braincase short, narrow, an- 191 gular, pronounced sagittal and lambdoid crests form deep posterodorsal flanges. Dental formula 2/3 1/1 3/3 3/3 = 38; upper and lower canines very long, robust, and prominent; pronounced di- astemata on either side of C and posterior to c; short diastemata posterior to first upper and lower premolars. Rkmarks — Specimens were collected in undis- turbed lowland and primary montane forest. Ten- rec ecaudatus has been previously reported in parcel 1 of the reserve and thought to occur in parcel 2 on the basis of published distributional maps (O'Connor et al., 1987). This species is broadly distributed across Madagascar and occurs in a wide variety of habitats, including humid and deciduous forests as well as spiny bush. It seems to thrive in disturbed areas, even in the presence of human hunting pressure and forest exploitation (Ganzhorn et al., 1990, 1996; Raharivololona, 1996). Subfamily Geogalinae Geogale aurita Milne Edwards & A. Grandidier, 1872 Holotype — MNHN no. 267: unsexed specimen in alcohol, skull extracted, collection date un- known (Rode, 1942). Type Locality — "Morondova (= Morondava) et Tullear" (= Toliara). Referred Material— FMNH 156350-156352, 156551-156553: 7.5 km ENE of Hazofotsy, 24°49'S, 46°36'E, 120 m. Key Features — Small, shrew-like, similar in overall appearance to Microgale but pinnae large and prominent, tail covered with fine hairs, HF short relative to HB (see Table 13-1 for dimen- sions). Pelage soft, very short, and not dense. Dorsal fur coloration varies from light gray to a light reddish brown, ventrum buffy white. Crani- um elongated, dorsal profile straight, inclining gradually from rostrum to occiput; rostrum nar- row; zygomatic plate broad, with prominent an- terior margin; interobital region narrow and elon- gated; braincase short, angular, lambdoid crest pronounced. Dental formula 2/2 1/1 3/2 3/3 = 34; II large, slightly proodont, distostyle well devel- oped, small diastema present between first and second upper incisors; second incisor and C small, similar in height to distostyle of II; P2 and P3 very small, less than distostyle of C in height; talons of P4, M 1 , and M2 anteroposteriorly com- 192 pressed; M3 highly compressed anteroposteriorly, buccolingually elongated; il moderately large, slightly procumbent, second lower incisor slightly shorter than posterior cuspid of il; c reduced in size; first lower premolar very small, second low- er premolar with well-developed protoconid; tal- onid of m3 reduced to a single cuspid. Measurements — External measurements are presented in Table 13-1. Population Structure and Reproduction — The testes of one male with convoluted epididy- mides (FMNH 1 5635 1 ) measured 8X5 mm. One pregnant female (FMNH 156552) contained two embryos in the left and three in the right oviduct that measured 3 mm in crown to rump length. Litters of up to five individuals have been report- ed in this species (Stephenson, 1993). Mammary formula: 1-1-2 (n = 1), 2-0-2 (n = 1). Remarks — All specimens were collected in slightly disturbed spiny forest. This species is known to occur in a variety of dry forest types in southern and western Madagascar. The occurrence of Geogale in parcel 2 of the RNI d'Andohahela represents the eastern known limit of this species in southeastern Madagascar. It has been recorded in spiny bush, deciduous forest, and transitional habitats between these two forest types (Nicoll & Langrand, 1989; Goodman & Ganzhorn, 1994; Ganzhorn et al., 1996; Goodman & Rasoloarison, 1997). Subfamily Oryzorictinae Oryzorictes hova A. Grandidier, 1870 Oryzorictes talpoides G. Grandidier and Petit, 1930 Holotype— MNHN type no. 262, CG 1887- 874: adult female in alcohol, skull extracted, col- lection date unknown. Type Locality — "Ankaye (= Ankay, along the Mangoro River near Lac Alaotra) et Antsianak" ( = Antsianaka, region to the east of Lac Alaotra; Viette, 1991). Referred Material— FMNH 156485-156488, 156601-156603: 20 km SE of Andranandambo, 24°33'S, 46°43'E, 1875 m. Key Features — Pelage soft, slightly iridescent; TL approximately half HB (TL:HB mean 49.51 ± 2.70, range 46.0-53.9, n = 5); forefeet with very enlarged claws; broad naked rhinarium; eyes very small; ears small, concealed in pelage. Skull moderately robust, premaxillae dorsolateral^ FIELDIANA: ZOOLOGY flared, braincase short, broad and deep, lambdoid crests well marked; 13 very small, approximately as tall as distostyle of 12; diastema present be- tween 13 and C, forming a sulcus to accommodate the distal tip of c; upper and lower canines mark- edly taller than all other teeth; distostyle of C very small. Measurements — External measurements are i presented in Table 13-1. Population Structure and Reproduction — The sex ratio of females to males was 1:2. No Juveniles were present in the sample collected. | The testes of one male (FMNH 156601 ) with con- voluted epididymides measured 6x5 mm. One female (FMNH 156602) was pregnant, carrying one embryo in the left and two in the right oviduct that measured 5 mm in crown to rump length. Another female (FMNH 156603) with permanent I dentition was imperforate. Remarks — All specimens collected during the 1 1995 survey were taken in transitional primary I forest between upper montane and sclerophyllous ii forest, either in valley bottoms or on ridge crests. The only exception is the remains of an Oryzo- 1 rides hova that were recovered from the digestive I system of a Pseudoxyrhopus snake captured at the 1200 m site. Other records from southeastern Madagascar include one animal taken at about 50 ijm at Bezavona near the base of Pic St. Louis, 1 km west of Tolagnaro (USNM 577052) and two at 750 m in the Marosohy Forest, along the north- ieastern border of parcel 1 of the RNI • d'Andohahela (USNM 578789, 578913). This species has a broad distribution across the humid portions of the island and is known to occur in lowland marsh and rice paddy habitats up to high- elevation montane and sclerophyllous forest. Taxonomic Comments — In 1870 A. Grandidier described a new genus and species of lipotyphlan from the northern portion of eastern Madagascar as Oryzorictes hova. He characterized the genus as having a long rhinarium terminating in small nostrils; very small eyes; medium-sized round ears; forefeet with four digits, three with large and strongly recurved claws; hind feet with five digits; and a dental formula of 3/3 1/1 6/6 (premolars and molars combined). As pointed out by G. Grandidier and Petit (1930). the holotype of Oryzorictes hova possess- es a highly reduced fifth digit on the forelimbs, not just four digits as stated by A. Grandidier (1870). This mistake was reiterated by several au- thors (e.g., Dobson, 1882). Subsequent to the pub- lication of the O. hova description, a second spe- cies in this genus, O. talpoides, with five digits on the forefoot, was named by G. Grandidier and Petit (1930). The type locality of O. talpoides is near Marovoay in northwestern Madagascar, not far from Mahajanga. A third member of this ge- nus, O. tetradactylus, was named by Milne Ed- wards and A. Grandidier in 1882; this species has four digits on the forelimbs. A further discrepancy between the characters noted in the description of Oryzorictes hova and the actual type specimen has to do with the dental formula. Although A. Grandidier (1870) clearly stated that the type specimen had an 13, there is no evidence of this tooth, its root, or any erupting structure. This difference in incisor count was not noted by G. Grandidier and Petit (1930) or Dob- son ( 1 882) in their descriptions of O. talpoides, but it was used as one of the distinguishing char- acters by Genest and Petter (1975) to differentiate O. hova and O. talpoides. The latter authors noted that O. hova lacked an 13. On the basis of tooth eruption and suture ossification, the holotype of O. hova represents an adult individual. After examination and measurement of the ho- lotypes of Oryzorictes hova and O. talpoides (MNHN 264) and nearly 30 specimens of five- toed Oryzorictes collected from a variety of lo- calities on Madagascar, no individual other than the holotype of O. hova has been found lacking the 13. Furthermore, cranial measurements of the holotypes of O. hova, O. talpoides, and five-toed Oryzorictes show broad overlap, and these species cannot be differentiated based on these characters (Table 13-2). Another character that has been proposed to differentiate Oryzorictes hova and O. talpoides is the size of the pollex (Grandidier & Petit, 1930; Genest & Petter, 1975). This character is difficult to measure in dried specimens, and it is impos- sible to compare such measurements with speci- mens preserved in fluid. On the basis of the col- lection from Andohahela and material housed in BM(NH) that was collected from a variety of lo- calities in the eastern humid forest, however, there is considerable variation in this character at the population level as well as the regional level. Therefore we conclude that it is of no use in dif- ferentiating among types of five-toed Oryzorictes. Each of the specimens we have examined in museum collections that was previously deter- mined as Oryzorictes hova possesses an 13 (e.g., FMNH 5641; BMNH 94.290. 97.9.1.77) and on the basis of the Genest and Petter (1975) key is referable to O. talpoides. Furthermore, there are GOODMAN ET AL.: LIPOTYPHLA 193 SO (N +1 r*: <~1 o __ o m (n rt en +1 T II .SsJ Si so to (N 00 rl rn — ON „ r- oo 00 C4 in on © wi CN >/S +1 I/I — -H __ _ o ^ +1 ^ II : so H -41 I II — — i rN s Si so J1 ^1 — o 2 c_ a U p-pi 00 O^oo o w"~' o w 00 u a. 3 - > s. — 53 X> U XI o .2 C2 02 E 2 * £ w II 03 m £ O ob I— I c . .. o X) c Si oo cd N ^ 00 u c "~ t/J 3 °Z 4) U o- c5 oo few published records of O. hova. Stephenson ( 1 994b) reported capturing individuals of O. hova in the RS d'Analamazaotra (18°28'S, 48°28'E) and in the forest of Anandrivola (15°46'S, 49°36'E), although how the animals were identi- fied was not mentioned explicitly. He further not- ed that the specimen obtained at Anandrivola was close to a site from which O. talpoides had been reported (Heim de Balsac, 1972) and that the two species might be locally sympatric. On the basis of material available to us we con- clude that the various characters proposed to al- low one to distinguish between Oryzorictes hova and O. talpoides show broad overlap and are not useful to differentiate these two taxa. It is odd that the only known specimen of an adult five-toed Oryzorictes lacking the 13 is the holotype of O. hova, but such dental anomalies have been ob- served in other members of the Tenrecidae (Dob- son, 1882; Leche, 1907; Heim de Balsac, 1972). On the basis of current evidence, we conclude that O. talpoides and O. hova represent the same spe- cies and that the former name is a junior synonym of the latter. 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- sileo (Ankafana = Ankafina, Fianarantsoa, Fian- arantsoa Province, 21°12'S, 47°12'E; see Mac- Phee, 1987, Carleton & Schmidt, 1990). Referred Material— FMNH 156355, 156554, 156555, UA-SMG 7517, 7558: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m; FMNH 156356-156372, 156431, 156556-156563; UA- MP 29, 30, UA-SMG 7647: 15 km NW of Emi- niminy, 24°34'S, 46°43'E, 1500 m; FMNH 156373-15410, 156564, 156565, UA-SMG 7664, 7665, 7673, 7674, 7685: 20 km SE of Andran- andambo, 24°33'S, 46°43'E, 1875 m. Key Features — Medium sized (see Table 13- 1); TL moderately short, shorter or subequal to HB. Pelage dark brown dorsally, gray ventrally with a reddish brown wash; tail bicolored, dark brown dorsally, sharply demarcated from paler reddish buff underside. Skull medium in size, ros- trum elongated; pronounced diastemata separate teeth of upper anterior dentition from II to P3, also on either side of c and p2. All elements of 194 FIELDIANA: ZOOLOGY A talonid of m3 present, including hypoconid, en- toconid ridge, talonid basin, and entoconid. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of females to males was 1:1.3. The ratio of juveniles to adults was 1:1.2. The upper and lower last two molars (respectively M2, M3, m2, and m3) were still in the process of erupting in one juvenile specimen, and M3 and m3 were still erupting in three other specimens. The other juveniles showed a complete deciduous dentition, with all molars erupted; the exception was a sin- gle specimen in which the deciduous upper and lower third incisors (dI3 and di3) had been shed and their permanent replacements (13 and i3) were erupting. One adult male with convoluted epidid- ymides had testes measuring 8X7 mm. One adult female was pregnant with two embryos in the left and one in the right oviduct. Mammary formula: 0-1-2 (n = 4), 0-2-1 (n = 1), 1-0-2 (n = 1). Remarks — This species as currently defined {sensu Jenkins et al., 1996) has a broad distribu- tion across the eastern humid forest from the RNI d'Andohahela north to at least the RS d'Anjanaharibe-Sud and occurs from lowland for- est to summital zones above the tree line at 2450 m (Langrand & Goodman. 1997; Goodman & Jenkins, 1998). Microgale dobsoni Thomas, 1884 Nesogale dobsoni Thomas, 1918 Holotype— BM (NH) 84.10.20.1: immature male, in alcohol, skull extracted. Collected Febru- ary or March 1884 by W. Waters. Type Lwality — Nandesen Forest. Central Bet- sileo (Nandihizana, 10 miles S. of Ambusitra — i manuscript note in Thomas's private copy of orig- inal description. Nandihizana, ca. 20 miles [30 km] SSW of Ambositra; see MacPhee, 1987. Es- timated as ca. 20°50'S, 47°10'E). Referred Material— FMNH 156468. 156598: 8 km NW of Eminiminy, 24°37'S, 46°45'E, 440 m; FMNH 156413: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m; FMNH 156414- 156417. 156469, 156470, 156567: 15 km NW of Eminiminy, 24°34'S, 46°43'E, 1500 m; FMNH 156418-156421, 156471, 156566, UA-SMG 7666, 7688: 20 km SE of Andranandambo, 24°33'S, 46°43'E. 1875 m. Key Features — Large, TL subequal to or lon- ger than HB. Dorsal pelage brown, venter gray with buff wash. Skull large and robust, sutures fused and obscure; rostrum moderately broad, in- terorbital region long; braincase angular, superior articular facets very prominent, lambdoid crest well developed; occipital region reduced in size. Diastemata between 1 1 and 12 and between 13 and C; II larger than 12, i2 considerably larger than c; talonid of m3 reduced, hypoconid low, hypo- conulid prominent, entoconid ridge and talonid basin poorly defined, entoconid absent. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of males to females was 1:7. No juveniles were collected. Two of the females were lactating. Mammary formula: 0-2-1 (n = 1), 1-0- 2 (n = 1), 1-1-2 (n = 3), 1-2-1 (n = 1). Remarks — The species was trapped along the complete elevational transect between 440 and 1875 m, although no individual was captured in the 810 m zone. Most specimens were trapped on the ground, but one was obtained 2.5 m above ground on a 3-cm-diameter horizontal branch. Microgale dobsoni is broadly distributed in the eastern humid forest and is known from numerous sites in the geographical zone between the RNI d'Andohahela north to at least the RS d'Anjanaharibe-Sud (MacPhee, 1987; Jenkins et al., 1996; Goodman & Jenkins, 1998). It is a widespread species that appears to be tolerant of habitat disturbance (Goodman et al., 1996b). Microgale fotsifotsy Jenkins et al., 1997 Holotype— UMMZ 168468: adult male in al- cohol, skull extracted, collected 13 January 1992 by Christopher J. Rax worthy. Type Locality — Camp 2, Antomboka River, Fitsahana. Pare National de la Montagne d'Ambre, Antsiranana Fivondronana, Antsiranana Province, 12°29'S, 49°10'E, 650 m. Referred Material— FMNH 156568. 156569: 8 km NW of Eminiminy, 24°37'S, 46°45'E, 440 m; FMNH 156424: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m; FMNH 156570: 15 km NW of Eminiminy, 24°34'S, 46°43'E, 1500 m. Key Features — Dorsal pelage pale grayish brown, soft in texture, venter with lighter buffy wash; digits of fore- and hind feet and extreme tail tip contrastingly lighter colored; pinnae prom- inent and conspicuous; fifth digit of hind foot elongated, scarcely shorter than second. Skull with maxillary process of zygoma at right angles GOODMAN ET AL.: LIPOTYPHLA 195 CM in (N £ CM £ — , *r> co w o w NO W r- w NO W On u-) NL^ r- w CM ^ CM "* "X IT) CM co CM 00 CM 00 l-H CO CM O f s CM w CM CM - o no d^ 5^ •-4 d*> d^ d — • d d d d d d d cm — d t d rf r^ On On On H +i 7 + l i + 1 1 o + 1 1 + 1 1 +l i +1 l + 1 I +1 l +1 1 NO +1 1 +1 1 -J On m r- rt _ o co CO in oo "■ : co r- 8* S^n on ^od &S 5^2 $2 32 2d t-; ^ ~ ££ r~ r-' ON On' od d d d d "" '-' en CO w m50 oo 50 m C2- ^H ""> ON C- CO "^ t ^T on 50 >/-i c r- w VO W CO w CM CO W co — in CM ■<*• CM Tt — Tfr CM — i d00- ©^ ©^ ^ d^. d^ d d d d ©^ ©^ d — r-< d d d — : NO NO ON 00 00 On On + 1 1 + 1 1 + 1 1 o +1 1 +l^ +i 7 + 1 I + 1 1 + 1 1 +l3 00 + 1 i +1 1 Tt CO oo t-~ ON t-- , ON ON d , o in 3 NO 3*; 3°o od ^^ 8^ ^2 £* ^loo On S°°' q° ^2 oo 2 NO no ON 00 00 d ON ON — c d d in 1/0 UN ~ ^4 CO CM w ^ "" On w + 1 I O £nO in ,—. oo — ^^ NO W r- co o oo do ^d CM CM +11+11 P- NO £oo £^ m 50, ^ C- oo © i ""! + 1 7 + r- in " ■>*' CO *-< Tfr NO © CO © CM CM CM d°° ~- r- I +1 I ^ i^ NO NO +11+11 ON CO f- — <"- _; *>^ ^CM CM — CM CM +i 7 +i 7 CM On in -h oo CO 00 — 00 © cm o — ; +1 I +1 _ o (NNO ^ o -* ~ CM w + l i n ^ 00 © nC in r- in in co X s — ' i — i s — ' f— 1 N — , in in NO CM 00 in NO +1 1 © + 1 CM + 1 1 in o in OO m r- no 3 © Tf NO a r^ m in C «n NO 3 0, a ~ 5S BC c 3 «a, 3 00 ^^ ON ^ — I no ro m NO NO +11+11 CM CM in no oo uo Tf o 00 — CM — — - ^ no n^ CO ON CM CM +11+1 I no in O NO © NO CO © Tt ^+ oo r^ +11+11 o m no r^ © no + 1 l ON in r^ no in © NO + 1 I . o 80 C/S in CO Tf d od +l7 CM in c> . CM ^r oo ON CO NO NO ^~- ^^ co in i- ' ^-^ CM CM CM ON ~ ^f CM CM + 1 1 © +i 7 + 1 1 ^t in in CO — i CO 00 r^ r- • ^+ 00 — in CO NO in ^^ ^^ r- n^ P W d oo ^- d co — ' OO +1 l o +1 l +1 1 r^ m in —I NO o — CO ^H (^ ™" o CO — CM NO ■^f in CM r-H O -H CM r- •^t in O CO ^t 00 cm in 0O NO ~-> ^-^ © ^^ r- ^ 00 n— ' 00 w o -^ NO w ^ s in ^ ON w — H 00 -^ 00 © r- © 00 d on ^ © d on i— i — < r- d © CM CM + 1 1 +1 1 +1 l +1 1 + 1 l + 1 1 +1 1 o + 1 1 + 1 1 in rfr in m r^ 00 t- ^r 00 Tt- r-i r- -h © I— CM — © ^H 00 — r^ Tt „ r- — NO —" CM CO od o 00 ' ' in —; ^^ NO NO NO ^^ ON r^ — i ON ^^ — ^ OO in ^h ^+ o in CO 00 On in r- in CO CO -* ^ NO ^ NO r- CO ^ •* ^ CM ^ ■* -^ oo m CO NO in ^^ P w m ^ ^ © — i CO NO in ^^ CO ^ P^ NO _ CO _< ^ t-. -^ CM ^ cm m CM NO — ^ CO »-< m — n in CM ON ■* — NO NO od NO CO O CM — ' CM 00 NO NO oo NO t~- t~- 00 + 1 1 + 1 J_ + 1 1 © © + 1 1 + 1 1 + 1 1 + 1 1 + 1 1 + 1 1 +1 1 o + 1 1 +1 1 00 o 00 ON o NO r^ rf in CO o in "* NO t-- 00 00 m cm ON O C: ^o © in P "* © ^r 00 ON © oo \o r- oo CM CM ^p r- On —' CO — c p ON in *"" CO "" CO ON r^ ^^ NO CO Tt in CO m m NO NO NO NO 00 i— < CO NO NO — i— < ON 00 ^r in — ^o NO ^ 00 ^ ^r »-< CM NO ON ON +1 l + 1 1 On 00 o r- O 00 o in -— d So cin © en a _ ■* _ © r~ d vd n r> +11+11 Ci — -2 ©^ OS w sq w — >T) ^^ 00 d © © OV n + 1 1 +1 1 r~ r~ m — © — © ^ 00 — m Tf oo w r» w rt O — o oo r-» +11 +11 rn SO r- so w> so ro sO -rt Os OS 00 +1 I +1 I v© O r- oo in oo Os >o n — d n (N + 1 1 (N m (N m en n Os" »r> •>* O (N + 1 1 (N r- O sq 00 © s r- VO © Os en + 1 1 in (N Os r~ sd o t« 3 — e« ■5 ^ i» o N • — '35 w c — .2 Q. Q. £ X S3 « [/I |_ is .as eg *-* U "O _' -a * 3 2 C3 flj e^ f- C/3 to long axis of cranium; braincase broad and short. Third upper and lower incisors small, 13 slightly greater in crown height than distostyle of 12, i3 subequal in height to posterior accessory cusp of i2; i2 greater in breadth than il; C and c respectively greater in crown height than P3 and p3- Measurements — External and cranial measure- ments are presented in Table 13-3. Variation — The small samples available sug- gest that the population from RNI d'Andohahela averages smaller than those from further north in Madagascar. Population Structure and Reproduction — The sex ratio of females to males was 1:2. The ratio of juveniles to adults was 1:3. One male (FMNH 156569) with permanent dentition was reproductively immature, with nonconvoluted ep- ididymides. The adult female had one placental scar in the right oviduct and two in the left. Mam- mary formula: 0-1-2 (n = 1). Remarks — As with so many of the recent find- ings concerning new taxa of Microgale and in- formation on their geographical range, M. fotsi- fotsy has a broad distribution and is known from sites running the complete length of the eastern humid forest, including the RNI d'Andohahela, RNI d'Andringitra, PN de Ranomafana, RS de Zahamena, RS d'Ambatovaky, RNI de Marojejy, RS d'Anjanaharibe-Sud, and PN de la Montagne d'Ambre (Jenkins et al., 1997; Goodman & Jen- kins, 1998). Two individuals (USNM 578787, 578887) have also been recorded (Jenkins et al., 1997) from Marosohy Forest, along the north- eastern boundary of parcel 1 of the RNI d'Andohahela, between 650 and 700 m. In gen- eral this species is captured in low numbers and across an elevational range from lowland to lower montane forest. For example, in the RNI d'Andohahela up to two individuals were trapped within each elevational zone studied between 440 and 1200 m; in the RNI d'Andringitra two indi- viduals were captured in the 1210 m zone; and in the PN de la Montagne d'Ambre M. fotsifotsy has been taken between 650 and 1 1 50 m (Jenkins et al., 1997; Goodman et al., 1996a). Microgale gracilis (Major, 1896) Oryzoryctes [sic] gracilis Major, 1896 Leptogale gracilis Thomas, 1918 Holotype— BM(NH) 97.9.1.78: adult of un- determined sex, skin and skull. Collected Novem- ber 1894 by C. I. Forsyth Major. GOODMAN ET AL.: LIPOTYPHLA 197 Type Locality — Ambohimitombo Forest (Am- bohimitombo town, 43 km [by road] SE of Am- bositra, 10 km into eastern forest; Fianarantsoa, Fianarantsoa; 20°43'S, 47°26'E; see MacPhee, 1987. MacPhee 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 lo- cality by Major [1897] is 1400-1500 m). Referred Material — FMNH 156573: 15 km NW of Eminiminy, 24°34'S, 46°43'E, 1500 m; FMNH 156422, 156423, 156425, 156426, 156571: 20 km SE of Andranandambo, 24°33'S, 46°43'E, 1875 m. Key Features — Size large, TL shorter than HB. Pelage dark brown dorsally with buff speck- ling. Muzzle very long; large, naked rhinarium anteriorly reticulated, striae on posterior region in- complete. Eyes very small; ears small, partially concealed by pelage. Forefeet broad, claws en- larged. Skull very elongated and gracile; rostrum slender, markedly attenuated; braincase rounded, moderately broad and long. Dentition reduced; upper incisors subequal in height, incisors and ca- nine very slender; extensive diastemata between all anterior teeth, particularly P2 and P3; talons on molars very reduced, resembling cingula; tal- onid of m3 slightly reduced, entoconid lacking. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of females to males was 1:5 in the small sample available. Only adults were collect- ed. Testes of two males with convoluted epidid- ymides (FMNH 156571 and 156573) measured respectively 10 X 5 mm and 12X7 mm. Remarks — MacPhee (1987) noted that this spe- cies was very rare in collections. In more recent years it has been recorded at a range of sites in the eastern humid forest (arranged from south to north): RNI d'Andohahela; RNI d'Andringitra (Jenkins et al., 1996); PN de Ranomafana (USNM 449179; Nicoll & Rathbun, 1990); and Foret de Nosiarivo, Ankaratra Massif (Goodman et al., 1996b). The distribution of Microgale gracilis is apparently confined to the southern half of the eastern humid forest. by Steven M. Goodman and Christopher J. Rax- worthy. Type Locality — Thirty-eight km S of Ambala- vao. Reserve Naturelle Integrate d'Andringitra, on ridge E of Volotsangana River, Fianarantsoa Prov- ince, 22°H'39"S, 46°58'16"E, 1625 m. Referred Material— FMNH 156427, 156428, 156572, 156573: 15 km NW of Eminiminy, 24°34'S, 46°43'E, 1500 m; FMNH 156429, 156574: 20 km SE of Andranandambo, 24°33'S, 46°43'E, 1875 m. Key Features — Large, TL shorter than HB. Dorsal pelage dark brown, dark gray-brown ven- trum. Muzzle very long, forming a proboscis; rhi- narium very large with transversely striated naked region. Eyes very small. Ears small, virtually con- cealed in pelage. Forefeet broad, claws enlarged. Skull long, moderately gracile; rostrum slender and elongated; braincase short and broad. Denti- tion moderately reduced with long diastemata be- tween all anterior teeth from II to P3 and i2 to p3; talonid of m3 slightly reduced; talonid basin, hypoconid, hypoconulid, and entoconid ridge present, entoconid indicated. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of females to males was 1:4 in the small sample available. Only adults were collect- ed, but one of these (FMNH 156574, a male with permanent dentition) was not in reproductive con- dition and possessed nonconvoluted epididy- mides. The testes of another male (FMNH 156572) with convoluted epididymides measured 6X4 mm. Mammary formula: 1-1-2 (n = 1). Remarks — This recently described species is known from a variety of localities in the eastern humid forest (south to north): RNI d'Andohahela, RNI d'Andringitra (Jenkins et al., 1996), PN de Ranomafana (UA), near Fanovana (Jenkins et al., 1996), Foret d'Andranomay, Anjozorobe (Good- man et al., 1998), and the RS d'Anjanaharibe-Sud (Goodman & Jenkins, 1998). It occurs in montane to sclerophyllous forest and is unknown from lowland humid forest. Microgale gymnorhyncha Jenkins et al., 1996 Microgale gracilis (Major): MacPhee, 1987, in part Microgale longicaudata Thomas, 1882 Microgale majori Thomas, 1918: MacPhee, 1987 Holotype— FMNH 151807: adult female in al- cohol, skull extracted, collected 13 December 1993 Holotype— BM(NH) 82.3.1.15: adult female, body in alcohol, skull extracted, collected mid- 198 FIELDIANA: ZOOLOGY February 1879 to mid-March 1880 by the Rev- erend 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 156576: 8 km NW of Eminiminy, 24°37'S, 46°45'E, 440 m; FMNH 156449, 156450, 156579-156581: 12.5 km NW of Eminiminy, 24°35'S, 46°44'E, 810 m; FMNH 156583, UA-SMG 7542, 7548: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m; FMNH 156430-156433, 156455, 156584: 15 km NW of Eminiminy, 24°34'S, 46°43'E, 1500 m; FMNH 156434-156448, 156451, 156452, 156578, 156593, UA-SMG 7680, 7720-7724, UA-MP 36: 20 km SE of Andranandambo, 24°33'S, 46°43'E, 1875 m. Key Features — Small in size, TL very long, more than twice as long as HB; distal portion of tail naked and transversely wrinkled on dorsal surface; fifth hind digit elongated, subequal in length to second digit. Dorsal pelage reddish brown, venter gray with bright reddish buff or buff wash. Skull small, rostrum moderately short; braincase moderately narrow and long. Diaste- mata present between II and 12, on either side of C and P2; well-developed anterior and posterior accessory cusps present on 12, C, and P2; C sub- equal to or taller than II ; P4 scarcely greater in crown height than P3; p2 caniniform; talonid of m3 with low hypoconid, hypoconulid well devel- oped, narrow talonid basin, reduced entoconid ridge, and entoconid absent. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of males to females was 1:1.3. The ratio of adults to juveniles was 1:3.6. The third molars were not fully erupted in two juvenile specimens. All juveniles had completely decidu- ous anterior dentitions, except for two in which the di3s were in the process of replacement by the permanent teeth. One male with a deciduous den- tition (FMNH 156576) was reproductively mature with convoluted epididymides and testes measur- ing 4X3 mm. Mammary formula: 1-0-2 (n = 1), 1-1-2 (n = 1), 1-2-2 (n = 1). Remarks — This species is broadly distributed in the eastern humid forests from the RNI d'Andohahela north to the PN de la Montagne d'Ambre (MacPhee, 1987; Jenkins, 1993; Rax- worthy & Nussbaum, 1994; Jenkins et al., 1996). An individual tentatively assigned to the Micro- gale longicaudata species group has been report- ed from the dry deciduous forest near Morondava (Ade, 1996). Throughout much of its range, this species is sympatric with another long-tailed species, Mi- crogale principula. These two species have re- markable modification of the tail for prehensile activity, and they may occupy some aspects of the same arboreal niche (Eisenberg & Gould, 1970; Goodman & Jenkins, 1998). In the RS d'Anjanaharibe-Sud these two species replace one another along an elevational gradient, with M. principula occurring in lower-lying forest and M. longicaudata on the upper slopes (Goodman & Jenkins, 1998). Although there is a tendency for the same pattern to occur in the humid forest of the RNI d'Andohahela, the two species over- lap broadly in elevation between 440 and 1200 m. On the basis of capture results, M. principula was more common in the 440 m zone, whereas M. longicaudata was more common at higher el- evations (Tables 13-4 and 13-5). Both species have also been recorded from PN de Mantady (Jenkins, 1993). Microgale parvula G. Grandidier, 1934 Microgale 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. Probably Montagne d'Ambre; see Jenkins et al., 1997]. Referred Material— FMNH 156456, 156585, 156586: 8 km NW of Eminiminy, 24°37'S, 46°45'E, 440 m; FMNH 156457, 156458: 12.5 km NW of Eminiminy, 24°35'S, 46°44'E, 810 m; FMNH 156587, 156588, UA-SMG 7543: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m; FMNH 156459, 156589: 15 km NW of Emini- miny, 24°34'S, 46°43'E, 1500 m; FMNH 156590: 20 km SE of Andranandambo, 24°33'S, 46°43'E, 1875 m. Key Features — Very small, TL slightly shorter than HB. Dorsal pelage dark brown, ventral pel- age dark gray-brown, tail uniform dark gray. Skull very small and delicate, rostrum slender, braincase shallow and long, occipital condyles posterodor- sally orientated. Diastemata between II and 12, on either side of C and P2, and between c and p2; anterior and posterior accessory cusps present on GOODMAN ET AL.: LIPOTYPHLA 199 Table 13-4. Capture results for small mammals trapped in pitfall lines in the humid forest of parcel 1 (440, 810, 1200, 1500, and 1875 m) and the spiny bush of parcel 2 (120 m) of the RNI d'Andohahela. Line Character 1 2 3 4 5 6 Descriptive information Lowland Transitional lowland/montane Forest type 440 m 810 m Altitude (m) 420 430 440 800 830 840 Line placement! V S R V S R First sample day (day/month) 20/10 20/10 20/10 29/10 29/10 29/10 Last sample day (day/month) 26/10 26/10 26/10 5/11 5/11 5/11 Total pitfall nights 77 77 77 88 88 88 Species sampled MAMMALIA Insectivora Tenrec ecaudatus — — — — — — Microgale cowani — — — — — — M. dobsoni — 1 — — — — M. fotsifotsy — 2 — — — — M. gracilis — — — — — — M. gymtwrhyncha — — — — — — M. longicaudata — 1 — 2 2 1 M. parvula 1 2 — — 2 — M. principula 3 1 1 — — 2 M. soricoides — — — 1 — — M. thomasi 3 2 — 1 1 2 Oryzorictes hova — — — — — — Geogale aurita — — — — — — Suncus madgascariensis — — — — — — Rodentia Eliurus spp. — — 1 — — 1 Capture results Total number of small mammal captures 7 9 2 4 5 6 Total number of Microgale captures 7 9 1 4 5 5 Total number of Insectivora species 3 6 1 3 3 3 Total number of Microgale species 3 6 1 3 3 3 Capture rate of Microgale 9.1% 1 1 .7% 3.3% 4.5% 5.7% 5.7% * The site in the spiny bush lacked the topographical relief found in parcel 1 humid forest, and the designation of the position categories is slightly exaggerated. t V = valley; S = slope; R = ridge. 12, 13, and P2; talonid of m3 with well-developed hypoconulid but reduced hypoconid, entoconid, and entoconid ridge, and narrow, shallow talonid basin. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of females to males was 1:1.5. The ratio of juveniles to adults was 1:1. The anterior dentition of all juveniles was completely decidu- ous, with all molars erupted. One adult female was lactating. Mammary formula: 1-0-2 (n = 2). Remarks — This species is now known to occur in a variety of forested localities ranging from the RNI d'Andohahela in the south to PN de la Mon- tagne d'Ambre in the far north (Raxworthy & Nussbaum, 1994; Goodman et al., 1996a; Jenkins et al., 1997). It has also been recorded in anthro- pogenic habitats, including isolated and frag- mented small forest blocks (Goodman et al., 1996b). In the RNI d'Andohahela (parcel 1) it was recorded across the complete elevational range of the transect from 440 to 1875 m. At other sites surveyed along elevational gradients, this species also tends to be captured in pitfall buckets in relatively low numbers across a broad altitu- 200 FIELDIANA: ZOOLOGY Table 13-4. Extended. Line 7 8 9 10 11 12 13 14 15 16 17 18 Upper montane Montane Montane sc lerophyllous Spiny bush 1200 m 1500 m 1875 m 120 m* 1225 1200 1215 1500 1500 1550 1890 1900 1825 130 120 110 R V S V S R V R S S R V 7/11 7/11 7/11 17/11 18/11 18/11 27/11 27/11 28/11 8/12 8/12 8/12 15/11 15/11 15/11 25/11 25/11 25/1 1 3/12 3/12 3/12 13/12 13/12 13/12 99 99 99 99 88 88 77 77 66 66 66 66 7 1 14 1 12 4 20 1 42 4 1 1 4 1 1 3 2 1 1 1 1 12 5 1 1 2 1 1 1 2 5 15 5 8 6 15 21 19 41 59 26 5 5 4 4 6 6 15 20 19 39 54 26 0 0 0 4 5 4 6 6 5 7 7 5 2 2 2 4 4 4 6 6 5 6 6 5 0 0 0 4.0% 6.1% 6.1% 17.0% 22.7% 21.6% 50.1% 70.1% 39.4% 0.0% 0.0% 0.0% dinal range (Raxworthy & Nussbaum, 1994; Goodman et al., 1996a; Goodman & Jenkins, 1998). Another similarly sized species, Microgale pu- silla, is not known to occur sympatrically with M. parvula at any site. The former species also has a broad distribution, occurring across much of the eastern humid forest and at a variety of elevations (MacPhee, 1987). The only record we are aware of for M. pusilla in extreme southeastern Mada- gascar is a specimen (USNM 578862) collected in the littoral forest of Manafiafy (24°47'S, 47°12'E). This site is within about 35 km of those in the RNI d'Andohahela in which M. parvula was collected, and on the basis of current infor- mation, these two allopatric species apparently have a complicated patchwork distributional pat- tern. Microgale principula Thomas, 1926 Microgale sorella Thomas, 1926: 1987 MacPhee, Holotypk— BM (NH) 25.8.3.15: adult female, body in alcohol, skull extracted, collected by C. Lamberton. Typk Locality — Midongy du Sud, SE Mada- gascar (Midongy Atsimo, 23°35'S, 47°01'E; see MacPhee, 1987). Referred Material— FMNH 156453, 156460, 156461, 156575, 156577: 8 km NW of Emini- GOODMAN ET AL.: LIPOTYPHLA 201 Table 13-5. Summary of capture results for insectivores trapped in pitfall lines in the humid forest (parcel 1) and spiny bush (parcel 2) of the RNI d'Andohahela. Elevation (m) Parcel 2 120 Parcel 1 Species 440 810 1200 1500 1875 Geogale aurita Suncus madagascariensis Microgale principula M. thomasi 7 7 5 2 2 4 1 2 M. fotsifotsy M. longicaudata M. parvula 2 1 2 5 2 3 2 1 5 23 1 Tenrec ecaudatus 1 M. cowani 3 26 73 M. soricoides 1 1 3 3 M. dobsoni 5 4 M. gracilis M. gymnorhyncha Oryzorictes hova 1 2 4 2 6 Total specimens captured Total Microgale captured Total species captured Capture success 14 0 2 7.1% 12 12 5 6.1% 14 14 5 7.1% 13 12 7 6.6% 43 43 7 21.7% 116 110 8 58.6% The capture rate is standardized for the first 6 nights that three pitfall lines (1 1 buckets each) were in place within each elevational zone. miny, 24°37'S, 46°45'E, 440 m; FMNH 156454, 156591: 12.5 km NW of Eminiminy, 24°35'S, 46°44'E, 810 m; FMNH 156592: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m. Key Features — Medium sized, TL very long, more than twice as long as HB; distal portion of tail naked and transversely wrinkled on dorsal surface; fifth hind digit elongated, subequal in length to second digit. Pelage distinctly bicolored, reddish brown dorsally, gray with buff wash ven- trally. Skull medium in size, rostrum moderately short and broad, braincase moderately narrow. Short diastemata between 1 1 and 12, and on either side of C and P2, with 12 and 13 more or less in contact; well-developed anterior and posterior ac- cessory cusps present on 12, C, and P2; II greater in crown height than C; P4 distinctly greater in crown height than P3; p2 moderately caniniform; talonid of m3 with low hypoconid, well-devel- oped hypoconulid, broad talonid basin, reduced entoconid ridge, and entoconid absent. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of females to males was 1:4. The ratio of juveniles to adults was 1:1.7. The anterior dentition of all three juveniles was fully decidu- ous. The testes of one adult male (FMNH 156592) measured 6X4 mm, with convoluted epididy- mides. Mammary formula: 1-0-2 (n = 1). Remarks — One specimen was trapped 1 m above the ground on a 4-cm-diameter horizontal branch in a vine tangle, although most others were collected in pitfall traps. As mentioned un- der the species account for Microgale longicau- data, Microgale principula was more common in the 440 m zone than at higher altitudes. On the basis of our trapping results, the upper elevation- al limit of this species is approximately 1200 m. It is known from a variety of sites in the eastern humid forest from the RNI d'Andohahela north to at least the RS d'Anjanaharibe-Sud (Goodman & Jenkins, 1998). Microgale soricoides Jenkins, 1993 Holotype — BM (NH) 91.565: adult male in al- cohol, skull extracted, collected 13 April 1991 by Christopher J. Raxworthy. Type Locality — Mantady National Park, ca. 15 km N of Perinet (Andasibe), 18°51'S, 48°27'E, in primary rain forest, 1100-1150 m. Referred Material— FMNH 156594: 12.5 km 202 FIELDIANA: ZOOLOGY NW of Eminiminy, 24°35'S, 46°44'E, 810 m; FMNH 156462, 156583, 156595: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m; FMNH 156463-156465, 156596: 15 km NW of Emini- miny, 24°34'S, 46°43'E, 1500 m; FMNH 156466, 156467, 156597, UA-SMG 7668, 7676, 7683: 20 km SE of Andranandambo, 24°33'S, 46°43'E, 1875 m. Key Features — Size large, TL subequal to or longer than HB. Pelage light buff brown dorsally, gray-brown ventrally with reddish buff wash. Skull moderately large and robust, rostrum and interorbital region broad, braincase short and broad; supraoccipital ridge present. First upper II markedly robust and proodont; il and i2 robust and procumbent, i2 smaller than il but larger than c; first upper and lower premolars very small, with a single root; talonid of m3 reduced to very low hypoconid, oblique crest, and prominent hy- poconulid. Measurements — External and cranial measure- ments are presented in Table 13-3. Variation — The early replacement of the first upper incisor, described below, may be abnor- mal. There are, however, indications from other populations that this eruption sequence, atypical for most species of Microgale, is the norm for this species. Specimens from RNI d'Andohahela are smaller on average than those collected from sites further north in Madagascar (see Jenkins, 1993; Jenkins et al., 1996; Goodman & Jenkins, 1998). Population Structure and Reproduction — Of the adults in this small sample only females were collected. The ratio of juveniles to adults was 1:4. In both juveniles, the anterior dentition, with the exception of II, was deciduous, but M3 and m3 in one specimen were still in the process of erupting. Two of the adult females were preg- nant and three were lactating. Mammary formula: 0-1-2 (n = 1), 0-2-1 (n = 2), 1-0-2 (n = 2), 1-1- 2 (n = 3). Remarks — Most specimens were caught on the ground in Sherman or pitfall traps, although one was trapped 2 m above ground on a 2-cm-diam- eter horizontal liana that ran from the ground to a height of 7 m. This recently described species is broadly distrib- uted in the eastern humid forest; it is known to occur in a variety of sites from the RNI d'Andohahela north to at least the RS d'Anjanaharibe-Sud, includ- ing RNI d'Andringitra and PN de Mantady (Jenkins, 1993; Goodman et al., 1996c; Goodman & Jenkins, 1998). Microgale thomasi Major, 1896 Holotype— BM (NH) 97.9.1.108: adult male; skin, skull, and skeleton, collected 19 July 1895 by C. I. Forsyth Major. Type Locality — Ampitambe Forest (The po- sition of this locality has never been satisfactorily fixed, although the best estimate to date is that proposed by Carleton & Schmidt, 1990, in Fia- narantsoa Province at ca. 20°22'S, 47°46'E, >900 m.). Referred Material— FMNH 156472-156478: 8 km NW of Eminiminy, 24°37'S, 46°45'E, 440 m; FMNH 156353, 156354, 156479-156481, 156599, 156600: 12.5 km NW of Eminiminy, 24°35'S, 46°44'E, 810 m; FMNH 156482-156484: 13.5 km NW of Eminiminy, 24°35'S, 46°44'E, 1200 m. Key Features — Moderately large Microgale (see Table 13-3), TL shorter than HB, ears mod- erately prominent. Dorsal pelage speckled dark rufous brown, paler ventrally; tail bicolored at least proximally in most specimens, dark brown dorsally, buff ventrally, with moderately long, dense scale hairs. Skull moderately robust, brain- case deep, short and broad; basisphenoid stepped between anterior ends of tympanic bullae. First upper and lower premolars large; hypoconid of m3 prominent, entoconid and entoconid ridge present, talonid basin well marked. Measurements — External and cranial measure- ments are presented in Table 13-3. Population Structure and Reproduction — The sex ratio of females to males was 1:2. The ratio of juveniles to adults was 1:1.8. The juve- niles showed variation in dental development; in one the anterior dentition was fully deciduous, in another i3 was erupting, in a third II and 13 were erupting, and in the fourth all the permanent an- terior teeth were erupting or had erupted, with the exception of the second deciduous upper and low- er premolars (dP3 and dp3). One adult female was lactating and one juvenile female (FMNH 156482) with deciduous dentition was pregnant with one embryo in each oviduct. Conversely, one male with a fully permanent dentition (FMNH 156600) was reproductively immature with non- convoluted epididymides. Mammary formula: 0- 1-2 (n = 1), 0-2-1 (n = 1), 0-2-2 (n = 1). Remarks — Prior to the recent wave of biological inventories on Madagascar, Microgale thomasi was poorly known in museum collections and seemed to be restricted to the southern portion of the east- ern humid forest (MacPhee, 1987). In the humid forests of the RNI d'Andohahela this species was GOODMAN ET AL.: LIPOTYPHLA 203 relatively common from low to mid-altitudes. It was also commonly captured with pitfall tech- niques in the Marosohy Forest, along the north- eastern boundary trail of parcel 1 of the RNI d'Andohahela (USNM 578772-578783). It is also known to occur in and around the PN de Rano- mafana (USNM 448876, 448877, 448917, 449193-449196). The northern limit of this species is apparently the Perinet area (Heim de Balsac, 1972), and it is also known from the humid forests to the east and northeast of Ambositra ( Ampitambe and Ivohimanitra). Micro gale monticola, a species morphologically similar to M. thomasi, occurs fur- ther north, in the vicinity of the RS d'Anjanaharibe-Sud and within the elevational range of 1550-1950 m (Goodman & Jenkins, 1998). Microgale thomasi and M. monticola are believed to be allopatric. Family Soricidae Suncus madagascariensis (Coquerel, 1848) Holotype — MNHN no. 96: unsexed mounted specimen, skull removed and subsequently lost, brought back to France by M. Coquerel in 1847, collection date unknown (Rode, 1942). Type Locality — Nossi-Be (= Nosy Be). Referred Material— FMNH 156493, 156494, 156604, 156605: 7.5 ENE of Hazofotsy, 24°49'S, 46°36'E, 120 m. Key Features — Extremely small (see Table 13-1). Pelage soft and short; tail with long, fine- bristle hairs. Skull very small and delicate; ros- trum short and narrow; interorbital region narrow, short; braincase relatively long, shallow, and broad. Dental formula 3/1 1/0 2/2 3/3 = 30 (no- menclature of anterior unicuspid dentition unre- solved); II well developed, proodont, principal cusp hook-shaped; upper unicuspids decrease in size from moderately large anteriormost (12) to very small fourth unicuspid; talon of P4 broad; cusp pattern of trigon M2 and M3 dilambodont, M3 anteroposteriorly compressed; il large, pro- cumbent; c absent; first lower unicuspid small, second lower unicuspid approximately half height of ml; m3 talonid reduced to a single cuspid but talonid basin present. Measurements — External measurements are presented in Table 13-1. Population Structure and Reproduction — All specimens collected were male; two of these (FMNH 156604 and 156605) had convoluted ep- ididymides measuring respectively 3X2 and 4 X 2 mm. Remarks — All specimens were collected in slightly disturbed spiny forest using pitfall traps. The taxonomic status of this form is unclear. Heim de Balsac and Meester (1977) tentatively placed this animal as a subspecies of Suncus etruscus, but more recently Hutterer (1993) ele- vated it to a species. It is uncertain if this animal was introduced to Madagascar or is part of the native fauna. If it was introduced, the origin of the founding population is unknown. These ques- tions are probably best resolved with biochemical and karyotypical studies. Analysis and Discussion General A total of 1,254 pitfall bucket-days was accrued during the survey of the RNI d'Andohahela be- tween 26 October and 13 December 1995. These were divided between the two habitat types as 23 1 bucket-days at 440 m, 264 at 810 m, 297 at 1200 m, 275 at 1500 m, and 220 at 1875 m in the hu- mid forest (parcel 1), and 198 bucket-days at 120 m in the spiny forest (parcel 2) (Tables 13-4 and 13-5). There were 233 small mammals captured in parcel 1, including 220 Microgale, seven Ory- zorictes, two Tenrec, and three species of Eliurus (for rodents, see Chapter 14). Fourteen small mammals were caught in parcel 2, including sev- en individuals each of Geogale aurita and Suncus madagascariensis. Furthermore, 4,573 trap-nights were accrued in the humid forest using small mammal traps with a standard baiting regime (Chapter 14), and 27 (0.60 %) lipotyphlans were captured. After 685 trap nights in the spiny bush, one (0.15 %) tenrecine was caught. The following species were obtained in standard museum traps: parcel 1 — Microgale dobsoni, M. gymnorhyncha, M. principula, M. soricoides, M. thomasi, Ory- zorictes hova, and Setifer setosus; parcel 2 — Echinops telfairi. All species of Microgale re- corded within each elevational zone were cap- tured in the pitfall traps, and no species was taken exclusively in the small mammal traps. The single exception was an individual of M. dobsoni caught at 1200 m; this constituted our single record for the species within that elevational zone. The combined trapping results, with pitfalls and standard live traps, located 15 species of Tenre- 204 FIELDIANA: ZOOLOGY cidae (Echinops, Setifer, Tenrec, Geogale aurita, Oryzorictes hova, and 10 species of Microgale) and one species of Soricidae (Suncus madagas- cariensis) within the RNI d'Andohahela. The only lipotyphlans previously documented from the re- serve are Tenrec and Setifer (O'Connor et al., 1987; Nicoll & Langrand, 1989). Before proceed- ing with the analysis of the trapping results, it is important to establish whether the sampling effort was sufficient to reflect some measure of com- pleteness for the survey and the actual lipotyphlan species richness within each elevational zone. Species Accumulation Curves The total number of species known from each elevational zone was plotted as a function of sam- pling effort (33 pitfall buckets per 24-hr period) to produce species accumulation curves (Fig. 13- la). An examination of these curves shows that the accumulation of previously unrecorded spe- cies was slow in the 440 m zone, and by the end of the sampling period a plateau had not been reached. This pattern is in contrast to those from virtually all of the other elevational zones sampled in parcel 1; an asymptote was reached in the 810 m zone after 66 pitfall bucket-days (total of five species in 264 pitfall bucket-days), in the 1200 m zone after 231 pitfall bucket-days (total of eight species in 297 pitfall bucket-days), in the 1500 m zone after 264 pitfall bucket-days (total of eight species in 297 pitfall bucket-days), and in the 1875 m zone after 154 pitfall bucket-days (total of eight species in 220 pitfall bucket-days). In the spiny bush (parcel 2) the plateau in the species accumulation curve was reached after 2 nights of pitfall operation (total of two species in 198 buck- et-days). In both parcels 1 and 2 the flattening of the species accumulation curve within each ele- vational zone that occurred with additional trap- ping effort did not generally coincide with a de- cline in overall pitfall trap success (Fig. 13- lb). The number of lipotyphlans captured within each elevational zone did tend to decline, however, dur- ing the time the lines were in place. We feel that our overall trapping results, partic- ularly with pitfall techniques, closely parallel the actual species richness of lipotyphlans within each elevational zone. (For further discussion of this point see Goodman & Jenkins, 1998, p. 155.) The only species that has been previously documented from the immediate region of the RNI d'Andohahela and was not encountered during the 1995 survey is Microgale talazaci. This animal has been collected in the Nahampoana Forest, from a site resting on lateritic soils at 100 m (USNM 577053-57) and in the Marosohy Forest, between 700 and 725 m, along the northeastern boundary of parcel 1 of the RNI d'Andohahela (USNM 578746, 578747). Given these records we assume that it also occurs within parcel 1 of the RNI d'Andohahela, but was not captured in our trapping devices. An additional species, Hemicen- tetes, was previously reported as possibly occur- ring in the reserve (O'Connor et al., 1987). Other studies of the Malagasy Lipotyphla using pitfall traps indicate that this technique is effective for documenting species richness in eastern humid forest (Raxworthy & Nussbaum, 1994; Goodman et al., 1996c). Trapping Success and Abundance Summary information on the pitfall trapping is given in Tables 13-4 and 13-5. Three pitfall lines were in operation within each elevational zone for a minimum of 6 days, and there was considerable variation in the capture rate of lipotyphlans within and between lines. For Microgale in humid forest the capture success rate at 440 m varied from 3.3% to 11.7% (average 6.1%), at 810 m from 4.5% to 5.7% (average 7.1%), at 1200 m from 4.0% to 6.1% (average 6.6%), at 1500 m from 17.0% to 22.7% (average 21.7%), and at 1875 m from 39.4% to 70.1% (average 58.6%). In the spiny forest portion of parcel 2 the capture suc- cess rate for the single site sampled (120 m) var- ied from 6.1% to 7.6% (average 7.1%). At the 1500 and 1875 m sites in the RNI d'Andohahela, the pitfall capture success was ex- ceptionally high. In the PN de la Montagne d'Ambre, capture rates of comparable magnitude (range 60.6-77.9%) have been reported for an area of montane forest between 1300 and 1380 m, just below the summit (1475 m) of the mountain (Goodman et al., 1996a). At lower elevations on the same mountain, capture rates as high as 50.4% have been calculated for 1250 m in mid-montane forest and between 55.8% and 58.4% in the 980- 1010 m zone (Raxworthy & Nussbaum, 1994; Goodman et al., 1996a). Relationship Between Rainfall and Capture Rates During the course of numerous field seasons in Madagascar trapping lipotyphlans with pitfall GOODMAN ET AL.: LIPOTYPHLA 205 Bucket-nights ~ — 1 -T -A ,,•'' '//''.*■•' oV- ...o >' '*o. O"- ■;-<•"■ -0- *" "O* 0i J p M i 1 1 V l V 33 66 99 132 165 198 231 264 297 Bucket-nights Fig. 13-1. Species accumulation curves (A) and pitfall trap success (B) plotted for each elevational zone in the RNI d'Andohahela against the total number of bucket-nights. The pitfall lines placed in parcel 2 (spiny bush) were 206 FIELDIANA: ZOOLOGY techniques it has been our impression that capture rates are higher after rainfall. To determine wheth- er there was a relationship between these variables during the RNI d'Andohahela survey, a series of Kendall's rank correlations was computed for each elevational zone, with the v variable being the number of animals captured and the x variable the amount of precipitation during the previous 24 hr. This procedure mirrors that used by Goodman et al. (1996c) to evaluate the same relationship in the RNI d'Andringitra. In the RNI d'Andohahela no significant corre- lation was found between capture rate and rain- fall. The next stage in the analysis was to perform a regression analysis for each elevational zone us- ing the same variable configuration as in Ken- dall's rank correlations. In all cases there was a positive trend between rainfall and capture rates, but the results were not statistically significant. In general there was little rainfall during the inven- tory, particularly at the 440, 810, and 1200 m sites (see Chapter 1). Perhaps if the inventory had been slightly later in the year and during the principal rainy season (usually starting in December) there would have been a stronger correlation between these variables. Scansoriality in Species of Microgale of M. principula was obtained on a 4-cm-diameter horizontal branch running through vine tangle and 1 m off the ground. This long-tailed species, which Eisenberg and Gould (1970) placed in their climbers and springers class, had been noted pre- viously to have prehensile tail adaptations (Tho- mas, 1926). Although this species and M. longi- caudata show modifications of the tail for pre- hensile use (Goodman & Jenkins, 1998), the vast majority of the M. principula and all of the M. longicaudata animals captured in the RNI d'Andohahela were in pitfall buckets. Thus, these two species spend considerable time moving on the ground. One M. dobsoni was obtained 2.5 m above ground on a 3-cm-diameter horizontal branch. This relatively short-tailed species had been classified previously as having moderate climbing ability (Eisenberg & Gould, 1970). An individual of the moderately long-tailed species M. soricoides was trapped 2 m above the ground on a 2-cm-diameter horizontal liana that extended from ground level to about 7 m. TL/HB in this species is 1.2 ± 0.09 (n = 12, range 1.0-1.3), and HF/HB is 0.22 ± 0.01 (n = 12, range 0.20-0.23); these values would place it in Eisenberg and Gould's (1970) class of animals with climbing ability. Although there are few data available on arboreality in the Tenrecidae, no species is known to live exclusively off the ground. Using morphometric correlations associated with tail, body, and hind foot measurements, Ei- senberg and Gould (1970) divided Microgale spe- cies into four locomotor forms: fossorial or sem- ifossorial; surface foragers with moderate climb- ing ability; surface foragers and climbers; and climbers and springers. Because our principal method of trapping lipotyphlans was the use of pitfall buckets, which only capture species mov- ing along the ground, much of our information could not be used to address the question of ad- aptations toward arboreality in this group. About 40% of the small mammal traps were installed above the ground within each elevational zone, however, and 4,395 trap-nights were accrued (see Chapter 14); these data are relevant to the ques- tion of arboreality in Microgale. Single individuals of three Microgale species were captured in mammal live traps. A specimen Altitudinal Variation Within Parcel 1 On the basis of the trapping regimens used dur- ing the survey, we are able to present some details on the elevational ranges of the various Lipotyph- la found in parcels 1 and 2 of the RNI d'Andohahela (Table 13-6). Setifer setosus was found only in lowland forest, at an altitude of 440 m, whereas Tenrec ecaudatus was observed or trapped between 440 and 1200 m. Species of Mi- crogale were found at all altitudes within the dif- ferent types of humid forest. M. principula and M. thomasi were documented in lowland, transi- tional, and montane forest between 440 and 1200 m, with more specimens of both species collected at the lower altitudes. Only two species, M. lon- gicaudata and M. parvula, were found at all al- titudes and forest types from 440 to 1875 m. Mi- ax 120 m and those in parcel 1 (humid forest) at 440. 810. 1200. 1500. and 1875 m. Information from the three lines at each zone is combined. GOODMAN ET AL.. LIPOTYPHLA 207 Table 13-6. The known elevational distribution of Lipotyphla in the humid forest (parcel 1) and spiny bush (parcel 2) of the RNI d'Andohahela. Species Elevation (m) Parcel 2 120 Parcel 1 440 810 1200 1500 1875 Geogale aurita Suncus madagascariensis Echinops telfairi Setifer setosus Microgale principula Microgale thomasi Microgale fotsifotsy Microgale longicaudata Microgale parvula Tenrec ecaudatus Microgale cowani Microgale soricoides Microgale dobsoni Microgale gracilis Microgale gymnorhyncha Oryzorictes hova Total number of species Total number of Microgale x x [x] X X X X [x] Information is based on all trapping techniques. Inferred occurrence within an elevational swath [in brackets] is based on the presence of a species in the zones above and below the one in question. Sight records are designated with the letter S. * The only record of this species in the 1 200 m zone was an individual recovered from the digestive system of a snake (see text p. 193). This record is not included in the species total for the 1200 m zone. crogale longicaudata was considerably more abundant in transitional upper montane/sclero- phyllous forest, at 1875 m (63.2%), than in low- land forest, at 440 m (2.7%), based on pitfall cap- tures. Microgale cowani and M. soricoides oc- curred from transitional lowland/montane forest at 810 m to transitional upper montane/sclerophyl- lous forest at 1875 m. Both species were also most abundant at higher altitudes. In M. cowani 7 1 .6% of the specimens captured in pitfalls during the survey were obtained at 1875 m, 24.5% at 1500 m, and 2.9% at 1200 m. Microgale dobsoni was found from 1200 to 1875 m, whereas both M. gracilis and M. gymnorhyncha were collected only at the highest altitudes, from 1500 to 1875 m. Microgale fotsifotsy was the only species ap- parently showing a disjunct distribution, with very small samples collected at 410, 1200, and 1500 m; this is attributed to a sampling anomaly. Ory- zorictes hova was collected only at 1875 m. Habitat Separation As defined by numerous proposed systems for the vegetational classification of Madagascar (re- viewed by Lowry et al., 1997; see Chapter 2), the two parcels of the RNI d' Andohahela investigated during this study are floristically different. Re- gardless of the names applied to these two habitat types, parcel 1 can be characterized as humid (rain) forest and parcel 2 as xerophytic bush. Be- tween the two parcels, which are separated by a distance of about 20 km, there is a remarkable change in the climate and flora across the rain shadow of the Anosyenne Mountains. Rain sys- tems moving in from the eastern coast of Mada- gascar release their precipitation along the eastern slopes of the Anosyenne Mountains, and as Donque (1972, p. 136) has noted, "the boundary between the semi-arid climate and the tropical damp climate of the south-east coast is extremely sharp." This ecotone between wet and dry is known to be a major barrier to dispersal for cer- tain groups of land vertebrates (Goodman et al., 1997). Tenrec ecaudatus, Oryzorictes hova, and all species of Microgale were taken only in the hu- mid forest of parcel 1 , and Echinops telfairi, Geo- gale aurita, and Suncus madagascariensis were captured only in the spiny forest of parcel 2. We found no species in common between parcel 1 and 208 FIELDIANA: ZOOLOGY Table 13-7. Faunal similarity coefficients for Lipotyphla communities in the various altitudinal and ecological zones of parcels 1 (120 m) and 2 (440, 810, 1200, 1500, and 1875 m) in the RNI d'Andohahela. 120 m 440 m 810 m 1200 m 1500 m 1875 m 120 m 0.0 0.0 0.0 0.0 0.0 440 m 0.0 — 0.88 0.88 0.50 0.38 810 m 0.0 0.78 — 1.0 0.63 0.50 1200 m 0.0 0.70 0.89 — 0.75 0.63 1500 m 0.0 0.33 0.45 0.55 — 0.88 1875 m 0.0 0.23 0.33 0.42 0.78 — The similarity coefficients derived from Simpson's Index are above the diagonal; those derived from the Jaccard Index are below the diagonal. 2 of the reserve. The only possible exception to this statement is Tenrec ecaudatus, which has been reported from both parcels 1 and 2 (O'Connor et al., 1987). The western versant of the Anosyenne Mountains, along the abrupt gra- dient passing westward from wet to dry, thus ap- pears to represent a dramatic barrier to dispersal and mixing of the lipotyphlan faunas within these two parcels. Despite the modern dramatic faunal turnover along the western slopes of the Anosyenne Moun- tains, this boundary (or at least the site of the boundary) was not constant over recent geological time, and there have been marked fluctuations in the regional biological communities. The Grotte d'Andrahomana, a well-known Holocene subfos- sil cave site in the region, is located about 50 km west-southwest of Tolagnaro in a region that is now largely spiny bush. This site, which was ex- cavated in the early portion of this century (Gran- didier, 1902; Goodman & Rakotondravony, 1996), contained a rich assortment of vertebrate material. Among these remains were many species of ex- tinct and extant lemurs, some with affinities to the humid forest and others to the dry forest (Godfrey et al., 1997). The type specimen of Microgale de- caryi G. Grandidier, 1928 (a synonym of M. prin- cipula according to MacPhee, 1987) was exca- vated from this cave. The remains of an extinct rodent, Hypogeomys australis, whose sole extant congener lives in dry deciduous forest in the Mo- rondava region, have been excavated from the cave as well (Grandidier, 1903). Radiocarbon dat- ing of Hypogeomys remains from the Grotte d'Andrahomana shows that these animals existed in the region 4,440 ± 60 BP (Goodman & Rako- tondravony, 1996). Along the eastern slopes of parcel 1 the shifts in habitat between the various elevational zones are part of a continuum of vegetational change as a function of altitude and microclimate shifts. In or- GOODMAN ET AL.: LIPOTYPHLA der to further assess the faunistic relationships of the lipotyphlan faunas between each of the eleva- tional zones, we calculated two different similarity indices, Simpson's Index and the Jaccard Index: Simpson's Index = Jaccard Index = N N, + N2 - C where N, = the number of species at site 1 (the smaller fauna), N: = the number of species at site 2, and C = the number of species common to both sites. The coefficients derived from these indices, presented in Table 13-7, were used in a cluster algorithm ("Phylip," written by J. Felsenstein us- ing the Fitch-Margoliash method with contempo- rary tips). The results of the cluster analyses are presented in Figure 13-2. Both indices show the same re- lationship. Within the elevational zones studied in the reserve, the spiny bush region at 120 m con- tains a fauna that is unique with regard to the other sites in the humid forest. Three distinct groups occur in parcel 1 : ( 1 ) the lipotyphlan spe- cies found in the lowland forest, at 440 m, (2) the lower montane community, between 810 and 1200 m, and (3) the upper montane to sclero- phyllous forest group, between 1500 m and the summital zone at 1875 m. Comparison Between Mountains Over the past decade or so there has been con- siderable instability in the alpha-level taxonomy of the genus Microgale, and the number of spe- cies recognized has varied considerably (cf. MacPhee, 1987; Nicoll & Rathbun, 1990; Ste- phenson, 1995). Our knowledge of the geograph- ical ranges of Microgale has grown substantially 209 0.06 0.32 0.12 — 1875 m A 0.06 — 1500 m 0.001 0.06 1200 m 0.06 ttU 1 1 1 0.50 0.21 ( 0.18 — 1875 m B 0. 12 — 1500 m 0.0 5 0.07 1200 m - flirim 0. 16 AACl m °-50 1onm Fig. 13-2. Cluster analysis of faunal similarity of in- sectivore communities occurring in a range of elevation- al zones in parcel 1 (humid forest, sites from 440 to 1875 m) and parcel 2 (spiny bush, site at 120 m) in the RNI d'Andohahela. Simpson's Index (A), and the Jac- card Index (B) were used. The data are based on Table 13-7, which is derived from Table 13-6. over the past few years, largely through the efforts of field surveys that involved specimen collection. The increase in research on Microgale — and Ten- recidae in general — has brought with it a clearer picture of species limits within this group. With the use of comparable field techniques, direct comparisons are now possible between the lipotyphlans from the eastern humid forests of RNI d'Andringitra, RS d'Anjanaharibe-Sud, or PN de la Montagne d'Ambre and those from RNI d'Andohahela (parcel 1 only) (Table 13-8). These four localities span the complete latitudinal swath of eastern humid forests on the island. From south to north, there have been 13 lipotyphlan species (including 10 Microgale spp.) documented in the RNI d'Andohahela (parcel 1), 15 species (includ- ing 11 Microgale) in the RNI d'Andringitra (Goodman et al., 1996c; Jenkins et al., 1996; Goodman, unpubl. data), 15 species (including 11 Microgale) in the RS d'Anjanaharibe-Sud (Good- man & Jenkins, 1998), and nine species (includ- ing six Microgale) in the PN de la Montagne d'Ambre (Raxworthy & Nussbaum, 1994; Good- man et al., 1996a; Jenkins et al., 1997). Thus, with the exception of the reduced species richness in the Montagne d'Ambre region, there is consider- able consistency in the level of lipotyphlan diver- sity within the humid forests across the 10 degrees of latitude from Andohahela to Anjanaharibe-Sud. It is possible that the reduced species richness from Montagne d'Ambre is related to recent vol- canic activity and isolation from other humid for- ests in recent geological times (Goodman et al., 1996a). On the basis of our current knowledge with re- gard to patterns of geographical variation and Mi- crogale species limits, a number of generaliza- tions can be made. Several animals are found across a vast area of the eastern humid forest from Andohahela to Anjanaharibe-Sud (those species whose names are followed by an asterisk occur north to Montagne d'Ambre): M. cowani, M. dob- soni, M. fotsifotsy, * M. gymnorhyncha, M. longi- caudata* M. parvula* and M. soricoides. Other species are known from the southern portion of the humid forest zone: M. gracilis, M. principula, and M. thomasi; and some are known from the northern portion of this zone: M. brevicaudata and M. talazaci. One species, M. taiva, appears to be centrally distributed. Microgale drouhardi has an apparently discontinuous distribution, based on the sites sampled. Finally, two species, M. dryas and M. monticola, apparently have limited geo- graphical ranges. Relationships Between Species Richness and Abundance The data from these four sites allow further elaboration of the relationships among Microgale species richness, density, and variation along ele- vational and latitudinal gradients. Numerous hy- potheses have been presented to explain patterns of diversity along gradients (e.g., Brown, 1988; Colwell & Hurtt, 1994; Rosenzweig, 1995; Rah- bek, 1997). One that has received considerable at- tention is that changes in species richness along elevational gradients closely mirror environmental productivity (Ricklefs & Schluter, 1993; Rosen- zweig, 1995). For tropical forests there is evi- dence that productivity is highest at mid-eleva- tions (Janzen et al., 1976). Our purpose here is not to test this hypothesis, 210 FIELDIANA: ZOOLOGY Table 13-8. Distribution of Lipotyphla on four different mountains in the eastern humid forest of Madagascar. Species Site: Elevational range (m): Latitude: RS PN de la RNI RNI d'Anjanaharibe- Montagne d'Andohahela* d'Andringitrai" Sud$ d'Ambre^ 440-1875 m 720-2450 m 875-1950 m 340-1350 m 25°S 22°S 15°S 12°S Soricidae Suncus murinus\\ Tenrecidae Hemicentetes nigriceps Hemicentetes sp. Microgale brevicaudata Microgale cowani Microgale dobsoni Microgale fotsifotsy Microgale gracilis Microgale gymnorhyncha Microgale longicaudata Microgale drouhardi Microgale dryas Microgale monticola Microgale panula Microgale principula Microgale taiva Microgale talazaci Microgale thomasi Microgale soricoides Oryzorictes hova Oryzorictes tetradactylus Setifer setosus Tenrec ecaudatus Total number of species Total number of Microgale spp. + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 13 15 15 9 10 11 11 6 * Includes only species from parcel 1 . t Information derived from Jenkins et al. (1996). Goodman et al. (1996c), Langrand and Goodman (1997). and Goodman (unpubl. data). X Information derived from Goodman and Jenkins (1998). •I Information derived from Raxworthy and Nussbaum (1994). Goodman et al. (1996a). and Jenkins et al.. (1997). || Species introduced to Madagascar. but simply to evaluate aspects of variation that may indicate nonuniform patterns within our data set (Fig. 13-3). With the exception of PN de la Montagne d'Ambre, all of these site surveys were conducted during the same season of different years. For two sites. RNI d'Andringitra (Fig. 13- 3b) and RS d'Anjanaharibe-Sud (Fig. 13-3c), the species richness of Microgale shows clear peaks in the 1200 m zone or at mid-elevation. Further- more, as measured by percent trap capture, Mi- crogale abundance closely mirrors the mid-ele- vational hump in species richness. Thus species richness and small mammal densities parallel one another at these two sites at mid-latitudes in the eastern humid forest. When these same parame- ters are examined for the RNI d'Andohahela (Fig. 13-3a) and PN de la Montagne d'Ambre (Fig. 13- 3d), sites in the extreme south and north (respec- tively) of the eastern humid forest, a mid-eleva- tional hump in species richness is not present and abundance increases as a function of altitude. There is thus no evidence in these two cases that species richness mirrors changes in abundance (Terborgh, 1977; Graham, 1990). The factors that give rise to these two different patterns along ele- vational gradients in Malagasy humid forest are unknown, but they may be related to variation in meteorological (temperature, cloud cover, and rainfall) patterns on slopes or along the 12° of latitude spanning the eastern humid forest, topo- graphical features that may give rise to variation in orographic precipitation, soil types, food re- sources or resource diversity, effects of radiant- heat, or some aspects of productivity (Donque, GOODMAN ET AL.: LIPOTYPHLA 211 60 50 40 30- 20- 10- 0 "D Microgale species ^> •O percent capture rate / 1 1 1 1 1 440 810 1200 1500 1875 Elevation (m) 15- <>..... ""••o 10- 5- o 4/ o- i i i 720 810 1210 1625 Elevation (m) 1 o- % 10- 'o... o 5- o- i i ! I 875 Elevation (m) 340 1000 1350 Elevation (m) Fig. 13-3. Plots of Microgale species diversity and capture rates (percentage of capture) along elevational gra- dients at four sites on Madagascar: RNI d'Andohahela (A), RNI d'Andringitra (B), RS d'Anjanaharibe-Sud (C), and PN de la Montagne d'Ambre (D). The data are derived from the same sources mentioned in the footnotes to Table 13-8. 1972; Brown, 1973; Janzen et al., 1976; Pendry & Proctor, 1996a,b). The considerable differences that have been noted in patterns of elevational dis- tribution between different organisms in different regions of the world may be in part related to different phylogenetic histories (Wright & Cald- eron, 1995; Patterson et al., in press), and thus a search for a common denominator to explain these different patterns may prove intractable. Acknowledgments This fieldwork could not have been accom- plished without the help of World Wide Fund for Nature, Tolagnaro, particularly Lala Andriamana- rivo and Mark Fenn. We are grateful to officials of the Direction des Eaux et Forets and Association Nationale pour la Gestion des Aires Protegees for permits to enter the reserve and collect specimens. Michael Carleton of the National Museum of Nat- ural History, Washington, D.C., and Laurent Gran- jon of the Museum National d'Histoire Naturelle, Paris, kindly made specimens from their collec- tions available for this study. For comments on an earlier version of this chapter we are grateful to Howard Whidden and an anonymous reviewer. Literature Cited Ade, M. 1996. Morphological observations on a Mi- crogale specimen (Insectivora, Tenrecidae) from west- 212 FIELDIANA: ZOOLOGY ern Madagascar, pp. 251-255. In Ganzhorn, J. U., and J. -P. Sorg, eds.. Ecology and economy of a tropical dry forest in Madagascar. Primate Report, Special Is- sue 46-1, Deutsches Primatenzentrum, Gottingen, 382 pp. 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. Brown, W. L., Jr. 1973. A comparison of the Hylean and Congo-West African rain forest ant faunas, pp. 161-185. In Meggers, B. J., E. S. Ayensu, and W. D. Duckworth, eds.. Tropical forest ecosystems in Africa and South America: A comparative review. Smithson- ian Institution Press, Washington, D.C.. 350 pp. Butler. P. M. 1988. Phylogeny of the insectivores, pp. 117-141. In Benton, M. J., ed.. The phylogeny and classification of the tetrapods, vol. 2. Clarendon Press, Oxford. 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 Novi- tates. 2987: 1-36. Colwell, R. K., and G. C. Hurtt. 1994. Nonbiological gradients in species richness and a spurious Rapoport effect. American Naturalist, 144: 570-595. Coquerel, C. 1848. Note sur une espece nouvelle de musaraigne trouvee a Madagascar. Annales des Sci- ences Naturelles, Zoologie, 9: 193-198. Dobson. G.-E. 1882. A monograph of the Insectivora, part I. John van Voorst, London. Donque, G. 1972. The climatology of Madagascar, pp. 87-144. In Battistini, R.. and G. Richard-Vindard. eds., Biogeography and ecology in Madagascar. W. Junk, The Hague, 765 pp. Eisenberg. J. F, and E. Gould. 1970. The tenrecs: A study in mammalian behavior and evolution. Smith- sonian Contributions to Zoology, 27: 1-138. Ganzhorn, J. U., A. W. Ganzhorn, J. -P. Abraham, L. Andriamanarivo, and A. Ramananjatovo. 1990. The impact of selective logging on forest structure and tenrec populations in western Madagascar. Oecologia (Berlin), 84: 126-133. Ganzhorn, J. U.. S. Sommer. J. -P. Abraham, M. Ade, B. M. Raharivololona, E. R. Rakotovao, C. Rak- otondrasoa, and R. Randriamarosoa. 1996. Mam- mals of the Kirindy Forest with special emphasis on Hypogeomys antimena and the effects of logging on the small mammal fauna, pp. 215-232. In Ganzhorn, J. U.. and J. -P. Sorg, eds.. Ecology and economy of a tropical dry forest in Madagascar. Primate Report, Special Issue 46-1, Deutsches Primatenzentrum, Got- tingen, 382 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. Godfrey. L. R., W. L. Jungers, K. E. Reed, E. L. Si- mons, and P. S. Chatrath. 1997. Subfossil lemurs: Inferences about past and present primate communi- ties in Madagascar, pp. 218-256. In Goodman, S. M., and B. D. Patterson, eds.. Natural change and human impact in Madagascar. Smithsonian Institution Press, Washington, D.C, 432 pp. Goodman, S. M., A. Andrianarimisa, L. E. Olson, and V. Soarimalala. 1996a. Patterns of elevational dis- tribution of birds and small mammals in the humid forests of Montagne d'Ambre, Madagascar. Ecotropi- ca, 2: 87-98. Goodman, S. M., and J. U. Ganzhorn. 1994. Les petits mammiferes, pp. 58-63. In Goodman. S. M., and O. Langrand, eds., Inventairc biologique de forct dc Zombitse. World Wide Fund for Nature-Centre d' Information et de Documentation Scientifiquc et Technique, Antananarivo. Recherches pour le Devel- opment, Serie Sciences biologiques. No. Special, 106 pp. Goodman, S. M., and P. D. Jenkins. 1998. The insec- tivores of the Reserve Spcciale d'Anjanaharibe-Sud, pp. 139-161. In Goodman, S. M., ed., A floral and faunal inventory of the Reserve Spcciale d'Anjanaharibe-Sud. Madagascar: With reference to elevational variation. Fieldiana: Zoology, n.s. 90: 1- 246. Goodman, S. M.. M. Pidgeon, A. F A. Hawkins, and T. S. Schulenberg. 1997. The birds of southeastern Madagascar. Fieldiana: Zoology, n.s. 87: 1-132. Goodman, S. M., and D. Rakotondravony. 1996. The Holocenc distribution of Hypogeomys (Rodentia: Muridae: Nesomyinae) on Madagascar, pp. 283-293. In Lourcnco, W. R., ed., Biogeographie de Madagas- car. Editions de l'ORSTOM. Paris. 588 pp. Goodman, S. M., D. Rakotondravony, G. Schatz. and L. Wilme. 1996b. Species richness of forest-dwelling birds, rodents and insectivores in a planted forest of native trees: A test case from the Ankaratra, Mada- gascar. Ecotropica. 2: 109-120. Goodman, S. M.. D. Rakotondravony. L. E. Olson. E. Razafimahatratra, and V. Soarimalala. 1998. Les insectivores et les rongeurs, pp. 80-93. In Rakoton- dravony, D., and S. M. Goodman, eds.. Inventairc biologique — Forct d'Andranomay. Anjo/.orobe. World Wide Fund for Nature-Centre d' Information et dc Documentation Scientifiquc et Technique. Antanana- rivo, Recherches pour 1c Devcloppmcnt. Serie Sci- ences biologiques. No. 13, 1 10 pp. Goodman, S. M.. and R. Rasoi.oarison. 1997. Les pe- tits mammiferes. pp. 145-157. In Langrand. O., and S. M. Goodman, eds., Inventairc biologique Forct de Vohibasia et d'Isoky-Vohimena. World Wide Fund for Nature-Centre d' Information et dc Documentation Scientifiquc et Technique. Antananarivo. Recherches pour le Development, Serie Sciences biologiques. No. 12. 197 pp. Goodman. S. M.. C. J. Raxworthy, and P. D. Jenkins. 1996c. Inscctivorc ecology in the Reserve Naturcllc Integrate d'Andringitra. Madagascar, pp. 218-230. In Goodman. S. M., ed., A floral and faunal inventory of the eastern slopes of the Reserve Naturcllc Intc- grale d'Andringitra. Madagascar. With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1- 319. GOODMAN ET AL.: LIPOTYPHLA 213 Graham, G. L. 1990. Bats versus birds: Comparison among Peruvian volant vertebrate faunas along an ele- vational gradient. Journal of Biogeography, 17: 657- 668. Grandidier, A. 1870. Description de quelques animaux nouveaux, decouverts a Madagascar, en novembre 1869. Revue et Magasin de Zoologie, 22: 49-54. Grandidier, G. 1902. Observations sur les lemuriens disparus de Madagascar. Collections Alluaud, Gau- bert, Grandidier. Bulletin Museum d'Histoire Natu- relle, Paris, 7: 497-505 and 587-592. . 1903. Description de V Hypogeomys australis, une nouvelle espece de rongeur sub-fossile de Mada- gascar. Bulletin Museum d'Histoire Naturelle, Paris, 9: 13-15. 1928. Description de deux nouveaux mammi- feres insectivores de Madagascar. Bulletin Museum National d'Histoire Naturelle, Paris, ser. 2, 34: 63-70. . 1934. Deux nouveaux mammiferes insecti- vores de Madagascar. Bulletin du Museum National d'Histoire Naturelle, ser. 2, 6: 474-477. Grandidier, G., and G. Petit. 1930. Description d'une espece nouvelle d'Insectivore Malgache, suivie de re- marques critiques sur le genre Oryzorictes. Bulletin du Museum National d'Histoire Naturelle, 2nd series, 2: 498-505. 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. Heim de Balsac, H., and J. Meester. 1977. Order In- sectivora, part 1, pp. 1-29. In Meester, J., and H. W. Setzer, eds., The mammals of Africa: An identification manual. Smithsonian Institution Press, Washington, D.C. Humbert, H. 1955. Les territoires phytogeographiques de Madagascar. Annee Biologique (serie 3), 31: 439- 448. 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. Janzen, D. H., M. Ataroff, M. Farinas, S. Reyes, N. Rincon, A. Soler, P. Soriano, and M. Vera. 1976. Changes in the arthropod community along an ele- vational transect in the Venezuelan Andes. Biotropica, 8: 193-203. Jenkins, P. D. 1988. A new species of Microgale (In- sectivora: Tenrecidae) from northeastern Madagascar. American Museum Novitates, 2910: 1-7. . 1993. A new species of Microgale (Insecti- vora: Tenrecidae) from Eastern Madagascar with an unusual dentition. American Museum Novitates, 3067: 1-11. Jenkins, P. D., S. M. Goodman, and C. J. Raxworthy. 1996. The shrew tenrecs {Microgale) of the Reserve Naturelle Integrale d'Andringitra, Madagascar, pp. 191-217. In Goodman, S. M., ed., A floral and faunal inventory of the eastern slopes of the Reserve Natu- relle Integrale d'Andringitra, Madagascar: With ref- erence to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. Jenkins, P. D., C. J. Raxworthy, and R. Nussbaum. 1997. A new species of Microgale (Insectivora, Ten- recidae), with comments on the status of four other taxa of shrew tenrecs. Bulletin of the Natural History Museum, London (Zoology), 63: 1-12. Langrand, O., and S. M. Goodman. 1997. Inventoire biologique des oiseaux et des micro-mammiferes des zones sommitales de la Reserve Naturelle Integrale d'Andringitra. Akon'ny Ala, 20: 39-54. Leche, W. 1907. Zur Entwicklungsgeschichte des Zahnsystems der Saugetiere zugleich ein Beitrag zur Stammesgeschichte dieser Tiergruppe. Zweiter Teil: Phylogenie. Zweites Heft: Die Familien der Centeti- dae, Solenodontidae und Chrysochloridae. E. Schweizerbartsche, Stuttgart, 157 pp. + 4 pi. LOWRY, P. P., II, G. E. SCHATZ, AND P. B. PHILLIPSON. 1997. The classification of natural and anthropogenic vegetation in Madagascar, pp. 93-123. In Goodman, S. M., and B. D. Patterson, eds., Natural change and human impact in Madagascar. Smithsonian Institution Press, Washington, D.C, 432 pp. 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. 1896. Diagnoses of new mam- mals from Madagascar. Annals and Magazine of Nat- ural History, ser. 6, 18: 318-325. . 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, 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. . 1882. Description d'une nouvelle espece d' insectivore de Madagascar. Le Naturaliste, 2: 55. 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. Nicoll, M. E., and G. B. Rathbun. 1990. African In- sectivora and elephant-shrews: An action-plan for their conservation. International Union for the Con- servation of Nature and Natural Resources, Gland, Switzerland, iv + 53 pp. O'Connor, S., M. Pidgeon, and Z. Randria. 1987. Un programme de conservation pour la Reserve d'Andohahela, pp. 31-36. In Mittermeier, R. A., L. A. 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, Gland, 167 pp. Patterson, B. P., D. F Stotz, S. Solari, J. W Fitzpat- rick, and V Pacheco. In press. Contrasting patterns of elevational zonation for birds and mammals in the Andes of southeastern Peru. Journal of Biogeography. 214 FIELDIANA: ZOOLOGY Pendry, C. A., and J. Proctor. 1996a. Altitudinal zo- nation of rain forest on Bukit Belalong. Brunei: Soils, forest structure and floristics. Journal of Tropical Ecol- ogy, 13: 221-241. . 1996b. The causes of altitudinal zonation of rain forests on Bukit Belalong. Brunei. Journal of Ecology, 84: 407-418. Raharivololona, B. M. 1996. Impact de l'exploitation selective de la foret et la variation saisonniere sur la composition de la population de rongeurs et d'insectivores dans une foret seche de l'ouest de Mad- agascar. Diplome d' Etudes Approfondies, Departe- ment de Paleontologie et d' Anthropologic Biologique, Faculte des Sciences, Universite d'Antananarivo. Rahbek, C. 1997. The relationship among area, eleva- tion, and regional species richness in neotropical birds. The American Naturalist, 149: 875-902. 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. Rickelfs. R. E., and D. Schluter, eds. 1993. Species diversity in ecological communities. University of Chicago Press, Chicago, 414 pp. Rode, P. 1942. Catalogue des types de mammiferes du Museum National d'Histoire Naturelle. Ordre des in- sectivores. Bulletin Museum National d'Histoire Na- turelle, Paris, ser. 2. 14: 307-314. 382-387. Rosenzweig, M. L. 1995. Species diversity in space and time. Cambridge University Press, New York, 436 pp. Schreber. J. C. D. von. 1777. Die Saiigethiere in Ab- bildungen nach der Natur mit Beschreibungen, vol. 3. Leipzig. Stephenson, P. J. 1993. Reproductive biology of the Large-eared Tenrec, Geogale aurita (Insectivora: Ten- recidae). Mammalia. 57: 553-563. . 1994a. Seasonality effects on small mammal trap success in Madagascar. Journal of Tropical Ecol- ogy, 10: 439-444. . 1994b. Notes on the biology of the fossorial tenrec. Orvzorictes hova (Insectivora: Tenrecidae). Mammalia.' 58: 312-315. 1995. Taxonomy of shrew-tenrecs (Microgale spp.) from eastern and central Madagascar. Journal of Zoology, London. 235: 339-350. Swindler, D. R. 1976. Dentition of living primates. Ac- ademic Press. London, 308 pp. Terborgh. J. 1977. Bird species diversity on an Andean elevational gradient. Ecology, 58: 1007-1019. 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, ser. 5, 14: 337-338. . 1918. On the arrangement of the small Ten- recidae hitherto referred to Oryzorictes and Micro- villi'. Annals and Magazine of Natural History, ser. 9. 1: 302-307. . 1926. On some small mammals from Mada- GOODMAN ET AL.: LIPOTYPHLA gascar. Annals and Magazine of Natural History, ser. 9. 17: 250-252. ViETTE, P. 1991. Principales localities ou des Insectes ont ete recueillis a Madagascar. Faune de Madagascar, supplement 2. Private printing. Wright. S. J., and O. Calderon. 1995. Phylogenetic patterns among tropical flowering phenologies. Jour- nal of Ecology. 83: 937-948. Appendix 13-1. Key to the Genera of Lipotyphla Occurring in KM d'Andohahela 1. Buccal morphology of Ml and M2 dilambo- dont (W-shaped) Suncus Upper molars zalambodont (V-shaped) ... 2 2. Pelage spinous; tail very short 3 Pelage soft, lacking spines; tail medium to long 5 3. Dorsal surface covered with a mixture of spines and long, coarse hair Tenrec Close-set, sharp spines cover dorsal surface . . 4 4. Dental formula 2/2 1/1 3/3 3/3 = 36; II small Setifer Dental formula 2/2 1/1 3/3 2/2 = 32; II en- larged Echinops 5. Dental formula 2/2 1/1 3/2 3/3 = 34; II and il much larger than small canines . . Geogale Dental formula 3/3 1/1 3/3 3/3 = 40; canines not reduced in size relative to II and il . . 6 6. Body robust, forelimbs robust, forefeet broad with enlarged, stout claws; C longer than II Oryzorictes Body slender to moderately robust, forelimbs not enlarged, forefeet slender to moderately broad, claws short to moderately lengthened; 1 1 longer or subequal in length to C Microgale Appendix 13-2. Key to the Species of Microgale Occurring in RNI d'Andohahela Measurements (mm) refer to dental adults. 1. Size very small: HB < 65, CIL < 16.9; dark brown dorsal and ventral pelage . . M. parvula Size larger: HB > 65, CIL > 18.6; pelage not dark brown dorsally and ventrally 2 215 2. Ratio of TL: HB > 1.7 3 Ratio of TL: HB < 1.4 4 3. Size smaller: HB < 67, WT < 7.9, CIL < 20.4 M. longicaudata Size larger: HB > 69, WT > 9.5, CIL > 22.0 M. principula 4. Digits and tail tip contrastingly paler than body, tail, and feet M. fotsifotsy Tail tip and digits not obviously paler than rest of body 5 5. Size very large: HB > 95, CIL > 30.0; i2 > c M. dobsoni Size smaller: HB < 96, CIL < 27.5; i2 sub- equal or > c 6 6. Proboscis long, large rhinarium extends pos- terodorsally onto muzzle; forefeet broad, fore- claws enlarged 7 Small rhinarium confined to anterior of short proboscis; forefeet slender without lengthened foreclaws 8 7. Posterior region of rhinarium with transverse striae; BL < 8.0 M. gymnorhyncha Posterior region of rhinarium reticulated; BL > 8.5 M. gracilis 8. 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. Size smaller: HB < 80, WT < 16, CIL < 23.2; p2 not robust M. cowani Size larger: HB > 85, WT > 19, CIL > 26.0; p2 markedly robust M. thomasi 216 FIELDIANA: ZOOLOGY Chapter 14 Rodents of the Reserve Naturelle Integrate d'Andohahela, Madagascar Steven M. Goodman,1 Michael D. Carleton,2 and Mark Pidgeon3 Abstract Between late October and late December 1995 a study was made of the rodents occurring in the three forested parcels that form the Reserve Naturelle Integrale d'Andohahela. In parcel 1, composed of humid forest, five elevational zones between 440 and 1875 m were surveyed. Seven species of rodents belonging to the endemic subfamily Nesomyinae (Eliurus majori, E. minor, E. tanala, E. webbi, Gymnuromys roberti, Monticolomys koopmani, and Nesomys rufus), as well as a member of the introduced subfamily Murinae (Rattus rattus), were collected. In parcel 2, including dry spiny bush (xerophilous) and degraded riverine gallery forest, one Nesomyinae (Eliurus myoxinus) and Rattus rattus were obtained. Rodents trapped in parcel 3, a transitional forest type between humid portions of parcel 1 and dry areas of parcel 2, included Eliurus myoxinus and the two introduced Murinae Rattus norvegicus and R. rattus. The highest diversity of native rodents in parcel 1 was found at 1 200 m in montane forest (five species), and the lowest diversity at 440 m in lowland forest (one species) and at 1875 m in sclerophyllous forest (two species). No species of rodent was found to occur across the complete elevational range of the survey. Eliurus minor and Rattus rattus were found between 810 and 1875 m. One species was restricted to the lowland forest (Eliurus webbi) and another to the sclerophyllous forest (Monticolomys koopmani). In drier forests of parcel 2 and 3 the indigenous rodents showed low levels of diversity and density. Over the 20 km between the western side of parcel 1 and the eastern side of parcel 2 there is a complete turnover in native rodent species. Resume Entre fin Octobre et fin Decembre 1995, une etude a ete entreprise sur les rongeurs se trouvant a l'interieur des trois Parcelles boisees qui forment la Reserve Naturelle Integrale d'Andohahela. Au niveau de la Parcelle 1, composee d'une foret humide, cinq zones d'altitudes comprises entre 440 m et 1875 m ont ete etudiees. On a releve sept especes de rongeurs appartenant a la sous-famille endemique Nesomyinae (Eliurus majori, E. minor, E. tanala, E. webbi, Gymnuromys roberti, Monticolomys koopmani, et Nesomys rufus), ainsi qu'un membre d'une sous-famille introduite Murinae (Rattus rattus). Au niveau de la Parcelle 2. comprenant des broussailles epineuses et seches et une foret galerie humide. une Nesomyinae (Eliurus myoxinus) et Rattus rattus ont ete relevees. Des rongeurs attrapes par des pieges dans la Parcelle 3, ou on trouve un type de foret intermediate entre les parties humides de la Parcelle 1 et les 'Field Museum of Natural History, Roosevelt Road at Lake Shore Drive. Chicago. IL 60605-2496. U.S.A. 2Division of Mammals. National Museum of Natural History. Smithsonian Institution. Washington. D.C. 20560, U.S.A. 'Route de St. Cergue. 1270 Trelex. Switzerland. GOODMAN ET AL. : RODENTS 2 1 7 zones seches de la Parcelle 2, comprennent Eliurus myoxinus et les deux especes introduites Murinae Rattus norvegicus et R. rattus. La diversite la plus importante des rongeurs indigenes dans la Parcelle 1 a ete rencontree a une altitude de 1200 m dans la foret de montagne (cinq especes), et la diversite la plus faible se trouve a une altitude de 400 m dans la foret des zones basses (une espece) et egalement a une altitude de 1875 m dans la foret sclerophylle (2 especes). On n'a trouve aucune espece commune de rongeur a travers la totalite de la region etudiee d' altitudes differentes. On a recontre Eliurus minor et Rattus rattus a des altitudes comprises entre 810 m et 1875 m. Une espece est limitee aux forets des zones de basse altitude {Eliurus webbi) et une autre aux forets sclerophylles (Monticolomys koopmani). Dans les forets plus seches des Parcelles 2 et 3, les rongeurs indigenes enregistrent des faibles niveaux de diversite et de densite. Sur les 20 km entre la limite a l'Ouest de la Parcelle 1 et la limite a l'Est de la Parcelle 2, on remarque une apparition totalement composee d'especes de rongeurs indigenes. Introduction As the number of studies on the small mam- mals of Madagascar has increased over the past decade, taxonomic and distributional knowledge of the island's endemic rodents, subfamily Neso- myinae, has expanded accordingly (Carleton, 1994; Carleton & Goodman, 1996, 1998; Good- man & Carleton, 1996, 1998; Goodman et al., 1996a; Stephenson 1993, 1994, 1995). Several studies have looked at species turnover along ele- vational transects and now information is avail- able from a variety of key sites in eastern humid forest, each set within a different mountain system (Fig. 14-1): Pare National (PN) de la Montagne d'Ambre, at 12.5°S (Raxworthy & Nussbaum, 1994; Goodman et al., 1996a, 1997a); Reserve Speciale (RS) d'Anjanaharibe-Sud, at 14°S (Carleton & Goodman, 1998; Goodman & Carle- ton, 1998); and the Reserve Naturelle Integrate (RNI) d'Andringitra at 22°S (Carleton & Good- man, 1996; Goodman & Carleton, 1996; Lan- grand & Goodman, 1997). Together with data available for other eastern reserves (e.g., PN de Ranomafana and RS d'Analamazaotra), informa- tion on nesomyine rodent diversity now covers much of the eastern humid forest, a long but nar- row biome that once spanned most of Madagas- car's rugged eastern versant. A notable omission to this geographical coverage involves those mountains and forests in extreme southeastern Madagascar. The 1995 biological inventory of the RNI d'Andohahela, a protected area including the southern part of the Anosyenne Mountains, re- dresses this regional void and advances systematic understanding of the native rodent fauna. Much of the striking habitat diversity of Mad- agascar is concentrated in the island's southeast- ern corner, where an abrupt ecotone demarcates wet and dry environments (Goodman et al., 1997b). The ecological contrast is largely influ- enced by the north-south alignment of the Ano- syenne Mountains, which act as a rain barrier for weather systems moving in an east to west direc- tion from the Indian Ocean (Battistini, 1964). Di- minished precipitation associated with this rain shadow dramatically affects floristic structure and composition across this narrow zone, with changes evident over a distance of just a few ki- lometers. Thus, on the eastern or windward slopes of the Anosyenne Mountains, Madagascar's east- ern humid forests reach their southern limits; al- though south of the Tropic of Capricorn, the hu- mid forests are typically tropical in structure and species composition (White, 1983; see also Chap- ters 2 and 4 herein). From a few exposed ridges just kilometers to the west, on the leeward side of this range, one can view dry forest (spiny bush) with its characteristic baobab (Adansonia) trees and thick stands of cactus-like Didiereaceae. The nearby coastal plain retains stands of evergreen littoral forest, a once widespread woodland for- mation; such remnants have low canopies relative to lowland humid forest and rest on sandy, organ- ically poor soils. The forests of southeastern Mad- agascar become progressively drier from north to south and abruptly so from east to west, changing to deciduous thicket and bush. The RNI d'Andohahela is composed of three disjunct parcels that are separated by a minimum distance of 20 km (Fig. 14-2). Each parcel con- tains one or more distinct forest types: parcel 1 — eastern humid forest, including lowland forest to high mountain sclerophyllous forest; parcel 2 — spiny forest, with remnants of riverine gallery for- est; and parcel 3 — transitional forest, containing a 218 FIELDIANA: ZOOLOGY Fig. 14-1. Topography of Madagascar, illustrating places mentioned in the text, most of which have been surveyed for rodents. Note the relative isolation of the northern highlands, especially Montagne d'Ambre, the extensive area occupied by the Central High Plateau, and the extreme position of the RNI d*Andohahela near the terminus of the southern highlands. imixture of elements from humid and dry forests (see Chapters 2 and 4 for detailed floristic descrip- tions of parcels 1 and 2). During the 1995 inven- tory an elevational transect was conducted at five stations (440-1875 m) in parcel 1, and single sites were sampled in parcels 2 (120 m) and 3 (170 m). This study presents the results of those surveys, documents the rodent diversity of the RNI d'Andohahela, and summarizes attendant natural history and biogeographic information. In addi- tion, data on the nesomyine rodent community within the RNI d'Andohahela are integrated with those now available from comparable inventories, which collectively embrace a broad expanse throughout Madagascar's great eastern forest from approximately 12.5° to 24.5° south latitude. Previous Work in the Region Between 1929 and 1931, the Mission Zoolo- gique Franco-Anglo-Americaine (MZFAA) con- GOODMAN ET AL.: RODENTS 219 KEY ■ town or village * peak ♦ forest Marovony ♦ Soavary Analalava* Manantenina- Anosyenne Mountains _•«*> & \ Vohibaka jEsomony ■ ^-^^--1 i I Pic TrafonaombyA 24° 30'- ♦Marosoh\ Ivohibe Eminiminv -^- ^Ranomafana \// Bemangidy Tranomaro V 7 N /Pic Andohahela * / Col de / _^J_X-a/^ / /^Vohimena / Tanatana^- J \l / I Mountains Hazofotsy /) \\ / ♦ Manafiafy J Isaka-Ivondro parcel 1 Mandrare/rparcei24 River — ' ' f Behara L Berentyl .parcel 3 Ankapoky n/ Manantantehy Bevilany Amboasary Ranopiso Manambarc Lac Anony ' 47° 00' I Lac Erombo Petriky Andrahomana Cave + Itapera Mandena Nahampoana *^ Tolagnaro ^w 25° 00'- INV 20 30 _1 1_ I** 0 0' 40 I V 50 km _1 Fig. 14-2. Southeastern Madagascar, illustrating the three parcels of the RNI d'Andohahela, mountain ranges, and other local place names and geographical features referenced in the text. Areas shaded in gray denote elevations above 700 m and emphasize the principally north-south orientation of the Anosyenne and Vohimena mountains. ducted the most ambitious geographical survey of Malagasy birds and mammals to date, but the itin- erary of this group of scientists did not include the extreme southeastern corner of the island. One field team visited a forest 20 km west of Vondrozo (Rand, 1936), the MZFAA site closest to the RNI d'Andohahela but a distant 200 km north of parcel 1, and coastal plain near Manombo, just to the south of Farafangana. Rodents collected at these places include Eliurus minor, E. webbi, Gymnu- romys roberti, and Nesomys audeberti, vouchers of which are preserved in the American Museum of Natural History (AMNH; New York), the Brit- 220 ish Museum of Natural History (BMNH; London; now The Natural History Museum), or the Mu- seum National d'Histoire Naturelle (MNHN; Par- is). In 1944, Cecil S. Webb, a resourceful naturalist and collector sponsored by the BMNH, visited a few areas to the west of Tolagnaro, where he ob- tained small mammals, particularly rodents (Carleton & Schmidt, 1990). He worked two lo- calities in the vicinity of the RNI d'Andohahela: 5 mi (8.0 km) E of Bevilany, which is just south- west of parcel 3; and 7 mi (11.3 km) NE of Lac Anony, near the coast and farther to the southwest FIELDIANA: ZOOLOGY of the reserve (Fig. 14-2). The rodents collected consist of Eliurus myoxinus and Macrotarsomys bastardi, specimens of which are deposited in BMNH. As part of a study of mammals and their role as disease vectors, Harry Hoogstraal visited sev- eral places in southeastern Madagascar in Decem- ber 1948 (Hoogstraal 1953; Uilenberg et al., 1979). His stops included Mandena and the forest of Bemangidy (approximately 72 km north of To- lagnaro), along the eastern slope of the Vohimena Mountains (Fig. 14-2). Hoogstraal made a small general collection of open country and forest- dwelling mammals that is housed in the Field Mu- seum of Natural History (FMNH; Chicago) and the National Museum of Natural History (USNM; Washington, D.C.); among them is a single spec- imen of Nesomys audeberti taken in original for- est near Bemangidy. On the basis of two recon- naissance trips in 1989 and 1990, S.M.G. found that the area still contains relatively undisturbed humid forest, particularly on the slopes below Pic Ivohibe. In 1989 and 1990, a field team was assembled to assess the potential biological impact of a pro- posed mining project in southeastern Madagascar. This faunal inventory is referred to as the QIT- FER project, and G. Ken Creighton coordinated the small mammal studies in collaboration with E. Raholimavo, D. Rakotondravony, and J. Ryan. Numerous localities, containing a variety of hab- itat types, were visited, but emphasis was devoted to the humid littoral forests of Mandena and Man- afiafy, north of Tolagnaro, and to the dry littoral forest of Petriky, west of Tolagnaro (Fig. 14-2). These three sites are adjacent to or part of the proposed mining sites, but the QIT-FER team sur- veyed other places in their vicinity in order to provide a regional faunal perspective. Two low- land humid forests in the Vohimena Mountains and near Tolagnaro were surveyed; the elevational sampling at both of these sites ranged from 100 to 450 m, including the Manantantely Forest at the southern end of the chain and the Naham- poana Forest on the eastern slopes. To augment information on elevational variation, the humid forest of Marosohy, along the northeastern bound- ary trail of parcel 1 of the RNI d' Andohahela, was studied at 425 and 725 m. Finally, they visited the spiny bush forest of Ankapoky, just south of par- cel 2 and west of parcel 3 of the RNI d' Andohahela; this site is very close to Webb's old locality at 5 mi E of Bevilany (Fig. 14-2). Specimens collected during this survey are depos- ited in the Departement de Biologie Animale, Universite d'Antananarivo, Antananarivo (UAD- BA), and the USNM. We have examined the nesomyine rodents col- lected during these several field studies and have generally referenced them to augment the taxo- nomic and distributional context of material pro- cured during the 1995 survey of the RNI d' Andohahela. Materials and Methods This study is based on fieldwork conducted be- tween 19 October and 29 December 1995 by S.M.G. and M.P. Taxonomic determinations were verified by M.D.C., who also undertook system- atic comparisons. Field Methods and Trapping Protocol The general field protocols follow those previ- ously outlined in parallel reports on the rodents of the RNI d'Andringitra (Goodman & Carleton, 1996) and the RS d'Anjanaharibe-Sud (Goodman & Carleton, 1998). Information is presented on the rodent faunas of all three parcels within the RNI d'Andohahela. Five altitudinal zones (440, 810, 1200, 1500, and 1875 m) were studied within the humid forest of the reserve (parcel 1). The areas surveyed at 440 and 810 m showed signs of both ancient and recent human disturbance, whereas the other three sites were in seemingly undisturbed habitat. Immediately after termination of the elevational transect in parcel 1 , the survey group moved to the spiny bush area of the reserve (parcel 2) and trapped one site in disturbed habitat at 120 m. For all sites visited in parcels 1 and 2, trap lines were maintained for a minimum of 7 nights (Table 14-1). Each trap line, numbered se- quentially starting with the 440 m zone, consisted of Sherman live traps (9 X 3.5 X 3 in.) and Na- tional live traps (16 X 5 x 5 in.), at a ratio of 4: 1. Traps were baited daily, generally between 1500 and 1700 hr, with finely ground peanut but- ter; lines were visited and traps inspected at least twice daily, once at dawn and again in late after- noon. At each of these sites, sampling was also conducted with pitfall traps, but this technique yielded few rodents (see Chapter 13). In late De- cember 1995, M.P. visited the transitional forest of parcel 3 and trapped for 5 nights at 170 m. GOODMAN ET AL.: RODENTS 221 Table 14-1. Summary of trap lines in the RNI d'Andohahela.* Elevation No. of traps Length (m) of line Mean distance (m) between traps Mean height (m) above ground Parcel 1 440 m (20-27 Oct) Line 1 Line 2 85 50 894 490 7.7 9.5 6.42(1-38) 3.50(3-21) 1.4 1.5 + + 0.65 (0.2-2.5), n = 30 (35%) 0.76 (0.3-3.0), n = 17 (34%) 810 m (29 Oct-5 Nov) Line 3 Line 4 50 75 548 510 10.3 7.5 -+■ + 3.75(1-22) 4.66(0.5-21) 1.8 1.4 + + 0.73(0.2-3), n = 21 (42%) 0.59 (0.3-3), n = 28 (37%) 1200 m (8-16 Nov) Line 5 Line 6 50 75 373 364 7.6 4.9 -f- 2.81 (4-17) 3.11 (1-18) 1.8 1.6 + 0.82 (0.2-3), n = 23 (46%) 0.46 (0.5-3), n = 27 (36%) 1500 m (17-26 Nov) Line 7 Line 8 50 75 375 454 7.5 6.1 + 3.51 (0.5-17) 3.63(1-17) 1.8 2.0 4- ■+- 1.13(0.1-4), n = 18 (36%) 0.46(1.5-3), n - 17 (23%) 1875 m (27 Nov-4 Dec) Line 9 Line 10 50 75 255 415 6.3 5.6 -t- -t- 3.12(0.5-14) 3.48(1-13) 1.9 1.7 + + 0.53(1-2.8), n - 31 (62%) 0.62 (0.5-3), n = 28 (37%) Parcel 2 120 m (8-14 Dec) Line 1 1 Line 12 50 75 553 760 11.1 10.1 + 5.18(2-21) 4.31 (1-19) 1.9 1.8 ■+- ■+■ 0.47 (0.2-2.4), n = 22 (44%) 0.62 (0.5-2.5), n = 27 (36%) * Each line consisted of National and Sherman live traps in a ratio of 4:1 (see page 221). Descriptive statistics are presented as the mean ± standard deviation, with the range in parentheses. This visit was intended only to assess what ro- dents commonly occurred in this forest type, not to comprehensively document species diversity as for other sites visited within this reserve. A trap-night is defined as one live trap in use for a 24 hr period (dawn to dawn). The total num- ber of trap-nights accrued in each elevation varied slightly; consequently, the first 500 trap-nights in an elevational zone are considered the "standard- ized" trapping regimen in order to facilitate com- parisons among the sites sampled. Standing bio- mass of a species is based on the total catch of individuals during a standardized trapping regi- men multiplied by the average body weight of the species (Table 14-2). We depended exclusively on live trap techniques during this inventory for rea- sons explained previously (Goodman & Carleton, 1998, p. 201). To quantify differences in the spatial distribu- tion of small mammal captures, several trapping variables were systematically recorded for each trap installed: (1) type of trap; (2) total length of trap line; (3) distance between traps; and (4) spe- cific placement of trap, including its substrate, surrounding forest structure, and position on or height above the ground. Categorization of micro- habitat was simplified from a more detailed sys- tem used earlier (Goodman & Carleton, 1996), as follows: On Ground — ( 1 ) In leaf litter, generally in area of open understory; (2) under decomposed downed trees or woody vegetation; (3) by tree root or trunk, with or without cavity or hole; or (4) miscellaneous, including placement under ex- posed rocks or boulders, at base of rock face, at entrance of hole in ground, in thick herbaceous vegetation, or on moss-covered rocks. Above Ground — (1') On liana, limb, or trunk of <10 cm diameter in horizontal to vertical po- sition; (2') on liana, limb, or trunk of >10 cm diameter in horizontal to vertical position; (3') on limbs or trunks suspended by lianas; or (4') mis- cellaneous, including placement on bamboo stalks, in small cavities at junctions of tree limbs, or on large moss-covered rocks. Specimens and Measurements Captured animals were prepared as standard museum skins with associated skulls and partial skeletons, as fluid-preserved carcasses, or as full skeletons. Whole carcasses were wrapped in fine cheesecloth before immersion in formalin to pre- 222 FIELDIANA: ZOOLOGY Table 14-2. External measurements and sample statistics for adult rodents collected in the RNI d'Andohahela. Species TOTL HBL TL HFL EL WT Rattus norvegicus 322 157 154 32 23 96 Rattus rattus 393.5 168.3 206.6 34.1 25.1 144.6 10.3 8.5 8.6 2.0 1.4 7.4 370-405 160-181 190-216 32-39 23-28 132-155 (n = 10) (n = 8) (n = 10) (n = 10) (n - 10) (n = 10) Eliurus rnajori 354, 358 155.2 189. 190 30.2 22.4 99.6 8.5 1.6 2.4 11.5 145-168 29-33 20-25 86.5-113 (n = 5) (n = 5) (n = 5) (n = 5) Eliurus minor 239.8 106.5 128.0 21.8 18.8 38.5 7.2 4.1 5.5 0.7 1.4 4.4 230-254 99-113 115-137 21-23 15-21 31.0-47.0 (n = 15) (n = 22) (n = 15) (n = 22) (n = 22) (n = 22) Eliurus myoxinus 283.7 130.0 147.7 27.7 22.7 65.8 37.8 14.8 23.1 1.5 0.6 19.7 251-325 120-147 126-172 26-29 22-23 48.0-87.0 (n = 3) (n = 3) (n = 3) (n = 3) (n = 3) (n = 3) Eliurus tanala 339.5 149.3 181.7 30.5 23.6 94.3 14.8 5.5 11.1 1.4 1.3 13.7 303-362 131-159 152-197 28-34 21-25 76.5-134 (n = 15) (n = 18) (n = 16) (n = 20) (n = 19) (n - 20) Eliurus webbi 331.4 146.1 175.0 30.2 23.3 88.3 12.7 5.3 11.7 0.9 1.2 6.6 309-347 134-154 154-191 29-32 21-25 77.0-99.5 (n = 12) (n = 18) (n = 12) (n = 18) (n = 17) (n = 18) Gymnuromys roberti 360 157 184 37 23 120 Monticolomys koopmani 227.4 86.2 134.4 23.6 18.8 22.6 7.2 1.9 5.9 0.6 0.8 3.8 217-236 84-89 127-142 23-24 1 8-20 18.5-27.5 (n = 5) (n = 5) (n = 5) (n = 5) (n = 5) (n = 5) Nesomys rufus 364.0 184.0 169.7 44.3 26.3 182.0 16.5 7.5 11.0 1.0 1.0 12.7 345-375 177-191 159-181 43-45 25-27 164-191 (n = 3) (n = 4) (n = 3) (n = 4) (n = 4) (n = 4) Abbreviations are explained in the Materials and Methods section. The sample statistics arc given as the mean, standard deviation, and range, with the number of animals in parentheses. vent loss or mixing of ectoparasites between their specific hosts. A large proportion of the captured rodents during our work in the RNI d'Andohahela were prepared as vouchers. This material is housed in the FMNH, and a representative series has been returned to the UADBA. Specimens de- posited immediately after the survey in the latter institution have not yet been catalogued and are individually referenced by the collector's field numbers (UA-SMG or UA-MP). To confirm tax- onomic identifications, nesomyine holdings in other museums (see Appendix 22-1, p. 283, in Goodman & Carleton, 1996) were also consulted, including the holotypes of all described forms of Nesomyinae except 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. Measurement abbreviations and definitions 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) GOODMAN ET AL.: RODENTS 223 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 distal tip of the pinna WT (weight): measured with Pesola spring scales, to ±0.5 g for animals <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 previously (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 auditory 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 rostrum ONL, occiptonasal length PPB, posterior breadth of the bony palate PPL, postpalatal length WM 1 , width of the first upper molar 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 possess fully erupt- ed, though sometimes unworn, third molars. Where sample sizes permitted, two-sample Mests and one-way analyses of variance were applied to the mensural variables, with sex as the categorical variable. Analytical routines were carried out us- ing Systat (version 6.01, 1996). The mammae for- mula is presented as the number of paired post- axial, abdominal, or inguinal teats. To assess faunal similarities among geographic regions, we used the Jaccard Index, a matching coefficient that scales similarity to range from 0 to 1: Jaccard Index = Nc N, + N2 - Nc' (1) where N, = the number of species at site 1 (the smaller fauna), N2 = the number of species at site 2, and Nc = the number of species common to both sites. The indices from these pairwise com- parisons were clustered using UPGMA as imple- mented by Systat. Accounts of Species Information on the natural history and eleva- tional range of each rodent species captured dur- ing the survey of the RNI d'Andohahela is pre- sented under the subheadings Distribution, Ecol- ogy and Reproduction, Comments (when nec- essary), and Specimens Examined. The last subheading includes only material collected in the reserve during the 1995 mission. More detailed criteria for species identification and discussion of alpha-level taxonomic problems are presented in Carleton (1994), Carleton and Goodman (1996, 1998), and Goodman and Carleton (1996, 1998). External measurements and masses are given here for the rodent species captured during the 1995 survey to aid researchers in identification of Mal- agasy rodents and to provide baseline data for some of the analyses presented in the Discussion section. Family Muridae: Subfamily Murinae Rattus norvegicus (Berkenhout, 1769) Distribution — This introduced species was trapped only within parcel 3. A single individual taken near the main road between Tolagnaro and Amboasary-Sud was obtained in a line that yield- ed several specimens of Rattus rattus and Eliurus myoxinus. In general, R. norvegicus is a commen- sal species that is particularly common in urban centers and rural communities, and notably un- common in agricultural fields and at the forest edge (Malzy, 1964; Rakotondravony, 1992). Ecology and Reproduction — The single in- dividual, an adult female with small mammae and a perforated vagina, entered a trap set on a 5 -cm- diameter branch (2.2 m above ground) of a large emergent tree. 224 FIELDIANA: ZOOLOGY Table 14-3. Trap success for Rattus rattus in the humid forest of parcel 1 of the RNI d'Andohahela. Capture rate Elevation Number Number of per 100 (m) captured trap-nights trap-nights 440 0 845 0.0 810 4 850 0.47 1200 0 775 0.0 1500 5 1,050 0.48 1875 2 875 0.23 Specimens Examined — Parcel 3, 6 km SE of Bevilany, 25°01.3'S, 46°38.8'E, 170 m (FMNH 156536). Rattus rattus (Linnaeus, 1758) Distribution — Rattus rattus occurs in all three parcels of the RNI d'Andohahela. This introduced species is known from a variety of habitats on the island, including pristine forest (Stephenson, 1993; Goodman, 1995; Goodman & Carleton, 1996, 1998). Ecology and Reproduction — Previous work along elevational transects in the eastern humid forest indicated that Rattus rattus is more com- mon at mid-elevations than in lowland forest or near the summit (Goodman et al., 1997a; Good- man & Carleton, 1996, 1998). This pattern did not hold within the RNI d'Andohahela, where no marked change was recorded in the density of this species when present, as measured by trap suc- cess, along the elevational gradient (Table 14-3). During the period of our field study, R. rattus was encountered less frequently in the RNI d'Andohahela than on other mountains with sim- ilar botanical communities and altitudinal strat- ification. For example, at 1350 m on Montagne id'Ambre, R. rattus was ubiquitous, totaling 85% iof all trap captures (Goodman et al., 1997a); in ithe same highland zone of the RNI d'Andohahela, | none were captured at 1200 m, and only five in- jdividuals were trapped in 1,050 trap-nights at 1 1500 m (Table 14-3). Like its altitudinal pattern on those other mountains, however, R. rattus was found to occur in the deep forest of parcel 1 but iwas not trapped at the lowest elevational zone I closest to the forest edge. Populations of the spe- cies appear to be most abundant in the zone be- tween 1500 and 1625 m, as judged from trap cap- tures in the RNI d'Andohahela, RNI d'Andringitra, and RS d'Anjanaharibe-Sud (Table 14-3; Good- man & Carleton, 1996. 1998). In the dry deciduous Kirindy Forest, near Mo- rondava, R. rattus is rare in the large relatively intact forest block (Ganzhorn et al., 1996). This rat is common in small satellite forests, however, and its presence is significantly and negatively as- sociated with the occurrence of the nesomyines Eliurus myoxinus and Macrotarsomys bastardi. This explanation fits our observations of the dry forests sampled in the RNI d'Andohahela. In par- cel 2 the single R. rattus taken in 1,023 trap- nights inhabited disturbed gallery forest along a small river, and in parcel 3 all individuals cap- tured were in close proximity to a stream. Earlier trapping (144 trap-nights using Sherman traps baited with banana) in parcel 3 conducted in late January 1990 yielded three E. myoxinus and no Rattus (Pidgeon, unpubl. data). Of the 1 1 R. rattus captured within parcel 1 and for which there is information on trap placement, all except one were taken on the ground (Table 14-4). The majority of trap sets were placed near streambeds, rocky outcrops, or at the base of large trees with buttressed roots. The single arboreal capture involved a trap placed on an 8-cm-diam- eter tree trunk (2.5 m above ground) leaning at 30° into a liana tangle. Twelve specimens of R. rattus were examined for reproductive condition: six males with scrotal testes, two males with abdominal testes, three adult females with mammae enlarged or actively lactating, and one subadult female. The mammae formulae varied from 1-1-2 (n = 1) to 1-2-2 (n = 2). Comments — Rattus rattus constitutes a sub- stantial portion of prey taken by Asia madagas- cahensis in southeastern Madagascar. At a roost in the Nahampoana Forest, within the ecotone be- tween lowland humid forest and an agricultural area, about 40% of the prey consumed by this owl is R. rattus (Goodman et al., 1993). Southeastern Madagascar has been settled and occupied by humans since at least the 9th century (Wright & Rakotoarisoa, 1997). Tolagnaro (Fort Dauphin) has been a major port of international shipping since the 16th century, allowing more than 400 years for colonization of the region by seagoing Rattus. Etienne de Flacourt, a represen- tative of the French Compagnie des Indes Orien- tates, based at Fort Dauphin in the latter half of the 17th century, noted (1658) that rats and mice were common everywhere and caused consider- able destruction of grain in houses and agricul- tural fields. Areas in and around parcel 1 of the RNI d'Andohahela have witnessed human pres- GOODMAN ET AL.: RODENTS 225 Table 14-4. Microhabitat occurrences of rodent species by elevation in parcel 1 of the RNI d'Andohahela. Ground location* Elevation and species Trap position No. On Above taken ground ground Leaf litter Under By rotten roots, wood trunks Misc. Aboveground location Vine, limb, or Limbs, Sus- trunk trunks pended <10 cm >10 cm trunks Misc. 440 m Trap distribution 88 47 20 15 26 27 25 15 3 4 Eliurus webbi 13 1 5 6 1 810 m Trap distribution 76 49 21 5 17 33 26 15 1 7 Rattus rattus Eliurus minor Eliurus tanala Eliurus webbi Nesomys rufus 4 2 8 6 1 1 1 1 1 3 2 1 1 2 3 2 3 1200 m Trap distribution 74 51 17 4 31 22 21 20 5 5 Eliurus majori Eliurus minor Eliurus tanala Gymnuromys roberti Nesomys rufus 2 9 5 1 2 1 1 3 1 1 1 2 3 2 1 3 1500 m Trap distribution 81 44 8 17 30 26 13 25 0 6 Rattus rattus Eliurus majori Eliurus minor Eliurus tanala 5 4 7 11 2 1 1 1 2 3 1 5 2 3 1 3 2 1875 m Trap distribution 64 61 15 3 19 27 17 41 2 1 Rattus rattus Eliurus minor Monticolomys koopmani 2 3 5 4 1 2 1 1 1 Totals: 440-1875 m Trap distribution 383 252 81 44 123 135 102 116 11 23 Rattus rattus Eliurus majori Eliurus minor Eliurus tanala Eliurus webbi Gymnuromys roberti Monticolomys koopmani Nesomys rufus 11 6 21 26 19 1 5 3 10 2 5 15 4 1 5 3 1 4 16 11 15 2 2 1 4 1 1 1 1 1 3 5 1 7 2 1 7 2 1 1 1 1 2 7 4 5 1 6 7 9 1 3 1 Total captured 92 45 47 10 3 10 22 19 23 4 1 * See p. 222 for habitat definitions. ence since the 15th or 16th centuries (Razanaba- hiny, 1995; Goodman & Rakotoarisoa, 1998; Rakotoarisoa, 1998). Thus, even given the long- term presence of Rattus in the region, the species has not yet exploited humid forest habitats en masse, as it has some other areas of the island (Goodman et al., 1997a). The factors influencing this colonizing variation are unknown and warrant detailed examination, particularly in view of the likelihood that R. rattus may be displacing certain endemic rodents in some regions (Goodman, 1995; Ganzhorn et al., 1996). 226 FIELDIANA: ZOOLOGY Specimens Examined — Parcel 1. 12.5 km NW of Eminiminy, 24°35.6'S, 46°44.3'E, 810 m (FMNH 156537; UA-MP 21; UA-SMG 7475, 7483); parcel 1, 15.0 km NW of Eminiminy, 24°34.2'S, 46°43.9'E, 1500 m (FMNH 156538- 156540, 156544, 156582); parcel 1, 20.0 km SE of Andranondambo, 24°33.7'S, 46°43.3'E, 1875 m (FMNH 156541); parcel 2, 7.5 km ENE of Ha- zofotsy, 24°49.0'S, 46°36.6'E, 120 m (FMNH 156543); parcel 3, 6 km SE of Bevilany, 25°01.3'S, 46°38.8'E, 170 m (FMNH 156545- 156548). Family Muridae: Subfamily Nesomyinae Eliurus majori Thomas, 1895 Distribution — Specimens from the RNI d'Andohahela extend the range of Eliurus majori 260 km further south of its previously known lim- it in the RNI d'Andringitra (Carleton, 1994; Goodman & Carleton. 1996). This rodent is now known to occur in wet forest on numerous moun- tains along the length of the island, from PN de la Montagne d'Ambre in the north, through RS d'Anjanaharibe-Sud, Anjozorobe, Ambohimitam- bo, and RNI d'Andringitra, to RNI d'Andohahela in the south (Fig. 14-1; Carleton, 1994; Goodman & Carleton, 1996, 1998; Goodman et al., 1996a; Goodman et al., 1998). Its occurrence in the An- osyenne Mountains, at 1200 and 1500 m of parcel 1, conforms to the general altitudinal setting doc- umented elsewhere — a belt of middle to upper montane forest from 1000 to 2000 m (Goodman & Carleton. 1996. 1998). Ecology and Reproduction — Four of five E. majori were captured in a variety of arboreal placements (Table 14-4). a result similar to that reported for the species in the RNI d'Andringitra and RS d'Anjanaharibe-Sud. In the RNI d'Andohahela, successful trap sets included rela- tively thin branches and lianas of <10 cm diam- eter, a large branch of 25 cm diameter, and a sec- tion of a fallen tree trunk suspended in a vine tangle. Two animals were obtained in traps placed on the ground, both at the base of large boulders or rocky outcrops that sheltered holes and hol- lows. Of the five individuals captured, three are adult males with large scrotal testes and convoluted ep- ididymides, and two are adult females with prom- inent mammae. One female has three placental scars. The mammae total six (n = 2), distributed GOODMAN ET AL.: RODENTS as one pair postaxial, one pair abdominal, and one pair inguinal, a formula consistent with individu- als from the RS d'Anjanaharibe-Sud and with the genus (Carleton, 1994; Goodman & Carleton, 1998). Specimens Examined — Parcel 1, 13.5 km NW of Eminiminy, 24°35.0'S, 46°44.1'E, 1200 m (FMNH 156503, 156615); parcel 1, 15.0 km NW of Eminiminy, 24°34.2'S, 46°43.9'E, 1500 m (FMNH 156616, 156617. 156658). Eliurus minor Major, 1896a Distribution — The 1995 inventory found Eli- urus minor to occur broadly in forest between 810 and 1875 m (Table 14-5). This elevational range resembles the distribution in the RNI d'Andringitra, where the species was recorded from 720 to 1625 m (Goodman & Carleton, 1996). Carleton (1994) reported the presence of E. minor at 20 km west of Vondrozo (500 m), a MZFAA locality visited in 1929 (Rand, 1932). More recently, the species was collected by the QIT team in the Marosohy Forest between 350 and 450 m, a place near the northeastern bound- ary trail of parcel 1, RNI d'Andohahela. Eliurus minor is now known throughout the eastern hu- mid forest, from Montagne d'Ambre in the north to the RNI d'Andohahela in the south (Carleton, 1994; Goodman et al., 1996a). Ecology and Reproduction — Of 21 E. minor obtained in traps, 16 (76%) were taken in arboreal sets, of which seven were on vines and branches of <10 cm diameter (Table 14-4). The five E. mi- nor caught in ground traps were in a variety of situations, such as next to tree root buttresses, at the opening of a tunnel system in the soil adjacent to a dead and hollow tree, at the base of rotten and fallen tree trunks, and in relatively open un- derstory dominated by a herbaceous growth of Acanthaceae. Furthermore, two E. minor were ob- tained in pitfall traps (see Chapter 13). On the basis of trap captures along small branches and vines as well as terrestrial sets, this species ap- pears to be predominantly scansorial. Most individuals of E. minor trapped in the 810. 1200. and 1500 m zones were adults (Table 14-6), and most showed signs of active breeding. Counts of embryos and placental scars of four fe- males uniformly disclosed a litter size of three. No apparent synchrony was noted in the repro- ductive cycles of individuals captured within an elevational zone. Mammae counts for the nine fe- 227 Table 14-5. Elevational occurrence (m) of rodents within the three parcels of the RNI d'Andohahela based on the 1995 inventory. 1* 2* 120 3* Species 440 810 1200 1500 1875 170 Murinae Rattus norvegicus Rattus rattus Nesomyinae Eliurus majori Eliurus minor Eliurus myoxinus Eliurus tanala Eliurus webbi Gymnuromys roberti Monticolomys koopmani Nesomys rufus Total species Native species + + + + + + + + + + + + + + + + 5 5 4 3 2 3 4 5 3 2 1 1 * Parcel 1, humid forest ranging from lowland to sclerophyllous forest; parcel 2, spiny bush; parcel 3, transitional between humid forest and spiny bush. The floristic composition of these forest types is detailed in Chapter 2. males for which information is available varied from 1-1-1 (n = 2) to 1-0-2 (n = 6) and 0-2-1 (n = 1). Comments — Remains of Eliurus minor have been identified in regurgitated pellets from Asio madagascariensis within the Nahampoana Forest north of Tolagnaro (Goodman et al., 1993). The site is at near the edge of lowland forest and en- croaching agricultural fields. Specimens Examined — Parcel 1, 12.5 km NW of Eminiminy, 24°35.6'S, 46°44.3'E, 810 m (FMNH 156618-156620); parcel 1, 13.5 km NW of Eminiminy, 24°35.0'S, 46°44.1'E, 1200 m (FMNH 156504, 156505, 156621-156624; UA-MP 23; UA-SMG 7523, 7551); parcel 1, 15.0 km NW of Eminiminy, 24°34.2'S, 46°43.9'E, 1500 m (FMNH 156506-156510, 156626, 156629, 156533; UA-MP 32); parcel 1, 20.0 km SE of Andranondambo, 24°33.7'S, 46°43.3'E, 1875 m (FMNH 156625, 156627, 156628). Eliurus myoxinus Milne Edwards, 1885 Distribution — Eliurus myoxinus is a wide- spread inhabitant of spiny bush and deciduous forest of western and southern Madagascar (Carleton, 1994; Goodman & Ganzhorn, 1994; Ganzhorn et al., 1996; Goodman & Rasoloarison, 1997). In the RNI d'Andohahela, this species is limited to the spiny bush forest of parcel 2 and transitional forest of parcel 3. Webb had earlier collected E. myoxinus from the hills east of Be- vilany, within 5-6 km of the southern boundary of parcel 2 (Carleton, 1994), and Creighton more recently obtained it in the Petriky Forest, 5-7 km SE of Manambaro, the easternmost limit so far known for the species. Ecology and Reproduction — All three E. myoxinus obtained during the 1995 mission were caught in arboreal trap sets, both on trunks and branches of >10 cm diameter (two specimens) and <10 cm diameter (one). Our findings under- score Webb's (1954) perception that the species is arboreal. In parcel 3 one animal was taken on a horizontal branch of a fruiting Apocynaceae tree next to a Dypsis decaryi (Arecaceae) palm tree endemic to parcel 3 and its immediate vicinity. It has been proposed that Eliurus consumes the fruits of this latter tree (J. Ratsirarson, pers. comm.). With just three specimens captured, little can be said about the reproductive season of E. myox- inus. Two of the animals trapped are females, one with large and the other with small mammae; the third individual is an adult male with scrotal tes- tes. The mammae formula is typical of Eliurus. Specimens Examined — Parcel 2, 7.5 km ENE of Hazofotsy, 24°49.0'S, 46°36.6'E, 120 m (FMNH 156630); parcel 3, 6 km SE of Bevilany, 25°01.3'S, 46°38.8'E, 170 m (FMNH 156511, 156312). 228 FIELDIANA: ZOOLOGY c S-o. f- O vC C". i/~, Q O Q ri ^? f^ \5 — — ^y ^ o ?5 — o — — fN ^ ^ ^t I — C) c-. -o y: r M •- E.S 11 o S3 T3 '> Eliurus ta tin la Major, 1896a Distribution — This species is widespread in the eastern humid forest, previously reported from the forest of Didy, near Lac Alaotra, as far south as the Vinanitelo region (Carleton, 1994). Eliurus tanala actually occurs further north than Didy; it was recently obtained between 875 and 1260 m in the RS d'Anjanaharibe-Sud (Goodman & Carleton, 1998). In the RNI d'Andohahela, E. tan- ala was recovered from pristine forest in parcel 1, between 810 and 1500 m (Table 14-5), but was not recorded from the dry lowland formations in parcels 2 and 3. The species was also collected during the QIT-FER studies in the Marosohy For- est (800 m) and Manantantely Forest (100-450 m). Together these localities in the Anosyenne Mountains and Vohimena Mountains extend the southern distributional limit of the species. This limit appears to coincide with the southernmost projection of middle montane forest in these high- lands. Ecology and Reproduction — Eleven of 26 E. tanala obtained in trap lines involved sets above the ground, and 15 were on the ground (Table 14- 4). About one-third of the arboreal sets were on lianas and branches of <10 cm diameter, whereas the remainder were on substrates with diameters of >10 cm. Terrestrial stations included a variety of microhabitats, such as under boulders, by fallen and rotten logs, in front of openings between ex- posed roots and cavities at tree bases, and in open understory forest with either a dense growth of Acanthaceae ground cover or thick leaf litter. The nearly even proportion of arboreal versus ground captures contrasts with trapping results in the RS d'Anjanaharibe-Sud, where only one of six E. tanala was captured in an arboreal setting (Good- man & Carleton, 1998). As in the RS d'Anjanaharibe-Sud, the distri- bution of this species along trap lines appeared to be clumped (Goodman & Carleton, 1998). For ex- ample, in the 1500 m zone of the RNI d'Andohahela, five of eight E. tanala were cap- tured within a 30 m section of a trap line that was 454 m long. No differences in level of breeding activity were apparent within and between the various ele- vational zones during the 2-month period of the survey. Animals trapped at most sites varied in age and condition — from young males with ab- dominal testes to adult males with large scrotal testes and convoluted epididymides, and from im- mature females with imperforate vaginas to adults GOODMAN ET AL.: RODENTS 229 carrying embryos or bearing placental scars. No female, however, was actively lactating. The ratio of adult to subadult individuals was 4:4 at 810 m, 4:1 at 1200 m, and 6:5 at 1500 m (Table 14-6). Of the females dissected, one had four embryos measuring 19 mm crown-rump length and one had three placental scars. The mammae formula was 1-0-2 (n = 7), although one individual devi- ated from this and had a formula of 1-1-1. Comments — The examples of E. tanala from the RNI d'Andohahela average smaller in most external and cranial dimensions compared to more northern populations, particularly those from the RS d'Analamazaotra (18.5°S) and RS d'Anjanaharibe- Sud (14.7°S). In their uniformly gray venter and general size, the Andohahela specimens more closely resemble those from the RNI d'Andringitra (22.3°S) and the vicinity of Vinan- itelo (21.7°S). The latter place is the type locality of the species. Carleton and Goodman (1998) at- tributed these slight size and pelage differences to geographically clinal variation, from smaller in the south to larger in the north. Specimens Examined — Parcel 1, 12.5 km NW of Eminiminy, 24°35.6'S, 46°44.3'E, 810 m (FMNH 156514, 156528, 156631-156636, 156641; UA-SMG 7496); parcel 1, 13.5 km NW of Eminiminy, 24°35.0'S, 46°44.1'E, 1200 m (FMNH 156637, 156638; UA-MP 24, 25; UA- SMG 7544, 7549); parcel 1, 15.0 km NW of Em- iniminy, 24°34.2'S, 46°43.9'E, 1500 m (FMNH 156515-156521, 156531, 156532, 156639, 156640). Eliurus webbi Ellerman, 1949 Distribution — Although it was discovered rel- atively late in the taxonomic history of Nesomyi- nae (Ellerman, 1949), and even then acknowl- edged only as a subspecies, E. webbi is emerging as one of the most common and geographically widespread rodents of the eastern humid forest. As of 1994, the species was reported from the region of Montagne d'Ambre in the north, and south to the area around Vondrozo and Manombo (south of Farafangana) (Carleton, 1994). The spe- cies generally occupies lowland rain forest, rang- ing in elevation from near sea level to about 800 m, but exceptional records include 1000 m on Montagne d'Ambre and 1525 m on the mountains east of Ivohibe (Carleton, 1994; Goodman et al., 1996a). Within the RNI d'Andohahela, E. webbi inhab- its lowland forest of parcel 1 between 440 and 810 m. At the lower elevation it was the only species of rodent, native or exotic, trapped during the survey. Other recent fieldwork in southeastern Madagascar has demonstrated that this species is ubiquitous in the remaining relatively intact lit- toral forests (Manafiafy, Itapera, and Mandena); in lowland forests resting on lateritic soils. (Ma- rovony, Analalava, and Nahampoana); and in marginally upland areas of forest also resting on lateritic soils (Manantantely and Marosohy). At the Marosohy locality, near the northeastern boundary of parcel 1 of the RNI d'Andohahela, the QIT-FER team trapped this species between 350 and 900 m. Ecology and Reproduction — There was con- siderable disparity between the 440 and 810 m zones in the proportion of arboreal versus ground captures of E. webbi. At 440 m, 12 of 13 traps yielding E. webbi were off the ground, whereas at 810 m only three of six were so positioned (Table 14-4). At 720 m in the RNI d'Andringitra, six of 10 captures issued from arboreal sets; at 810 m this proportion was 11 of 13 (Goodman & Carleton, 1996). Individuals of E. webbi were not captured in sympatry with E. tanala at either the 440 m site in the RNI d'Andohahela or the 720 m site in the RNI d'Andringitra, but at the next higher elevational station, 810 m in both reserves, the two species were encountered in the same trap lines. Whether consistent patterns exist in the prevalence of terrestrial versus arboreal activities by E. webbi, corresponding to the presence or ab- sence of E. tanala, will require verification through longer term field studies at single sites. Of the 12 E. webbi captured in arboreal sets within the 440 m zone, one-half were on lianas, branches, or trunks of >10 cm diameter, whereas all three taken in the 810-m zone were on sub- strates of >10 cm diameter (Table 14-4). Individ- ual E. webbi entered live traps placed in a variety of terrestrial microhabitats, from sites in and around boulders and rock outcrops, to forest with open understory but thick herbaceous vegetatior or dense leaf litter, and near tree roots. In the 81C m zone one subadult was captured in a pitfall trap (see Chapter 13). At 440 m, seven of nine males captured pos- sessed abdominal testes and three of five females were reproductively inactive. One female with three placental scars had been lactating. In con- trast, all three male E. webbi captured at 810 m had scrotal testes, and four of five females exhib ited signs of recent or active reproduction. A max 230 FIELDIANA: ZOOLOGY imum of only 16 days separated trapping activi- ties at these two elevational zones, suggesting a difference in the incidence of breeding of E. web- bi and perhaps a slightly delayed onset at the higher locality. The mammae formula is consis- tently 1-0-2 (n = 6), except for one individual that appears to lack the postaxial pair and to possess an extra abdominal set (0-2-1). The breeding schedule appears to vary consid- erably among populations of E. webbi, based on this survey and those in the RNI d'Andringitra and RS d'Anjanaharibe-Sud (Goodman & Carle- ton, 1996, 1998). In the RNI d'Andringitra (22°S) at 720 and 810 m there was little breeding activity in November and December, whereas in the RS d'Anjanaharibe-Sud (14°S) all eight individuals captured in middle to late October showed some signs of ongoing reproduction. The factors that influence this variation are not known. Comments — Eliunis webbi remains have been identified in pellets from Asio madagascariensis collected in the Nahampoana Forest north of To- lagnaro (Goodman et al., 1993). The pellets were collected below a roost within lowland humid for- est and close to the ecotone along a cleared ag- ricultural zone. Specimens Examined — Parcel 1, 8 km NW of Eminiminy, 24°37.6'S, 46°45.9'E, 440 m (FMNH 156513, 156522-156527, 156642-156644; UA- SMG 7419, 7427, 7438, 7444); parcel 1, 12.5 km NW of Eminiminy, 24°35.6'S, 46°44.3'E, 810 m (FMNH 156529, 156530, 156659; UA-MP 19; UA-SMG 7474, 7478, 7481). Gymnuromys roberti Major, 1896b Distribution — Carleton and Schmidt (1990) summarized the distribution of this relatively rare, or at least seldom collected, species based on only five localities in eastern humid forest that span an elevational belt of 500-900 m. Renewed field ef- forts have obtained Gymnuromys roberti from ad- ditional sites: RS d'Anjanaharibe-Sud at 1260 m (Goodman & Carleton, 1 998), near Anjozorobe at about 1300 m (Goodman et al., 1998), and the RNI d'Andringitra between 720 and 1625 m (Goodman & Carleton, 1996). Intensive small mammal surveys in the humid forests of the PN de la Montagne d'Ambre have yet to uncover the species so far north (Goodman et al., 1996a, 1997a), leading those authors to suspect that the northern distributional limit is the highland com- plex around Tsaratanana. In parcel 1 of the RNI d'Andohahela, one in- dividual of G. roberti was captured in the 1200 m zone. This record extends the southern range of the species 200 km beyond its previously re- ported limit, the MZFAA site 20 km west of Von- drozo (Carleton & Schmidt, 1990). Like Eliurus tanala and Nesomys rufus, G. roberti emerges as another widespread component species of mid-el- evation eastern humid forest, although it is appar- ently not nearly as common as those rodents. Ecology and Reproduction — The single in- dividual, a female, was trapped on the ground, along a small mammal runway and next to a clus- ter of boulders. The animal has large mammae ( 1 - 0-2) and two placental scars but contains no em- bryos. Specimens Examined — Parcel 1, 13.5 km NW of Eminiminy, 24°35.0'S, 46°44.1'E, 1200 m (FMNH 156614). Monticolomys koopmani Carleton & Goodman, 1996 Distribution — The description of this recently named genus and species is based on four speci- mens. One (the holotype) was collected in May 1 929 on the Ankaratra Massif, around 1 800 m and above the Manjakatompo Forestry Station, and three individuals were obtained in 1993 at 1625 m in sclerophyllous forest of the RNI d'Andringitra (Carleton & Goodman, 1996). In February 1996 Goodman and colleagues revisited the Ankaratra Massif and found the species to persist in the No- siarivo Forest at 2000 m (Goodman et al., 1996b). Monticolomys koopmani occurs in parcel 1 of the RNI d'Andohahela at the upper limit of sclero- phyllous forest at 1875 m, just below Pic Trafon- aomby. Its presence in Andohahela extends the known distribution of the species about 260 km south of Andringitra. Ecology and Reproduction — All five individ- uals of M. koopmani collected had triggered live traps placed on the ground (Table 14-4) — on slopes with open understory (three captures), in an area of open understory with thick leaf litter, and along a runway under a moss-covered rock outcrop. In the RNI d'Andringitra, two of three specimens were taken on a liana and the third in a pitfall bucket; the individual recently collected from the Ankaratra Massif was trapped on the ground (Goodman et al., 1996b). These results collectively portray the species as predominantly terrestrial but accessing the lower strata of the for- GOODMAN ET AL.: RODENTS 231 est. Because traps were seldom placed more than 3 m above the ground (Table 14-1) it cannot be determined whether Monticolomys or other small mammals utilize the mid-strata and canopy por- tions of Malagasy forest. At the 1875 m site of parcel 1, individuals of M. koopmani appear to be clumped or distributed very locally. Four of five specimens were cap- tured within a 24 m section of a trap line that stretched for 255 m. Two females were taken on consecutive days at one trap station, and a male was captured in a trap 9 m away on one of those days. Microhabitat preferences of the species can- not be gleaned from the coarse habitat variables quantified. Two of the three males captured had scrotal tes- tes that measured 12 X 8 mm, with convoluted epididymides on one adult. The third male had partially descended testes that measured 5X3 mm, with nonconvoluted epididymides. Both cap- tured females were adults, but neither bore signs of ongoing reproductive activity. Each had six mammae, distributed as postaxial, abdominal, and inguinal pairs. Comments — Monticolomys koopmani is now known from three mountains (Ankaratra, Andrin- gitra, and Andohahela) in middle and southern Madagascar; a distance of 600 km separates its northernmost and southernmost occurrences in the upper reaches (1625-2000 m) of the eastern hu- mid forest biome. The fragmented distribution of the species conforms to the High Mountain Do- main, a phytogeographical unit dominated by sclerophyllous plants that occupy isolated peaks from Andohahela in the south to Tsaratanana in the north (Humbert, 1955). Although populations of Monticolomys are now patchily distributed, palynological data suggest that upper montane vegetation once stretched between these peaks as recently as the early Holocene (Burney, 1987, 1997). Its presence in parcel 1 of the RNI d' Andohahela likely approximates the southern geographical limit of the species, but its northern extent remains unknown. No examples of Monti- colomys were recovered during a small mammal survey of sclerophyllous montane forest in the RS d'Anjanaharibe-Sud, northeastern Madagascar. On this massif, another new genus and species of diminutive nesomyine rodent, Voalavo gymno- caudus Carleton and Goodman, 1998, was dis- covered, filling apparently the same upper mon- tane niche as that documented for Monticolomys. Notwithstanding their present-day discontinui- ty, the few locality samples of Monticolomys are remarkably similar in size, form, and pelage color. The univariate ranges of most variables, especial- ly craniodental, overlap appreciably (Table 14-7; uniformity of protocol for external dimensions be- tween two different collectors is, of course, sus- pect), and multivariate analysis of craniodental measurements (PCA) divulged no interpretable discrimination according to geographical origin (results not figured). Although the total number of specimens (10) is insufficient to critically assess patterns of differentiation, the evidence at hand portrays only isolated segments of a single, for- merly continuous, montane species. Specimens Examined — Parcel 1, 20.0 km SE of Andranondambo, 24°33.7'S, 46°43.3'E, 1875 m (FMNH 156534, 156660-156663). Nesomys rufus Peters, 1870 Distribution — Nesomys rufus has a broad dis- tribution in eastern, northern, and northwestern forests of the island, occurring from 900 to 2300 m (Carleton & Schmidt, 1990). This relatively large, richly colored nesomyine is not recorded from the PN de la Montagne d'Ambre in the ex- treme north (Goodman et al., 1996a), but it does inhabit the middle and upper montane forest of the RS d'Anjanaharibe-Sud (Goodman & Carle- ton, 1998). The southernmost locality of the spe- cies had previously been documented in the RNI d'Andringitra, where it was recorded within an elevational range from 810 to 1625 m (Goodman & Carleton, 1996). The several vouchers of N. rufus obtained at 810 and 1200 m in parcel 1, RNI d' Andohahela (Table 14-5), amplify the southern range limits of the species. Furthermore, individual Nesomys, presumably referable to N. rufus, were observed on a few occasions at 1430 and 1500 m (E Haw- kins, pers. comm.). This species overlaps with the lower-elevational form, N audeberti, between 900 and 1000 m in the PN de Ranomafana, where the two are probably the most common members of the local rodent community (Ryan et al., 1993). The QIT-FER small mammal surveys in 1989 and 1990 did not encounter N. rufus but did trap N. audeberti at several sites in the southeastern re- gion (Marovony Forest and Manantantely Forest). Ecology and Reproduction — Previous trap- ping results and mark-and-release ecological stud- ies indicate that TV. rufus is a strictly terrestrial rat (Ryan et al., 1993; Goodman & Carleton, 1996, 1998). The three individuals collected in live traps 232 FIELDIANA: ZOOLOGY Table 14-7. Selected external and craniodental measurements of Monticolomys koopmani from the three massifs of its known geographic occurrence. Ankaratra Andringitra (n = 3)t Andohahela ariable Holotype* 156211t (n = 5)H 3TL 205.0 245.0 236.3 ±3.2 234.0-240.0 227.4 ± 7.2 217.0-236.0 BL 89.0 85.0 98.0 ± 3.6 94.0-101.0 86.2 ± 1.9 84.0-89.0 L 116.0 150.0 138.0 ± 4.6 134.0-143.0 134.4 ± 5.9 127.0-142.0 FL 24.0 23.5 24.3 ± 0.6 24.0-25.0 23.6 ±0.5 23.0-24.0 L 15.0 19.0 18.3 ± 0.6 18.0-19.0 18.8 ± 0.8 18.0-20.0 T — 21.0 26.3 ± 1.1 25.0-27.0 23.0 ± 3.8 19.0-28.0 NL 26.3 27.1 27.8 ± 0.3 27.5-28.1 27.5 ± 0.4 26.9-28.0 B — 13.7 13.6 ± 0.5 13.1-14.1 13.6 ± 0.3 13.3-14.1 BC 12.5 11.2 12.7 ± 0.5 12.4-13.2 11.2 ± 0.1 11.1-11.3 )B 3.9 4.0 4.0 ± 0.1 3.9-4.0 4.0 ± 0.1 3.9-4.1 R 8.5 9.3 9.9 ± 0.3 9.6-10.2 9.5 ± 0.1 9.4-9.6 R 4.2 4.6 4.7 ± 0.2 4.5-4.9 4.6 ± 0.1 4.4-4.7 3L 8.7 9.1 9.5 ± 0.2 9.3-9.7 9.4 ± 0.3 9.1-9.7 BP 3.7 4.2 4.2 ± 0.2 4.0-4.4 3.8 ± 0.3 3.4-4.0 [F 5.3 4.6 5.2 ± 0.1 5.1-5.3 5.1 ± 0.2 4.9-5.4 IF 2.0 1.9 2.0 ± 0.1 1.9-2.0 2.0 ± 0.1 1.9-2.1 D 7.2 7.7 7.9 ± 0.1 7.7-8.0 7.6 ± 0.2 7.3-8.0 Mis 5.7 5.5 5.9 ± 0.1 5.8-6.0 5.7 ± 0.1 5.6-5.9 AB 4.4 4.9 4.8 ± 0.1 4.7-4.9 4.7 ± 0.1 4.6-4.9 ZP 2.2 2.2 2.2 ± 0.1 2.1-2.2 2.3 ± 0.2 2.0-2.4 OC 6.4 6.1 6.6 ± 0.2 6.5-6.8 6.4 ± 0.1 6.3-6.6 Ml-3 3.70 3.35 3.53 ± 0.12 3.42-3.65 3.52 ± 0.06 3.42-3.57 rMl 1.09 1.02 1.09 ± 0.03 1.07-1.12 1.09 ± 0.02 1.07-1.12 * AMNH 100727; Antananarivo Province, Manjakatompo, 1800 m. t FMNH 15621 1; Antananarivo Province, Foret de Nosiarivo, 2000 m. t FMNH 151727, 151899, 151900; Fianarantsoa Province, RNI d'Andringitra, 38 km S of Ambalavao, 1625 m. 1FMNH 156534, 156660-156663; Toliara Province, RNI d'Andohahela, 20 km SE of Andranondambo, 1875 m. Sample parameters are mean ± SD, and range. Refer to Materials and Methods section for abbreviations of variables. came from sites in open understory with thick herbaceous vegetation or leaf litter, on a slope un- der a dense growth of ferns, and by a large open- ing descending into a cavity under the roots of a fallen tree. A fourth individual was obtained in a National live trap set for carnivores. Population densities of this species, as indexed by trap returns, seem to vary appreciably among GOODMAN ET AL.: RODENTS 233 Table 14-8. Number of individuals and percent trap success based on all small mammals captured during the first 500 trap-nights within humid forest (parcel 1) and spiny bush (parcel 2) of the RNI d'Andohahela. Order and species Lipotyphla Microgale cowani Microgale dobsoni Microgale gymnorhyncha Microgale soricoides Microgale thomasi Oryzorictes hova Tenrec ecaudatus Rodentia Murinae Rattus rattus Nesomyinae Eliurus majori Eliurus minor Eliurus myoxinus Eliurus tanala Eliurus webbi Gymnuromys roberti Monticolomys koopmani Nesomys rufus Carnivora Galidictis fasciata Total individuals Percent trap success Total Rodentia Percent rodent trap success Total Nesomyinae Percent nesomyine trap success Humid forest Spiny bush 440 m 810 m 1200 m 1500 m 1875 m Total 120 m 2 1 3 1 5 4 1 11 1 8 2 4 14 10 5 1 1 1 2 2 15 1 2 1 3 10 20 14 15 12 71 2 2.0 4.0 2.8 3.0 2.4 2.8 0.4 10 18 11 11 4 54 2 2.0 3.6 2.2 2.2 0.8 2.2 0.4 10 14 11 9 3 47 1 2.0 2.8 2.2 1.8 0.6 1.9 0.2 sites in the eastern humid forest. In the RS d'Anjanaharibe-Sud, the greatest trap success was in the 1260 m zone, with eight animals captured in 500 trap-nights (1.6%) (Goodman & Carleton, 1998); in the RNI d'Andringitra the best results were 11 individuals in 625 trap-nights at 1210 m (1.8%) and 14 animals in 650 trap-nights at 1625 m (2.2%) (Goodman & Carleton, 1996). There were only two individuals in 500 trap-nights (0.4%) in the RNI d'Andohahela (Table 14-8). These biological inventories followed the same trapping protocol and were conducted at approx- imately the same time of the year. Along with additional information from the PN de Ranoma- fana, the data thus suggest that Nesomys rufus is more common in the middle segment of eastern humid forest but is more rare in the southern por- tion of this biome. Such an inference does not appear to be a spurious reflection of chance vari- ation in trap success. When Nesomys are present in any numbers within a forest, their diurnal ac- tivities, animated scuffling along the forest floor, and relatively large size readily draw an observ- er's attention. The species was both visibly rare and infrequently trapped in parcel 1 of the RNI d'Andohahela. Both of the male N rufus that were captured had large scrotal testes, and both females each contained two embryos measuring between 8 and 10 mm in crown-rump length. A similarly com- mon incidence of breeding activity was also doc- umented during the RNI d'Andringitra and RS d'Anjanaharibe-Sud surveys (Goodman & Carle- ton, 1996, 1998). Thus, depressed reproduction does not solely account for the variation in trap success between these localities and the appar- ently low population level noted in the RNI d'Andohahela. The two females have a mammae count of 0- 1 - 1 . Specimens Examined — Parcel 1, 12.5 km NW of Eminiminy, 24°35.6'S, 46°44.3'E, 810 m (FMNH 156645, 156646); parcel 1, 13.5 km NW 234 FIELDIANA: ZOOLOGY of Eminiminy, 24°35.0'S, 46°44.1'E, 1200 m (FMNH 156535, 156647). Discussion During fieldwork conducted between 20 Octo- ber and 4 December 1995, small mammals were censused in all three parcels that compose the RNI d'Andohahela. Previous faunal reports for the re- serve do not list any species of Nesomyinae (O'Connor et al., 1987; Nicoll & Langrand, 1989). The 92 specimens (captured in small mam- mal live traps) of the eight nesomyine species re- ported herein thus represent the first vouchered evidence of native rodents within the boundaries of the reserve (Table 14-5). Unlike recent field studies of the RNI d'Andringitra and RS d'Anjanaharibe-Sud (Carleton & Goodman, 1996, 1998), however, the biological survey of the RNI d'Andohahela did not uncover any mammals new to science. This may be an indication that our un- derstanding of Malagasy rodent diversity is ap- proaching biological reality, at least for those ne- somyines that inhabit the eastern humid forest (see Carleton & Goodman, 1998, for elaboration). The specimens listed above form the empirical basis for the discussions that follow, and those gathered through the earlier field efforts of Webb, Hoogstraal, MZFAA, and QIT-FER are generally referenced where appropriate to clarify and cor- roborate our summaries. We concentrate on re- sults obtained for parcels 1 and 2, given the longer period and greater number of trap-nights devoted to those surveys. Trapping Effort and Sampling Confidence Collecting effort was disproportionate among the reserve's three parcels, totaling 4,935 trap- nights in parcel 1, 750 in parcel 2, and only 150 in parcel 3 (Table 14-1). In parcel 1, eight species of rodents (Table 14-5), all of which are endemic Nesomyinae except Rattus rattus, were docu- mented over a 1435 m elevational band that en- compassed lowland forest (440 m), a zone tran- sitional between lowland and montane forest (810 and 1200 m), montane forest (1500 m), and scle- rophyllous forest (1875 m). In contrast, indige- nous rodent diversity is notably lower in the drier plant communities that dominate parcels 2 and 3 of the reserve (Table 14-5). In the spiny bush and gallery forest (120 m) of parcel 2, only two rodent species were captured: one nesomyine (Eliurus myoxinus) and one introduced murine {Rattus rat- tus). The brief visit to transitional forest of parcel 3 (170 m) also yielded just one species of Neso- myinae (E. myoxinus), as well as the two Rattus commensals (R. rattus and R. norvegicus). Ten- recid insectivores regularly entered live-trap lines intended for rodents (Table 14-8; and see Chapter 13); carnivores did so occasionally, but no lemur species was captured in such traps. Trap success of small mammals — native and in- troduced rodents, insectivores, and carnivores — varied little among the five elevational stations of parcel 1, ranging from 2% (440 m) to 4% (810 m) as calculated for the standardized trapping pe- riod (Table 14-8). The rates based on overall trap success, the grand total of 4,395 trap-nights, differ only trivially: 2.1% (18 animals in 845 trap- nights) at 440 m, 3.1% (26 animals in 850 trap- nights) at 810 m, 2.8% (22 animals in 775 trap- nights) at 1200 m, 3.0% (31 animals in 1,050 trap- nights) at 1500 m, and 2.9% (25 animals in 875 trap-nights) at 1875 m. The relative homogeneity of trap returns across elevations in the RNI d'Andohahela contrasts with results recorded in similar montane transects conducted elsewhere in Madagascar. For example, in the RNI d'Andringitra and RS d'Anjanaharibe-Sud, capture rates were distinctly higher over mid-elevations (ca. 1000- 1500 m), ranging from 6 to 10% (Goodman & Carleton, 1996, 1998). The trapping results from RNI d'Andohahela conform with those previously reported in one respect: the poorest success was obtained at the lowest elevation surveyed in each transect, sites (440-875 m) situated within low- land rain forest or near its transition to montane forest. Small mammal trap success in parcel 2 was especially meager, 0.4% realized during 750 trap- nights (Table 14-8), but the short duration of trap- ping and the paucity of comparable field studies in such dry habitats do not permit meaningful generalizations. The cumulative number of rodent species cap- tured at each elevational zone of parcel 1 became asymptotic well before the termination of a trap- ping session (Fig. 14-3, top). The expenditure of sampling effort at which these various plateaus were attained varies considerably — 250 trap- nights at 440, 1200, and 1500 m; 500 trap-nights at 1875 m; and 625 trap-nights at 810 m. Simi- larly, there was substantial variability in capture results over the course of trapping at each eleva- tion; there was no suggestion of orderly diminish- GOODMAN ET AL.: RODENTS 235 o 7 6- 5- 4 3- 2- 1 - o o o o -o- o- o / (k a a a- a a a a 0 / ^ffl -ffl -ffl p El ffi ET -a- -O D- 1 1 1 1 1 1 1 1 0 125 250 375 500 625 750 875 1000 1125 1250 Trap nights 10 T3 CD q- Z3 U. CO 8- o CO 7- CO 6- Ti > 5- -a _c 4- ., o 3- i_ CD •?- JQ E 12 1 - y o-i, o- ---o-- 500 750 Trap nights 1000 — i — 1250 1500 Fig. 14-3. Plots of trap-nights against the cumulative number of rodent species obtained (top) and against the number of individuals trapped (bottom) for the five elevations surveyed within parcel 1, RNI d'Andohahela (results include all nesomyine species and Rattus rattus). ing rates as the survey progressed, nor was there any correspondence to the asymptotic levels that were eventually reached (Fig. 14-3, bottom). The leveling of these species accumulation curves therefore does not appear to result simply from reduction in trap success. Although species accumulation curves stabi- lized over the 7 weeks of fieldwork, the knowl- edge of nesomyine distributional patterns gained over the past decade predicts that the number of forms resident in the reserve will be increased by at least two, perhaps three. These include one spe- cies associated with dry habitats, Macrotarsomys bastardi, and two species found in moist areas, Brachytarsomys albicauda and Nesomys audeber- ti. Macrotarsomys bastardi is a denizen of dry de- ciduous forests and arid bush and thicket forma- tions in western and southern Madagascar (Carle- ton & Schmidt, 1990). Webb had collected the species at 5 mi E Bevilany (his series is in BMNH), and the QIT-FER team discovered it in the Petriky Forest, 5-7 km southeast of Manam- baro (series are in UADBA and USNM). Parcels 2 and 3 of the RNI d'Andohahela are close to these localities, particularly the Bevilany site, and they contain habitat suitable for Macrotarsomys. Furthermore, the presence of Eliurus myoxinus within parcels 2 and 3 (see species account) un- derscores our expectation that M. bastardi will yet be found, because the two species commonly oc- cur at the same localities elsewhere in Madagascar 236 FIELDIANA: ZOOLOGY (Carleton & Schmidt, 1990). At other dry forest sites, its capture has proved highly variable and may relate to the cyclical availability of foods and corresponding population densities (Randrianjafy Rasoloarisoa, 1993; Ganzhorn et al., 1996; Good- man & Rasoloarison, 1997). Under such circum- stances, a species such as M. bastardi could have been easily overlooked during the abbreviated fieldwork in parcels 2 and 3. Brachytarsomys albicauda was not trapped in the RNI d'Andohahela, but there is some evidence that this large, arboreal species is resident there. A villager from Enosiary, at the eastern limit of parcel 1, assisted the 1995 mission. He remem- bered that several years earlier, while traversing at night the humid forest along the Col de Tana- tana between Isaka-Ivondro and Eminiminy, he had encountered a lemur-like animal moving along a system of vines. It was captured approx- imately 1 m off the ground, killed, roasted, and eaten.1 His description of the animal aptly fits the salient characters of B. albicauda. The Col de Tanatana is at about 750 m, which is within the known altitudinal range of Brachy- tarsomys albicauda (Carleton & Schmidt, 1990). To date the southern limit of the species has been noted as eastern humid forest in the vicinity of Vinanitelo and Ikongo (Carleton & Schmidt, 1990), approximately 300 km northeast of the Col de Tanatana. Curiously, no specimens of Brach- ytarsomys were recovered in the survey of the RNI d'Andringitra (Goodman & Carleton, 1996), nor did the QIT-FER team collect any in their sampling of several lowland and littoral forests in southern Madagascar. A museum specimen, its identification confirmed by comparison to type material, would do much to resolve the question of the species' existence in the RNI d'Andohahela. The third species of possible occurrence in the RNI d'Andohahela is Nesomys audeberti. Like N. rufus, it is a large and conspicuous diurnal rodent. This nesomyine has been previously recorded in southeastern Madagascar, from lowland forest covering lateritic soils. Sites include a ridge west ' Normally the consumption of rodent meat is strictly taboo to most Malagasy, but lemur is eaten by certain cultural groups (Ruud, 1970). The Enosiary guide re- counted the moment after the animal was killed and the factors that prompted his decision to eat it. He was con- vinced that it was a lemur by the way it moved and its general behavior, and he was very hungry. He was somewhat perplexed by the animal's rodent-like teeth, particularly the front incisors, but he finally decided that it was lemur-like enough to be properly edible. of Vondrozo at 500 m (MZFAA series in AMNH, BMNH, and MNHN), Manombo near sea level (MZFAA specimen in MNHN), the Bemangidy Forest below 100 m (Hoogstraal specimen in USNM), and the Marovony Forest at 50 m and Manantantely Forest at 100-450 m (QIT-FER se- ries in UADBA and USNM; Fig. 14-2). Our fail- ure to catch N. audeberti in parcel 1 and the same result for the QIT-FER team in the Marosohy For- est along parcel 1 may indicate that its preferred habitats are not found within this section of the RNI d'Andohahela. Populations of N. audeberti may be contained in other sections of parcel 1, such as the magnificent low-lying forest, with its strikingly different plant composition, that covers the headwaters of the Manampanihy Basin. A fourth nesomyine, a giant species known only from subfossil remains, deserves mention here. Hypogeomys australis, a congener of H. an- timena, from west-central Madagascar, was named based on material excavated from Andra- homana Cave, 40 km west of Tolagnaro (Gran- didier, 1903). A bone of//, australis from the type locality has yielded a radiocarbon date of 4,440 ± 60 BP (Goodman & Rakotondravony, 1996). If this species had survived into recent times, such a unique and imposing animal would certainly be retained in the rich folklore of the Antanosy and Antandroy tribes, if not mentioned in the earliest historical accounts (Flacourt, 1658). Along with Rattus rattus and R. norvegicus, a third commensal murine, Mus musculus, is com- mon in lowland villages and agricultural settings, but we lack evidence that it enters natural forest within the reserve. In the Nahampoana Forest, north of Tolagnaro, this rodent was identified in regurgitated pellets from Asio madagascariensis collected at the edge of lowland forest not far from agricultural lands (Goodman et al., 1993). This owl is believed to be largely forest-dwelling (Langrand, 1995), in which case the Mus could have been killed within the forest or along its edge. In summary, we are confident that the seven nesomyine species actually trapped in parcel 1, a tract of eastern humid forest, represent nearly all of those plausibly expected to inhabit forest at this latitude. The completeness of documentation for parcels 2 and 3 of the RNI d'Andohahela is jus- tifiably suspect, given the short duration of field- work at those sites and our awareness of the prob- able occurrence of one other species within such dry habitats in southeastern Madagascar. GOODMAN ET AL.: RODENTS 237 Elevation and Rodent Associations Elevational Distribution — No rodent species was trapped at all five stations along the eleva- tional transect of parcel 1, RNI d'Andohahela (Ta- ble 14-5). Eliurus minor, recorded at the four sites between 810 and 1875 m, appears to occupy the broadest band among native species within the re- serve, a finding that agrees with its elevational occurrence as noted elsewhere on the island (Carleton & Schmidt, 1990; Goodman & Carle- ton, 1996). Eliurus tanala populates middle ele- vations (810, 1200, and 1500 m) and is found sympatrically with E. webbi at the lower zone and with E. majori at the two upslope sites. Other spe- cies were obtained in just one or two of the tran- sect sites. Such examples of restricted occurrence may be only an artifact of unsuccessful trapping (Gymnuromys roberti and Nesomys rufus) or mir- ror real affinity to a particular vegetation class (E. webbi in lowland rain forest, E. majori in montane forest, and Monticolomys koopmani in sclero- phyllous montane forest). Introduced Rattus rattus was trapped in essentially pristine forest at all sites except those at 440 and 1200 m; further trap- ping will likely reveal the species as ubiquitous in primary forest of the reserve. The elevational juxtaposition of the three large Eliurus species in parcel 1 corroborates survey results for other forested mountain slopes in east- ern Madagascar. Eliurus webbi was not only rel- atively abundant at 440 m but the only species recorded at that elevation. At 810 m, however, it was outnumbered by individuals of E. tanala by a ratio of 8:5. At 1200 m E. webbi is not part of the local rodent fauna, but E. tanala and E. majori do co-occur there and at 1500 m (Table 14-5). The altitudinally contiguous allopatry observed for E. webbi (440-810 m) and E. majori (1200-1500 m), with populations of E. tanala overlapping each of those species, repeats the pattern found in both the RNI d'Andringitra and RS d'Anjanaharibe- Sud (Goodman & Carleton, 1996, 1998). It is in- teresting that to date the only place where E. web- bi and E. majori have been collected in sympatry is at 1000 m in the PN de la Montagne d'Ambre, an outlier mountain not inhabited by E. tanala (Goodman et al., 1996a). Whether these instances of altitudinal replacement among congeneric ro- dent species represent interference competition or reflect fine-grained microhabitat requirements will require detailed long-term ecological investiga- tion. As noted above, the elevational fidelities of cer- tain rodents in parcel 1 naturally correspond to previously recognized, broadly defined forest types. For example, Eliurus webbi occurs in low- land rain forest; E. tanala, E. majori, and Neso- mys rufus are encountered in the lowland-montane transitional zone through the upper reaches of montane forest (the infrequently trapped Gymnu- romys roberti may belong here as well); Monti- colomys koopmani inhabits sclerophyllous mon- tane forest; and E. minor, apparently the most eu- rytopic nesomyine, ranges broadly across all of these elevations and forest associations. Although the montane rodent communities of Madagascar exhibit some altitudinal zonation, they seem to lack the sharply defined stratification documented in other tropical areas (Pearson & Ralph, 1978; Rupp, 1980; Patterson et al., in press). In part, this modest zonal partitioning may simply reflect the absolutely fewer rodent species found on Mada- gascar as compared to their greater numbers along the eastern flanks of the Andes in South America or on the slopes of the rift mountains in East Af- rica. A notable omission, however, among Madagas- car's endemic rodents consists of those species that live exclusively in open habitats above the tree line. Such an alpine contingent forms a rich and distinctive subset of rodent communities that occur on high mountains elsewhere in the New and Old World tropics — for instance, the paramo and puna zones of the Andes (Pearson & Ralph, 1978), the alpine heathlands of the Ethiopian highlands (Rupp, 1980), or the meadowed pla- teaus that crown New Guinea's mountainous backbone (Flannery, 1995). No evidence presently exists that such an alpine rodent community differentiated on Madagascar. This absence seems to be an accident of Malagasy geology; there are no massive tablelands rising above 3000 m that would suit the formation of high-altitude grasslands and foster the evolution of pastoral rodents, their numbers consequently enhancing the zonation of species along an ele- vational profile. The alpine region surrounding the island's second highest mountain, Pic Boby (2658 m) in the RNI d'Andringitra, contains only the single indigenous species Brachyuromys betsi- leoensis (Langrand & Goodman, 1997). This vole-like form, with small, rounded pinnae and short tail, occurs throughout heathland areas (>2000 m) of the Andringitra Massif but also penetrates contiguous portions of montane and sclerophyllous forest (1700-1900 m); on the same massif, its congener B. ramirohitra, equally vole- 238 FIELDIANA: ZOOLOGY o - A Montagne d'Ambre 7- ,.Qt — -O— ■ Andringitra s 6_ 0" ■ -9* \ D Andohahela 'o / - - >/v> Anjanaharibe-Sud - Mt. d'Ambre/ /T] Andohahela i i 1500 2000 2500 3000 Elevation (m) at Summit Fig. 14-5. Regression of the elevation with the most native rodent species and height of summit for four reserves within the eastern humid forest biome. The highly significant relationship suggests the interplay of physical relief and orographic factors such as rainfall and cloud cover in explaining the maximal overlap of rodent species along tropical mountain slopes (see text for discussion). richness recurs among these four mountains, there is no consistent height at which diversity of native rodents peaks (Fig. 14-4). Zones that hold the largest number of species vary from 1000 m on Montagne d'Ambre to 1625 m on the Andringitra Massif. Nor is there an orderly north to south trend, suggestive of slight elevational shifts in vegetational communities along a latitudinal gra- dient, in the height where maximum species abun- dance was recorded. The two mountains with spe- cies diversity maxima at lower elevations lie at the opposite ends of Madagascar (Montagne d'Ambre, 1000 m, and Andohahela, 1200 m), with higher species counts at higher elevations in between (Anjanaharibe-Sud, 1260 m, and An- dringitra, 1625 m). Instead, there exists a strong, positive correla- tion (R2 = 0.99, F = 179.9, P < 0.01) between absolute height at the summit and that elevation on the mountain with the most species (Fig. 14- 5). For each of these four Malagasy mountains, the site with the greatest species richness occurs within the belt of wet montane rain forest, a zone below the nearly perennial cloud cap and the sub- alpine (sclerophyllous montane) and alpine habi- tats it envelopes. Productivity of rodent populations, as indexed by our measures of abundance and biomass (Table 14-9), inconsistently or obscurely corresponds to elevation with highest species diversity. The 810 m zone of the RNI d' Andohahela, with three ne- somyines in contrast to five at 1200 m, supports the largest standing crop of rodents, with or with- out the inclusion of data from Rattus (Fig. 14-6). This disconnection between elevation of greatest species richness and that of the largest total bio- mass departs from results obtained in the RNI d'Andringitra and RS d' Anjanaharibe-Sud, where all measures of rodent abundance — species diver- sity, population density, and biomass — were log- ically and closely coupled within a single eleva- tion (Goodman & Carleton, 1996, 1998). One fac- tor for the biomass-diversity disagreement within Andohahela may involve the inordinately low population density of Nesomys rufus, a large-bod- ied rodent that was numerically abundant in mon- tane zones of Andringitra and Anjanaharibe-Sud and whose heft (adult weight = 150-200 g) con- tributed substantially to biomass estimates within those reserves. Nor do elevational zones with the largest value of rodent biomass clearly relate to summit height for the three Malagasy reserves surveyed during approximately the same yearly interval (data from PN de la Montagne d'Ambre, obtained during a different time period, are excluded). Again, a non- relationship is generally disclosed whether indi- viduals of Rattus are included or excluded. Our 240 FIELDIANA: ZOOLOGY Table 14-9. Estimated biomass (g) of rodents trapped along an elevational transect in parcel 1 of the RNI d'Andohahela. Elevation (m) Species 440 810 1200 1500 1875 578 199 289 100 289 39 193 154 39 754 189 377 883 442 120 182 45 1 4 5 4 3 883 1,813 883 920 373 Rattus rattus Eliurus majori Eliurus minor Eliurus tanala Eliurus webbi Gymnuromys roberti Monticolomys koopmani Nesomys rufus Number of species Total biomass (g) Total biomass (g) excluding Rattus 883 1,235 883 631 84 The numerical data represent summations from the average weight of adults captured over the first 500 trap-nights within each zone. biomass comparisons among elevations and be- tween major reserves hinge critically on line-tran- sect, removal-trapping results, which can be no- toriously happenstance and embrace unknown but likely high levels of sampling error. Furthermore, some studies have shown that nesomyine species undergo seasonal reproductive cycles and popu- lation fluctuations (Rakotondravony, 1992; Ste- phenson, 1994; Goodman et al., 1997a), annual phenomena which our short-term calculations of biomass would overlook. Whether rodent popu- lation density and biomass on Madagascar's mountains predictably conform to some larger pattern will require specifically designed field in- vestigation, conducted over a longer term and pre- ferrably employing a grid-based, mark-and-re- lease scheme. In summary, the peak of rodent species rich- ness in Madagascar's eastern humid forest consis- tently occurs within middle to upper montane vegetation, a band which coincides with that por- tion of a mountain just below the zone of frequent cloud cover. The absolute height of a mountain, and its physical effect upon cloud formation and rainfall pattern, influences the vertical position of the montane zone and accounts for the significant positive correlation between elevation with the most rodent species and height at the summit (Fig. 14-5). In turn, topography and the related oro- graphic variables of rainfall and cloud cover strongly modulate processes such as water trans- fer, penetrance of solar radiation, nutrient cycling, and richness of the soil macrofauna. Acting in concert, such variables may locally optimize eco- logical productivity along a mountainside and partly explain the mid-elevational bulge in rodent species richness. Intra- and Interspecific Differences in Re- production— Over the past few years new infor- mation has become available on the reproductive ecology of nesomyine rodents. Most of this work is associated with differences in the age and re- productive state of rodents along elevational tran- sects (Goodman & Carleton, 1996, 1998). We hasten to add that although there have been some advances, we know virtually nothing about the cues these animals use to commence breeding cy- cles. Here we summarize information on the re- productive condition of rodents trapped in the RNI d'Andohahela and compare this information to parallel transects on the slopes of the RNI d'Andringitra and the RS d'Anjanaharibe-Sud. Levels of rodent reproduction were generally high on the slopes of parcel 1 of the RNI d'Andohahela, with signs of active breeding evi- dent in 36% (440 m) to 94% (1220 m) of the animals trapped within an elevational zone (Table 14-6). Moreover, no progressive trend with ele- vation is suggested for these data. In the RNI d'Andringitra there was a directional pattern of increasing reproductive activity at higher alti- tudes, ranging from only 25% of the rodents sam- pled at 720 m to over 80% at 1625 m. This trend is not evident in the data from RS d'Anjanaharibe- Sud. These three elevational transects were made at slightly different periods: parcel 1 of RNI d'Andohahela from late October through early GOODMAN ET AL.: RODENTS 241 5UUU- 4000- S ,.C N-o^ © _; 00 © vD © m + 17+17 ^ (^ "^ f-^ "^" 00 in tt vq © en ON nD — — 1 1 1 — m -st- ci m n Tt " 00 ^ — n rr r- 00' CO ■ rt I ON ON - _- o '-. — © © _ ; 00 r- o ri "■> p~ co IT) > > m f~- (N 0 a> 1 rnrf © CO — 1 h rf T3 X) co u-i b « co ^H rt 03 r -^^0+ .= 11 SS5S •a c « c ^ 9 = * ^3 co S 3 3 F 0 £ & ° & § « .2 S +1 s a. . u >< — t/i II iu .a Q u 254 FIELDIANA: ZOOLOGY abdominal testes; the other was a female with an open pubic symphysis and large mammae. Specimens Examined From the Reserve — Par- cel 1, 12.5 km NW of Eminiminy, 810 m (FMNH 156499); parcel 1, 13.5 km NW of Eminiminy, 1200 m (FMNH 156500). Miniopterus manavi Thomas, 1906 Two individuals of this species were obtained in parcel 1 of the RNI d'Andohahela. One was netted at 810 m within 10 min after dusk over a small tributary of the Andranohahela River. The bare skin around the eye was dull yellow. The second individual was found by A. Raselimanana at 1500 m in a night roost under a small rock overhang. These specimens are assigned to Mini- opterus manavi on the basis of differences in mea- surements of Malagasy Miniopterus spp. outlined by Peterson et al. (1995) and Hill (1993). Miniop- terus manavi has also been collected in the Ma- rosohy Forest near the northern limit of the re- serve (parcel 1) at about 350 m (USNM 578744, 578745). It has also been reported along the Man- dromodromotra River, 16 km NE of Tolagnaro (Peterson et al., 1995) and near Tolagnaro (Hill, 1993). In the USNM there are series of this spe- cies taken at other sites in southeastern Madagas- car that include the Nahampoana Forest, 7 km NNW of Tolagnaro, between 100 and 450 m (USNM 577102-577119, 577129-577131, 577297-577299, 577302-577307); the Manantan- tely Forest, 8-12 km WNW of Tolagnaro, be- tween 100 and 450 m (USNM 577096-577101, 577296); the Itapera Forest, approximately 21 km NW of Tolagnaro, and near sea level (USNM 577128); and along the Itaranta River, 20 km WNW of Ranopiso, at 20 m (USNM 577120- 577122, 577124, 577125). Reproduction — The male had slightly de- scended scrotal testes, and the female had large mammae. Specimens Examined from the Reserve — Par- cel 1, 12.5 km NW of Eminiminy, 810 m (FMNH 156501); parcel 1, 15.0 km NW of Eminiminy, 1500 m (FMNH 156502). Family Molossidae Mormopterus jugularis (Peters, 1865) Two individuals of Mormopterus jugularis were netted over a small river at the edge of par- cel 2 of the RNI d'Andohahela. On one side of the river was heavily degraded gallery forest and on the other bank slightly degraded spiny bush. Mormopterus jugularis has also been collected in the humid portion of southeastern Madagascar north of Manantenina (USNM 577161-577172, 577178-577188) and in drier areas near the Itar- anta Forest (USNM 577132-577134), near Ber- aketa (USNM 577361), and in the Ankapoky For- est (USNM 577313, 577314, 577321, 577322, 577336). This species has a broad distribution across the island and generally roosts in buildings (Peterson et al., 1995). Reproduction — The two individuals captured were females, one with large mammae and the other with slightly enlarged mammae. Specimens Examined From the Reserve — Par- cel 2, 7.5 km ENE of Hazofotsy, 120 m (FMNH 156612, 156613). Other Regional Records Several other species of Megachiroptera and Microchiroptera bats have been reported or col- lected in the Tolagnaro area and around the vari- ous parcels of the RNI d'Andohahela. The holo- type of Pteropus rufus princeps, a subspecies not currently recognized, was collected in the Tolag- naro region (Andersen, 1908). This species has been found near Tolagnaro (Peterson et al., 1995), Bemangidy (USNM 317901, 317902), and Man- antenina (USNM 578832, 578833). Colonies are also known from Berenty along the Mandrare River (Nicoll & Langrand, 1989) and in the Man- afiafy (St. Luce) forest (Goodman, pers. obs.). Ei- dolon dupreanum has been reported from the To- lagnaro region and the Grotte d'Andrahomana, along the coast south of Ranopiso (Peterson et al., 1995). Among the Microchiroptera several other spe- cies have been reported from the region. Myzo- poda aurita, a member of the endemic family My- zopodidae, was obtained at Antanifotsy, 8 km N of Tolagnaro (Gopfert & Wasserthal, 1995); 4 km SSE and 2-3 km NW of Manafiafy (USNM 578742, 578743, 578856-578858); along the Mandromodromotra River, 15 km NE of Tolag- naro (Peterson et al., 1995); near Bemangidy (FMNH 85237, 92832, 92833); and in the Ana- lalava Forest (USNM 577065). The Bemangidy specimens were found "in curled-up central leaf of very large traveller's palm [Ravenala mada- GOODMAN: BATS 255 Table 15-2. A comparison of the forest-dwelling bat faunas of several sites in the eastern humid forests o1 Madagascar. Site, latitude, and elevational range sampled Species Andohahela* 24°36'S 400-1875 m Andringitrat 22°13'S 720-1625 m Anjanaharibe-Sudt 14°45'S 875-1950 m MarojejyU 14°26'S -500-700 m Rousettus madagascariensis + + Emballonura atrata Hipposideros commersoni + + Triaenops rufus Myotis goudoti + + Scotophilus robustus Miniopterus fraterculus Miniopterus manavi + Miniopterus minor + Miniopterus spp. Myzopoda aurita Tadarida pumila Total number of species 4 4 + * Restricted to the humid forest portions of the reserve and based on information presented in this chapter. t Goodman (1996). X Goodman (1998). ^ Pont and Armstrong (1990). || Apparently several different species were netted and no voucher specimens were collected. Field determinations for this genus are unreliable. gascariensis, family Strelitziaceae] near river" (H. Hoogstraal, field notes, FMNH). The horse- shoe-shaped pads on the sole and thumb appar- ently aid this species in moving along and cling- ing to the leaves. At numerous other sites on the island Myzopoda has been captured in heavily de- graded areas and at the edge of marshes where Ravenala is a dominant species (Schliemann & Maas, 1978; Gopfert & Wasserthal, 1995). There is some evidence, however, that Ravenala is not the only plant used for roost sites. I captured two of the USNM specimens from Manafiafy men- tioned above at dusk in mist-nets placed at un- derstory level and within relatively intact littoral forest. These nets were at least 500 m from any Ravenala, and it was my impression that the in- dividuals were leaving a nearby roosting site and flying in the direction of the open marshland. Other species of Microchiroptera known from the area include Emballonura atrata (Emballon- uridae), collected near the Marovony Forest (USNM 577062, 577063, 577257, 577258), north of Manantenina, and near Fanjahira, on the west- ern side of the Vohimena Mountains (Peterson et al., 1995); Triaenops rufus (Hipposideridae), along the Itaranta River (USNM 577064), and near Tolagnaro (Peterson et al., 1995); Miniopte- rus majori (Vespertilionidae), at Manantantely (USNM 577075) and along the Itaranta River (USNM 577076-577078); and Mormopterus ju- gularis (Molossidae), near Tolagnaro (Peterson et al., 1995). Also, Pipistrellus nanus (= Pipistrellus sp., sensu Peterson et al., 1995) and Miniopterus {minor) manavi have been reported from the RNI d' Andohahela (Nicoll & Langrand, 1989). Discussion Over the past few years a number of bat inven- tories have been conducted in the eastern humid forests of Madagascar, and sufficient data are now available to permit preliminary comparisons be- tween the faunas of these different sites. It must be noted that although the data from the RNI d' Andohahela, RNI d'Andringitra, and RS d'Anjanaharibe-Sud are comparable with respect to sampling effort (Goodman, 1996, 1998, here- in), the species lists that were generated should not be considered complete. During these surveys no effort was made to employ canopy nets, which are useful for capturing high-flying species, nor were any searches made for roost sites (Voss & Emmons, 1996). Other chiropteran species will undoubtedly be found at these forest sites with further sampling. Even with these limitations, however, several clear patterns have emerged. 256 FIELDIANA: ZOOLOGY On the basis of faunal inventories, the forest- dwelling bat community of the eastern humid for- est of Madagascar shows low species richness and little species turnover across nearly the complete length of the island and nearly 12° of latitude (Ta- ble 15-2). The majority of bats classified here as forest-dwelling species also occur outside of this habitat, and they presumably forage in open areas. Probably the most intensive survey of bats in the humid forests of Madagascar was conducted by Pont and Armstrong (1990) in the RNI de Ma- rojejy, where 23 bats of eight species were cap- tured in 8,071 net meter hr. Of this capture effort, 1,063 net meter hr (13% of the total) accrued in primary forest formations where not a single bat was captured. Although this reserve had more in- tensive sampling effort and higher species rich- ness compared to the other three sites presented in Table 15-2, there are few differences among these four reserves in the forest-dwelling portions of their bat faunas. Peterson et al. (1995) recognized 29 different bat species on the island. This figure includes spe- cies living in open areas, as human commensals, and in humid and dry forests. When compared to other large Old World tropical islands of similar surface area, such as Borneo and New Guinea (Payne et al., 1985; Flannery, 1990) the bat fauna of Madagascar is depauperate. In particular, there is no evidence of a distinct obligate forest-dwell- ing bat community on Madagascar. The use of harp traps or very fine mesh mist-nets during in- ventories might reveal a greater diversity of for- est-dwelling bats than that demonstrated through the use of standard mist-nets. Acknowledgments For comments on an earlier draft of this paper I am grateful to two anonymous reviewers. Literature Cited Andersen, K. 1908. Twenty new forms of Pteropus. Annals & Magazine of Natural History, 8th series. 2: 361-370. Bergmans, W. 1994. Taxonomy and biogeography of African fruit bats (Mammalia, Megachiroptera), 4: The genus Rousettus Gray, 1821. Beaufortia, 44: 79- 126. Dorst, J. 1947a. Les chauves-souris de la faune Mal- gache. Bulletin du Museum National d'Histoire Na- turelle, series 2, 19: 306-313. . 1947b. Essai d'une clef de determination des chauve-souris Malgaches. Memoires de lTnstitut Scientifique de Madagascar, series A, 1: 81-88. . 1948. Biogeographie des Chiropteres malgaches. Memoires de l'lnstitut Scientifique de Madagascar, series A, 1: 193-198. Flannery, T 1990. The mammals of New Guinea. Robert Brown and Associates, Carina, Australia. Goodman, S. M. 1996. Results of a bat survey of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagascar, pp. 284-288. In Good- man. S. M., ed., A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagascar: With reference to eleva- tional variation. Fieldiana: Zoology, n.s. 85: 1-319. . 1998. Notes on the bats of the Reserve Spe- ciale d'Anjanaharibe-Sud, Madagascar, pp. 223-226. In Goodman, S. M., ed., A floral and faunal inventory of the Reserve Speciale d'Anjanaharibe-Sud. Mada- gascar: With reference to elevational variation. Field- iana: Zoology, n.s. 90: 1-246. -, M. Pidgeon, A. F. A. Hawkins, and T S. Schu- lenberg. 1997. The birds of southeastern Madagas- car. Fieldiana: Zoology, n.s. 87: 1-132. Gopfert, M. C, and L. T Wasserthal. 1995. Notes on echolocation calls, food and roosting behaviour of the Old World sucker-footed bat Myzopoda aurita (Chiroptera, Myzopodidae). Zeitschrift fur Saugetier- kunde, 60: 1-8. Hill, J. E. 1993. Long-fingered bats of the genus Mini- opterus (Chiroptera: Vespertilionidae) from Madagas- car. Mammalia, 57: 401-405. 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. Payne, J., C. M. Francis, and K. Phillips. 1985. A field guide to the mammals of Borneo. The Sabah Society, Sabah, Malaysia. Peterson, R. L., J. L. Eger, and L. Mitchell. 1995. Chiropteres, vol. 84. Faune de Madagascar. Museum National d'Histoire Naturelle, Paris, 204 pp. Pont, S. M., and J. D. Armstrong. 1990. A study of the bat fauna of the Reserve Naturelle Integral de Ma- rojejy in north-east Madagascar. Report of the Aber- deen University expedition to Madagascar 1989. De- partment of Zoology, University of Aberdeen, Aber- deen, 57 pp. Rasolozaka, I. N. 1994. Lcs micro-chiropteres, pp. 64-67. In Goodman, S. M., and O. Langrand, eds., Inventaire biologique Foret de Zombitse. Recherches pour le Developpement, Serie Sciences biologiques. No. Special. Centre d' Information et de Documenta- tion Scientifique et Technique, Antananarivo. Schliemann, H.. and B. Maas. 1978. Myzopoda aurita. Mammalian Species. 116: 1-2. Voss, R. S., and L. H. Emmons. 1996. Mammalian di- versity in Neotropical lowland rainforests: A prelim- inary assessment. Bulletin of the American Museum of Natural History, 230: 1-1 15. GOODMAN: BATS 257 Chapter 16 Carnivora of the Reserve Naturelle Integrate d'Andohahela, Madagascar Steven M. Goodman1 and Mark Pidgeon2 Abstract The humid forests of parcel 1 of the Reserve Naturelle Integrale d'Andohahela contain 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 diversities of native animals were at 440 and 1200 m, with four species each. The known carnivore species of the reserve are identical to those of two other sites in the southern portion of the eastern humid forest. In general, most species of carnivores found in humid forests of Madagascar are known from scattered localities along most of the latitudinal length of this habitat on the island. Resume Les forets humides de la Parcelle 1 de la Reserve Naturelle Integrale d'Andohahela renfer- ment cinq carnivores originaires de la region (Galidia elegans, Galidictis fasciata, Cryptoprocta ferox, Eupleres goudotii, et Fossa fossana) et trois carnivores introduits (Canis lupus, Felis silvestris, et Viverricula indica). La plus grande diversite des animaux indigenes se trouve a 440 m et 1200 m avec quatre especes. Les especes de carnivores connues dans la reserve sont identiques a celles trouvees dans deux autres sites dans la partie Sud de la foret humide de l'Est. En general, la plupart des especes rencontrees dans les forets humides de Madagascar sont connues a partir des endroits disperses le long d'une grande partie de la longueur latitu- dinale de cet habitat de l'ile. Introduction Little information is available on the carni- vores occurring in the southeastern portion of Madagascar. In his monograph on the carni- vores of the island, Albignac (1973) plotted re- cords on distributional maps for only one spe- cies south of the Mananara River. Two more re- cent summaries of the known vertebrate fauna of the Reserve Naturelle Integrale (RNI) 1 Field Museum of Natural History, Roosevelt Road at Lake Shore Drive. Chicago, IL 60605-2496, U.S.A. 2 Route de St. Cergue, 1 270 Trelex, Switzerland. d'Andohahela listed several species of carni- vores (O'Connor et al., 1987; Nicoll & Lan- grand, 1989); in most cases these records were based not on direct observations but on com- munications with local forest guards and people living around the reserve. In this chapter we re- view records of Carnivora in southeastern Mad- agascar and present information on the carni- vores of the RNI d'Andohahela that was gath- ered during the 1995 expedition at five sites along an elevational gradient in the humid for- est of parcel 1 and at a sixth location at 120 m in the spiny forest of parcel 2. We follow the taxonomy of Wozencraft (1993). GOODMAN & PIDGEON: CARNIVORA 259 Materials and Methods Information on the Carnivora of the RNI d' Andohahela was gathered as part of a biological inventory of the reserve (see Chapter 1 for de- tails). In parcel 1 the transect zones were centered at 440, 810, 1200, 1500, and 1875 m. The ele- vational swath within each zone was ±75 m from the mid-point. Data on carnivores were gathered using three techniques: direct observations by the survey participants, identification of scats, and systematic trapping. Within each elevational zone two systems of trap lines were used: one for rodents and the other for carnivores. Rodent lines were baited with pea- nut butter (see Chapter 14); carnivore lines were baited with rehydrated dried fish, chicken offal, and morsels of fresh river eel. The carnivore lines consisted of 13 National traps placed in the forest, near the camp, off and along trails, and beside rivers. Most of the trapped animals were weighed, checked for reproductive condition, marked, and released near the capture site. Markings consisted of clipping unique patterns of fur from discrete portions of the tail. Trapped animals saved as voucher specimens were prepared either as flat museum skins, with associated skulls and partial skeletons, or as fluid- preserved carcasses. This material is housed in the Field Museum of Natural History (FMNH), Chi- cago, and the Departement de Biologie Animale, Universite d' Antananarivo (UA). Other material reviewed for this paper includes specimens from southeastern Madagascar in the Museum National d'Histoire Naturelle (MNHN), Paris; National Museum of Natural History (USNM; formerly the United States National Museum), Washington, D.C; and The Natural History Museum (BMNH; formerly the British Museum of Natural History), London. Five measurements, in millimeters (mm) or grams (g), were taken for each specimen in the flesh; their abbreviations and definitions are given below. TOTL total length of body and tail: from tip of nose to end of last caudal vertebra (not including terminal hair tuft) TL tail length: from base of tail (held at right angles to the body) to end of last caudal vertebra (not including terminal hair tuft) HFL hind foot length: from heel to tip of lon- gest toe (not including claw) EL ear length: from basal notch to distal tip of pinna WT weight: measured with Pesola spring scales, to ±5 g for animals < 1,000 g and to ±10 g for those > 1,000 g. Six cranial and four dental dimensions were measured with a digital caliper to the nearest 0.1 mm. These measurements, and their abbrevia- tions, follow. BBC breadth of the braincase: greatest distance across the mastoids along an axis perpen- dicular to the cranium CBL condylobasal length: least distance from the posteriormost portions of the occipital condyles to the anteriormost portion of the premaxillae CM length of maxillary toothrow: from canine to posteriormost molar PL palatal length: from indentation at the pos- terior edge of the palatine bones to ante- rior edge of the incisor alveoli POB postorbital breadth: least distance across the frontal bones posterior to the postor- bital processes LTR length of mandibular toothrow: from an- terior edge of first incisor to posterior edge of last molar ML length of mandible: from mandibular con- dyle to anteriormost edge of the mandible UTR length of maxillary toothrow: from ante- rior edge of incisors to posterior edge of last molar WC width across the upper canines: measured across the base of the canine alveoli ZB zygomatic breadth: the greatest distance across the zygomatic arches measured per- pendicular to the long axis of the skull Results Eight species of carnivores were recorded in the RNI d' Andohahela; five were endemic to the island and three were introduced (Table 16-1). All of these species were recorded in parcel 1. The highest diversity of carnivores was in the 440 m zone, with four native and two introduced species. No native and three introduced carnivores were documented during this inventory in parcel 2. Three of the five endemic carnivores recorded from the reserve were captured in traps; records 260 FIELDIANA: ZOOLOGY Table 16-1. The distribution of carnivores along an elevational gradient in humid forest (parcel 1) and spiny bush (parcel 2) of the RNI d'Andohahela. Carnivore species Parcel 1 440 m 810 m 1200 m 1500 m Parcel 2 1875 m 120 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 species Total number of endemic species + + + + + + + + + + + + + + + + 6 3 4 2 2 3 4 3 4 2 2 0 Introduced to Madagascar. of the other two species are based on scats or observations. Trap success with the carnivore lines was generally low and varied from 0.0 to 7.7% (Table 16-2). Quantification of trap success for carnivores captured in the rodent trap lines is presented in Chapter 14. Cranial and external measurements of captured individuals and those preserved as voucher specimens are given in Ta- ble 16-3. no clear evidence of feral dogs in habitats deep within the reserve. Our observations of this ani- mal from the RNI d'Andohahela were confined in parcel 1 to the 440 m site, which was close to the forest edge, and in parcel 2 to the 120 m site, not far from several villages (Table 16-1). Comments — Domestic dogs are regularly used by local people for hunting, principally of Tenrec ecaudatus and Potamochoerus larvatus, and as guards for cattle and property. Species Accounts Family Canidae Canis lupus Linnaeus, 1758 Distribution — Domestic dogs were confined to areas in and around human habitation. We found Family Felidae Felis silvestris Schreber, 1775 Distribution — There are records of cats that phenotypically resemble African wild cats from a variety of localities on Madagascar, particularly Table 16-2. Trap capture rates for the carnivore lines established in parcels 1 and 2 of the RNI d'Andohahela. meter Parcel 1 . Parcel 2 Descriptive para 440 m 810 m 1200 m 1500 m 1875 m 120 m Cumulative number of Animals captured Trap-success trap-n ghts 91 1 1.1% 91 7 7.7% 104 3 2.9% 104 4 3.8% 78 0 0.0% 78 0 0.0% Captures of Galidia elegans Galidictis fasciata 1 7 1 1 1 3 — — Fossa fossana — — 1 — — — GOODMAN & PIDGEON: CARNIVORA 261 o CM >n — 0\ On + 1 I Wi en — i in m in Tf 00 00 r- m co «NO + i7^0noo — i CM -h ■* en rn en ■* oo (N en d 00 CM 00 .+1 I Tf I^ Tf in .0 ± 1. 69-72 74, 75 73 67 en on m r- no no o °) m oo cm no in cm cm no oo + 11 -en ~ O r~ CM CM •n no en fsj (N CM NO CM CI ON VO on ■<*• en CM CM CM *n cm - in en o ^o no oo oo + 11 .NO On ^ — Tt m tj- en CM r^ On' O «"> CM NO CM On en en in r^ no >n in On en rfr en ON ON .+1 I CM tJ- ON ("- 00 ON oo ON "J fN ■* NO CM NO —i 3 T3 idy adult idy adult idy subad 2 11 11 3 c c 0+ a """ II C ^ wO+ *o 0+ — 03 00 60 00 c c c ,«fl *o 0+ 3 u B o C III D 1) Q) — — CO 3 3 X) "3 "3 £ < CQ 03 CQ "a "O T3 3 ^ d K -O T3 c 0 262 FIELDIANA: ZOOLOGY u O ON no d m r- — + 1 I d — ""> C"S pr; +1 I rn Tt TT _ — m — NO r-; d on fS r, r; + 1 I 00 00 _ <^ o-, Tt r- — ~ r~ (N no + 1 I r-^ On ^-nc^ ON NO NO — p~, o nC ■* E 2 u 0 c: "O '■— - ill 1 _ « u J -2 "o _fe sj +1 M ^ 2 = ^ • c vf V. ^ -O « C w 00 « •£ « S -o .5 u 3 NO — S. O o — u j- Q. .5 .• 0 u }) c — * ,, E 00 S ^ > £.on a ^ - Q 00 i GOODMAN & PIDGE0N: CAPvNIVORA 263 western deciduous forest and southern spiny bush (M. Nicoll, pers. comm.). In several cases it is unclear whether these individuals represent feral domestic cats or a remnant population of intro- duced African wild cats. All of our records of this species in the RNI d'Andohahela are from the for- est edge or outside the forest at 440 m in parcel 1 and 120 m in parcel 2 (Table 16-1). Reports of wild cats in the region date from at least the latter half of the 17th century (Flacourt, 1658). Family Herpestidae Subfamily Galidiinae Galidia elegans elegans I. Geoffroy Saint- Hilaire, 1837 Distribution — Galidia elegans is one of the most broadly distributed native carnivores on Madagascar, occurring throughout the length of the eastern humid forest from Tolagnaro to Mon- tagne d'Ambre, as well as in the Sambirano and western forests (Albignac, 1973; O'Connor et al., 1987). It was the most commonly observed and captured carnivore in parcel 1 of the RNI d'Andohahela. This species was recorded over the complete altitudinal breadth of the transect from 440 to 1875 m (Table 16-1). It was not found in parcel 2 of the reserve, but Tandroy villagers have reported occasional sightings of Galidia in the spiny bush during the short wet season (T Saotsy, pers. comm.). Ecology and Reproduction — Galidia elegans was often observed in and around our camps and was attracted to the sounds and smells of the kitchen and laboratory areas. Trap captures and sight records showed it to be most common in the 810 m zone, where seven individuals were cap- tured in 91 trap-nights (7.7%) in the carnivore lines (Table 16-2). This species was also captured in all other elevational zones (except 1875 m) in National traps set for rodents and baited with pea- nut butter (see Chapter 14). An individual that was captured, marked, and released in the 810 m zone was observed several days later in the 1200 m zone. There was a straight line displacement of approximately 2.5 km between the two sites. Nine of the 10 Galidia captured in the carnivore lines were males, and about half of them had scrotal testes. The large number of Galidia elegans trapped and observed during the survey certainly reflects a high overall local abundance. It is presumably the most common carnivore in the reserve. The more frequent sightings of Galidia compared to other carnivore species are largely due to its bold and inquisitive nature; certainly they reflect its abundance, diurnal activity, and utilization of trails and river margins. Another contributory fac- tor may thus be that our camps were always close to water. On several occasions this species was observed moving through the mid-canopy, up to 10 m off the ground, and passing between large canopy trees via touching limbs. Arboreal behav- ior was previously repotted for this species (Rand, 1935; Goodman, 1996). When moving about ar- boreally, a G. elegans would often tear open clumps of epiphytes clinging to tree branches or stick its head into small crevices or holes. Al- though no observations were made of food items being taken, this may be how the carnivore finds and feeds on the small nocturnal primates (Micro- cebus and Cheirogaleus) that it is known to prey upon (Goodman et al., 1993; Wright & Martin, 1995). This carnivore was also observed being mobbed by two species of birds, Phyllastrephus zosterops and Xenopirostris polleni (F. Hawkins, in litt.). On one occasion G. elegans was observed hunting at the edge of a stream and diving into the water in pursuit of amphibian prey (A. Rase- limanana, pers. comm.). Comments — Currently three subspecies of Galidia elegans are recognized, with the nominate form occurring from Tolagnaro north to at least the Antalaha region (Albignac, 1973). On the basis of specimens captured in the RNI d'Andohahela and comparison to material housed in museums from the eastern humid forests, the population living in and around the reserve shows noticeable differences in pelage coloration from typical G. e. elegans. The red pelage is often much darker and more saturated with a deep ru- fous-red color than more northerly populations of G. e. elegans. Further, the ventrum of most indi- viduals from the reserve is distinctly black, and this color often extends down the inner surface of the hind and forelegs and anteriorly to the chin. It is not yet clear whether this variation in pelage coloration is correlated with mensural or osteo- logical differences from other eastern humid for- est populations or is part of a broad latitudinal cline. The form occurring in the dry deciduous forest of the west, G. e. occidentalis Albignac, 1971, apparently shares some pelage characters with the Andohahela populations. Specimens Examined — Parcel 1, 12.5 km NW of Eminiminy, 24°35.6'S, 46°44.3'E, 810 m 264 FIELDIANA: ZOOLOGY (FMNH 156649, 156651); parcel 1, 15.0 km NW of Eminiminy, 24°34.2'S, 46°43.9'E, 1500 m (FMNH 156650); Bemangidy, approximately 72 km N of Tolagnaro, 24°34'S, 47°14'E (FMNH 85192-85194, 85873). Galidictis fasciata fasciata (Gmelin, 1788) Distribution — In the RNI d'Andohahela Gali- dictis fasciata was observed or captured in the 440, 810, 1200, and 1500 m zones. There are widely scattered records of this species in the cen- tral portion of the eastern humid forest from the RNI de Zahamena south to the region between Farafangana and Manakara (N. Rakotoarison, pers. coram.; Albignac, 1973). The previous southernmost record appears to be Vondrozo (BMNH 1935.01.08.305, MNHN 1932-3539). Our records from the RNI d'Andohahela extend the distribution of G. fasciata about 250 km fur- ther south. Ecology and Reproduction — This carnivore was occasionally seen at night on trails within the forest and in and around our camps. An adult fe- male and a young female were taken in the same trap set on consecutive days. The trap site was on the ground in a passageway under a fallen and rotten log. This species was trapped nearly as of- ten as it was seen, almost certainly because of its secretive nocturnal behavior. The four individuals captured included an adult male, two adult females, and a subadult female. The adult female had a single pair of small in- guinal mammae, and the adult male had scrotal testes measuring 25 X 12 mm with slightly con- voluted epididymides. The subadult female showed no signs of reproductive activity. Comments — Trapping showed Galidictis fas- ciata to be most common in the 1500 m zone. Although it is difficult to compare relative den- sities of carnivores along our elevational transect, G. fasciata seems to be the second most common species of carnivore in parcel 1 of the RNI d'Andohahela. It occurs from lowland to montane forest. Galidictis f fasciata is the subspecies found in the southern portion of the eastern humid forest (Albignac, 1973, p. 46, contra fig. 5). The local Tanosy people living near the eastern border of parcel 1 call this species vontsira fotsy. Specimens Examined — Parcel 1, 13.5 km NW of Eminiminy, 24°35.0'S, 46°44.1'E, 1200 m (FMNH 156652); parcel 1, 15.0 km NW of Em- iniminy, 24°34.2'S, 46°43.9'E, 1500 m (FMNH 156549, 156653). Family Viverridae Subfamily Cryptoproctinae Cryptoprocta ferox Bennett, 1833 Distribution — We found evidence of Crypto- procta ferox at two sites in the RNI d'Andohahela. Scats identifiable to this species were found on the trail below the 440 m camp, at 1850 m at the base of the hump of Pic Trafon- aomby, and at about 1800 m in an open area by a water hole. This species has a broad distribution across much of Madagascar, in and near natural forested habitats from sea level to high mountain zones above the forest line (Albignac, 1973; Goodman et al., 1997). Previous records of C. fe- rox from the RNI d'Andohahela include reports from forest guardians or local villagers for parcels 1 and 2 (O'Connor et al., 1987). People living in rural areas near forested areas know this animal well and note that it occasionally feeds on village chickens. Comments — The scat found near the 440 m camp contained fur remains of Hapalemur gri- seus. Cryptoprocta has been reported feeding on a wide variety of lemur species (Goodman et al., 1993; Rasoloarison et al. 1995; Goodman et al., 1997; Wright et al., 1997); but as far as we can determine this is the first evidence of it preying on H. griseus. The scat from the base of Trafon- aomby held bones of at least two Oryzorictes hova, one Microgale thomasi, and two M. cowani. All of these lipotyphlans were trapped in the 1875 m zone and weighed less than 35 g (see Chapter 13). Cryptoprocta living in other high mountain zones show a dramatic drop in mean body size of prey in comparison to those in lower lying areas of forest (Goodman et al., 1997). Subfamily Euplerinae Eupleres goudotii Doyere, 1835 Distribution — Eupleres goudotii was seen once during the 1995 inventory of the RNI d'Andohahela. On 15 November a single individ- ual was observed by F. Hawkins in the 1200 m zone. The animal was moving through a valley bottom at 1615 hr. GOODMAN & PIDGEON: CARNIVORA 265 The distribution of Eupleres goudotii is poorly known, but it has been recorded from scattered localities in the eastern humid forest from Mon- tagne d'Ambre (Albignac, 1973; Nicoll & Lan- grand, 1989) south to the RNI d'Andringitra (Goodman, 1996). It has also been reported in dry forest from the Sambirano south to Baly (Albig- nac, 1973; Hawkins, 1994). Fossa fossana (Miiller, 1776) Distribution — This species was recorded dur- ing the 1995 inventory of the RNI d'Andohahela in the 440, 810, and 1200 m zones. It has been reported from numerous localities in the eastern humid forest from the Antalaha area south to the RNI d'Andringitra and Reserve Speciale (RS) du Pic dTvohibe (Albignac, 1973; Goodman, 1996), as well as near Vondrozo (Rand, 1935). Forest guardians and local villagers living near parcel 1 of the RNI d'Andohahela noted the presence of this species in the reserve (O'Connor et al., 1987). We have also observed this species in the littoral forest near Manafiafy. Ecology and Reproduction — The single in- dividual captured was a subadult male with slight- ly scrotal testes measuring 5X3 mm and non- convoluted epididymides. The animal was in the process of replacing the deciduous upper incisors with permanent teeth. All other records of this species in the RNI d'Andohahela were observa- tions made at night. At least twice this species was observed pilfering food stocks in our camps during the night. Comments — In November 1948 Harry Hoogs- traal collected three Fossa fossana in the vicinity of Bemangidy. We examined and measured two of these specimens (Table 16-3). Inscriptions on the specimen labels indicate that one was "trapped with meat bait at edge of original forest" and another "was trapped with fruit bait beside stream at edge of original forest." Specimens Examined — Parcel 1, 13.5 km NW of Eminiminy, 24°35.0'S, 46°44.1'E, 1200 m (FMNH 156648); Bemangidy, approximately 72 km N of Tolagnaro, 24°34'S, 47°14'E (FMNH 85195, 85196; USNM 318107). Subfamily Viverrinae Viverricula indica (Desmarest, 1804) Distribution — This introduced species has been previously reported from parcel 3 of the RNI d'Andohahela (O'Connor et al., 1987); we found it only in parcel 2 during the 1995 mission. Vi- verricula indica is broadly distributed across many areas of the island, particularly in close proximity to rural settlements. It is found in a va- riety of environments from the suburbs of Anta- nanarivo to lowland agricultural areas, humid for- est, and spiny bush habitat. We commonly ob- served this species in the littoral forests near Man- afiafy. Comments — In December 1948 Harry Hoogs- traal collected a specimen of Viverricula indica near Bemangidy that, based on notes inscribed on the label, was "trapped in manioc field [with] fish bait." Specimens Examined — Bemangidy, approxi- mately 72 km N of Tolagnaro, 24°34'S, 47°14'E (FMNH 85201). Discussion The only carnivore species previously reported from the reserve but not documented during the 1995 inventory is Salanoia concolor. (I. Geoffroy Saint-Hilaire, 1837) (O'Connor et al., 1987), a poorly known animal that is apparently confined to humid forest (Albignac, 1973). The report of S. concolor in the RNI d'Andohahela was based on information from a person living close to par- cel 1. This species is rather nondescript and may have been confused with another species of car- nivore, notably Galidia elegans. The presence of S. concolor in the reserve thus needs confirma- tion. The southernmost documented record of S. concolor is from the Perinet area (Albignac, 1973). Five of the eight carnivore species known from parcel 1 of the RNI d'Andohahela are endemic to Madagascar, and the other three are introduced. A comparison of sites in the eastern humid forest that have been relatively well surveyed for car- nivores shows a similar fauna across a broad lat- itudinal area (Table 16-4). The sites of RNI d'Andohahela, RNI d'Andringitra, and the Pare National (PN) de Ranomafana have identical na- tive carnivore faunas. The Reserve de Biosphere (RB) de Mananara has the same native carnivore faunas as these other three sites, with the addition of Salanoia concolor. No native carnivore was recorded in parcel 2 of the RNI d'Andohahela during the 1995 inven- tory, but there is a report of Cryptoprocta ferox 266 FIELDIANA: ZOOLOGY Table 16-4. The known distribution of native carnivores at several sites in the eastern humid forests of Madagascar. RNI RNI PN RB d'Andohahela d'Andringitra de Ranomafana de Mananara Latitude: 24°S 22°S 21°S 16°S Species Source of in formation*: 1 2 3 4 Galidia elegans + + + + Galidictis fasciata + + + + Salanoia concolor - - - + Cryptoprocta ferox + + + + Eupleres goudotii + + + + Fossa fossana + + + + Total number of endemic species 5 5 5 6 * 1 = this study; 2 and Langrand (1989). Goodman (1996); 3 = Nicoll and Langrand (1989) and P. Wright (pers. comm.); 4 = Nicoll occurring there (O'Connor et al., 1987), and there have been local sightings of Galidia elegans (T. Saotsy, pers. comm.). On the basis of current dis- tributional information, no other species of Mal- agasy carnivore would be expected to occur in this parcel. Cryptoprocta is the only native car- nivore that occurs in both parcels 1 and 2. These two sites are separated by an air distance of about 20 km. Three species of introduced carnivores were recorded in parcel 2; two of these are human commensals and may occasionally revert to being feral (Canis lupus and Felis silvestris), and the third lives completely in a feral state (Viverricula indicd). There is little evidence that introduced carnivores are able to penetrate far into humid for- est habitat, and these animals do not pose any apparent threat to the native carnivore popula- tions. Acknowledgments We acknowledge the help of the various mis- sion participants who shared their carnivore ob- servations with us. L. Granjon, Museum National d'Histoire Naturelle, Paris; P. Jenkins, The Natural History Museum, London; and M. Carleton, Na- tional Museum of Natural History, Washington, D.C., kindly made specimens available for study. Two anonymous reviewers provided most useful comments on an earlier version of this chapter. Literature Cited Albignac, R. 1971. Une nouvelle sous-espece de Gali- dia elegans: G. elegans occidentalis (Viverridae de Madagascar). Mise au point de la repartition geograp- hique de l'espece. Mammalia, 35: 307-310. 1973. Faune de Madagascar. 36. Mammiferes: Carnivores. Office de la Recherche Scientifique et Technique Outre-Mer and Centre National de la Re- cherche Scientifique, Paris, 210 pp. + plates. Flacourt, E. de. 1658. [reprinted in 1995]. Histoire de la Grande Isle de Madagascar. Edition annotee et pre- sentee par Claude Allibert. INALCO-Karthala, Paris. 656 pp. Goodman, S. M. 1996. The carnivores of the Reserve Naturelle Integrale d'Andringitra, Madagascar, pp. 289-292. In Goodman, S. M., ed., A floral and faunal inventory of the eastern slopes of the Reserve Natu- relle Integrale d'Andringitra, Madagascar: With ref- erence to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. , O. Langrand, and B. P. N. Rasolonandra- sana. 1997. The food habits of Cryptoprocta ferox in the high mountain zone of the Andringitra Massif, Madagascar (Carnivora, Viverridae). Mammalia, 61: 185-192. Goodman, S. M., S. O'Connor, and O. Langrand. 1993. A review of predation on lemurs: Implications for the evolution of social behavior in small, nocturnal primates, pp. 51-66. In Kappeler, P. M., and J. U. Ganzhorn, eds., Lemur social systems and their eco- logical basis. Plenum Press, New York. Hawkins, F 1994. Eupleres goudotii in west Malagasy deciduous forest. Small Carnivore Conservation, 11: 20. 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. O'Connor, S., M. Pidgeon, and Z. Randria. 1987. Un programme de conservation pour la Reserve d'Andohahela, pp. 31-36. In Mittermeier, R. A., L. A. 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, Gland. Rand, A. L. 1935. On the habits of some Madagascar mammals. Journal of Mammalogy, 16: 89-104. Rasoloarison, R. M, B. P. N. Rasolonandrasana, J. | GOODMAN & PIDGEON: CARNIVORA 267 U. Ganzhorn, and S. M. Goodman. 1995. Predation on vertebrates in the Kirindy Forest, western Mada- gascar. Ecotropica, 1: 59-65. Wozencraft, W. C. 1993. Order Carnivora, pp. 279- 348. In D. E. Wilson and D. M. Reeder, eds.. Mammal species of the world: A taxonomic and geographic ref- erence, 2nd ed. Smithsonian Institution Press, Wash- ington, D.C. Wright, P. C, and L. B. Martin. 1995. Predation, pol- lination and torpor in two nocturnal prosimians: Chei- rogaleus major and Microcebus rufus in the rain foresi of Madagascar, pp. 45-60. In Alterman, L., G. A Doyle, and M. K. Izard, eds., Creatures of the dark: The nocturnal prosimians. Plenum Press, New York. Wright, P. C., S. K. Heckscher, and A. E. Dunham 1997. Predation on Milne-Edward's Sifaka (Propithe- cus diadema edwardsi) by the Fossa (Cryptoproctc ferox) in the rain forest of southeastern Madagascar Folia Primatologia, 68: 34-43. 268 FIELDIANA: ZOOLOGY Chapter 17 Lemurs of the Reserve Naturelle Integrate d'Andohahela, Madagascar Anna T. C. Feistner1 and Jutta Schmid2 Abstract An inventory of the lemur fauna of the Reserve Naturelle Integrate d'Andohahela was con- ducted between 18 October and 16 December 1995. Study sites in parcel 1 of the reserve were located at five different elevations, 440, 810, 1200, 1500, and 1875 m, up an altitudinal gradient in humid forest. In addition, one site was located at 120 m elevation within the spiny forest portion of the reserve (parcel 2). At each site the presence and relative abundance of lemur species were estimated using the line transect method. A total of eight lemur taxa were recorded in parcel 1 (Propithecus verreauxi verreaiixi, Eulemur fidvus collaris, Hapalemur griseus, Mi- crocebus rufus, Cheirogaleus major, Avahi laniger, Lepilemur mustelinus, and Daubentonia madagascariensis). Species richness (eight species) was highest at the lowest elevation (440 m) and decreased to a relatively constant level (four to six species) for the four higher eleva- tional zones. The single observed group of P. v. verreauxi, which is generally a species of the dry deciduous forest and spiny thorn scrub, contained some individuals of the dark (P. v. 'majorf) morph. Distinctive signs of feeding by Daubentonia were recorded at every eleva- tional level. Four species of lemurs were sighted in the spiny forest (P. v. verreauxi. Lemur catta, Microcebus murinus, and Lepilemur leucopus). A fifth species, Phaner furcifer ssp., was recorded by vocalizations only. In both wet and dry forest parcels, further efforts are necessary to ensure long-term protection of the RNI d'Andohahela and prevent the degradation of its fauna and flora. There was evidence that E. f. collaris is hunted within the reserve, although the RNI d'Andohahela is the only protected area in Madagascar where this taxon occurs nat- urally. Resume Un inventaire des lemuriens de la Reserve Naturelle Integrate d'Andohahela a ete effectue entre le 18 octobre et 16 decembre 1995. Les sites d'etudes sont situes a cinq differentes altitudes, 440 m, 810 m, 1200 m, 1500 m et 1875 m, le long de la pente Est de la Parcelle 1 de la Reserve, une foret humide de 63.000 ha. En outre, un site est situe a une altitude de 120 m dans une foret epineuse de 12.920 ha de la Parcelle 2. Dans chaque site, une presence et une abondance relative d'especes de lemuriens ont ete estimees en utilisant la methode de transect lineaire. Un total de 8 taxa de lemuriens est inventorie dans la foret humide (Propi- thecus verreauxi verreauxi, Eulemur fulvus collaris, Hapalemur griseus, Microcebus rufus, Cheirogaleus major, Avahi laniger, Lepilemur mustelinus, et Daubentonia madagascariensis). La zone la plus riche en especes se trouve a la plus basse altitude (440 m). Cette richesse 1 Jersey Wildlife Preservation Trust, Les Augres Manor, Trinity, Jersey JE3 58R Channel Islands. 2 Deutsches Primatenzentrum, Kellnerweg 4. 37077 Gottingen. Germany, and Abteilung fiir Verhaltenphysiologie, Beim Kupferhammer 8. 72070 Tubingen. Germany. FEISTNER & SCHMID: LEMURS 269 diminue mais ensuite reste constante pour les quatre zones de niveau superieur. Le seul groupe de Propithecus verreauxi verreauxi, generalement une espece rencontree en foret seche et broussailles epineuses, contient quelques individus de couleur foncee (' major?). Des signes distinctifs d' alimentation par Daubentonia sont inventories a chaque niveau altitudinal de la foret humide. Dans la foret epineuse, on a distingue quatre especes de lemuriens (P. v. ver- reauxi, Lemur catta, Microcebus murinus, et Lepilemur leucopus). Une cinquieme espece, Pha- ner furcifer ssp., a ete inventorize par vocalisations seulement. Dans les Parcelles de foret humide et seche, des efforts supplementaires doivent etre deployes afin d' assurer une protection a long terme de la Reserve et empecher la degradation de sa faune et de sa flore. On a decele des preuves de chasse de E.f. collaris a l'interieur de la Reserve, pourtant la RNI d'Andohahela est la seule aire protegee a Madagascar ou le taxon apparait d'une maniere naturelle. Introduction Madagascar has been isolated from other land masses for some 88 million years; consequently its flora and fauna have evolved in isolation (Sto- rey et al., 1995). Endemism is extremely high. Madagascar is considered one of the world's ma- jor "hotspots" of biodiversity (Myers, 1988, 1990). One of the major adaptive radiations in Madagascar is among the primates. The primate fauna is highly diverse (comprising about 32 spe- cies) and entirely endemic (the lemurs found on the Comoro Islands almost certainly were intro- duced). Madagascar is also a high priority for conservation efforts because many of its unique environments are under threat. The primate fauna is restricted to forest zones (eastern humid forests, western dry forests, and southern spiny forests). These forests are disap- pearing as land is cleared for agricultural purpos- es, as trees are felled to provide building materials and charcoal, and as a result of commercial log- ging. Moreover, some lemurs are also hunted for food. In order to develop effective conservation strat- egies for the unique habitats in Madagascar, in- formation on the distribution and abundance of their faunas and floras is crucial. Madagascar al- ready has a system of protected areas, but accu- rate inventories and effective protection are lack- ing for many of the reserves. As part of a program to assess biodiversity within the protected areas of Madagascar, inter- national teams of scientists have been undertaking floral and faunal inventories in a variety of humid and dry forest sites (Goodman & Langrand, 1994; Goodman, 1996; Langrand & Goodman, 1997; Goodman, 1998). These surveys not only provide information important for conservation planning, but they also help in developing an understanding of patterns of diversity and variation with latitude, relief, and elevation. The work reported here was carried out in the Reserve Naturelle Integrate (RNI) d'Andohahela. This reserve is of particular interest and impor- tance because it includes the southernmost tropi- cal humid forests in the Old World, traverses the east-west divide in Madagascar, and encompasses areas of dry spiny forest. The scant information on lemur diversity and density in the RNI d'Andohahela is inconsistent (Nicoll & Langrand, 1989; Mittermeier et al., 1992, 1994) in that the published primate species lists contain different species (see especially Mittermeier et al., 1992, 1994). The reserve is known to be important for at least one lemur, the Collared Brown Lemur Eu- lemur fulvus collaris, because it represents the only protected area in which this taxon occurs (but see Sterling & Ramaroson, 1996). The aim of our work on lemurs was to assess the diversity and abundance of that fauna in the RNI d'Andohahela. Most of our effort was con- centrated in the eastern humid forest (parcel 1), where assessments were undertaken at a variety of elevations on the eastern slopes of the Ando- hahela Massif. To complete the primate inventory, assessments were also undertaken in the dry west- ern spiny forest (parcel 2). Materials and Methods In order to enhance the validity of comparisons between different protected areas, methods for rapid assessment of the primate fauna were stan- dardized across forest sites. The methods used in the RNI d'Andohahela were the same as those used in the RNI d'Andringitra (Sterling & Ra- maroson, 1996), the Reserve Speciale (RS) 270 FIELDIANA: ZOOLOGY Table 17-1. Description of the different habitat types found along the trails walked during censuses at each site in the RNI d'Andohahela. Trails at 440-1875 m were located in the humid forest of parcel 1; those at 120 m were in the spiny forest of parcel 2. Trail Trail length (m) 440a 1200 440b 1150 810a 1750 810b 600 1200a 950 1200b 1000 1500a 1200 1500b 725 1500c 365 1875a 190 1875b 1875c 120a 810 690 3500 120b 3000 120c 900 General habitat characteristics Open forest with a high proportion of lianas (canopy 11-15 m); slope: dense steep forest (canopy 8-12 m) Open forest with dense lianas; slope: forest (canopy 10-12 m); open ridge forest. (degraded) riverine forest Riverine forest: dense ground vegetation, bamboo, tree ferns (canopy 5-6 m); open forest (canopy 12-15 m) Slope: steep forest (canopy 8-10 m); ridge: open forest (canopy 15-20 m). bamboo and grassy bamboo: rocks on one side: slope: riverine forest Open forest: no ground vegetation (canopy 12-15 m): ridge: mossy forest (canopy 8- 10 m). bamboo, tree ferns (canopy 8 m); slope: steep and humid forest (canopy 6 m) Open forest (canopy 10-12 m): ground vegetation, leaf litter, liana: slope: rocks, open forest; valley: open riverine forest, grassy bamboo Ridge: open forest (canopy 6-8 m), no ground vegetation: bamboo and grassy bam- boo Slope: steep and open forest (canopy 10 m); ridge: grassy bamboo, mossy, few trees (canopy 4-5 m) Ridge: open forest (canopy 8-10 m), thick ground vegetation; tree ferns (canopy 8 m) Ridge: mountain forest (canopy 2-4 m). mossy, ground vegetation 30-50 cm. rocky. moss, sedge Ridge: moss forest (canopy 4-8 m). rocky, lianas Ridge: moss and bamboo forest, sedge plateau Main road: degraded spiny forest (canopy 2-3 m), sisal patches; streamside: spiny and gallery forest with baobabs (canopy 5-10 m) Slope: spiny forest; ridge: spiny forest (canopy 2-3 m); main road: patchy spiny for- est, degraded Degraded spiny forest d'Anjanaharibe-Sud (Schmid & Smolker, 1998), and on the Masoala Peninsula (Sterling & Rako- toarison, 1998). Study Sites Censuses for lemurs were conducted at five el- evations in the humid forest along the eastern slope of the Andohahela Massif (parcel 1) be- tween 18 October and 5 December 1995. In ad- dition, an inventory of lemur species was under- taken at one site in the spiny forest of parcel 2 of the same reserve between 8 and 14 December 1995. The team, consisting of two researchers ex- perienced in observing Malagasy primates, visited each of the six sites for a minimum of 7 days. Survey sites of parcel 1 were located at five different elevations. All transects were within 75 m elevation above or below sites centered at 440, 810, 1200, and 1500 m. Due to the lack of water, the fifth camp site was located at 1710 m but the transect zone was centered at 1875 m. The sixth site, in the spiny forest of parcel 2, was located east of Hazofotsy at an altitude of 120 m (see Chapter 1). The line transect method was employed (Na- tional Research Council, 1981). At each location two or three trails of varying lengths (humid for- est: 190-1750 m; spiny forest: 900-4000 m) were used for lemur surveys. We utilized preexisting trails left by bush pigs, cows, and people, as well as newly cut trails. We attempted to select trails that covered a variety of forest habitats, including ridges, slopes, valleys, and stream/river courses. A general description of the forest type along each trail is given in Table 17-1. Each trail was marked in 20 or 25 m cumulative intervals with flagging tape. Two researchers walked either singly or in tan- dem slowly (ca. 0.6 km/hr) along the trails to cen- sus lemurs. Diurnal censuses took place at times of increased lemur activity, in the morning (0600- 1130 hr) and in the afternoon (1500-1730 hr). FEISTNER & SCHMID: LEMURS 271 Walks were always separated by a time interval of at least 6 hr when trails were censused twice per day during daylight hours. For nocturnal sam- ples, trails were walked after dark, between 1 830 and 2230 hr. Whenever possible, trails were walked from the direction opposite that of the pre- vious sample to reduce potential biases (Brock- elman & Ali, 1987). When the terrain was diffi- cult, we paused fairly regularly (approximately every 25 m) to watch and listen for signs of pri- mate presence, such as vocalizations or move- ments in the vegetation. At night the dim light of a headlamp was used to pick up the eyeshine from the reflective tapetum lucidum of nocturnal le- murs. Once detected, a more powerful handheld flashlight and binoculars (7 X 40) were used for species identification. For all observations we noted the species, time of contact, position on the transect, elevation, dis- tance from the observer, angle, height from the ground, and habitat type. The perpendicular dis- tance from the trail was estimated for the first in- dividual seen in each group. Whenever possible, we also recorded the number of individuals, age/ sex composition, and general behavior. No more than 10 min were spent for any single sighting. In the humid forest, censusing along each transect was repeated three to four times (except trail 1 875c) for nocturnal transects and nine to thirteen times (cf . trail 1 875c) for diurnal transects. In the spiny forest, trails were censused one to three times at night and three to nine times in daylight. We did not conduct census walks when the view- ing distance was restricted to less than 15 m, as was sometimes the case owing to dense fog or heavy rain. Inadequate sample sizes, such as the few rep- etitions of each transect and the relatively short distances covered at each site, prevented us from determining density values (Whitesides et al., 1988; Ganzhorn, 1992, 1994; Sterling & Rama- roson, 1996; Schmid & Smolker, 1998). The mean numbers of lemur sightings per km transect were calculated, however. In addition, the mean detec- tion distance, perpendicular to the trail, at which lemurs were seen was compiled for each species and trail. For each species where we did not find statistically significant differences across trails per site, a single average detection distance was cal- culated. For diurnal censuses the mean number of groups and for nocturnal censuses the mean num- ber of individuals observed within the transects were determined. Lemurs heard but not seen dur- ing census walks or those seen outside census walks by us or other researchers were not includ- ed in calculations of encounter rate for either di- urnal or nocturnal surveys. Species accumulation curves were based on sightings and vocalizations during transect walks. Additional Observations Apart from the systematic transect surveys, oth- er general observations were made during the day. Within each elevational zone we explored the for- est away from the trail system used for transects to look for secondary signs of the presence of certain primates. These included characteristic feeding signs of Daubentonia madagascariensis (gnaw marks from excavation of dead wood, liv- ing branches, or bamboo) or sleeping sites for nocturnal species (e.g., nests for Daubentonia or tree holes for Cheirogaleus or Microcebus). Within the 1875 m zone there were many signs of Daubentonia feeding on bamboo stems. We therefore decided to investigate the stems to see if we could locate their potential prey. A bamboo stem was cut less than 30 cm above the ground. Each segment of the stem, until it became very thin and leaves sprouted, was opened, and the contents, if any, were examined. Any inverte- brates contained within the stem were collected and preserved in alcohol. In two areas located ap- proximately 60 m and 120 m from trail 1875a, respectively, a total of 25 intact bamboo stems (5 m apart), ranging in length from 2.5 to 6.0 m, were examined. In total, 333 segments were opened. Results Species Accumulation Curves Species accumulation curves are shown in Fig- ure 17-1. For nocturnal censuses at the survey sites in the humid forest of the RNI d'Andohahela, no additional species were record- ed in the 440 and 810 m zones after 3 hr of ob- servation (Fig. 17-la,b). In the 1500 and 1875 m zones all lemurs were recorded after only 2 hr of observation (Fig. 17-ld,e), whereas at 1200 m all nocturnal lemurs were recorded after 1 hr of cen- sus walks (Fig. 17-lc). The species accumulation curve for diurnal lemurs was much more variable. At 440 m the curve did not plateau until 30 hr of 272 FIELDIANA: ZOOLOGY observation (Fig. 17- la). This was due to the un- expected sighting of Propithecus v. verreauxi that occurred on the last day of censusing. At 810 m all diurnal lemurs were encountered within 2 hr, whereas at 1200 m 12 hr were required (Fig. 17- lb, c). At 1500 m only 2 hr were needed, but at the highest elevation (1875 m) 9 hr of censusing was necessary before all species were recorded (Fig. 17-ld,e). At the survey site in the spiny forest, the spe- cies accumulation curve for lemurs recorded dur- ing nocturnal censuses reached its plateau after only 1 hr of observation (Fig. 17- If)- For diurnal censuses, we recorded two species after 4 hr of censusing, and although walks for day censusing totaled 52 hr, no additional species were seen (Fig. 17-lf). of these were diurnal {Propithecus verreauxi ver- reauxi and Lemur catta) and three were nocturnal {Microcebus murinus, Lepilemur leucopus, and Phaner furcifer ssp.) (Table 17-2). Sleeping P. v. verreauxi were occasionally encountered during nocturnal censuses. Propithecus v. verreauxi, M. rufus, and L mustelinus were observed directly. Lemur catta was identified from vocalizations heard during the transect walks, although a group was encountered during general observations. Phaner furcifer ssp. was identified only from vo- calizations. Table 17-5 lists the lengths of tran- sects, numbers of census walks, and observed species for night and day censuses in the spiny forest. During night surveys a total of 18.9 km was walked, with 65.7 km walked during the day surveys. Species Diversity In total, one diurnal species {Propithecus ver- reauxi verreauxi), two cathemeral species {Eule- mur fulvus collaris and Hapalemur griseus), and five typically nocturnal species {Microcebus ru- fus, Cheirogaleus major, Avahi laniger, Lepile- mur mustelinus, and Daubentonia madagascari- ensis) were found in five elevational zones of par- cel 1 of the RNI d'Andohahela (Table 17-2). All taxa were directly observed except for Dauben- tonia, which was recorded indirectly by from its characteristic feeding signs (Duckworth, 1993; Erickson, 1995). Species diversity was highest in the 440 m zone, where all eight lemur species recorded in parcel 1 were found. Five lemur spe- ! cies were present in the 810 and 1200 m zones, : six species in the 1500 m zone, and only four in I the 1875 m zone. Eulemur f. collaris, M. rufus, i C. major, and D. madagascariensis were found at j all elevations. Hapalemur griseus was absent from the highest elevational zone. Avahi laniger was seen in the 440 m and 1500 m zones, and L. mustelinus was only present at 440 m. Propithe- cus v. verreauxi is generally considered a dry for- j est species, and our single record of it at 440 m ! was exceptional. The lengths of transects, num- ! bers of census walks, and species recorded in the various elevational zones of parcel 1 are listed in Table 17-3 for nocturnal censuses and in Table 17- 4 for diurnal censuses. In total, we walked 39.2 km during nocturnal censuses and 117.2 km dur- ing diurnal censuses. Five lemur species were recorded in the spiny forest of parcel 2 of the RNI d'Andohahela; two Species Descriptions Pelage and morphological characteristics as well as reports on activity or social behavior were derived for the lemurs from both systematic and general field observations. Sexual differences in the morphological characters are restricted to di- urnal species. Any observations related to breed- ing activity are also included under the species descriptions. Microcebus rufus (Rufous Mouse Lemur) Microcebus rufus was found at all altitudes in the humid forest. The mean encounter rate re- mained approximately constant across the eleva- tional range from 440 to 1500 m (Table 17-3), but it dropped off at 1 875 m, where there were only two sightings for 4.5 km of trail walked. The mouse lemurs had relatively small ears; they were brown to rufous on the dorsum and whitish on the ventrum. The head was reddish in color and the nose was not very prominent. Microcebus rufus were almost always seen singly and were very active, with rapid movements. They were gener- ally found in dense bushy vegetation, as well as in the taller trees of the forest. Microcebus murinus (Grey Mouse Lemur) Census results showed that Microcebus muri- nus was the most common nocturnal lemur spe- FEISTNER & SCHMID: LEMURS 273 10 15 20 25 30 Hours of observations (hr) 35 40 4 6 8 10 Hours of observations (hr) D) 1500 m 4 r 12 hr -© 29 hr 0 5 10 15 20 25 30 Hours of observations (hr) 11 hr 4 6 8 10 Hours of observations (hr) 35 33 hr 33 Hours of observations (hr) 18 20 Hours of observations (hr) 52 hr 18 52 Fig. 17-1. Species accumulation curves as a function of observation hours for nocturnal (•) and diurnal (O) lemui species censused in the RNI d'Andohahela. Humid forest at 440 m (A), 810 m (B), 1200 m (C), 1500 m (D), and 1800 m (E) elevation zones, spiny forest at 120 m (F). 274 FIELDIANA: ZOOLOGY Table 17-2. The primate species found in the humid forest (parcel 1) and spiny bush (parcel 2) of RNI d'Andohahela listed by elevational zone. Species were recorded during survey walks or during additional observa- tions. Parcel 1 Parcel 2 Species 440 m 810 m 1200 m 1500 m 1875 m 120 m Microcebus rufus + + + + + Microcebus murinus + Cheirogaleus major + + + + + Phaner furcifer ssp. + * Avahi laniger + + Lepilemur mustelinus + Lepilemur leucopus + Daubentonia madagascariensis fd fd fd fd fd Hapalemur griseus + + + + Lemur carta + Eulemur fulvus collaris + + + + + Propithecus verreauxi verreauxi + + Total numbers of species 8 5 5 6 4 5 A plus sign ( + ) indicates that the given species was present: the abbreviation fd indicates that the presence of this species was determined by analyzing feeding damage to bamboo or wood. * The presence of this species was determined only through the identification of its vocalizations. cies in parcel 2 of the RNI d'Andohahela (Table 17-5). The body pelage was gray on the dorsum and whitish on the ventrum. A thin black dorsal stripe was often visible. The ears of M. murinus were large and prominent, in contrast to the small- er and more concealed ears of M. rufus. Cheirogaleus major (Greater Dwarf Lemur) Cheirogaleus major was observed from 440 to 1500 m by direct sightings during census walks in parcel 1 (Table 17-2) and was the most com- mon primate species at these elevations. One in- Table 17-3. Mean number of sightings per km transect and mean detection distances (m) (± standard deviation) of species (individuals) seen during nocturnal censuses in each elevational zone (parcel 1) in the RNI d'Andohahela. Length of Number transects of (m) censuses Microcebus rufus Cheirogaleus major Avahi laniger Lepilemur mustelinus species Trail Mean (±SD) detection distance (m) Number of 440a 1200 4 1.3 (9.0 ± 4.0) 2.9 (7.0 ± 3.4) - 440b 1150 4 1.7 (5.7 ± 3.4) 2.8 (8.3 ± 2.3) 0.4 (8.0) 810a 1750 4 0.9 (2.3 ± 1.5) 0.7 (2.3 ± 2.3) - 810b 600 4 - - - 1200a 950 4 0.5 (3.5 ± 0.7) 0.8 (5.0 ± 1.0) - 1200b 1000 4 - 1.5 (5.7 ± 2.8) - 1500a 1200 4 0.6 (1.0 ± 0.0) 0.2 (10.0) 0.6 (1.0) 1500b 725 3 2.3 (1.2 ± 2.1) 1.8 (6.3 ± 1.3) - 1500c 365 3 - 0.9 (1.0) - 1875a 190 3 - - - 1875b 810 4 0.6 (2.3 ±1.1) - - 1875c 690 1 - - - Overall mean detection distance (m) 4.0 ± 3.7 n = 30 6.2 : 3.2 40 3.8 ± 3. n = 5 7.0 ± 1.2 Detection distances are given perpendicular to the trail, n the species was not detected. number of individuals. A minus sign (-) indicates that FEISTNER & SCHMID: LEMURS 275 Table 17-4. Mean number of sightings per km transect and mean detection distances (m) (± standard deviation) of species (individuals) seen during diurnal censuses in each elevational zone (parcel 1) in the RNI d'Andohahela. Trail Mean (±SD) detection distance (m) Length of Number of transects (m) censuses Propithecus verreauxi verreauxi (+ imajorV) Eulemur fulvus collaris Hapalemur griseus Number of species 440a 1200 11 440b 1150 11 810a 1750 11 810b 600 10 1200a 950 13 1200b 1000 12 1500a 1200 13 1500b 725 12 1500c 365 11 1875a 190 9 1875b 810 10 1875c 690 5 0.1 (10.0) 0.2 (8.0 ± 1.0) 0.1 (15.0) 2 - 0.2 (6.2 ± 5.0) 2 0.2 (5.0 ± 6.2) 0.3 (2.8 ± 3.2) 2 - 0.2 (12.0) 1 0.7 (7.0 ± 6.3) 0.1 (3.0) 2 0.3 (9.3 ± 7.4) 0.1 (3.0) 2 - 0.2 (4.3 ± 3.5) 1 - 0.1 (3.0) 1 - 0.2 (6.0) 1 - - 0 0.1 (10.0) Overall mean detection distance (m) 10.0 n = 1 7.3 ± 5.6 n = 18 5.0 ± 4.3 n = 18 Detection distances are given perpendicular to the trail, n = number of individuals. A minus sign (— ) indicates that the species was not detected. dividual was also seen in the 1875 m zone outside survey walks (F. Hawkins, pers. coram.). The dor- sal pelage of C. major was gray-brown with red- dish tinges, and its underparts were paler, almost white. The head was gray-brown, and a small patch of white fur was located between the eyes, which were surrounded by marked dark rings. Most of the C. major observed had very fat tails. Cheirogaleus major was generally observed alone. However, in five (15%) of 40 sightings two individuals were seen less than 3 m from each other in the same tree. We never saw more than two individuals in close proximity. Avahi laniger (Eastern Woolly Lemur) Avahi laniger was only sighted twice in parcel 1: once at 440 m and once at 1500 m (Table 17- 3). It was never heard calling. Dorsal pelage was gray and the ventral parts were paler gray. The tail was gray-brown and darkened toward the tip. The face was brownish with a whitish band and distinct white patches above the eyes. The fur on the cheeks and throat was lighter colored. Two adults were seen in the same tree at 440 m. Two adults with a single offspring were observed at 1500 m. The group huddled together in the char- acteristic vertical resting posture in the fork of a tree, and the infant was carried on the back of one adult. Lepilemur mustelinus (Weasel Sportive Lemur) Lepilemur mustelinus was also detected only twice, with each sighting at 440 m (Table 17-3). This species and Avahi laniger were the rarest species in parcel 1 of the RNI d'Andohahela. On each occasion an adult L. mustelinus with a single offspring on its back was seen. The dorsal and ventral fur of the two adult individuals was brown, and the tail color darkened distally. Lepilemur leucopus (White-footed Sportive Lemur) Lepilemur leucopus was very abundant in the spiny forest of parcel 2. The dorsal fur of L. leu- copus was gray, with the underparts paler and sometimes tipped with brown. The arms and the tail were both reddish brown, whereas the thighs often showed gray tops. No infants were seen. 276 FIELDIANA: ZOOLOGY Table 17-5. Mean number of sightings, per kilometer transect, of diurnal (groups) and nocturnal (individuals) lemurs and mean detection distances (± standard deviation) of species seen in censuses in the spiny forest (parcel 2) of the RNI d'Andohahela. Length of transects (m) Nocturnal censusus Diurnal censuses No. of censuses Mean (±SD) detection distance (m) No. of species No. of censuses Mean (±SD) detection distance (m) . No. Propithecus v. verreauxi of Trail Lepilemur leucopus Microcebus murinus species 120a 120b 120c 3500 3000 900 3 2 1 1.1 (4.9 ± 2.0) 0.6 (2.2 ± 2.1) 0.1 (2) 0.7 (4.2 ± 1.5) 2.2 (4.0 ± 5.7) 2 2 1 9 9 3 1.3 (12.2 ± 5.6) 0.03 (13.5 ± 9.2) 1 1 0 Total detection distance (m) Total detection distance (m) 4.7 ± 2.1 3.2 ± 2.5 n = 13 n = 12 12.6 ± 5.9 n = 7 Detection distances (m) are given perpendicular to the trail, n = number of individuals. A minus sign ( that the species was not detected. ) indicates Daubentonia madagascariensis (Aye-aye) Although we did not observe Daubentonia mad- agascariensis directly, their presence was evident in parcel 1 of the RNI d'Andohahela. Two types of damage attributable to this species were recorded between 440 and 1875 m. First, holes with notches around them that resembled incisor gnaw-marks were found in living and dead tree trunks in all ele- vational zones. Similar holes were also found in rot- ten pieces of wood on the ground. Second, we also found holes in bamboo stems wherever this plant was abundant. The bamboo internodes had been punctured and ripped back, producing a hole 1-2 cm long X 0.5 cm wide that exposed the cavity within (as illustrated in Duckworth, 1993). Tooth- marks were present at the top of the rip. Occasion- ally the ripped-back slivers of bamboo were still attached to the holes. They were often moist and fresh, indicating that the holes had been gnawed re- cently. Most of the bamboo stems had several traces of excavation that were found at heights of 1-5 m. At 810 m we found seeds of Canarium mada- gascariense (Burseraceae), a known food item for Daubentonia madagascariensis (Iwano & Iwaka- wa, 1988; Sterling et al., 1994; Goodman & Ster- ling, 1996), and checked them for feeding traces. All 37 seeds found and collected had been opened by rodents, and none exhibited traces character- istic of Daubentonia. Hapalemur griseus (Bamboo Lemur) Hapalemur griseus was found between 440 and 1500 m along the eastern slope of the Andohahela Massif (Table 17-4). Its dorsal fur was dark gray, slightly tipped with olive-brown. The ventral fur was more brownish in color, and the crown was brownish red. These pelage characteristics did not correspond completely with those of typical H. g. griseus (Tattersall, 1982; Mittermeier et al., 1994). The individuals seen in the humid forest of RNI d'Andohahela were considerably larger than typ- ical H. g. griseus and somewhat darker in pelage. Hapalemur griseus was very cryptic. Often only two or three individuals were seen, but vo- calizations indicated the presence of other group members. Group sizes ranged from one to six in- dividuals, corresponding to previous reports with regard to this species (Petter et al., 1977; Tatter- sall, 1982; Pollock, 1986). Adults, subadults, and juveniles of less than 1 year (about 50% of adult body size) were seen. On one occasion an adult female was observed carrying a small brown-col- ored infant (head + body ca. 13-15 cm) on her back. Hapalemur griseus were observed feeding on bamboo. Their vocalizations were heard throughout the day, with a peak before dawn and in the early morning. Lemur catta (Ring-tailed Lemur) Lemur catta was heard but not seen during cen- sus walks in parcel 2 of the reserve. A group of L. catta was encountered, however, outside of the survey walks. The group consisted of six to eight individuals, with multiple adult males and only one clearly identified female. One well-grown in- FEISTNER & SCHMID: LEMURS 277 fant was seen riding on the female's back. One of the males had a damaged/infected left eye. The dorsal fur of L. catta was gray, as were the limbs and haunches. The face was white, with a black nose and dark triangular patches around the eyes. The tail showed the typical alternating black and white bands. Some members of the group were foraging on the ground, whereas the others stayed in the trees. Reaction to our presence varied from alarm calling to immediate flight. Eulemur fulvus collaris (Collared Brown Lemur) Eulemur fulvus collaris was found at altitudes of 440, 810, 1200, and 1875 m in parcel 1 of the RNI d'Andohahela. Surprisingly, we did not find them in the 1500 m zone. These lemurs showed a distinct sexual dimorphism in pelage coloration. Males had a brownish gray coat with a lighter colored ventrum, and a dark stripe down the back was often visible. The head, muzzle, and forehead were black. Males possessed a thick and dense beard that was orange to rufous in color and sometimes very bright. The dorsal fur color of female E. f. collaris was more rufous to brown, and the face was gray compared to that of the males, which was black. The beard of females was also reddish but shorter and less dense than that of males. Both sexes had completely dark tails. The four individuals seen at 1875 m had distinctly thicker pelage and a woollier appear- ance than those seen at the lower elevations. The single male we spotted in this zone also had a very bright orange beard and cheeks. The small numbers observed and the few re- peated sightings of Eulemur f. collaris within each transect zone made it impossible to determine the actual number of groups, and we are unable to give precise details on group size. When all re- cords of this species are summarized, however, group size ranged from three to seven adults, with multiple males and females. Adults were seen car- rying single offspring in virtually every group. In- fants were always carried by adult females, with the single exception of an infant seen riding on the back of a male. No twins were observed. This lemur was recorded only during diurnal censuses, but other observations during this sur- vey showed the species to be active at night. Eu- lemur fulvus collaris should thus be considered cathemeral, pending further investigation. Propithecus verreauxi verreauxi (Verreaux's Sifaka) Propithecus verreauxi verreauxi, generally considered to be a species of only dry forest (Tat- tersall, 1982; Harcourt & Thornback, 1990), was seen in the humid forest at parcel 1 as well as in the spiny forest of parcel 2. The pelage coloration of the P. v. verreauxi found in the humid forest did not differ from that of those seen in the spiny forest. On the dorsum, outer thighs, and forelimbs the body fur was white. The inner and ventral parts were darker and somewhat grayish, and the lemurs had a brownish cap and a black face. In the 440 m zone of parcel 1 , however, we recorded one group of six individual Propithecus that con- sisted of three typical white P. v. verreauxi forms and three "dark," or melanistic forms, previously called 'P. v. majorV (Mittermeier et al., 1994). The three 'dark' individuals were also predomi- nantly white, but they had dark brown fur on the back (dorsum) and on the inside of the extremi- ties. Their ventral fur was brownish, compared to the grayish colored ventrum of the "white" ani- mals. The face was black and the head cap was reddish brown to brown. All individuals were adult-sized, and one individual of the P. v. ver- reauxi "white form" carried a single infant on its back. Some members of the group started produc- ing the "sifaka" vocalization when one observer approached. In general, however, the group re- mained relatively calm, and no flight reaction oc- curred during the 10 min of observation. Opened Bamboo Segments The majority of the 333 bamboo segments ex- amined were empty. Eighteen (5.4%) segments, however, contained some form of animal life: white caterpillars (2-3 cm; n = 7), green cater- pillars (n = 4), flatworms (n = 1), spiders (n = 2), and a frog (n = 1). In three sections (0.9%) we found large (3-5 cm) insect larvae, which are a preferred food item of Daubentonia (Sterling et al., 1994; Erickson, 1995). Water was found in- side 33 (9.9%) of the 333 bamboo segments. Discussion During the surveys in the humid forest (parcel 1) of RNI d'Andohahela eight lemur species were 278 FIELDIANA: ZOOLOGY recorded: Microcebus rufus, Cheirogaleus major, Avahi laniger, Lepilemur mustelinus, Daubento- nia madagascariensis, Hapalemur griseus, Eule- mur fulvus collaris, and Propithecus verreauxi verreauxi (both white and melanistic). Previous workers have listed a variety of species present in the reserve (O'Connor et al., 1987; Nicoll & Lan- grand, 1989; Mittermeier et al., 1992, 1994). In addition to the species reported here, Mittermeier et al. (1992, p. 11; 1994, p. 280) listed Propithe- cus diadema edwardsi, Propithecus diadema dia- dema. Lemur catta, Eulemur rubriventer, Varecia variegata variegata, and Indri indri as occurring in this parcel. Their lists are clearly inaccurate. The geographical limit of P. d. edwardsi, E. rub- riventer, and Varecia is north of the Mananara River near Farafangana, and the southern limit of Indri and P. d. diadema distribution is the Man- goro River, some 5° degrees latitude and 600 km north of the RNI d' Andohahela. Lemur catta gen- erally occurs in the dry and spiny forests at rela- tively low altitudes (Tattersall, 1982; Harcourt & Thornback, 1990), but it was recently recorded by Goodman and Langrand (1996) above the forest line in the open summit zone (2520 m) of the RNI d'Andringitra. During the 48 days we spent sur- veying in the humid forest of the RNI d' Andohahela, however, we neither saw nor heard L catta. Of particular note was the observation of the generally strict dry forest lemur Propithecus ver- reauxi verreauxi during a diurnal transect in the 440 m zone of parcel 1. So far, the southeastern limit of the range of this Propithecus subspecies is the spiny forest of parcel 2 of the RNI d' Andohahela. Thus it is likely that some individ- uals from the western portion of the reserve occur in both forest types; whether their movements are seasonal is unknown. Additionally noteworthy was the fact that the sole group of Propithecus observed in parcel 1 contained three of the char- acteristic white forms and three melanistic forms (Tattersall, 1982, 1986). The melanistic form, pre- viously regarded as subspecies 'majori of P. ver- reauxi (Elliot, 1907, 1913 in Tattersall, 1986), sometimes occurs within groups containing typi- cal P. v. verreauxi (Petter et al., 1977; Tattersall, 1986), but its taxonomic status is poorly known. It seems clear that 'majorV represents a polymor- phic melanistic form of P. v. verreauxi. More work will define precisely the ecological require- ments and needs of P. v. verreauxi in the humid forest. Avahi laniger is found in numerous protected areas (Mittermeier et al., 1994), providing a good picture of its latitudinal distribution. On the east- ern slopes of the RNI d'Andringitra it was re- corded up to 1625 m (Sterling & Ramaroson, 1996), on the RS d'Anjanaharibe-Sud up to 1260 m (Schmid & Smolker, 1998), and in parcel 1 of the RNI d' Andohahela up to 1500 m. The results from these studies indicate that A. laniger occu- pies a considerable elevational range. The Hapalemur griseus individuals seen at the lower altitudes in parcel 1 of the RNI d' Andohahela were considerably larger than typical H. g. griseus and somewhat darker in pelage. One possibility is that they represent individuals of a putative south- ern form, H. g. meridionalis (Waiter et al., 1987). The six animals upon which the description is based were captured 'about 10 km north of Fort Dauphin' (Waiter et al., 1987, p. 51). H. g. mer- idionalis was described as having standard exter- nal features of H. g. griseus, with the exception of a darker coat. Head and body measurements were 30 and 32 cm; tail measurements were 36 and 3 1 cm. The karyotype of H. g. meridionalis was 2n = 56, in comparison to 2n = 54 of H. g. griseus (Waiter et al., 1987). Confirmation of the taxonomic status of the lemurs we observed in parcel 1 was not possible, however, and will re- quire additional morphometric and genetic data. Thus we refer to the observed bamboo lemurs simply as Hapalemur griseus. The encounter rate of Lepilemur mustelinus was extremely low, with only two sightings in the lowest transect zone. The people living in the vi- cinity of the reserve never mentioned hunting Le- pilemur, and it therefore seems unlikely that this is the cause of their rarity. Lepilemur are folivo- rous lemurs with very specialized microhabitats (Ganzhorn, 1989). The average number of Eulemur fulvus collaris sightings increased with elevation (except at 1500 m). We received reports that several hundred of these lemurs are hunted in the lower portions of the reserve each year (B. Randriamampionina, pers. comm.). The 440 m zone was within a 3 hr walk of the nearest village, a fact that might ex- plain the low encounter rate and the flight reaction of this species. Hunting pressure on brown lemurs is widespread, and it is known to affect group size and behavior (Duckworth et al., 1995; Schmid & Smolker, 1998). It is difficult to explain the ab- sence of E. f. collaris in the 1 500 m zone because they were seen within all of the lower zones as well as at 1875 m. We suspect that they would have been detected with a more intensive sam- FEISTNER & SCHMID: LEMURS 279 pling effort. The thick pelage of the Eulemur ob- served at the summit camp suggests adaptation to the high mountain and meteorologically more ex- treme zone of the massif. Signs of Daubentonia were recorded at every site censused, from 390 m to 1920 m. This last elevation may represent the highest altitude at which this species has ever been recorded. Just below the summit of Pic Trafonaomby there was a large amount of bamboo, and there were nu- merous signs of Daubentonia feeding on prey cached in the stems of this plant. Identical traces were found in bamboo at the RNI de Marojejy (Duckworth et al., 1995). Whenever bamboo was abundant we found numerous bamboo stems that had been damaged by aye-ayes, presumably prey- ing on insect larvae. It was not surprising that we found few insect larvae in the bamboo segments we opened, because these segments may well have been investigated by aye-ayes just prior to our arrival and were left intact because they were empty. Pollock et al. (1985) reported Daubentonia feeding on shoots of bamboo. It has been suggested that the distribution of Daubentonia is closely tied to that of Canarium (Iwano et al., 1991; Sterling & Ramaroson, 1996), and that the seeds are, at least during the hot dry season, one of this species's most commonly eat- en foods (Sterling et al., 1994). In the humid for- est of the RNI d'Andohahela, however, Canarium was exceptionally rare. Only two Canarium trees were found in 5 ha of plots censused across the elevational gradient, one at 810 m and one at 1200 m. None of the 37 seeds examined bore traces of Daubentonia. Daubentonia density may thus be low in parcel 1 due to an important food source being locally rare, or Canarium may not be as critical to Daubentonia in Andohahela as studies from Nosy Mangabe suggest. For the spiny forest of parcel 2 of the RNI d'Andohahela, Mittermeier et al. (1994) listed six lemur species. From this list we recorded five spe- cies: Microcebus murium, Lepilemur leucopus, Lemur catta, and Propithecus verreauxi verreauxi by direct observations and Phaner by vocaliza- tions. We did not locate the sixth species, Chei- rogaleus medius. The apparent absence of this last species is surprising because it has been recorded previously in Andohahela (O'Connor et al., 1987; Nicoll & Langrand. 1989), although it is reported to be rare in both Berenty and Beza Mahafaly (Sussman & Richard, 1986; Mittermeier et al., 1994). It seems unlikely that the absence of C. medius is due to sampling error. During the sur- vey all nocturnal species present were generally recorded within 3 hr of censusing. Thus, the 18 hours of nocturnal census work at parcel 2 should have been sufficient to record Cheirogaleus, had it been there in any reasonable numbers. It may be difficult, however, to obtain a complete species list for a region within a short time span if lemur activity is strongly affected by seasonal variation. This is particularly so for C. medius, which un- dergoes a period of prolonged torpor from April or May until mid-September (Martin, 1972; Pet- ter-Rousseaux, 1980). The survey in parcel 2 was undertaken between 8 and 1 3 December, however, so hibernation cannot account for this species' ab- sence. The ambient and physiological conditions determining the end of hibernation are still un- clear, but rainfall seems to have a strong influence on emergence date (J. Fietz, pers. comm.). At the beginning of the 6 days that we spent in the spiny forest, we had exceptionally heavy rain, so C. medius could be expected to have emerged and to have been recorded during our surveys. Phaner furcifer was heard on two separate oc- casions; four different individuals could be local- ized on one of these occasions. Vocalizations are an important part of social communication in this species, and individuals are generally in constant vocal contact (Charles-Dominique & Petter, 1980). We never observed Phaner directly, per- haps because they occur at heights of 12-15 m, where they are difficult to detect (Petter et al., 1971). In fact, there are few records of Phaner being sighted in parcel 2, and records of its oc- currence have largely relied upon the identifica- tion of its vocalizations (Russell & McGeorge, 1977; O'Connor et al., 1987). Phaner furcifer can frequently be heard calling near the village of Ha- zofotsy (Feistner, unpubl. data; M. Pidgeon & S. Goodman, pers. comm.). It is not known which subspecies occurs in this region. Rapid biodiversity assessment techniques to de- cipher the patterns of species distribution and population abundance are problematic for several reasons. First, sampling times often differ from site to site, as do climatic conditions. For instance, during our survey work in the 810 m zone of par- cel 1 , heavy rain and fog interfered with sampling on half of the census nights, and few lemur sight- ings were made. Second, in upland areas the trails often cross bare, rocky terrain that is unlikely to support lemurs. At some sites in this survey, the steep relief meant that transect lengths were short to keep within the elevational limits. In lieu of repeating short transects several times, fewer rep- 280 FIELDIANA: ZOOLOGY etitions of much longer transects would have been preferable. In the 1200 m zone the same groups of Eulemur f. collaris were encountered repeat- edly on one of the transects, violating the as- sumption of independent sampling necessary for calculation of density (Brockelman & Ali, 1987; Whitesides et al., 1988). Overall, small sample sizes mean that it is difficult to accurately deter- mine group size and composition for diurnal le- mur species. In addition, rare species may not be detected at all within the available sampling time. These factors need to be considered when analyz- ing and interpreting rapid census data. The lemur diversity of only eight species along the eastern slope of Andohahela is low in com- parison to other humid forest sites at lower lati- tudes, where reliable data are available. For ex- ample, 13 species have been recorded in the RNI d'Andringitra (Sterling & Ramaroson, 1996), 12 in the Pare National de Ranomafana (Mittermeier et al., 1994), 13 in the RNI de Zahamena (Mit- termeier et al., 1992), 10 on the Masoala Penin- sula (Sterling & Rakotoarison, 1998), and 11 in the RS d'Anjanaharibe-Sud (Schmid & Smolker, 1998). Only one species of Hapalemur is listed for parcel 1 of the RNI d'Andohahela, compared to the eastern slope of the RNI d'Andringitra, which supports three Hapalemur species. When the species observed in the western spiny forest of parcel 2 are added, the RNI d'Andohahela has a total of 13 lemur taxa. Apart from the relatively low diversity of le- murs, the abundance of several species was strik- ingly low. Although we did not calculate actual lemur densities, the encounter rate for both diur- nal and nocturnal humid forest species was ex- ceptionally low. For example, in the 1875 m zone there was on average one nocturnal lemur sighting per 2.3 km of transect. The lowest encounter rate for diurnal lemurs was in parcel 1 at 1500 m, where there was an average of one sighting per 5.5 km walked. Despite the comparatively low diversity and density of lemur species, it is clear that Andohah- ela is important for the conservation of both wet and dry forest primates. The eastern humid forest of the RNI d'Andohahela (parcel 1) is the only protected area in Madagascar where Eulemur ful- vus collaris occurs naturally. Despite this it was reported to us that large numbers of this taxon are hunted every year. In the spiny forest, the envi- ronment was being degraded by people and their livestock. We often encountered people, goats, zebu, dogs, and even cats in the reserve. Grazing and browsing by the ungulates prevents plant re- generation and growth, and trampling crushes and degrades the flora and soils. Domestic carnivores and feral rats can have a significant impact on bird and reptile faunas (Iverson, 1978; Case et al., 1992). The ongoing deforestation of Andohahela's western spiny forest also threatens the eastern hu- mid forest slopes of the massif. In the 1875 m zone, the sclerophyllous forest below Pic Trafon- aomby (1956 m) had been burned just a few days before our arrival. Blackened bamboo stems and tree trunks were left, still smoking from the fire. Below the summit zone was a plateau system that had clearly been modified by burning and cattle grazing for some considerable time. Human dis- turbance had thus already advanced up the west- ern slope of the Andohahela Massif, where most of the forest was gone already. In order to draw up conservation management plans for Madagascar's unique flora and fauna, it is important to investigate the diversity and dis- tribution of endemic plants and animals. Even with such apparently well-studied animals as le- murs, distributional and elevational limits are poorly known for many taxa. Biological invento- ries are important in providing this information. Indeed such work has already resulted in the res- urrection of synonymized species and the descrip- tion of new lemur taxa (Schmid & Kappeler, 1994; Zimmerman et al., 1998). Given the speed with which habitats are being degraded, conser- vation decisions may need to be made rapidly. Devising efficient methods of undertaking biolog- ical inventories, including rapid assessments of primate distribution, population dynamics, and conservation status, is especially important when time is short, resources are limited, and the need for remedial conservation action is acute. Literature Cited Brockelman, W. Y., and R. Ali. 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. Case. T. J.. D. T. Bolger, and A. D. Richman. 1992. Reptilian extinctions: The last ten thousand years, pp. 91-125. //; Fiedler. P. L., and S. K. Jain, eds.. Con- servation biology. Chapman and Hall. New York. Charles-Dominique, P., and J. -J. Petter. 1980. Ecol- ogy and social life of Phaner furcifer, pp. 75-95. In Charles-Dominique. P., H. M. Cooper. A. Hladik, C. M. Hladik, E. Pages, G. F. Pariente, A. Petter-Rous- seaux, J. -J. Petter. and A. Schilling, eds.. Nocturnal FEISTNER & SCHMID: LEMURS 281 Malagasy primates: Ecology, physiology and behav- ior. Academic Press, New York. Duckworth, J. W. 1993. Feeding damage left in bam- boos, probably by aye-ayes (Daubentonia madagas- cariensis). International Journal of Primatology, 14: 927-931. Duckworth, J. W., M. I. Evans, A. F. A. Hawkins, R. J. Safford, and R. J. Wilkinson. 1995. The lemurs of Marojejy Strict Nature Reserve, Madagascar: A sta- tus overview with notes on ecology and threats. In- ternational Journal of Primatology, 16: 545-559. Erickson, C. J. 1995. Feeding sites for extractive for- aging by the aye-aye, Daubentonia madagascariensis. American Journal of Primatology, 35: 235-240. Ganzhorn, J. U. 1988. Food partitioning among Mal- agasy primates. Oecologia, 75: 436-450. . 1989. Primate species separation in relation to secondary plant chemicals. Human Evolution, 4(2): 125-132. 1992. Leaf chemistry and the biomass of fo- livorous primates in tropical forests: Test of a hypoth- esis. Oecologia, 91: 540-547. 1994. Les lemuriens, pp. 70-72. In Goodman, S. M., and O. Langrand, eds., Inventaire Biologique: Foret de Zombitse. Recherches pour le Developpe- ment. Serie Sciences biologiques. Centre dTnformation et de Documentation Scientifiques et Technique, An- tananarivo, Madagascar, No. Special. . 1995. Low-level forest disturbance effects on primary production, leaf chemistry, and lemur popu- lations. Ecology, 76(7): 2084-2096. Goodman, S. M., ed. 1996. A floral and faunal inven- tory of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagascar: With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. , ed. 1998. A floral and faunal inventory of the Reserve Speciale d'Anjanaharibe-Sud, Madagascar: With reference to elevational variation. Fieldiana: Zo- ology, n.s. 90: 1-246. Goodman, S. M., and O. Langrand. 1994. Inventaire biologique foret de Zombitse. Recherches pour le De- veloppement, Serie Sciences biologiques. Centre dTnformation et de Documentation Scientifiques et Technique, Antananarivo, Madagascar, No. Special. . 1996. A high mountain population of the ring- tailed lemur Lemur catta on the Andringitra Massif, Madagascar. Oryx, 30(4): 259-268. Goodman, S. M., and E. J. Sterling. 1996. The utili- zation of Canahum (Burseraccae) seeds by vertebrates in the Reserve Naturelle Integrale d'Andringitra, Madagascar, pp. 83-89. In Goodman, S. M., ed., A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagas- car: With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. Harcourt, C, and J. Thornback. 1990. Lemurs of Madagascar and the Comores. The IUCN Red Data Book. IUCN, Gland and Cambridge. Iverson, J. B. 1978. The impact of feral cats and dogs on populations of the West Indian iguana Cyclura carinata. Biological Conservation, 14: 63-73. Iwano, T, and C. Iwakawa. 1988. Feeding behavior of the aye-aye (Daubentonia madagascariensis) on nuts of ramy (Canarium madagascariense). Folia Prima- tologica, 50: 136-142. 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. Langrand, O., and S. M. Goodman. 1997. Inventaire biologique Foret de Vohibasia et dTsoky-Vohimena. Recherches pour le Developpement, Serie Sciences biologiques. Centre dTnformation et de Documenta- tion Scientifiques et Technique, Antananarivo, Mada- gascar, No. 12. Martin, R. D. 1972. A preliminary field-study of the lesser mouse lemur (Microcebus murinus J. F Miller 1777). Zeitschrift fiir Tierpsychologie, Supplement 9: 43-89. Mittermeier, R. A., W. R. Konstant, M. E. Nicoll, and O. Langrand. 1992. Lemurs of Madagascar: An action plan for their conservation. 1993-1999. IUCN/ SSC Primate Specialist Group. Gland, Switzerland. Mittermeier, R. A., I. Tattersall, W. R. Konstant, D. M. Meyers, and R. B. Mast. 1994. Lemurs of Mad- agascar. Conservation International, Washington, D.C. Myers, N. 1988. Threatened biotas: "Hotspots" in tropical forests. Environmentalist, 8(3): 1-20. . 1990. The biodiversity challenge: Expanded hot-spots analysis. Environmentalist, 10(4): 243-256. National Research Council. 1981. Techniques for the study of primate population ecology. National Acad- emy Press, Washington, D.C. Nicoll, M. E., and O. Langrand. 1989. Madagascar: Revue de la Conservation et des Aires Protegees. World Wide Fund for Nature, Gland. 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., Priorities en Matiere de Conservation des Especes a Madagascar. Occasional Papers of the IUCN Species Survival Commission, no. 2, Gland, Switzerland. Petter, J-J., R. Albignac, and Y. Rumpler. 1977. Mammiferes lemuriens (Primates prosimiens). Faune de Madagascar, 44. ORSTOM-CNRS, Paris. Petter, J. -J., A. Schilling, and G. Pariente. 1971. Ob- servations ethologiques sur deux lemuriens malgaches nocturnes, Phaner furcifer et Microcebus coquereli. Terre et Vie, 3: 287-327. Petter-Rousseaux. A. 1980. Seasonal activity rhythms, reproduction, and body weight variations in five sympatric nocturnal prosimians in simulated light and climatic conditions, pp. 137-152. In Charles- Dominique, P., H. M. Cooper, A. Hladik, C. M. Hla- dik, E. Pages, G. F. Pariente, A. Petter-Rousseaux, J.- J. Petter, and A. Schilling, eds., Nocturnal Malagasy primates: Ecology, physiology and behavior. Academ- ic Press, New York. Pollock, J. I. 1986. A note on the ecology and behav- ior of Hapalemur griseus. Primate Conservation, 7: 97-100. 282 FIELDIANA: ZOOLOGY Pollock, J. I., I. D. Constable, R. A., Mittermeier, J. Ratsirarison, and H. Simons. 1985. A note of the diet and feeding behavior of the aye-aye Daubentonia madagascariensis. International Journal of Primatol- ogy, 6: 435-447. Russell, R. J., and McGeorge, L. W. 1977. Distribu- tion of Phaner (Primates, Lemuriformes, Cheirogali- dae, Phanerinae) in southeast Madagascar. Journal of Biogeography, 4: 169-170. Schmid, J., and P. M. Kappeler. 1994. Sympatric mouse lemurs (Microcebus spp.) in western Madagas- car. Folia Primatologica, 63: 162-170. Schmid, J., and R. Smolker. 1998. Lemurs in the Re- serve Speciale d'Anjanaharibe-Sud. Madagascar, pp. 227-238. In Goodman, S. M., ed.. A floral and faunal inventory of the Reserve Speciale d'Anjanaharibe- Sud, Madagascar: With reference to elevational vari- ation. Fieldiana: Zoology, n.s. 90: 1-246. Sterling, E. J., E. S. Dierenfeld, C. J. Ashbourne, and A. T. C. Feistner. 1994. Dietary intake, food com- position and nutrient intake in wild and captive pop- ulations of Daubentonia madagascariensis. Folia Pri- matologica, 62: 115-124. Sterling, E. J., and N. Rakotoarison. 1998. Rapid assessment of primate species richness and density on the Masoala Peninsula, eastern Madagascar. Folia Pri- matologica, 69(Suppl. 1): 109-116. Sterling, E. J., and M. G. Ramaroson. 1996. Rapid assessment of the primate fauna of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagascar, pp. 293-305. In Goodman, S. M., ed., A floral and faunal inventory of the eastern slopes of the Reserve Naturelle Integrale d'Andringitra, Madagas- car: With reference to elevational variation. Fieldiana: Zoology, n.s. 85: 1-319. Storey, M., J. J. Mahoney, A. D. Saunders, R. A. Dun- can, S. P. Kelly, and M. F Coffin. 1995. Timing of hot spot-related volcanism and the breakup of Mada- gascar and India. Science 267: 852-855. Sussman, R. W, and A. F Richard. 1986. Lemur con- servation in Madagascar: The status of lemurs in the south. Primate Conservation 7: 86-92. Tattersall, I. 1982. The primates of Madagascar. Co- lumbia University Press, New York. . 1986. Notes on the distribution and taxonomic status of some subspecies of Propithecus in Madagas- car. Folia Primatologica, 46: 51-63. Warter, S., G. Randrianasolo, B. Dutrillaux, and Y. Rumper. 1987. Cytogenetic study of a new subspe- cies of Hapalemur griseus. Folia Primatologica, 48: 50-55. 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. Zimmerman, E., S. Cepok, N. Rakotoarison, V. Ziete- mann, and U. Radespiel. 1998. Sympatric mouse le- murs in north-west Madagascar: A new rufous mouse lemur species (Microcebus ravelobensis). Folia Pri- matologica, 69: 106-114. FEISTNER & SCHMID: LEMURS 283 Gazetteer of Localities Mentioned in the Text* Longitude E Latitude S Elevation Locality 0 ' 0 ' (m) Alaotra, Lac 48 30 17 30 Ambatomaniha, Col d' 46 45 24 46 -1000 Ambatovaky, RS -49 15 -16 42 Amboanemba 46 28 24 41 225 Amboasary-Sud 46 24 25 03 Ambohimitambo see Ambohimitombo Ambohimitombo 47 23 20 43 Ambohitantely, RS -47 16 -18 09 1450-1660 Ambositra 47 15 20 30 Ambatorongorongo 46 47 25 04 Ambovombe 46 05 25 11 Ampamakiesiny, Col d' 46 50 24 31 1375 Ampanihy 44 45 24 42 Ampitambet 47 46 20 22 Anadabolava 46 19 24 13 200 Analalava Forest 47 19 24 13 20-50 Analamazaotra, RS 48 28 18 28 930-1040 Analavelona 44 12 22 41 Anandrano 46 59 24 57 10 Anatranatra 46 39 24 18 325 Andaza 46 34 24 03 315 Andohahela, Pic 46 42 24 38 1935 Andrahomana Cave (Grotte) 46 40 25 50 Andrandrivola Forest 49 36 15 46 450-625 Andranohela River -46 47 -24 38 Andranomay Forest 47 57 18 29 1300 Andranomintina 50 19 15 44 Andranondambo 46 35 24 25 Andringitra, RNI 46 54 22 14 Anjanaharibe-Sud, RS -49 26 -14 42 500-2064 Anjozorobe 47 52 18 24 Ankafana see Ankafina Ankaftna 47 12 21 12 Ankapoky Forest 46 31 24 59 Ankarafantsika, RNI -46 57 -16 09 80-330 Ankaratra 47 12 19 25 Ankazomanga 46 37 24 03 430 Ankazondrano see ' Irafonaomby Ankepotsy 46 43 24 33 1550 Anony, Lac 46 31 25 08 0-10 Anosibe An'ala 48 03 18 55 Anosy River not located Antalaha 50 16 14 53 Antanifotsy 46 58 24 59 Antsirana see Antsiranana Antsiranana 49 17 12 16 Antsovela 46 28 25 05 20 Bakika see Ebakika Baly 45 17 16 05 Beampingaratra 46 51 24 28 Beampingaratra Mountains see Anosyenne Mountains Belavenoka 47 05 24 50 20 Bemangidy 47 14 24 34 -100 Bemangily see Bemangidy Beraketa 45 42 24 11 Berenty, RP 46 17 24 59 Berohanga 46 36 24 39 550 Besomosoy 46 28 24 06 275 GAZETTEER 285 Gazetteer Longitude E Latitude S Locality Elevation (m) Betanimena Betenina Betroka Bevilany Bevoay Bezavona Camp 1 (1995) Camp 2 (1995) Camp 3 (1995) Camp 4 (1995) Camp 5 (1995) Camp 6 (1995) Didy Diego-Suarez Ebakika River Efaho River Ejeda Elakelaka Eminiminy Enakara Enosiary Erombo, Lac Esira Esomony Etsilitsily Ezoambo Fanjahira Fanovana Farafangana Fenoevo Fort-Dauphin Grand Lavasoa Hazofotsy Ifarantsa Ifotaka Ihazofotsy Ikongo Forest Iloty Imanombo Imonty Isaka-Ivondro Isedro, Col de Isedro Trail Itapera Itaranta River Itrafanaomby Ivohimanitra Kalambatritra, RS Kirindy Forest Lakato Mahajanga Mahamavo Mahamavo, Col de Majunga Malahelo Forest Manafiafy Manajary Manakara Manambaro 286 46 39 24 46 25 24 46 06 23 46 36 25 46 49 24 46 58 25 46 45.9 24 46 44.3 24 46 44.1 24 46 43.9 24 46 43.3 24 46 36.6 24 48 32 18 See Antsiranana 47 10 24 46 52 25 44 31 24 not located 46 49 24 46 54 24 46 49 24 46 37 25 46 43 24 46 38 24 46 46 24 46 52 24 46 54 24 48 34 18 47 50 22 46 53 24 46 See Tolagnaro 45 25 46 33 24 46 52 24 46 08 24 43 see Hazofotsy 33 21 46 37 24 45 46 24 46 41 24 46 52 24 see Ambatomaniha, Col 46 46 24 47 07 24 46 29 25 47 see Trafonaomby 25 20 46 29 23 44 43 20 48 22 19 46 19 15 46 43 24 46 42 24 46 see Mahajanga 47 25 47 11 24 48 20 21 48 01 22 46 49 25 48 25 16 01 40 01 37.6 35.6 35.0 34.2 33.7 49.0 07 43 48 20 41 37 40 09 20 30 50 49 55 55 49 42 06 49 56 48 47 38 28 49 48 46 53 01 42 28 03 03 43 46 38 120 325 -100 100 440 810 1200 1500 1875 120 100 0-15 400 530 25 75 823 100 20 60 525 215 175 50 0-20 1300-1500 1050 370 25 24 21 22 25 05 45 0-20 13 08 02 FIELDIANA: ZOOLOGY Gazetteer Longitude E Latitude S Locality Elevation (m) Manampanihy River Mananara, RB Mananara River Manangotry, Col de Manantantely Forest Manantenina Mandena Mandrare River Mandromodromotra River Mangoro River Manjakatompo Manombo, RS Mantadia Mantady, PN Maroantsetra Marojejy, RNI Maromby Marotoko River Marosohy Forest Marosohy, Col de Marovoay Marovony Forest Masoala Peninsula Montagne d'Ambre Montagne d'Ambre, PN Moramanga Morondava Nahampoana Nandihizana Nosiarivo Forest Nossi-Be iNosy Be ! Perinet j Petriky • Pic Boby iPic d'lvohibe (Ivohibe) i Pic d'lvohibe (Bemangidy) iPic d'lvohibe. RS i Pic Maromokotro i Pic St. Louis Ranoholo River jRanomafana, PN Ranomafana Atsimo Ranomafana du Sud j Ranomafana-Sud ! Ranomafana-Tanosy Ranopiso iRanopiso, Col de Sakamalio Sakatay -46 49 46 46 46 47 47 46 47 48 47 47 48 49 -49 46 46 46 46 46 47 -50 -49 -49 48 44 46 -47 47 48 48 46 46 46 47 -46 48 46 47 46 46 46 46 58 44 33 52 55 19 00 24 02 45 26 44 see Mantady 27 44 15 35 39 49 48 39 20 10 10 10 12 17 58 10 18 see Nosy Be 15 25 53 53 57 12 59 58 58 27 16 24 24 24 24 24 25 24 20 19 23 18 15 14 24 24 24 24 16 24 15 12 12 18 20 24 20 19 13 18 25 22 22 24 -22 14 25 see Andranohela River 28 21 see Ranomafana de Sud 57 24 see Ranomafana du Sud see Ranomafana du Sud 42 25 39 25 41 24 see Hazofotsy 35 23 50 45 59 17 58 03 55 00 20 02 51 26 26 24 44 34 32 06 05 38 37 37 56 17 58 50 21 20 56 04 11 31 33 32 01 01 16 34 04 02 32 -830 50-600 30 0-20 0-137 75-2133 345 275 350-1300 -1300 50-100 75-300 0-40 2658 2876 530 -300 750 Sedro see Isedro Soanierana 46 52 24 48 20 Ste. Luce see Manafiafy Talakifeno 46 41 24 50 145 Tanandava 47 03 24 24 i Tanatana. Col de 46 51 24 44 -750 Tapera see I tapera Tarantsy River 46 34 25 00 -70 Tolagnaro 46 59 25 01 0-40 GAZETTEER 287 Gazetteer Locality Trafonaomby, Pic Tranomaro Tsaratanana, RNI Tsilotsilo, Col de Tsivory Varavara, Pic Vinanitelo Vohibaka Vohimena Mountains Vohisandria Vondrozo Zahamena, RNI Zombitse, PN Longitude E Latitude S Elevation o ' 0 ' (m) 46 44 24 33 1956 46 29 24 36 -48 51 see -13 Etsilitsily 59 227-2876 46 05 24 04 46 43 24 30 47 16 21 43 46 46 24 32 -47 00 -24 50 46 39 25 10 47 20 22 49 -48 50 -17 40 750-1512 -44 40 -22 47 485-825 * For geographical localities such as rivers, large reserves, and mountain ranges an intersection of coordinates is given that allows for easy location on maps. The information in the body of the gazetteer is based partially on data from U.S. Board on Geographic Names (1955) and Viette (1991). Abbreviations for protected area designations: PN = Pare National, RB = Reserve de Biosphere, RNI = Reserve Naturelle Integrate, RP = Reserve Privee, and RS = Reserve Speciale. t Based on Carleton and Schmidt (1990). Literature Cited 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. United States Board on Geographic Names. 1955. Madagascar, Reunion and the Comoro Islands. Gaz- etteer no. 2. U.S. Government Printing Office, Wash- ington, D.C. Viette, P. 1991. Principales localities ou des Insectes ont ete recueillis a Madagascar. Faune de Madagascar, supplement 2. Private printing. 288 FIELDIANA: ZOOLOGY Index to Scientific Names Acanthaceae 15, 17, 18, 19, 20, 23, 69, 227 Accipiter francesii 180, 184 henstii 180 Acridotheres tristis 182 Acrocephalus newtoni 181 Actiniopteris radiata 49 Actitis hypoleuca 180 Adansonia za 21, 22. 69 Adiantum 218 capillus-veneris 48. 49 madagascariense var. prolongatum 28 phanerophlebium 28 Ad/'na microcepahala 69, 96 Afroptilum 115. 122 decipiens 122 electropterum 122 gi/terfi 117-119, 122, 123 mathildae 116-117. 122. 123 Agapornis cana 180 AgaMr/fl 19. 21. 60. 75. 93 Agavaceae 73. 74, 81. 82, 84. 86, 87, 91 Aglyptodactylus madagascariensis 160, 168 Atoma 19 Albizia 13. 76. 84. 87 gumifera 76 vintsioides 181 Alectroenas madagascariensis 180. 183 Allophxllus 15, 16, 60, 79, 87, 89 cofcfo- 79. 84 Alluaudia 21. 48. 68 ascendens 22, 67, 69, 95 dumosa 95 humbertii 96 procera 22, 67. 95 A/o? 69 divaricata 23 humbertii 19 vaombe 23 Amauropelta bergiana 28. 44 Amblyopone 134. 137. 141 Amblyoponini 134, 137 Amphibia 160, 163, 164. 166. 171 Amphighssus 161, 165, 167, 168, 169. 170 ano.vvfns/s 161, 169 igneocaudatus 1 6 1 macrocercus 161. 163. 169 melanopleura 161, 163, 168, 169 omaticeps 156, 157, 159, 161, 163, 165 punctatus 161, 163, 169 splendidus 171 Anacardiaceae 67, 68, 73. 74, 80. 81, 84, 85. 86, 88, 89, 90,91, 94, 95 Anas erythrorhyncha 1 80 Androngo trivittatus 170, 171 Angiopteris madagascariensis 28 Angraecum 23 Annonaceae 60, 73, 74, 80, 81, 84, 85, 86, 88, 89, 90, 91 Anochetus grandidieri 134, 138 Anodonthyla boulengeri 160, 168 nigrigularis 160, 163, 167 roiixae 167 Anthocleista 18, 19 madagascariensis 76, 89 Anthostema 61, 64 Antidesma 15, 75, 89 petiolare 75, 84, 90 Antirrhoea 13 Antrophyum bivittatum 28 boryanum 28, 44 malgassicum 28, 44 Aphaenogaster 133, 141 Aphloia 19 theaeformis 13, 60, 61, 76, 85, 91, 92 Apocynaceae 23, 70, 73, 74, 81, 88, 89, 90, 94, 95 Apodocephala pauciflora 19 A/7WS barbatus 181 me/6a 181 Aquifoliaceae 73, 74, 80. 81, 82, 83. 84, 86, 89, 92 Araliaceae 18, 19, 20, 58, 59, 73, 74, 80, 81, 82, 84, 86, 87, 88. 89, 90,91, 92, 93 Ardea cinerea 180 purpurea 180 Ardeola idae 180 ralloides 180 Arecaceae 73, 74, 80, 81, 82, 84, 88, 89, 92. 228 Arthropteris monocarpa 28 orientalis var. subbiaurita 28 Arundinaria 18 Asclepiadaceae 23, 69. 70. 94, 96 Asio madagascariensis 181, 225, 227, 231, 237 Asplenium 13, 17. 34. 40, 41, 42, 43 aethiopicum 28, 45 auritum 28 bipartitum 28, 44, 47 blastophorum 28, 44 cuneatum 28, 44, 47 dregeanum 28 erectum var. erectum 28 erectum var. zeyheri 28 friesiorum 28, 45. 47 herpetopteris var. herpetopteris 28 herpetopteris var. massoulae 28, 44 inaequilaterale 28, 44, 47 lividum 28, 36 mannii 28 mWws 13, 28,44, 47 normale 28 pellucidum 28 petiolulatum 28 pool ii 28, 32 poolii var. linearipinnatum 28 poolii var. poolii 28 prionites 28 protensum 28 rutifolium 28 sandersonii 28, 32 theciferum 28 thunbergii 28 unilaterale 28 variabile var. paucijugum 28, 36, 37, 44, 47 virchowii 28, 35 viviparioides 28, 36 Asteraceae 20, 73, 74, 82, 83, 92, 93 Atelornis crossleyi 181, 183, 185 pittoides 181 Athyrium scandicinum 28 A va/i/ /an/^r 269, 273, 275, 276, 279 A viceda madagascariensis 180, 184 Baetidae 115-124 Bakerella 23 clavata 19 Balsaminaceae 19 Bauhinia 23 hildebrandtii 96 Belschmedia 60, 61. 76, 92 Belvisia spicata 28. 44 Bignoniaceae 73. 74, 80, 82, 86, 88, 91, 94. 95 289 Blaeseodactylus sakalava 161 Blechnum 19, 32, 41, 43 attenuatum 28 attenuation van giganteum 28, 44 australe 36 bakeri 28 humbertii 36 ivohibense 28 madagascariense 28, 37 punctulatum 28, 35, 36, 45 simillimum 28 tabulare 20, 28 Blotiella pubescens 28 Boa dumerilii 161 manditra 161, 168, 169 Boidae 161 Boophis 161 albilabris 160, 168 albipunctatus 160, 168 andohahela 160, 165, 166 boehmei 160, 168 2-9410