Skip to main content

Full text of "A Floral and faunal inventory of the eastern slopes of the Réserve naturelle intégrale d'Andringitra, Madagascar : with reference to elevational variation"

See other formats


UNlvtR- OF 
ILLINOIS RY 

«URBANA 1PAIGN 
BIC _' 



yvul? ^o r i^o.^S" l*\s«©i* boelb £«vJer 




l Floral and Faunal 
he Eastern Slopes c 
Jaturelle [ntegrale d 
Vith Refer* ice to 



teven A 



A Floral and Faunal Inventory of the 

Eastern Slopes of the Reserve Naturelle 

Integrate d'Andringitra, Madagascar 



& 



WWF 






FIELDIANA 

Zoology 



NEW SERIES, NO. 85 



A Floral and Faunal Inventory of 

the Eastern Slopes of the Reserve 

Naturelle Integrate d'Andringitra, Madagascar: 

With Reference to Elevational Variation 

Steven M. Goodman, Editor 

Department of Zoology 

Field Museum of Natural History 

Roosevelt Road at Lake Shore Drive 

Chicago, Illinois 60615 

U.S.A. 

World Wide Fund for Nature 
Aires Protigies 
B. P. 738 

Antananarivo (101) 
Madagascar 






Accepted February 28, 1996 
Published September 30, 1996 
Publication 1480 



PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY 



© 1996 Field Museum of Natural History 

ISSN 0015-0754 

PRINTED IN THE UNITED STATES OF AMERICA 



Contents 



Preface vii 

1. Description of the 1993 Biological Inventory of the Reserve Naturelle Integrate d'Andrin- 
gitra, Madagascar 1 

Steven M. Goodman 
Appendix 1-1. Participants in the Project (Field and Laboratory) 5 

2. Description of the Reserve Naturelle Integrale d'Andringitra, Madagascar 7 

Steven M. Goodman and Beverley A. Lewis 

3. Meteorology 20 

Steven M. Goodman and Aristide Andrianarimisa 

4. A Study of the Botanical Structure, Composition, and Diversity of the Eastern Slopes of the 
Reserve Naturelle Integrate d'Andringitra, Madagascar 24 

Beverley A. Lewis, Peter B. Phillipson, Michele Andrianarisata, Grace Rahajasoa, 
Pierre J. Rakotomalaza, Michel Randriambololona, and John F. McDonagh 

Appendix 4-1. Checklist of Vascular Plant Species Recorded from Andringitra 55 

Appendix 4-2. Vertebrate Food Plants on the Eastern Slopes of the RNI d'Andringitra 72 

5. The Pteridophytes of the Eastern Slopes of the Reserve Naturelle Integrate d'Andrin- 
gitra 76 

France Rakotondrainibe and Fidele Raharimalala 

6. The Utilization of Canarium (Burseraceae) Seeds by Vertebrates in the RNI d'Andringitra, 
Madagascar 83 

Steven M. Goodman and Eleanor J. Sterling 

7. Millipedes (Diplopoda) from the Eastern Slopes of the Reserve Naturelle Integrale d'An- 
dringitra, Madagascar 90 

Henrik Enghoff 

8. Ant Diversity Patterns Along an Elevational Gradient in the Reserve Naturelle Integrate 

d' Andringitra, Madagascar 93 

Brian L Fisher 

9. Spatial Distribution of Some Aquatic Insects in the Reserve Naturelle Intdgrale d'Andrin- 
gitra, Madagascar 1 09 

Francois-Marie Gibon, Jean-Marc Elouard, and Michel Sartori 

10. New Heptageniidae (Insecta: Ephemeroptera) from the Reserve Naturelle Integrale d'An- 
dringitra, Madagascar 121 

Michel Sartori and Jean-Marc Elouard 

11. Two New Species of Simulium (Diptera: Simuliidae) from the Reserve Naturelle Integrale 
d'Andringitra, Madagascar 131 

Jean-Marc Elouard, Theogene Pilaka, and Fabienne Ranaivoharindriaka 

12. Parasitic and Commensal Arthropods of Some Birds and Mammals of the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 136 

Barry M. OConnor 

13. Blood Parasites from Birds in the Reserve Naturelle Integrale d'Andringitra, Madagascar 142 

Ellis C. Greiner, Michael S. Putnam, and Steven M. Goodman 

14. Elevational Variation in Soil Macroinvertebrates on the Eastern Slopes of the Reserve Na- 
turelle Integrale d'Andringitra, Madagascar 144 

Steven M. Goodman, Philip P. Parrillo, Sam James, and Petra Sierwald 

15. Shrimps (Crustacea: Decapoda: Atyidae) of the Reserve Naturelle Integrale d'Andringitra, 
Madagascar 152 

Mahefason Richard Andriamihaja 



16. Crayfish (Parastacidae) and Crabs (Potamonidae) of the Reserve Naturelle Integrate d'An- 
dringitra, Madagascar 155 

Bako Rabeharisoa 

17. Amphibians and Reptiles of the Reserve Naturelle Integrate d'Andringitra, Madagascar: A 
Study of Elevational Distribution and Local Endemicity 158 

Christopher J. Raxworthy and Ronald A. Nussbaum 

18. The Birds of the Eastern Slopes of the Reserve Naturelle Integrate d'Andringitra, Madagas- 
car 171 

Steven M. Goodman and Michael S. Putnam 

19. The Shrew Tenrecs (Microgale) (Insectivora: Tenrecidae) of the Reserve Naturelle Integrate 
d'Andringitra, Madagascar 191 

Paulina D. Jenkins, Steven M. Goodman, and Christopher J. Raxworthy 
Appendix 19-1. Key to the Species of Microgale Occurring in RNI d'Andringitra 217 

20. Insectivore Ecology in the Reserve Naturelle Integrate d'Andringitra, Madagascar 218 

Steven M. Goodman, Christopher J. Raxworthy, and Paulina D. Jenkins 

21. Systematic Studies of Madagascar's Endemic Rodents (Muroidea: Nesomyinae): A New 
Genus and Species from the Central Highlands 231 

Michael D. Carleton and Steven M. Goodman 

22. The Rodents of the Reserve Naturelle Integrate d'Andringitra, Madagascar 257 

Steven M. Goodman and Michael D. Carleton 
Appendix 22-1. Comparative Material 283 

23. Results of a Bat Survey of the Eastern Slopes of the Reserve Naturelle Integrate d'Andrin- 
gitra, Madagascar 284 

Steven M. Goodman 

24. The Carnivores of the Reserve Naturelle Integrate d'Andringitra, Madagascar 289 

Steven M. Goodman 

25. Rapid Assessment of the Primate Fauna of the Eastern Slopes of the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 293 

Eleanor J. Sterling and Marie Gisele Ramaroson 

Gazetteer of Localities Mentioned in the Text 306 

Index (Common Names, Scientific Names, and Subject) 309 



VI 



Preface 

In late 1992, Olivier Langrand and Sheila 
O'Connor, of World Wide Fund for Nature 
(WWF), Madagascar, and I discussed the need to 
conduct systematic multidisciplinary biological 
inventories of poorly known forested areas of 
Madagascar. These surveys would provide critical 
information on the biological diversity of re- 
serves, basic data on the natural history of a wide 
variety of organisms, baseline information to as- 
sess changes in the biota of these areas, and the 
means to get Malagasy and foreign researchers 
together for collaborative field projects. Further, it 
was considered critical that the initial results of 
each survey be published in monograph form and 
as quickly as possible, so that the information 
could be made available to Malagasy officials and 
the conservation community. 

For decades, researchers have been conducting 
biological inventories on Madagascar. Often these 
surveys consist of relatively small groups of field- 
workers, encompassing one or two disciplines, 
and are conducted in areas with permanent infra- 
structure. Further, methodologies are often incon- 
sistent between studies, thus limiting comparisons 
between sites. Our main questions were whether 
it was feasible and practical to conduct surveys 
with a large multidisciplinary and multinational 
group of biologists in remote areas and whether 
the data that could be gathered during relatively 
rapid surveys would be useful. 

As a test case, in April 1993 a biological in- 
ventory of the Zombitse Forest, near Sakaraha, 
was organized. This site was relatively easy to 
survey because of its proximity to roads and other 
infrastructure. The group consisted of 18 biolo- 
gists — nine from Madagascar and the balance 
from another five countries. In general, the results 
of this survey, which have already been published 
elsewhere, 1 were satisfactory. 

After the successful mission to the Zombitse 
Forest, it was clear that such projects were feasi- 
ble and generated useful information. We decided 
to conduct inventories along altitudinal gradients 
in mountainous areas of the island, particularly 
where WWF has ongoing conservation projects. 
The months from September to December, tucked 
in between the end of the austral winter and the 



'Goodman, S. M., and O. Langrand, eds. 1994. In- 
ventaire biologique foret de Zombitse. Recherches pour 
le Developpement, Serie Sciences biologiques, No. Spe- 
cial. Centre d' Information et de Documentation Scien- 
tifique et Technique, Antananarivo, Madagascar. 



beginning of the rainy season, are the best period 
to conduct such surveys. It is the only time of the 
year when more or less optimal conditions exist 
for multidisciplinary groups to conduct biological 
inventories. For example, this is the period in 
which many plants begin to flower and fruit; birds 
are breeding, singing, and conspicuous; and the 
activity of reptiles and amphibians increases ow- 
ing to warming days and greater rainfall. 

The first transect, in 1993, was conducted on 
the eastern slopes of the Reserve Naturelle Inte- 
grate (RNI) d'Andringitra. The results of this tran- 
sect are reported in this volume. The 1994 survey 
was conducted in the Reserve Speciale d'Anja- 
naharibe-Sud in the northeast (the results of that 
survey are now being organized as a parallel 
monograph to this one), and the 1995 transect was 
conducted in the RNI d'Andohahela in the south- 
east. 

The 1993 biological inventory of the RNI 
d'Andringitra was made possible by funding from 
the German government through Kreditanstalt fur 
Wiederaufbau (KfW) to WWF as part of an in- 
tegrated conservation and development project 
undertaken to protect the Andringitra and Pic 
d'lvohibe areas. 

The success of the 1993 mission rests largely 
with the people of Ambalamanenjana. The local 
village committee was extremely congenial to 
work with and provided remarkable assistance in 
arranging guides, camp assistants, and porters. We 
extend our sincere thanks to the people of Am- 
balamanenjana. 

We are indebted to WWF staff in both Anta- 
nanarivo and Ambalavao for their help in orga- 
nizing this mission; in particular, we thank O. 
Langrand, S. O'Connor, H. Rabetaliana, P. 
Schachenmann, and C. Wilm6. For permits to 
work in the reserve we are grateful to the Direc- 
tion des Eaux et Forets and the Association Na- 
tional pour la Gestion des Aires Protegees, es- 
pecially G. Rakotonarivo, C. Ravaoarinoromanga, 
and M. H. Faramalala. 

A mission such as the RNI d'Andringitra in- 
ventory is never simple or easy, and the compan- 
ionship and endurance of the participants are 
greatly acknowledged. The relatively rapid deter- 
mination of material gathered during the mission, 
analysis of data, and assemblage of this report by 
both field and laboratory researchers attest to the 
perseverance of those involved. 

We owe a great debt to the reviewers, who crit- 
ically evaluated the chapters presented in this vol- 
ume, in several cases multiple papers, and often 



VII 



under considerable time constraints. The endpaper itor of Fieldiana, and M. Pannell, Managing Ed- 
map was drawn by MYE, Antananarivo, using a itor of the Field Museum Press, took on the task 
modified base map "Carte de vegetation de la of producing this volume with enthusiasm and 
R.N.I, no. 5 d'Andringitra," prepared by T. An- skill. Partial subsidy for publication of this mono- 
driamanga and J. Rahantamalala. M. Razafimpa- graph was provided by KfW. 
hanana drafted many of the French-language re- 
sumes, which were subsequently edited by O. S. M. Goodman 
Langrand. J. Sedlock, J. Weinstein, and D. White 

helped immensely in preparing the illustrations July 1996 

for this volume. Finally, W. Burger, Scientific Ed- Chicago, Illinois 



Vlll 



Chapter 1 

Description of the 1993 Biological Inventory of 
the Reserve Naturelle Integrate 
d'Andringitra, Madagascar 

Steven M. Goodman 



The Reserve Naturelle Integrate (RNI) d'An- 
dringitra is located in south-central Madagascar 
(endpaper map) between 22°07'-22°2rS latitude 
and 46°47'-47°02'E longitude. It comprises an 
area of approximately 31,160 ha over the eleva- 
tional range of 650-2658 m. The reserve was cre- 
ated in 1927, and the limits were demarcated in 
1966 (Nicoll & Langrand, 1989). 

The main portion of the reserve is a flank of 
high mountains that skirts the Central High Plateau 
and the eastern escarpment leading down to the 
Indian Ocean. The notable peaks within the reserve 
are Pic Boby (2658 m) and Pic Bory (2630 m). On 
the basis of the combination of elevational range, 
size, and the remarkable diversity of habitats and 
plant communities, the RNI d'Andringitra has been 
considered an area of significant biological diver- 
sity (Nicoll & Langrand, 1989) and identified as a 
priority site under the National Environmental Ac- 
tion Plan. It is a region with numerous types of 
habitats — large continuous and mostly undisturbed 
rain forests to the east, dry forest to the west, and 
magnificent crystalline mountain ranges, mountain 
meadows, heathlands, and escarpments on the high 
plateau. Most previous scientific studies in the re- 
gion have been conducted at the upper elevational 
limit of the humid forest and on the high plateau. 
These two biomes are different from the lower- 
lying humid forest on the eastern slopes of the 
massif. The latter area, which until now had re- 
mained unstudied, was the focus of a biological 
inventory at the end of 1993, the results of which 
are presented in this volume. 

A multinational and multidisciplinary group of 



biologists studied the fauna and flora of the east- 
ern slopes of the reserve between September and 
December 1993. Our main interests were to doc- 
ument the biological richness of the area and to 
study species distribution and turnover as a func- 
tion of elevation. Further, we wanted to test the 
utility of rapid assessment methods for a wide va- 
riety of organisms. Four camps were placed at 
different elevations (720, 810, 1210, and 1625 m), 
and transects were centered around these camps 
at an altitudinal limit of ±75 m elevation. Two 
separate field groups worked these elevational 
zones; each group generally moved between 
camps in unison and worked a respective zone for 
the same period. The only exception was the 
aquatic insect specialists, who needed substantial- 
ly less time per elevational transect. 

We also gathered data to test one additional hy- 
pothesis within the context of a general inventory. 
The 720 m zone was in a forested area with con- 
siderable human disturbance, and the 810 m zone 
was mostly undisturbed. A comparison of these 
two zones allowed us to test the hypothesis that 
forest degradation within the same altitudinal 
swath has little effect on the various organisms 
studied. Further, the data from the 810, 1210, and 
1625 m sites allowed examination of the distri- 
bution and density of organisms along an eleva- 
tional gradient. 



Abbreviations Used in the Text 

ANGAP Association Nationale pour la Gestion 
des Aires Prot6g6es 



FIELDIANA: ZOOLOGY, N.S., NO. 85, SEPTEMBER 30, 1996, PP. 1-319 



CNRS 


Centre National de la Recherche 




Scientifique 


DEF 


Direction des Eaux et Forets 


FTM 


Foiben-Taosarintanin'i Madagasikara 




(Institut National de Geodesie et Car- 




tographic) 


FMNH 


Field Museum of Natural History 


KfW 


Kreditanstalt fur Wiederaufbau 


LRSAE 


Laboratoire de Recherche sur les Sys- 




temes Aquatique et leur Environne- 




ment 


MBG 


Missouri Botanical Garden 


MNHN 


Museum National d'Histoire Natu- 




relle, Paris 


ORSTOM Office de la Recherche Scientifique et 




Technique Outre-mer 


PBZT 


Pare Botanique et Zoologique de 




Tsimbazaza 


PN 


Pare National 


RB 


Reserve de Biosphere 


RCP 


La Recherche Cooperative sur Pro- 




gramme No. 225, under the Centre 




National de la Recherche Scientifique 


RNI 


Reserve Naturelle Integrate 


RS 


Reserve Speciale 


UMMZ 


University of Michigan Museum of 




Zoology 


WWF 


World Wide Fund for Nature 



Transect Sites 

Coordinates were determined with the use of a 
Geographical Position System, the geographical 
names from various maps (Foiben-Taosarintanin'i 
Madagasikara [Institut National de Geodesie et 
Cartographie]; FTM, 1979) and consultation with 
local people. The altitude was determined with al- 
timeters. The positions of each camp and the major 
trail systems are shown on the endpaper map. 

720 m (camp 1) — Fianarantsoa Province, approx- 
imately 45 km S of Ambalavao, east bank Ian- 
tara River, along Ambalamanenjana-Ambatom- 
boay trail, edge of RNI d'Andringitra, 
22°13'20"S, 47°01'29"E. 

810 m (camp 2) — Fianarantsoa Province, approx- 
imately 43 km S of Ambalavao, RNI d'Andrin- 
gitra, junction of the Sahanivoraky and Saha- 
vatoy rivers, 22°13'40"S, 47°00'13"E. 

1210 m (camp 3) — Fianarantsoa Province, ap- 
proximately 40 km S of Ambalavao, RNI 
d'Andringitra, along Volotsangana River, a trib- 



utary of the Sahavatoy River, 22°13'22"S, 
46°58'18"E. 
1625 m (camp 4) — Fianarantsoa Province, ap- 
proximately 38 km S of Ambalavao, RNI 
d'Andringitra, on ridge above and east of Vo- 
lotsangana River, 22°11'39"S, 46°58'16"E. 



Itinerary of the 1993 Expedition 

Reconnaissance — Reconnaissance (10-15 Sep- 
tember 1993) was conducted by Goodman to lo- 
cate the zone on the eastern slope appropriate for 
an elevational transect. He was accompanied by 
Schachenmann and Rabetaliana for a portion of 
the trip. 

First Field Trip — During the first field trip (23 
September- 1 November 1993), scientific mem- 
bers included Fisher (ants), Goodman (birds and 
bats), Putnam (birds), Ramaroson (lemurs), and 
Sterling (lemurs). The dates for each transect fol- 
low: 23 September-4 October, 720 m transect; 5- 
13 October, 810 m transect; 14-21 October, 1210 
m transect; and 22-31 October, 1625 m transect. 

Second Field Trip — During the second field trip 
(14 November- 18 December 1993), scientific mem- 
bers included Andrianarisata (botany), Elouard 
(aquatic insects), Gibon (aquatic insects), Goodman 
(small mammals and carnivores), Lewis (botany), 
McDonagh (botany), Rabibisoa (reptiles and am- 
phibians), Rahajasoa (botany), Raharilala (botany), 
Rakotomalaza (botany), Randriambololona (bota- 
ny), Raxworthy (reptiles and amphibians), and Ra- 
zafimanantsoa (reptiles and amphibians). The dates 
for each transect follow: 14-21 November, 720 m 
transect; 22-29 November, 810 m transect; 30 No- 
vember-7 December, 1210 m transect; and 8 De- 
cember- 17 December, 1625 m transect. 

The only survey reported here that was not con- 
ducted in the latter portion of 1993 is the pteri- 
dophyte study, conducted by Rakotondrainibe and 
Raharimalala (Chapter 5). In May and June 1995, 
these two researchers visited the eastern slopes of 
the reserve and conducted an elevational transect. 
Their camp and research sites were the same as 
those used by the 1993 group. A list of all field 
and laboratory participants in this project is pre- 
sented in Appendix 1-1. 



Logistics and Trail Systems 

The eastern slopes of the RNI d'Andringitra are 
isolated and not directly accessible by motor ve- 



FIELDIANA: ZOOLOGY 




Fig. 1-1. Camp 1, at 720 m, along the Iantara River. A view from the small settlement of Ambarongy northwest 
across the Iantara River. The river marks the eastern boundary of RNI d'Andringitra. The forest clearing in the 
foreground has been converted into irrigated terraces. Across the river and within the reserve is a recent tavy. In the 
lower right corner is camp 1 . (Photograph by R Schachenmann.) 



hide. We drove from Ambalavao along the road 
leading to Analafolaka (endpaper map). About 40 
km from Ambalavao and a few kilometers before 
Miarinarivo there is a road that ends in the village 
of Ambalamanenjana. This was the village "at the 
end of the road" where we stored supplies and 
from which most of our guides, assistants, and 
porters originated. From Ambalamanenjana we 
walked along a more or less southerly trail that 
for the first 3-4 hours passed through deforested 
swamps, wet meadows, and open savannah, then 
entered forest of varying levels of disturbance, 
crossed the Col de Vohitsatsiva just west and 
slightly below Pic Vohipia (1761 m), and then 
dropped into the Iantara River valley near its con- 
fluence with the Korokoto River; the latter point 
is the northeastern boundary of the RNI d'An- 
dringitra. The trail continues south along the Ian- 
tara River, marking the eastern limit of the re- 
serve, to Ambatamboay and then to villages fur- 
ther south. Our first camp (720 m) was along this 
trail. The direct distance from Ambalamanenjana 



to the northeastern reserve limit is about 12 km; 
it was another 6 km south to our first camp. 

Our first camp was just north of the junction of 
the Sahavatoy and Iantara rivers, at the edge of a 
4-6 ha clearing associated with the small settle- 
ment of Ambarongy, a community of two or three 
families of slash-and-burn (tavy in Malagasy) ag- 
riculturalists (Fig. 1-1). Between the Col de Vo- 
hitsatsiva and our 720 m camp, the trail passed 
through more or less continuous forest of various 
levels of degradation. The forested area along the 
Iantara River and the low hills just to the east 
contained an assortment of trails that were suit- 
able for most of our transects. Between Ambaron- 
gy and Ambatamboay, the main Iantara trail 
leaves the forest and passes through areas of tavy, 
settlements, and secondary forest. 

We used the Sahavatoy Valley as our access 
into the higher elevations of the reserve; the val- 
ley has its origin near Pic Bory (2630 m). A trail 
near our first camp traversed the Iantara River and 
passed obliquely toward the Sahavatoy River. Our 



GOODMAN: DESCRIPTION OF THE 1993 BIOLOGICAL INVENTORY 



second camp (810 m) was on the north bank of 
the Sahavatoy River at the junction of the Saha- 
nivoraky River, about 2.5 km (trail distance) away 
from camp 1. The trail continued as a poorly de- 
fined cattle path for another 1 .5 km along the Sa- 
havatoy River, all within the elevational transect 
zone. The area around the 810 m camp contained 
a trail system sufficient for the inventory. The for- 
est in the vicinity of the camp was relatively un- 
disturbed except for a clearing along the main trail 
about 1 km below the camp. Further, in various 
areas, mostly below the camp, there were clear 
signs of cattle browsing. Within the reserve, on 
the southern bank of the Sahavatoy River and 
closer to the confluence with the Iantara River, 
there was an active tavy of several hectares within 
the reserve limits. 

Together with several local people, and with 
permission from the Direction des Eaux et Forets 
(DEF), we cut a trail leading from the end of the 
Sahavatoy River trail, above the 810 m camp, 
along a projectory paralleling the north side of the 
river valley. The new trail, which was just wide 
enough for people to pass single file, climbed 
steeply up a ridge to about 1 1 00 m and passed 
through an area with numerous hills and deep val- 
leys. The trail continued until it crossed the Vo- 
lotsangana River, where we established the third 
camp, at 1210 m, in a small hollow along the west 
bank. There was no sign of human disturbance in 
this area, and it was necessary to establish all of 
our own trails within this transect zone. Below our 
camp, the Volotsangana River plunged dramati- 
cally into the Sahavatoy River — within less than 
l A km of ground distance it dropped about 200 m 
in elevation. 

Our fourth camp, at 1625 m, was established 
along a north-bearing trail that led up a steep 
ridge on the eastern side of the Volotsangana Riv- 
er, not far from our 1210 m camp. This trail, 
which we cut, climbed steeply; within about 2 km 
of the bifurcation with the trail leading between 
camps 2 and 3 it leveled off to about 1550 m, 
which was within the lower limit of the 1625 m 
transect. This main ridge trail continued for an- 
other 1.5 km past our fourth camp and was the 
main path of the 1625 m transect. 

We were able to cut a route that climbed up to 
the high plateau region of the reserve. This path 
passed along a saddle ridge, above the origin of 
the Volotsangana River, to the west of Ampasi- 
potsy, and allowed access to the upper Kimora 
River valley and Pic Bory. The herpetologists 
used this trail to explore the upper regions of the 



mountain and establish a fifth camp along the Ki- 
mora River at 2075 m. Other researchers also vis- 
ited the high plateau area of the reserve. 



Previous Research Conducted 
in the Reserve 

Little scientific research has been conducted on 
the Andringitra Massif (Paulian, 1958). The no- 
table exception is the 1970-1971 Recherche 
Cooperative sur Programme (RCP) No. 225, un- 
der the Centre National de la Recherche Scienti- 
fique (CNRS), mission to the reserve (Paulian et 
al., 1971). The principal studies conducted during 
this expedition were on the geomorphology, cli- 
mate, and vegetational structure of the high moun- 
tain portion of the reserve. Several zoologists ac- 
companied the group and made collections (see 
subsequent chapters in this monograph for a re- 
view of this information). Because much of the 
zoological information from the RCP expedition 
has not been formally published (e.g., that on rep- 
tiles, rodents, and insectivores) but is critical to 
our understanding of the high mountain fauna, de- 
tails on the positions of the RCP camps (Paulian 
et al., 1971) are presented here (endpaper map). 

Camp 1 (1650 m) — Manjarivolo, at the southern 
end of the Andringitra chain and to the north 
of Pic d'lvohibe. The area covered included 
dense humid middle-altitude forest at 1600 m 
and dense humid montane forest between 1 600 
and 1850 m. The expedition occupied this site 
between 23 October and 4 November 1970. 

Camp 2 (2030 m) — Plateau d'Andohariana, in the 
northern sector of the mountain complex, be- 
low and north of the main flank of high peaks 
and above Antanifotsy. The vegetation of this 
zone is largely herbaceous and ericoid. This 
camp was occupied between approximately 8 
and 23 November 1970. 

Camp 3 (2470 m) — Cuvette Boby, just below Pic 
Boby (2658 m). Mostly rock formations with 
rupicolous plant communities are found there. 
Workers stayed at this site between 23 and 29 
November 1970. During the same period the 
area around Varavarana, close to Pic Bory, and 
Pic Ivangomena (2556 m), in the same contour 
as Pic Boby, were explored. 

Camp 4 (2000 m) — At Marositry in the north- 
eastern portion of the high mountain complex. 
The vegetational cover was mostly ericoid, with 



FIELDIANA: ZOOLOGY 



vestigial areas of dense sclerophyllous plants, 
largely composed of Agauria. This camp was 
occupied between 30 November and 5 Decem- 
ber 1970. 

Camp 5 (1995 m) — Anjavidilava, just below the 
"signal" d'Anjavidilava (2030 m), and along 
the Eaux et Forets weather station trail. Several 
types of vegetation occur in this area — near the 
summit of Anjavidilava, sclerophyllous forest 
largely composed of Philippia, ericoid bush, 
and rupicolous vegetation; to the south and in 
the Kimora Valley, the upper portion of dense 
middle-altitude humid forest; on the slopes 
northwest of Anjavidilava, a combination of 
montane and middle-altitude dense humid for- 
est; and to the northeast of the massif of Vo- 
hidray, a combination of ericoid bush and ru- 
picolous vegetation. This zone and RCP camp 
6 were occupied between 17 December 1970 
and 16 January 1971. 

Camp 6 (1530 m) — Ambalamarovandana, on the 
west side of the Vohidray flank of the main 
mountain chain. The forest in this area was 
classified as middle-altitude humid forest. 



Literature Cited 

FTM. 1979. Sendrisoa no. 55, precarte, 1:100,000. Foi- 
ben Taosarintanin'i Madagasikara, Antananarivo. 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: 
Revue de la conservation et des aires protegees. World 
Wide Fund for Nature, Gland, xvii + 374 pp. 

Paulian, R. 1958. L'Andringitra. Revue de Madagascar, 
troisieme trimestre, nouvelle s6rie, 3: 51-54. 

Paulian, R, J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. RCP 225. Etudes des eco- 
systemes montagnards dans le region malgache. I. Le 
massif de l'Andringitra. 1970-1971. Geomorphologie, 
climatologie et groupements veggtaux. Bulletin de la 
Soci<<t6 d'Ecologie. 11(2-3): 198-226. 



Appendix 1-1. 

Participants in the Project 
(Field and Laboratory) 

A total of 35 scientists and field-workers from 
eight different countries were involved in this 
multidisciplinary study. Their number included 
field participants listed in the previous section as 
well as researchers responsible for some of the 
laboratory studies. The names and addresses of all 
scientific participants follow: 



Andriamihaja, M. R., Laboratoire de Recherche 
sur les Systemes Aquatique et leur Environne- 
ment, LRSAE, B.P. 434, Antananarivo (101), 
Madagascar. 

Andrianarimisa, A., D6partement de Biologie An- 
imate, B.P. 906, University d' Antananarivo, 
Antananarivo (101), Madagascar. 

Andrianarisata, M., % Missouri Botanical Gar- 
den, B.P. 3391, Antananarivo (101), Madagas- 
car. 

Carleton, M. D., Division of Mammals, National 
Museum of Natural History, Smithsonian Insti- 
tution, Washington, D.C. 20560, U.S.A. 

Elouard, J.-M., ORSTOM and Laboratoire de Re- 
cherche sur les Systemes Aquatique et leur En- 
vironnement, B.P. 434, Antananarivo (101), 
Madagascar. 

Enghoff, H., Zoologisk Museum, Universitetspar- 
ken 15, DK-2100 Copenhagen 0, Denmark. 

Fisher, B. L., Department of Entomology, Uni- 
versity of California, Davis, California 95616, 
U.S.A. 

Gibon, F.-M., 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, U.S.A., and WWF, Aires 
Protegees, B.P. 738, Antananarivo (101), Mad- 
agascar. 

Greiner, E. C, Institute of Food and Agricultural 
Sciences, College of Veterinary Medicine, De- 
partment of Infectious Diseases, Building 471, 
Mowry Road, Gainesville, Florida 32611, 
U.S.A. 

James, S., Department of Biology, FB 1056, Ma- 
harishi International University, Fairfield, Iowa 
52557, U.S.A. 

Jenkins, P. D., Mammal Group, The Natural His- 
tory Museum, Cromwell Road, London SW7 
5BD, United Kingdom. 

Lewis, B. A., Missouri Botanical Garden, B.P. 
3391, Antananarivo (101), Madagascar. 

McDonagh, J. F, Wye College, London Univer- 
sity, Wye, Kent, TN25 5AH, United Kingdom. 

Nussbaum, R. A., Museum of Zoology, Univer- 
sity of Michigan, Ann Arbor, Michigan 48109- 
1079, U.S.A. 

OConnor, B. M., Museum of Zoology, University 
of Michigan, Ann Arbor, Michigan 48109- 
1079, U.S.A. 

Parrillo, P. P., Field Museum of Natural History, 



GOODMAN: DESCRIPTION OF THE 1993 BIOLOGICAL INVENTORY 



Roosevelt Road at Lake Shore Drive, Chicago, 

Illinois 60605, U.S.A. 
Phillipson, P. B., Department of Botany, Rhodes 

University, Grahamstown, South Africa. 
Pilaka, T., Laboratoire de Recherche sur les Sys- 

temes Aquatique et leur Environnement, B.P. 

434, Antananarivo (101), Madagascar. 
Putnam, M. S., Department of Zoology, Birge 

Hall, 430 Lincoln Drive, Madison, Wisconsin 

53706, U.S.A. 
Rabeharisoa, B., Ecole des Sciences Agrono- 

miques, B.P. 175, Antananarivo (101), Mada- 
gascar. 
Rabibisoa, N., Departement de Biologie Animale, 

University d' Antananarivo, Antananarivo 

(101), Madagascar. 

Rahajasoa, G., % Missouri Botanical Garden, B.P. 

3391, Antananarivo (101), Madagascar. 
Raharilala, J., Pare Botanique et Zoologique de 

Tsimbazaza, B.P. 4096, Antananarivo (101), 

Madagascar. 

Raharimalala, E, Faculte des Sciences, Universite 
d' Antananarivo, B.P. 906, Antananarivo (101), 
Madagascar. 

Rakotomalaza, P. J., Missouri Botanical Garden, 
B.P. 3391, Antananarivo (101), Madagascar. 

Rakotondrainibe, E, Ecole Pratique des Hautes 
Etudes, 16 rue Buffon, 75005 Paris, France. 



Ramaroson, M. G., World Wide Fund for Nature, 
B.P. 738, Antananarivo (101), Madagascar. 

Ranaivoharindriaka, E, Laboratoire de Recherche 
sur les Systemes Aquatique et leur Environne- 
ment, B.P. 434, Antananarivo (101), Madagas- 
car. 

Randriambololona, M., % Missouri Botanical 
Garden, B.P. 3391, Antananarivo (101), Mada- 
gascar. 

Raxworthy, C. J., Museum of Zoology, The Uni- 
versity of Michigan, Ann Arbor, Michigan 
48109-1079, U.S.A. Current address: Center 
for Environmental Research and Conservation, 
Columbia University, 1200 Amsterdam Ave- 
nue, New York, NY 10027. 

Razafimanantsoa, A., % Project Montagne 
d'Ambre, Joffreville, Antsiranana (202), Mad- 
agascar. 

Sartori, M., Musee Cantonal de Zoologie, CP 448, 
CH-1000, Lausanne 17, Switzerland. 

Sierwald, P., Field Museum of Natural History, 
Roosevelt Road at Lake Shore Drive, Chicago, 
Illinois 60605, U.S.A. 

Sterling, E. J., Deutsches Primatenzentrum, Kell- 
nerweg 4, Gottingen 37077, Germany. Current 
address: Center for Biodiversity and Conser- 
vation, American Museum of Natural History, 
Central Park West at 79th Street, New York, 
NY 10024. 



FIELDIANA: ZOOLOGY 






Chapter 2 

Description of the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 

Steven M. Goodman and Beverley A. Lewis 



Setting and Geology 

The RNI d'Andringitra lies between 22°07' and 
22°21'S and between 46°47' and 47°02'E within 
the Fianarantsoa Province. The central portion of 
the reserve is about 140 km N of the Tropic of 
Capricorn, 40 km S of Ambalavao, and 90 km 
SSW of the city of Fianarantsoa. The only road 
that passes close the reserve originates in Amba- 
lavao (via Sendrisoa) and terminates to the north- 
western side at Antanifotsy (1470 m). From An- 
tanifotsy it takes several hours to walk to the edge 
of the reserve and nearly a full day to walk to the 
upper summit zone (Fig. 2-1). This route is the 
principal one used by most visitors to the reserve. 

The geology of the region is complex, and the 
following discussion is largely based on works 
published by Paulian et al. (1971) and Petit 
(1971). The main mountain chain of Andringitra 
is a magnificent series of seven peaks and pin- 
nacles that originates from a crystalline plateau 
oriented along a north-south axis (Fig. 2-2). The 
four principal peaks are Pic Boby (2658 m), Pic 
Bory (2630 m), Pic Soaindra (2630 m), and Pic 
Ivangomena (2556 m). The main central portion 
of the formation is composed of syenite-granites, 
the eastern flank of granite-migmatite (an injec- 
tion gneiss), and the southern portion of granites. 
The last formation is not continuous with Pic 
d'lvohibe. The main rock complex of the massif 
is part of the Vohibory System, which is believed 
to date from the Precambrian. This is in contrast 
to some of the larger massifs on the island (e.g., 
Ankaratra, Itasy, and Tsaratanana), which had 
their origin during the late Tertiary/Quaternary 
(Battistini, 1972; Brenon, 1972). 



The main central mountain chain has drainages 
to the west and east that provide critical water 
sources for numerous lowland agricultural com- 
munities (Chaperon et al., 1993). To the east are 
various smaller rivers, such as the Korokoto, Sa- 
havatoy, and Lalangina, that drain into the Iantara, 
which then merges into the Manampatra (= Man- 
ampatrana) River. This river is a major source of 
water for agricultural lands in the Farafangana ba- 
sin and drains an area of over 4000 km 2 . The nu- 
merous small tributaries of the Zomandao River 
start on the northern and western sides of the re- 
serve, and this river joins the Mangoky River, 
which drains into the Mozambique Channel. The 
Manarara River has its origin on the southwestern 
side of the reserve. 



Floristics 

Plants on the more accessible western side and 
summit area of the reserve have been relatively 
well collected (see Paulian et al., 1971, for a his- 
tory of botanical research). The forest on the east 
side of the reserve was apparently unknown prior 
to the present study. The plant communities of the 
RNI d'Andringitra are exceptional for any single 
reserve on Madagascar in that they encompass 
three distinct floristic zones: the Eastern Domain, 
or lowland moist forest; the Central Domain, or 
mid-altitude moist forest (also referred to as mid- 
elevation forest); and the High Mountain Domain, 
or dwarf montane flora and heathlands (Humbert, 
1955). Like other high mountain protected areas 
on Madagascar, such as RNI d'Marojejy and RNI 



GOODMAN & LEWIS: DESCRIPTION OF THE RNI D'ANDRINGITRA 




Fig. 2-1. Plateau d'Andohariana at about 2100 m elevation on the northwestern side of the RNI d'Andringitra 
just east of the main granitic ridge. The view is toward Pic Bory. In the foreground is the endemic Aloe andringi- 
trensis. (Photograph by P. Schachenmann.) 



d'Tsaratanana, RNI d'Andringitra has its own par- 
ticular floral assemblage and is rich in endemic 
species. Detailed analyses of the botanical com- 
munities and floral structure of each of the tran- 
sect zones studied during the 1993 survey are pre- 
sented in Chapter 4. In this section the character- 
istic aspects of the flora of the various elevational 
zones are reviewed. For the sake of brevity we 
have not presented author abbreviations after 
plant names; this information can be found in Ap- 
pendix 4-1 of Chapter 4. 

Our initial camp was along the Iantara River, 
just outside the western boundary of the reserve 
(Fig. 2-3). The local forest is in close proximity 
to a tavy (area of slash-and-burn agriculture) and 
a heavily used footpath that forms the western 
boundary of the reserve and links villages to the 
north and south of the reserve. A number of tree 
stumps were found in this forest (the trees pre- 
sumably having been cut for timber), and there 
were fewer massive trees here than in forest inside 
the reserve, suggesting that trees had been selec- 
tively removed. Additionally, no large Dalbergia 



(Fabaceae) trees were found outside the reserve. 
This species, known as rosewood or palisandre, is 
a valuable commercial tree, sought after for its 
dark color and dense grain. Wild honey was lo- 
cally exploited, and large forest trees were felled 
for access to hives. Additionally, this disturbed 
forest had abundant Ravenala madagascariensis, 
the traveler's palm. This species occurs in both 
primary and secondary humid forest (Kress et al., 
1994). In the forest outside of the reserve, Rav- 
enala had colonized large tracts of disturbed and 
cleared areas. 

The floristic composition of the lowland moist 
forest on the eastern slopes of RNI d'Andringitra, 
at about 800 m, is as follows (Figs. 2-4, 2-5). The 
emergent trees were generally buttressed, 25-30 
m in height, and dominated by Canarium mada- 
gascariense ssp. obtusifolium and Sloanea rho- 
dantha var. rhodantha form "quadriloba." Can- 
opy trees 15-20 m in height belonged to Laura- 
ceae (Ocotea, Cryptocarya, Apollonias madagas- 
cariensis), Euphorbiaceae (Uapaca), Oleaceae 
(Noronhia), Myrtaceae (Syzygium), Clusiaceae 



FIELDIANA: ZOOLOGY 




Fig. 2-2. The western flank of the high mountain granitic ridge of the Andringitra Range. This view is from the 
southwest toward the western flank of the ridge and near Pic Boby. Note the erosion pattern of parallel folds in the 
rock. (Photograph by P. Schachenmann.) 



(Ochrocarpus cerasifer, Calophyllum panicula- 
tum, C. drouhardi, and Garcinia), Monimiaceae 
(Tambourissa thouvenotii, Decaryodendron per- 
rieri, Ephippiandra), Anacardiaceae (Protorhus 
sericea) Araliaceae (Schefftera myriantha), Mo- 
raceae (Pachytrophe dimepate), Sapindaceae 
(Blighia), and Sterculiaceae (Dombeya). Contrary 
to the statement by Nicoll and Langrand (1989), 
Dalbergia baroni was not a common tree in the 
eastern forest between 700 and 800 m. Smaller 
trees, 10-15 m in height, included Euphorbiaceae 
(Drypetes), Fabaceae (Albizia, Abrus precatorius, 
Dalbergia), Rubiaceae (Psychotria and Tricaly- 
sia), Ebenaceae (Diospyros), Aquifoliaceae (Ilex 
mitis), Cyatheaceae (Cyathea), and Dracaenaceae 
(Dracaena xiphophylla). 

Common shrubs consisted of Myrsinaceae (On- 
costemum) and Rubiaceae (Schismatoclada). The 
herb layer, composed of bamboo (Nastus), some- 
times formed a dense undergrowth, along with 
Poaceae and Cyperaceae. Melastomataceae (Gra- 
vesia) were also common, along with Balsami- 
naceae (Impatiens), Euphorbiaceae, and ferns (As- 



plenium, Blechnum, and Schizaea dicotoma). Ep- 
iphytes included Melastomataceae (Medinilla), Pi- 
peraceae (Piper), numerous pteridophytes (e.g., 
Platycerium, Asplenium), Orchidaceae (e.g., Bul- 
bophyllum, Angraecum), Cactaceae (Rhipsalis 
baccifera), and Araceae (Pothos scandens). 

The riverine forest bordering the Iantara River 
is rich in Moraceae. Ficus trichopoda is an abun- 
dant shrub up to 1.5 m tall, and was noted to be 
an important food plant for birds and lemurs. Fi- 
cus brachyclada, F. lutea, and F. tiliifolia trees 
were also found. Flacourtiaceae (Apholia theifor- 
mis) was a common riverine tree, along with Clu- 
siaceae (Calophyllum and Ochrocarpus), Stercu- 
liaceae (Dombeya ivohibeensis ssp. ivohibeensis), 
Ericaceae (Agauria), Anacardiaceae (Rhus 
thouarsii), Cunoniaceae (Weinmannia mammea), 
and Pandanaceae (Pandanus). The parasitic Lo- 
ranthaceae, Bakerella clavata var. peralata, was 
common and a nectar source for several birds. 
Plants growing on the rocks alongside the river 
included a dwarf Pandanus, Xerophyta dasyli- 



GOODMAN & LEWIS: DESCRIPTION OF THE RNI D' ANDRINGITRA 




Fig. 2-3. The Iantara River at the level of our camp 1 (720 m). The river marks the boundary of the RNI 
d'Andringitra. (Photograph by M. S. Putnam.) 



rioides (Velloziaceae), Cynorkis (Orchidaceae), 
and Utricularia (Lentibulariaceae). 

The terrain in the 800 m zone along the river 
valleys is largely flat, and this area is bordered by 
ridges and small hills. In general there are few 
areas of exposed bedrock or outcrops. In a few 
places massive isolated boulders were found in 
the forest. The riverine valley soils are largely al- 
luvium, with a relatively shallow organic layer. 

A change in floral composition occurs with in- 
creasing elevation. For example, the tall-but- 
tressed Canarium trees were not found above 
1000 m. Additionally, between 800 and 1200 m, 
Sloanea rhodantha var. rhodantha form "querci- 
folia" was present, in contrast to the form "quad- 
riloba," which occurred at lower elevations. The 
higher-altitude Sloanea was readily distinguished 
from lower-altitude trees by smaller leaves with 
serrate margins and an acuminate apex. The form 
"quadriloba" has entire, obovate leaves. During 
the month of November the lower-elevation forest 
(700-800 m) had fewer plants in flower or fruit 
than during the month of December, when the 



higher-elevation forest (1200-1650 m) was col- 
lected. 

The montane rain forest of about 1200 m is 
characterized by the presence of Podocarpus 
madagascariensis (Podocarpaceae) (Fig. 2-6). As 
at 800 m, Lauraceae (Ocotea, Cryptocarya) were 
still dominant canopy trees, along with Elaeocar- 
paceae {Sloanea rhodantha var. rhodantha form 
"quercifolia") and Clusiaceae (Garcinia, Sym- 
phonia, Calophyllum drouhardi). Smaller trees 
(10-15 m tall) included Euphorbiaceae {Croton, 
Drypetes, Deuteromallotus), Erythroxylaceae (Er- 
ythroxylum pervillei), Annonaceae {Polyalthia 
humbertii, Xylopia), Icacinaceae (Cassinopsis to- 
mentosa), Myrtaceae (Syzygium), Monimiaceae 
(Tambourissa), Ebenaceae (Diospyros), Logania- 
ceae {Anthocleista madagascariensis), and Rubi- 
aceae (Schismatoclada, Gaertnera, Canthium, 
Psychotria). 

The shrub layer was rich in Oncostemum, in- 
cluding O. microsphaerum (Myrsinaceae), as well 
as Rubiaceae {Psychotria) and Rutaceae {Vepris 
fitoravina). Herbs such as Gravesia (Melastoma- 



10 



FIELDIANA: ZOOLOGY 




Fig. 2-4. The Sahanivoraky River just below our camp 2 (810 m) and above its junction with the Sahavatoy 
River. (Photograph by S. M. Goodman.) 



taceae) and Orthosiphon (Lamiaceae) were com- 
mon, along with Streptocarpus suffruticosus var. 
sericeus (Gesneriaceae). The lianascent bamboo, 
Nastus, was abundant in certain areas. Epiphytes 
included Melastomataceae (Medinilla), Pipera- 
ceae (Piper and Peperomia), Orchidaceae (es- 
pecially Bulbophyllum and Angraecum), and 



ferns. The parasites Bakerella (Loranthaceae) and 
Viscum (Viscaceae) were also common. 

Starting at about 1200 m, considerable differ- 
ences were noted in the vegetation between east 
and west versants of ridges. The western slopes 
face the Central High Plateau of the main moun- 
tain massif and weather systems, which form near 



GOODMAN & LEWIS: DESCRIPTION OF THE RNI D'ANDRINGITRA 



11 




Fig. 2-5. Moist montane forest at 800 m elevation near our camp 2. The photograph was taken in the relatively 
flat plain running parallel to the Sahavatoy River. (Photograph by M. S. Putnam.) 



the mountain summits and then descend into these 
valleys (see Chapter 3). Differences in the herb 
layer were especially noticeable. For example, 
Nastus was less common, whereas Cyperaceae 
and Hymenophyllaceae were more abundant, on 
the wetter, west-facing slopes. Epiphytic moss and 
lichen were also more common on the western 
slopes. 

In the riverine forest along the Sahavatoy River 
(1200 m), the following trees were seen: Pittospo- 
raceae (Pittosporum), Cunoniaceae (Weinmannia 
mammea), Anacardiaceae {Micronychia madagas- 
cariensis, Rhus thouarsii), Euphorbiaceae (Ma- 
caranga cuspidata), Flacourtiaceae (Casearia el- 
liptica), Loganiaceae (Strychnos madagascarien- 
sis), Rutaceae, and Bignoniaceae (Ophiocolea flo- 
ribunda). Ficus politoria was the only member of 
this genus found along the river. Three members 
of the Gesneriaceae, Streptocarpus suffruticosus 
var. suffruticosus, S. hilsenbergii, and Didymocar- 
pus madagascariensis, occurred by the river 
banks. 

At 1200 m and above there was a greater num- 



ber of epiphytes than in lower-lying areas. Further, 
mosses and lichens covering tree trunks and 
crown branches increased substantially, whereas 
there was a noticeable decrease in the number and 
size of lianas (Fig. 2-7). 

Between 1000 and 1200 m there was a clear 
change in soils. The fine alluvium was replaced 
by distinctly coarser soils, and exposed bedrock 
and outcrops were relatively common. In some 
areas the terrain above 1000 m became steep, and 
vertical rock faces up to 200 m in height ap- 
peared. Within these virgin areas there were large 
gaps, up to 0.5 ha, carved out of natural forest, 
that were covered with various types of secondary 
vegetation. These gaps were presumably the result 
of natural rockfalls and avalanches, and perhaps 
occasional heavy winds. The pioneering plants of 
these disturbed areas were generally ferns. River 
valleys were often choked with massive and 
rounded boulders, and the first waterfalls ap- 
peared in this zone. The changes between 1000 
and 1 200 m marked the shift between the lowland 
and montane zones of the mountain. 



12 



FIELDIANA: ZOOLOGY 




Fig. 2-6. A view from near our third camp ( 1 200 m) looking west through the treetops of montane rain forest. 
The plateau in the background rises about 400 m and is part of the high mountain granitic complex. (Photograph by 
M. S. Putnam.) 



The ridge-top sclerophyllous forest at 1625 m 
experienced frequent mist and cloud cover and 
was characterized by a low canopy height (5-10 
m) and abundant pendant mosses and lichens (Fig. 
2-8). Areas of this forest were dominated by 
dense stands of bamboo (Fig. 2-9), Arundinaria 



and Nastus (Poaceae), and intermixed Podocarpus 
madagascariensis (Podocarpaceae), Weinmannia 
(Cunoniaceae), Pandanus (Pandanaceae), and 
Symphonia (Clusiaceae). Tree leaves were typi- 
cally small and coriaceous, presumably an adap- 
tation to reduce transpiration associated with the 



GOODMAN & LEWIS: DESCRIPTION OF THE RNI DANDRINGITRA 



13 




Fig. 2-7. Understory of montane rain forest at approximately 1200 m and below our camp 3. Note the water 
coursing over the rocks in the foreground and the relatively heavy epiphyte loads on the standing vegetation. (Pho- 
tograph by M. S. Putnam.) 



extreme wind effects and temperatures on the 
ridge. Outside the bamboo-dominated areas, other 
trees were common, such as Ericaceae (Vaccin- 
ium), Rubiaceae {Gaertnera, Danais fragrans, 
Mussaenda arcuata, Saldinia), Asteraceae (Bra- 
chylaena), Flacourtiaceae (Ludia madagascarien- 



sis, Aphloia), Lauraceae (Cryptocarya crassifolia 
and Ocotea), Euphorbiaceae {Macaranga echino- 
carpa), Monimiaceae (Ephippiandra microphyl- 
la), Verbenaceae (Vitex, Clerodendrum), and Ica- 
ciaceae (Cassiopsis). 

The terrain becomes progressively steeper and 



14 



FIELDIANA: ZOOLOGY 




Fig. 2-8. View of ridge-top sclerophyllous forest at approximately 1625 m and near our camp 4. (Photograph by 
M. S. Putnam.) 



precipitous in the vicinity of and above 1600 m. 
In this zone, sides of ridges often end as sheer 
cliffs that plunge 200-400 m into river valleys, 
and large rock outcrops and isolated boulders 
were relatively common. The soils were coarse 
and often filled with gravel and larger rocks. From 
a vantage point near 1625 m, the high plateau 



cliffs across the Volotsangana valley, several hun- 
dred meters higher in elevation, were clearly vis- 
ible (Fig. 2-10). After even relatively light rains, 
roaring waterfalls would appear at the edge of 
these cliffs, draining water from the large surface 
area of the high plateau. 

On the eastern slopes of the central crystalline 



GOODMAN & LEWIS: DESCRIPTION OF THE RNI D'ANDRINGITRA 



15 




Fig. 2-9. View from camp 4 (1650 m) of forest at the edge of a cliff face dropping down to Volotsangana River 
valley. Note the dense stands of bamboo (Arundinaria and Nastus) intermixed with Pandanus and emergent trees. 
(Photograph by S. M. Goodman.) 



ridge of the mountain chain, the tree line gener- 
ally occurs between 1950 and 2050 m. The forest 
ended abruptly in a mixture of heathland, open 
scrub, and exposed rocks (Fig. 2-11). In some 
higher-lying valleys, up to 2150 m elevation, there 
were isolated patches of montane forest. Shrubs 
found in areas above the tree line included Eleao- 



carpaceae (Eleaocarpus hildebrandtii and E. sub- 
serratus), Rubiaceae (Sardinia and Canthiwn par- 
vistipula), Araliacaeae (Polyscias), Asteraceae 
(Vernonia), Scrophulariaceae (Halleria tetrago- 
na), Thymeleaceae (Peddiea involucrata), and 
Myrsinaceae (Oncostemum). Parasitic Viscum was 
common, as was Bakerella clavata var. lenticel- 



16 



FIELDIANA: ZOOLOGY 




Fig. 2-10. View at 1700 m looking west across the Volotsangana River valley toward the high mountain plateau. 
Natural landslides and avalanches are a regular feature of this landscape, and areas of secondary scrub and regen- 
erating forest were common between 1200 m and the tree line (1950-2050 m). (Photograph by C. J. Raxworthy.) 



lata. The most common herbs were Ericaceae 
(Vaccinium secundiflorum), Euphorbiaceae (Phyl- 
lanthus), and Balsaminaceae (Impatiens). 

The heathland vegetation was typically 60 cm to 
2 m tall and was dominated by Philippia (Erica- 
ceae). Other common plants included the endemic 
Melasomataceae Rousseauxia andringitrensis and 
Asteraceae, such as Helichrysum retrorsum var. 
empetroides, H. cryptomerioides, H. danguyanum, 
and Senecio myricaefolius. In this ericoid zone, a 
Mentha (Lamiaceae) was abundant. Other herbs in- 
cluded Geranium andringitrense (Geraniaceae) and 
Euphorbia emirnensis (Euphorbiaceae). On the 
granite rocks, Aloe (Aloaceae) and Kalanchoe 
(Crassulaceae) were seen, but they were not in 
flower in December. These xerophytes are adapted 
to the extreme climate of the high mountain area 
of the reserve. The shrub Alberta minor (Rubi- 
aceae) was also frequent. 

The last few hundred meters of altitude to the 
summits of Pic Bory (2630 m) and Pic Boby 
(2658 m) are largely composed of areas of open 



rock covered with lichens, with some herbaceous 
and woody vegetation (Fig. 2-12). Paulian et al. 
(1971) noted a forest of 4-m-tall bamboo (Arun- 
dinaria), intermixed with the dominant ericoids 
Philippia and Agauria, on Pic Bory at 2550 m. 
The moss Sphagnum and the lichen Cladonia 
were also common. The summit flora was also 
rich in Cyperaceae, Asteraceae (particularly Hel- 
ichrysum), Velloziaceae (Xerophyta dasylirioi- 
des), Gramineae (Panicum cupressifolium), and 
Restionaceae (Restio madagascariensis). Further, 
several types of orchids and Vaccinium are re- 
ported from this zone. (See Paulian et al. [1971] 
for a review of the higher mountain vegetation.) 



Acknowledgment 

We thank Jeanine Raharilala (Pare Botanique et 
Zoologique de Tsimbazaza, PBZT) for her work 
in the field and help with field identification. 



GOODMAN & LEWIS: DESCRIPTION OF THE RNI D'ANDRINGITRA 



17 




Fig. 2-11. View of the tree line at approximately 2100 m. Note the mosaic pattern of vegetation from the upper 
elevational limit of sclerophyllous forest in the foreground to the Philippia- and Agawn'a-dominated ericoid heathland 
on the hills behind the first ridge. (Photograph by C. J. Raxworthy.) 




Fig. 2-12. View of open grassland area at 2300 m in the vicinity of Pic Bory. Note the areas of lichen-covered 
open rock intermixed with herbaceous and woody vegetation. (Photograph by C. J. Raxworthy.) 



18 



FIELDIANA: ZOOLOGY 



Literature Cited 

Battistini, R. 1972. Madagascar relief and main types 
of landscape, pp. 1-25. In Battistini, R., and G. Rich- 
ard- Vindard, eds., Biogeography and ecology in Mad- 
agascar. W. Junk, The Hague. 

Brenon, P. 1972. The geology of Madagascar, pp. 27- 
86. In Battistini, R., and G. Richard-Vindard, eds., 
Biogeography and ecology in Madagascar. W. Junk, 
The Hague. 

Chaperon, P., J. Danloux, and L. Ferry. 1993. 
Fleuves et rivieres de Madagascar. Monographies Hy- 
drologiques no. 10, ORSTOM, Paris, 874 pp. 

Humbert, H. 1955. Les territoires phytogeographiques 
de Madagascar. In Les divisions ecologiques du 
monde. Colloques Internationale, Centre National de 
Recherche Scientifique, 59: 195-204. 

Kress, W. J., G. E. Schatz, M. Andrianifahanana, and 



H. Simons Morland. 1994. Lemur pollination in 
Madagascar. American Journal of Botany, 8: 542-55 1 . 

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. 

Pauuan, R. 1958. L'Andringitra. Revue de Madagascar, 
troisieme trimestre, nouvelle seYie, 3: 51-54. 



, J. M. Betsch, J. L. Guillaumet, C. Blanc, and 

P. Griveaud. 1971. RCP 225. Etudes des ecosyste- 
mes montagnards dans le region malgache. I. Le mas- 
sif de l'Andringitra. 1970-1971. Geomorphologie, cli- 
matologie et groupements v6g6taux. Bulletin de la So- 
ci6t6 d'Ecologie, 11(2-3): 198-226. 

Petit, M. 1971. Le modele' de type karstique des reliefs 
granitiques (chaine de 1'Andringitra, Madagascar). 
Revue de Geographic 19: 51-105. 



GOODMAN & LEWIS: DESCRIPTION OF THE RNI D'ANDRINGITRA 



19 



Chapter 3 
Meteorology 



Steven M. Goodman and Aristide Andrianarimisa 



Introduction 

The remarkable variation in the biological com- 
munities of the Reserve Naturelle Integrate (RNI) 
d'Andringitra is related to the massif's orographic 
position and elevational gradient. Tropical weath- 
er systems sweeping westward across the Indian 
Ocean pass over the eastern coast of Madagascar 
and inland on to the Central High Plateau 
(Donque, 1975). The high mountains of the island 
act as a partial barrier to the passage of these sys- 
tems further westward. The Andringitra Massif is 
a primary example of such a barrier, and the east- 
ern side receives more precipitation than the west. 
The vegetational communities are distinctly dif- 
ferent between these two slopes. 

Paulian et al. (1971) presented detailed infor- 
mation on the meteorological patterns of the mas- 
sif on the basis of data collected over a period of 
10 years at four meteorological stations. During 
the 1993 expedition daily measurements were 
made of temperature and precipitation. In this 
chapter we review information presented by Pau- 
lian et al. (1971) and analyze additional data from 
a meteorological station as well as data gathered 
between September and December 1993. 



Data 

The weather data presented by Paulian et al. 
(1971) were recorded between 1960 and 1969 on 
a daily basis at the following stations. The first 
station, near Antanifotsy (1470 m), was in an un- 
forested zone, to the north of the reserve and to 
the north of the main mountain chain. The second 



station, along the Andohariana Plateau (1980 m), 
was in the central portion of the reserve and to 
the north of Pic Boby. The vegetation in this zone 
was dominated by Philippia bush and areas of ex- 
posed rock. The third station was near Anjavidi- 
lava (2025 m), along the eastern edge of the main 
mountain chain and north of the Kimora River. 
The local vegetation was a mixture of sclero- 
phyllous plants. Below the site was the upper limit 
of the montane forest. The fourth station was in 
the Cuvette du Boby, at 2470 m, in the central 
point of the massif and just below the highest 
peak. This area was largely exposed rock with 
some rupicolous vegetation and stands of Philip- 
pia bush. Through the courtesy of Petera Rasoja, 
Chef secteur Nord du Cantonnement Forestier, 
RNI d'Andringitra, and Peter Schachenmann, we 
also present further meteorological data gathered 
at the Antanifotsy station between 1979 and 1990. 
Data were collected on the minimum and max- 
imum daily temperatures (°C) and daily precipi- 
tation (mm) during the 1993 field expeditions to 
the eastern slopes of the RNI d'Andringitra. These 
data were collected each morning between 0700 
and 0830. 



Results 

1993 Expedition 

The minimum and maximum temperature and 
precipitation data gathered between 23 September 
and 16 December 1993 are presented in Table 3-1; 
they are divided into the periods during which 
each of the transect zones was visited. In general, 



20 



FIELDIANA: ZOOLOGY 



Table 3-1. Summary of minimum and maximum 
temperatures and precipitation during 1993 expedition 
to RNI d'Andringitra. 



Periods of 
measurement 


Temperature (°C)* 


Rainfall 


at each camp 


Minimum Maximum 


(mm)t 


720 m 








23 Sept.-4 Oct. 


12, 12-16 


11, 17-31 


5, 0.5-70 




12.7, 3.7 


26.2, 5.1 


16.3, 30.1 


14-21 Nov. 


8, 9-16 


8, 22-36 


2, 9.5-60 




13.9, 2.4 


29.6, 4.7 




810 m 








5-13 Oct. 


9, 9-16 


8, 22-31 


2, 0.3-0.5 




12.8, 2.4 


27.4, 3.8 




22-29 Nov. 


8, 15-19 


8, 23-30 


3, 1.5-40 




16.8, 1.7 


26.3, 2.5 




1,210 m 








14-21 Oct. 


8, 10-14 


8, 15-26 


2, 0.5-8.0 




11.3, 1.8 


21.6,4.9 




30 Nov.-7 Dec. 


8, 13-18 


8, 20-31 


5, 1.0-11.5 




14.9, 1.9 


28.1, 3.9 


3.6, 4.5 


1,625 m 








22-30 Oct. 


8, 10-12 


8, 15-21 


4, 2.0-17.5 




11.1,0.6 


17.1, 1.9 


9.5, 6.9 


8-16 Dec. 


9, 9-15 


9, 16-26 


6,0.5-31 




12.3, 2.0 


21.1,4.3 


7.3, 11.8 



* Data are presented as number of records, range, 
mean, and standard deviation. 

t Data are presented as number of days with rain, 
range, mean, and standard deviation. 



the average daily minimum and maximum tem- 
perature decreased with altitude. Precipitation was 
highly variable and not clearly correlated with el- 
evation, although there were proportionately more 
cloudy and rainy days at higher elevations, partic- 
ularly in the 1625 m zone. The heaviest rain 
shower experienced during the expedition was in 
the late afternoon of 3 1 September, when 70 mm 
of rain fell within a 3-hour period. 

In general, between the first and second visits 
to each transect zone there was an increase in the 
average daily minimum and maximum tempera- 
ture. Precipitation was highly variable between 
visits, and in general there were more rainy or 
cloudy days within each transect zone during the 
second expedition to the area. 



Antanifotsy (1979-1990) 

An analysis of weather data collected between 
1979 and 1990 at the Antanifotsy (1470 m) me- 
teorological station (Table 3-2) shows consider- 
able seasonal variation in rainfall and temperature. 
When comparing these results with data gathered 



during 1993, it should be kept in mind that An- 
tanifotsy is at the north end of the complex and 
lies between two major flanks of the mountain. 
The ridge to the east of Antanifotsy is the Vohi- 
dray chain, with several summits over 2000 m. 
Antanifotsy is in the rain shadow, and thus re- 
ceives on average less rain per year than the east- 
ern side of the ridge. 

At Antanifotsy the rainy season generally be- 
gins in November and continues through the end 
of February or the early part of March. The yearly 
variability in monthly rainfall is considerable. For 
example, during the month of January the mean 
rainfall is 233 mm, although in some years only 
54 mm or as much as 568 mm has fallen. Between 
May and August precipitation declines substan- 
tially, and on average there is less than 15 mm of 
rainfall per month. Once again this is highly vari- 
able, because in some years there is no precipi- 
tation during this period and in other years there 
is up to 50 mm. The percentage of days per month 
with rain tends to parallel the pattern of general 
rainfall, although once again this is variable. 

At Antanifotsy the warmest months are No- 
vember through March, coinciding with the rainy 
season. The daily high temperature seldom ex- 
ceeds 30°C and the daily low is seldom less than 
7°C. The standard deviation of temperatures dur- 
ing this period is generally less than 2.2°C. The 
period from late March to April is transitional be- 
tween the warm and cold seasons. By May, daily 
minimum and maximum temperatures drop on av- 
erage more than 3°C. Between June and Septem- 
ber there are nights below freezing, and daily high 
temperatures are on average about 20°C. Snow 
has been reported on the upper peaks (Donque, 
1972). The months of September and October rep- 
resent the period of change between the cold and 
warm seasons. 



Records from 1960 to 1969— Paulian et al. 
(1971) 

The meteorological patterns described by Pau- 
lian et al. (1971) for the four weather stations at 
various points on the massif are in general com- 
parable to that described above with respect to 
seasonal variation in temperature and precipita- 
tion. The peak in rainfall at Antanifotsy was in 
November and December, whereas it peaked at 
the other three sites in January. The average total 
annual precipitation at the four stations was: An- 
tanifotsy, 1,236 mm; Andohariana, 1,581 mm; 



GOODMAN & ANDRIANARIMISA: METEOROLOGY 



21 



Table 3-2. Meteorological data by month from Antanifotsy on the western side of the Andringitra Massif (at 
1,470 m) between 1979 and 1990.* 







Percent of 


Minimum 


Maximum 




Rainfall 


days/month 


temperature 


temperature 


Month 


(mm) 


with rain 


(°C) 


(°C) 


January 


232.7 ± 180.88 


47.0 ± 22.41 


14.7 ± 2.60 


25.3 ± 1.83 




(54.2-567.5) 


(19.4-100) 


(7.8-16.9) 


(17.8-29.5) 


February 


223.9 ± 77.62 


47.4 ± 14.85 


14.1 ± 1.53 


24.8 ± 1.84 




(72.8-344.9) 


(17.9-75) 


(7.9-16.9) 


(18.1-28.9) 


March 


142.4 ± 89.46 


36.1 ± 15.46 


13.3 ± 1.68 


24.6 ± 1.60 




(13.3-339.3) 


(9.7-64.5) 


(8.2-16.8) 


(18.6-29.2) 


April 


35.3 ± 25.99 


17.5 ± 8.06 


11.3 ± 1.92 


23.8 ± 2.24 




(6.8-90.3) 


(10-36.7) 


(6.0-17.1) 


(16.2-29.5) 


May 


13.2 ± 14.35 


8.6 ± 4.83 


8.5 ± 2.31 


21.8 ±2.17 




(0-50.2) 


(0-16.1) 


(1.2-14.4) 


(15.8-29.8) 


June 


5.1 ± 6.83 


10.3 ± 11.23 


6.4 ± 2.37 


19.8 ±2.10 




(0-22.5) 


(0-30) 


(-1.2-12.1) 


(14.0-24.5) 


July 


11.2 ± 10.13 


9.1 ± 6.28 


5.9 ±2.12 


19.5 ± 2.10 




(0-30.9) 


(0-19.4) 


(-2.4-10.9) 


(12.0-24.9) 


August 


13.9 ± 13.24 


13.7 ± 7.03 


5.9 ± 2.02 


20.5 ± 6.61 




(1.1-42.8) 


(3.2-25.8) 


(-4.0-11.5) 


(10.9-28.6) 


September 


18.7 ± 18.3 


11.9 ±9.37 


7.4 ± 2.32 


23.2 ± 9.77 




(0-54.4) 


(0-26.7) 


(-1.8-12.2) 


(12.7-30.2) 


October 


73.0 ± 40.02 


24.2 ± 10.25 


10.2 ± 2.05 


25.1 ±2.55 




(12.4-127.6) 


(6.5-38.7) 


(4.1-18.5) 


(17.2-31.1) 


November 


112.3 ±66.3 


36.9 ± 13.67 


12.0 ± 1.60 


25.2 ± 2.30 




(28.1-216.8) 


(16.7-53.3) 


(6.9-16.0) 


(19.0-31.6) 


December 


187.8 ± 89.25 


44.1 ± 16.24 


13.4 ± 1.61 


25.2 ±2.11 




(54.3-321.0) 


(12.9-61.3) 


(7.6-16.8) 


(18.6-30.0) 


Yearly summary 


1,057 ± 310.6 
(468-1,686) 









Data are presented as mean ± standard deviation (minimum-maximum). 



Cuvette Boby, 2,086 mm; and Anjavidilava, 2,625 
mm. Thus, the easternmost site receives the great- 
est annual precipitation, whereas the site at lower 
elevation and on the west side of the Vohidray 
Chain has the lowest annual rainfall. Paulian et al. 
(1971) found considerable variability in tempera- 
tures between sites that is correlated with eleva- 
tion rather than their east-west position. 



Discussion 

Paulian et al. (1971) classified the regional 
forms of precipitation into three types: (1) Storms. 
Rain systems move into the massif from both the 
east and west. Those systems coming from the 
Indian Ocean, associated with the dominant north- 
east trade winds, tend not to pass over the high 
mountain portion of the reserve and thus they de- 
posit their moisture on the eastern side of the mas- 
sif. Donque (1972) pointed out that areas on the 
eastern escarpment and high plateau with pro- 
nounced relief receive higher rainfall on the wind- 



ward (eastern) side. In areas of the Andringitra 
Massif where the upper portion of the mountain 
chain is lower, for example Andrianony, the sys- 
tems pass over to the west side. Clouds originat- 
ing from the west generally flow over the main 
massif and down the eastern side. During the 
rainy season these storms often bring violent 
rains. The variation in rainfall reported from the 
four meteorological stations clearly reflects their 
relative position to the types of weather systems 
that move across the Andringitra Massif. (2) Fog 
and drizzle. The higher altitudes on the eastern 
side of the massif are often shrouded in fog from 
the late afternoon through the night. This cloud 
cover regularly results in nightly drizzle. (3) Hail. 
During the warm season hailstorms are not infre- 
quent; they occur approximately two times per 
week during the months of November and De- 
cember. 

It is striking that at Antanifotsy the mean 
monthly rainfall during the dry season was less 
during the period from 1979 to 1990 than from 
1960 to 1969. Further, the mean annual rainfall 
declined from 1236 mm between 1960 and 1969 



22 



FIELDIANA: ZOOLOGY 






to 1057 mm between 1979 and 1990. Because 
only the mean values during the former period 
were given by Paulian et al. (1971), it is not pos- 
sible to determine whether these differences are 
statistically significant. However, there appears to 
be a trend during the past few decades toward 
decreased rainfall. The year 1990 marked a period 
of drought in southern Madagascar, and the min- 
imum rainfall recorded at Antanifotsy that year 
was 467 mm. 

The meteorology of the RNI d'Andringitra can 
be characterized as follows: (1) The eastern slopes 
show a pattern of rainfall and temperature parallel 
to the pattern seen in eastern humid forests 
(Donque, 1975). (2) Further toward the west and 
on the higher portions of the massif, there is a 
decrease in average temperature and an increase 
in seasonal variability. (3) The western slopes of 
the region are distinctly drier than other areas of 
the reserve. 



Acknowledgments 

We are grateful to Petera Rasoja for allowing 
us access to the 1979-1990 weather registers 
from Antanifotsy, and to Peter Schachenmann for 
facilitating this. 



Literature Cited 

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. 

. 1975. Contribution geographique a l'6tude du 

climat de Madagascar. Nouvelle Imprimerie des Arts 
Graphiques, Antananarivo, vii + 478 pp. 

Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. RCP 225. Etudes des eco- 
systemes montagnards dans le region malgache. I. Le 
massif de l'Andringitra. 1970-1971. G6omorphologie, 
climatologie et groupements v6g6taux. Bulletin de la 
Soci&e" d'Ecologie, 11(2-3): 198-226. 



GOODMAN & ANDRIANARIMISA: METEOROLOGY 



23 



Chapter 4 

A Study of the Botanical Structure, Composition, 
and Diversity of the Eastern Slopes of the 
Reserve Naturelle Integrate d'Andringitra, 
Madagascar 

Beverley A. Lewis, Peter B. Phillipson, Michele Andrianarisata, 
Grace Rahajasoa, Pierre J. Rakotomalaza, Michel Randriambololona, 
and John F. McDonagh 



Abstract 

A rapid quantitative assessment of botanical structure, composition, and diversity was con- 
ducted in the rain forest on the eastern side of the Reserve Naturelle Integrale (RNI) d'An- 
dringitra. The reserve is situated on the southeastern edge of the central plateau of Madagascar. 
Two hypotheses were tested. The first was that there would be a difference in the structure, 
composition, and species diversity between forest outside the reserve and forest inside the 
reserve boundaries owing to differences in levels of disturbance. The second was that there 
would be changes in structure, composition, and species diversity in the forest along an ele- 
vational gradient from 810 m to 1625 m. Our results show that there were fewer large trees 
outside the reserve, suggesting selective removal, and there was a trend toward reduced floristic 
diversity in the disturbed forest. With increasing elevation, canopy height decreased, and the 
amount of epiphytic bryophytes and lichens increased. There was no indication that floristic 
diversity decreased with increasing elevation, although there were changes in the floristic com- 
position. Our findings are compared with similar work done in Madagascar, Africa, and the 
Neotropics. A vegetation classification system for the forest sampled is proposed. 

Resume 

Un inventaire quantitatif rapide de la structure, de la composition et de la diversite botanique 
a ete effectue dans la foret pluviale orientale de la Reserve Naturelle Integrale (RNI) d'An- 
dringitra (No. 5). Au cours de cet inventaire, deux hypotheses ont ete testees: La premiere c'est 
qu'il y aurait une difference dans la structure et la diversite des especes forestieres a l'exterieur 
de la reserve et a l'interieur des limites de la reserve due a des niveaux de perturbation 
differents. La seconde c'est qu'il y aurait des changements dans la diversite specifiques et dans 
la structure forestiere entre 810 m et 1625 m d' altitude. Les resultats ont montre qu'il y avait 
moins de grands arbres en dehors de la reserve, ce qui suggere qu'il y a eu une exploitation 
forestiere selective et que la diversite floristique a une tendance a diminuer au sein des forets 
d6gradees. A mesure que Ton monte en altitude, la hauteur de la voute forestiere diminue et 
le nombre de bryophytes epiphytes et de lichens augmente. II n'y a aucune indication relative 
a la diminution de la diversite floristique en fonction de l'altitude, malgre des changements 
observes dans la composition floristique. Les resultats de nos recherches sont compares a des 
travaux similaires qui ont ete effectues a Madagascar, en Afrique et en Amerique neotropicale. 
Nous proposons ici un systeme de classification de la vegetation de la foret echantillonnee. 

24 FIELDIANA: ZOOLOGY 



Introduction 

There has been little quantitative botanical 
work done in Madagascar. There is a shortage not 
only of ecological data useful for zoologists, con- 
servation planners, and many other disciplines, 
but also of basic information on plant community 
structure and diversity within an area of forest. In 
an effort to address this problem, the Missouri 
Botanical Garden (MBG) Madagascar Program 
has begun to perform quantitative assessments in 
many parts of the country. The present study rep- 
resents one of these assessments. 

The aim of this work was to make a quantita- 
tive assessment of the botanical structure, com- 
position, and diversity of the rain forest of the 
Reserve Naturelle Integrate (RNI) d'Andringitra 
within approximately one calendar month. The re- 
serve is situated on the southeastern edge of the 
central plateau of Madagascar. It includes the is- 
land's second highest peak and has an elevational 
range from 700 to 2658 m. It is remarkable for 
encompassing three of the four floristic domains 
of the Eastern Malagasy Region recognized by 
Humbert (1955): namely, the Eastern, the Central, 
and the High Mountain domains. Each domain 
has certain floristic features and is characterized 
by different vegetation types, including primary 
forest types with distinct structural features. The 
vegetation of the Eastern Malagasy Region was 
reviewed by White (1983), who integrated treat- 
ments by earlier authors and provided a consistent 
terminology for the different vegetation types. 

The vegetation of the RNI d'Andringitra was 
discussed by Perrier de la Bathie (1927) and by 
Paulian et al. (1971). Both studies focused pri- 
marily on the non-forest vegetation at higher al- 
titudes than the present study, although Paulian et 
al. included descriptions of some forest vegetation 
at several sites in the reserve. Perrier de la Bathie 
provided a list of species endemic to Andringitra 
(see Appendix 4-1). However, they did not visit 
the area that is the subject of the present study. 

As with all high mountain areas in Madagascar, 
the Andringitra Massif has its own particular spe- 
cies assemblage and is rich in species believed to 
be endemic to the reserve and its immediate sur- 
roundings (Phillipson, 1994). The more accessible 
western side and summit area of the reserve have 
been relatively well collected botanically (see 
Paulian et al., 1971, for a history of botanical 
work). The forest on the eastern side of the re- 
serve was virtually unexplored prior to the present 
study. 



The botanical survey work was carried out be- 
tween 14 November and 18 December 1993, and 
it was split into 7- to 8-day work periods within 
each of four transect zones. Transects 1 and 2 
were at approximately the same elevation (720 m 
and 810 m, respectively); the former was outside 
and the latter within the reserve. At transect 1 
there was clear evidence of recent exploitation of 
the forest. People from the nearby village had re- 
moved wood and other plant products, and live- 
stock were allowed to graze within the forest. At 
transect 2 there was no evidence of recent distur- 
bance, although here and in other areas within the 
reserve it was impossible to be certain that no 
exploitation had ever taken place in the past. 
However, it was hypothesized that any differences 
detected between the results from transect 1 and 
2 would be a consequence of this disturbance. 
Transects 2, 3, and 4 were each separated by ap- 
proximately 400 m in elevation (810, 1210, and 
1625 m, respectively). In the vicinity of transect 
2, there were some minor signs of human-related 
disturbance, particularly cattle browsing. Tran- 
sects 3 and 4 were well within the undisturbed 
forest of the reserve. 

The specific objectives of the present study 
were (1) to provide a quantatitive account of the 
vegetation of the study area; (2) to compare the 
forest around transects 1 and 2 and assess whether 
differences in structure, species composition, or 
species diversity, could be accounted for by forest 
exploitation; (3) to determine whether the vege- 
tation changed along the elevational gradient be- 
tween transects 2 and 4; (4) to assess the influence 
of aspect, slope, and topography on forest struc- 
ture, composition, and diversity; and (5) to com- 
pare the forests of the eastern slopes of the RNI 
d'Andringitra with other Malagasy forests and 
rain forests in other parts of the world. 



Classification of Eastern 
Malagasy Forests 

The vegetation of the Eastern Region of Mad- 
agascar was divided into a number of broad cat- 
egories by White (1983), who synthesized treat- 
ments by earlier authors, notably Perrier de la 
Bathie (1921), Humbert and Cours Darne (1965), 
and Koechlin et al. (1974), and provided consis- 
tent terminology. According to White, the primary 
forest types correspond approximately to broad al- 
titudinal bands, as follows: up to 800 m — "low- 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



25 



land rain forest"; 800 to 1300 m — "moist mon- 
tane forest"; 1300 to 2300 m — "sclerophyllous 
montane forest". 

Local conditions may cause the vegetation to 
deviate from this generalization; for example, 
moist montane forest may extend as high as 2000 
m on suitable soils in sheltered localities. These 
three forest types show the following trends from 
low to high altitude: (1) decreasing stature, (2) 
fewer straight unbranched boled trees, (3) less 
stratification, (4) more species with sclerophyllous 
leaves, (5) more epiphytes, (6) more bryophytes 
and lichens, (7) a better developed and more di- 
verse herb layer, and (8) floristic changes. Guil- 
laumet (1983) gives more precise information on 
the floristic changes. In many areas the region dis- 
turbance and degradation have resulted in second- 
ary communities replacing primary vegetation 
types. 

On the drier western slopes, between 800 and 
1600 m a floristically distinct community occurs. 
It is dominated by Uapaca bojeri and referred to 
as "Tapia forest." At higher altitudes, this forest 
is replaced by "montane bushland and thicket." 
On rock outcrops "rupicolous shrublands" occur 
at all altitudes. Although these vegetation types 
may occur within the RNI d'Andringitra, they 
would not be expected to occur in the present 
study area. 

Paulian et al. (1971) classified the forest com- 
munities at several sites on and around the An- 
dringitra Massif as follows: 

(1) "Foret dense humide de moyenne altitudes" 
(three sites between 1300 and 1650 m). This 
is a dense, closed type of forest with many 
layers. The herb layer is generally composed 
of species with large leaves, and grasses are 
usually absent. 

(2) "Foret dense humide de montagne" (seven 
sites between 1650 and 2000 m). This forest 
type differs from low- and mid-elevation for- 
est in the smaller stature of the trees. Addi- 
tionally, trees in the canopy layer are more 
likely to have sclerophyllous leaves and an 
abundance of epiphytes (individuals and spe- 
cies), especially bryophytes and lichens. The 
herb layer is dense with ferns and herbs with 
large, soft leaves. 

(3) "Foret dense scle'rophylle de montagne" 
(three sites at approximately 2000 m). This 
type of forest has a canopy height of 10 m, 
with the majority of the leaves sclerophyllous. 
The middle layer is composed of shrubs with 



small leaves, and there is an abundance of 
terrestrial lichens and bryophytes. 

These three forest types correspond approxi- 
mately to the lowland, montane, and sclerophyl- 
lous forests of White (1983) on the basis of the 
few characteristics given. However, each appears 
to occur at considerably higher elevations than 
would be expected. Little precise information on 
differential or diagnostic species of the different 
vegetation types or detailed information on spe- 
cific plant communities throughout the Eastern 
Malagasy Region is available. 



Sampling Methods 
Rationale 

There is no accepted standard method for the 
assessment of forest structure and tropical plant 
communities, and comparisons between different 
studies are often difficult or impossible. The fol- 
lowing is a discussion of the rationale behind the 
sampling protocol used at RNI d'Andringitra. 

Of the ecological work that has been done in 
Madagascar, several plot-based sampling methods 
have been used (e.g., Schatz & Malcomber, in 
press), whereas line transect methods have been 
adopted in other studies (Gentry, 1988; Du Puy 
et al., 1994). The MBG is currently engaged in a 
long-term study of the rain forests in Madagascar; 
a component of this study has been the establish- 
ment of three 1-ha permanent rectangular plots in 
Pare National (PN) de Ranomafana (Schatz & 
Malcomber, in press), two in the Masoala Penin- 
sula, one north of Fenoarivo at Tampolo, and one 
north of Tolagnaro at Ste. Luce (G. E. Schatz, 
pers. comm.). Within these plots, measuring 500 
X 20 m, all stems greater than 10 cm diameter at 
breast height (dbh) are sampled. Specimens of all 
species recorded from each plot are collected, se- 
lecting fertile material whenever possible, for 
identification and to provide voucher specimens. 
Non-fertile material frequently presents problems 
because it is often difficult and extremely time- 
consuming to identify beyond family or genus. 
This problem is further compounded when the 
area to be sampled has not been adequately ex- 
plored botanically, and by the generally poor state 
of our knowledge of the Malagasy flora. In such 
cases, however, a conservative estimate of the 
numbers of different species in a genus collected 



26 



FIELDIANA: ZOOLOGY 



from the sample area can be made on the basis of 
vegetative features, even if binomial names can- 
not reliably be assigned to each specimen; these 
species are termed "morphospecies." This infor- 
mation is valuable because the number of species 
in an area is a simple and appropriate indicator of 
diversity (Whittaker, 1977; Gentry, 1988). 

Large plots (^1 ha) are usually used when the 
aim is to include virtually all of the species oc- 
curring in a local community, an approach that 
may be necessary if only one sample of that com- 
munity is being taken. The selection of represen- 
tative sites for such plots is difficult and may be 
subjective, requiring the researcher to be familiar 
with the area in question. Small numbers of large 
(1 ha) permanent plots are therefore appropriate 
for long-term studies, particularly when there is a 
monitoring component to the work or if the land- 
scape is relatively uniform. This approach is also 
time-consuming; approximately 3 weeks may be 
required for a team of four trained personnel to 
mark, measure, climb, and collect the large num- 
ber of trees that typically occur in a 1 ha parcel 
of tropical forest. In sampling large areas of veg- 
etation over a limited time period, the aims and 
approach are different, and the sampling of a 
greater number of smaller plots is preferable. This 
is particularly true if the area under study is to- 
pographically diverse and spans a broad eleva- 
tional range, as is the case along the eastern slopes 
of RNI d'Andringitra. It was not our aim to record 
all the species in a single community, but rather 
to sample in a way that allowed estimation of the 
variability occurring within the study site that 
could be used for comparative purposes. 

The RNI d'Andringitra study was designed spe- 
cifically to detect and quantify differences within 
the vegetation as a function of disturbance and 
elevation. The results were analyzed using statis- 
tical methods and replicate sampling in each tran- 
sect zone. Extrapolating from data on another 
Malagasy forest at similar altitudes, PN de Ran- 
omafana (Schatz & Malcomber, in press), it was 
calculated that approximately 20-30 trees with 
dbh measurements of 10 cm or above could be 
expected to fall within a 10 X 20 m plot. This 
was considered to be the minimum size for ade- 
quate sampling. In substantially smaller plots, the 
presence of particularly large individuals or 
clumped distribution of certain species would bias 
the results. Also, because it is inevitable that 
boundary errors occur due to uncertainty of 
whether trees on the boundaries should be includ- 
ed or excluded, in smaller plots a higher propor- 



tion of the trees will be sited on the boundaries, 
thus potentially increasing the sampling errors. 
Four or five replicates are usually considered suf- 
ficient for biological studies where variation will 
not be excessively high (Watt, 1993). Increasing 
replication gives increasing precision but also in- 
creases workload. It was considered that five 10 
X 20 m plots in each transect zone would be a 
manageable workload, given time and worker 
constraints. 

A sampling technique, used in a wide range of 
tropical forests, in which all plants with dbh mea- 
surements of at least 2.5 cm are recorded in 0. 1 
ha transects consisting of 10 2 x 50 m belt tran- 
sects is presented in detail by Gentry (1982, 
1988). In Madagascar, Gentry used this sampling 
technique at Pdrinet, Ampijoroa, and on Nosy 
Mangabe, but it was not followed in the present 
study because sampling would have required 
more time at each transect zone than was avail- 
able. 

Line transects represent another technique for 
sampling plant communities; they have the poten- 
tial to maximize sampling effort in different com- 
munities with heterogeneous vegetation. Du Puy 
et al. (1994) used line transects during a rapid 
assessment of an area of deciduous forest in 
southwestern Madagascar (Zombitse Forest). 
Three 500 m transects were laid out in the forest, 
with 10 10 X 5 m quadrats placed along each 
transect line at 50 m intervals. Height, dbh mea- 
surements, and field identifications were made for 
each tree of at least 10 cm dbh. Although this is 
a useful technique, and one of the best means of 
rapidly measuring diversity within a forest, it is 
difficult to compare data from line transects with 
those from the plot work that is currently being 
done throughout Madagascar. Additionally, con- 
sidering the problems mentioned above with plot 
boundaries, it is difficult to extrapolate any mea- 
surements on a per-hectare basis with confidence 
from such small quadrats. 



Sampling Procedure 

Five 10 X 20 m plots were established and 
sampled at each of the four transect zones in the 
RNI d'Andringitra. The plots were selected to in- 
clude a representative range of microhabitats seen 
at each elevation; for example, plots of differing 
aspect, slope, and topography (ridge top, mid- 
slope, and valley bottom) were included. The 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



27 



plots were situated within a radius of 1 km from 
the base camp, with a distance of at least 50 m 
separating each plot. The elevational range at each 
transect was centered at the elevation of the camp 
± 75 m. Plots were often placed near to, or over- 
lapping, transects established for other investiga- 
tions being done as part of the survey. 

The plots were measured and marked out with 
flagging tape. Trees greater than 10 cm dbh were 
numbered, and a voucher specimen of each spe- 
cies was collected. The following measurements 
were made for each tree: (1) dbh (cm); (2) height 
of tree (m) — estimated from ground level; (3) 
crown diameter (m) — determined by standing be- 
low the tip of a typical branch and estimating the 
distance to the trunk, and multiplying by two; and 
(4) number of lianas using the main stem (trunk) 
of the tree for support. 

In addition, the following information was 
gathered for each plot: (1) site information — ele- 
vation, aspect, slope, and site position; (2) abun- 
dance of vascular plant epiphytes (i) on tree 
trunks and (ii) in tree crowns (estimated on a scale 
of 0-3, where indicated absence, 1 occasional, 
2 common, and 3 abundant); and (3) presence of 
epiphytic bryophytes and lichens (i) on tree trunks 
and (ii) in tree crowns (estimated on the same 
scale as other epiphytes). 

Fertile voucher specimens were collected in 
sets of up to 10 duplicates, which will be widely 
distributed. Sterile voucher specimens were col- 
lected in two duplicate sets; one was deposited at 
the herbarium of the Pare Botanique et Zoolo- 
gique de Tsimbazaza (PBZT), Madagascar (acro- 
nym: TAN), and one will be distributed to spe- 
cialists for further identification. The voucher 
specimens were identified, as accurately as pos- 
sible, at TAN. 

In addition to plot work, several days were 
spent in each transect zone making general col- 
lections of fertile material. This was an essential 
part of the work because it allowed us to augment 
the botanical information obtained from the study 
plots. It also provided fertile material to enable 
identification of a number of sterile plot vouchers, 
and it contributed valuable material from an area 
previously uncollected. A checklist of vascular 
plant species recorded in the RNI d'Andringitra 
has been compiled (Appendix 4-1). This list is not 
comprehensive. It is merely a compilation of the 
species and morphospecies recognized in the pres- 
ent study and information readily available from 
other sources. The checklist records voucher spec- 
imens collected during the present study and may 



serve to stimulate more exhaustive searches in the 
literature, in herbaria, and in the field. A search 
of information held in the MBG computer data- 
base for taxa with "Andringitra" forming part of 
their original description, or with an epithet de- 
rived from the word, and a search for specimens 
from the reserve, also produced numerous re- 
cords. 



Data Analysis 

Statistical Analysis 

Basal area calculations are widely used to give 
a general impression of tree biomass (Gentry, 
1993; Schatz & Malcomber, in press). The basal 
area of an individual tree is the cross-sectional 
area of its stem (trunk) at breast height. The equa- 
tion for the calculation of basal area of a tree is 
given below; these calculations were made for 
each tree and then summed to give values per plot 
and transect zone. 



basal area (m 2 ) = 77 



(dbh(cm)) 2 
2 x 100 



Statistical analyses were performed separately 
to test the two hypotheses given in the introduc- 
tion (p. 25). A two-way analysis of variance 
(ANOVA) was performed to test for significant 
differences in the variables between transect 
zones 1 and 2, and a second ANOVA test was 
used to test for significant differences in the vari- 
ables between transect zones 2, 3, and 4. Except 
where stated otherwise, significance at P s 0.05 
was tested. For each ANOVA the following vari- 
ables were tested: tree basal area, tree height, 
number of trees, crown diameter, number of li- 
anas, abundance of epiphytes (vascular and bryo- 
phytes or lichens) on tree trunks and in the tree 
crowns, number of families, and number of spe- 
cies present. Raw plot data were used for number 
of trees, number of species, and number of fam- 
ilies, but for all other variables, mean values were 
calculated for each plot within each transect zone; 
these mean values were used in the ANOVA. 
Where significant differences were found, a 
means separation test was used (Duncan's Multi- 
ple Range Test) to establish which means differed 
significantly from each other. 

Where appropriate, regression lines were fitted 
to the data to test relationships between measured 
variables, and correlation tests were performed to 



28 



FIELDIANA: ZOOLOGY 






determine whether any two measured variables 
were correlated. 

Multivariate Analysis 

The following three multivariate analysis tech- 
niques were applied to the collected data: ( 1 ) two- 
way indicator species analysis (TWINSPAN) 
(Hill, 1979a); (2) detrended correspondence anal- 
ysis (DECORANA) (Hill, 1979b); and (3) canon- 
ical correspondence (CANOCO) analysis (Ter 
Braak, 1986). All three analyses were conducted 
independently using the canopy (crown diameter) 
data and the dbh data. 

Twinspan — TWINSPAN provides a hierarchi- 
cal clustering of plot data. The method provides 
an estimation of the similarity between plots by 
comparing the characteristics of each in terms of 
species composition and the importance of each 
species as a component in each plot. In addition 
to providing a quantitative analysis of the com- 
position of the vegetation for each plot, TWIN- 
SPAN provides an assessment of the variability of 
the plots within and between each transect zone. 
This method has been widely employed in vege- 
tation analysis in many ecosystems, and it gen- 
erates a "two-way" (plot-by-species) table. This 
is similar to a conventional "Braun-Blanquet Ta- 
ble" (Gauch, 1982) and is used to generate a den- 
drogram representing the hierarchical relationship 
between the data for each plot. TWINSPAN also 
facilitates the detection of differential and other 
diagnostic species, that is, respectively, species 
whose occurrence in the sample plots corresponds 
exactly with the clusters of plots generated by the 
analysis at a particular hierarchical level and spe- 
cies that are only present in most but not all plots 
within a particular cluster. Such species may be 
useful indicators of the specific plant communities 
recognized. The analysis was conducted using the 
TWINSPAN computer program for the MS-DOS 
operating system, developed and copyrighted by 
Microcomputer Power (111 Clover Lane, Ithaca, 
NY 14850, USA). 

Decorana — DECORANA provides a means of 
ordinating sample plots and species on a scatter 
diagram against axes obtained from an iterative 
process derived from reciprocal averaging 
(Gauch, 1982). Like TWINSPAN, DECORANA 
has been widely employed in quantitative vege- 
tation studies and serves to assess the extent of 
clustering of plots in term of species composition 
and importance. It is expected that the relative 



position of plot clusters along the principal axis 
(x) would correspond to trends between the dif- 
ferent plots. These trends may relate to environ- 
mental gradients, successionary relationships, or 
other factors. Plotting the corresponding ordina- 
tion for the individual species on equivalent axes 
provides a means of detecting species with cor- 
responding ordination values. Correlation be- 
tween clusters of species and clusters of samples 
on the plotted graphs is used to determine char- 
acteristic species for each vegetation type. These 
species are those that tend to correspond to par- 
ticular clusters of plots, but which are less strong- 
ly indicative of a particular plot cluster than the 
differential and diagnostic species revealed by 
TWINSPAN. DECORANA is usually performed 
in conjunction with TWINSPAN, and it further 
assists in recognizing, distinguishing, and char- 
acterizing distinct plant communities. The analy- 
sis was conducted using the detrended correspon- 
dence analysis option within the CANOCO com- 
puter program, written by C. J. F. Ter Braak (In- 
stitute of Applied Computer Science, Box 100, 
6700 AC Wageningen, The Netherlands). 

Canoco — CANOCO analysis also provides a 
means of ordinating sample plots and species on 
a scatter diagram, like DECORANA. However, 
with CANOCO the iteration takes into account 
environmental data on a plot-by-plot basis. Thus, 
axes may be drawn that represent specific envi- 
ronmental gradients, thereby indicating how these 
may relate to the different vegetation types or 
communities that have been recognized. The 
lengths of these axes indicate the relative impor- 
tance of the environmental variables in influenc- 
ing the vegetation patterns discovered. The inclu- 
sion or exclusion of specific environmental vari- 
ables in the analysis helps to determine which en- 
vironmental factors correlate strongly with the 
botanical data and therefore play significant roles 
in shaping the type of plant community that oc- 
curs at a specific site. Interpretation of the species 
ordination generated by CANOCO provides in- 
sight into the relationship between the occurrence 
of individual species and the environmental vari- 
ables, thus revealing possible indicator species. 
Like DECORANA, it can also provide a means 
of detecting diagnostic species for particular veg- 
etation types. A criticism of CANOCO analysis 
has been that the underlying assumptions of lin- 
earity of species responses and environmental ef- 
fects may be too stringent a requirement for much 
ecological data (Gauch, 1982). The environmental 
variables analyzed were: (1) disturbance: plots 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



29 



Table 4- 1 . Environmental data for each plot. 



Transect no. Plot no. 



Disturbed Elevation (m) 



Aspect 



Slope 



Topography 



1 

2 

3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 

15 

16 

17 

18 

19 

20 



Yes 
Yes 
Yes 
Yes 
Yes 
No 
No 
No 
No 
No 
No 
No 
No 
No 
No 
No 
No 
No 
No 
No 



715 

740 

750 

785 

775 

815 

810 

860 

820 

820 

1,245 

1,250 

1,310 

1,280 

1,220 

1,660 

1,620 

1,550 

1,630 

1,620 



SW 
WSW 
WSW 
WNW 

W 

SE 

SW 

S 

SE 
W 

NW 

S 
NW 
NW 

W 

w 

E 
W 



35° 
35° 
35° 
40° 
15° 
35° 

0° 
30° 
25° 

0° 
40° 
45° 
40° 
35° 
40° 
30° 
40° 
35° 
45° 
40° 



Valley 

Valley 

Ridge 

Ridge 

Ridge 

Valley 

Valley 

Ridge 

Valley 

Valley 

Ridge 

Ridge 

Ridge 

Valley 

Valley 

Ridge 

Mid-slope 

Mid-slope 

Ridge 

Ridge 



around transect 1 were classified as disturbed and 
plots around transects 2, 3, and 4 were classified 
as undisturbed (plots in transect 2 were not in dis- 
turbed forest, although areas within this zone 
showed signs of cutting and cattle browsing); (2) 
altitude: absolute values in m; (3) aspect: separate 
east-west and north-south components were 
scored, each on a 3-point scale, the median value 
corresponding to no slope with respect to the par- 
ticular axis; (4) slope angle: reduced to six class- 
es, 0-9°, 10-19°, 20-29°, 30-39°, >40°; and (5) 
plot topography: classified as valley, mid-slope, or 
ridge. The analysis was conducted using the 
CANOCO computer program. 



Results 

Environmental Data 

The environmental data recorded for each sam- 
ple plot are given in Table 4- 1 . 

Structural Characteristics — Trees 

Data on the structural characteristics of the 
trees recorded in the plots at each transect, with 
analyses of variance, are given in Table 4-2. 

Numbers of Trees — The mean number of trees 
with dbh >10 cm in the plots for each transect 



30 



zone is given in Table 4-2. It ranged from a min- 
imum of 21.4 stems at transect 2 to 28.0 stems at 
transect 3, equivalent to 1,070 and 1,400 
stems/ha, respectively. There were no significant 
differences in number of trees either between 
transects 1 and 2 or between transects 2, 3, and 
4. The number of trees differed considerably 
among the individual plots, ranging from 16 trees 
(in plot 2 of transect 1) to 34 trees (in plot 13 of 
transect 3). 

dbh Measurements — Trees with dbh measure- 
ments >25 cm are listed for each transect in Ta- 
bles 4-3 through 4-6. The dbh values were sorted 
into size classes at intervals of 5 cm, and histo- 
grams of frequency were plotted for each transect 
(Fig. 4-1). This shows that the majority of the 
trees at all four transect zones had values in the 
smallest class, falling between 10 and 14.9 cm. 
The numbers of trees falling within the larger size 
classes decreased sharply at all elevations sam- 
pled within the reserve; only a few trees were in 
each size class with in which the dbh was >35 
cm. 

Clear differences in tree dbh were found be- 
tween transect zones (Fig. 4-1). Forest outside the 
reserve boundaries, in the plots at the 720 m tran- 
sect zone, had no trees with dbh >35 cm and only 
1 1 trees with dbh >25 cm. Forest within the re- 
serve, in the 810 m transect zone, had 21 trees 
with a dbh >25 cm, nine of which had a dbh >35 
cm. Indeed, the largest tree recorded in the study, 
Sloanea rhodantha var. rhodantha form "quadri- 

FIELDIANA: ZOOLOGY 






§ 
St 

r 



-«5 



- s 



Tt Os <N <N 
<*' Tt ■*' "*' 

I I 

oo r'i f^i r^ 
<N ci c""5 ci 



o o o o 
+1 +1 +1 +1 
**! "1 •"": **! 

CS ^ f*1 CI 



vO <N ' . 

— — Ov 
I I I 

• CCS 

p s0 m 
(so- 
il +1 +1 
so r» ov 
•«t — r^ 



rn «■« >n <n 

od so so 00 

C> 00 SO SO 

l I II 

r-- p so r-; 

r^ od od (N 

-MMN 



'00 00 



r^ po oo o 

r~ <n — — 

+1 +1 +1 +1 

*n rf — fsj 

so ro os en 
fS rf Tt •* 



os o on r» 
r^ — d © 

— <n <n rsi 
1 i _L J. 

© <N f) 00 

io r-' >rJ Tt 

^^ y y 

<n ^ o r» ** *- 

— — rj rsi 

+1 +1 +1 +1 
oo v> r-; © 
«ri oo od r~ 



■* r^ -<r fs) 

& rsi & f> 

so r» <A o 

— — <N (S 

SO «Q ^ p V3 00 
prj SO !"•; SO Jr y> 

r~ rn c«S ■** 

+1 +1 +1 +1 

00 rf O 00 



^ -. 6< P*> 

— EG-* 

E E O «/-> < < = 

oo^>>^ 

— oj o •<* < < J 



E 
I 

3 

I 
'E 
I 

u 

60 

« ■ "3 <f 

1 £«> 

S o- «= «s w 
Ji C u u Jg 

£ w C E £ 

i) ii 5 J C 

S.g c c § 

+i S JS js -S 

<« ° <o « a 

g i*. y O C 

E I § i is 

— -^ — — _ 
£ ° ° s 

— c .2 .52 H 

^ ■•" '3 '35 tn 

2 sB5»l 

c * 2 2 wi 
•- Si § S -a 

s ' a © © i 

"3 2 a 12 — 

a 1 1 1 Q 

W t« o» 4J 00 

3 - £* DC _ 



loba" (dbh 81.1 cm), was found in this transect 
zone. 

Transect 3 (1210 m) also had 21 trees with a 
dbh >25 cm. Transect 4 (1625 m) had 65% of 
trees in the smallest size class (10-14.9 cm dbh) 
and few that exceeded 25 cm dbh, although this 
zone had the largest number of trees (five) with a 
dbh >50 cm. 

There were no significant differences in dbh 
values between transect 1 and 2 or between tran- 
sects 2, 3, and 4 (see Table 4-2). It should, how- 
ever, be noted that the intra-transect variability of 
this measurement was high, particularly at tran- 
sect 2, and the differences between dbh measure- 
ments at transects 1 and 2 were almost significant 
at the 5% probability level. 

Basal Area — The lowest total basal area was 
calculated for transect 1 at 26.5 m 2 /ha. The basal 
areas for transects 2 through 4 ranged from 43.2 
to 49.1 m 2 /ha (Table 4-2). There were no statis- 
tically significant differences between transects 
among the total basal area values; this result was 
expected because they are calculated from the dbh 
measurements. 

Tree Height — Tree height was segregated into 
five size classes for each of the transects (Fig. 
4-2). Although no attempt has been made to sta- 
tistically examine the correlation between tree 
height and tree dbh, some general trends can be 
drawn from the data. For example, in transect 4 
the trees of a particular dbh class tend to be short- 
er than trees of similar dbh in the lower transects 
(compare Fig. 4- 1 with Fig. 4-2). In transect 3 the 
same effect is apparent, although it is less well 
defined. Calculating the mean height/dbh values 
for all trees with dbh >25 cm for transects 1 
through 4 (values from Tables 4-3 through 4-7), 
values of 57.3, 58.8, 49.0, and 37.1, respectively, 
are obtained, confirming that at higher elevations 
these trees tend to have a lower height-to-dbh ra- 
tio. 

Tree height was significantly greater in transect 
zone 2, within the reserve boundaries, than out- 
side the reserve in transect zone 1 (P ^ 0.05). 
Furthermore, the number of tall trees (>20 m) 
was lower in transect zone 1 than zone 2. 

Seventy-five percent of the trees in transect 4 
were within the 5-10 m height range, indicating 
a much lower canopy. Tree height decreased sig- 
nificantly (P ^ 0.05) from transect zone 2 to 3 
and from transect 3 to 4, indicating a clear de- 
crease in tree height with increasing elevation. 
Mean tree height as a function of elevation in 
transects 2 through 4 showed that a significant 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



31 



Table 4-3. Transect 1 (720 m). Trees with dbh >25 cm arranged in decreasing size. 







dbh 


Basal area 


Height 


Family 


Genus/species 


(cm) 


(m 2 ) 


(m) 


Lauraceae 


sp. 3 


34.8 


0.095 


20 


Lauraceae 


sp. 10 


31.3 


0.077 


12 


Clusiaceae 


Mamtnea sp. 6 


30.3 


0.072 


15 


Not recorded 




30.0 


0.071 


20 


Rubiaceae 


Gaertnera sp. 3 


29.7 


0.069 


14 


Lauraceae 


Cryptocarya sp. 3 


28.4 


0.063 


16 


Lauraceae 


sp. 16 


27.2 


0.058 


15 


Lauraceae 


Cryptocarya madagascariensis 


25.5 


0.051 


18 


Burseraceae 


Canarium madagascarien.se 


25.2 


0.050 


18 


Lauraceae 


sp. 18 


25.0 


0.058 


15 



negative relationship existed (R 2 = 0.79, P < 
0.001; Fig. 4-3). 

Crown Diameter — Mean values for crown di- 
ameter ranged from 3.3 (transects 1 and 4) to 4.1 
m (transect 2). Crown diameter was significantly 
greater in transect zone 2 than in zone 4 (P < 
0.05), but otherwise none of the differences be- 
tween transects were statistically significant based 
on the two ANOVA tests (Table 4-2). 

Structural Characteristics — Lianas 
and Epiphytes 

Number of Lianas — The mean number of li- 
anas per tree ranged from 0.4 to 1.2 in the four 



transect zones (Table 4-7). None of the individual 
lianas was greater than 10 cm dbh. The difference 
in mean number of lianas per tree between tran- 
sects 1 and 2 or between 2, 3, and 4 was not 
significantly different. As with tree height, obser- 
vations suggest that this parameter is influenced 
by microhabitat. 

Epiphytes — Using the visual scale in tree 
crowns, mean values of vascular epiphytes ranged 
from 1.2 (transect 1) to 2.2 (transect 4) and on 
tree trunks from 0.6 (transects 1 and 3) to 1.0 
(transects 2 and 4) (Table 4-7). There were gen- 
erally more epiphytes in the crowns of trees than 
on the trunks. No statistically significant differ- 
ence was observed in the numbers of vascular ep- 
iphytes between transects. 



Table 4-4. Transect 2 (810 m). Trees with dbh >25 cm arranged in decreasing size. 







dbh 


Basal area 


Height 


Family 


Genus/species 


(cm) 


(m 2 ) 


(m) 


Elaeocarpaceae 


Sloanea "quadriloba" 


81.1 


0.517 


28 


Sapindaceae 


sp. 4 


75.8 


0.451 


27 


Lauraceae 


Cryptocarya sp. 3 


61.5 


0.297 


30 


Clusiaceae 


Calophyllum drouhardi 


54.8 


0.236 


25 


Elaeocarpaceae 


Sloanea "quadriloba" 


48.8 


0.187 


18 


Elaeocarpaceae 


Sloanea "quadriloba" 


42.0 


0.139 


16 


Lauraceae 


Cryptocarya sp. 3 


36.2 


0.103 


28 


Fabaceae 


Dalbergia sp. 1 


36.1 


0.102 


20 


Myrtaceae 


Syzygium sp. 2 


35.8 


0.101 


22 


Cunoniaceae 


Weinmannia sp. 6 


34.7 


0.095 


25 


Clusiaceae 


Calophyllum drouhardi 


31.8 


0.079 


18 


Sterculiaceae 


Dombeya sp. 3 


31.7 


0.079 


16 


Monimiaceae 


Tambourissa thouvenotii 


30.9 


0.075 


12 


Lauraceae 


sp. 10 


30.5 


0.073 


20 


Cunoniaceae 


Weinmannia sp. 1 


29.2 


0.067 


20 


Cunoniaceae 


Weinmannia sp. 1 


28.8 


0.065 


18 


Moraceae 


Streblus dimepate 


28.2 


0.062 


20 


Lauraceae 


Cryptocarya sp. 3 


26.5 


0.055 


25 


Lauraceae 


Cryptocarya sp. 3 


25.6 


0.051 


22 


Violaceae 


Rinorea cf. arborea 


25.0 


0.049 


14 


Lauraceae 


Cryptocarya sp. 3 


25.0 


0.049 


20 



Note: records of Sloanea quadriloba refer to Sloanea rhodantha var. rhodantha form "quadriloba. 



32 



FIELDIANA: ZOOLOGY 



Table 4-5. Transect 3 (1,210 m). Trees with dbh >25 cm arranged in decreasing size. 







dbh 


Basal area 


Height 


Family 


Genus/species 


(cm) 


(m 1 ) 


(m) 


Euphorbiaceae 


Cleistanthus boivinianus 


70.0 


0.385 


20 


Clusiaceae 


Calophyllum sp. 1 


57.5 


0.260 


20 


Lauraceae 


sp. 1 


54.2 


0.231 


20 


Clusiaceae 


Calophyllum drouhardi 


47.0 


0.173 


22 


Clusiaceae 


Calophyllum drouhardi 


44.5 


0.156 


20 


Aquifoliaceae 


Hex mitis 


38.6 


0.117 


22 


Euphorbiaceae 


Mallotus capuronii 


35.4 


0.098 


16 


Myrtaceae 


Syzygium sp. 2 


35.2 


0.097 


17 


Euphorbiaceae 


Cleistanthus boivinianus 


34.0 


0.091 


16 


Euphorbiaceae 


Cleistanthus boivinianus 


31.4 


0.077 


16 


Aquifoliaceae 


Ilex mitis 


31.3 


0.077 


14 


Myrtaceae 


Syzygium sp. 16 


31.0 


0.075 


18 


Araliaceae 


Schefflera myriantha 


30.9 


0.075 


14 


Annonaceae 


Polyalthia humbertii 


30.5 


0.073 


20 


Euphorbiaceae 


Cleistanthus boivinianus 


30.1 


0.071 


17 


Araliaceae 


Polyscias sp. 4 


29.9 


0.070 


16 


Euphorbiaceae 


Mallotus cf. capuronii 


29.5 


0.068 


16 


Rutaceae 


Evodia sp. 1 


29.2 


0.067 


12 


Icacinaceae 


sp. 3 


27.9 


0.061 


14 


Lauraceae 


sp. 2 


27.9 


0.061 


15 


Lauraceae 


Cryptocarya crassifolia 


27.7 


0.060 


18 



Mean values for estimates of bryophytes and 
lichens in tree crowns ranged from 0.6 (transect 
1) to 3.0, the highest value possible, in transect 4, 
and on tree trunks from 0.4 (transects 1 and 3) to 
2.6 (transect 4). Differences in these values were 
not statistically significant between transects 1 
and 2, but the value for tree trunks was signifi- 
cantly higher for transect 4 than for transects 2 
and 3 (P ^ 0.01). For tree crowns the value for 
transect 4 was significantly higher than that for 
transect 2 (P < 0.05) (Table 4-7). The abundance 
of epiphytic bryophytes and lichens increased 
with elevation. 

Casual observations suggested that epiphytes 
were locally abundant in certain areas of forest in 
all four transect zones. Local variation in micro- 
environmental factors such as shelter, humidity, 
and host plant characteristics also influence epi- 
phyte distribution. 



Forest Floristic Composition 

Diversity — The number of plant families and 
species found in each plot and the number of spe- 
cies in each plot new to the transect are given in 
Table 4-8; the statistical analysis of this informa- 
tion is given in Table 4-9. 

Forty-four plant species (dbh ^ 10 cm) were 
found in the plots in transect 1 (Table 4-10), 54 



species in transect 2 (Table 4-11), 49 species in 
transect 3 (Table 4-12), and 51 species in transect 
4 (Table 4- 1 3). A smaller number of families were 
also represented in transect 1 (20) than at any of 
the other three transects (26-28). However, the 
lower numbers of species and families represented 
in transect 1 compared to transect 2 were not sta- 
tistically significant (Table 4-9), due to the high 
level of variation among the plots. There was also 
no statistical difference in species or family di- 
versity with increasing elevation (Table 4-9). 

In order to determine whether the five 10 X 20 
m plots were effective in sampling the flora of 
each elevational zone, the total cumulative num- 
bers of species per transect were plotted against 
plot number (Fig. 4-4). The extent to which the 
resultant species accumulation curves reach an as- 
ymptote indicates how adequately the five plots 
represent the entire floristic diversity of each tran- 
sect zone. On the basis of these plots, it is clear 
that the survey does not adequately represent the 
species diversity in any of the elevation zones, 
because all of the curves fail to show any sign of 
approaching a plateau at the fifth plot. 

Floristics — The contribution of each plant 
family to the vegetation in terms of numbers of 
stems and total basal area at each transect is given 
in Tables 4-10 through 4-13. Figures 4-5 and 4-6 
show the relative abundance of plant families cal- 
culated from the number of stems present in a 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



33 



Table 4-6. Transect 4 (1,625 m). Trees with dbh >25 cm arranged in decreasing size. 



Family 



Genus/species 



dbh 
(cm) 



Basal area 

(m 2 ) 



Height 

(m) 



Lauraceae 

Violaceae 

Myrtaceae 

Lauraceae 

Lauraceae 

Lauraceae 

Araliaceae 

Sapotaceae 

Canellaceae 

Cunoniaceae 

Lauraceae 

Verbenaceae 

Lauraceae 

Rutaceae 

Lauraceae 

Sterculiaceae 

Lauraceae 

Lauraceae 



Ocotea sp. 2 
Rinorea sp. 2 
Syzygium sp. 9 
Ocotea sp. 2 
sp. 2 

Ocotea sp. 2 
Schefflera myriantha 
Faucheria sp. 1 
Cinnamosma fragrans 
Weinmannia sp. 7 
Ocotea sp. 2 
Wtex sp. 1 

Cryptocarya crassifolia 
Evodia sp. 2 
Ocotea sp. 2 
Dombeya sp. 2 
Ocotea sp. 2 
Cryptocarya sp. 1 



67.5 


0.358 


14 


57.5 


0.260 


16 


56.0 


0.246 


11 


54.7 


0.235 


15 


51.4 


0.207 


15 


47.0 


0.173 


11 


35.0 


0.096 


12 


33.5 


0.088 


12 


33.4 


0.088 


13 


32.5 


0.083 


12 


31.3 


0.077 


12 


30.6 


0.074 


14 


29.8 


0.070 


13 


29.5 


0.068 


14 


27.8 


0.061 


14 


27.0 


0.057 


15 


26.0 


0.053 


12 


26.0 


0.053 


12 



100 



I 

PL, 





Transect 1 (720 m) 
Transect 2 (810 m) 
Transect 3 (1210m) 
LZZ3 Transect 4 (1625 m) 



10-14.9 15-19.9 20-24.9 25-29.9 30-34.9 35-39.9 40-44.9 45-49.9 >50.0 
DBH size class range (cm) 

Fig. 4-1. Frequency by dbh (diameter at breast height) size class for trees of > 10 cm dbh in each transect. 



34 



FIELDIANA: ZOOLOGY 



u 



120 



100 



80 



60 



40 



20 



I— "^ Transect 1 (720 m) 

LZH Transect 2 (810 m) 

H Transect 3 (1210 m) 

EI3 Transect 4 (1625 m) 



= LB 




<5 



5-10 10-15 15-20 

Tree height class (m) 



>20 



Fig. 4-2. Frequency of trees of s 10 cm dbh in each height class in each transect. 



Table 4-7. Structural characteristics of the lianas and epiphytes in each transect zone, representing mean values 
of five plots per transect (mean values given ± standard deviation, with the minimum and maximum values in 
parentheses), with analyses of variance. 



Transect 



No. of lianas 
per tree 



Epiphytes in 
tree crown* 



Epiphytes on 
tree trunk* 



Moss and lichens Moss and lichens 
in tree crown* on tree trunk* 



1 (710 m) 

2 (810 m) 

3 (1,210 m) 
4(1,625 m) 
ANOVA 1/2+ 
ANOVA 2/3/4$ 
LSD§ 



0.8 ±0.21 (0.6-1.1) 
1.2 ±0.85 (0.1-2.2) 
0.4 ±0.18 (0.2-0.6) 
0.9 ±0.49 (0.3-1.4) 

NS 
NS 



1.2 ±0.45 (1-2) 
1.6 ±0.90 (1-3) 
2.0 ±0.71 (1-3) 
2.2 ± 0.45 (2-3) 

NS 
NS 



0.6 ±0.55 (0-1) 
1.0 ±0.71 (0-2) 
0.6 ±0.55 (0-1) 
1.0 ±0 (1-1) 

NS 

NS 



0.6 ± 1.3(0-3) 

1.2 ± 1.3(0-3) 

2.0 ± (2-2) 

3.0 ±0 (3-3) 

NS 
** 

1.1 



0.4 ± 0.89 (0-2) 

0.8 ± 1.10(0-2) 

0.4 ±0.55 (0-1) 

2.6 ± 0.55 (2-3) 

NS 
*** 

1.3 



* Values attributed on a 0-3 scale (see text for explanation). 

+ Results of the analysis of variance between transects 1 and 2: NS, not significant; **, significant at P ^ 0.05 
level. 

t Results of the analysis of variance between transects 2, 3, and 4: NS, not significant; **, significant at P ^ 0.05 
level; ***, significant at P ^ 0.01 level. 

§ LSD, least significant difference between means (ANOVA 2/3/4). 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



35 




1000 1200 

Elevation (m) 



Fig. 4-3. Regression line showing tree height re- 
gressed against elevation for plots in transects 2 through 
4. Calculated using mean tree height values for each 
plot. 



transect zone and the basal areas, respectively. In 
the forest outside the reserve, at 720 m (transect 
zone 1), the plant family with the largest number 
of stems was the Violaceae, comprising a single 
species, Rinorea cf. arborea, an understorey or 
canopy tree. Lauraceae (Cryptocarya and Ocotea) 
was the next most common family, followed by 
Myrtaceae (Syzygium), and Rubiaceae (Gaert- 
nera, Psychotria, and Tricalysia) (Table 4-10). 
The basal area data (Fig. 4-6) from transect zone 
1 clearly reflect the status of Lauraceae as large 



canopy trees because they form 30.6% of the total 
basal area (Table 4-10). 

Although not found in any of the plots, Rav- 
enala madagascariensis Sonn. (Strelitziaceae) 
was frequent in the forest outside of the reserve. 
This species is common in Malagasy rain forest 
up to 1000 m elevation. It occurs primarily in ar- 
eas with significant rainfall, both in primary rain 
forest and open secondary growth. It invades dis- 
turbed forest and cleared areas (Kress et al., 
1994). At 720 m (transect 1), R. madagascariensis 
was abundant in disturbed areas of forest. 

The forest inside the reserve boundaries at 810 
m (transect 2) was similar in floristic composition 
to that found outside at 720 m (transect 1). The 
greatest number of stems were Violaceae (Rino- 
rea), followed by Lauraceae (Cryptocarya and 
Ocotea), Myrtaceae (Syzygium), and Rubiaceae 
(Gaertnera, Psychotria, and Tricalysia) (Table 
4-11). The basal area data (Fig. 4-6) show that 
Elaeocarpaceae (Sloanea) formed nearly 20% of 
the total basal area. Sloanea are large, emergent, 
and often buttressed trees. Lauraceae were the 
second most important family, as estimated by to- 
tal basal area, forming almost 19% of the total. 
Sapindaceae formed 11% of the total basal area, 
though this can be traced to the occurrence of a 
single 75.8 cm dbh tree. Twelve families repre- 
sented in zone 2 were not found in transect zone 



Table 4-8. Family and species diversity at each plot. 













Percent of 






No. of 


No. of 


No. of species 


species new 


Transect 


Plot 


families 


species 


new to transect 


to transect 


1 (720 m) 


1 


6 


8 


8 


100 




2 


9 


11 


8 


72.7 




3 


9 


10 


9 


90.0 




4 


9 


14 


8 


57.1 




5 


7 


13 


10 


76.9 


2 (810 m) 


6 


10 


10 


10 


100 




7 


15 


18 


16 


88.9 




8 


10 


15 


14 


93.3 




9 


10 


11 


7 


63.6 




10 


9 


12 


7 


58.3 


3 (1,210 m) 


11 


11 


12 


12 


100 




12 


9 


10 


7 


70.0 




13 


13 


21 


14 


66.7 




14 


15 


21 


12 


57.1 




15 


11 


12 


8 


66.7 


4(1,625 m) 


16 


13 


14 


14 


100 




17 


10 


16 


12 


75.0 




18 


15 


21 


13 


61.9 




19 


10 


14 


5 


35.7 




20 


14 


14 


8 


57.1 



36 



FIELDIANA: ZOOLOGY 






Table 4-9. Family and species diversity of the trees in each transect zone, representing mean values of five plots 
per transect, with analyses of variance. 



Transect 



No. of families per plot* 



No. of species per plot*t 



1 (720 m) 

2 (810 m) 

3 (1,210 m) 
4(1,625 m) 
ANOVA 1/2$ 
ANOVA 2/3/4§ 



8.2 ± 1.30(6-9) 
10.2 ±2.78 (8-15) 
11.0 ±2.12 (8-14) 
12.4 ± 1.82(10-14) 

NS 

NS 



11.2 ±2.39 (8-14) 
13.2 ± 3.27(10-18) 
15.2 ±5.36(10-21) 
15.8 ±3.03 (14-21) 

NS 
NS 



* Mean values given ± standard deviation, with minimum and maximum values in parentheses. 

t Includes morphospecies. 

$ Results of the analysis of variance between transects 1 and 2: NS, not significant. 

§ Results of the analysis of variance between transects 2, 3, and 4: NS, not significant. 



1 , whereas only four that were represented in tran- 
sect zone 1 were absent in zone 2. 

In transect 3 (1210 m), Clusiaceae (Mammea, 
Symphonic and Calophyllum) had the largest 
number of stems; together with Euphorbiaceae 
(Croton, Cleistanthus, and Mallotus) they ac- 
counted for 43% of the trees sampled (Table 4-12, 
Fig. 4-3). Other important families included Lau- 
raceae (Ocotea and Cryptocarya), Myrtaceae (Sy- 
zygium), Rubiaceae (Schismatoclada, Gaertnera, 
Canthium, and Psychotria), and Araliaceae (Po- 
lyscias and Schefflera). The basal area data and 
stem data show the same trend, with Clusiaceae 
and Euphorbiaceae forming nearly 43% of the to- 



tal basal area. Elaeocarpaceae form 10% of the 
basal area, but only 4% of the total number of 
stems. Five families were represented that were 
not recorded in transect zones 1 and 2, whereas 
nine families were absent that were found at lower 
elevation. 

At transect 4 (1625 m) there was a dramatic 
shift in forest composition. Lauraceae formed 
1 1 % of the total number of stems sampled. The 
other most important families were Podocarpa- 
ceae (Podocarpus), Araliaceae (Polyscias and 
Schefflera), Cunoniaceae (Weinmannia), and Pan- 
danaceae (Pandanus) (Table 4-13, Fig. 4-3). Eu- 
phorbiaceae were distinctly less abundant than at 



Table 4-10. Plant families in transect 1 (720 m) listed in descending order of stem number. Information on basal 
area and no. of species present is also provided. 









Percent 




Basal 


Percent 






Ranked by 


No. of 


of total 


Ranked by 


area 


of total 


No. of 


Family 


stem no. 


stems 


stem no. 


basal area 


(m 2 ) 


basal area 


species 


Violaceae 


1 


34 


28.6 


2 


0.47 


17.7 


1 


Lauraceae 


2 


26 


21.8 


1 


0.86 


32.5 


9 


Myrtaceae 


3 


15 


12.6 


3 


0.34 


12.8 


6 


Rubiaceae 


4 


7 


5.9 


4 


0.17 


6.4 


5 


Clusiaceae 


5 


6 


5.0 


5 


0.15 


5.7 


4 


Monimiaceae 


6 


6 


5.0 


6 


0.10 


3.8 


2 


Elaeocarpaceae 


7 


4 


3.4 


7 


0.08 


3.0 


1 


Sapindaceae 


8 


4 


3.4 


9 


0.06 


2.3 


1 


Oleaceae 


9 


4 


3.4 


10 


0.05 


1.9 


2 


Sapotaceae 


10 


2 


1.7 


12 


0.05 


1.9 


2 


Apocynaceae 


11 


2 


1.7 


17 


0.03 


1.1 


2 


Not determined 


12 




0.8 


8 


0.07 


2.6 


1 


Burseraceae 


13 




0.8 


11 


0.05 


1.9 


1 


Tiliaceae 


14 




0.8 


13 


0.04 


1.5 


1 


Cunoniaceae 


15 




0.8 


14 


0.04 


1.5 


1 


Fabaceae 


16 




0.8 


15 


0.03 


1.1 


1 


Sterculiaceae 


17 




0.8 


16 


0.03 


1.1 


1 


Annonaceae 


18 




0.8 


18 


0.01 


0.4 


1 


Melastomataceae 


19 




0.8 


19 


0.01 


0.4 


1 


Araliaceae 


20 




0.8 


20 


0.01 


0.4 


1 






119 


100 




2.65 


100 


44 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



37 



Table 4-11. Plant families in transect 2 (810 m) listed in descending order of stem number. Information on basal 
area and no. of species present is also provided. 









Percent 






Percent 






Ranked by 


No. of 


of total 


Ranked by 


Basal 


of total 


No. of 


Family 


stem no. 


stems 


stem no. 


basal area 


area (m 2 ) 


basal area 


species 


Violaceae 


1 


17 


15.9 


7 


0.20 


4.5 


1 


Lauraceae 


2 


16 


15.0 


2 


0.82 


18.9 


9 


Myrtaceae 


3 


10 


9.4 


6 


0.26 


6.0 


4 


Rubiaceae 


4 


9 


8.4 


10 


0.10 


2.3 


6 


Clusiaceae 


5 


7 


6.5 


4 


0.38 


8.7 


4 


Cunoniaceae 


6 


6 


5.6 


5 


0.35 


8.0 


2 


Eleocarpaceae 


7 


4 


3.7 


1 


0.86 


19.8 


1 


Euphorbiaceae 


8 


4 


3.7 


12 


0.07 


1.7 


2 


Sapindaceae 


9 


3 


2.8 


3 


0.47 


10.8 


3 


Moraceae 


10 


3 


2.8 


11 


0.09 


2.1 


1 


Araliaceae 


11 


3 


2.8 


14 


0.07 


1.6 


2 


Anacardiaceae 


12 


3 


2.8 


15 


0.07 


1.6 


1 


Dracaenaceae 


13 


3 


2.8 


16 


0.06 


1.4 


1 


Sterculiaceae 


14 


2 


1.9 


8 


0.12 


2.8 


2 


Fabaceae 


15 


2 


1.9 


9 


0.11 


2.5 


1 


Cyathaceae 


16 


2 


1.9 


20 


0.02 


0.5 


2 


Icacinaceae 


17 


2 


1.9 


22 


0.02 


0.5 


1 


Monimiaceae 


18 




0.9 


13 


0.07 


1.6 


1 


Sapotaceae 


19 




0.9 


17 


0.04 


2.1 


1 


Rutaceae 


20 




0.9 


18 


0.04 


2.1 


1 


Annonaceae 


21 




0.9 


19 


0.03 


0.7 


1 


Tiliaceae 


22 




0.9 


21 


0.02 


0.5 


1 


Dichapetalaceae 


23 




0.9 


23 


0.02 


0.5 


1 


Myrsinaceae 


24 




0.9 


24 


0.02 


0.5 


1 


Rhopalocarpaceae 


25 




0.9 


25 


0.01 


0.3 


1 


Aquifoliaceae 


26 




0.9 


26 


0.01 


0.3 


1 


Melanophyllaceae 


27 




0.9 


27 


0.01 


0.3 


1 


Burseraceae 


28 




0.9 


28 


0.01 


0.3 


1 






107 


100 




4.35 


100 


54 



lower elevations. With regard to basal area, Lau- 
raceae formed 34%, reflecting their stature as can- 
opy species (Table 4-13). Four of the six largest 
trees found in transect zone 4 were Lauraceae (Ta- 
ble 4-6). Cyatheaceae, Pandanaceae, and Podo- 
carpaceae were comparatively unimportant in 
terms of basal area (Table 4-13, Fig. 4-6), reflect- 
ing their small size as understorey trees, even 
though they were relatively common in the plots. 
Five families were recorded in zone 4 that had 
not been found at lower elevations, and eight fam- 
ilies that were present in transect zone 3 were ab- 
sent from zone 4. 



TWINSPAN 

Canopy Data — The dendrogram obtained from 
the crown diameter data with TWINSPAN is giv- 
en in Figure 4-7a. The first-level division sepa- 
rated the low-altitude plots (transects 1 and 2) 
from the high/mid-altitude plots (transects 3 and 



4), with the exception of plot 7, which was 
grouped with the mid-altitude plots. The second- 
level division within the mid/high-altitude cluster 
(including plot 7) partitioned the mid-altitude 
(transect 3) plots and plot 7 from the high-altitude 
(transect 4) plots. The third-level division divided 
plot 7 from the five plots of transect 3, and it 
separated plots 16 and 17 from plots 18, 19, and 
20 of transect 4. A fourth-level division separated 
plots 1 1 and 12 from plots 13, 14, and 15 of tran- 
sect 3. Within the low-altitude plots (excluding 
plot 7), plots 2 and 1 were each separated from 
the rest by second- and third-level divisions, re- 
spectively. Plots 4 and 5 were separated from the 
rest by a fourth-level division, whereas plots 3, 6, 
and 10 were separated from plots 8 and 9 by a 
fifth-level division. 

dbh Data — The dendrogram obtained from the 
dbh data using TWINSPAN is given in Figure 
4-7b. The first-level division separated perfectly 
the low-altitude plots (transects 1 and 2) from the 
high/mid-altitude plots (transects 3 and 4). The 



38 



FIELDIANA: ZOOLOGY 



Table 4-12. Plant families in transect 3 (1210 m) listed in descending order of stem number. Information on 
basal area and no. of species present in also provided. 









Percent 




Basal 


Percent 






Ranked by 


No. of 


of total 


Ranked by 


area 


of total 


No. of 


Family 


stem no. 


stems 


stem no. 


basal area 


(m 1 ) 


basal area 


species 


Clusiaceae 


1 


31 


22.3 


1 


1.18 


24.0 


4 


Euphorbiaceae 


2 


29 


20.9 


2 


0.92 


18.7 


3 


Lauraceae 


3 


9 


6.5 


4 


0.45 


9.1 


4 


Myrtaceae 


4 


9 


6.5 


5 


0.40 


8.1 


4 


Rubiaceae 


5 


8 


5.8 


11 


0.10 


2.0 


4 


Araliaceae 


6 


7 


5.0 


6 


0.26 


5.3 


4 


Eleocarpaceae 


7 


5 


3.6 


3 


0.50 


10.2 


2 


Monimiaceae 


8 


5 


3.6 


10 


0.13 


2.6 


1 


Aquifoliaceae 


9 


4 


2.9 


7 


0.25 


5.1 


1 


Icacinaceae 


10 


4 


2.9 


9 


0.13 


2.6 


1 


Ebenaceae 


11 


4 


2.9 


12 


0.07 


1.4 


1 


Annonaceae 


12 


3 


2.2 


8 


0.14 


2.8 


2 


Sterculiaceae 


13 


3 


2.2 


14 


0.06 


1.2 


1 


Oleaceae 


14 


3 


2.2 


15 


0.05 


1.0 


2 


Sapindaceae 


15 


2 


1.4 


16 


0.04 


0.8 


2 


Cunoniaceae 


16 


2 


1.4 


17 


0.03 


0.6 


2 


Erythroxylaceae 


17 


2 


1.4 


18 


0.02 


0.4 


2 


Myrsinaceae 


18 


2 


1.4 


20 


0.02 


0.4 


1 


Rutaceae 


19 


1 


0.7 


13 


0.07 


1.4 


1 


Rhamnaceae 


20 


1 


0.7 


19 


0.02 


0.4 


1 


Loganiaceae 


21 


1 


0.7 


21 


0.01 


0.2 


1 


Flacourtiaceae 


22 


1 


0.7 


22 


0.01 


0.2 


1 


Anacardiaceae 


23 


1 


0.7 


23 


0.01 


0.2 


1 


Cyathaceae 


24 


1 


0.7 


24 


0.01 


0.2 


1 


Not determined 


25 


1 


0.7 


25 


0.01 


0.2 


1 


Melastomataceae 


26 


1 


0.7 


26 


0.01 


0.2 


1 






140 


100 




4.92 


100 


49 



second-level division within the mid/high-altitude 
cluster divided perfectly the mid-altitude (transect 
3) plots from the high-altitude (transect 4) plots. 
The third-level division separated plot 15 from 
plots 11, 12, 13, and 14 of transect 3, and it sep- 
arated plots 16 and 17 from plots 18, 19, and 20 
of transect 4. Within the low-altitude plots, a sec- 
ond-level division divided plots 1, 2, and 5 from 
the rest. A third-level division separated plot 7, 
and a fourth-level division clustered plots 3, 4, 
and 6 together and plots 8, 9, and 10 together. 

Differential and Diagnostic Species — No dif- 
ferential species were recorded that separated the 
low-altitude plots (transects 1 and 2) from the 
mid/high-altitude plots (transects 3 and 4), even if 
the anomalous plot 7 was excluded. However, Ri- 
norea cf. arborea may be regarded as a diagnostic 
species, being present in seven of the 10 plots of 
transects 1 and 2 and absent from plots at tran- 
sects 3 and 4. 

Among plot clusters generated by TWINSPAN 
at lower division levels with either the canopy or 
the dbh data for the low-altitude transects, only 
the plot 8-9 cluster (crown data) had a differential 



species (Weinmannia sp. 1). Other plot clusters 
with diagnostic species were: plot cluster 3-4-10 
(canopy data) Antidesma petiolare and Gaertnera 
sp. 1; plot cluster 1-2-5 (dbh data) Cryptocarya 
sp. 3; and plot cluster 8-9-10 (dbh data) Dalber- 
gia sp. 1, Syzygium sp. 2, and Weinmannia sp. 1. 

Differential and diagnostic species for the plots 
in transect 3 were Cleistanthus boivinianus (all 
five plots), Mallotus cf. capuronii (four plots), and 
Lauraceae sp. 1 (three plots). Among the plot 
clusters at higher division levels, only the plot 1 1- 
13 cluster (canopy data) had differential species 
(Croton sp. 2 and Schismatoclada sp. 2). 

Differential and diagnostic species for the plots 
of transect 4 were Pandanus sp. 2 (all five plots), 
Weinmannia sp. 7 (four plots), Ocotea sp. 2 (three 
plots), Polyscias sp. 2, and the two diagnostic spe- 
cies of plot cluster 18-19-20 (listed below). Dif- 
ferential species for the two plot clusters that were 
obtained from both the canopy and the dbh data 
are: plot cluster 16-17 Casearia sp. 1, Croton sp. 
1, and Pandanus sp. 2; and plot cluster 18-19-20 
Podocarpus madagascariensis and Syzygium sp. 
4. The numbers of differential species for the plots 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



39 



Table 4-13. Plant families in transect 4 (1625 m) listed in descending order of stem number. Information on 
basal area and no. of species present is also provided. 









Percent 




Basal 


Percent 






Ranked by 


No. of 


of total 


Ranked by 


area 


of total 


No. of 


Family 


stem no. 


stems 


stem no. 


basal area 


(m 2 ) 


basal area 


species 


Lauraceae 


1 


16 


11.0 


1 


1.49 


34.5 


5 


Podocarpaceae 


2 


13 


9.0 


9 


0.16 


3.7 


1 


Araliaceae 


3 


11 


7.6 


2 


0.37 


6.3 


4 


Cunoniaceae 


4 


11 


7.6 


4 


0.27 


6.3 


3 


Pandanaceae 


5 


10 


6.9 


12 


0.10 


2.3 


2 


Flacourtiaceae 


6 


9 


6.2 


7 


0.19 


4.4 


2 


Cyathaceae 


7 


8 


5.5 


14 


0.09 


2.1 


1 


Myrtaceae 


8 


7 


4.8 


3 


0.33 


7.6 


3 


Rubiaceae 


9 


7 


4.8 


10 


0.16 


3.7 


6 


Sterculiaceae 


10 


6 


4.1 


6 


0.21 


4.9 


2 


Euphorbiaceae 


11 


6 


4.1 


11 


0.11 


2.5 


3 


Asteraceae 


12 


5 


3.5 


17 


0.06 


1.4 


1 


Sapotaceae 


13 


3 


2.1 


8 


0.17 


3.9 


1 


Elaeocarpaceae 


14 


3 


2.1 


16 


0.07 


1.6 


2 


Icacinaceae 


15 


3 


2.1 


19 


0.04 


0.9 


1 


Monimiaceae 


16 


3 


2.1 


20 


0.03 


0.7 


2 


Rutaceae 


17 


2 


1.4 


13 


0.10 


2.3 


2 


Verbenaceae 


18 


2 


1.4 


15 


0.08 


1.9 


1 


Myrsinaceae 


19 


2 


1.4 


21 


0.02 


0.5 


2 


Dichapetalaceae 


20 




0.7 


5 


0.26 


6.0 


1 


Aquifoliaceae 


21 




0.7 


18 


0.04 


0.9 


1 


Clusiaceae 


22 




0.7 


22 


0.02 


0.5 


1 


Erythroxylaceae 


23 




0.7 


23 


0.02 


0.5 


1 


Melanophyllaceae 


24 




0.7 


24 


0.02 


0.5 


1 


Canellaceae 


25 




0.7 


25 


0.01 


0.2 


1 


Ericaceae 


26 




0.7 


26 


0.01 


0.2 


1 






134 


100 




4.32 


100 


51 



that were separated from all other plots in the 
analyses were as follows: plot 1 (canopy data), 
four species; plot 2 (canopy data), six species; 
plot 7 (canopy and dbh data), 10 species; and plot 
15 (dbh data), three species. 



C 30- 

3 

2 

u 20 
'O 




transect 1 (720 m) 
transect 2 (810 m) 
transect 3 (1210 m) 
transect 4 (1625 m) 



1 2 3 4 5 6 

Plot number (replicate) 

Fig. 4-4. Species accumulation curves for transects 
1 through 4 in the RNI d'Andringitra. 



DECORANA 

The ordinations generated by DECORANA for 
the dbh data are given in Figure 4-8a (plots) and 
4-8b (species). Similar ordinations were obtained 
with the canopy (crown diameter) data. In both 
cases the five plots of transects 3 and the five plots 
of transect 4 are revealed as closely clustered 
points, but the 10 plots of transects 1 and 2 are 
not clearly separated and are much more scat- 
tered. Plots 7 and 8 are close to each other but 
are relatively far from the remainder of the tran- 
sect 1 and 2 plots. Plot 1 is shown to be an ex- 
treme outlier, although its spatial separation from 
the rest of the transect 1 and 2 plots is primarily 
with respect to the Y-axis; this is considered to 
be much less significant in DECORANA than 
separation against the X-axis (Gauch, 1982). The 
results of the canopy and dbh data differ only in 
subtle details. The canopy data placed plot 2 as 
an outlier among the transect 1 and 2 plots, 
whereas the dbh data placed plot 9 in an outlying 
position. The clustering of samples shows a close 
correspondence with the TWINSPAN results. 



40 



FIELDIANA: ZOOLOGY 



Olea Sa P J >< 




Myrt 



15 



others 

25 



720 m (119 trees) 




810 m (107 trees) 




others 
36 




1210 m (140 trees) 1625 m (134 trees) 

Fig. 4-5. Number of stems for each family. Calculated on a per transect basis. 



The characteristic species for each cluster de- 
termined from DECORANA also show little dif- 
ference between the canopy and dbh data. Char- 
acteristic species for the transect 1 and 2 plot clus- 
ter (excluding the outlier plots 2 and 9 and the 
extreme outlier plots 1 , 7, and 8) are given below 



(differences between the results from canopy data 
and dbh data are indicated): 

Albizia gummifera 
Antidesma petiolare 
Calophyllum paniculatum 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



41 



Olea 


others 






others 




0.05 v 


0.30__- 






0.60^-n 


bbb^ Eleo 


fc&'.JJs 


mL'A 


^k Laur 
Ik 0.81 


v u Rubi y 
Faba o.iUfl 

0.11 ^^^ MM 




^^0.86 


0.08 — -M 


*%kJ 




Ster /^ 


fc^\\// ^ 




Moni__ — — fl 






0,12 feeLi 


sSS^ 


HHHffii^^ 


o.io K 


^r. 




Myrt 1 


^^^mwi 


K^ 


Clus v= 


a^^H WW* ^ 




0.26 HBB 




L ^H 


o.i5 yt 


^r~ ' 




Euph^ 




1 \^H 

■ / ' ' V ^^B ••4' 


Rubi 






0.27 ^ 




1 ' ^'^^^ 


0.17 


X^£T^ 


^^ Euph 


Cunc 


i >^3B 


1- I 'V zS 




Myrt 
0.34 


^^^ 0.51 


0.35 


Clus 

0.38 


*^ — Sapi 
0.47 



720 m 



810 m 



Icac Moni 
0.14 ^ 0. 




1210 m 




Flac 
0.19 



Dich 



Myrt 



0.26 Cuno 3 ' 4 
0.27 

1625 m 



Fig. 4-6. Dominant plant families. Calculated from basal areas (mVO.l ha). 



Canarium madagascariense 
Cryptocarya madagascariensis 
Cryptocarya sp. 3 
Cyathea sp. 2 
Dichapetalum sp. 1 
Ephippiandra sp. 1 
Gaertnera sp. 1 
Chrysophyllum sp. 1 



Grewia sp. 2 

Lauraceae sp. 3 

Lauraceae sp. 5 

Lauraceae sp. 1 1 

Lauraceae sp. 12 

Lauraceae sp. 13 

Lauraceae sp. 14 

Mammea sp. 3 (dbh data only) 



42 



FIELDIANA: ZOOLOGY 




Fig. 4-7a (top). Dendrogram obtained from TWIN- 
SPAN using canopy data. 

Fig. 4-7b (bottom). Dendrogram obtained from 
TWINSPAN using dbh data. 

Mammea sp. 4 

Mammea sp. 5 

Noronhia sp. 7 

Oncostemum sp. 13 

Rinorea cf. arborea 

Sapindaceae sp. 1 

Syzygium sp. 5 (dbh data only) 

Syzygium sp. 6 

Syzygium sp. 8 

Syzygium sp. 1 1 

Syzygium sp. 14 

Syzygium sp. 17 

Weinmannia sp. 5 

The characteristic species of plot 2 are: 



1000 
800 
000 
400 
200 


-200 



PI 1-15 



PI6-20 



~& t 



(2\ W 



P2A10 



-000 -400 



<S 



PI 6-20 




P2-6,I0 



Fig. 4-8a (top). DECORANA ordination of sample 
plots for dbh data. PI, plot 1; P2-6.10, plots 2, 3, 4, 5, 
6, and 10; P9, plot 9; P7,8, plots 7 and 8; PI 1-15, plots 
11, 12, 13, 14, and 15; P16-20, plots 16, 17, 18, 19, and 
20. 

Fig. 4-8b (bottom). DECORANA ordination of spe- 
cies for dbh data, with corresponding clusters from or- 
dination of plots superimposed. PI, plot 1; P2-6,10, 
plots 2, 3, 4, 5, 6, and 10; P9, plot 9; P7,8, plots 7 and 
8; PI 1-15, plots 11, 12, 13, 14, and 15; P16-20, plots 
16, 17, 18, 19, and 20. 



Apocynaceae sp. 1 
Apocynaceae sp. 2 
Breonia sp. 
Labramia sp. 1 
Memecylon cf. longipetalum 
Polyscias sp. 3 

The characteristic species of plot 9 are: 

Dombeya sp. 3 

Dracaena xiphophylla 

Mammea sp. 4 

Psychotria sp. 2 (dbh data only) 

Rubiaceae sp. 1 

Sapindaceae sp. 2 

Streblus dimepate (dbh data only) 

Uapaca sp. 1 



The characteristic species of plot 1 , transect 1 , 
are: 

Lauraceae sp. 16 
Mammea sp. 6 
Noronhia sp. 6 
Psychotria sp. 5 

The characteristic species of plots 7 and 8, tran- 
sect 2, are: 

Apodytes sp. 1 
Canthium sp. 5 
Cyathea sp. 3 
Grewia sp. 1 
Lauraceae sp. 4 
Lauraceae sp. 6 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



43 



Lauraceae sp. 8 
Lauraceae sp. 9 
Melanophylla cf. alnifolia 
Protorhus sp. 1 
Psychotria sp. 1 
Psychotria sp. 6 
Rhopalocarpus sp. 1 
Sapindaceae sp. 4 (dbh data only) 
Vepris sp. 1 

The characteristic species of transect 3 are: 

Anthocleista madagascariensis 

Calophyllwn sp. 1 

Canthium sp. 4 

Cassinopsis tomentosa 

Cleistanthus boivinianus 

Croton sp. 2 

Diospyros sp. 4 

Dombeya cf. amplifolia 

Dombeya sp. 4 

Erythroxylum capitatum (canopy data only) 

Evodia sp. 1 

Gaertnera sp. 5 

Garcinia sp. 1 

Icacinaceae sp. 3 

Lauraceae sp. 1 

Lauraceae sp. 2 (canopy data only) 

Lauraceae sp. 15 

Macphersonia sp. 1 

Mallotus cf. capuronii 

Memecylon cf. ivohibense 

Micronychia madagascariensis 

Micronychia tsiramiramy 

Noronhia sp. 3 

Noronhia sp. 4 

Polyalthia humbertii 

Polyscias sp. 4 

Polyscias sp. 6 

Polyscias sp. 8 

Psychotria macrochlamys 

Rhamnaceae sp. 1 

Rinorea sp. 1 

Schismatoclada sp. 2 

Symphonia sp. 1 

Syzygium sp. 12 

Syzygium sp. 15 

Syzygium sp. 16 

Tambourissa cf. parvifolia 

The characteristic species of transect 4 are: 

Asteraceae sp. 1 
Casearia sp. 1 



Cinnamosma fragrans Baill. 

Clerodendrum sp. 2 

Croton sp. 1 

Cryptocarya sp. 1 

Cyathea sp. 1 

Dombeya sp. 1 

Dombeya sp. 2 

Euphorbiaceae sp. 1 

Evodia sp. 2 

Evodia sp. 3 

Faucheria sp. 1 

Gaertnera sp. 7 

Gaertnera sp. 8 

Icacinaceae sp. 1 

Lauraceae sp. 7 

Macaranga echinocarpa 

Melanophylla sp. 1 

Ocotea sp. 2 

Oncostemum sp. 15 

Pandanus sp. 2 

Podocarpus madagascariensis 

Polyscias sp. 1 

Polyscias sp. 2 

Polyscias sp. 5 

Polyscias sp. 9 

Psychotria sp. 4 

Rinorea sp. 2 

Rubiaceae sp. 4 

Syzygium sp. 4 

Syzygium sp. 7 

Syzygium sp. 9 

Syzygium sp. 10 

Tambourissa sp. 1 

Tambourissa sp. 2 

Tricalysia sp. 1 

Vaccinium sp. 1 

Vwcum sp. 1 

V7tex sp. 1 

Weinmannia sp. 3 

Weinmannia sp. 7 



CANOCO 

All Environmental Variables — The ordina- 
tions of plots generated by CANOCO for the dbh 
data using all six environmental variables are giv- 
en in Figure 4-9. The plots (1 through 10) of the 
low-altitude transects (1 and 2) are not closely 
clustered, but unlike the DECORANA results, the 
plots of the two transects do not overlap. Among 
the plots of transect 1 , plots 1 , 2, and 5 and plots 
3 and 4 cluster closely. Among the plots (6 
through 10) of transect 2, all are relatively well 



44 



FIELDIANA: ZOOLOGY 



300- 
















200- 




< 




e 
|| 


\ V 3 






ioo- 


A 




ri 


\ 


f v- c \ 








V.A 














-100- 


n 










^ d 


*" b 


d 


•200- 


. 


















*J 














•300- 
-400- 


y 




a 











•500 -400 -300 -200 -100 



100 200 300 400 500 



Fig. 4-9. CANOCO ordination of sample plots for 
dbh data, related to all environmental variables. 1 . tran- 
sect 1; 2, transect 2; 3, transect 3; 4, transect 4; a, dis- 
turbance (-172, -276); b, altitude (262, -24); c, slope 
(138, 71); d, S/N aspect (86, -63); e, E/W aspect (-27, 
152); f, position (62, 54). 



spaced. The sample plots of transect 3 are more 
closely clustered than the low-altitude transects; 
among these, plots 11, 12, and 13 are particularly 
close. The sample plots of transect 4 are all close- 
ly clustered. Of the six environmental variables 
considered, the extent of disturbance is the most 
significant factor influencing the ordination of the 
data, whereas altitude is the second most impor- 
tant. The other four variables are less significant; 
in decreasing order these are: the angle of slope, 
east/west aspect, north/south aspect, and topog- 
raphy. Similar ordinations were obtained with the 
canopy data. Ordinations of the species were also 
conducted, but the results are not presented be- 
cause they are less instructive than those for a 
reduced number of environmental variables, as re- 
ported below. 

Disturbance and Altitude Only — CANOCO 
analysis was also undertaken for both the tree dbh 
and crown diameter data sets, including only the 
two most significant environmental variables, dis- 
turbance and altitude. The ordinations of plots and 
species generated by CANOCO for the dbh data 
are given in Figure 4-10. The ordinations of plots 
(Fig. 4- 10a) confirm the clustering obtained for 
the full set of environmental variables. Whereas 
the plots fall into three distinct classes — low-, 
mid-, and high-altitude groups with respect to the 
axis for altitude — the plots at low altitude show 
some spread along the disturbance axis. The plots 
of transect 1 (classed as disturbed forest) do not 
overlap those of transect 2 (classed as undisturbed 
forest); however, there is considerable variation 
with respect to the disturbance axis within both 
transects. At transect 1 , sample plots 1 and 5 show 




-400 -200 200 400 000 




-400 -200 



200 400 000 



Fig. 4- 10a (top). CANOCO ordination of samples 
for dbh data, related to disturbance and elevation only. 
1, transect 1; 2, transect 2; 3, transect 3; 4, transect 4; 
a, disturbance (-180, 364); b, altitude (269, 46). 

Fig. 4- 10b (bottom). CANOCO ordination of spe- 
cies for dbh data, related to disturbance and elevation 
only. 1, low elevation/high disturbance; 2, low eleva- 
tion/moderate disturbance; 3, low elevation/slight dis- 
turbance; 4, low elevation/low disturbance; 5, low/mid- 
elevation intermediates; 6, mid-elevation; 7, mid/high- 
elevation intermediates; 8, high elevation; 9, Tricalysia 
sp. 2; 10, Xylopia sp. 1; 11, Rinorea cf. arborea; a, 
disturbance (-180, 364); b, altitude (269, 46). 



a higher value for disturbance than plots 2, 3, and 
4; at transect 2, plots 6 and 10 show higher values 
than plots 7, 8, and 9. Transects 3 and 4 show 
little effect of disturbance. The ordination of spe- 
cies (Fig. 4- 10b) shows clusters of species that 
correspond to the three altitude classes, and other 
species intermediate between them. Similarly, 
with respect to the disturbance gradient, two main 
concentrations of species are found at opposite 
extremes, and scattered or weakly clustered as in- 
termediate groups. The ordinations obtained with 
the canopy data gave similar results, and these are 
not presented. 

The clusters that correspond to the mid- and 
high-elevation plots (transects 3 and 4) consist of 
essentially the same species as revealed by DE- 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



45 



CORANA (see pp. 40-44) as characteristic of 
these vegetation types. Intermediate to these clus- 
ters are found the following (mid/high-elevation) 
species: 

Aphloia theiformis 

Cryptocarya crassifolia 

Elaeocarpus subserratus 

Erythroxylum sp. 2 

Oncostemum sp. 14 

Sloanea rhodantha from "quercifolia" 

Weinmannia sp. 2 

The following species are intermediate between 
the low- and mid-elevation species clusters: 

Polyscias sp. 7 
Syzygium sp. 2 

The species that form the low-elevation/low- 
disturbance cluster are: 

Antidesma petiolare 
Apodytes sp. 1 
Canthium sp. 5 
Cyathea sp. 2 
Cyathea sp. 3 
Dalbergia sp. 1 
Dichapetalum sp. 1 
Dombeya sp. 3 
Dracaena xiphophylla 
Gaertnera sp. 1 
Grewia sp. 1 
Lauraceae sp. 4 
Lauraceae sp. 5 
Lauraceae sp. 6 
Lauraceae sp. 8 
Lauraceae sp. 9 
Lauraceae sp. 14 
Mammea sp. 3 
Mammea sp. 4 
Melanophylla cf. alnifolia 
Oncostemum sp. 13 
Protorhus sp. 1 
Psychotria sp. 1 
Psychotria sp. 6 
Rhopalocarpus sp. 1 
Rubiaceae sp. 1 
Sapindaceae sp. 2 
Sapindaceae sp. 4 
Streblus dimepate 
Syzygium sp. 3 
Syzygium sp. 13 
Uapaca sp. 1 



Vepris sp. 1 
Weinmannia sp. 1 
Weinmannia sp. 6 

The following species form a low-eleva- 
tion/slight-disturbance group: 

Chrysophyllum sp. 1 

Cryptocarya sp. 2 

Dombeya cf. spectabilis 

Lauraceae sp. 10 

Rinorea cf. arborea (dbh data only) 

Sloanea rhodantha form "quadriloba" 

Syzygium sp. 14 

The following species form a low-eleva- 
tion/moderate-disturbance group: 

Mammea sp. 3 

Psychotria sp. 2 

Rinorea cf. arborea (canopy data only) 

Sapindaceae sp. 1 

Tambourissa thouvenotii 

The following species are indicative of high 
disturbance: 

Albizia gummifera 

Apocynaceae sp. 1 

Apocynaceae sp. 2 

Breonia sp. 

Calophyllum paniculatum 

Canarium madagascariense 

Cryptocarya madagascariensis 

Cryptocarya sp. 2 

Cryptocarya sp. 3 

Ephippiandra sp. 1 

Gaertnera sp. 3 

Grewia sp. 1 

Labramia sp. 1 

Lauraceae sp. 3 

Lauraceae sp. 11 

Lauraceae sp. 12 

Lauraceae sp. 13 

Lauraceae sp. 16 

Mammea sp. 5 

Mammea sp. 6 

Memecylon cf. longipetalum 

Noronhia sp. 6 

Noronhia sp. 7 

Polyscias sp. 3 

Psychotria sp. 5 

Syzygium sp. 5 

Syzygium sp. 6 



46 



FIELDIANA: ZOOLOGY 






bOO 
400 










300- 










200- 


* 




* 




100- 


d 


J^ ^ 


_ c 




-100 


• • 


\ * * 






-200 


A 






"~ a 


-300 










-400- 
-KXV 




* b 







•400-300-200 



100 200 300 400 500 




-400-300-200 



Fig. 4-1 la (top). CANOCO ordination of samples 
for dbh data, related to aspect, slope, and position only, 
a, slope (437, -158); b, position (120, -451); c, S/N 
aspect (214, -68); d, E/W aspect (-93, 61). 

Fig. 4-1 lb (bottom). CANOCO ordination of spe- 
cies for dbh data, related to aspect, slope, and position 
only, a, slope (437, -158); b, position (120, -451); c, 
S/N aspect (214, -68); d, E/W aspect (-93, 61). 






Syzygium sp. 8 
Syzygium sp. 1 1 
Syzygium sp. 17 
Weinmannia sp. 5 

Two species fell outside of this cluster, Trica- 
lysia sp. 2 and Xylopia sp. 1. 

Disturbance and Altitude Omitted — CAN- 
OCO analysis was also conducted for both dbh 
and crown diameter data sets, excluding the two 
most significant environmental variables, distur- 
bance and altitude. The ordinations of plots and 
species generated by CANOCO for the dbh data 
are given in Figure 4-11. The ordination of plots 
(Fig. 4- 11 a) and species (Fig. 4-1 lb) both show 
no appreciable clustering. Similar results were ob- 
tained using the canopy data. Additional analyses 
were attempted omitting the environmental vari- 
ables aspect, slope angle, and topography in dif- 
ferent combinations for each data set. In each of 



these, there was no appreciable clustering of sam- 
ples or species. 



Discussion 

Forest Structure 

Numbers of Trees — Although there was no 
statistically significant difference in the density of 
trees between transect zones as measured by the 
numbers of stems per plot, there was a consider- 
able amount of plot-to-plot variation, ranging 
from 16 (transect zone 1) to 34 trees (transect 
zones 1 and 3) in a plot. This was probably a 
function of microhabitat variation. There was no 
correlation between numbers of trees and mean 
tree width (as measured by dbh) or mean tree 
height in the plot, even though such a correlation 
might have been expected if there was competi- 
tion for space. 

dbh Measurements and Basal Areas — Tran- 
sect zone 1 has the smallest number of large trees, 
and this, considered with the fact that it also has 
the lowest value for basal area (26.5 m 2 /ha), sug- 
gests that large trees have been selectively re- 
moved. For example, large Dalbergia trees were 
found inside the reserve in transect zone 2, but 
they were missing in transect zone 1. It is impor- 
tant to note that Dalbergia spp. yield valuable 
hardwood timber (Jenkins, 1987). 

It is perhaps surprising that the differences in 
dbh and basal area values between transects 1 and 
2 were not significant, even though the mean bas- 
al area in transect 2 was more than 60% greater 
than that in transect 1. This can be attributed to 
the high variation in the dbh measurements within 
the transect 2 plots. It is likely, with the addition 
of more or larger plots, that significant differences 
would be found. 

The basal area values for transect zones 2, 3, 
and 4 are similar (43.4, 49.1, and 43.2 m 2 /ha, re- 
spectively), indicating that there were no large 
differences in tree biomass between these three 
transects and that no change occurs within the 
range of elevations sampled. Considering the high 
proportion of small trees in transect zone 4 (65% 
in the smallest dbh size class) it may initially 
seem surprising that the differences in dbh and 
basal area were not statistically significant. It 
should be noted, however, that there were also 
more large trees (>50 cm dbh) in the plots in 
transect zone 4 than at any other zone, and that 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



47 



the numbers of trees in many of the intermediate 
dbh size classes were comparable to those at tran- 
sect zone 2. The low numbers of large trees and 
the small size of the plots sampled suggest that 
there is a large random component involved. A 
single large tree could introduce a large bias into 
the results. It is also possible that selective felling 
or other disturbance has taken place in one or 
more of the transect zones 2, 3, and 4. 

Tree Height — Tree height was significantly 
lower outside the reserve at 720 m (transect 1) 
than inside the reserve at 810 m (transect 2). This 
also suggests that the largest trees have been se- 
lectively removed from the forest outside the re- 
serve. 

Tree height is one of the physical characteristics 
thought to be most affected by increasing eleva- 
tion. Significant differences were found between 
transect zones 2, 3, and 4, with mean height de- 
creasing from 15 m at 720 m to 8 m at 1625 m. 
The heights of the tallest trees within each tran- 
sect zone show similar trends; trees over 22 m 
were only found in transect 2, where they consti- 
tuted 6% of the trees recorded. In transect 4, the 
tallest tree was 16 m in height. Within a particular 
area, tree height is strongly influenced by envi- 
ronmental factors such as degree of exposure and 
substrate characteristics. For example, tall trees 
were found where the land was flat, whereas the 
trees on ridge tops were always somewhat stunted 
compared to those in the surrounding vegetation, 
regardless of elevation. Exposed and unstable ar- 
eas are more susceptible to wind damage and land 
slips; these phenomena will reduce the average 
age of the vegetation on these areas. The differ- 
ences in tree heights at the different elevations 
correspond closely with the characteristic main 
canopy levels of lowland, montane, and sclero- 
phyllous forest types, respectively (White, 1983). 

Crown Diameter — The visual estimate of this 
parameter could be made rapidly, but it was not 
precise. No decrease in crown diameter was de- 
tected with increasing elevation. This may be due 
to two things. First, crown diameter is extremely 
difficult to estimate from the ground because it is 
often obscured by understorey trees (our results 
may simply reflect inaccuracies). Second, there 
was also considerable variation within transect 
zones that may have masked statistically signifi- 
cant differences. 

Numbers of Lianas — There were no significant 
differences in the numbers of lianas found in the 
different elevation zones. From our observations 



it would appear that the presence of lianas is more 
a function of microhabitat than elevation. 

Epiphytes — The forest in transect zone 4 (1625 
m) was particularly rich in epiphytic bryophytes 
and lichens, having a significantly greater quantity 
than the forests at lower elevations in both the tree 
crowns and on the tree trunks. This is a charac- 
teristic of sclerophyllous montane forest (White, 
1983) that occurs at elevations where the plants 
are almost continually wet and the air is humid. 
However, there was also forest rich in epiphytic 
bryophytes and lichens in transect zones 2 and 3 
in areas close to streams. The high humidity as- 
sociated with montane and cloud forest can also 
occur locally near streams at lower elevations. 
The effects of exposure related to elevation were 
in some cases similar to those at lower elevation 
due to aspect or site. For example, exposed low- 
altitude ridges often have flora and structure sim- 
ilar to that of high-altitude sheltered forest. 

Comparison with Other Forests — An assess- 
ment of structural differences between RNI 
d'Andringitra and other humid forests on the is- 
land can be done using a comparison of the den- 
sities and relative frequencies of different size 
classes of trees. The combined plots in each ele- 
vation zone in the RNI d'Andringitra total 0.1 ha, 
and by extrapolation there was an estimated range 
from 1,070 (at 810 m) to 1,400 (at 1210 m) stems 
per ha. This appears to be consistent with other 
Malagasy rain forests. At Vohiparara, PN de Ran- 
omafana (elevation 1100-1200 m), there was a 
total of 1,093 stems/ha (S. Malcomber, pers. 
comm.). Extrapolating from the results of the 0. 1 
ha surveys done by Gentry (1993), there were 
about 1,220 stems per ha at Peri net and about 
1,600 stems per ha at Nosy Mangabe. This con- 
trasts with much lower numbers in African moist 
forest, where an average of 590 trees per ha are 
found, and the Neotropics (moist and wet forest), 
which have an average of 640 trees per ha (Gen- 
try, 1993). 

Our results for basal area at 810, 1210, and 
1625 m are higher than the 35 m 2 /ha found at 
Vatoharanana (1000-1100 m) in the PN de Ran- 
omafana (Schatz & Malcomber, in press), al- 
though the tree size distribution is consistent be- 
tween the two sites. Gentry (1993) reports that 
basal area values in Africa average 34 m 2 /ha and 
35-36 m 2 /ha in the Neotropics. Gentry (1993) 
also noted that there was a comparatively high 
density of large trees (>1 m dbh) in Gabon, and 
such trees, even if they occur in low numbers, 
make up a large proportion of the basal area and 



48 



FIELDIANA: ZOOLOGY 



biomass of the forest. In the RNI d'Andringitra 
the maximum dbh of any tree measured in the 
plots was 81.1 cm. Thus, there were few very 
large trees in this forest relative to rain forests in 
mainland Africa. This was supported by casual 
observations outside the sample plots, and it may 
indicate a faster turnover time, due perhaps to 
high winds or substrate instability caused by high 
rainfall and/or shallow soil, or a lack of species 
capable of reaching very large proportions in the 
RNI d'Andringitra. 

No lianas with stems greater than 10 cm dbh 
were found in the RNI d'Andringitra. In the PN 
de Ranomafana, only two lianas with stems great- 
er than 10 cm dbh were encountered in the per- 
manent 1 ha plot (Schatz & Malcomber, in press). 
This compares with an average of 9.25 lianas of 
greater than 10 cm dbh in Gabonese plots (Reits- 
ma, 1988). Our results confirm the observation of 
Schatz & Malcomber (in press) that Malagasy for- 
est lacks the large lianas of some African rain 
forests. Gentry (1993) also noted unusually high 
densities of small dbh (^2.5 cm) lianas elsewhere 
in Malagasy forests (P6rinet and Nosy Mangabe). 

Structural differences between forests on dif- 
ferent continents are only presently being studied 
and analyzed, but they may have a profound im- 
pact on other organisms that occur in these habi- 
tats. For example, differences in liana size and 
density may have been a critical factor selecting 
for a variety of locomotor adaptations among can- 
opy vertebrates on different continents (Gentry, 
1988). 



Diversity 

There was an overall reduction in numbers of 
tree species and there were fewer families in the 
forest outside the reserve (transect zone 1) in 
comparison to the forest within the reserve at 
nearly the same elevation (transect zone 2). How- 
ever, the high degree of variation between plots 
in each transect zone suggests that different com- 
munities exist with different levels of diversity at 
this elevation. This finding has important conser- 
vation implications, and we recommend that fol- 
low-up work be done to confirm the validity of 
this observation. 

The average number of species over 10 cm dbh 
in a 0.1 ha sample of primary forest at RNI 
d'Andringitra (transects 2-4) was 74 (Table 4-7). 
This compares with 76 at P6rinet and 80 at Nosy 
Mangabe (Gentry, 1993). Gentry (1993) also not- 



ed that African moist forests have an average of 
33 species of trees of more than or equal to 10 
cm dbh in a 0. 1 ha sample; in African dry forests 
there is an average of 21 species, and Malagasy 
dry forest (Ankarafantsika) has 28 species. This 
underlines the high diversity of Malagasy forests 
in comparison with those in Africa. 

Between 720 and 1625 m there was no appre- 
ciable reduction in floristic diversity with increas- 
ing elevation. In a study on Andean forests, Gen- 
try (1988) obtained a similar result for forests up 
to 1700 m; a clear decrease in diversity occurred 
above this elevation. 



Floristics 

In the forest at transect zones 1 and 2, Elaeo- 
carpaceae (Sloanea rhodantha var. rhodantha 
form "quadriloba") and Burseraceae (Canarium 
madagascariense ssp. obtusifolium) formed large 
forest emergents, and members of the Lauraceae, 
Myrtaceae, Violaceae (Rinorea), Monimiaceae 
(Tambourissa, Ephippiandra, and Decaryden- 
drori) were the most common canopy and under- 
storey trees. 

With increasing elevation there was a change 
in floristic composition. For example, Canarium 
reached its upper limit at approximately 1000 m; 
similarly, Sloanea rhodantha var. rhodantha 
form "quadriloba" was replaced by S. rhodantha 
var. rhodantha form "quercifolia" between 810 
and 1210 m. This form of Sloanea rhodantha is 
readily distinguished from the form "quadrilo- 
ba" by its smaller leaves with serrate margins 
and acute acuminate apices, whereas the latter 
has larger, entire leaves with obtuse to emargin- 
ate non-acuminate leaves. Tirel (1985) reported 
the widespread occurrence in the eastern Mala- 
gasy region of a boundary at about 1200 m be- 
tween different forms of S. rhodantha var. rho- 
dantha. Although these forms appear to be suf- 
ficiently distinct to merit specific rank, Tirel not- 
ed the existence of intermediates in some areas 
and has discussed the complex variation in the 
species (1985). Pending a more detailed taxo- 
nomic investigation, it is advisable to provision- 
ally accept Tirel's conclusions. 

Between 810 and 1210 m there were also 
some changes in family composition, with Clu- 
siaceae becoming more prominent above 1000 
m; in particular, the genera Symphonia and Gar- 
cinia beame more abundant at higher elevations. 
These findings are consistent with those of Gen- 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



49 



try (1988), who speculated that 1000 m marks a 
transition from low- to mid-elevation forest in 
Madagascar. 

In their study of a 1 ha plot at Vatoharanana, 
PN de Ranomafana (elevation 1000-1100 m), 
based on numbers of stems, Schatz and Malcom- 
ber (in press) found the Monimiaceae to be the 
most abundant family; together with the Laura- 
ceae it accounted for 25% of the stems. Other 
prominent families at Vatoharanana include the 
Myrtaceae, Elaeocarpaceae, Sterculiaceae, and 
Clusiaceae. In comparison, the floristic composi- 
tion of the forests on the eastern slopes of the RNI 
d'Andringitra at 800-1200 m (transect zones 2 
and 3) differs in having fewer Monimiaceae but 
more Euphorbiaceae; otherwise it is similar, with 
the Lauraceae well represented. Schatz and Mal- 
comber (in press) also found that over half of the 
largest trees (>50 cm dbh) belonged either to 
Elaeocarpaceae (Sloanea) or Lauraceae (Crypto- 
carya and Ocotea). This agrees with our results 
from the 810 m zone. 

Between 1210 and 1625 m another shift in flo- 
ristic composition was detected, perhaps marking 
a change from mid- to high-elevation forest. This 
higher elevation forest (surrounding transect zone 
4) is characterized by Podocarpus, Weinmannia, 
Pandanus, Cyathea, and Asteraceae. The lowest 
elevational record of Vaccinium (Ericaceae) was 
in this zone. 

According to White (1983), the following flo- 
ristic characteristics help to differentiate lowland, 
montane, and sclerophyllous forests: (1) Rinorea 
and Canarium are common in lowland forest. (2) 
Melastomataceae are poorly represented in scle- 
rophyllous forest. (3) Araliaceae, Asteraceae, Er- 
ythroxylaceae, Verbenaceae, and Podocarpus are 
usually more common at higher altitudes. (4) Clu- 
siaceae, Fabaceae, Myrtaceae, Rubiaceae, and 
Sapindaceae are generally rarer in sclerophyllous 
forest. (5) Ephippiandra (Monimiaceae) is en- 
demic to moist montane and sclerophyllous for- 
ests. 

In general, forests within the three elevational 
zones sampled in the RNI d'Andringitra, corre- 
spond to the three forest types in White's (1983) 
classification. However, it should be noted that 
transects 1 and 2 lie near the upper elevational 
limit of the lowland forest and transect 4 lies near 
the lower elevational limit of the sclerophyllous 
forest. Additionally, we propose the following 
modifications: (1) Ephippiandra is not restricted 
to Malagasy high elevation forest, because it was 
found at 800 m (transect zone 2) and has been 



recorded at 300 m on the Masoala Peninsula (G. 
Schatz, pers. comm.). (2) Melastomataceae, which 
are underrepresented in the plot data because the 
majority of species do not produce woody stems 
of more than 10 cm diameter, were as well rep- 
resented in transect zone 4 as elsewhere. (3) There 
was no appreciable decrease in the quantity of 
Myrtaceae and Rubiaceae in transect zone 4. 

The following species were recorded at plots in 
all three altitudinal bands; they are therefore eco- 
logical transgressors in the forests of the eastern 
slopes of the RNI d'Andringitra: 

Calophyllum drouhardii 
Ilex mitis 
Lauraceae sp. 2 
Schefflera myriantha 
Sloanea rhodantha 

As in other Malagasy rain forests, few Faba- 
ceae occur in the forests of the RNI d'Andringitra. 
In the sample plots, the family comprised only 
0.6% of the stems and 0.86% of the total basal 
area. This is in marked contrast to African and 
Neotropical moist forests, and Malagasy dry for- 
ests, where Fabaceae are virtually always the 
dominant family (White, 1983; Gentry, 1988). In 
southeast Asian rain forests, Dipterocarpaceae are 
the dominant family. In Malagasy rain forest, the 
situation regarding dominance is not so clear, due 
to floristic variability along elevational gradients. 
However, taken as a whole, Lauraceae is the dom- 
inant family, both in terms of numbers of stems 
and total basal area. In the forests of the eastern 
slopes of the RNI d'Andringitra, 13.4% of the 
stems belonged to species of Lauraceae, and these 
comprised 22.29% of the total basal area. The 
eastern slopes of the RNI d'Andringitra are poor 
in palm species. 

Environmental Influences 

Elevation — All three methods of multivar 
iate analysis showed a clear correlation betweei 
the plot data and plot elevation. Although th< 
plots at 720 and 810 m, both of which include( 
both disturbed and undisturbed forest, exhibited 
a wide range of variation in composition, the? 
were in general quite distinct from those a 
higher altitudes. Plots at 1210 and 1625 m wer< 
strongly clustered, indicating relatively unifom 
composition in both zones, particularly at 162 
m. 



50 



FIELDIANA: ZOOLOG' 



These results confirm elevation as the most 
important natural environmental variable in the 
forests of the study area. The strong clustering 
of the data is not surprising, however, because 
plot elevation was the primary basis for sample 
stratification. If plots were to be sampled at in- 
termediate elevations, the discrete clusters of 
points obtained for each transect zone would 
probably be linked. 

Disturbance — Disturbance was revealed by 
CANOCO to be the most significant environmen- 
tal variable influencing the composition of the for- 
ests. Clearly, disturbance is a complex phenome- 
non; its nature and history are both critical factors 
in correctly interpreting its consequences. Reduc- 
ing disturbance to a single binary character — 
"disturbed" or "not disturbed" — based on sub- 
jective visual assessment is a gross oversimplifi- 
cation. The results of TWINSPAN and DECOR- 
ANA confirm that the plots at 720 and 810 m do 
not clearly fall into two clear categories based on 
their perceived disturbance, and in some cases 
clustering of points between the two zones is 
stronger than within each zone. Three factors 
should be considered in this regard: (1) the extent 
and nature of the disturbance may vary; (2) plots 
at 810 m did show some signs of disturbance; and 
(3) the analyses may have detected different plant 
communities that exist at that altitude and have 
remained somewhat similar in composition, de- 
spite disturbance to some. The importance of each 
of these factors could only be determined by more 
detailed study. 

Other Environmental Variables — The envi- 
ronmental variables aspect, slope angle, and to- 
pography appear to be much less influential to the 
vegetation than elevation and disturbance. Sample 
plots that are similar with respect to these vari- 
ables do not appear to be particularly similar bo- 
tanically (compare plots 5, 12, 16, 18, and 20). A 
more structured approach to sampling within each 
elevational band with respect to selection of plots 
with similar aspect, slope, and topography would 
permit a better assessment of the influence of 
these variables. It must be presumed that aspect, 
slope, and topography may be partly responsible 
for the differences between plots within each tran- 
sect zone, but more plots would need to be sam- 
pled in each transect to test this adequately. Other 
unmeasured environmental variables, such as mi- 
croclimate and soil characteristics, may be partly 
responsible for differences within and between the 
transect zones. 



Plant Communities 

The effects of increasing elevation on local 
microclimate are largely responsible for the dif- 
ferent vegetation types found on a mountain. 
With increasing elevation, mean temperatures 
decrease, and humidity, mist, and cloud cover in- 
crease; the latter, in turn, reduce the amount of 
light reaching the vegetation. One of the aims of 
multivariate analysis in plant ecology is to rec- 
ognize distinct associations and communities of 
species and the hierarchical relationship between 
them, based on the similarity between sample 
plots. A second aim is to assess the correlation 
of measured environmental variables with these 
communities. 

At present no standard systematic classification 
of the vegetation of Madagascar at the community 
level is available, in large part owing to the pau- 
city of basic research for most parts of the island. 
In the present study the analyses have clearly re- 
vealed clusters of similar plots among the 20 plots 
sampled, and the characteristics of each cluster 
may represent specific plant communities. The 
sample plots tended to cluster most closely with 
other plots in the same elevational band, suggest- 
ing that different plant communities occur in each 
elevational zone. The present analysis suggests 
that these communities are quite distinct in nature, 
but they may simply represent sets of samples 
along an elevational gradient, rather than truly 
distinct communities. 

Plots at 720 and 810 m have produced some- 
what confusing results; the effects of disturbance 
have distorted the pattern of primary vegetation 
communities. The results of the multivariate anal- 
yses of associations of plant species on the eastern 
slopes of the RNI d'Andringitra can be summa- 
rized as follows. 

( 1 ) Rinorea cf. arboreal Sloanea rhodantha var. 
rhodantha form "quadriloba" forest (plots 3, 4, 
5, 6, 9, and 10; 750-820 m). This forest is char- 
acterized by the dominance of Rinorea cf. arbo- 
rea. It is the most important species in terms of 
numbers of individuals (12 out of 28 trees in plot 
4), and it made the highest total contribution to 
the plot with respect to crown diameter and dbh. 
Individual trees of this species were never large, 
with crown diameter rarely exceeding 3 m and 
dbh less than 20 cm. Sloanea rhodantha form 
"quadriloba" is associated with Rinorea cf. ar- 
borea, although it was absent from plots 9 and 
10. In all plots, Sloanea was less important in 
terms of number of individuals and total crown 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



51 



and dbh measurements than Rinorea, except in 
plot 6, where it waas codominant. In plot 9, Wein- 
mannia sp. 6 is codominant with the Rinorea; 
elsewhere this species is only found in plot 8. In 
plot 5, Syzygium sp. 6 was numerous and was the 
second most important species; it was not record- 
ed in any other plots at any altitude. This forest 
type occurred in a range of situations, and this 
environmental diversity may account for some of 
the differences recorded between each of the sam- 
ple plots. The results of the CANOCO analysis 
suggest that this community corresponds to some 
degree of disturbance. In plots within the same 
elevational band that the analysis reveals as un- 
disturbed, Rinorea cf. arborea is absent (plot 7), 
infrequent (plot 8), or only codominant (plot 9). 
It may be a species that proliferates after mild 
disturbance. 

(2) Lauraceae sp. 16 forest (plot 1; 715 m). The 
vegetation at plot 1 was overwhelmingly domi- 
nated by this species, which was not recorded 
elsewhere. In plot 1 there was Rinorea/Sloanea, 
the widespread Xylopia sp. 1, and three other spe- 
cies unique to the plot. Species diversity was the 
lowest of all the plots, with only eight species 
recorded; this result may be the result of distur- 
bance. The CANOCO analysis supports this con- 
clusion, although plot 1 contained some relatively 
large individual trees. The Lauraceae sp. 16 forest 
occurred on a southwest-facing 35° slope at the 
bottom of a valley, an environmental setting sim- 
ilar to that of plot 2. 

(3) Lauraceae sp. 10 forest (plot 2; 740 m). The 
vegetation at plot 2 was dominated by this spe- 
cies, with Tambourissa thouvenotii and Syzygium 
sp. 5 of secondary importance. These and some 
of the other species also occur in Rinorea cf. ar- 
borealSloanea forest, although six species were 
unique to plot 2. The dominant Lauraceae sp. 10 
was found only in plots 2 and 7, both valley bot- 
tom plots. 

(4) Sloanea rhodantha var. rhodantha form 
"quadriloba"/Lauraceae sp. 10 forest (plot 7; 810 
m). The vegetation of plot 7 was dominated by 
Sloanea rhodantha form "quadriloba" and Lau- 
raceae sp. 10. However, few other species of ei- 
ther the Rinorea/Sloanea or the Lauraceae sp. 10 
were recorded. Calophyllum drouhardii was of 
secondary importance; it occurred mainly in for- 
ests at higher elevations. Three other species 
found at higher altitudes, Ilex mitis, Schefflera my- 
riantha, and Sapindaceae sp. 4, were also present, 
but 10 of the species recorded at plot 7 did not 
occur elsewhere. Plot 7 was valley bottom forest 



on level ground, and although its environmental 
setting was similar to that of plot 10, no species 
were common to both plots. 

(5) Syzygium sp. 2 forest (plot 8; 860 m). The 
vegetation in plot 8 was dominated by Syzygium 
sp. 2, although Dalbergia sp. 1, Syzygium sp. 3, 
and Calophyllum drouhardii were also important. 
Syzygium sp. 2 and C. drouhardii are important 
components of the forest in transect zone 3, 
whereas Dalbergia sp. 1 and Syzygium sp. 3 were 
also found only in Rinorea/Sloanea forest in plot 
6. Four of the species at plot 8 were not recorded 
elsewhere. The Syzygium sp. 2 forest was on a 
southwest-facing ridge forest; in this exposed sit- 
uation it would be expected that some species of 
higher altitudes might be present. Although plot 3 
is in a similar situation, plots 3 and 8 only had 
two species in common, Sapindaceae sp. 1 and 
Syzygium sp. 14. 

(6) Cleistanthus boivinianus forest (plots 11, 
12, 13, 14, and 15; 1220-1310 m). The plots at 
1210 m were characterized by the presence of 
Cleistanthus boivinianus in all five plots, Mallotus 
cf. capuronii in four of the five plots, and Lau- 
raceae sp. 1 in three of the five plots. These spe- 
cies were absent from plots at all other elevations. 
Cleistanthus boivinianus was dominant only in 
plot 14; it was codominant with Ilex mitis in plot 
15, whereas the other two species were never 
present in more than moderate quantity. In plots 
11 and 12, Garcinia sp. 1 was dominant, com- 
prising 18 of the 52 trees recorded in the two 
plots. In plot 13, Calophyllum drouhardii was 
dominant, but the Cleistanthus was also impor- 
tant. Calophyllum drouhardii appears to be a 
common species with a wide environmental tol- 
erance, occurring in forest at each altitude of dif- 
ferent aspect, slope, and topography, and also in 
disturbed forest. 

(7) Weinmannia sp. 1/Ocotea sp. 2/Podocarpus 
madagascariensis forest (plots 16, 17, 18, 19, and 
20; 1550-1660 m). The forest around 1625 m ap- 
pears to be a mosaic of species. Only Pandanus 
sp. 2 occurred in all five plots, but it was never 
prevalent. The following species are also impor- 
tant: Weinmannia sp. 7 (four plots), Ocotea sp. 2 
(three plots), and Podocarpus madagascariensis 
(three plots). Each of these species is dominant or 
codominant in each of the plots except plot 16, 
where the Weinmannia occurs, but is much less 
important than certain other species, notably Ma- 
caranga echinocarpa and Asteraceae sp. 1. Al- 
though this plot has seven species that were not 
recorded elsewhere, the analyses suggest that it 



52 



FIELDIANA: ZOOLOGY 



should not be separated from the other plots at 
1625 m. 



Methodological Improvements 



On the basis of the results obtained from the 
eastern slopes of the RNI d'Andringitra, several 
recommendations can be made to improve the 
methodology for botanical rapid assessment stud- 
ies. The high degree of variation between plots in 
each transect zone, and the high proportion of pre- 
viously unrecorded species found in the final plot 
sampled in each zone, indicate that plot size and/ 
or the number of plots sampled were inadequate. 
The minimum plot size and sample number that 
will provide an acceptable result for a particular 
type of vegetation depends on the structural and 
floristic variability of the vegetation and the type 
of study that is being undertaken. As discussed 
in the introduction, it was unreasonable to aim 
for complete or near complete sampling of plant 
communities as diverse and complex as a Mal- 
agasy rain forest during a rapid assessment. 
Rather, a relative measure of diversity and de- 
scription of the main communities, at least for 
comparative purposes, can be obtained by sam- 
pling a proportion of the vegetation. For the pur- 
poses of rapid assessment, the standardization of 
methods is desirable, and it is recommended that 
the exercise of determining optimum plot size 
using the established method of sampling nested 
quadrats would improve the reliability of future 
attempts at rapid assessment of forest commu- 
nities in Madagascar. This approach should be 
coupled with a study to determine the minimum 
number of plots required to provide an adequate 
vegetation survey. 

The stratification of sampling on the basis of 
disturbance and elevation served to highlight the 
importance of these environmental variables in 
the forest of the Eastern Malagasy Region. How- 
ever, a more structured approach to the sampling 
of vegetation on different aspects, slopes, and to- 
pography (and other environmental variables) 
would have yielded more complete information 
about the influence of these variables. Thus, every 
I effort should be taken to sample an equivalent set 
of sites, within each elevational zone, subject to 
the constraints of the landscape. 



Summary 

The vegetation in the study area (720-1625 m) 
in the RNI d'Andringitra is a complex mixture of 
forest types and individual plant communities; 
this study confirms the exceptional diversity of 
forest in the Eastern Malagasy Region. Structural 
and floristic differences were found in forests out- 
side (720 m) and inside the reserve (810 m). 
These differences are probably due to disturbance, 
although the results suggest that disturbance of 
the forest has occurred within the reserve bound- 
aries, at least within some plots. The forest out- 
side the reserve had fewer large trees, indicating 
that the largest individuals had been selectively 
removed, and a lower floristic diversity than plots 
within the reserve. 

In the three elevation bands sampled (810, 
1210, and 1625 m), the forest had distinctive 
structural and floristic features. With increasing 
elevation the canopy height decreased and the 
amount of hanging moss and lichen increased. 
These were the only structural parameters mea- 
sured that were significantly correlated with in- 
creasing elevation. There was no indication that 
overall floristic diversity decreased with elevation 
within the range sampled, although there were 
marked changes in the floristic composition with 
increasing elevation. The forest, at ±800, ±1200, 
and ±1600 m, corresponded closely with White's 
(1983) lowland, montane, and sclerophyllous 
montane forest types, respectively. Although un- 
doubtedly influencing the vegetation at each sam- 
ple plot, the slope, aspect, and topography showed 
no correlation with the composition of the forest 
across the elevational gradient, based on crown 
and dbh measurements. 

Our results on the floristic diversity of the for- 
ests on eastern slopes of the RNI d'Andringitra 
agree with those for certain other rain forest ar- 
eas in Madagascar in demonstrating extremely 
high species richness compared with similar for- 
est types in Africa. It is hoped that this study 
will stimulate further work on the diversity and 
structure of Malagasy rain forests, and that it will 
contribute toward the development of a system- 
atic classification of the vegetation at a commu- 
nity level. We also hope that this study will add 
information to the larger picture of the structural 
differences and similarities between forests on 
different continents. Additionally, we hope to 
stimulate discussion as to the best methodology 
for rapid assessment studies of rain forest areas. 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



53 



Acknowledgments 

We thank DEF and ANGAP for permission to 
work in the RNI d'Andringitra, and we thank 
WWF for providing logistical support. We are 
grateful to Albert Randrianjafy (Director, PBZT) 
and Philippe Morat (Director, Museum National 
D'Histoire Naturelle, Paris) for providing research 
facilities. 

We also thank Jeanine Raharilala (PBZT) for 
her work in the field and help with field identifi- 
cations. We are grateful to Tony Palmer (Range 
and Forest Institute) and Ted Avis and Rachel 
Judd (Rhodes University) for their advice and as- 
sistance with the multivariate analyses. We are 
particularly indebted to Steven Goodman for ex- 
tensive comments on earlier drafts, and to Pete 
Lowry, George Schatz, Simon Malcomber, Mich- 
elle Zjhra and three anonymous reviewers for 
their comments and suggestions on the manu- 
script. The work was supported through a grant 
from KfW, the Claiborne/Ortenberg Foundation, 
and the John D. and Catherine T MacArthur 
Foundation. 



Literature Cited 

Barbour, M. G., J. H. Burk, and W. D. Pitts. 1980. 
Terrestrial plant ecology. Benjamin/Cummings Pub- 
lishing Company, Menlo Park, California. 

Brummitt, R. K. 1992. Vascular plant families and gen- 
era. Royal Botanic Gardens, Kew. 

Du Puy, B., J. P. Abraham, and A. J. Cooke. 1994. 
Les plantes, pp. 15-29. In Goodman, S. M., and O. 
Langrand, eds., Inventaire Biologique Foret de Zom- 
bitse. Recherches pour le Developpement, Serie Sci- 
ences Biologiques no. Special. Centre d' Information 
et de Documentation Scientifique et Technique, An- 
tananarivo. 

Gauch, H. G. 1982. Multivariate analysis in community 
ecology. Cambridge University Press, Cambridge, 
U.K., 298 pp. 

Gentry, A. H. 1982. Patterns of neotropical plant spe- 
cies diversity. Evolutionary Biology, 15: 1-84. 

. 1988. Changes in plant community diversity 

and floristic composition on environmental and geo- 
graphical gradients. Annals of the Missouri Botanical 
Garden, 75: 1-34. 

. 1993. Diversity and floristic composition of 

lowland tropical forest in Africa and South America, 
pp. 500-547. In Goldblatt, P., ed., Biological relations 
between Africa and South America. Yale University 
Press, New Haven, Connecticut. 

Guillaumet, J.-L. 1983. Forets et fourres de montagne 

a Madagascar. Candollea, 38: 481-502. 
Hell, M. O. 1979a. TWINSPAN— A Fortran program 

for arranging multivariate data in an ordered two-way 



table by classification of the individuals and attributes. 
Cornell University, Ithaca, New York, 48 pp. 

1979b. DECORANA— A Fortran program for 



detrended correspondence analysis and reciprocal 
averaging. Cornell University, Ithaca, New York, 30 
pp. 

Humbert, H. 1955. Les territoires phytogeographiques 
de Madagascar. In Les divisions ecologiques du mon- 
de. Colloques Internationale du Centre National de 
Recherche Scientifique LIX. Annee biologique, sen 3, 
31(5-6): 439-448. 

Humbert, H., and G. Cours Darne. 1965. Notice de 
la carte Madagascar. Centre National de la Recherche, 
Scientifique et de l'Office de la Recherche Scienti- 
fique et Technique Outre-Mer, 164 pp. 

Jenkins, M. D., ed. 1987. Madagascar. An environmen- 
tal profile. International Union for the Conservation 
of Nature, Gland, 374 pp. 

Koechlin, J., J. L. Guillaumet, and P. Morat. 1974. 
Flore et v6getation de Madagascar. J. Cramer, Vaduz, 
Liechtenstein, 687 pp. 

Kress, W. J., G. E. Schatz, M. Andrianifahanana, 
and H. Simons Morland. 1994. Lemur pollination 
in Madagascar. American Journal of Botany, 8: 542- 
551. 

Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. Etude des ecosystemes mon- 
tagnards dans la region Malgache. I. — Le massif de 
l'Andringitra. 1970-1971. G6omorphologie, climato- 
logie et groupements vegetaux. Bulletin de la Societe 
d'Ecologie, 11(2-3): 198-226. 

Perrier De La Bathie, H. 1921. La vegetation mal- 
gache. Annales du Musee Colonial de Marseille, ser. 
3, 9: 1-268. 

. 1927. Le Tsaratanana, l'Ankaratra et l'Andrin- 



gitra. Memoires de l'Academie Malgache, 3: 1-71. 

Phillipson, P. B. 1994. Madagascar. In Davis, S. D., V 
H. Heywood, and A. C. Hamilton, eds. Centres of 
plant diversity: A guide and strategy for their conser- 
vation. Volume 1: Europe, Africa, South West Asia 
and the Middle East. IUCN Publications Unit, Cam- 
bridge, U.K. 

Reitsma, J. M. 1988. Vegetation forestiere du Gabon. 
Tropenbos Technical ser. 1: 1-142. 

Schatz, G. E., and S. T. Malcomber. In press. Floristic 
composition of one hectare plots in Ranomafana Na- 
tional Park. In Wright, P. C, ed., Biodiversity in Ran- 
omafana National Park, Madagascar. State University 
of New York Press, Albany. 

Ter Braak, C. J. F 1986. Canonical correspondence 
analysis: A new eigenvector technique for multivariate 
direct gradient analysis. Ecology, 67: 1167-1179. 

Tirel, C. 1985. Elaeocarpaceae. Flore de Madagascai 
et des Comores. 125. Museum National d'Histoire Na- 
turelle, Paris, 53 pp. 

Watt, T. A. 1993. Introductory statistics for biolog) 
students. Chapman & Hall, London, 185 pp. 

White, F 1983. The vegetation of Africa. UNESCO 
Paris, 356 pp. 

Whittaker, R. H. 1977. Evolution of species diversity 
in land communities. Evolutionary Biology, 10: 1-87 



54 



FIELDIANA: ZOOLOGY 



Appendix 4-1. 

Checklist of Vascular Plant Species 
Recorded from Andringitra 

This checklist is arranged in three parts: 
Ferns and Fern Allies, Gymnosperms, and An- 
giosperms, each ordered alphabetically by plant 
family (following Brummitt, 1992, with irreg- 
ular names corrected). Localities given in pa- 
rentheses refer to information accompanying 
the relevant taxonomic descriptions of taxa. In- 
formation obtained from Perrier de la Bathie 
(1927) on endemics recorded from Andringitra 
is cited, and species mentioned by Paulian et al. 
(1971) are cited with relevant page numbers. 
Nomenclature is brought up to date as far as 
possible. Lewis specimens with numbers in pa- 
rentheses preceded by "s" are sterile voucher 
specimens. Voucher specimens are generally 
lodged at TAN and MO. Morphospecies without 
generic allocation are placed immediately be- 
low each family name. This list was generated 
with the help of the Conspectus database pro- 
ject, coordinated by George Schatz, Missouri 
Botanical Garden, St. Louis, Missouri. Several 
specimens are cited from the Reserve Naturelle 
(RN) series. For the sake of conciseness the 
1993 collections are designated by camp num- 
ber. The precise localities for each transect zone 
are: 

Camp 1: Approximately 45 km S Ambalavao, 
east bank Iantara River, along the Ambala- 
manenjana-Ambatamboay trail, 22°13'20"S, 
47°0r29"E, 720 m. 

Camp 2: Approximately 43 km S Ambalavao, 
junction of the Sahanivoraky and Sahavatoy 
rivers, 22°13'40"S, 47°00'13"E, 810 m. 

Camp 3: Approximately 40 km S Ambalavao 
along tributary of Sahavatoy River, 
22°13'22" S, 46°58'18"E, 1210 m. 

Camp 4: Approximately 38 km S Ambalavao, 
on ridge above headwaters of Sahavatoy Riv- 
er, 22°H'39"S, 46°58'16"E, 1625 m. 



Part 1: Ferns and Fern Allies 
Adiantaceae 



Pityrogramma humbertii C. Chr. — "Rochers de la 
zone sommitale 2300-2650 m" (Paulian et al., 
1971, p. 248). 

Pteris cretica L. — Approximately 5 km SE An- 
tanifotsy, the reserve headquarters, 1500 m: 
Schatz 2662, with P. Goldblatt, A. Rakotozafy, 
and J. Randrianasolo. 



Aspleniaceae 

Asplenium marojejyense Tardieu — 50 km S Am- 
balavao, near abandoned meteo station above 
Ambalamarina, high altitude moss/lichen rain 
forest-ericaceous bush, Philippia dominant, 
1975 m: Nicoll 257. 

Asplenium poolii Baker — 50 km S Ambalavao, 
near abandoned meteo station above Ambala- 
marina, high altitude moss/lichen rain forest- 
ericaceous bush, Philippia dominant, 1975 m: 
Nicoll 258. 

Asplenium preussii Hieron. ex Brause — Approxi- 
mately 5 km SE Antanifotsy, the reserve head- 
quarters, 1500 m: Schatz 2661, with P. Gold- 
blatt, A. Rakotozafy, and J. Randrianasolo. 

Asplenium rutifolium (Berg.) Kunze — 50 km S 
Ambalavao, near abandoned meteo station 
above Ambalamarina, high altitude moss/lichen 
rain forest-ericaceous bush, Philippia domi- 
nant, 1975 m: Nicoll 254. 

Asplenium sandersonii Hook. — Approximately 5 
km SE Antanifotsy, the reserve headquarters, 
1500 m: Schatz 2657, with P. Goldblatt, A. 
Rakotozafy, and J. Randrianasolo. 



Cyatheaceae 

Cyathea dregei Kuntze — "V6g6tation liee a la 
presence d'eau 2000-2300 m" (Paulian et al., 
1971, p. 243). 

Cyathea sp. 1 — Camp 4: Lewis (s 407). 

Cyathea sp. 2 — Camp 2: Lewis (s 127). 

Cyathea sp. 3 — Camp 2: Lewis (s 184). 



Grammitidaceae 



Cheilanthes horizontalipinnata. Bonap. — Rocail- 
les pres de la cime (Perrier de la Bathie, 1927). 



Xiphopteris hildebrandtii (Hieron.) Tardieu- 
Camp 3: Lewis 971. 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSfTION, AND DIVERSITY 



55 



Hymenophyllaceae 

Hymenophyllum compactum Bonap. — Foret a 
mousses, vers 2000 m, versant Est (Perrier de 
la Bathie, 1927). 

Hymenophyllum sp. 1 — Foret a mousses, vers 
2000 m, versant Est (Perrier de la Bathie, 
1927). 

Trichomanes angustilaciniatum Bonap. — Rocail- 
les vers 2000 m versant Est (Perrier de la Bath- 
ie, 1927). 



specified) — "Foret a strate herbacee [below] 

1970 m" (Paulian et al., 1971, p. 254). 
Podocarpus madagascariensis Baker var. mada- 

gascariensis — Camp 3: Lewis 941. 
Podocarpus madagascariensis Baker var. rotun- 

dus Laurent — Camp 4: Lewis (s 432). 



Part 3: Angiosperms 
Acanthaceae 



Lomariopsidaceae 

Elaphoglossum achroalepis (Baker) C. Chr. — 
Epiphytic on riverine trees, Camp 2: Lewis 877. 



Lycopodiaceae 

Lycopodium cernuum L. — 50 km S Ambalavao, 
near abandoned meteo station above Ambala- 
marina, high altitude moss/lichen rain forest- 
ericaceous bush, Philippia dominant, 1975 m: 
Nicoll 264. 

Lycopodium clavatum L. — "Hauts fourre arbus- 
tifs 2000-2300 m" (Paulian et al., 1971, p. 
241); "Foret dense sclerophylle de montagne 
2000 m" (Paulian et al., 1971, p. 250). 



Polypodiaceae 

Loxogramme lanceolata (Sw.) C. Presl — 50 km S 
Ambalavao, near abandoned meteo station 
above Ambalamarina, high altitude moss/lichen 
rain forest-ericaceous bush, Philippia domi- 
nant, 1975 m: Nicoll 256; approximately 5 km 
SE Antanifotsy, the reserve headquarters, 1500 
m: Schatz 2660, with P. Goldblatt, A. Rakoto- 
zafy, and J. Randrianasolo. 

Pleopeltis excavata (Bory ex Willd.) Sledge — 50 
km S Ambalavao, near abandoned meteo sta- 
tion above Ambalamarina, high altitude 
moss/lichen rain forest-ericaceous bush, Phi- 
lippia dominant, 1975 m: Nicoll 260. 



Part 2: Gymnosperms 
Podocarpaceae 

Podocarpus madagascariensis Baker (var. not 



Mendoncia flagellaris Benoist — Camp 1: Lewis 

759. 
Mendoncia sp. 1 — Camp 3: Lewis 944. 



Aloaceae 

Aloe andringitrensis H. Perrier (Madagascar: tres 
commune, aussi bien sur le versant Ouest, que 
sur le versant Est); Rocailles, de 2000 mala 
cime (Perrier de la Bathie, 1927); "Vegetation 
rupicole 2000-2300 m" (Paulian et al., 1971, 
p. 242). 

Aloe capitata Baker "Vegetation rupicole 2000- 
2300 m" (Paulian et al., 1971, p. 242). 

Aloe decorseii H. Perrier (Madagascar: massif 
d'Andringitra) — Rocailles, vers 1700 m, ver- 
sant Sud-Ouest Est (Perrier de la Bathie, 1927). 

Aloe haworthioides Baker var. aurantiaca H. Per- 
rier — Rocailles, vers 2000 m, versant Est (Per- 
rier de la Bathie, 1927). 

Aloe macroclada Baker "Vegetation rupicole 
2000-2300 m" (Paulian et al., 1971, p. 242). 



Anacardiaceae 

Micronychia macrophylla H. Perrier — Camp 1: 
Lewis 803. 

Micronychia madagascariensis Oliv. — Camp 3: 
Lewis (s 311), 961. 

Micronychia tsiramiramy H. Perrier — Camp 3: 
Lewis 778, 976. 

Protorhus sp. 1 — Camp 2: Lewis (s 138, 140) 
881, 912. 

Rhus taratana (Baker) H. Perrier — Camp 1 : Lew- 
is 805. 

Rhus thouarsii (Engl.) H. Perrier — Camp 1 : Lewi; 
778; Camp 3: Lewis 976. 



56 



FIELDIANA: ZOOLOGY 



Annonaceae 

Artabotrys mabifolius Diels — Camp 2: Lewis 844. 

Polyalthia humbertii Cavaco & Keraudren — 
Camp 3: Lewis (s 341, 343). 

Xylopia sp. 1 — Camp 1: Lewis (s 7); Camp 2: 
Lewis (s 190); Camp 3: Lewis (s 259). 



Anthericaceae 

Arthropodium caesioides H. Perrier — "Prai- 
ries/prairies altimontaines 1700-2000 m" (Pau- 
lian et ah, 1971, pp. 236, 256); "Foret dense 
scleYophylle de montagne [above] 1950 m" 
(Paulian et ah, 1971, p. 256). 

Apiaceae 

sp. 1 — Camp 3: Lewis 967. 

Phellobium madagascariensis Baker "Prairies al- 
timontaines 2000-2300 m" (Paulian et ah, 
1971, p. 241), "V6g6tation liee a la presence 
d'eau 2000-2300 m" (Paulian et ah, 1971, p. 
243). 

Sanicula europaea L. "Foret dense de moyenne 
altitude 1500-1650 m" (Paulian et ah, 1971, p. 
250). 



Apocynaceae 

sp. 1 — Camp 1: Lewis (s 30). 

sp. 2 — Camp 1 : Lewis (s 27). 

Landolphia myrtifolia (Poir.) Markgraf — Camp 1 : 
Lewis 795. 

Mascarenhasia arborescens DC. — Camp 1: Lew- 
is 760; Camp 2: Lewis 873. 

Tabernaemontana ciliata Pichon — Camp 2: Lew- 
is 919. 



Aquifoliaceae 

Ilex mitis (L.) Radlk. — Camp 2: Lewis (s 149), 
893; Camp 3: Lewis (s 344, 352); Camp 4: 
Lewis (s 463). 



Araceae 

Pothos scandens L. — Camp 1: Lewis 777. 

Araliaceae 

Polyscias sp. 1 — Camp 4: Lewis (s 447, 464), 
1047. 

Polyscias sp. 2 — Camp 4: Lewis (s 387, 419, 
434), 843. 

Polyscias sp. 3 — Camp 1: Lewis (s 34), 835. 

Polyscias sp. 4 — Camp 3: Lewis (s 353), 965. 

Polyscias sp. 5 — Camp 4: Lewis (s 399, 427). 

Polyscias sp. 6 — Camp 2: Lewis 926; Camp 3: 
Lewis (s 336), 979. 

Polyscias sp. 7 — Camp 2: Lewis (s 180); Camp 
3: Lewis (s 295, 291). 

Polyscias sp. 8 — Camp 3: Lewis (s 229). 

Polyscias sp. 9 — Camp 4: Lewis (s 423). 

Schefflera longipedicellata (Lecomte) Bernardi — 
Approximately 5 km SE Antanifotsy, the re- 
serve headquarters, 1500 m: Schatz 2674, with 
P. Goldblatt, A. Rakotozafy, and J. Randriana- 
solo. 

Schefflera monophylla (Baker) Bernardi — Ap- 
proximately 5 km SE Antanifotsy, the reserve 
headquarters, 1500 m: Schatz 2677, with P. 
Goldblatt, A. Rakotozafy, and J. Randrianasolo. 

Schefflera myriantha (Baker) Drake — Camp 2: 
Lewis (s 139); Camp 3: Lewis (s 317); Camp 
4: Lewis (s 497), 1028. 

Schefflera sp. 1 — Camp 3: Lewis 981. 

Schefflera sp. 2— Camp 4: Lewis 1099. 

Arecaceae 

Dypsis linearis (Becc.) Beentje & J. Dransf. — 
Camp 1: Lewis 774; Camp 2: Lewis 845. 

Dypsis sp. 1 — Camp 2: Lewis 864. 

Ravenea glauca Jum. & H. Perrier — Bois des 
pentes occidentals, de 1600 a 2000 m (Perrier 
de la Bathie, 1927). 

Asclepiadaceae 

Asclepias fruticosa L. — S Ambalavao above vil- 
lage of Antanifotsy, partially disturbed, proba- 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



57 



bly burned areas dominated by grasses, 1400 

m: Lowry 4513. 
Ceropegia perrieri Choux — Foret a mousses, vers 

1600 metres, versant Est (Perrier de la Bathie, 

1927). 
Cynanchum andringitrense Choux — "Foret 

d'Ambohiby, alt. 1800 m", Leandri 761. 
Cynanchum pycnoneuroides Choux — Rocailles, 

1 800 m, versant Sud-Oeust (Perrier de la Bath- 
ie, 1927). 
Cynanchum lineare (Bello) Alain — Andringitra, 

1800 m: Liede 2863, with J. Conrad. 

Cynanchum obovatum Choux — Camp 2: Lewis 

896. 
Cynanchum papillatum Choux — Andringitra, 

2000 m: Liede 2862, with J. Conrad. 

Tylophora sylvatica Decne. — Camp 2: Lewis 909. 



Asteraceae 

sp. 1 — Camp 4: Lewis (s 384). 

sp. 2 — Open ericaceous vegetation, on E side 

Ampasipotsy, above Camp 4, 2000 m: Lewis 

1074. 

Apodocephala pauciflora Baker — Camp 2: Lewis 
886. 

Brachylaena ramiflora (DC.) Humbert — 50 km 
S Ambalavao, near abandoned meteo station 
above Ambalamarina, high altitude 
moss/lichen rain forest-ericaceous bush, Phi- 
lippia dominant, 1975 m: Nicoll 239; "Vege- 
tation rupicole 2000-2300 m" (Paulian et al., 
1971, p. 242). 

Conyza andringitrana Humbert — Bords de tor- 
rents, vers 1200 m, versant Est (Perrier de la 
Bathie, 1927). 

Helichrysum abeitifolium Humbert ssp. gracilifol- 
ium (Humbert) Humbert — Rocailles de torrents, 
vers 1600 m (Perrier de la Bathie, 1927). 

Helichrysum adhaerens (DC.) R. Vig. & Humbert 
var. subulatum Humbert (Centre: massif de 
1' Andringitra). 

Helichrysum aff. cryptomerioides Baker — "La 
cuvette de Boby 2500 m" (Paulian et al., 1971, 
p. 247). 

Helichrysum attenuatum Humbert — Rocailles, de 
1500 a 2500 m (Perrier de la Bathie, 1927). 

Helichrysum cf. baronii Humbert — Growing on 
rocks in open ericaceous vegetation, on E side 



Ampasipotsy, above Camp 4, 2000 m: Lewis 
1072. 

Helichrysum barorum Humbert (massif d' Andrin- 
gitra, au pied de l'escarpement a Tangle NE du 
massif a Test de la riviere Antsifo). 

Helichrysum bracteiferum (DC.) Humbert var. an- 
dringitranum Humbert — "Praires altimontaines 
2000-2300 m" (Paulian et al., 1971, p. 241). 

Helichrysum calocladum Humbert — Tourbieres, 
vers. 2000 m (Perrier de la Bathie, 1927); "Ve- 
getation liee a la presence d'eau 2000-2300 
m" (Paulian et al., 1971, p. 242). 

Helichrysum cryptomerioides Baker — Growing 
on rocks in open ericaceous vegetation, on E 
side Ampasipotsy, above Camp 4, 2000 m: 
Lewis 1067. 

Helichrysum danguyanum Humbert — Rocailles, 
de 2000 a 2500 m (Perrier de la Bathie, 1927); 
growing on rocks in open ericaceous vegeta- 
tion, on east side of Ampasipotsy, above Camp 
4, 2000 m: Lewis 1073. 

Helichrysum hypnoides (Bojer ex DC.) A. Ca- 
mus — "Praires altimontaines 2000-2300 m" 
(Paulian et al., 1971, p. 241). 

Helichrysum minutiflorum Humbert — Rocailles, 
vers 2500 m (Perrier de la Bathie, 1927). 

Helichrysum mirabile Humbert — Rocailles, de 
1600 a 2400 m (Perrier de la Bathie, 1927). 

Helichrysum retrorsum DC. — Growing on rocks 
in open ericaceous vegetation, on E side Am- 
pasipotsy, above Camp 4, 2000 m: Lewis 1066. 

Helichrysum stilpnocephalum Humbert — Rocail- 
les, de 2000 a 2500 m (Perrier de la Bathie, 
1927). 

Helichrysum tomentosum Humbert — Rocailles, de 
2200 a 2600 m (Perrier de la Bathie, 1927). 

Rochonia aspera Humbert — Rocailles, au-dessus 
de 2000 m (Perrier de la Bathie, 1927). 

Senecio andringitrensis Humbert — Rocailles, de 
1600 mala cime (Perrier de la Bathie, 1927). 

Senecio caniculatus Bojer — "Vegetation rupico- 
le" (Paulian et al., 1971, p. 257). 

Senecio denisii Humbert — Rocailles, de 2000 m a 
la cime (Perrier de la Bathie, 1927). 

Senecio emirnensis DC. var. angavonensis (Bojer 
ex DC.) Humbert — 50 km S Ambalavao, near 
abandoned meteo station above Ambalamari- 
na, high altitude moss/lichen rain forest-eri- 
caceous bush, Philippia dominant, 1975 m: 
Nicoll 247. 



58 



FIELDIANA: ZOOLOGY 



Senecia latibracteatus Humbert — Rocailles, au- 
dessus de 2400 m (Perrier de la Bathie, 1927). 

Senecio melastomaefolius Baker — "Vdgdtation 
rupicole" 1800 m" (Paulian et al., 1971, p. 
237). 

Senecio myricaefolius (Bojer ex DC.) Humbert — 
Camp 4: Lewis 1076. 

Senecio vaingaindrani Scott-Elliott — Camp 1: 

Lewis 818. 
Stoebe cryptophylla Baker — "Hauts fourre arbus- 

tifs 2000-2300 m" (Paulian et al., 1971, p. 

238). 
Stoebe pachyclada Humbert — Broussailles 6ricoi- 

des, de 2000 a 2400 m, versant Est (Perrier de 

la Bathie, 1927). 
"Hauts fourre' arbustifs 2000-2300 m" (Paulian 

et al., 1971, p. 238). 
Syncephalum arbutifolium (Baker) Humbert — 

Broussailles 6ricoides, entre 1600 et 2400 m 

(Perrier de la Bathie, 1927). 
Syncephalum candidum Humbert — Rocailles, vers 

2500 m (Perrier de la Bathie, 1927). 
Vernonia alleizettei Humbert var. rienanensis 

Humbert — Camp 2: Goodman 6402. 
Vernonia rubicundus Klatt. — Ericaceous vegeta- 
tion, on E side Ampasipotsy, above Camp 4, 

1800 m: Lewis 1061. 
Vernonia sp. 1 — Camp 4: Lewis 1048. 
Vernonia sp. 2 — Camp 3: Lewis 954. 



Balsaminaceae 

Impatiens sp. 1 — Camp 2: Lewis 855. 
Impatiens sp. 2 — Camp 1: Lewis 753; Camp 3: 

Lewis 937. 
Impatiens sp. 3 — Camp 4: Lewis 1112; ericaceous 

vegetation, on E side Ampasipotsy, above 

Camp 4, 1800 m: Lewis 1058. 



Bignoniaceae 

Ophiocolea floribunda H. Perrier — Camp 1: Lew- 
is 771; Camp 3: Lewis 990, 1015. 



Brassicaceae 

Nasturtium sp. 1 — Rocailles, de 2000 m a la cime 
(Perrier de la Bathie, 1927). 



Burseraceae 

Canarium madagascariense Engl. ssp. obtusifol- 
ium (Scott-Elliott) Leenhouts — Camp 1: Lewis 
(s 86); Camp 2: Lewis (s 134). 



Cactaceae 

Rhipsalis baccifera (Sol. ex J.M. Mill.) Stearn — 
"Foret a strate herbacee 1695-1840 m" (Pau- 
lian et al., 1971, p. 254); Camp 2: Lewis 914. 



Campanulaceae 

Lobelia agrestis E. Wimm. — Camp 1 : Lewis 788. 

Cannellaceae 

Cinnamosma fragrans Baill. sp. 1 — Camp 4: 
Lewis (s 450, 496), 1077. 

Clusiaceae 

Calophyllum drouhardii H. Perrier — Camp 2: 
Lewis (s 146, 179); Camp 3: Lewis (s 284, 289, 
296, 306, 307, 309, 316, 345). 

Calophyllum paniculatum Stevens — Camp 1: 
Lewis (s 35, 39), 775. 

Calophyllum sp. 1 — Camp 3: Lewis (s 279). 

Garcinia sp. 1 — Camp 2: Lewis 195; Camp 3: 
Lewis (s 230, 232, 238, 246, 254, 257, 260, 
271, 274, 278, 280). 

Mammea sp. 1 — Camp 4: Lewis 1088. 

Mammea sp. 2 — Camp 4: Lewis 1063. 

Mammea sp. 3 — Camp 1: Lewis (s 20); Camp 2: 
Lewis (s 219). 

Mammea sp. 4 — Camp 1: Lewis 792; Camp 2: 
Lewis (s 129), 882. 

Mammea sp. 5 — Camp 1: Lewis (s 46), 766; 
Camp 2: Lewis 875. 

Mammea sp. 6 (Ochrocarpus cerasifer H. Per- 
rier) — Camp 1: Lewis (s 11), 782. 

Psorospermum sp. 1 — Camp 3: Lewis 983. 

Psorospermum sp. 2 — Camp 1: Lewis 817. 

Psorospermum sp. 3 — Camp 1: Lewis 815. 

Symphonia microphylla (Cambess.) Benth. & 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



59 



Hook. f. ex Vesque ssp. microphylla — 50 km S 
Ambalavao, near abandoned meteo station 
above Ambalamarina, high altitude moss/lichen 
forest-ericaceous bush, Philippia dominant, 
1975 m: Nicoll 249. 
Symphonia sp. 1 — Camp 3: Lewis (s 262, 273, 
321). 



Commelinaceae 

Coleotrype goudotii C.B. Clarke — Camp 1: Lewis 
839. 



Convolvulaceae 



balamarina, high altitude moss/lichen rain for- 
est-ericaceous bush, Philippia dominant, 1975 
m: Nicoll 240. 

Weinmannia mammea Bernardi — Camp 1: Lewis 
781; Camp 3: Lewis 960. 

Weinmannia sp. 1 — Camp 2: Lewis (s 176, 202). 

Weinmannia sp. 2 — Camp 3: Lewis (s 298); 
Camp 4: Lewis (s 413). 

Weinmannia sp. 3 — Camp 4: Lewis (s 465). 
Weinmannia sp. 4 — Camp 2: Lewis (s 224). 
Weinmannia sp. 5 — Camp 1: Lewis (s 53). 
Weinmannia sp. 6 — Camp 2: Lewis (s 181, 204). 
Weinmannia sp. 7 — Camp 4: Lewis (s 386, 428, 
451,453,473,491), 1022. 



Hubertia andringitrensis (Humbert) C. Jeffrey — 
Perrier de la Bathie 2784, 13687. 



Crassulaceae 

Crassula sp. 1 — Growing on rocks, Camp 3: 
Lewis 973. 

Kalanchoe bergeri Raym.-Hamet & H. Perrier — 
de 2000 m a la cime (Perrier de la Bathie, 
1927); "Fissures et corniches 2300-2650 m" 
(Paulian et al., 1971, p. 247). 

Kalanchoe gracilipes (Baker) Baill. — Camp 2: 
Goodman 6380. 

Kalanchoe jongmansi Raym.-Hamet & H. Per- 
rier — de 2000 mala cime (Perrier de la Bathie, 
1927). 

Kalanchoe mangini Raym.-Hamet & H. Perrier — 
de 2000 m a la cime (Perrier de la Bathie, 
1927). 

Kalanchoe miniata Hils. & Bojer var. andringi- 
trensis — H. Perrier. Bois vers 1700 m, versant 
Ouest (Perrier de la Bathie, 1927). 

Kalanchoe sp. 1 — Camp 1: Lewis 800. 

Sedum madagascariense H. Perrier — de 2000 m 
a la cime (Perrier de la Bathie, 1927); "Flore 
xerophile d'altitude 2300-2650 m" (Paulian et 
al., 1971, p. 247). 



Cyperaceae 

Carex andringitrensis Cherm. — Rocailles, vers 
2000 m jusqu' a la cime (Perrier de la Bathie, 
1927). 

Coleochloa setifolia (Ridley) Gilly — "Vegetation 
rupicole 1800 m" (Paulian et al., 1971, p. 237). 

Costularia cf. purpurea Cherm. — Camp 4: Lewis 
1090. 

Cyperus nemoralis Cherm. — Rocailles, vers 2000 
m (Perrier de la Bathie, 1927). 

Cyperus calochrous Cherm. — Rocailles, vers 
2000 m (Perrier de la Bathie, 1927). 

Cyperus micrantherus Cherm. — Rocailles, vers 
2000 m versant Ouest (Perrier de la Bathie, 

1927). 

Mariscus andringitrensis Cherm. — Rocailles, vers 
2000 m (Perrier de la Bathie, 1927). 

Mariscus viguieri Cherm. var. condensatus 
Cherm. — Bords des ruisseaux, vers 2000 m 
versant Est (Perrier de la Bathie, 1927). 

Pycreus reductus Cherm. — Marais, au-dessus de 
1600 m, versant Est (Perrier de la Bathie, 
1927). 

Scleria andringitrensis Cherm. — Rocailles, vers 
2000 m versant Est (Perrier de la Bathie, 1927). 



Cunoniaceae 

Weinmannia eriocarpa Tul. — 50 km S Ambala- 
vao, near abandoned meteo station above Ara- 



Dichapetalaceae 

Dichapetalum sp. 1 — Camp 2: Lewis (s 123), 862. 
Dichapetalum sp. 2 — Camp 1: Lewis 762. 



60 



FIELDIANA: ZOOLOGY 



Dracaenaceae 

Dracaena xiphophylla Baker — Camp 2: Lewis (s 
188), 913. 



Ebenaceae 

Diospyros sp. 1 — Camp 2: Lewis 908; Camp 3: 

Lewis 1019. 
Diospyros sp. 2 — Camp 1: Lewis 831; Camp 2: 

Lewis 871. 
Diospyros sp. 3 — Camp 1 : Lewis 799. 
Diospyros sp. 4 — Camp 3: Lewis 301, 312, 313. 



Elaeocarpaceae 

Elaeocarpus capuronii Tirel — Camp 1: Lewis 
830. 

Elaeocarpus hildebrandtii Baill. — Camp 4: Lewis 
1045; Andringitra, canton de Sendirisoa: Rak- 
oto RN6499. 

Elaeocarpus subserratus Baker — Camp 3: Lewis 
(s 325, 364); Camp 4: Lewis (s 431), 1052. 

Sloanea rhodantha (Baker) Capuron var. rhodan- 
tha form "quadriloba" — Camp 1 : Lewis (s 44, 
68, 114); Camp 2: Lewis (s 141); Andringitra: 
Leandri et al., 3383. 

Sloanea rhodantha (Baker) Capuron var. rhodan- 
tha form "quercifolia" — Camp 3: Lewis (s 
286, 347, 360), 1006; Camp 4: (s 448), 1104, 
1 105; Massif de 1' Andringitra, foret d'Imaiso 
1500-2000 m: Leandri et al. 3383 (form un- 
certain). 



Ericaceae 

Agauria salicifolia Hook. f. — "V6g6tation rupi- 
cole 2000-2300 m" (Paulian et al., 1971, p. 
242); "Foret dense scleYophylle de montagne 
2000 m" (Paulian et al., 1971, p. 250). 

Agauria sp. 1 — Camp 1: Lewis 761. 

Philippia andringitrensis H. Perrier — Broussailles 
encoides de la cime, depuis 2400 m (Perrier de 
la Bathie, 1927). 

Philippia floribunda Benth. ssp. pilulifera (Hum- 
bert) H. Perrier — Broussailles encoides de la 
cime, depuis 2400 m (Perrier de la Bathie, 
1927). 



Philippia pilosa Baker — "V6g6tation liee a la 
pr6sence d'eau 2000-2300 m" (Paulian et al., 
1971, p. 243). 

Philippia spinifera H. Perrier — Broussailles eYi- 
coi'des de la cime, depuis 2400 m (Perrier de la 
Bathie, 1927). 

Philippia viguieri H. Perrier — Broussailles 6ricoi- 
des de la cime, depuis 2400 m (Perrier de la 
Bathie, 1927). 

Philippia sp. 1 — Ericaceous vegetation above the 
tree line, on east side of Ampasipotsy, above 
Camp 4, 1800-2000 m: Lewis 1051, 1065. 

Vaccinium emirnense Hook. — Camp 2: Lewis 
905. 

Vaccinium secundiflorum Hook. — Ericaceous 
vegetation above the tree line, on E side Am- 
pasipotsy, above Camp 4, 1800-2000 m: Lewis 
1062, Goodman 6450; 50 km S Ambalavao, 
near abandoned meteo station above Ambala- 
marina, high altitude moss/lichen rain forest- 
ericaceous bush, Philippia dominant, 1975 m: 
Nicoll289, 290, and 291. 

Vaccinium sp. 1 — Camp 4: Lewis (s 383), 1079. 



Eriocaulaceae 

Eriocaulon fenestratum Bojer — "Veg&ation liee 
a la presence d'eau 2000-2300 m" (Paulian et 
al., 1971, p. 242). 



Erythroxylaceae 

Erythroxylum capitatum Baker — Camp 2: Lewis 

894; Camp 3: Lewis (s 241). 
Erythroxylum sp. 1 — Camp 4: Lewis 1102. 
Erythroxylum sp. 2 — Camp 3: Lewis (s 265); 

Camp 4: Lewis (s 478). 



Euphorbiaceae 

sp. 1 — Camp 4: Lewis (s 439). 

Acalypha andringitrensis Leandri (Madagascar: 

Centre, massif de 1'Andringitra, vallees de la 

Riambava et de 1'Antsifotra). 
Amyrea humbertii Leandri — Camp 1 : Lewis 826. 
Antidesma petiolare Tul. — Camp 2: Lewis (s 124, 

216), 824. 
Blotia tanalorum Leandri — Camp 1: Lewis 824. 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



61 



Claoxylon sp. 1 — Camp 2: Lewis 861. 
Claoxylon sp. 2 — Camp 2: Lewis 863. 
Cleistanthus boivinianus (Baill.) Mull. Arg. — 

Camp 3: Lewis (s 240, 251, 267, 268, 269, 270, 

277, 283, 318, 328, 334, 339, 348, 350), 994. 
Croton myriaster Baker var. austromadecassa 

Leandri (Centre: Andringitra, Ikongo, Ando- 

hahelo, Beampingaratra). 
Croton nitidulus Baker var. tandrokensis Leandri 

(Massif d' Andringitra, col du Tandroka, versant 

Est; 1200 m). 

Croton sp. 1 — Camp 3: Lewis 987; Camp 4: Lew- 
is (s 418), 1023. 
Croton sp. 2— Camp 3: Lewis (s 234, 239, 250). 
Croton sp. 3 — Camp 2: Lewis 878, 931. 

Drypetes madagascariensis (Lam.) Humbert & 
Leandri — Camp 2: Lewis 917. 

Euphorbia emirnensis Baker — Open ericaceous 

vegetation: Lewis 1068. 
Macaranga ankafinensis Baill. — Camp 1: Lewis 

752; Camp 2: Lewis 849; Camp 3: Lewis 939, 

993. 
Macaranga cf. decaryana Leandri — Camp 2: 

Lewis 859; Camp 3: Lewis 963; Camp 4: Lewis 

1110. 
Macaranga cuspidata Baill. — Camp 3: Lewis 

1009. 
Macaranga echinocarpa Baker — Camp 3: Lewis 

1017; Camp 4: Lewis (s 369, 391, 392). 

Macaranga oblongifoia Baill. — Camp 4: Good- 
man 6458. 
Macaranga sp. 1 — Camp 2: Lewis 925. 

Macaranga sp. 2 — Camp 1: Lewis 809; Camp 2: 
Lewis 918. 

Mallotus cf. capuronii Leandri — Camp 2: Lewis 
927; Camp 3: Lewis 287, 276, 323, 338, 952. 

Phyllanthus fusco-luridus Mull. Arg. ssp. villosus 
Leandri (Andringitra). 

Phyllanthus iratsiensis Leandri (Massif de 1' An- 
dringitra, Iratsy, vallees de la Riambava). 

Phyllanthus monticola Leandri (Andringitra, foret 
a mousses). 

Phyllanthus sp. 1 — Camp 2: Lewis 891. 

Phyllanthus sp. 2 — Ericaceous vegetation, on E 

side Ampasipotsy, above Camp 4, 1800 m: 

Lewis 1056. 

Phyllanthus sp. 3 — Camp 1: Lewis 822. 

Savia andringitrana Leandri (base nord du Pic 
d'lvohibe). 



Suregada baronii (Moore) Croizat — Camp 1: 

Lewis 833. 
Uapaca sp. 1 — Camp 2: Lewis (s 192). 



Fabaceae 

Abrus precatorius L. — Camp 1 : Lewis 779. 
Abrus sp. 1 — Camp 1: Lewis 802. 
Albizia gummifera (J.E Gmel) C.A. Sm. — Camp 
1: Lewis (s 90), 767. 

Albizia sp. 1 — Camp 1: Lewis 798. 

Crotalaria andringitrensis R. Vig. — (Andringitra 

range). 
Dalbergia sp. 1 — Camp 2: Lewis (s 182, 217). 

Desmodium repandum DC. — "Foret dense de 
moyenne altitude 1500-1650 m" (Paulian et 
al., 1971, p. 250). 

Dolichos minutiflorus R. Vig. (Massif de 1' An- 
dringitra). 

Indigofera andringitrensis R. Vig. (Massif de 
F Andringitra; 1400-2000 m). 

Mimosa andringitrensis R. Vig. (Massif de 1' An- 
dringitra; 1400-2000 m)— Bois, vers 1600 m 
(Perrier de la Bathie, 1927). 

Mimosa descarpentriesii R. Vig. (Massif de 
I' Andringitra; 1800-2300 m). 

Mundulea andringitrensis R. Vi< . (Massif de 
l'Andringitra) — "Hauts fourre arbustifs 2000- 
2300 m" (Paulian et al., 1971, p. 238); "Ve- 
getation rupicole 2000-2300 m" (Paulian et al., 
1971, p. 242). 

Mundulea splendens R. Vig. (environs d'Ambos- 
itra a l'Andringitra; 1400-2000 m). 

Smithia chamaecrista Benth. var. stipulata R. Vig. 
(versant Ouest de l'Andringitra, 800-1600 m). 

Strongylodon sp. 1 — Camp 2: Lewis 858. 



Flacourtiaceae 

Aphloia theiformis Benn — Camp 2: Lewis 785, 
879; Camp 3: Lewis (s 275); Camp 4: Lewis (s 
410, 414), 1027. 

Casearia elliptica Tul. — Camp 3: Lewis 1018. 
Casearia nigrescens Tul. — Camp 2: Lewis 930. 

Casearia nigrescens Tul. var. lucida (Tul.) Sleu- 
mer — Approximately 5 km SE Antanifotsy, the 
reserve headquarters, 1500 m: Schatz 2678, 



62 



FIELDIANA: ZOOLOGY 



with P. Goldblatt, A. Rakotozafy, and J. Ran- 

drianasolo. 
Casearia sp. 1 — Camp 4: Lewis (s 371, 377, 398), 

1024. 
Scolopia sp. 1 — Camp 3: Lewis 1013. 
Scolopia sp. 2 — Camp 4: Lewis 1080. 



Gentianaceae 

Exacum dipterum Klack. — Versant est du massif 
de I'Andringitra, 1200 m: Perrier de la Bathie 
9069; Andringitra, Col du Tandroka, versant E 
du massif d' Andringitra, 1600 m: Perrier de la 
Bathie 9067. 

Exacum emirnense (Baker) Schinz — Andringitra, 
Sendrisoa: Rakoto RN6496. 

Exacum naviculare Klack. — Andringitra, 2000 m: 
Perrier de la Bathie 9080bis. 

Ornichia madagascariensis (Baker) Klack. ssp. 
pubescens (Baker) Klack. — Andringitra: Raza- 
findrakoto, RN3501; Camp 1: Lewis 811. 

Swertia rosulata (Baker) Klack. Andringitra, can- 
ton de Sendrisoa: Randriamiera RN5579. 

Tachiadenus platypterus Baker ssp. platypterus — 
N. du massif d' Andringitra 1600 m: Rakoto 
RN6475; Andringitra, canton de Vohitsaoka; 
Perrier de la Bathie 14468. 



Streptocarpus hilsenbergii R. Br. — Growing on 

rocks, Camp 3: Lewis 1008. 
Streptocarpus macropodus B.L. Burtt — Camp 4: 

Lewis 1038, 1043. 
Streptocarpus suffruticosus Humbert "Foret a 

strate herbac£e [below] 1970 m" (Paulian et al., 

1971, p. 254); Growing on rocks, Camp 3: 

Lewis 955, 962. 



Hyacinthaceae 

Rhodocodon urgineoides Baker — Camp 2: Lewis 
876. 



Icacinaceae 

sp. 1 — Camp 3: Lewis 982, 992; Camp 4: Lewis 
1035. 

sp. 2 — Camp 1: Lewis 836. 

sp. 3 — Camp 3: Lewis (s 248). 

Apodytes sp. 1 — Camp 2: Lewis (s 155). 

Apodytes sp. 2 — Camp 4: Lewis 1082. 

Cassinopsis tomentosa H. Perrier — Camp 3: Lew- 
is (s 297, 288), 945. 



Iridaceae 



Geraniaceae 

Geranium andringitrense H. Perrier — Rocailles, 
de 2000 m a la cime (Perrier de la Bathie, 
1927); "Foret dense sclerophylle de montagne 
2000 m" (Paulian et al., 1971, p. 250); erica- 
ceous vegetation above the tree line, on E side 
Ampasipotsy, above Camp 4, 1800-2000 m: 
Lewis 1069. 

Pelargonium madagascariense Baker — Rocailles, 
de 2000 a 2400 m (Perrier de la Bathie, 1927). 



Gesneriaceae 

Didymocarpus madagascariensis C.B. Clarke — 
Growing on rocks, Camp 3: Lewis 972. 

Didymocarpus vestitus Baker var. lanceolatus 
Humbert ex B.L. Burtt (versant sud du massif 
de I'Andringitra, riviere Ihovika, bassin de la 
Matitanana). 



Aristea angustifolia Baker — Slopes of massif, be- 
low steep cliffs, 6-8 km S Antanifotsy, along 
path to Pic Boby, 1900-2050 m: Goldblatt and 
Schatz 8971; Massif d' Andringitra, 1800-2200 
m: Perrier de la Bathie 14410. 

Aristea cladocarpa Baker — Camp 1: Lewis 804; 
50 km S Ambalavao, near abandoned meteo 
station above Ambalamarina, high altitude 
moss/lichen rain forest-ericaceous bush, Phi- 
lippia dominant, 1975 m; grassy/rocky summit 
with orchids, 1975 m: Nicoll 241. 

Aristea humbertii H. Perrier — Massif de I'Andrin- 
gitra, Iratsy: vallees de la Riambava et de 
1'Antsifotra et montagnes environnantes 2000- 
2500 m; Humbert 3790. 

Aristea madagascariense Baker — Andringitra, 
Canton de Sendrisoa: Henri RN4827; Andrin- 
gitra: Razafindrakoto RN2408. 

Gladiolus andringitrae Goldblatt — Andringitra, 
Boby: Homolle 1189; Massif de I'Andringitra 
2400 m: Perrier de la Bathie 13655; Massif de 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



63 



1' Andringitra 2000-2400 m; Perrier de la Bath- 
ie 14483; Andringitra Forest Reserve, slopes of 
massif, below steep cliffs, 6-8 km S Antani- 
fotsy, along path to Pic Boby, 1900-2050 m: 
Goldblatt and Schatz 8974. 

Gladiolus bojeri (Baker) Goldblatt — Plateau 
d'Andohariana. Andringitra 2000-2100 m: 
Guillaumet 3754; Massif de l'Andringitra, Ir- 
atsy, vallees de la Riambava et de l'Antsifotra 
et montagnes environnantes 1600-2000 m: 
Humbert 3655; Andringitra Forest Reserve, 
slopes of massif, below steep cliffs, 6-8 km S 
Antanifotsy, along path to Pic Boby, 1900- 
2050 m: Goldblatt and Schatz 8970. 

Gladiolus dalenii Geel — Andringitra, Canton de 
Sendrisoa, Henri RN5554; Andringitra: Raza- 
findrakoto RN2412. 



Lamiaceae 



sp. 14 — Camp 2: Lewis (s 226). 

sp. 15— Camp 3: Lewis (s 243, 329, 332); Camp 

4: Lewis (s 461). 
sp. 16 — Camp 1: Lewis (s 3). 
Cryptocarya crassifolia Baker — Camp 3: Lewis 

340; Camp 4: Lewis (s 374, 503), 1081. 
Cryptocarya madagascariensis (Baill.) Kos- 

term. — Camp 1: Lewis (s 99). 
Cryptocarya oppositifolia Kosterm. — Camp 4: 

Lewis 1085. 
Cryptocarya sp. 1 — Camp 4: Lewis 1031. 
Cryptocarya sp. 2 — Camp 1: Lewis (s 19, 57); 

Camp 2: Lewis (s 213). 
Cryptocarya sp. 3 — Camp 1: Lewis (s 8, 112); 

Camp 2: Lewis 911. 
Ocotea sp. 1 — Camp 3: Lewis 1005. 
Ocotea sp. 2 — Camp 2: Lewis 889; Camp 4: Lew- 
is (s 406, 411, 412, 435, 466), 1026. 



Micromeria sp. 1 — Ericaceous vegetation, on east 
side of Ampasipotsy, above Camp 4, 2000 m: 
Lewis 1064. 

Orthosiphon sp. 1 — Camp 4: Lewis 1108. 

Orthosiphon sp. 2 — Camp 3: Lewis 933. 

Perrierastrum oreophilum Guillaumin (Massif 
d' Andringitra). 

Plectranthus sp. 1 — Camp 3: Goodman 6408. 

Solenostemon bojeri (Benth.) Guillaumet & Cor- 
net — Camp 4: Goodman 6455. 



Lauraceae 



Lentibulariaceae 

Utricularia humbertiana H. Perrier var. andrin- 
gitrensis H. Perrier (Massif de l'Andringitra, 
2500 m). 

Utricularia livida E. Mey. — 50 km S Ambalavao, 
near abandoned meteo station above Ambala- 
marina, high altitude moss/lichen rain forest- 
ericaceous bush, Philippia dominant, 1975 m: 
Nicoll 243. 

Utricularia mauroyae H. Perrier (Massif de 
l'Andringitra, sur les cimes, vers 2500 m). 



sp. 1— Camp 3: Lewis (s 261, 302, 314, 342). 
sp. 2 — Camp 2: Lewis (s 165); Camp 3: Lewis (s 

300, 367); Camp 4: Lewis (s 420). 
sp. 3 — Camp 1: Lewis (s 106). 
sp. A — Camp 2: Lewis (s 173). 
sp. 5 — Camp 2: Lewis (s 212). 
sp. 6 — Camp 2: Lewis (s 162). 
sp. 7 — Camp 4: Lewis (s 401). 
sp. 8 — Camp 2: Lewis (s 137). 
sp. 9— Camp 2: Lewis (s 150, 159). 
sp. 10 — Camp 1: Lewis (s 22); Camp 2: Lewis (s 

144). 
sp. 11 — Camp 1: Lewis (s 65). 
sp. 12 — Camp 1: Lewis (s 66). 
sp. 13 — Camp 1: Lewis (s 72). 



Loganiaceae 

Anthocleista madagascariensis Baker — Camp 2: 
Lewis 897; Camp 3: Lewis 337; Camp 4: Lewis 
1089. 

Buddleja indica Lam. — Camp 1: Lewis 821. 

Nuxia involucrata A.DC. — Camp 1: Lewis 773. 

Strychnos sp. 1 — Camp 3: Lewis 975. 



Loranthaceae 

Bakerella clavata (Desrouss.) Balle (var. not de- 
termined) — Camp 3: Lewis 950; Camp 4: Lew- 
is 1025. 

Bakerella clavata (Desrouss.) Balle var. lenticel- 



64 



FIELDIANA: ZOOLOGY 



lata (Baker) Balle — Camp 4: Lewis 1029, 
1041, Goodman 6453; ericaceous vegetation 
above the tree line, on E side Ampasipotsy, 
above Camp 4, 1800-2000 m: Lewis 1057. 

Bakerella clavata (Desrouss.) Balle var. perala- 
ta — Camp 1: Lewis 801. 

Bakerella tandrokensis (Lecomte) Balle (Andrin- 
gitra) — Ericaceous vegetation above the tree 
line, on east side of Ampasipotsy, above Camp 
4, 2050 m: Goodman 6459. 



Lythraceae 

Rotala sphagnoides H. Perrier (Massif d'Andrin- 
gitra). 



Malpighiaceae 

Tristellateia madagascariensis Poir. — Camp 2: 
Lewis 902. 



Malvaceae 

Kosteletzkya macrantha Hochr. — Savoka a Philip- 
pia, ravins de 1500 a 2000 m, versant Ouest 
(Perrier de la Bathie, 1927). 

Kosteletzkya malvocaerulea Hochr. — Savoka a 
Philippia, ravins de 1500 a 2000 m, versant 
Ouest (Perrier de la Bathie, 1927). 



Gravesia calliantha Jum. & H. Perrier — vers 

1700 m Rocailles, versant SE. (Perrier de la 

Bathie, 1927); Camp 2: Goodman 6409. 
Gravesia cf. vestita (Baker) H. Perrier — Camp 2: 

Lewis 884. 
Gravesia inappendiculata H. Perrier — Camp 4: 

Goodman 6456. 
Gravesia sp. 1 — Camp 3: Lewis 996. 
Gravesia sp. 2 — Camp 3: Lewis 935. 
Gravesia sp. 3 — Camp 1: Lewis 837. 
Medinilla ericarum Jum. & H. Perrier — Camp 2: 

Goodman 6407. 
Medinilla sp. 1 — Camp 2: Lewis 890; Camp 3: 

Lewis 953; Camp 4: Lewis 1059, 1093. 
Medinilla sp. 2 — Camp 3: Lewis 977. 
Medinilla sp. 3 — Camp 2: Lewis 856. 
Medinilla sp. 4 — Camp 1: Lewis 813. 
Medinilla torrentium Jum. & H. Perrier — Bois, 

vers 1 700 m, versant SE. (Perrier de la Bathie, 

1927); Camp 4: Goodman 6429. 
Memecylon cf. ivohibense Jacq.-FeL — Camp 3: 

Lewis (s 346) 892. 
Memecylon cf. longipetalum H. Perrier — Camp 1 : 

Lewis (s 21); Camp 2: Lewis 868. 
Memecylon cf. perangustum Jacq.-FeL — Camp 1 : 

Lewis 806. 
Osbeckia andringitrensis H. Perrier — Rocailles, 

vers la cime (Perrier de la Bathie, 1927). 
Osbeckia bicolor H. Perrier — Bois, vers 2000 m, 

versant Est (Perrier de la Bathie, 1927). 
Osbeckia mandrarensis H. Perrier — Camp 4: 

Lewis 1097. 



Melanophyllaceae 

Melanophylla cf. alnifolia Baker — Camp 2: Lewis 

847, 860. 
Melanophylla cf. humbertiana Keraudren — Camp 

2: Lewis 895. 
Melanophylla sp. 1 — Camp 4: Lewis (s 422). 



Melastomataceae 



Meliaceae 

Turraea sp. 1 — Camp 1: Lewis 838. 

Menispermaceae 

Burasaia madagascariensis Thouars — Camp 1: 

Lewis 829. 
Cyclea sp. 1 — Camp 1: Lewis 765. 



Dichaetanthera matitanensis Jum. & H. Perrier — 
Foret a mousses, vers 1 800 m, versant Est (Per- 
rier de la Bathie, 1927). 

Dichaetanthera sp. 1 — Camp 1: Lewis 763. 

Dichaetanthera sp. 2 — Camp 2: Lewis 854. 



Menyanthaceae 

Limnanthemum aff. indicum Griseb. — "Vegeta- 
tion liee a la presence d'eau 2000-2300 m" 
(Paulian et al., 1971, p. 242). 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



65 



Monimiaceae 

Decaryodendron perrieri Cavaco — Camp 2: Lew- 
is 901. 

Ephippiandra microphylla (Perk.) Cavaco — 
Camp 4: Lewis 1100. 

Ephippiandra sp. 1 — Camp 1: Lewis (s 40). 

Tambourissa cf. parvifolia Baker — Camp 3: Lew- 
is (s 315), 942, 951, 964. 

Tambourissa sp. 1 — Camp 4: Lewis (s 452). 

Tambourissa sp. 2 — Camp 4: Lewis (s 402). 

Tambourissa sp. 3 — Camp 1: Lewis 816; Camp 
2: Lewis 874, 898. 

Tambourissa thouvenotii Danguy — Camp 1 : Lew- 
is 789, 834; Camp 2: Lewis (s 214). 



Moraceae 

Ficus brachyclada Baker — Camp 1: Lewis 825; 

Camp 2: Lewis 841, 888. 
Ficus lutea Vahl — Camp 1: Lewis 820. 
Ficus politoria Lam. — Camp 3: Lewis 966; Camp 

4: Lewis 1103. 
Ficus reflexa Thunb. — Camp 1: Goodman 6353. 
Ficus rubra Vahl — Camp 1: Lewis 724; Camp 2: 

Lewis 866. 
Ficus tiliifolia Baker — Camp 1 : Lewis 790; Camp 

2: Lewis 904. 
Ficus trichopoda Baker — Camp 1: Lewis 769. 
Ficus sp. 1 — Camp 3: Lewis 988. 
Streblus dimepate (Bureau) C.C. Berg — Camp 2: 

Lewis (s 133, 145, 200). 



Myrothamnaceae 

Myrothamnus moschatus (Baill.) Baill. — Andrin- 
gitra, canton de Vohitsaoka: Rakotovao 
RN6555. 



marina, mid-altitude rain forest with high stem 
density and canopy at 25-30 m, 1700 m: Nicoll 
305. 

Monoporus simplex H. Perrier — 50 km S Amba- 
lavao, near abandoned meteo station above 
Ambalamarina, high altitude moss/lichen rain 
forest-ericaceous bush, Philippia dominant, 
1975 m: Nicoll 233. 

Monoporus sp. 1 — Camp 2: Lewis 842. 

Monoporus sp. 2 — Camp 2: Lewis 870. 

Oncostemon ankifiense Mez. — Camp 4: Goodman 
6430. 

Oncostemon botryoides Baker — Camp 1: Lewis 
783. 

Oncostemon microsphaerum Baker — Camp 3: 
Lewis 999. 

Oncostemon racemiferum Mez. — Camp 1: Good- 
man 6352. 

Oncostemon sp. 1 — Camp 2: Lewis 923. 

Oncostemon sp. 2 — Camp 3: Lewis 1003. 

Oncostemon sp. 3 — Camp 2: Lewis 929. 

Oncostemon sp. A — Camp 3: Lewis 1004. 

Oncostemon sp. 5 — Camp 1: Lewis 812. 

Oncostemon sp. 6 — Camp 4: Lewis 1096. 

Oncostemon sp. 7 — Camp 4: Lewis 1044, 1101. 

Oncostemon sp. 8 — Camp 4: Lewis 1060. 

Oncostemon sp. 9 — Camp 3: Lewis 1010. 

Oncostemon sp. 10 — Camp 3: Lewis 957; Camp 
4: Lewis 1106. 

Oncostemon sp. 11 — Camp 2: Lewis 921. 

Oncostemon sp. 12 — Camp 1: Lewis 764. 

Oncostemon sp. 13 — Camp 2: Lewis (s 208); 
Camp 3: Lewis 986. 

Oncostemon sp. 14 — Camp 3: Lewis 361; Camp 
4: Lewis 1086. 

Oncostemon sp. 15 — Camp 4: Lewis 1036. 

Oncostemon sp. 16 — Camp 4: Lewis 1040. 



Myrtaceae 



Myrsinaceae 

Embelia concinna Baker — Camp 3: Lewis 989. 
Embelia madagascariensis A. DC. — Camp 1: 
Lewis 757. 

Maesa lanceolata Forssk. — 50 km S Ambalavao, 
near abandoned meteo station near Ambala- 



Syzygium sp. 1 — Camp 2: Lewis 928. 

Syzygium sp. 2 — Camp 2: Lewis (s 170); Camp 

3: Lewis (s 228, 249, 281, 335). 
Syzygium sp. 3 — Camp 2: Lewis (s 168, 221). 
Syzygium sp. 4 — Camp 4: Lewis (s 456, 468, 

500). 
Syzygium sp. 5 — Camp 1: Lewis (s 31, 63). 
Syzygium sp. 6 — Camp 1: Lewis (s 113). 



66 



FIELDIANA: ZOOLOGY 



Syzygium sp. 7 — Camp 4: Lewis (s 385). 
Syzygium sp. 8 — Camp 1: Lewis (s 70). 
Syzygium sp. 9 — Camp 4: Lewis (s 426). 
Syzygium sp. 10 — Camp 4: Lewis (s 372). 
Syzygium sp. 11 — Camp 1: Lewis (s 87); Camp 

2: Lewis 899. 
Syzygium sp. 12 — Camp 3: Lewis (s 303). 
Syzygium sp. 13 — Camp 1: Lewis 776. 
Syzygium sp. 14 — Camp 1: Lewis (s 52, 79), 828; 

Camp 2: Lewis (s 227), 867, 922. 
Syzygium sp. 15 — Camp 3: Lewis (s 258). 
Syzygium sp. 16 — Camp 3: Lewis (s 293). 
Syzygium sp. 17 — Camp 1: Lewis (s 92). 



Ochnaceae 

Campylospermum obtusifolium Tiegh. — Camp 1: 
Lewis 794, 807. 



Oleaceae 

Noronhia sp. 1 — Camp 2: Lewis 900. 

Noronhia sp. 2 — Camp 3: Lewis 1007; Camp 4: 
Lewis 1107. 

Noronhia sp. 3 — Camp 3: Lewis (s 294). 
Noronhia sp. 4 — Camp 3: Lewis (s 299, 326), 
947. 

Noronhia sp. 5 — Camp 3: Lewis 1011. 
Noronhia sp. 6 — Camp 1: Lewis (s 5). 
Noronhia sp. 7 — Camp 1: Lewis (s 37). 



Orchidaceae 

Angraecum oblongifolium Toill. — Gen. & Bosser 
(Andringitra). 

Angraecum sp. 1 — Camp 3: Lewis 948. 

Angraecum sp. 2 — Camp 4: Lewis 1091. 

Benthamia exilis Schltr. var. exilis (Massif d' An- 
dringitra, 2200 m alt.; 2400 m alt.) — Rocailles, 
de 2000 m a la cime (Perreir de la Bathie, 
1927). 

Benthamia exilis Schltr. var. tenuissima Schltr. 

(Massif d' Andringitra, 2200 m alt.). 
Benthamia macro Schltr. (Massif d'Andringitra, 

2200 m alt.). 

Benthamia monophylla Schltr. (Massif d'Andrin- 



gitra, Buissons ericoides, vers 2600 m) — Ro- 
cailles vers 2000 m et au-dessus (Perreir de la 
Bathie, 1927); 50 km S Ambalavao, near 
abandoned meteo station above Ambalamari- 
na, high altitude moss/lichen rain forest-eri- 
caceous bush, Philippia dominant, 1975 m: 
Nicoll 262. 

Benthamia oreodorum Schltr. — Foret a mousses, 
vers 1 800 m, versant Est (Perrier de la Bathie, 
1927). 

Benthamia praecox Schltr. — 50 km S Ambalavao, 
near abandoned meteo station above Ambala- 
marina, high altitude moss/lichen rain forest- 
ericaceous bush, Philippia dominant, 1975 m: 
Nicoll 288. 

Bulbophyllum andringitranum Schltr. — Rocailles, 
au-dessus de 2000 m, versant Ouest (Perrier de 
la Bathie, 1927). 

Bulbophyllum sp. 1 — Camp 2: Lewis 857. 

Calanthe repens Schltr. var. pauliani Ursch & 
Toill. — Gen. (Andringitra). 

Cynorkis alborubra Schltr. (Massif d'Andringitra, 
2600 m alt.) — Rocailles, au-dessus de 2000 m 
(Perrier de la Bathie, 1927). 

Cynorkis andingitrana Schltr. — Rocailles, au-des- 
sus de 2000 m (Perrier de la Bathie, 1927). 

Cynorkis angustipetala Ridl. var. amabilis 
(Schltr.) H. Perrier — Rocailles, au-dessus de 
2000 m (Perrier de la Bathie, 1927); on trail 
from Antanifotsy to Pic Boby, high altitude 
meadow by mountain stream, 2000 m: Nicoll 
323. 

Cynorkis baronii Rolfe (Shady places on humus 
on summit of Andringitra Mts) — On trail from 
Antanifotsy to Pic Boby, high altitude meadow 
by mountain stream, 2000 m: Nicoll 324. 

Cynorkis bathiei Schltr. (Massif d'Andringitra, 
2400 m alt.) — Rocailles, au-dessus de 2000 m 
(Perrier de la Bathie, 1927). 

Cynorkis bella Schltr. — Rocailles, au-dessus de 
2000 m (Perrier de la Bathie, 1927). 

Cynorkis filiformis H. Perrier — Rocailles, au-des- 
sus de 2000 m (Perrier de la Bathie, 1927). 

Cynorkis gibbosa Ridl. — On trail from Antanifot- 
sy to Pic Boby, high altitude meadow by moun- 
tain stream, 2000 m: Nicoll 327. 

Cynorkis gigas Schltr. — Rocailles, au-dessus de 
2000 m (Perrier de la Bathie, 1927). 

Cynorkis inversa Schltr. — Rocailles, au-dessus de 
2000 m (Perrier de la Bathie, 1927). 

Cynorkis jumelleana Schltr. var. gracillima 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



67 



Schltr. — Rocailles, au-dessus de 2000 m (Per- 
rier de la Bathie, 1927). 

Cynorkis jumelleana Schltr. var. jumelleana — Ro- 
cailles, au-dessus de 2000 m (Perrier de la 
Bathie, 1927). 

Cynorkis lilacina Ridley var. pulchra (Schltr.) H. 
Perrier — Rocailles, au-dessus de 2000 m (Per- 
rier de la Bathie, 1927). 

Cynorkis ochroglossa Schltr. — Rocailles, au-des- 
sus de 2000 m (Perrier de la Bathie, 1927). 

Cynorkis quinqueloba H. Perrier — 50 km S Am- 
balavao, near abandoned meteo station above 
Ambalamarina, high altitude moss/lichen rain 
forest-ericaceous bush, Philippia dominant, 
1975 m: Nicoll 244. 

Cynorkis saxicola Schltr. — Rocailles, au-dessus 
de 2000 m (Perrier de la Bathie, 1927). 

Cynorkis sp. 1 — Camp 1: Lewis 810. 

Disa andringitrana Schltr. — Broussailles ericoi- 
des, 2200 m (Perrier de la Bathie, 1927). 

Habenaria decaryana H. Perrier — Camp 3: Lewis 
949. 

Liparis andringitrana Schltr. — Rocailles, 2000 m, 
versant Est (Perrier de la Bathie, 1927). 

Liparis densa Schltr. — Rocailles, 2000 m, versant 
Est (Perrier de la Bathie, 1927). 

Liparis henrici Schltr. — Foret a mousses, vers 
1800 m, versant Est (Perrier de la Bathie, 
1927). 

Liparis latilabris Schltr. — Foret a mousses, vers 
1800 m, versant Est (Perrier de la Bathie, 
1927). 

Liparis listeroides Schltr. — Foret a mousses, vers 
1800 m, versant Est (Perrier de la Bathie, 
1927). 

Liparis sp. 1 — Camp 2: Goodman 6381. 

Polystachya oreocharis Schltr. — 50 km S Amba- 
lavao, near abandoned meteo station above 
Ambalamarina, high altitude moss/lichen rain 
forest-ericaceous bush, Philippia dominant, 
1975 m: Nicoll 248. 

Satyrium trinerve Lindl. — On trail from Antani- 
fotsy to Pic Boby, high altitude meadow by 
mountain stream, 2000 m: Nicoll 326. 

Tylostigma foliosum Schltr. (Massif d'Andringitra, 
2500 m alt.) — Savoka a Philippia, vers 2000 m, 
versant Est (Perrier de la Bathie, 1927). 



Pandanaceae 

Pandanus sp. 1 — Camp 1: Lewis 768. 
Pandanus sp. 2 — Camp 4: Lewis 1032. 
Pandanus sp. 3 — Camp 1: Lewis 787. 
Pandanus sp. 4 — Camp 2: Lewis 848. 

Passifloraceae 

Deidamia bicolor H. Perrier — Camp 1: Lewis 
840. 



Phormiaceae 

Dianella ensifolia (L.) Redoute — "Foret dense 
sclerophylle de montagne [above] 1950 m" 
(Paulian et al., 1971, p. 256). 



Piperaceae 

Peperomia sp. 1 — Camp 3: Lewis 946. 
Peperomia sp. 2 — Camp 3: Lewis 978. 
Piper sp. 1 — Camp 2: Lewis 872. 



Pittosporaceae 

Pittosporum sp. 1 — Camp 3: Lewis 956. 
Pittosporum sp. 2 — Camp 1: Lewis 770. 
Pittosporum sp. 3 — Camp 4: Lewis 1039. 
Pittosporum sp. 4 — Camp 2: Lewis 903. 



Poaceae 

Andropogon trichozygus Baker — "Prairies alti- 
montaines 2000-2300 m" (Paulian et al., 1971, 
p. 242); "Vegetation liee a la presence d'eau 
2000-2300 m" (Paulian et al., 1971, p. 242). 

Arundinella nepalensis Trin. — "Vegetation liee a 
la presence d'eau 2000-2300 m" (Paulian et 
al., 1971, p. 242). 

Brachiaria dimorpha A. Camus. (Massif d'An- 
dringitra) — Foret a mousses, vers 2000 m (Per- 
rier de la Bathie, 1927); "Prairies altimontaines 
2000-2300 m" (Paulian et al., 1971, p. 242). 

Dichanthium andringitrensis A. Camus — Rocail- 



68 



FIELDIANA: ZOOLOGY 



les, 2000 m, versant Ouest (Perrier de la Bathie, 
1927). 

Digitaria perrieri A. Camus — Rocailles, 2000 m 
(Perrier de la Bathie, 1927). 

Festuca perrieri A. Camus sp. 1 — pres des eaux, 
vers 2000 m versant Est (Perrier de la Bathie, 
1927). 

Lasiorrachis viguieri (A. Camus) Bosser — "Prai- 
ries altimontaines 2000-2300 m" (Paulian et 
al., 1971, p. 242). 

Lecomtella madagascariensis A. Camus (Massif 
d'Andringitra, 1600-2400 m); versant Ouest, 
de 1600 m a 2400 m (Perrier de la Bathie, 
1927). 

Nastus decaryanus A. Camus (Massif d'Andrin- 
gitra). 

Nastus elongatus A Camus — Bois, vers 2000 m, 
versant Est (Perrier de la Bathie, 1927). 

Panicum aequinerve Nees — 50 km S Ambalavao, 
near abandoned meteo station above Ambala- 
marina, high altitude moss/lichen rain forest- 
ericaceous bush, Philippia dominant, 1975 m: 
Nicoll 295. 

Panicum ambositrense A. Camus — "Foret dense 
scterophylle de montagne [above] 1950 m" 
(Paulian et al., 1971, p. 256). 

Panicum cupressiforme A. Camus — "V6g6tation 
liee a la presence d'eau 2000-2300 m" (Pau- 
lian et al., 1971, p. 243); "Groupements her- 
baces 2000-2500 m" (Paulian et al., 1971, p. 
247); "La cuvette de Boby 2500 m" (Paulian 
et al., 1971, p. 247). 

Panicum neoperrieri A. Camus (Massif d'Andrin- 
gitra, 1600 m). 

Panicum spergulifolium A. Camus (Massif 
d'Andringitra, 2200-2500 m) — Rocailles, vers 
2000 m a la cime (Perrier de la Bathie, 1927); 
"Prairies altimontaines 2000-2300 m" (Pauli- 
an et al., 1971, p. 241); "V6g6tation H6e a la 
presence d'eau 2000-2300 m" (Paulian et al., 
1971, p. 242). 

Panicum subhystrix A. Camus sp. 1 — Rocailles, 
vers 2000 m (Perrier de la Bathie, 1927). 

Pentaschistis andringitrensis A. Camus (Massif 
d'Andringitra). 

Poa perrieri A. Camus sp. 1 — pres des eaux, vers 
2000 m versant Est (Perrier de la Bathie, 1927). 

Proteaceae 

Faurea forficuliflora H. Perrier (Massif d'Andrin- 
gitra, 2400 m). 



Ranunculaceae 

Clematis mauritiana Lam. var. sulfurea R. Vig. & 
H. Perrier (Massif d'Andringitra, vers 2200 m). 



Restionaceae 

Restio madagascariensis Cherm. — "Groupements 
herbac^s 2100-2500 m" (Paulian et al., 1971, 
p. 247); "La cuvette de Boby 2500 m" (Paulian 
et al., 1971, p. 247). 



Rhamnaceae 

sp. 1— Camp 3: Lewis (s 349), 959, 1 109. 

Rhopalocarpaceae 

Rhopalocarpus sp. 1 — Camp 2: Lewis (s 147) 

Rosaceae 

Alchemilla andringitrensis R. Vig. & de Wild. — 

Rocailles, de 2000 m a la cime (Perrier de la 

Bathie, 1927). 
Rubus rosaefolius Sm. — "Foret dense scleYophyl- 

le de montagne 2000 m" (Paulian et al., 1971, 

p. 250). 

Rubiaceae 

sp. 1— Camp 2: Lewis 205. 

sp. 2— Camp 3: Lewis 970. 

sp. 3 — Camp 2: Lewis 906. 

sp. 4 — Camp 4: Lewis (s 424). 

sp. 5 — Camp 2: Lewis 924. 

sp. 6 — Camp 3: Lewis 1000. 

sp. 7 — Camp 3: Lewis 968. 

Alberta humblotii Drake — Camp 2: Lewis 916. 

Alberta minor Baill. — Open ericaceous vegeta- 
tion, on E side Ampasipotsy, above Camp 4, 
1800-2000 m: Lewis 1070. 

Anthospermum ibityense Puff — 50 km S Amba- 
lavao, near abandoned meteo station above 
Ambalamarina, high altitude moss/lichen rain 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



69 



forest-ericaceous bush, Philippia dominant, 

1975 m: Nicoll 292. 
Anthospermum madagascariense Homolle ex Puff 

(Andringitra Massif, brousse ericoide vers 2600 

m). 
Breonia sp. — Camp 1: Lewis 755. 

Canthium andrigitrense Cavaco (Massif de 1' An- 
dringitra, Iratsy, vallees de la Riambava et de 
l'Antsifotra). 

Canthium parvistipula Arenes & Cavaco — Camp 
1: Lewis 780; Camp 4: Lewis 1049. 

Canthium sp. 1 — Camp 3: Lewis 984. 

Canthium sp. 2 — Camp 4: Lewis 1087. 

Canthium sp. 3 — Camp 3: Lewis 997. 

Canthium sp. A — Camp 3: Lewis (s 331). 

Canthium sp. 5 — Camp 2: Lewis 171. 

Chassalia ambodirianensis Bremek. (Massif de 
l'Andringitra, foret d'Ambodipaiso, 1100 m). 

Chassalia bojeri Bremek. var. bojeri — 50 km S 
Ambalavao, near abandoned meteo station 
above Ambalamarina, high altitude moss/lichen 
rain forest-ericaceous bush, Philippia domi- 
nant, 1975 m: Nicoll 301. 

Chassalia sp. 1 — Camp 3: Goodman 6421. 

Danais fragrans Gaertn. — Camp 4: Lewis 1098. 

Danais paludosa Baker — Camp 2: Lewis 920; 
Camp 3: Lewis 1012. 

Gaertnera sp. 1 — Camp 1: Lewis 756, 819; Camp 
2: Lewis (s 136, 211). 

Gaertnera sp. 2 — Camp 2: Lewis 880; Camp 3: 
Lewis 958. 

Gaertnera sp. 3 — Camp 1: Lewis (s 2). 

Gaertnera sp. A — Camp 4: Lewis 1092. 

Gaertnera sp. 5 — Camp 1: Lewis 786; Camp 2: 

Lewis 869; Camp 3: Lewis (s 322, 330). 
Gaertnera sp. 6 — Camp 1: Lewis 808. 

Gaertnera sp. 7 — Camp 4: Lewis (s 440), 1030, 
1084. 

Gaertnera sp. 8 — Camp 4: Lewis (s 489). 

Galium andringitrense Homolle ex Puff. 

Mapouria assimilis Bremek. (Andringitra, 1600 
m). 

Mussaenda arcuata Poir. — Camp 3: Lewis 969; 
Camp 4: Lewis 1078, 1094. 

Mussaenda sp. 1 — Camp 1: Lewis 791. 

Otiophora pauciftora Baker — 50 km S Ambala- 
vao, near abandoned meteo station above Am- 
balamarina, high altitude moss/lichen rain for- 



est-ericaceous bush, Philippia dominant, 1975 
m: Nicoll 263. 

Paederia andringitrensis Homolle. 

Pauridiantha paucinervis (Hiern) Bremek. ssp. 
lyallii (Baker) Verde. — Camp 2: Goodman 
6374, 6378; Lewis 851; Camp 3: Lewis 1001; 
Camp 4: Goodman 6443, 6454; 50 km S Am- 
balavao, near abandoned meteo station above 
Ambalamarina, high altitude moss/lichen rain 
forest-ericaceous bush, Philippia dominant, 
1975 m: Nicoll 299. 

Psychotria macrochlamys (Bremek) combined. — 
Camp 3: Lewis (s 290, 292). 

Psychotria sp. 1 — Camp 2: Lewis (s 172). 

Psychotria sp. 2 — Camp 1: Lewis (s 103); Camp 
2: Lewis (s 183). 

Psychotria sp. 3 — Camp 3: Lewis 934. 

Psychotria sp. A — Camp 4: Lewis (s 415), 1095. 

Psychotria sp. 5 — Camp 1: Lewis 751. 

Psychotria sp. 6 — Camp 2: Lewis (s 142). 

Saldinia sp. 1 — Camp 4: Lewis 1033, 1046. 

Schismatoclada sp. 1 — Camp 2: Lewis 852, 887. 

Schismatoclada sp. 2 — Camp 2: Lewis 932; 
Camp 3: Lewis (s 231, 285). 

Tricalysia cf. cryptocalyx Baker — Camp 3: Lewis 

991, 936. 
Tricalysia sp. 1 — Camp 4: Lewis (s 405). 
Tricalysia sp. 2 — Camp 1: Lewis (s 116); Camp 

4: Lewis (s 379). 



Rutaceae 

Evodia sp. 1 — Camp 3: Lewis (s 282). 

Evodia sp. 2 — Camp 4: Lewis (s 454). 

Evodia sp. 3 — Camp 4: Lewis (s 433). 

Teclea sp. 1 — Camp 3: Lewis 985, 995. 

Vepris fitoravina H. Perrier — Camp 1: Lewis 797; 
Camp 3: Lewis 998. 

Vepris sp. 1 — Camp 2: Lewis (s 164). 

Vepris sp. 2 — Camp 3: Lewis 1021. 



Santalaceae 

Thesium pseudocystoseiroides Cavaco & Kerau- 
dren (Massif de l'Andringitra, 2000-2500 m). 



70 



FIELDIANA: ZOOLOGY 



Sapindaceae 

sp. 1 — Camp 1: Lewis (s 49, 78); Camp 2: Lewis 

(s 166). 
sp. 2 — Camp 2: Lewis (s 197). 
sp. 3 — Camp 1: Lewis 832. 
sp. 4 — Camp 2: Lewis (s 152); Camp 3: Lewis 

940. 
Allophylus sp. 1 — Camp 2: Lewis 846. 
Doratoxylon sp. 1 — Camp 4: Lewis 1111. 
Macphersonia sp. 1 — Camp 3: Lewis (s 319). 
Tina sp. 1 — Camp 2: Lewis 910. 



Sapotaceae 

Chrysophyllum sp. 1 — Camp 1: Lewis (s 77). 
Faucheria sp. 1 — Camp 4: Lewis (s 430, 460). 
Labramia sp. 1 — Camp 1: Lewis (s 24). 



Scrophulariaceae 

Craterostigma perrieri Bonati — pres des eaux, 
vers 2000 m (Perrier de la Bathie, 1927). 

Halleria tetragona Baker — Sclerophyllous forest 
with canopy height 4-5 m, on E side Ampasi- 
potsy, above Camp 4, 1800-2000 m: Lewis 
1053. 

Hydrotriche mayacoides A. Raynal (Massif de 
l'Andringitra, mare, 1600 m). 

Ilysanthes longipes Bonati — pres des eaux, vers 
1200 m (Perrier de la Bathie, 1927). 

Sopubia trifida Buch.-Hamet — Open, level areas 
above village of Antanifotsy, partially dis- 
turbed, probably burned areas dominated by 
grasses, 1400 m: Lowry 4515. 



Solanaceae 

Solanum sp. 1 — Camp 4: Lewis 1083. 



Dombeya cf. amplifolia Arenes — Camp 3: Lewis 
(s 242, 320). 

Dombeya cf. spectabilis Bojer — Camp 2: Lewis 
(s 67, 156). 

Dombeya ivohibeensis Arenes — Camp 1: Lewis 
796. 

Dombeya leiomacrantha Hochr. ssp. leiomacran- 
tha var. angustata (Hochr.) Arenes (Massif de 
l'Andringitra, Iratsy, vallee de la Riambava et 
de T Antsifotra, 2000-2500 m). 

Dombeya leiomacrantha Hochr. ssp. leiomacran- 
tha var. dissecta (Hochr.) Arenes (Massif de 
l'Andringitra, 2000 m). 

Dombeya leiomacrantha Hochr. ssp. leiomacran- 
tha var. leiomacrantha (Massif de l'Andringi- 
tra, granite, 1600 m). 

Dombeya leucomacrantha Hochr. var. crassibrac- 
tea Hochr. (Massif de l'Andringitra, ravins, 
2400 m). 

Dombeya leucomacrantha Hochr. var. leucoma- 
crantha (Massif de l'Andringitra, 2400 m, 
brousse ericoide). 

Dombeya muscosa Hochr. (partie ouest de massif 
d'Andringitra, granite, 1600 m). 

Dombeya sp. 1 — Camp 4: Lewis 1042 (s 378). 

Dombeya sp. 2 — Camp 4: Lewis (s 445, 449, 457, 
488). 

Dombeya sp. 3 — Camp 2: Lewis (s 196). 

Dombeya sp. 4 — Camp 3: Lewis 324. 



Strelitziaceae 

Ravenala madagascariensis Sonn. — Camp 1: 
Lewis sight record; Camp 2: Lewis sight rec- 
ord. 



Theaceae 

Asteropeia rhopaloides (Baker) Baill. var. angus- 
tata H. Perrier (W du massif d'Andringitra). 



Sterculiaceae 

Byttneria melleri Baker — Camp 1: Lewis 814. 
Dombeya borraginopsis Hochr. ( W du massif de 
l'Andringitra, pres des torrents, 1200 m). 



Thymelaeaceae 

Peddiea involucrata Baker — Ericaceous vegeta- 
tion, on E side Ampasipotsy, above Camp 4, 
1800-2000 m: Lewis 1054; 50 km S Ambala- 
vao, near abandoned meteo station above Am- 



LEVVIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



71 



balamarina, high altitude moss/lichen rain for- 
est-ericaceous bush, Philippia dominant, 1975 
m: Nicoll 296. 



Rinorea urschii H. Perrier — Camp 1: Lewis 827. 

Viola abyssinica Steud. — "Foret a strate herbacee 
1695-1840 m" (Paulian et al., 1971, p. 254). 



Tiliaceae 

Grewia sp. 1 — Camp 2: Lewis (s 154). 
Grewia sp. 2 — Camp 3: Lewis 1020. 

Urticaceae 

Boehmeria sp. 1 — Camp 1: Lewis 758. 

Elatostema subfavoswn Leandri (Massif de l'An- 
dringitra, vers 1100 m). 

Elatostema sp. 1 — Camp 2: Lewis 865. 

Pilea rivularis Wedd. — 50 km S Ambalavao, near 
abandoned meteo station above Ambalamarina, 
high altitude moss/lichen rain forest-ericaceous 
bush, Philippia dominant, 1975 m: Nicoll 252. 

Urera sp. 1 — Camp 1: Lewis 754. 



Viscaceae 

Viscum humbertii Lecomte "Madagascar: Centre 
et Hautes Montagnes, Andringitra." 

Viscum sp. 1 — Camp 3: Lewis 943; Camp 4: Lew- 
is (s 469). 

Viscum sp. 2 — Camp 3: Lewis 938. 

Viscum sp. 3 — Camp 3: Lewis 1014. 

Viscum sp. A — Camp 4: Lewis 1055. 

Viscum sp. 5 — Camp 3: Lewis 980. 

Viscum sp. 6 — Ericaceous vegetation above the 
tree line, on E side Ampasipotsy, above Camp 
4, 1800-2000 m: Lewis 1050. 



Xyridaceae 



Velloziaceae 

Xerophyta dasylirioides Baker (van not deter- 
mined) — Camp 1: Lewis 772; "Vegetation rup- 
icole 1800-2000 m" (Paulian et al., 1971, p. 
237, 257); "Flore xerophile d'altitude 2300- 
2650 m" (Paulian et al., 1971, p. 247). 

Xerophyta dasylirioides Baker var. andringitren- 
sis H. Perrier (Andringitra Massif). 



Verbenaceae 

Clerodendrum sp. 1 — Camp 2: Lewis 915. 
Clerodendrum sp. 2 — Camp 4: Lewis (s 490). 
Vitex sp. 1 — Camp 3: Lewis 1016: Camp 4: Lewis 

1034 (s 462). 
Vitex sp. 2 — Camp 4: Lewis 1075. 



Violaceae 

Rinorea cf. arborea Baill. — Camp 1 : Lewis (s 36, 
58, 88, 93); Camp 2: Lewis (s 177, 187, 215). 
Rinorea sp. 1 — Camp 3: Lewis (s 263). 
Rinorea sp. 2 — Camp 4: Lewis (s 394). 



Xyris madagascariensis Malme — "Prairies alti- 
montaines 1700 m" (Paulian et al., 1971, p. 
236). 



Appendix 4-2. 

Vertebrate Food Plants on the Eastern 
Slopes of the RNI d'Andringitra 

Steven M. Goodman, Beverley A. Lewis, 
Peter B. Phillipson, and Jeannine Raharilala 

During our meanderings around the forest of 
the RNI d'Andringitra we had the occasion to ob- 
serve a wide variety of vertebrates feeding on var- 
ious plants. In most cases, specimens of these 
food plants were collected. The identifications of 
this material were confirmed by Phillipson and 
Raharilala. When collected, the field numbers of 
voucher specimens from the catalogs of S. M. 
Goodman (SMG) and B. Lewis (BL) are present- 
ed. Voucher specimens have been deposited with 
Missouri Botanical Garden (MO). 



72 



FIELDIANA: ZOOLOGY 



Annonaceae 

Artabotrys mabifolius Diels — At 720 m, fruits on 
ground eaten by Nesomys rufus (BL 844). 



Asteraceae 

Vernonia alleizettei Humbert var. rienanensis 
Humbert — At about 1300 m, Zosterops mad- 
eraspatana was observed picking at fruits 
(SMG 6402). 



cifolia" at 1650 m elevation — whole fruits on 
ground gnawed open by Nesomys rufus in order 
to extract seeds. 



Ericaceae 

Vaccinium secundiflorum Hook. — At 1625 m, 
Hypsipetes madagascariensis was seen eating 
green-colored fruits (SMG 6450). 



Euphorbiaceae 



Burseraceae 

Canarium madagascariense Engl. ssp. obtusifol- 
ium (Scott-Elliott) Leenhouts — At 720 m, sev- 
eral nuts were found below this tree that had 
been opened by Daubentonia madagascarien- 
sis. Also, nuts were regularly found on forest 
floor that had rodent gnaw marks. At 810 m, 
22 nuts were excavated from a burrow of Gym- 
nuromys roberti, 21 of which were gnawed 
open. Further, Canarium stones, mixed with 
Cryptocarya fruits, were found in a Nesomys 
rufus burrow. At this same elevation, nuts were 
regularly encountered on the forest floor that 
had been opened and consumed by Daubenton- 
ia madagascariensis and numerous species of 
rodents (see Chapter 6). 



Canellaceae 

Cinnamosma fragrans Baill. — At 1650 m, Cora- 
copsis nigra was observed feeding on fruits 
(BL 1077). They would remove exocarp from 
seed and consume the latter. 



Clusiaceae 

Symphonia sp. — Coracopsis nigra was observed 
feeding on unopen blossoms at 870 m in un- 
disturbed forest (no specimen). 



Macaranga oblongifolia Baill. — At 1650 m, Phi- 
lepitta castanea was seen feeding on small or- 
ange-colored buds (SMG 6458). 

Uapaca sp. — At 720 m, in slightly disturbed for- 
est, whole fruits were eaten by Eulemur fulvus. 



Lamiaceae 

Plectranthus sp. 1 — At 1210 m, Nectarinia soui- 
manga was seen visiting lavender-colored flow- 
ers (SMG 6408). 



Lauraceae 

Cryptocarya sp. — At 720 m and 810 m, whole 
ripe and unripe fruits were eaten by Eulemur 
fulvus. Animals would feed in canopy and on 
outer branches of tree, often filling their 
cheeks, and then rest in the same tree or near- 
by tree consuming fruits. Feces contained in- 
tact fruits minus exocarp. Fallen fruits were 
also eaten by Eliurus spp. and Nesomys rufus. 
Captive Eliurus preferred fruits with red exo- 
carps. Excavated burrows of Eliurus webbi at 
720 m and Nesomys rufus at 810 m contained 
quantities of cached and partially eaten fruits 
of this genus. Alectroenas madagascariensis 
was observed at 810 m feeding on whole fruits 
(no specimen). 



Elaeocarpaceae 

Sloanea rhodantha (Baker) Capuron var. rhodan- 
tha form "quadriloba" at 810 and form "quer- 



Loranthaceae 

Bakerella clavata (Desr.) Balle var. lenticellata 
(Baker) Balle — At about 1625 m, Neodrepanis 
hypoxantha, Nectarinia souimanga, and Zoster- 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 



73 



ops maderaspatana were seen feeding on ver- 
milion-colored flowers (SMG 6453 and BL 
1029). 

Bakarella clavata (Desr.) Balle var. peralata (Le- 
comte) Balle — At 720 m, Neodrepanis corus- 
cans and Zosterops maderaspatana and at 1210 
m, Nectarinia souimanga were observed visit- 
ing the dull reddish-pink flowers of this plant. 

Bakerella tandrokensis (Lecomte) Balle — At 
2050 m, Neodrepanis hypoxantha, Nectarinia 
souimanga, and N. notata were seen feeding on 
dull reddish-pink flowers (SMG 6459). 



Melastomataceae 



Myrsinaceae 

Maesa lanceolata Forssk. — At 740 m, fruits were 
eaten by Alectroenas madagascariensis. 

Oncostemum ankifiense Mez — At 1625 m, Zos- 
terops maderaspatana was seen plucking and 
eating flower buds (SMG 6430). 

Oncostemum racemiferum Mez — Tree near forest 
edge at 730 m; whole ripe fruits were con- 
sumed by Alectroenas madagascariensis and 
Hypsipetes madagascariensis. Neomixis tenella 
and Zosterops maderaspatana were also at- 
tracted to fruits, but it was not clear if they were 
actually eating fruits or insects attracted to them 
(SMG 6352). 



Gravesia calliantha Jum. & H. Perrier — At 1210 
m, Philepitta castanea was seen feeding on 
fruits (SMG 6409). 

Medinilla ericarum Jum. & H. Perrier — At 1210 
m, Hypsipetes madagascariensis and Philepitta 
castanea were seen feeding on orangish-red 
fruits (SMG 6407). 

Medinilla torrentum Jum. & H. Perrier — At about 
1650 m, Nectarinia souimanga was seen visit- 
ing white-colored flowers and apparently pierc- 
ing unopened flowers with its bill (SMG 6429). 



Monimiaceae 



Myrtaceae 

Syzygium sp. 1 3 — Whole ripe fruits were eaten by 
Eulemur fulvus, at about 710 m, at ecotone be- 
tween forest edge and tavy (BL 776). 



Orchidaceae 

Liparis sp. 1 — At 810 m, Neodrepanis coruscans 
and Nectarinia souimanga were seen visiting 
flowers (SMG 6381). 



Tambourissa cf. parvifolia Baker — At 1240 m, 
whole fruits were eaten by Eulemur fulvus (BL 
964). 



Moraceae 



Pittosporaceae 

Pittosporum sp. — At 800 m, whole fruits were 
eaten by Philepitta castanea (no specimen). 



Ficus brachyclada Barker — Fruits were eaten by 
birds at 810 m (BL 841). 

Ficus reflexa Thunb. — Whole fruits were eaten by 
Alectroenas madagascariensis and Streptopelia 
picturata at edge of clearing at 710 m (SMG 
6353). 

Ficus rubra Vahl — Fruits were eaten by birds at 
810 m (BL 866). 

Ficus trichopoda Baker — Fruits were eaten by 
birds at 720 m (BL 769). 

Streblus dimepate (Bureau) C.C. Berg — At 730 
m, whole fruits were eaten by Alectroenas mad- 
agascariensis in slightly disturbed forest. 



Poaceae 

Nastus sp. — At 810, 1250, and 1700 m, bamboos 
were found that had been gnawed open by Dau- 
bentonia madagascariensis, presumbly to ex- 
tract insect larvae (no specimen). 



Podocarpaceae 

Podocarpus madagascariensis Baker — At 810 m, 
Eulemur fulvus was observed eating fruits (no 
specimen). 



74 



FIELDIANA: ZOOLOGY 



Rubiaceae Pauridiantha paucinervis (Hiern) Bremek. ssp. 

lyallii (Baker) Verde. — At 810 m, small red 

Chassalia sp. 1 — At 1210 m, Coracopsis nigra, fruits were eaten by Philepitta castanea, Hyp- 

Philepitta castanea, and Hypsipetes madagas- sipetes madagascariensis, and Zosterops mad- 

cariensis were eating whole bright red fruits eraspatana (SMG 6374, 6378), and at 1625 m, 

(SMG 6421). by Philepitta castanea (SMG 6454). 



LEWIS ET AL.: BOTANICAL STRUCTURE, COMPOSITION, AND DIVERSITY 75 



Chapter 5 

The Pteridophytes of the Eastern Slopes of the 
Reserve Naturelle Integrate d'Andringitra, 
Madagascar 

France Rakotondrainibe and Fidele Raharimalala 



Abstract 

An elevational survey of the pteridophytes occurring along the eastern slopes of the Reserve 
Naturelle Integrale d'Andringitra was conducted. A total of 188 species were recorded. Within 
the transect zones, 76 species were collected at 720 m, 115 at 800-900 m, 96 at 1200-1300 
m, and 74 at 1600-1700 m. On the basis of species turnover of ferns, the transition zone 
between lowland and montane forest is between 700 and 900 m, and the transition between 
montane and sclerophyllous forest is between 1550 and 1600 m. 

Resume 

Une etude sur la repartition altitudinale de Pteridophytes a ete menee sur le versant Est de 
la Reserve Naturelle Integrale d'Andringitra. 188 especes au total ont ete recensees. Le long 
du transect altitudinal, 76 especes ont ete observees a 720 m, 1 1 5 entre 800 et 900 m, 96 entre 
1200 et 1300 m et 74 entre 1600 et 1700 m. La composition floristique des communautes de 
fougeres permet de situer la zone de transition entre la foret de basse altitude et la foret de 
montagne a 700-900 m et celle entre la foret de montagne et la foret sclerophylle a 1500- 
1600 m. 



Introduction 



Methods 



Our current knowledge of the taxonomy of 
Malagasy pteridophytes and their distribution is 
largely based on the work of Christensen (1932), 
Perrier de la Bathie (1932), and Tardieu-Blot 
(1951-1971). Using this information, Tardieu- 
Blot (1948), Leroy (1978), and Jacobsen (1991) 
discussed the origin and the affinities of this group 
of plants. In recent years, under the direction of 
biological inventories associated with conserva- 
tion projects, a considerable amount of new in- 
formation has been gathered on the island's pte- 
ridophytes. This new material advances our 
knowledge of this flora. 



Between 13 May and 7 June 1995 we con- 
ducted a survey of the pteridophytes of the Re- 
serve Naturelle Integrale (RNI) d'Andringitra, us- 
ing the same transect as the late 1993 mission (see 
Chapter 1). The 720 m zone was occupied be- 
tween 13 and 16 May, the 810 m zone between 
17 and 22 May, the 1210 m zone between 23 and 
30 May, and the 1625 m zone between 31 May 
and 5 June. Further, general collections were 
made in secondary grassland and disturbed forest 
between 700 and 1 150 m. Voucher specimens are 
deposited in the Museum National d'Histoire Na- 
turelle, Paris (P), and FOFIFA, Antananarivo 
(TEF). 



76 



FIELDIANA: ZOOLOGY 



The collections made represent the general pte- 
ridophyte flora of the eastern slopes of the re- 
serve. The majority of plants were terrestrial and/ 
or epiphytic (between 0.5 and 8 m above the 
ground). A few species, approximately 20, were 
found between 15 and 20 m above the ground. A 
considerable portion of the information on epi- 
phytic canopy pteridophytes was obtained from 
fallen emergent trees. Further, canopy plants were 
identified with binoculars. 

The current nomenclature for the Cyatheaceae 
is not parallel between continents and is in need 
of major revision. Thus, we have chosen to utilize 
a somewhat out-of-date system that reflects the 
standard taxonomy for Madagascar. 



Results and Discussion 

Table 5-1 summarizes the pteridophyte inven- 
tory in an altitudinal range of 700-1700 m on the 
eastern slopes of the RNI d'Andringitra. Several 
vegetation types of the Central Domain (for phy- 
togeographical division of Madagascar and forest 
nomenclature, see Humbert, 1955, and White, 
1983) were studied: (1) disturbed forests and sec- 
ondary grasslands located outside of the reserve 
(columns I, II, and III of Table 5-1), and (2) the 
mid-montane moist forest within the reserve and 
its two ecotones between 720 and 900 m with 
lowland forest and between 1500 and 1600 m 
with the sclerophyllous forest (column IV of Ta- 
ble 5-1). 

A total of 188 species of pteridophytes were 
found, among which 70 (37.2%) are endemic to 
Madagascar. Prior to the present study, 61 taxa of 
pteridophytes had been reported from the Andrin- 
gitra Massif (Tardieu-Blot, 1951-1971). This fig- 
ure, when compared to those of other reserves on 
the island, indicates a very high species richness, 
stemming from the diversity of vegetation in the 
area. For example, Rahajasoa (1993) identified 
155 taxa of pteridophytes in the Pare National 
(PN) de Ranomafana between 900 and 1200 m, 
Rakotondrainibe and Quansah (1994) found 169 
taxa in the Reserve Sp6ciale de Manongarivo be- 
tween 200 and 1 876 m, and Rakotondrainibe and 
Raharimalala (in press) identified 156 taxa along 
the western slopes of the Masoala Peninsula be- 
tween and 1 105 m. 

The disturbed forests and secondary grasslands 
outside the reserve boundaries still contain a high 
species diversity (105 taxa) in comparison to that 



within the protected area (164 taxa). However, the 
floristic composition of the endemic pteridophytes 
in the disturbed forests is 35.2%, compared to 
39.6% within the reserve. In the undisturbed for- 
est, the species richness gradually decreases with 
increasing altitude: 1 15 taxa are found at 800-900 
m, 96 taxa at 1200-1300 m, and 74 taxa at 1600- 
1700 m. 

The most characteristic species of the following 
altitudinal zones in the RNI d'Andringitra are: up 
to 900 m — Asplenium bipartitum, A. dregeanum, 
Christella distans, Ctenopteris excaudata, Micro- 
lepia madagascariensis, Pteris catoptera, Tectar- 
ia magnified, Trichomanes bipunctatum, and T. 
rotundifolium; 800-1300 m — Asplenium preusii, 
Elaphoglossum scolopendriforme, Gymnosphaera 
boivinii, and Gymnosphaera coursii; 1300-1700 
m — Elaphoglossum subsessile, Grammitis holo- 
phlebia, and Rumohra lokohensis; and 1600-1700 
m — Blechnum humbertii, Dryopteris remotipin- 
nula, Elaphoglossum aubertii, and Hymenophyl- 
lum peltatum. The ecotone of the lowland and 
montane forest occurs between 700 and 900 m. 
This transition is easily defined by a group of low- 
land ferns whose upper elevational limit is at 
about 900 m (Nephrolepis biserrata, Phymatoso- 
rus scolopendria, and Platycerium madagascar- 
iense) and a group of montane species whose low- 
er limit appears to be at approximately 900 m (As- 
plenium anisophyllum, Elaphoglossum forsythii- 
majoris, Oleandra distenta, and Trichomanes 
speciosum). The transition between montane and 
sclerophyllous forest is well demarcated between 
1500 and 1600 m by both abundance and distri- 
bution of several species (e.g., Blechnum humber- 
tii, Dryopteris remotipinnula, Elaphoglossum au- 
bertii, and Hymenophyllum peltatum). Indepen- 
dent of elevation, the distribution of Elaphoglos- 
sum spathulatum, E. petiolatum var. salicifolium, 
and Xiphopteris hildebrandtii is closely correlated 
with the microhabitat of rocks along riverbanks. 

The presence and/or abundance of the follow- 
ing taxa indicate some human or natural distur- 
bance: on the edge of the forest — Lycopodium 
cernuum, L. clavatum, Blechnum punctulatum, 
and Mohria cqffrorum; in clearings surrounded by 
forest — Hypolepis villoso-viscida and Pteris pseu- 
dolonchitis; in the secondary grasslands succeed- 
ing forest fires — Pteridium aquilinum, Sticherus 
flagellaris, and Dicranopteris linearis; and in dis- 
turbed forests, particularly along trails — Amauro- 
pelta bergiana, Ctenopteris villosissima, Cyathea 
melleri, Elaphoglossum deckenii var. rufidulum, 
and Selaginella lyallii. Several taxa of pterido- 



RAKOTONDRAINIBE & RAHARIMALALA: PTERIDOPHYTES OF EASTERN SLOPES 






77 



Table 5-1. The pteridophytes of the Andringitra Massif: elevational distribution and fioristic composition of 
populations in disturbed (columns I, II, and III) and undisturbed forest (column IV). 



Taxa 



Vegetation Types: I and II III 
Altitude (m): 700-1150 720 



rv rv iv 

800-900 1200-1300 1600-1700 



Amauropelta aff. oppositiformis (C.Chr.) Holttum 
Amauropelta aff. strigosa (Willd.) Holttum 
Amauropelta bergiana (Schltr.) Holttum 
Anthrophyum boryanum (Willd.) Kaulf. 
Antigramma virchowii (E. Kuhn) Tardieu 
Arthropteris monocarpa (Cordem.) C.Chr. 
Arthropteris parallela C.Chr. 
Asplenium aethiopicum (Burm.f.) Bech. 
Asplenium anisophyllum Kunze 
Asplenium auritum Sw. 
Asplenium bipartitum Bory 
Asplenium cancellatum Alston 
Asplenium cuneatum Lam. 
Asplenium dregeanum Kunze 
Asplenium rectum var. zeyheri 

(Pappe & Rawon) Alston & Schelpe 
Asplenium erectum Willd. var. erectum 
Asplenium friesiorum C.Chr. 
Asplenium herpetopteris Baker 

var. herpetopteris 
Asplenium herpetopteris 

var. massoulae (Bonap.) Tardieu 
Asplenium inaequilaterale Willd. 
Asplenium lividum E. Kuhn 
Asplenium mannii Hook. 
Asplenium nidus L. 
Asplenium normale D. Don 
Asplenium obscurum Blume 
Asplenium pellucidum Lam. 
Asplenium petiolulatum Mett. 
Asplenium poolii Baker 
Asplenium preusii Brause 
Asplenium protensum Schrad. 
Asplenium rutifolium (P.J.Bergius.) Kunze 
Asplenium sandersonii Hook. 
Asplenium theciferum (Kunth) Mett. 
Athyrium scandicinum (Willd.) C.Presl. 
Belvisia spicata (L.f.) Mirb. 
Blechnum attenuatum (Sw.) Mett. 
Blechnum attenuatum var. giganteum (Kaulf.) Bonap. 
Blechnum humbertii Tardieu 
Blechnum ivohibense C.Chr. 
Blechnum punctulatum Sw. 
Blechnum simillimum (Baker) Diels 

fa binerve (Hook.) Tardieu 
Blechnum simillimum (Baker) Tardieu 

var. simillimum 
Blechnum simillimum 

var. xiphophyllum (Baker) Tardieu 
Blotiella madagascariensis (Hook.) Tard. 
Cheilanthes madagascariensis Baker 
Christella dentta (Forssk.) Holttum 
Christella distans (Hook.) Tardieu 



** 
** 





o 


* 




** 




*** 






** 


*** 




o 







*** 


** 


o 


** 


**** 


o 
** 


* 


* 


*** 



*** 


*** 




** 




* 


o 


* 


*** 


*** 



** 



* 


** 




*** 


*** 


*** 


** 


** 


** 



** 


** 








o 


o 




* 


**# 





* 




*#* 


*#** 


**** 



** 
** 



78 



FIELDIANA: ZOOLOGV 



Table 5-1. Continued. 



Vegetation Types: I and II III IV IV IV 

Taxa Altitude (m): 700-1150 720 800-900 1200-1300 1600-1700 

Coniogramme madagascariensis C.Chr. * *** ** 

Ctenitis lanuginosa (Kaulf.) Copel. o 

Ctenitis magma (Baker) Tardieu * *** * 

Olenitis poolii (C.Chr.) Tardieu * 

Ctenopteris devoluta (Baker) Tardieu * * * * 

Ctenopteris elastica (Willd.) Copel. ** ** O 

Ctenopteris excaudata (Bonap.) Tardieu * ** *** o 

Ctenopteris rigescens (Bory) J.Sm. ** 

Ctenopteris villosissima (Hook.) Harley *** * * * 

Cyathea approximata Bonap. * ** 

Cyathea bellisqamata Bonap. ** 

Cyathea boivinii E.Kuhn * * 

Cyathea borbonica Desv. 

var. laevigata Bonap. * ** 

Cyathea decrescens E.Kuhn * *** 

Cyathea melleri (Baker) Domin *** * 

Cyathea melleri (Baker) Tardieu 

var. virescens Tardieu o 

Cyathea similis C.Chr. ** o 

Cyathea tsaratananensis Tardieu * *** *** *** ** 

Deparia bory ana (Willd.) Kato •*• *** 

Deparia parvisora (C.Chr.) Kato *** 

Dicranopteris linearis (Burm.f.) Underw. ** 

Diplazium zakamenense (Tardieu) Rakotondr. * * * 

Dryopteris mangindranensis Tard. * * * * o 

Dryopteris remotipinnula Bonap. *** 

Dryopteris subcrenulata (Baker) C.Chr. o 

Elaphoglossum achroalepis (Baker) C.Chr. ** 

Elaphoglossum acrostichoides (Hook.&Grev.) *** 

Elaphoglossum aff. ovalilimbatum Bonap. * *** * 

Elaphoglossum scolopendriforme Tardieu *** **** *** 

Elaphoglossum angulatum (Blume) T.Moore * 

Elaphoglossum aubertii (Desv.) T.Moore o *** 

Elaphoglossum coriaceum Bonap. o ** 

Elaphoglossum decaryanum Tardieu o ** 

Elaphoglossum deckenii (Kuhn) C.Chr. 

var. deckenii * 

Elaphoglossum deckenii var. rufidulum (Willd.) Tardieu *** ** * 

Elaphoglossum forsythii-majoris H.Christ. o *** 

Elaphoglossum humbertii C.Chr. ** o 

Elaphoglossum hybridum (Bory) Brack. ** *** o o 

Elaphoglossum leucolepis (Baker) Krajina * * o 

Elaphoglossum petiolatum (Sw.) J.B.Urb. 

ssp. salicifolium (Kaulf.) Schelpe o ** 

Elaphoglossum sieberi (Hook.&Grev.) T.Moore * ** o 

Elaphoglossum spathulatum (Bory) T.Moore *** 

Elaphoglossum sp. 6 O * ** 

Elaphoglossum sp. 7 * ** 

Elaphoglossum sp. 8 * *** 

Elaphoglossum subsessile (Willd.) Christ. ** *** *** 

Grammitis barbatula (Baker) Copel. o ** 

Grammitis cryptophlebia (Baker) Copel. * o 

Grammitis holophlebia (Baker) Copel. ** *** 

Gymnosphaera andohahelensis (Tardieu) Tardieu *** 



RAKOTONDRAINIBE & RAHARIMALALA: PTERIDOPHYTES OF EASTERN SLOPES 79 



Table 5-1. Continued. 



Taxa 



Vegetation Types: 
Altitude (m): 



I and II 
700-1150 



III 

720 



rv 

800-900 



rv iv 

1200-1300 1600-1700 



Gymnosphaera boivinii (Ettingsh.) Tardieu 
Gymnosphaera coursii (Tardieu) Tardieu 
Gymnosphaera sp. 1 (aff. boivinii) 
Histiopteris incisa (Thunb.) J.Sm. 
Huperzia megastachya (Baker) Tardieu 
Huperzia obtusifolia (Sw.) Rothm. 
Huperzia pecten (Baker) Tardieu 
Huperzia pichiana Tardieu 
Huperzia rubrica (Herter) Tardieu 
Huperzia sqarrosa (Forssk.) Trevis 
Huperzia verticillata (L.f.) Trevis 
Hymenophyllum capense Schrad. 
Hymenophyllum hirsutum (L.) Sw. 
Hymenophyllum humbertii C.Chr. 
Hymenophyllum peltatum (Poir.) Desv. 
Hymenophyllum perrierii Tardieu 
Hymenophyllum polyanthos (Sw.) Sw. 
Hymenophyllum polyanthos 

var. kuhnii (C.Chr.) Schelpe 
Hymenophyllum poolii Baker 
Hymenophyllum sibthorpioidews Mett. 
Hymenophyllum veronicoides C.Chr. 
Hypolepis villoso-viscida (A. Thouars) Tardieu 
Lindsaea flabellifolia (Baker) Kuhn 
Lindsaea madagascariensis Baker 
Lindsaea odorata Roxb. 
Lomariopsis pollicina (P. Willemet) Mett. 
Lonchitis occidentalis Baker 
Loxogramme lanceolata (Sw.) C.Presl. 
Lycopodium carolinianum L. 

var. affine (Bory) Pic.Serm. 
Lycopodium cernuum L. 
Lycopodium clavatum L. 

var. borbonicum Bory 
Lycopodium zanclophyllum Wilce 
Lygodium lanceolatum Desv. 
Marattia fraxinea J.F.Gmel 
Microlepia madagascariensis C.Presl. 
Microsorum punctatum (L.) Copel. 
Mohria caffrorum (L.) Desv. 
Nephrolepis biserrata (Sw.) Schott 
Nephrolepis undulata (Sw.) J.Sm. 
Odontosoria melleri (Hook.) C.Chr. 
Oleandra distenta Kunze 
Ophioglossum palmatum L. 
Osmunda regalis L. 
Pellaea angulosa (Bory) Baker 
Phymatosorus scolopendria (Burm.f.) Pic.Serm. 
Pityrogramma argentea (Willd.) Domin. 
Platycerium madagascariense Baker 
Pleopeltis excavata (Willd.) Sledge 
Pleopeltis macrocarpa (Willd.) Kaulf. 
Pneumatopteris subpennigera 

(C.Chr.) Holttum 



*** 


** 


** 


*** 


** 


*** 


* 




* 


* 


o 











o 









o 


o 








* 
o 


*** 


*** 


** 


* 


*** 


*** 


**** 


* 




* 


** 


*** 
*** 







* 




* 


** 


*** 


**** 






o 


* 


** 


** 


*** 


*** 



o 
** 
* 
o 
o 



** 






** 






*** 













*** 


*** 


** 


o 





* 




* 


*** 


*** 


*#* 


*** 


* 


*** 




*** 


*** 


*** 



o 



*** 

o 



80 



FIELDIANA: ZOOLOGY 



Table 5-1. Continued. 



Taxa 



Vegetation Types: 
Altitude (m): 



I and II 
700-1150 



III 

720 



rv 

800-900 



iv rv 

1200-1300 1600-1700 



Pneumatopteris unita Kunze 

Polystichum coursii Tardieu 

Pseudophegopteris cruciata (Willd.) Holttum 

Pseudophegopteris pulcher (Willd.) Holttum 

Pteridium aquilinum (L.) E.Kuhn 

Pteris catoptera Kunze 

Pteris cretica L. 

Pteris griseoviridis C.Chr. 

Pteris humbertii C.Chr. 

Pteris lane at folia J.Agardh 

Pteris pseudolonchitis Kuhn 

Rumohra adiantiformis (G. Forst.) Ching 

Rumohra lokohensis (Forssk.) Ching 

Saccoloma henriettae C.Chr. 

Schizaea dichotoma (L.) Sw. 

Selaginella fissidentoides Spring 

Selaginella goudotana Spring 

Selaginella lyallii Baker 

Selaginella pectinata (Willd.) Spring 

Sphaerostephanos arbuscula (Willd.) Holttum 

Sticherus flagellaris (Bory) St John 

Tectaria magnified (Bonap.) Tardieu 

Trichomanes bipunctatum Poir. 

Trichomanes bonapartei C.Chr. 

Trichomanes borbonicum Bosch 

Trichomanes cupressoides Desv. 

Trichomanes digitatum Sw. 

Trichomanes lenormandii Bosch 

Trichomanes longilabiatum Bonap. 

Trichomanes mannii Hook. 

Trichomanes meifolium Willd. 

Trichomanes montanum Hook. 

Trichomanes rigidum Sw. 

Trichomanes rotundifolium Bonap. 

Trichomanes speciosum Willd. 

Mi ta ha humblotii Hieron. 

Vittaria isoetifolia Bory 

Xiphopteris hildebrandtii (Hieron.) Tardieu 

Xiphopteris oosora (Baker) Alston 

var. micropecten C.Chr. 
Xiphopteris serrulata (Sw.) Kaulf. 
Xiphopteris sp. 2 



*** 


*** 


** 


• 


* 









*** 




* 


*** 


o 


** 


*** 


** 






*** 


**** 


*** 


*** 


** 




* 


*** 


* 




o 


*** 






*** 


** 







** 


*** 






*** 


*** 
o 






*** 


*** 


*** 


* 


* 


* 






* 




* 






* 


* 


*** 




* 








* 








o 








* 


•* 




* 


* 
*** 


** 




* 


*** 









** 


*** 


*** 
o 



Total number of taxa: 


188 




52 


76 


115 


96 


74 








105 






164 






Number of endemic taxa: 
Percentage: 


70 

37.52% 




37 
35.2% 






65 
39.6% 






I, savannah; II, relict forest; III, disturbed forest; 
that it is endemic to Madagascar; a taxon printed i 
southern part of the Central Domain of Malagasy. 
oRare. 
* Infrequent. 
** Frequent. 
*** Common. 
**** Very common. 


IV, undisturbed forest. A taxon printed in boldface type indicates 
in boldface type and underlined indicates that it is endemic to the 


RAKOTONDRAINIBE & RAHARIMALALA: PTERIDOPHYTES OF EASTERN SLOPES 




81 






phytes occur only in primary or slightly disturbed 
forest: Anthrophyum boryanum, Asplenium man- 
nii, A. anisophyllum, Blechnum ivohibense, and 
Trichomanes rotundifolium. 

Further survey work and collections are needed 
to develop our understanding of the relationships 
and altitudinal distribution of the Malagasy pte- 
ridophyte flora. Techniques are being developed 
to quantify measures of abundance along such 
gradients; these techniques will allow compari- 
sons between sites. Several students are currently 
writing theses associated with taxonomic revi- 
sions of this flora. 



Acknowledgments 

We are grateful to the staff of the World Wide 
Fund for Nature, particularly Peter Schachenmann 
and Hanta Rabetaliana, and the people of Amba- 
lamenjana for making our mission possible. The 
Laboratoire de Phytomorphologie of the Ecole 
Pratique des Hautes Etudes, under the direction of 
Dr. A. Le Thomas, provided partial financial sup- 
port for this project. 



Literature Cited 

Christensen, C. 1932. The pteridophyta of Madagas- 
car. Dansk Botanisk Arkiv, 80: 1-253. 
Humbert, H. 1955. Les territoires phytogeographiques 



de Madagascar. In Les divisions Fcologiques du mon- 
de. Colloques Internationaux du Centre National de 
Recherche Scientifique LIX. Annee biologique, se>. 3, 
31(5-6): 439-448. 

Jacob-sen, W. B. G. 1991. Phytogeographical analysis 
of the pteridophytes of some randomly selected areas 
of the world. Thaiszia, 1: 69-94. 

Leroy, J.-F. 1978. Composition, origin, and affinities of 
the Madagascan vascular flora. Annals of the Missouri 
Botanical Garden, 65: 535-589. 

Perrier de la Bathie, H. 1932. Distribution dans rile, 
pp. 207-221. In Christensen, C, ed. The pteridophyta 
of Madagascar. Dansk Botanisk Arkiv, 80: 1-253. 

Rahajasoa, G. M. 1993. Analyse de la Diversite des 
Pteridophytes dans le Pare national de Ranomafana: 
Inventaire et mode de repartition. Diplome d' Etudes 
Approfondies de Sciences Biologiques, option Ecol- 
ogie V6getale, Universite d'Antananarivo. 

Rakotondrainibe, E, and N. Quansah. 1994. The di- 
versity and originality of the pteridophytes of the for- 
est of Manongarivo Special Reserve (north-west Mad- 
agascar). Fern Gazette, 14: 259-267. 

Rakotondrainibe, E, and Raharimalala, E In press. 
Contribution a 1' etude de la flore des aires protegees 
de Madagascar: Les Pteridophytes de la presqu'ile 
Masoala. Centre d' Information et de Documentation 
Scientifique et Technique, Antananarivo, Serie Sci- 
ences biologiques. 

Tardieu-Blot, M. L. 1948. Le peuplement ptendolo- 
gique de Madagascar. Memoire de l'lnstitut Scienti- 
fique de Madagascar, sen B, 6: 219-243. 

Tardieu-Blot, M. L. 1951-1971. Les Pteridophytes. In 
Humbert, H., ed. Flore de Madagascar et des Co- 
mores, Fam. 1 a 13, Museum National d'Histoire Na- 
turelle, Paris, France. 

White, E 1983. The vegetation of Africa. UNESCO/ 
AETFAT/UNSO, Paris, 356 pp. 



82 



FIELDIANA: ZOOLOGY 



Chapter 6 

The Utilization of Canarium (Burseraceae) 
Seeds by Vertebrates in the Reserve 
Naturelle Integrale d'Andringitra, Madagascar 

Steven M. Goodman and Eleanor J. Sterling 






Abstract 

The aye-aye (Daubentonia madagascariensis) is known to feed extensively on the fat-rich 
seeds of Canarium, which are encased in an extremely hard endocarp. Endemic Nesomyinae 
rodents and introduced Rattus rattus captured in the forests of Reserve Naturelle Integrale 
d'Andringitra were held in captivity and fed Canarium seed capsules to see if they were able 
to penetrate the endocarp and consume the cotyledon. Five of the eight endemic rodents tested, 
as well as Rattus, fed on the seeds in captivity. Information is presented on size and pattern 
of feeding signs on Canarium seed capsules recovered from the forest floor and rodent burrows. 
Further, we review information on the utilization of Canarium fruits and seeds by other ver- 
tebrates and on possible competitive interactions by these animals for this food resource. 

Resume 

II est connu que 1' Aye-aye {Daubentonia madagascariensis) se nourrit en grande partie de 
graines de Canarium riches en lipides, qui sont enveloppees par un endocarpe tres dur. Dans 
les forets de la Reserve Naturelle Integrate d'Andringitra, nous avons capture des Nesomyinae, 
rongeurs end^miques, et Rattus rattus, espece introduite. Ces especes ont €t€ gardens en cap- 
tivity et ont €i€ nourries avec des fruits de Canarium afin de tester s'ils pouvaient percer 
1' endocarpe et consommer la graine. Sur les huit especes de rongeurs endemiques prenant part 
au test, cinq especes endemiques et Rattus rattus se sont nourris des graines lorsqu'ils etaient 
en captivite. On pr6sente ici des informations sur la taille et sur la fa?on de manger ces fruits 
de Canarium retrouves sur le sol et dans les terriers des rongeurs. De plus, nous passons en 
revue les informations relatives a l'utilisation des fruits et des graines de Canarium par d'autres 
especes de vertdbrds et pr6sentons de possibles interactions comp£titives entre ces animaux 
pour l'acces a cette ressource alimentaire. 



Introduction aye-aye {Daubentonia madagascariensis) is 

closely related to that of C. madagascariense 

Canarium madagascariense (Burseraceae) is a (Iwano & Iwakawa, 1988). Recent fieldwork on 

common tree at low elevations in the eastern hu- the Reserve Sp6ciale de Nosy Mangabe, an island 

mid forests of Madagascar (White, 1 983). Its dru- south of Maroantsetra, has shown that aye-ayes 

paceous fruits provide an important food resource spend from 30% to 89% of their foraging time 

for numerous vertebrates (Sterling, 1994). In fact, feeding on the seeds of Canarium (Sterling, 

it has been proposed that the distribution of the 1994). These fruits have a fleshy pericarp and an 

GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 83 



extremely hard three-loculed endocarp, the latter 
often surpassing 250 kg/mm 2 crushing resistance 
with a spring tester gauge (Sterling, 1994). Their 
nutritious cotyledon contains up to 60% fat, in 
addition to numerous minerals and fat-soluble vi- 
tamins (Sterling et al., 1994). In short, if an ani- 
mal has the ability to get through the endocarp of 
the seed, a valuable food resource lies within. 

In the Reserve Naturelle Intdgrale (RNI) 
d'Andringitra, where Canarium madagascariense 
Engl. ssp. obtusifolium (Scott Elliott) Leenhouts 
has been identified at altitudes of up to 1000 m 
(Chapter 4), the distinctive seed capsules of this 
tree (Fig. 6-1) were regularly encountered on the 
forest floor. Seed capsules were most often con- 
centrated under Canarium trees, but single ex- 
amples or clusters were found regularly on the 
ground a considerable distance from Canarium 
trees. In order to investigate the utilization of 
these seeds by the various mammalian vertebrates 
occurring in this forest, we collected random sam- 
ples of seed capsules found on the ground. Fur- 
ther, as part of the small mammal study (Chapter 
22), a wide variety of rodents were captured alive 
in mammal traps, and we presented whole Can- 
arium seed capsules to them in order to determine 
if they were able to penetrate the hard endocarp, 
if they left distinctive gnawing marks on the en- 
docarps, and if they ate the cotyledons. Finally, 
numerous rodent burrow systems were excavated, 
and cached seed capsules and fruits, including 
Canarium, were recovered. 

In this chapter we ask three questions: (1) What 
vertebrates in the RNI d'Andringitra feed on Can- 
arium fruits and seeds? (2) Do they leave distinc- 
tive gnawing marks? (3) Is there any evidence of 
differential utilization of this resource that might 
indicate competition between the different ani- 
mals feeding on the seeds? 



capsules of C. madagascariense and boivini are 
not distinguishable from one another (Perrier de 
la Bathie, 1946), for the purposes of this paper it 
is simpler to follow the taxonomy of Leenhouts 
(1959). 

Samples of Canarium seed capsules were 
picked up under trees using a quadrate method. 
Three 1-m 2 squares were placed at preset distanc- 
es (1 m east, 4 m west, and 2 m south) from the 
base of single trees at 720 and 810 m elevation. 
All seed capsules within the squares were picked 
up for later sorting and measuring. Also, several 
collections were made along trails and scattered 
localities in the forest. 

Techniques associated with small mammal trap- 
ping are described in Chapter 22. In each eleva- 
tional zone, live animals were housed in National 
traps in secluded areas covered by a tarpaulin at 
the periphery of the camp. Generally, two or three 
whole seed capsules of C. madagascariense, sup- 
plemented by wet cooked rice, were placed in 
each trap. Individual animals were never kept in 
captivity for more than 48 hours. Canarium seed 
capsules were transported to transect zones above 
the elevational limit of this tree; otherwise all 
seed capsules fed to captive animals were from 
the respective altitudinal zone. 

When seemingly active rodent ground burrows 
were encountered, traps were placed at the en- 
trance, and those burrows that yielded animals 
were subsequently excavated and searched for 
cached seed capsules or food remains. Opened 
seed capsules found on the ground or in burrows 
were divided into three classes according to how 
they were opened: aye-aye, rodent, and natural 
dehiscence (see Fig. 6-1). Measurements of Can- 
arium seed capsules, taken with dial calipers to 
the nearest 0.1 mm, include maximum length 
along the main axis and maximum width across 
the broadest portion of the midsection. 



Methods 

In the most recent revision of Malagasy Can- 
arium, Leenhouts (1959) recognized a single spe- 
cies, C. madagascariense. In earlier treatments of 
this group various numbers of species had been 
considered valid. For example, Perrier de la Ba- 
thie (1946) recognized two distinct taxa: C. mad- 
agascariense, a large leaf form, and C. boivini, a 
smaller leaf form. Recent field-workers recognize 
at least two species on the island (Andrianisa, 
1989; G. Schatz, pers. comm.). Because the seed 



Results and Discussion 

Feeding Experiments 

A total of nine species of captive rodents, in- 
cluding introduced Rattus rattus and eight species 
of endemic Nesomyinae, were fed the seed cap- 
sules of Canarium madagascariense (Table 6-1). 
Six of these rodents consumed Canarium seeds 
by gnawing a small hole in the central portion of 
the seed capsule and then extracting the cotyle- 



84 



FIELDIANA: ZOOLOGY 



Table 6-1. Results of feeding experiments of captive rodents in the RNI d'Andringitra. 









Gnaw 












marks, 












longitu- 






Species 


Mean mass 


Fed on seed 


dinal- 


Trapped on 


Trapped off 


(number of individuals tested) 


(g)* 


in captivity 


central 


ground 


ground 


Rattus rattus (3) 


103.4(11) 


Yes 


Yes 


Yes 


Yes 


Brachyuromys ramirohitra^ ( 1 ) 


117 (1) 


Yes 


Yes 


Yes 


No 


Eliurus majorif (2) 


96.6 (20) 


No 


— 


Yes 


Yes 


Eliurus minor (3) 


36.9 (25) 


No 


— 


Yes 


Yes 


Eliurus tanala (2) 


81.7(23) 


Yes 


Yes 


Yes 


Yes 


Eliurus webbi (2) 


72.3 (22) 


Yes 


Yes 


Yes 


Yes 


Gymnuromys roberti (2) 


119.3(3) 


Yes 


Yes 


Yes 


No 


Monticolomys koopmani\ ( 1 ) 


26.2 (3) 


No 


— 


Yes* 


Yes 


Nesomys rufus (4) 


158.5(30) 


Yes 


Yes 


Yes 


No 



* Mean body mass of adults based on data in Chapter 22; the figure in parentheses is the number of individuals 
weighed, 
t Species occurred above known elevational range of Canarium. 
% The single individual taken on the ground was in a pitfall bucket. 



don. Three of the rodent species tested occur 
above the known elevational limit of Canarium in 
the RNI d'Andringitra. One of these species, 
Brachyuromys ramirohitra, readily opened the 
seed capsules, whereas Eliurus majori and Mon- 
ticolomys koopmani did not consume the seeds in 
captivity. At other sites in the eastern humid for- 
est, B. ramirohitra is known to occur at about 900 
m (Carleton & Schmidt, 1990). Further, at other 
sites Canarium is known from elevations up to 
1800 m (Leenhouts, 1959). 

Of the six rodent species occurring within the 
elevational zone of Canarium, five consumed the 
seeds in captivity. The exception was Eliurus mi- 
nor, adults of which weigh on average 36.9 g, the 
smallest of any of the rodents known below 1000 




Fig. 6-1. Comparison of Canarium seeds (from left 
to right) opened by natural dehiscence, aye-aye, Neso- 
mys rufus, Rattus rattus, and Eliurus tanala. 



m in the RNI d'Andringitra. With the exception 
of Gymnuromys and Nesomys, which were cap- 
tured only on the ground, all species within this 
zone appear to be both terrestrial and arboreal 
(Table 6-1), and in principle are able to exploit 
fruits and seeds still on the trees as well as those 
that have fallen to the ground. The possibility ex- 
ists that arboreal rodents feed on fruits and seeds 
still attached to the Canarium trees. 

There has been some confusion in the literature 
about the gnawing marks made by aye-ayes on 
Canarium seed capsules. Without exception, aye- 
ayes on Nosy Mangabe remove the exocarp and 
pericarp by creating serrated edges all the way 
around one end of the fruit and removing a "cap" 
in one piece. They open the seed capsules by set- 
ting the superior incisors along the mid-endocarp, 
gnawing a groove with the inferior incisors into 
the apex, and then using the incisors as a lever to 
open the seed. Because they pry open the seeds 
from a gnawed groove at the apex, the point of 
breakage is always toward the mid-portion. They 
then extract the cotyledon with their thin middle 
finger (Sterling, 1993, 1994). Thus, the incisor 
groove at the apex is unique to seed capsules 
opened by aye-ayes (Fig. 6-1). These marks are 
easily distinguished from those of rodents. Ro- 
dents generally opened the seed capsules by 
gnawing a small, usually transverse, hole through 
the mid-endocarp along the central axis (Fig. 6- 1 ). 
In rare instances, the endocarp is gnawed open 
toward or at the apex of the seed capsule. In all 
cases the widths of the incisor gnaw marks are 
distinctly smaller than those of the aye-aye. 



GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 



85 





— 




e 




V 




w 




- 




U 




0, 




£ 


jU 


5 


"o 






£ 




— 




=i 




B 




V 




- 




•tori 




B 




V 




u 




- 


V 


ii 


S 


2- 


e 




V 




8 




.23 
2 


£ 




5 


13 


£ 


E 




a 








« 




z 


A 




"So 




fi 




^ 




- 




■— 




B 




V 




u 




i- 




V 




On 


V 




;►> 




? 


J3 


V 




< 


£ 




£ 








W> 




B 




0) 




- 




-*- 




B 




V 




u 




la 




V 




2. 




£ 


** 




s 


3 


V 







£ 


P£ 






£ 








=Jt 




S 




4> 




- 


c 


■a 


_o 


<-> 




■o 


1 


s 






© 













3 
C 


- 


a 


e 


B 




- 


c 


© 






a 


s 


u 






W 



ON ^O 

+i 7 

«"> ^ -* 

+l ^ ii 



SO ON 

_; on* 

+i 7 
■<* 

K2 



© ^d 
+1 J 



oc 



nC 



v© « 



3" 



--• **: co co 
*' 2} vo vd S 
+l ' ii +' ' ii 

2 co 2 ^ 

CO CO 



•£ V) 



ON 


On 




CO 




o 


CM 

+ 1 


cm 

1 




NO 






o 


— 




cm 




■■*- 




nC 




co 




o 


00 


— 








L< 


00 


cn 




00 




1 

E 


^ 


CM 


in 


(■* 


CM 




«* 




co 




3 


+ 1 


1 


li 




II 


C 


CM 

© 


-4; 


Z 


1-* 


z 


Z 


CO 




m 






'* 










U 

3 

15 


oo 






o 




> 


rr 






©' 




E 

3 

E 
1 


CM 






■ 




E 


ON 






1 




E 

3 

E 














oo 










c 


II H 




II 






a 


Z 10 




z 







— ; o 

IN 

+ 1 I 

>n 

22 



co oo 
+ 1 I 

§2 

CM 



co oo 

_; o 

CM 
+1 I 



co © 

_; r- 

+i 7 

.n<-i 



O^_oo"^ cn ^ tN r» 
*28<n2r-<n!2<»<n£2 

+1 i II +1 J. li +i i " +1 i ii 

^Z^Z^d Z ^ND 
2 CO ^ CO ON CO COcN 

^ ^ CO CO 



_: o 

CM 

+ 1 I 

CO^ 

NO « 



CM 00 

+i i 

On rt X 



ON ON 

— ; CM 

CM 

+ 1 I 



■ NO Z 

_< r- 

+1 7 

3d 



cm no t-» ""> oooo rtCO 
* o £ ri a" oo rr cm ^ ^ no gj 

+1^ ii +i ' 



+i i 



ii +i i 

cj co f^ co O co 



co 



I 



Oh 

?s 
*1 

VI 

C U 
i 5 

2, > 

3 (U 

£ -B 

St 



On the basis of the feeding experiments and the 
extensive experience of E.J.S. with aye-ayes, the 
seed capsules with central gnawed holes that 
Duckworth (1993) recorded as possible signs of 
aye-aye were most likely opened by a rodent. No 
difference was found in the gnaw marks between 
rodent species, although this would be worth fur- 
ther investigation, because the position of the 
opening along the axis may be species-specific. In 
summary, on the basis of our field observations 
and feeding experiments with captive rodents, it 
was possible to distinguish between seeds opened 
by the natural dehiscence of the capsules, by aye- 
ayes, and by rodents. 



Random Collections of Seed Capsules 

Information on the various random collections 
of Canarium seed capsules found within the RNI 
d'Andringitra at 720 and 810 m is presented in 
Table 6-2. There was considerable variation be- 
tween the four samples in the percentage of seed 
capsules unopened (range, 0-33.3%). In all four 
samples, rodents were the most common source 
of opened seed capsules; this varied from a min- 
imum of 38.1% to a maximum of 87.5%. Opened 
seed capsules allocated to aye-ayes were found in 
two of the four collections and never represented 
more than 12.5% of the total sample. Seed cap- 
sules that showed natural dehiscence represented 
between and 23.8% of each sample. Within the 
four samples, seed capsules that showed no signs 
of cotyledon predation by mammals (unopened or 
natural dehiscence) varied from to 57.1%. 

In Table 6-2 we present the length and width 
measurements of each sample; samples are also 
segregated by class of endocarp breakage. No sta- 
tistically significant difference was found within 
or between samples in the size of seed capsules 
eaten by aye-ayes and rodents, nor did the verte- 
brate-opened endocarps differ from the general 
size distribution of random samples. Thus, on the 
basis of the present data, we conclude that the 
local utilization of Canarium seeds is highly vari- 
able within this forest, although rodents appear to 
be the most common consumers of the cotyle- 
dons. 



Seed Capsules from Rodent Burrows 

In Table 6-3 we present information on seed 
capsules recovered from rodent burrows. At 810 



86 



FIELDIANA: ZOOLOGY 



Table 6-3. Measurements (mm) of Canarium mad- 
agascariense seed capsules collected from three rodent 
burrows in the RNI d'Andringitra. 





Measurements'" 


Species and elevation 


Length 


Width 


Gymnuromys roberti 


38.6 ± 5.9 


16.1 ± 1.7 


(810 m) 


23.1-49.2 


13.3-19.9 




(N = 21) 


(N = 21) 


Nesomys rufus 


47.4 ± 2.3 


19.0 ± 1.3 


(810 m) 


44.3-49.8 


17.6-20.5 




(N = 4) 


(N = 4) 


Nesomys rufus 


48.3 ± 2.8 


19.9 ± 2.2 


(810 m) 


43.1-50.4 


16.8-21.0 




(N = 22) 


(N = 22) 



* Presented as mean ± standard deviation, minimum- 
maximum measurements, and sample size in parenthe- 
ses. 



m elevation a burrow system of Gymnuromys ro- 
berti contained 22 Canarium seed capsules; 21 of 
these had been gnawed open and the cotyledons 
consumed, whereas one was intact (Table 6-2). 
There was no significant difference in the size of 
the seed capsules cached by Gymnuromys and 
those in the random 810 m collections. A burrow 
of Nesomys rufus at the same elevation and about 
120 m from that of the Gymnuromys contained 
four seed capsules of Canarium, all gnawed open; 
eight fruits of Cryptocarya (Lauraceae), six whole 
and two gnawed open; and one whole unidentified 
seed. The contents of a Nesomys rufus burrow at 
820 m elevation and about 500 m ground distance 
from the other excavated Nesomys burrow con- 
tained 22 Canarium seed capsules, 15 of which 
had been gnawed open. The pooled sample of 
seed capsules recovered from Nesomys burrows 
(N = 26) was longer than those in the random 
810 m sample (/ = 3.21, P = 0.002); there was 
no difference in width measurements. Thus, it 
would appear that Nesomys is preferentially se- 
lecting and caching the larger seed capsules it en- 
counters on the forest floor. An excavated Eliurus 
webbi ground burrow at 720 m contained 20 fruits 
of Cryptocarya, about half of which had been 
chewed open (Goodman, 1994). 



General Consumption of Canarium 
Seed Capsules 

In general, little information is available about 
the food habits of endemic Malagasy rodents. Ne- 
somys rufus, the only diurnal species of rodent 



within the RNI d'Andringitra, was observed to eat 
fallen seed capsules of Canarium, Artabotrys ma- 
bifolius (family Annonaceae), and Sloanea rho- 
danta (family Elaeocarpaceae). After a major 
storm passed through Vatoharanana, Pare Nation- 
al (PN) de Ranomafana, a large hollow tree was 
found toppled that contained a cavity within the 
main trunk packed with Canarium seed capsules, 
many of which had been gnawed open along the 
central axis; none of the intact seeds was sprout- 
ing (C. Hemingway, pers. comm.). On the basis 
of a description of the cavity, it probably belonged 
to Brachytarsomys albicauda, a species known to 
occur within this park (Nicoll & Langrand, 1989). 
In summary, although information in the literature 
on Nesomyinae food habits is meager, it is clear 
on the basis of our feeding experiments that nu- 
merous rodent species are able to open and feed 
on the seeds of Canarium. 

In a prior study on Nosy Mangabe, 204 Can- 
arium seeds that had been fed upon by aye-ayes 
were opened to determine the average number of 
locules per seed capsule and whether viable cot- 
yledons remain after an aye-aye has fed on the 
seed capsules (Sterling, unpubl. data). Thirty- 
eight percent of the fruits bore one locule, 52% 
bore two locules, and 10% bore three locules. 
With only one exception, aye-ayes had opened all 
the locules large enough for viable cotyledons and 
extracted the contents. In one seed a viable cot- 
yledon remained untouched. Aye-ayes only at- 
tempted to open aborted locules twice. 

The results of recent lemur feeding ecology 
studies have shown that several species consume 
the soft pericarp of Canarium fruits. Simons Mor- 
land (1991) found that Varecia variegata varie- 
gata on Nosy Mangabe consumes whole fruits of 
Canarium and probably passes whole seed cap- 
sules. She found considerable temporal variation 
in the utilization of these fruits. In November 
1987, Canarium fruit was the most common food 
resource, accounting for 34% of the diet by time 
spent feeding, whereas in December 1988, Can- 
arium fruit represented only 3% of the diet. On 
the Masoala Peninsula, Rigamonti (1993) noted 
that about 1 .5% of the feeding time of V. v. rubra 
was on Canarium fruits. 

Overdorff (1991) reported that Canarium fruits 
form a significant portion of the diet of Eulemur 
fulvus rufus and E. rubriventer in the PN de Rano- 
mafana, particularly between the months of Oc- 
tober and January, although there were consider- 
able differences between these species. Andrian- 
isa (1989) also noted that E. f albifrons on Nosy 



GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 



87 



Mangabe feeds on the fruits of Canarium. In ad- 
dition, bush pigs (Potamochoerus larvatus), a spe- 
cies probably introduced to Madagascar (Grubb, 
1993), frequently ingested fallen Canarium seeds 
on Nosy Mangabe (Sterling, unpubl.). 

It appears that Eulemur and Varecia are seed 
dispersers, whereas aye-ayes, rodents, and bush 
pigs are seed predators. Experimental tests of seed 
scarring did not increase germination rates of 
wild-collected Canarium seed capsules (Tsiza, 
1989). Unopened seed capsules cached in rodent 
burrows would also have the possibility of ger- 
minating, particularly because a significant por- 
tion of the seeds remain viable for up to 3 years 
in storage (Tsiza, 1989). 

There is evidence of competitive interactions 
within and between lemur species over the con- 
sumption of Canarium fruits and seeds. Simons 
Morland (1991) found that dominant female Va- 
recia displaced males and other females from 
food resources, including Canarium. Further, ag- 
gressive interactions between Varecia and Dau- 
bentonia have been observed on Nosy Mangabe 
when both species were feeding at night on fruits 
in the same Canarium tree (Sterling, 1993). 



Phenology of Canarium Fruiting 

Several studies have monitored the fruiting and 
flowering patterns of Canarium in eastern rain 
forests of Madagascar (Andrianisa, 1989; Over- 
dorff, 1991; Sterling, 1993). Some trees have mul- 
tiple crops per year, while others have single 
crops. Canarium trees at Ranomafana and on 
Nosy Mangabe were found to fruit asynchronous- 
ly. On Nosy Mangabe, fruit was found in all 
months of the year, although fewer fruits were 
available during the cold, wet season (June to Au- 
gust) than the rest of the year (Andrianisa, 1989; 
Sterling, 1993). 



and at least two species cached them in their bur- 
rows. On Nosy Mangabe, aye-ayes open seed cap- 
sules still in trees as well as those that have fallen 
to the ground. Thus, aye-ayes compete with other 
species of arboreal lemurs and potentially arboreal 
rodents when the fruits are still on the tree, and 
with terrestrial rodents and bush pigs when the 
fruits and seed capsules are on the ground. 

On the basis of current information, it is clear 
that there is overlap in the diet of the aye-aye and 
rodents on the eastern slopes of the RNI d'An- 
dringitra. There seems to be no clear difference in 
the sizes of Canarium seed capsules that they 
each exploit, however, although seed capsules 
cached in Nesomys burrows are larger than those 
found in random samples under trees. During cer- 
tain periods of the year, this resource is limited 
(Andrianisa, 1989; Sterling, 1993), and these an- 
imals may compete for the nutritious seeds. 



Competition Between Native Animals 
and Rattus rattus 

One of the surprising results of the small mam- 
mal survey of the RNI d'Andringitra was that a 
rodent introduced to Madagascar, Rattus rattus, 
has colonized areas of undisturbed forest (Chapter 
22). At other sites on the island, Rattus rattus ap- 
pears to be replacing native rodents (Goodman, 
1995). The finding that Rattus rattus readily fed 
on Canarium seeds, a resource that forms a sig- 
nificant proportion of the aye-aye's diet (Sterling, 
1994) and may also be important for several spe- 
cies of endemic rodents, suggests that the spread 
of Rattus rattus across the island may be having 
an effect on populations of aye-ayes and Neso- 
myinae rodents. This point is in need of imme- 
diate further research, because its implications are 
potentially catastrophic for endemic animals that 
have been the focus of so much conservation ef- 
fort on Madagascar. 



Competition Between Aye-ayes and 
Native Rodents? 

At the 720 and 810 m sites, Canarium mada- 
gascariense was a common emergent tree (Chap- 
ter 4). We found few Canarium seed capsules eat- 
en by aye-ayes; this may reflect the relative rare- 
ness of the primate in this forest, or perhaps we 
simply did not find a tree heavily visited by this 
animal. On the other hand, several species of en- 
demic rodents feed widely on these seed capsules, 



Acknowledgments 






We are grateful to Pete Philippson and Beverly 
Lewis for identifying various seed remains. Jodi 
Sedlock drew the figure. 

Literature Cited 

Andrianisa, J. A. 1989. Observations phenologiques 
dans la Reserve Sp6ciale de Nosy Mangabe, Maroant- 



88 



FIELDIANA: ZOOLOGY 



setra. M6moire de fin d' Etudes, Ecole Sup6rieure des 
Sciences Agronomiques, University d* Antananarivo. 

Carleton, M. D., and D. F. Schmidt. 1990. Systematic 
studies of Madagascar's endemic rodents (Muroidea: 
Nesomyinae): An annotated gazetteer of collecting lo- 
calities of known forms. American Museum Novita- 
tes, no. 2987: 1-36. 

Duckworth, J. W. 1993. Feeding damage left in bam- 
boos, probably by aye-ayes (Daubentonia madagas- 
cariensis). International Journal of Primatology, 14: 
927-931. 

Goodman, S. M. 1994. A description of the ground 
burrow of Eliurus webbi (Nesomyinae) and case of 
cohabitation with an endemic bird (Brachypteraciidae, 
Brachypteracias). Mammalia, 58: 670-672. 

. 1995. Rutins on Madagascar and the dilemma 

of protecting the endemic rodent fauna. Conservation 
Biology, 9: 450-453. 

Grubb, P. 1993. Order Artiodactyla, pp. 377-414. In 
Wilson, D. E., and D. M. Reeder, eds.. Mammal spe- 
cies of the world: A taxonomic and geographic ref- 
erence, 2nd ed. Smithsonian Institution Press, Wash- 
ington, D.C. 

Iwano, T, and C. Iwakawa. 1988. Feeding behaviour 
of the aye-aye (Daubentonia madagascariensis) on 
nuts of ramy (Canarium madagascariensis). Folia Pri- 
matologica, 50: 136-142. 

Leenhouts, P. W 1959. Revision of the Burseraceae of 
the Malaysian area in a wider sense. Xa. Canarium 
(Stickm.). Blumea, 9: 275-475. 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: 
Revue de la conservation et des aires prot£g6es. World 
Wide Fund for Nature, Gland, Switzerland, xvii + 374 
pp. 



Overdorff, D. J. 1991. Ecological correlates to social 
structure in two prosimian primates: Eulemur fulvus 
rufus and Eulemur rubriventer in Madagascar. Ph.D. 
thesis, Duke University, Durham, North Carolina. 

Perrier de la Bathie, H. 1946. Burs£rac6es (Bursera- 
ceae). Flore de Madagascar, Famille 106, pp. 1-50. 

Rigamonti, M. M. 1993. Home range and diet in Red 
Ruffed Lemurs (Varecia variegata rubra) on the Ma- 
soala Peninsula, Madagascar, pp. 25-39. In Kappeler, 
P. M., and J. U. Ganzhorn. eds.. Lemur social systems 
and their ecological basis. Plenum, New York. 

Simons Morland, H. 1991. Social organization and 
ecology of Black and White Ruffed Lemurs (Varecia 
variegata variegata) in lowland rain forest. Nosy 
Mangabe, Madagascar. Ph.D. thesis, Yale University, 
New Haven, Connecticut. 

Sterling, E. J. 1993. Behavioral ecology of the aye- 
aye (Daubentonia madagascariensis) on Nosy Man- 
gabe. Ph.D. thesis, Yale University, New Haven, Con- 
necticut. 

1994. Aye-ayes: Specialists on structurally de- 



fended resources. Folia Primatologica, 62: 142-154. 
E. S. Dierenfeld, C. J. Ashbourne, and A. T. 



C. Feistner. 1994. Dietary intake, food consumption 
and nutrient intake in wild and captive populations of 
Daubentonia madagascariensis. Folia Primatologica, 
62: 115-124. 

Tsiza, G. 1989. Essai de monographic sylvicole du 
ramy (Canarium madagascariensis) avec reference 
sp^ciale a la c6te est. M6moire de Fin d' Etudes, Ecole 
Sup6rieure des Sciences Agronomiques, University 
d' Antananarivo. 

White, F. 1983. The vegetation of Africa. UNESCO, 
Paris, 356 pp. 






GOODMAN & STERLING: UTILIZATION OF CANARIUM SEEDS BY VERTEBRATES 



89 



Chapter 7 

Millipedes (Diplopoda) from the Eastern 
Slopes of the Reserve Naturelle Integrate 
d'Andringitra, Madagascar 

Henrik Enghoff 



Abstract 

Eight species of millipedes (Diplopoda) were recorded in the Reserve Naturelle Integrate 
d'Andringitra. Two other species had been previously reported from the Andringitra Massif: 
Betshewna andringitrae Mauries, 1994, and B. orbatum Maudes, 1994. 

Resume 

Huit especes de myriapodes (Diplopoda) ont ete recensees dans la Reserve Naturelle Inte- 
grate d'Andringitra. Deux autres especes, Betcheuma andringitrae Mauries, 1994, etB. orbatum 
Mauries, 1994, venant du Massif d'Andringitra y avaient ete rdpertoriees precedemment. 



Introduction 

About 150 species of millipedes (Diplopoda) 
are known from Madagascar, and almost all are 
endemic. There are several endemic genera and 
two endemic higher taxa: Zoosphaeriini of the 
family Sphaerotheriidae (giant pill millipedes) and 
Spiromiminae of the family Pachybolidae (a 
group of spirobolidan millipedes). The giant pill 
millipedes (genus Globotherium, etc.) and the 
large cylindrical species of Sechelleptus (Spiros- 
treptidae) and Aphistogoniulus (Pachybolidae) are 
among the most conspicuous invertebrates of the 
island. 

The foundation of knowledge of Malagasy mil- 
lipedes was laid by the impressive monograph of 
Saussure and Zehntner (1902). Since then, little 
progress has been made. In addition to scattered 
records and remarks by various authors, there are 
three papers that have added significantly to our 
knowledge. 

Jeekel (1971) reviewed the endemic genus 
Aphistogoniulus (nine named species). In a sub- 



sequent paper, Jeekel (1974) studied relationships 
within the family Sphaerotheriidae and found the 
closest relative of the Malagasy forms (tribe 
Zoosphaeriini) to be the Arthrosphaerini of India 
and Sri Lanka. 

Recently, Mauries (1994) described the genus 
Betscheuma, which includes 13 species of tiny 
leaf litter and soil millipedes. The species, which 
were all described as new by Mauries (1994), oc- 
cur in mountain habitats (forests and "prairies"). 
Two of the new species were described from the 
Reserve Naturelle Integrate (RNI) d'Andringitra 
based on collections made in 1970-1971 during 
the Recherche Cooperative sur Programme (RCP) 
No. 225, mission to the Andringitra Massif (Pau- 
lian et al., 1971). 

Betscheuma andringitrae: Andringitra, Anjavi- 
dilava, "foret dense humide de montagne, 1800- 
1950 m," "foret dense sclerophylle de montagne 
a Philippia, 2000-2050 m." Andringitra, Maros- 
itry, "foret dense sclerophylle de montagne, 2000 
m." 

Betscheuma orbatum: Andringitra, "foret dense 



90 



FIELDIANA: ZOOLOGY 



Table 7-1. Elevational distribution of millipedes on the eastern slopes of the RNI d'Andringitra. 



Ia\ a 



720 m 



810 m 



1210 m 



1625 m 



Sphaerotheriidae 
undetermined 
Sechelleptus 

fulgens 
Sechelleptus 

metazonalis 
Sechelleptus 
sp. (nov.?) 
Spirostreptidae 
genus? sp.? 
Spirostreptidae 
genus? sp.? 
Aphis togon iulus 
aff. corallipes 
Pachybolidae 
genus? sp.? 



1M, 2F 



1M, IF 
3 juv. 

1M, 1 juv. 



5M, 4F 
1 juv. 
2M, 2F 



3M, IF 
1 juv.(?) 

5M, IF 



2M, 2F 
1 juv. 
1M, 5F 



IF 

4M, 3F 



M, male(s); F, female(s); juv., juvenile. 



sclerophylle de montagne a Philippia, 2000 m," 
"foret dense humide de montagne, 1950 m." An- 
dringitra, Pic Bory, zone sommitale, "fourre a 
bambous, 2550 m." Andringitra, Anjavidilava, 
"foret dense scleYophylle de montagne a Philip- 
pia, 2075 m." Andringitra, Andrianony, "foret 
dense humide de montagne, 1650 m." To the best 
of my knowledge, no other millipede has been 
recorded from the Andringitra Massif. 



Methods 

During the 1993 survey of the RNI d'Andrin- 
gitra, members of the expedition made general 
collections of millipedes. The majority of speci- 
mens were collected during the day on the ground 
or on vegetation; they are heavily biased toward 
large and colorful species. Descriptions of the 
vegetational communities associated with each al- 
titudinal zone are given in Chapter 2. Specimens 
were preserved in 70% alcohol. Four of the five 
unnamed species of Sphaerotheriidae, Spirostrep- 
tidae, and Pachybolidae are represented by adult 
males and are probably all undescribed. For three 
of them, the current state of the taxonomy of these 
families precludes even generic assignments. The 
collections are housed in the Field Museum of 
Natural History, with duplicates in the Zoological 
Museum, Copenhagen. 



Results 

The presently studied material, obtained in late 
1993 from the RNI d'Andringitra, comprises eight 
species (Table 7-1): 

Family Sphaerotheriidae 
Genus? sp.? 

Family Spirostreptidae 

Sechelleptus fulgens (Saussure & Zehntner, 
1901). Previously recorded (as Spirostreptus ful- 
gens) from Mandraka by Saussure and Zehntner 
(1901) and from Tolagnaro (Fort Dauphin) by 
Saussure and Zehntner (1902). 

Sechelleptus metazonalis (Saussure and Zehnt- 
ner, 1902). Previously recorded (as Spirostreptus 
metazonalis) from Col de Sakavalana by Saussure 
and Zehntner (1902). 

Sechelleptus sp. (nov.?) 
Genus? sp.? 1 
Genus? sp.? 2 

Family Pachybolidae 

Aphistogoniulus aff. corallipes (Saussure and 
Zehntner, 1902). Aphistogoniulus corallipes was 
recorded from Tolagnaro by Saussure and Zehnt- 
ner ( 1 902, as Spirobolus corallipes) and from Be- 
voalavo by Demange (1969, as Mystalides cor- 
allipes). 

Genus? sp.? 



ENGHOFF: MILLIPEDES FROM THE EASTERN SLOPES 



91 



With the negligible background information avail- 
able, no biogeographical interpretation of these 
results is possible. 



Acknowledgments 

I am grateful to Steve Goodman for letting me 
examine this material and for help with the manu- 
script. An anonymous reviewer provided impor- 
tant comments on an earlier version of this report. 



Literature Cited 

Demange, J.-M. 1969. Myriapodes recoltes a Madagas- 
car par M. L. Bigot. Bulletin Museum National 
d'Histoire Naturelle, Paris, 2nd ser, 41(2): 484-489. 

Jeekel, C. A. W. 1971. Notes on the genus Aphisto- 
goniulus Silvestri (Diplopoda, Spirobolida, Trigoniu- 



lidae). Bulletin Zoological Museum, University of 
Amsterdam, 2: 33-42. 

1974. The group taxonomy and geography of 



the Sphaerotheriida (Diplopoda). Symposium Zoolog- 
ical Society of London, 32: 41-52. 

Mauries, J. -P. 1994. Decouverte de Diplopodes Cras- 
pedosomides a Madagascar: Betscheuma n.g. de la 
famille gondwanienne des Pygmaeosomatidae Carl, 
1941 (Myriapoda, Diplopoda). Bulletin Museum Na- 
tional d'Histoire Naturelle, Paris, 4th seY, sect. A, 
16(1): 55-86. 

Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. RCP 225. Etudes des eco- 
systemes montagnards dans le region malgache. I. Le 
massif de l'Andringitra. 1970-1971. Geomorphologie, 
climatologie et groupements vegetaux. Bulletin Socie- 
te d'Ecologie, 11(2-3): 198-226. 

Saussure, H., and Zehntner, L. 1901. Myriopoden aus 
Madagaskar und Zanzibar gesammelt von Dr. A. 
Voeltzkow. Abhandlungen Senckenbergische Natur- 
forschende Gesellschaft, 26: 425-464. 

1902. Myriapodes de Madagas- 



car. Hist. phys. nat. politique Madagascar publiee par 
A. Grandidier, 53: 1-349, 15 pi. 



92 



FIELDIANA: ZOOLOGY 



Chapter 8 

Ant Diversity Patterns Along an Elevational 
Gradient in the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 

Brian L. Fisher 



Abstract 

Inventory methods designed to permit rapid, replicable, and quantitative sampling of the leaf 
litter ant fauna were tested in wet tropical forest in eastern Madagascar in the Reserve Naturelle 
Integrate d'Andringitra. Methods involved a combination of pitfall and leaf litter sampling 
along a 250 m transect. Surveys were conducted at four sites, located at 785, 825, 1275, and 
1680 m. From pitfall and leaf litter samples, I collected and identified 27,866 ants belonging 
to 114 species and 28 genera; general collecting yielded an additional 34 species. The first- 
order jackknife produced an estimate of a total of 132 species for the four sites collectively. 
Species accumulation curves demonstrate the efficacy of these techniques in sampling the 
majority of the ants in the leaf litter. The species collected and their relative abundance are 
presented. The composition of the fauna is compared with that of other tropical forest sites. 
Species diversity decreased with elevation. Species turnover and faunal similarity measures 
showed a division in ant communities between the lowest elevation sites (785 and 825 m) and 
the highest elevation sites (1275 and 1680 m) that corresponded to the cloud forest transition. 

Resume 

Dans la foret tropicale humide de la Reserve Naturelle Integrate d'Andringitra a Test de 
Madagascar, des techniques d'inventaire, concues pour permettre un dchantillonnage quantitatif 
rapide et r6p6titif des fourmis vivant dans la litiere de feruilles mortes ont 6te testees. Les 
techniques combinent l'utilisation de pieges "pitfall" et l'6chantillonnage de la litiere de fer- 
uilles mortes le long d'un transect de 250 m. Des inventaires ont 6te effecUtes dans quatre 
sites, localises respectivement a 785, 825, 1275 et 1680 m d'altitude. A partir des pieges 
"pitfall" et des echantillonnages effectu^s a partir de la litiere de feuilles mortes, 27866 fourmis 
appartenant a 1 14 especes et 28 genres ont 6te recueillis et identifies; une collecte ateatoire a 
permis de recolter 34 especes supptementaires. Le "first-order jackknife" a produit un total 
d'environ 132 especes pour quatre sites. Les courbes d'accumulation des especes ont d^montre" 
l'efficacite de ces techniques qui permettent de r6pertorier la majorite des fourmis dans la 
litiere de feuilles mortes. Les especes collectees et leur abondance relative sont presentees. La 
composition de la faune de fourmis est comparee a celle d'autres sites forestiers tropicaux. La 
diversite des especes diminue a mesure que 1'on monte en altitude. Le remplacement des 
especes et la similarite de la composition de la faune de fourmis montrent une division des 
communautes de fourmis entre les deux altitudes les moins eievees (785 et 825 m) et entre les 
deux altitudes les plus elevees (1275 et 1680 m). Ces derniers correspondent a la transition 
vers la foret d'altitude caracterisee par la presence de n£bulosite frequente. 

FISHER: ANT DIVERSITY PATTERNS 93 



Introduction 

Insects constitute 85% of the world's animal 
biodiversity (Groombridge, 1992) and deserve 
increased attention in regions of the world, such 
as Madagascar, where species-rich habitats are 
under threat. Conservation methods that priori- 
tize areas based on diversity patterns (species 
richness and local endemism) of birds and mam- 
mals and do not include insect diversity overlook 
most organisms and thus do not guarantee pres- 
ervation of the greatest diversity. Few studies 
have investigated the correlation between inver- 
tebrate and vertebrate diversity. Pearson and 
Cassola (1992) showed that for gridded squares, 
275 km on a side, across North America, the In- 
dian subcontinent, and Australia, species rich- 
ness levels of tiger beetles, birds, and butterflies 
were positively correlated. On a finer geograph- 
ical scale, however, there is evidence that verte- 
brate diversity is not an accurate indicator of in- 
vertebrate diversity. Prendergast et al. (1993) 
showed low overlap between highly diverse sites 
of butterflies, dragonflies, liverworts, aquatic 
plants, and breeding birds in Britain. At 32 sites 
in southeastern Australia, Yen (1987) found no 
correlation between the number of species of 
vertebrates and beetles. Scharff (1992) conclud- 
ed that sites in tropical rain forests of eastern 
Africa where linyphiid spiders showed the high- 
est diversity were not necessarily the same as 
those documented for birds, mammals, amphib- 
ians, and reptiles. Thus, arthropod patterns of di- 
versity cannot be assumed to be correlated with 
those of vertebrates, and methods are needed to 
quantify species richness and endemism in insect 
taxa. 

In this chapter, I propose and evaluate inven- 
tory methodologies to survey leaf litter ant di- 
versity along an elevational gradient. The ele- 
vational transect specifically addresses the fol- 
lowing question: To what degree does elevation 
influence diversity and abundance of ants? I ad- 
dress this question by comparing species rich- 
ness estimates, measures of faunal similarity, 
and species turnover (beta diversity) for ant 
species across four sampled elevations in the 
Reserve Naturelle Integrate (RNI) d'Andringi- 
tra. Future work will compare these patterns 
with those exhibited by a sympatric assembly 
of vertebrates. 



Study Organism and Methods 

Importance of Ants 

Madagascar's ant fauna is remarkable for its di- 
versity, endemism, ancient affinities, and ecolog- 
ical dominance. Because of its long isolation from 
other land masses, the level of endemism of ant 
species on Madagascar is high. In the only re- 
gional study of ants conducted on the island, D. 
M. Olson and P. S. Ward (pers. comm.) found 
92% of the ants at a dry forest site in western 
Madagascar to be endemic to the island. More- 
over, the ant fauna of Madagascar is one of the 
least understood. For example, two of the better 
known ant genera in the world still have over 50% 
of their species undescribed in Madagascar (Te- 
traponera: 24 undescribed out of 36 known, P. S. 
Ward, pers. comm.; Tetramorium: 33 undescribed 
out of 62 known, B. Bolton, pers. comm.). There 
are no valid species described from the region of 
RNI d'Andringitra. There is one form, Campono- 
tus hova becki var. altior Santschi, recorded from 
near the RNI d'Andringitra (Santschi, 1923; Per- 
rier de la Bathie, 1927), but this name is unavail- 
able taxonomically. The taxonomy of Campono- 
tus is in chaos, and the identity of this specimen 
cannot be understood until the complex of forms 
related to Camponotus hova have been thoroughly 
revised taxonomically. 



Assessment of Ant Diversity 

A primary objective of this project was to de- 
velop standardized methods for assessing terres- 
trial ant diversity that can be used in other tropical 
wet forests. The proposed method does not sam- 
ple the entire ant community (arboreal, subterra- 
nean, and terrestrial), but only those ant species 
that nest or forage in the forest leaf litter layer. 
Subterranean and arboreal ant faunas pose specific 
technical problems in developing standardized 
survey methods. Leaf litter lends itself to replic- 
able sampling with pitfall traps and leaf litter sift- 
ing. These two techniques are effective for col- 
lecting many ant species from a wide variety of 
habitats, and they provide data on abundance and 
diversity suitable for rigorous statistical analysis 
(Olson, 1991). Because we do not yet understand 
the relationship between leaf litter ant diversity 
patterns and the total ant fauna in any wet tropical 
forest, results from the following survey tech- 
nique may not be representative or indicative of 



94 



FIELDIANA: ZOOLOGY 




Fig. 8-1. Mini-Winkler sacks used for extracting leaf litter ants. At each elevation there were 26 mini- Winkler 
sacks, which were suspended under a tarp. See text for explanation of sampling methods. 



patterns of rarity, endemism, or species richness 
of arboreal or subterranean ant communities. 

In the RNI d'Andringitra, intensive ant surveys 
were conducted at four sites located at 785, 825, 
1275, and 1680 m. These fell within 75 m in el- 
evation of the project transects centered at 720, 
810, 1210, and 1625 m. At each elevation the sur- 
vey method employed 50 pitfalls and 50 leaf litter 
samples in parallel lines 10 m apart along a 250 
m transect. The site for the transect was chosen 
with the intent of sampling representative micro- 
habitats at each elevation. Pitfall traps were 
placed and the leaf litter samples gathered every 
5 m. Pitfall traps consisted of test tubes, 18 mm 
internal diameter by 150 mm long, that were part- 
ly filled with soapy water and a 5% ethylene gly- 
col solution, inserted into polyvinyl chloride 
(PVC) sleeves, and buried with the rim flush with 



the soil surface. Traps were left in place for 4 
days. 

I extracted arthropods from samples of leaf lit- 
ter using a modified form of the Winkler extractor 
(Besuchet et al., 1987; Ward, 1987; Olson, 1991). 
The leaf litter samples involved establishing two 

1 -m 2 plots on each side of the transect line, sep- 
arated by 1 m. The leaf litter inside each 1-m 2 
plot was collected and sifted through a sieve of 1 
cm grid size. Before sifting, the leaf litter material 
was minced using a machete to disturb ant nests 
in small twigs and decayed logs. Approximately 

2 liters of sifted litter were taken from the two 
1-m 2 plots. Ants were extracted from the sifted 
litter during a 48-hour period in modified Winkler 
sacks ("mini-Winkler"; Fig. 8-1). The standard 
Winkler sacks can hold up to four 2-liter samples 
of sifted litter; mini-Winkler sacks are designed 



FISHER: ANT DIVERSITY PATTERNS 



95 



to hold only one 2-liter sample. Each sifted litter 
sample was held in a mesh sack that was sus- 
pended in a larger cotton enclosure. The ants 
dropped out of the mesh sack and were collected 
in plastic bags ("Whirl Packs") containing 70% 
ethanol at the bottom of the mini-Winkler sack. 
At each elevational zone, a tarp was used to pro- 
tect the 25 mini-Winkler sacks from rain. Because 
insects are less effectively extracted from wet leaf 
litter, sifting was done at least 24 hours after sig- 
nificant rainfall. 

I also surveyed ants through general collecting, 
defined as any collection that is separate from the 
mini-Winkler sack or pitfall transects, including 
searching in rotten logs and stumps, in dead and 
live branches, in bamboo, on low vegetation, un- 
der canopy moss and epiphytes, under stones, and 
leaf litter sifting (four leaf litter samples were 
taken between 1800 and 2000 m elevation). At 
each transect site, general collections were con- 
ducted for an approximately 2-day period. These 
collections included samples of the arboreal ants 
found on low vegetation that were not sampled by 
pitfalls or leaf litter. Ants sampled using general 
collection methods, therefore, were not used in 
the analysis of the efficacy of the survey of the 
leaf litter ants, of faunal similarity, or beta diver- 
sity. 



Identification 

Specimens were identified to morphospecies 
based on characters previously established to be 
important at the species level for each genus. 
When possible, names were attached to these 
morphospecies by using taxonomic descriptions 
(Bolton, 1979) and by comparing specimens to 
material previously collected by Ward, Olson, and 
Fisher in Madagascar that was compared to type 
material. Specimens will be deposited at the Mu- 
seum of Comparative Zoology, Harvard Univer- 
sity, and a representative set will be returned to 
Madagascar. 



Data Analysis 

Evaluation of Sampling Method — Analyses 
of ant diversity patterns require an understanding 
of the extent to which the inventory technique 
samples the leaf litter ant community in each tran- 
sect zone. To assess the completeness of the sur- 
vey for the elevation sampled, I plotted cumula- 



tive species per sample curves for each elevation. 
Species accumulation is plotted as a function of 
the number of leaf litter and pitfall traps samples 
taken. For the analysis, each leaf litter sample was 
paired with the adjacent pitfall sample, collective- 
ly termed a station sample. Species accumulation 
curves for the 50 stations per transect and first- 
order jackknife estimates of total number of spe- 
cies in the local community from which the sam- 
ples were taken are plotted for each succeeding 
station. The first-order jackknife method is a non- 
parametric approach to improving the estimate of 
species richness and is based on the observed fre- 
quency of unique species (the jackknife estimator 
and standard deviation are defined in Heltshe & 
Forrester, 1983). For both species accumulation 
curves, sample order was randomized 100 times, 
and the means and standard deviation of the jack- 
knife estimates were computed for each succeed- 
ing station using the program Estimates (R. K. 
Colwell, unpubl.; see also Colwell & Coddington, 
1994). If the species accumulation or jackknife 
estimate curves appear to level off before 50 sta- 
tions, then the transect is arguably sufficient. Con- 
versely, if the curves do not flatten out, a longer 
sampling transect may be necessary to accurately 
compare diversity between elevations. 

Ant Diversity — Data on both species richness 
and abundance were used to assess the change in 
species composition along the elevational gradi- 
ent. Only records of ant workers were used in 
these calculations. Because alates may travel con- 
siderable distances during dispersal, their pres- 
ence does not necessarily signify the establish- 
ment of a colony of that species within the tran- 
sect zone. In addition, collections of queens and 
males dispersing from nearby nests at the time of 
the survey may not reflect relative abundance of 
the species. Because ants are colonial, abundance 
measures were based not on the total number of 
individual workers collected at each transect site 
but on species frequency (proportion of stations, 
out of 50, in which each species was collected at 
a site). 

For each elevation, I compared first-order jack- 
knife estimates of total species richness and 95% 
confidence limits. Similarity of the ant fauna of 
the different elevations was assessed using two 
different measures: (1) the Jaccard Index, based 
on presence/absence data only: Q = j/(a + b — 
j), where j = number of species found at both 
elevations, a = number of species at elevation A, 
and b = number of species at elevation B (Ma- 



96 



FIELDIANA: ZOOLOGY 






gurran, 1988); and (2) the simplified Morisita In- 
dex, which incorporates abundance data: 



('UU — 



2 1 (an, X bn,) 
(da + db)aN X bN 



where 



larif Ibn} 

da = — — — and db = 



aN 2 



bN 2 



where aN = total number of station/species oc- 
currences at elevation A, bN = total number of 
station/species occurrences at elevation B, an, = 
the number of stations occupied by the /th species 
at elevation A, and bn, = the number of stations 
occupied by the /th species at elevation B (Horn, 
1966; Wolda, 1981). 

Beta diversity (species turnover between ele- 
vations) was calculated in two ways. First, the 
beta diversity measure of Whittaker (1960) was 
used: beta-1 = (S/a) - 1, where S = the total 
number of species in the two elevations com- 
bined, and a = the mean number of species in 
each elevation. Because this measure does not dis- 
tinguish between species turnover and the loss of 
species along a gradient without adding new spe- 
cies, the measure of beta diversity developed by 
Harrison et al. (1992) was also calculated: beta-2 
= (S/a^) - 1, where S is the same as beta-1 
above and a^ is the maximum value of alpha 
diversity among the elevations compared. Finally, 
the proportions of species unique to an elevation 
were also compared. 



Results 

I collected and identified 30,708 ants, compris- 
ing 148 species and 28 genera, from general col- 
lections, leaf litter, and pitfall methods. These in- 
clude 705 queens and 441 males that were not 
used in any of the following analyses. Leaf litter 
and pitfall methods produced 27,866 worker ants 
belonging to 1 14 species. A list of ant species in 
the RNI d'Andringitra, based on all collecting 
techniques, and separated by elevation and tech- 
nique, is presented in Table 8-1. General collec- 
tions within each ±75 m transect zone are pre- 
sented. The 785 and 825 m general collections 
were separated by 3 km and did not overlap spa- 
tially. 

The total number of species was greatest in the 
lowest elevations sampled (89 at 785 m and 81 at 



825 m). These two zones also contained the great- 
est number of unique species, but the highest per- 
centage of species unique to an elevation was 
found at 1800-2000 m (50%). If the data for 785 
and 825 m are combined, 70% of the species (74 
of 105 spp.) are unique to the lowest elevations. 
The number of species and number of individuals 
collected from pitfall traps was low compared to 
mini-Winkler sack and general collection meth- 
ods. For example, at 785 m, 8,582 workers in 76 
species were collected from leaf litter samples, 
whereas only 218 workers in 19 species were col- 
lected from pitfall traps. Based on the number of 
ants extracted from the leaf litter samples over a 
48-hour period using mini-Winklers sacks the av- 
erage density of worker ants in the leaf litter layer 
was 67 per m 2 . 

The relative prevalence of the different subfam- 
ilies in the combined pitfall and leaf litter samples 
is shown in Figure 8-2. The fauna is dominated 
by Myrmicinae in both number of species and 
number of individuals, followed by Ponerinae. 

The abundance of ant species is presented in 
Table 8-2. Both the proportion of stations at which 
each species was collected and the number of in- 
dividuals collected are presented. Four out of 1 14 
species were present at every elevation, but the 
abundance of individual species often differed 
considerably from one site to the next. For ex- 
ample, Hypoponera sp. 1 had a low abundance at 
785 and 825 m (0.08 in both), while at 1275 and 
1680 m its abundance was much higher (0.88 and 
0.46, respectively). 

Species-accumulation curves and first-order 
jackknife estimates of species richness with their 
standard deviations are presented for each transect 
zone (Fig. 8-3a-d, Table 8-3). Jackknife and ob- 
served species accumulation curves leveled off, 
indicating that within the area of the survey, the 
techniques employed collected the majority of the 
ants foraging or living in the leaf litter. In a com- 
bined analysis of all elevations, the pitfall and 
mini-Winkler sack methods collected 86% of the 
total number of leaf litter ant species estimated by 
first-order jackknife technique to occur in the re- 
gion (Fig. 8-3e). 

Faunal similarity values (Table 8-4) based on 
presence/absence data (Jaccard's Index) and abun- 
dance (simplified Morisita Index) were lowest be- 
tween 1680 m and the two lowest elevations (785 
and 825 m). The highest value of similarity was 
between the 785 and 825 m sites. The 1275 m 
transects had a greater faunal similarity with the 



FISHER: ANT DIVERSITY PATTERNS 



97 



Table 8-1. Ant species list for the RNI d'Andringitra, including altitude and collection method. 















1800-2000 


Genus 


Species 


785 m 


825 m 


1275 m 


1680 m 


m 


CERAPACHYINAE 














Cerapachys 


1 
2 
3 


W, G 
W, G 


G 

W 

W, G 


G 








5 


W, G 


W 


W 








8 


W, G 


W, G 




W, G 






9 




G 








Simopone 


1 


G 










DOLICHODERINAE 














Technomyrmex 


mayri 


W 










FORMICINAE 














CAMPONOTINI 














Camponotus 


1 
2 
3 
4 
5 

10 
11 
13 
14 
15 
16 
17 


G 
G 


G 
P, G 


G 

W 

W, G 

W, G 

P 


G 
G 

P, W 


G 
G 

G 




hildebrandti 


G 


G 


G 






LAsmsn 














Paratrechina 


1 
2 
3 


W, P, G 
G 
G 


W, G 


W, P, G 








4 


W, G 


W, G 


W, G 


W 






5 


W, P, G 


W, P, G 










6 


W 










PLAGIOLEPIDINI 














Plagiolepis 


1 
3 


W, G 


W, P, G 




G 




MYRMICINAE 














CREMATOGASTRINI 














Crematogaster 


1 

2 

3 

lobata 




W 




G 


G 
G 




schenki 


W, G 


P 


P, W, G 




G 


DACETONINI 














Glamyromyrmex 


1 


W, G 


W 








Kyidris 


1 


W, G 


W 








Smithistruma 


2 
3 
5 


W, G 


G 


G 
G 






Strumigenys 


1 
2 
3 
4 
5 
6 
7 


W, G 
W, G 


G 


W 
W, G 


W, G 

W, G 
W 
W 






8 


W, P,G 


W 


W, G 








9 


W 


W 


W, G 








10 






W, G 








11 






W 








12 


W 


W 










13 


W 


W 










14 


W, P,G 


W, G 









98 



FIELDIANA: ZOOLOGY 



Table 8-1. Continued. 



Genus 



Species 



785 m 



825 m 



1275 m 



1800-2000 
1680 m m 



FORMICOXENINI 

Leptothorax 
PHEIDOLOGETONINI 

Oligomyrmex 

PHEIDOLINI 
Pheidole 



SOLENOPSIDINI 
Monomorium 



TETRAMORIINI 
Tetramorium 



15 


W 








16 


w 








17 






W 




18 


W, G 


W, G 






1 


G 


G 






1 


W, G 


W, G 






2 


W, P 


W, G 






1 
2 








W, P, G 


3 










6 






W, G 


W, P,G 


7 


W 


W 




W 


8 






W, P, G 




9 






W, PG 


W.PG 


10 






W, P 




11 


W, P 


W, P, G 






12 


w 


W, P 






13 


w 




G 




14 


w 


W, P,G 






15 


w 


W 






16 


w 


W 






17 


W, P 


W, PG 






18 








G 


19 










20 


w 








22 


w 








nemoralis 


W, P, G 


W, PG 






veteratrix 


W, P, G 


W, PG 


P 




longispinosa 


W, P,G 


W, PG 






1 






W, PG 


W, PG 


2 


W 


W 






3 






W, PG 


W, G 


4 








G 


5 


W 


W, G 






6 


W 


W, G 






7 


W, G 


W 






8 






W 


W, G 


9 






W 


W 


10 




W 


W, PG 




12 






P 




13 










14 


W, G 








1 




W 




W, PG 


2 






G 




3 


W 


W, G 






4 






P 


W 


5 


G 




G 


W.PG 


7 


W, G 


W 


W, G 




9 


W 








10 


W, P 


W, P,G 


W, PG 


W, PG 


11 


W, P 


W, P 






13 




P 






14 


W, P 


G 


W, P, G 




15 




W 






16 


W 


W 






17 






G 





FISHER: ANT DIVERSITY PATTERNS 



99 



Table 8-1. Continued. 

















1800-2000 


Genus 


Species 




785 m 


825 m 


1275 m 


1680 m 


m 




andrei 




W, P 


W 










cognatum 




W 


W, G 


W, G 








dysalum 




W, P, G 


W, P,G 










electrum 




W, P 


W, P 










marginatum 




W, P 


W, P 








PONERINAE 
















AMBLYOPONINI 
















Amblyopone 


2 
3 






W 


W 






Mystrium 


1 
2 




W 
G 


W, G 
G 








Prionopelta 


1 
3 




W, G 
W, G 


W, G 
W 








ECTATOMMINI 
















Discothyrea 


1 




W 


W, P 


G 






Proceratium 


1 




W 










PLATYTHYREINI 
















Platythyrea 


bicuspis 






G 








PONERINI 
















Anochetus 


grandidieri 




W, G 


W, G 








Hypoponera 


1 




W 


W 


W, G 


W, G 


G 




2 




W, G 


W 


W 








3 










G 


G 




6 




W 


W 










7 




W 


W 


W, G 


G 






8 








W 








9 




W 




W, G 


W, G 


G 




10 








W, G 


W, G 


G 




11 




W, G 


W 


W, G 


W, G 






12 




W 


W 










13 




W 


W 










14 




W, G 


G 










15 




W 












17 












G 




sakalava 




W, G 


W, G 


G 






Leptogenys 


1 

3 




P 


W 








Odontomachus 


coquereli 




W, G 


W, G 








Pachycondyla 


cambouei 
sikorae 




W, P,G 
G 


W, P 
G 








PSEUDOMYRMEC1 














Tetraponera 


exasciata 

grandidieri 

psw-S/ 

psw-92 




G 
G 

W, G 


W, G 
G 


G 








Total species: 


G 


47 


42 


36 


24 


20 




Total species 


P 


19 


19 


14 


8 






Total species: 


W 


76 


64 


35 


23 






Total species: All methods 


89 


81 


51 


31 


20 




Number (%) of unique species 


21 (24%) 


11(14%) 


9(18%) 


3(10%) 


10(50%) 




Total number of G collections 


55 


41 


24 


36 


16 




Number of workers: 


G 


479 


306 


382 


256 


273 




Number of workers 


P 


218 


255 


238 


127 






Number of workers: 


W 


8,582 


6,479 


5,467 


6,500 






Total number of workers 


9,279 


7,040 


6,087 


6,883 


273 



Only collections of worker ants are presented. Only general collections were conducted at 1800-2000 m; these 
included four leaf litter samples. P, from pitfall transect samples; W, from mini- Winkler sack and leaf litter transect 
samples; G, from general collection. A total of 148 ant species and 29,562 workers were collected. 



100 



FIELDIANA: ZOOLOGY 



CO 

o 

r- 



c 

CD 

O 

k_ 

CD 
CL 



70-i 

60- 

50- 

40- 

30- 

20 

10- 



oiO- 



__Q 



□ Species 
■ Individuals 



a 






f / 

c? <r «° 



# 



J 



/ 



Subfamily 



Fig. 8-2. Relative importance of the different sub- 
families in number of species and individuals for pitfall 
and leaf litter collections for all four elevations com- 
bined (general collections are excluded). Subfamily 
names are abbreviated (see Table 8- 1 ). 



1680 m transect than with the two lowest eleva- 
tions. 

Beta- 1 and beta-2 values showed similar trends 
in levels of species turnover between elevations 
(Table 8-5). The lowest levels of species turnover 
were between 785 and 825 m and between 1275 
and 1680 m. Beta-1 and beta-2 values differed 
from each other when comparing elevations with 
the greatest species turnover. Beta-1 values indi- 
cated that the highest species turnover was be- 
tween the lowest elevations (785 and 825 m) and 
1680 m, whereas beta-2 values indicated that this 
occurred between the lowest elevations and 1275 
m. Considering only altitudinally adjacent sites, 
however, the two measures were concordant. 



Discussion 

Efficacy of the Survey Technique 

The leveling off of the jackknife and observed 
species accumulation curves (Fig. 8-3) indicated 
that the mini-Winkler sack and pitfall methods 
used in this survey were effective and produced 
rapid, replicable, and quantitative results. This 
study showed that insects can be successfully in- 
cluded in rapid biological inventory programs. In- 



sects are not too numerous to survey and provide 
valuable information about patterns of species 
richness and abundance. Future surveys will be 
directed toward testing the efficacy of the ant sur- 
vey methods in other tropical habitats. 



Elevational Gradient 

The number of ant species decreased as a func- 
tion of elevation (Fig. 8-4). Olson (1994) docu- 
mented a similar rate of decrease of ant species 
richness in Panama (Fig. 8-4). Brown (1973) sug- 
gested that the reduction in ant diversity at higher 
elevations is the result of lower levels of radiant 
heat. For ants in montane forests, the most im- 
portant factor regulating colony survival may be 
clouds and high humidity, which prevent bright 
sunlight from raising the ground temperature to- 
ward the optimal level for larval development and 
for worker foraging activities. The biology of hu- 
mid forest ants is poorly known; therefore, it is 
unclear what traits allow certain species to persist 
in the cloud zone while others disappear. 

At higher elevations above the tree line (>2000 
m) in the RNI d'Andringitra, the leaf litter layer 
in open grassland/heath environment may receive 
more radiant heat during fog-free periods and thus 
support a greater ant diversity than at 1680 m. 
Although only 20 species were found in one af- 
ternoon of general collection at 1800-2000 m, 
compared with 31 species found using all meth- 
ods at 1680 m, I predict that a thorough study of 
the grassland/heath ant community at 2000 m will 
show a greater diversity of ant species than at 
1680 m. This would be consistent with Brown's 
(1973) hypothesis about the role of radiant energy 
in determining ant diversity. 



Fauna! Similarities and Species Turnover 

Faunal similarity and beta diversity measures 
(Tables 8-3 and 8-4) suggest a division of the RNI 
d'Andringitra ant fauna into two communities, 
one occurring in lowland forest (785 and 825 m) 
and the other in montane cloud forest (1275 and 
1680 m). Species turnover is greatest and faunal 
similarity is lowest between 1275 m and the two 
low-elevation sites. Although there is no detailed 
study of the temperature gradient along these el- 
evations, the montane cloud forest community 
zone corresponds to the area in which clouds ha- 
bitually develop on a daily basis (see also Chapter 



FISHER: ANT DIVERSITY PATTERNS 



101 



Table 8-2. Proportion of stations (out of 50 paired pitfall and leaf litter samples at each altitude) each species 
was recorded. The number of individual workers collected is given in parentheses. 



Genus 



Species 



785 m 



825 m 



1275 m 



1680 m 



CERAPACHYINAE 

Cerapachys 



PLAGIOLEPIDINI 

Plagiolepis 
MYRMICINAE 

CPvEMATOGASTRINI 

Crematogaster 

DACETONINI 

Glamyromyrmex 
Kyidris 
Smithistruma 
Strumigenys 



PHEIDOLOGETONINI 

Oligomyrmex 

PHEIDOLINI 

Pheidole 



1 


0.04 (90) 




2 




0.02 (3) 


3 


0.10(49) 


0.06 (23) 


5 


0.02 (14) 


0.02(1) 


8 


0.10(9) 


0.12(46) 



0.04 (30) 



0.80(146) 



0.54 (66) 



3 




0.32 (53) 


schenki 


0.02(1) 


0.02 (2) 


1 


0.06 (5) 


0.02(1) 


1 


0.16(151) 


0.14(130) 


3 


0.20(31) 




1 


0.02 (2) 




2 






3 


0.06 (6) 




4 






5 






6 






7 






8 


0.36 (56) 


0.08 (4) 


9 


0.06 (4) 


0.02(1) 


10 






11 






12 


0.06 (3) 


0.22 (21) 


13 


0.02(1) 


0.02(1) 


14 


0.24(41) 


0.16(11) 


15 


0.02(1) 




16 


0.02(1) 




17 






18 


0.98 (417) 


0.74(183) 


1 


0.64 (228) 


0.50(114) 


2 


0.08 (69) 


0.16(42) 


2 






6 






7 


0.02(1) 


0.02(1) 


8 






9 






10 






11 


0.66(188) 


0.74 (348) 


12 


0.08 (40) 


0.14(60) 


13 


0.30 (83) 





0.26(18) 



0.08 (5) 



0.72 (154) 
0.46 (78) 
0.24 (25) 
0.32 (32) 
0.06 (5) 



0.02(1) 



0.08 (30) 

0.76 (547) 
0.76 (892) 
0.44(102) 



0.12(21) 



DOLICHODERINAE 












Technomyrmex 


mayri 


0.04 (2) 








FORMICINAE 












CAMPONOTINI 












Camponotus 


5 
11 
13 
15 




0.02 (3) 


0.02(1) 

0.04 (2) 
0.02(1) 






17 






0.02(1) 


0.04 (2) 


LASIINI 












Paratrechina 


1 


0.82 (869) 


0.58 (546) 


0.38 (232) 






4 


0.16(32) 


0.62 (349) 


0.04 (4) 


0.02 (4) 




5 


0.22(132) 


0.80 (560) 








6 


0.02(1) 









0.06 (3) 

0.18(14) 
0.02(1) 
0.02 (2) 



0.60 (768) 
1.00(3,772) 
0.02 (7) 

0.38(531) 



102 



FIELDIANA: ZOOLOGY 



Table 8-2. Continued. 



Genus 



Species 



785 in 



825 m 



1275 m 



1680 m 



SOLENOPSIDINI 
Monomorium 



TETRAMORIINI 
Tetramorium 



PONERINAE 
AMBLYOPONINI 
Amblyopone 

Mystrium 
Prionopelta 

ECTATOMMINI 

Discothyrea 

Proceratium 
PONER1NI 

Anochetus 

Hypoponera 



14 


0.26 (325) 


0.52 (472) 






15 


0.02 (3) 


0.10(13) 






16 


0.10(20) 


0.16(24) 






17 


0.42(165) 


0.46 (332) 






20 


0.16(24) 








22 


0.02(1) 








nemoralis 


0.80 (828) 


0.56 (349) 






veteratrix 


0.60(147) 


0.54(186) 


0.02 (4) 




longispinosa 


0.30(171) 


0.66 (482) 






1 






0.14(29) 


0.44 (246) 


2 


0.08(10) 


0.28 (95) 






3 






0.28(165) 


0.04(31) 


5 


0.32(146) 


0.14(11) 






6 


0.88(1,862) 


0.82 (953) 






7 


0.58 (228) 


0.36(186) 






8 






0.42(191) 


0.44(158) 


9 






0.16(169) 


0.04(10) 


10 




0.12(14) 


0.42 (239) 




12 






0.02(1) 




14 


0.32 (63) 








1 




0.02 (2) 




0.86 (247) 


3 


0.02(1) 


0.80 (256) 






4 






0.02(1) 


0.04 (2) 


5 








0.16(13) 


7 


0.10(11) 


0.04 (4) 


0.08 (6) 




9 


0.12(37) 








10 


0.58(100) 


0.34 (46) 


0.62(157) 


0.50(179) 


11 


0.20 (20) 


0.06 (4) 






13 




0.02(1) 






14 


0.18(216) 




0.26 (24) 




15 




0.02(1) 






16 


0.08 (5) 


0.02(1) 






andrei 


0.12(44) 


0.08 (5) 






cognatum 


0.58 (99) 


0.22(15) 


0.22(31) 




dysalum 


0.34(110) 


0.30 (45) 






electrum 


0.16(12) 


0.06 (6) 






marginatum 


0.10(7) 


0.20(16) 






2 






0.06 (3) 




3 




0.02(1) 






1 


0.02(1) 


0.02(1) 






1 


0.56(122) 


0.18(12) 






3 


0.22(12) 


0.10(5) 






1 


0.02(1) 


0.04 (3) 






1 


0.02(1) 








grandidieri 


0.38 (96) 


0.30(113) 






1 


0.08 (45) 


0.08 (57) 


0.88 (896) 


0.46 (339) 


2 


0.44(154) 


0.04 (3) 


0.02 (26) 




6 


0.44 (236) 


0.02 (23) 






7 


0.20(16) 


0.24 (24) 


0.90 (875) 




8 






0.04(10) 




9 


0.02 (4) 




0.46 (92) 


0.60(119) 


10 






0.68 (275) 


0.02 (3) 


11 


0.72 (345) 


0.36 (48) 


0.76(351) 


0.46(155) 


12 


0.18(21) 


0.06 (3) 






13 


0.76 (227) 


0.88 (326) 






14 


0.04(10) 








15 


0.18(25) 









FISHER: ANT DIVERSITY PATTERNS 



103 



Table 8-2. Continued. 



Genus 



Species 



785 m 



825 m 



1275 m 



1680 m 



Leptogenys 

Odontomachus 
Pachycondyla 
PSEUDOMYRMECINAE 


sakalava 

1 

3 
coquereli 
cambouei 


0.20 (20) 

0.08 (4) 
0.06 (4) 
0.50 (66) 


0.12(12) 
0.02(1) 

0.02(1) 
0.06 (4) 


Tetraponera 


psw-87 
psw-92 


0.04 (2) 


0.08 (9) 



3). Further investigations of the high mountain 
grassland/heath ant community in Madagascar 
may also show a distinct ant community. Initial 
collection revealed 50% of the species unique to 
1800-2000 m in the RNI d'Andringitra. 

Additional ant surveys in eastern Madagascar 
will help to evaluate whether there exists a dis- 
tinct, homogeneous, and disjunct montane cloud 
forest plus grassland/heath ant community in 
Madagascar. Currently we lack distribution rec- 
ords on those elevation-specific species to deter- 
mine whether they are part of a cloud forest or 
grassland/montane community on other mountain 
ranges in eastern Madagascar. Because the heath 
community is limited to the highest mountains of 
Madagascar, this disjunct habitat may have spe- 
cies or clades that are more closely related to taxa 
in other heath communities than in adjacent cloud 
forest ants. 

Beta-1 and beta-2 values differed from each 
other in comparisons of sites with high species 
turnover (Table 8-5). Beta-1 and beta-2 indices 
differ in their propensity to emphasize either spe- 
cies replacement or species loss, the two means 
of obtaining species turnover. Species replace- 
ment is most prevalent at the lower elevations. For 
example, the measured species turnover between 



Table 8-3. The number of species collected and 
first-order jackknife estimates of total species richness, 
based on pitfall and leaf litter transects. Statistics are 
given for each altitude and for all elevations combined. 







Species 








richness 




Altitude (m) 


Observed 


estimate 


95% C.I. 


785 


77 


91.7 


0.363 


825 


67 


83.7 


0.395 


1,275 


39 


46.8 


0.247 


1,680 


23 


27.9 


0.242 


All elevations 


114 


131.9 


0.193 



95% C.I. indicates 95% confidence intervals. 



785 and 825 m is the result of a difference in 
species, not just the loss of species at 825 m. At 
higher elevations, however, there is species loss 
along the gradient without replacement. A nested 
dropout of species dominates the differences be- 
tween the 1275 and 1680 m ant communities. Un- 
der these circumstances, beta-2, which emphasiz- 
es species replacement over species loss, is the 
preferred measure of beta diversity (Harrison et 
al. 1992). 



Ant Fauna of RNI d'Andringitra 

In 1891, Forel provided keys and descriptions 
to 86 ant species known from Madagascar (Forel, 
1891). This was the last comprehensive treatment 
of the ant fauna of Madagascar. In 1922, Wheeler 
recorded 207 species from Madagascar (Wheeler, 
1922). Considering the number of undescribed 
taxa, the actual number of ant species in Mada- 
gascar may be on the order of 800. 

The chaotic state of the taxonomy of Malagasy 
ants makes it impossible to estimate the number 
of undescribed species listed in Table 8-1 without 
comparison to type material. There are a few gen- 
era, however, where recent revisions and type 
comparison make accurate estimates possible. Of 
the 19 species of Tetramorium collected in RNI 
d'Andringitra, only five are apparently described 
species. At least 17 of the 18 Strumigenys are un- 
described because there is only one available 
name for the entire Malagasy fauna (assuming 
that none of these specimens are tramp species). 
Two of the four Tetraponera are undescribed (P. 
S. Ward, pers. comm.). In addition, because there 
are no described species of Glamyromyrmex, Kyi- 
dris, Smithistruma, Leptothorax, Amblyopone, and 
Discothyrea from Madagascar, all species in these 
genera in Table 8-1 (eight total) are presumed to 
be undescribed. 

All 148 ant species in Table 8-1 are believed to 



104 



FIELDIANA: ZOOLOGY 





100- 


(a) 785 m 




90- 
80- 
70- 
60- 


x** 




50- 


i •♦* 


CD 


40- 


• ♦ 


O 


30- 


♦ 


CD 

Q_ 


20- 




C/) 


10- 






< 


i i i i i 

) 10 20 30 40 50 



90- 
80- 
70^ 
60- 
50- 
40- 
30- 
20- 
10- 
0- 


(b) 825 m 

♦ 
• ♦ 

♦ 


( 


I 1 I 1 I 

) 10 20 30 40 50 



CD 

n 

E 50-, 

c 

0) 40H 

> 



(c) 1275 m 



_C0 30 

t 20-I i ♦ 

o 

10-1 



lMW w 



^ 



i^ 1 !:: 

:♦♦♦* 



30 -. 
25- 
20- 

15 
10- 
5- 



(d) 1680 m 



- 1 1 1 1 1 

10 20 30 40 50 



J lip 



.iiSffi 



— i 1 1 1 1 

10 20 30 40 50 



Number of stations pooled 



140 



CO 

© 120 



(e) All elevations 



I 



co 100 

— 

o 



2 



80- 



E 

3 

C 60 

CD 

> 

j5 40 

3 

E 

8 *° 



? 

&* 



,#^::::s 

* 



20 



— I - 
40 



-I 1 1 1 1 1 

100 120 140 160 180 200 



60 80 

Number of stations pooled 



Fig. 8-3. Assessment of leaf litter ant sampling technique for each elevation (a-d) and for all elevations combined 
(e). The lower species accumulation curve in each chart plots the observed number of species as a function of the 
number of pooled stations sampled (stations = paired leaf litter and pitfall collections). The upper curve displays the 
non-parametric first-order jackltnife estimated total species richness (error bars are standard deviation) based on succes- 
sively larger number of samples from the data set (Heltshe & Forrester, 1983). Only every other point is shown in (a- 
d) and every fourth point is shown in (e). For all curves, each point is the mean of 100 estimates based on 100 
randomizations of sample accumulation order, calculated using the program Estimates (R. K. Colwell, unpublished). 



FISHER: ANT DIVERSITY PATTERNS 



105 



Table 8-4. Two measurements of faunal similarity 
between the four elevational zones sampled. 



Table 8-5. Beta-1 (above the diagonal) and beta-2 
(below the diagonal) diversity values of each pair of 
altitude sites. 



Elevation 


785 m 


825 m 


1275 m 


1680 m 






















785 m 




0.655 


0.160 


0.075 


Elevation 


785 m 


825 m 


1275 m 


1680 m 















825 m 


0.805 


— 


0.165 


0.084 


785 m 


— 


0.208 


0.724 


0.860 


1275 m 


0.229 


0.164 


— 


0.319 


825 m 


0.130 


— 


0.717 


0.844 


1680 m 


0.096 


0.068 


0.394 


— 


1275 m 


0.300 


0.358 


— 


0.516 












1680 m 


0.210 


0.239 


0.205 




Above the 


diaeona 


1 is the Jaccard Index of 


similarity 













(presence/absence data) and below the diagonal, the sim- 
plified Morisita Index of similarity (abundance data; 
Horn, 1966). Higher values represent greater similarity. 
Values in boldface represent comparisons of altitudinally 
adjacent transects. 



be endemic to Madagascar. This is greater than 
the estimated 92% endemism in ant species in the 
dry forest of Kirindy, western Madagascar (P. S. 
Ward, pers. comm.). The lower level of distur- 
bance and the lack of open habitats may have pre- 
vented the invasion of introduced species into the 
RNI d'Andringitra. 



Comparison with Other Faunas 

There is currently no other site in eastern Mad- 
agascar whose leaf litter ant fauna has been sam- 
pled with comparable intensity and identified to 
species. One comparison, however, can be made 
at the generic level. Most of the ant genera col- 
lected in RNI d'Andringitra (Table 8-1) were also 
collected in the RNI d'Andohahela in the Anosy 
Mountains, southeastern Madagascar, in a parallel 
survey conducted in 1992 (Fisher, unpubl.) The 
only exceptions are the genera Pilotrochus, Eu- 
tetramorium, and Aphaenogaster, which were en- 
countered only in RNI d'Andohahela, and Tech- 
nomyrmex, Leptothorax, and Odontomachus, 
which were collected only in RNI d'Andringitra. 

The relative prevalence of species from differ- 
ent subfamilies in the RNI d'Andringitra ant fauna 
(62% Myrmicinae and 22% Ponerinae; Fig. 8-1) 
shows striking similarities to other tropical forest 
leaf litter ant communities (Table 8-6). All of the 
following studies found the Myrmicinae to be the 
most common subfamily, followed by the Poner- 
inae, with an average Ponerinae/Myrmicinae spe- 
cies ratio of 0.336 ± 0.037 (SD) (Table 8-6): Bel- 
shaw and Bolton (1994), in moist tropical forest 
in Ghana; Levings (1983), in moist forest on Bar- 
ro Colorado Island, Panama; Olson (pers. comm.), 
in tropical wet forest in western Panama; Longino 
(1986), in tropical wet forest in Costa Rica; An- 



Higher values represent greater species turnover. Val- 
ues in boldface represent comparisons of altitudinally 
adjacent transects. Overall beta-1 diversity was 1.214 
and beta-2 diversity was 0.481. 



dersen and Majer (1991), in seasonally dry forest 
in Kimberley, northwestern Australia; and D. M. 
Olson and P. S. Ward (pers. comm.), in the dry 
Kirindy forest. The 1:3 ratio of Ponerinae to Myr- 
micinae in these studies suggests that the species 
richness of the Myrmicinae can be interpolated 
from the richness of the Ponerinae. In addition, if 
future studies in Madagascar confirm an approx- 
imate prevalence of 22% Ponerinae or 62% Myr- 
micinae, investigations directed at determining the 
species richness of leaf litter ants in an area may 
be able to use the number of Ponerinae or Myr- 
micinae species as an indicator for the entire fam- 
ily Formicidae. This extrapolation applies only for 
estimating the ground-dwelling, leaf litter ant fau- 
na. 



Potential Effects of Human Disturbance 

The 785 and 825 m transects differed slightly 
in ant composition, with low values of beta-1 and 



100-, 



O 80 

o 

<D 
</}• 60 



<D 40 
.Q 

E 

-5 204 




Andringtra 
Panama 



700 



— i 1 1 1 1 1 

900 1100 1300 1500 1700 1900 



Elevation (m) 



Fig. 8-4. The number of ant species as a function of 
elevation. RNI d'Andringitra data are from pitfall and 
mini-Winkler sack samples. Panama data are from leal 
litter samples (Olson, 1994). 



106 



FIELDIANA: ZOOLOGY 



' 



Table 8-6. The percentages of ant species collected in the Ponerinae and Myrmicinae subfamilies in six tropical 
forests. Only leaf litter and ground nesting ants, not arboreal were considered in the analysis. 







Myrmici- 










Ponerinae 


nae 








Location 


(%) 


(%) 


P/M 


Total spp. 


Source 


RNI d'Andringitra 


22 


62 


0.35 


114 


This study 


Ghana 


22 


63 


0.35 


176 


Belshaw & Bolton, 1994 


BCI, Panama 


23 


65 


0.36 


108 


Levings, 1983 


western Panama 


23 


67 


0.34 


196 


D. M. Olson, pers. comm. 


Costa Rica 


20 


70 


0.29 


134 


Longino, 1986 


Kimberley, Australia 


14 


37 


0.38 


115 


Anderson & Majer, 1991 


Kirindy, Madagascar 


15 


53 


0.28 


60 


P. S. Ward & D. M. Olson, 
pers. comm. 



Total species refers to the number of terrestrial ants at the location. The taxonomic ratio of species in the two 
subfamilies (P/M) is on average 0.336, with a standard deviation of 0.037. 



beta-2 (0.208 and 0.130, respectively). In addi- 
tion, the 825 m transect had fewer species and 
fewer individuals than at 785 m. These differ- 
ences could be the result of the distance or habitat 
differences between the two transect sites. The 
ability of species to disperse between the two 
transect sites could be affected by distance (3 
km), with potential barriers being the two rivers 
separating the transects, the Iantara and Sahani- 
voraky. The differences in diversity could also be 
the result of microhabitat differences in the forest. 
The 785 and 825 m transects were conducted in 
forest with similar structure, slope, and aspect, 
however, and the level of recent human distur- 
bance appeared to be comparable and low. One 
potentially important habitat difference is that the 
785 m transect began 75 m from a recently dis- 
turbed and cleared forest (tavy); this could affect 
abiotic conditions (relative humidity, soil temper- 
ature, and moisture content of soil and litter) that 
affect ant distributions within the forests (Adis, 
1988). None of the transects, nor an afternoon of 
collecting in the adjacent tavy, however, revealed 
any exotic or introduced ants. 



Summary 

This is the first intensive study of the leaf litter 
ant community in eastern Madagascar. When 
compared with data on other biotic groups, it will 
provide a unique basis for studying how species 
composition and diversity change with elevation 
and latitude. These comparative data are impor- 
tant for understanding intertaxon differences in di- 
versity patterns that should be factors critical for 
developing a conservation strategy based on di- 



versity. In addition, results of this study will pro- 
vide baseline data for monitoring future biological 
changes in this region of Madagascar. 



Acknowledgments 

This project was funded by a grant from Kredit- 
anstalt fur Wiederaufbau to World Wide Fund for 
Nature, Madagascar, National Geographic Society 
(no. 5152-93), and the National Science Founda- 
tion (INT 9319515). I am very grateful to S. Ra- 
zafimandimby for his expert field assistance, P. S. 
Ward for assistance in ant taxonomy, R. K. Col- 
well for help in data management and analysis, 
and D. Ahuatl-Cuautle, J. Defoe, and D. Rosen 
for data entry and collection curation. In addition, 
S. M. Goodman and P. S. Ward made valuable 
comments on the manuscript. 



Literature Cited 

Adis, J. 1988. On the abundance and density of terres- 
trial arthropods in central Amazonian dryland forests. 
Journal of Tropical Ecology, 4: 19-24. 

Andersen, A. N., and J. D. Majer. 1991. The structure 

and biogeography of rain forest ant communities in 

the Kimberley region of northwestern Australia, pp. 

333-346. In Mckenzie, N. L., R. B. Johnston, and P. 

G. Kendrick, eds., Kimberley rainforests of Australia. 

Surrey Beatty and Sons, Chipping Norton, Australia. 

490 pp. 
Belshaw, R., and B. Bolton. 1994. A survey of the 

leaf litter ant fauna in Ghana, West Africa (Hymenop- 

tera: Formicidae). Journal of Hymenoptera Research, 

3: 5-16. 
Besuchet, C, D. H. Burckhardt, and I. Lobl. 1987. 

The "Winkler/Moczarski" eclector for fungus and lit- 



FISHER: ANT DIVERSITY PATTERNS 



107 



ter Coleoptera. The Coleopterists Bulletin, 41: 393- 
394. 

Bolton, B. 1979. The ant tribe Tetramoriini (Hyme- 
noptera: Formicidae). The genus Tetramorium in the 
Malagasy region and in the New World. Bulletin of 
the British Museum (Natural History) (Entomology 
Series), 38: 129-181. 

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. 

Col well, R. K., and J. A. Coddington. 1994. Esti- 
mating terrestrial biodiversity through extrapolation. 
Philosophical Transactions of the Royal Society of 
London. Series B, 345: 101-118. 

Forel, A. 1891. Histoire naturelle des Hymenopteres. 
Deuxieme partie. Les formicides, pp. v + 237. In 
Grandidier, A. Histoire physique, naturelle, et poli- 
tique de Madagascar, Vol. 20(2). Impr. Nationale, Pa- 
ris. 

Groombridge, B., ed. 1992. Global biodiversity: Status 
of the Earth's living resources. World Conservation 
Monitoring Centre. Chapman and Hall, London. 

Harrison, S., S. J. Ross, and J. H. Lawton. 1992. Beta 
diversity on geographic gradients in Britain. Journal 
of Animal Ecology, 61: 151-158. 

Heltshe, J. F, and N. E. Forrester. 1983. Estimating 
species richness using the jackknife procedure. Bio- 
metrics, 39: 1-11. 

Horn, H. S. 1966. Measurement of 'overlap' in com- 
parative ecological studies. American Naturalist, 100: 
419-424. 

Levings, S. C. 1983. Seasonal, annual, and among-site 
variation in the ground ant community of a deciduous 
tropical forest: some causes of patchy species distri- 
butions. Ecological Monographs, 53: 435-455. 

Longino, J. T 1986. Expedition report: The ants of the 
Zona Protectora. Report filed with the Organization 
for Tropical Studies, San Jose, Costa Rica. 

Magurran, A. E. 1988. Ecological diversity and its 
measurement. Princeton University Press, Princeton, 
New Jersey, 179 pp. 



Olson, D. M. 1991. A comparison of the efficacy of 
litter sifting and pitfall traps for sampling leaf litter 
ants (Hymenoptera, Formicidae) in a tropical wet for- 
est, Costa Rica. Biotropica, 23: 166-172. 

. 1994. The distribution of leaf litter inverte- 
brates along a Neotropical altitudinal gradient. Journal 
of Tropical Ecology, 10: 129-150. 

Pearson, D. L., and F. Cassola. 1992. World-wide spe- 
cies richness patterns of tiger beetles (Coleoptera: Ci- 
cindelidae): Indicator taxon for biodiversity and con- 
servation studies. Conservation Biology, 6: 376-391. 

Perrier De La Bathie, H. 1927. Le Tsaratanana, 
l'Ankaratra et l'Andringitra. Memoires de l'Academie 
Malgache, 3: 4-71. 

Prendergast, J. R., R. M. Quinn, J. H. Lawton, B. C. 
Eversham, and D. W Gibbons. 1993. Rare species, 
the coincidence of diversity hotspots and conservation 
strategies. Nature, 365: 335-337. 

Santschi, F 1923. Descriptions de nouveaux Formici- 
des ethiopiens et notes diverses. I. Revue de Zoologie 
Africaine, 11: 259-295. 

Scharff, N. 1992. The linyphiid fauna of eastern Africa 
(Araneae: Linyphiidae)— -Distribution patterns, diver- 
sity and endemism. Biological Journal of the Linnean 
Society, 45: 117-154. 

Ward, P. S. 1987. Distribution of the introduced Ar- 
gentine ant {Iridomyrmex humilis) in natural habitats 
of the lower Sacramento Valley and its effects on the 
indigenous ant fauna. Hilgardia, 55(2): 1-16. 

Wheeler, W M. 1922. Ants of the American Museum 
Congo expedition. A contribution to the myrmecology 
of Africa. IX. A synonymic list of the ants of the 
Malagasy region. Bulletin of the American Museum 
of Natural History, 45: 1005-1055. 

Whittaker, R. M. 1960. Vegetation of the Siskiyou 
Mountains, Oregon and California. Ecological Mono- 
graphs, 30: 279-338. 

Wolda, H. 1981. Similarity indices, sample size and 
diversity. Oecologia, 50: 296-302. 

Yen, A. L. 1987. A preliminary assessment of the cor- 
relation between plant, vertebrate and Coleoptera 
communities in the Victorian mallee, pp. 73-88. In J. 
D. Majer, ed. The role of invertebrates in conservation 
and biological survey. Department of Conservation 
and Land Management, Western Australia. 



108 



FIELDIANA: ZOOLOGY 



Chapter 9 

Spatial Distribution of Some Aquatic 
Insects in the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 

Francois-Marie Gibon, Jean-Marc Elouard, 
and Michel Sartori 



Abstract 

The study of some groups of lotic insects (Trichoptera: Philopotamidae; Diptera: Simuliidae; 
Ephemeroptera: Leptophlebiidae and Euthyplociidae) of the Reserve Naturelle Integrate 
d'Andringitra highlights an important faunistic difference between the eastern forested slopes 
and the western "open" slopes within the reserve. Altitudinal zonation appears in the forest 
area, which separates localized "high-altitude species" and broadly distributed "low-altitude 
species." The species found on the western slopes generally have wider distributions on the 
island than those on the eastern slopes. 

Resume 

L'&ude de quelques groupes d'insectes lotiques de la Reserve Naturelle Integrate d'Andrin- 
gitra met en Evidence une profonde difference faunistique entre le versant oriental forestier et 
le versant occidental ouvert. Sur le versant forestier apparait une zonation altitudinale qui 
oppose des formes d' altitude tres localises et des formes de basse altitude un peu plus rdpan- 
dues. Les especes du versant occidental offrent genSralement une vaste aire de repartition. 



Introduction 

Among Malagasy freshwater organisms, the 
fish fauna is relatively well known (Kiener, 1963), 
and some information has been published for the 
macrocrustaceans, but there is little information 
on aquatic insects. Furthermore, for this latter 
group the information available is not evenly dis- 
tributed. For example, about 144 species of Odo- 
nata have been described from Madagascar 
(Schmidt, 1951; Fraser, 1956), compared to only 
24 Ephemeroptera, 22 Trichoptera, and 12 Dip- 
tera: Simuliidae. This is in contrast to the more 
than 1,600 species recorded for these latter three 



groups in continental Africa. On Madagascar, the 
low number of species probably reflects our cur- 
rent knowledge rather than the actual diversity. 

This lack of information has two important 
consequences concerning the present inventory of 
the Reserve Naturelle Integrate (RNI) d'Andring- 
itra. A large proportion of the aquatic insects col- 
lected are unknown to science, and with our cur- 
rent level of knowledge it is impossible to inter- 
pret the level of endemicity of these groups within 
the reserve. In this paper we focus roughly on the 
ecology of these organisms with reference to their 
elevational distribution, stream ecology, and gen- 
eral habitat specificity (see Malicky & Chantara- 
mongkol, 1993, for a review of these points). 



GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 



109 



Table 9- 1 . Altitudes, stream orders, and water tem- 
peratures of the sampling stations. 



Sam- 










pling 






Water 




station Altitude 


Stream 


tempera 




no. 


(m) 


orders 


ture ( e C) Biome 


Eastern slope 








1 


717 


5 


16 


Degraded forest 


2 


720 


4 


15 


Degraded forest 


3 


720 


5 


16 


Degraded forest 


4 


735 


4 


18 


Degraded forest 


5 


750 


3 


17 


Primary forest 


6 


750 


2 




Primary forest 


7 


900 


4 


16 


Primary forest 


8 


1180 


3 


14 


Primary forest 


9 


1210 


3 


14 


Primary forest 


10 


1625 


1 


11 


Primary forest 


11 


1630 


1 


11 


Primary forest 


Western 


slope 








12 


600 


5 


22 


Savannah 


13 


1400 


4 


21 


Savannah 


14 


1900 


2 


19 


Savannah 


15 


1950 


2 


19.5 


Savannah 


Study Sites 









This study was carried out in November 1993 
within the four elevational transect zones (720, 
810, 1210, and 1625 m) on the eastern slope of 
the massif (see Chapter 1 for a description of the 
sites) and in November 1994 at three additional 
sites in the summit zone and on the western side 
of the massif. The sampling of high-altitude rivers 
of the western slope was undertaken to distinguish 
the influences of different biomes (forest, degrad- 
ed forest, and savannah, etc.) from those of alti- 
tude and stream order. The major parameters for 
each collection site are listed in Table 9-1. 



Eastern Slope Stations 

720 m Elevational Zone — Hydrobiological 
collections were made at three sites within this 
zone. The Iantara River passes adjacent to the 720 
m camp and within a zone of partially degraded 
forest at the edge of swidden agriculture sites. The 
river is generally 10-15 m wide; in several places 
it is as wide as 30 m. For most of this length, the 
forest canopy does not overlap the river and the 
gallery forest is partially broken. The Iantara Riv- 
er is of stream order 5 (see Table 9-1). The main 
features of the river are strong flowing waters, al- 
ternating with basins and slow water flow. The 



Iantara River was sampled in two places: station 
3 (17 November 1993), located in the rockbound 
stretch upstream from the 720 m camp, and sta- 
tion 1 (16 November 1993), located 200 m down- 
stream from station 1 , just after the confluence of 
the Intara and Lalangina rivers. The vegetation 
along the Lalangina River, station 2 (17 Novem- 
ber 1993), a tributary of stream order 4, consisted 
largely of gallery forest. At the station 2 sampling 
site, about 40 m above the confluence, the river 
banks were overgrown with brush. 

810 m Elevational Zone — Three stations were 
sampled within this zone. Station 5 (21 November 
1993) was situated just below the 810 m camp, 
along the Sahanivoraky River, a tributary of the 
Sahavatoy. This watercourse (stream order 3) ran 
under a gallery forest with overlapping canopy. 
The river was torrent-like, running along a rela- 
tively steep route, broken by waterfalls. Rocky el- 
ements within the river were medium-sized to 
large boulders. Station 6 (21 November 1993) was 
in a small watercourse (stream order 2) inside the 
forest. The flow of water was very slow, with a 
discharge of a few liters per second. The substrate 
was gravel and sand. Station 4 (20 November 
1993) was along the Sahanivoraky River (stream 
order 4) and in an area of nonoverlapping gallery 
forest. The river course is chaotic, hindered by 
masses of rocks that act as numerous natural 
dams. 

1210 m Elevational Zone — Station 9 (22 No- 
vember 1993) was along a tributary of the Saha- 
vatoy River, known locally as the Volotsangana 
River, and situated close to the 1210 m camp. This 
waterway, of stream order 3, was bordered with 
a partly overlapping gallery forest. The water- 
course was characterized by large boulders, alter- 
nating with numerous waterfalls. The overall wa- 
terway was steep (> 45°). Station 8 (23 November 
1993) was along another river of the same order 
and parallel to the Volontsangana River, but at a 
slightly lower elevation; it was sampled only for 
Simuliidae. This river had a strong water flow and 
ran under an overlapping gallery forest. Its steep- 
ness was less than that of station 9, and the sub- 
strate was predominantly pebbles. 

1625 m Elevational Zone — Two steep water- 
courses were sampled. These were small streams 
cascading onto flagstone. Station 10 (25 Novem- 
ber 1993) was in a relatively open area, and water 
discharge was less than 1 liter per second. Station 
11 (24 November 1993) was in forest largely 
composed of bamboo and had a discharge of a 
few liters per second. 



110 



FIELDIANA: ZOOLOGY 



Western Slope Stations 

Four stations were sampled on the western 
slope, all of them along the Zomandao River, a 
confluent of the Mangoky Basin. Three sites were 
sampled during the November 1993 mission (sta- 
tions 13, 14, and 15; 28-30 November 1993), 
whereas station 12 was sampled during the No- 
vember 1992 field trip. It was retained because it 
lay on the Zomandao River at an altitude close to 
that of stations 1, 2, and 3 on the eastern slope. 
Station 12 was located near Ankaramena, along a 
sandy stretch of the Zomandao River (stream or- 
der 4), at 600 m. Here the watercourse was reg- 
ularly broken by rocks and was situated in a sa- 
vannah area. Station 13 was at the level of An- 
tanifotsy (1400 m); a rocky shelf and large boul- 
ders impeded the medium water flow (stream 
order 3). This river ran through a zone of savan- 
nah. Station 14 was on the western part of the 
high plateau of the reserve, at an altitude of 1900 
m. The site was near a small lake formed by the 
widening of the river (stream order 2), where it 
passed through a zone of grassy savannah inter- 
laced with Ericaceae bush. The river course was 
relatively level and was made up of flagstone, 
forming shelves, that alternated with deeper water 
zones that had stony-gravel bottoms. Station 15 
was located at the edge of the high plateau, at an 
altitude of 1950 m and a short distance from the 
source zone. At this site, the river ran through a 
zone of grassy savannah. Its steepness was slight 
(stream order 2), and the river bottom was made 
up of either flagstone or pebbles. 



Methods 

Sampling Techniques 

Three different methods of sampling were used: 
Evening Light Traps — This method was used 
to collect adults of aquatic insects with nocturnal 
flight activity. A large shallow pan was filled with 
water that was mixed with a tension-active agent 
(soap). The light sources were a white light gen- 
erated by a natural gas lamp (camping stove type), 
and an ultraviolet (UV) light produced by a bat- 
tery lamp with a UV tube. The trap was placed 
along the river bank, illuminated 10 minutes be- 
fore sunset, and switched off 1 hour later. Ephem- 
eroptera and other fragile insects were collected 
and preserved individually as they were caught in 



the trap. The other insects were filtered and pre- 
served in 70% ethanol. This method was highly 
efficient for the sampling of Trichoptera, less so 
for Ephemeroptera, and generally useless for Si- 
muliidae. 

Butterfly Net — This method was used pri- 
marily to capture adult Ephemeroptera, which 
swarm during the day and at dusk, and adult Odo- 
nata. 

Collection of Aquatic Stages — The different 
river substrates (pebbles, vegetation, fallen leaves) 
were sampled using a submerged net laid down- 
stream. Aquatic stages were also collected directly 
on the substrates. All substrate types were sam- 
pled thoroughly. In many cases the larval stages 
of various aquatic insects are known, and this 
technique complements and augments species col- 
lection by light traps. It was the only effective 
method for capturing Simuliidae. 



Results 

The collections made during the survey of the 
RNI d'Andringitra were sorted and sent to nu- 
merous specialists. Definitive results for numer- 
ous groups are not yet available, and the follow- 
ing results are far from complete for all aquatic 
insects. Descriptions of numerous new genera and 
species will be published elsewhere. 

In general our current knowledge of the species 
limits and geographical distribution of all aquatic 
insects is too imprecise for us to present a general 
review of the endemicity and ecological require- 
ments of this group. Herein we concentrate on 
three groups that we studied: Diptera (Simuli- 
idae), Ephemeroptera, and Trichoptera. 



Diptera: Simuliidae 

Simuliidae females in the imaginal stage are he- 
matophagous, whereas males are floricolous. Lar- 
vae live in well-oxygenated water. The identifi- 
cation of the Simulium species was relatively easy 
with pupae, particularly using characters associ- 
ated with the form and the number of gill fila- 
ments. The identification of larvae and adults to 
species level is more difficult; these life stages are 
seldom captured. 

Nine Simulium species were identified from the 
15 stations. Among these nine species, four were 
previously described {Simulium gyas, S. pentacer- 



GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 



111 



os, S. iphias, and S. imerinae), although the S. 
iphias is problematical and may comprise four 
distinct species (see below). Two species are new 
to science and are known only from the Andrin- 
gitra Massif (5. metecontae n. sp. and S. brunhesi 
n. sp.) (see Chapter 11). 

Within our material of Simulium iphias from 
the Andringitra Massif there are several morpho- 
logical species, each of which is distinguishable 
from the others by its number of pupal gill fila- 
ments (herein filaments are designated "f"), and 
each of which apparently lives in unique ecolog- 
ical conditions. Four Simulium iphias s.s.l. mor- 
phospecies are recognized from the reserve: 5. 
iphias 8f, S. iphias lOf, S. iphias 15f, and S. iphias 
19f. 



Distribution of Simulium in the Andringitra 
Area 

Species found on the eastern slopes of the re- 
serve included S. metecontae n. sp., 5. iphias lOf, 
S. pentaceros, and S. gyas; those on the western 
slopes included S. brunhesi n. sp., S. imerinae, S. 
iphias 8f, S. iphias 15f, and S. iphias 19f. No 
Simulium species occurring in the forested areas 
along the eastern slopes was found on the western 
side of the reserve; species within this genus thus 
appear to have relatively strict ecological require- 
ments (Fig. 9-1). 



Eastern Slope Simulium 

Simulium iphias lOf — This form was collected 
at stations 2, 4, and 9, of stream order 3, 4, and 
5 (not respective). It was not found at stations 3 
and 8. Stations 2, 4, and 9 had mean water flow 
ranging between 0.7 and 1.5 m/s. In contrast, wa- 
ter flow was slower at sampling stations 3 and 8. 

This species occurs all along the eastern coast 
of Madagascar, in small rivers running through 
zones of primary or degraded forest. It is absent 
from rivers and streams in completely degraded 
zones. 

Simulium metecontae n.sp. — This species is 
currently only known from the Andringitra Massif 
(see Chapter 11). It was first encountered in the 
820 m zone at the four stations (4, 5, 9, and 11) 
on rivers of stream order 1,3, and 4. Altitude and 
stream order may be critical parameters for the 
distribution of this species. Indeed, water flow is 



Wester 


n slope 












Eastern slope 




2000- 
1500- 
1000- 


m 




Diptera : Simuliidae 




|B,8f| 
°) 

14« 

B, 8f, l| 


8 t 


.« 


uP 




13« 

ST 


"^^10 




M, 10f, G 




/T 






2 


|M, I 


5] -^^^ 






V 


|15f, 19f| 






M, 10f,Gr 4 


n\ 


500- 










Stream order 1 ^^^^ Stream order 4 

Stream order 2 ■■■■■ Stream order 5 

^_^_ Stream order 3 *j> ■ sampled stations 













Fig. 9- 1 . Altitudinal distribution of Simulium in the 
RNI d' Andringitra on the basis of 1 1 sites on the eastern 
slopes and four sites on the western slopes. Key to spe- 
cies: I = Simulium imerinae, B = 5. brunhesi, G = S. 
gyas, M = 5. metecontae, P = S. pentaceros, 8f = S. 
iphias with eight filaments, lOf = S. iphias with 10 fil- 
aments, 15f = 5. iphias with 15 filaments, and 19f = S. 
iphias with 19 filaments. 



relatively slow for stations 6, 8, and 10, where 
this species was not collected. 

Simulium pentaceros — This species was found 
only at station 5. It is an infrequently collected 
species of Simulium, although it is also known 
from two forested rivers near Ranomafana, Ifan- 
adiana. It is difficult to make any general state- 
ments on its ecology. However, the two rivers 
near Ranomafana are of the same stream order as 
station 5 and at about 850 m. 

Simulium gyas — This species was found at 
three stations (4, 8, and 9) of stream order 4 and 
3, but not at station 5. S. gyas is broadly distrib- 
uted along the eastern coast from Antsiranana to 
Tolagnaro in small to medium-sized rivers run- 
ning through primary or degraded forest. It has a 
broad elevational distribution. 



Western Slope Simulium 

Simulium iphias 8f — This species was collected 
at stations 14 and 15, located on the high plateau 



112 



FIELDIANA: ZOOLOGY 



of the Andringitra Massif. The only other known 
record of this form is from the Ankaratra Massif, 
at the level of the source of the Sisaony River 
(1800 m). It is a Simulium of high altitude and 
open zones. 

Simulium brunhesi n.sp. — This species is cur- 
rently known only from the ericaceous savannah 
of the RNI d' Andringitra (see Chapter 11). It pre- 
sumably will be found at other high mountain 
sites on the island with similar habitats (e.g., Tsar- 
atanana and Ankaratra). 

Simulium imerinae — This species was collected 
at stations 13 and 14, along the Zomandao River 
and above 1300 m. It has been previously record- 
ed in numerous highland and foothill areas in the 
southern and eastern portions of the island. This 
species prefers small brooks running across flag- 
stone, with water flow of 1.2-2 m/s. 

Simulium iphias 15f — This species was found 
only at station 12, on the western slope. It has 
previously been collected in foothill areas in the 
southern and western parts of Madagascar from 
the Mandrare River to the Betsiboka River. 

Simulium iphias 19f — As with S. iphias 15f, S. 
iphias 19f was collected only at station 12. It in- 
habits areas with rapids on large highland rivers. 
Ankaramena is the lowest altitude at which it has 
been recorded. The cold waters of the RNI 
d' Andringitra may account for its low altitudinal 
distribution. 



Conclusions — Simuliidae 

The eastern and western slopes of the RNI 
d' Andringitra are inhabited by different species of 
Simulium. Water temperature and aquatic vegeta- 
tion (nutrients) are known to be important vari- 
ables accounting for these differences. Adult fe- 
male Simulium are hematophagous, and their 
feeding regimens may be strict and dependent on 
specific hosts (e.g., batrachians, reptiles, and 
mammals), perhaps at the family, genus, or spe- 
cies levels. Thus, local extirpation of vertebrates 
associated with deforestation may have a pro- 
found effect on the distribution of Simulium. The 
aquatic habitat of larvae and pupae may be only 
one of the distribution parameters for Simulium 
species. 

Species diversity is greatest in medium to large 
rivers on both the eastern and western slopes of 
the massif, particularly in waters of stream orders 
2, 3, and 4. The extreme stream orders 1 and 5 
are inhabited by only one Simulium species. In 



our sampling of Simulium in the RNI d' Andring- 
itra, five species were found on the western slopes 
and four species on the eastern slopes. A single 
species new to science was found on each slope. 
On the basis of variables associated with sam- 
pling stations (Table 9-1), it is difficult to state 
precisely what effects altitude and stream order 
have on Simulium distribution, because small and 
cold water bodies are located at the highest alti- 
tudes, and there is no large high-mountain river 
on the Andringitra Massif. 



Ephemeroptera 

Little is known about this order of insects in 
Madagascar (Demoulin, 1970). We are in the pro- 
cess of conducting inventories around the island. 
Many species were collected in the RNI d' An- 
dringitra, the majority of which are new to science 
and are currently being studied by several spe- 
cialists. Two families are considered here, Euthy- 
plociidae and Leptophlebiidae. It is important to 
remember that the standard emergence period for 
most mayflies is usually at the end of the rainy 
season (April and May), and that the Andringitra 
field trip was during the month of November. 



Euthyplociidae 

Only one genus of this family has been record- 
ed from Madagascar (Proboscidoplocia Demou- 
lin, 1966). Until recently, this genus was thought 
to be monospecific, represented by Proboscido- 
plocia sikorai (Vayssiere, 1895). 

Extensive studies are being jointly carried out 
by the Laboratoire de Recherche sur les Systemes 
Aquatiques et leur Environnement (LRSAE), An- 
tananarivo, and the Musee Cantonnal de Zoologie 
de Lausanne. This research has shown that at least 
10 Proboscidoplocia species occur on Madagas- 
car (Elouard et al., unpublished data). Among 
these 10 species, three are known from the An- 
dringitra Massif: Proboscidoplocia sikorai sensu 
stricto (s.s.) (stations 1, 4, 5, and 6); Probosci- 
doplocia vayssieri n. sp. (Elouard & Sartori, in 
press) (station 9); and a third species, recorded 
only at station 4, known only from females, and 
identified by eggs. The P. sikorai s.s. and P. vays- 
sieri n. sp. are easily distinguished by the shape 
of male genitalia. Proboscidoplocia pupae live 
under rocks in permanent water and were cap- 
tured with deep nets placed under rocks. Imagos 



GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 



113 



were caught by evening light traps. Most Ephem- 
eroidea appear in masses a few days per year. The 
November Andringitra mission coincided with 
one of these synchronous emergences. Specimens 
of males and females were captured at several sta- 
tions. 

Distribution of Euthyplociidae in the 
RNI d'Andringitra 

Proboscidoplocia were collected only on east- 
ern rivers (Fig. 9-2). The distribution of P. sikorai 
s.s. appears to be independent of stream order and 
forest cover. This species occurs in rivers of 
stream orders 5 and 2, passing through degraded 
and primary forest. Altitude appears to be a major 
factor for the separation of these two species, 
which are not elevationally sympatric on the mas- 
sif. Differences in elevation are related to water 
temperatures. No Proboscidoplocia adults or pu- 
pae were found in the 1625 m zone (stations 10 
and 11), although these sites were thoroughly in- 
spected. At this time it is impossible to affirm that 
their absence from the upper forest zone of the 
mountain is related to stream order or altitude. 
Proboscidoplocia sikorai s.s.d. has been recorded 
in the RNI de Marojejy (Fontaine, 1968), al- 
though no information is available on stream or- 
der. 



Leptophlebiidae 

Only five species of Leptophlebiidae are de- 
scribed from Madagascar. Two are known from 
the nymphal stage and three as imagos. On the 
basis of our recent collections from numerous 
sites and systematic studies at LRSAE, there are 
more than 50 species occurring on the island. This 
family is essentially forest-dwelling. One special- 
ist, W. L. Peters, is carrying out an extensive sys- 
tematic revision of this family in Madagascar. 

Twelve species of Leptophlebiidae were col- 
lected on the Andringitra Massif, 1 1 of which are 
new to science. We have classified these 12 taxa 
as subimaginal and imaginal morphospecies. The 
only named species is Nesophlebia adusta Peters 
and Edmunds, 1964, which is referred to here as 
Lepto-Q. 

Eastern Slope Leptophlebiidae 

Eleven species were collected along the eastern 
slopes of the RNI d'Andringitra. It is difficult to 



Western slope 



Eastern slope 



2000 -- 



1500 -- 



1000 



500 -- 



m 



Ephemeroptera : Euthyplociidae 
Proboscidoplocia 



J? 



®& 



^Vf * fi^y] "" 



Stream order 1 


■■m Stream order 4 


Stream order 2 


i^HHi Stream order S 


mm Stream order 3 


A = sampled stations 



Fig. 9-2. Altitudinal distribution of Proboscidoplo- 
cia in the RNI d'Andringitra on the basis of 1 1 sites on 
the eastern slopes and four sites on the western slopes. 
Key to species: V = Proboscidoplocia vayssieri, S = P. 
sikorai, and Sp3 = P. sp. 3. 



give precise details of habitat preferences of the 
various morphospecies on the basis of variables 
associated with collection stations. In general, 
only the following statements can be made: (1) At 
stations 1 and 5, five species were obtained at 
each station, and the species assemblages were 
totally different from one another. (2) No imago 
Leptophlebiidae was collected in the 1625 m zone 
(stations 10 and 11), although indeterminate lar- 
vae belonging to this family were found on these 
two small tributaries. The low nymphal densities 
unquestionably account for the absence of adults. 
(3) Lepto-P, Lepto-N, Lepto-S, and Lepto-AA 
were found only in the Iantara River. Stream order 
may be a critical parameter in the distribution oi 
these species. (4) Nesophlebia adusta (Lepto-Q) 
occurs in both the Iantara River (station 1) and 
one of its small tributaries (1210 m, station 9) 
For this species, stream order is of least impor- 
tance in the stock range established in the RNI 
d'Andringitra. Although this species was not col- 
lected at stations 2, 3, 4, and 5, its absence ai 
these stations is not proven. Once again, densit) 
and emergence periods may account for their lo 
cal absence at these sites. (5) Lepto-Y, Lepto-W 
and Lepto-Z were found at three stations of sim 



114 



FIELDIANA: ZOOLOGY 



Western slope 



Eastern slope 



14 « ► 



2000 -- 



1500 -- 



1000-- 



500 -- 



m 



. _ ! Ephemeroptera : Leptophlebiidae 




P,Q,N,S,AA 



Stream order 1 
. Stream order 2 
. Stream order 3 



h Stream order 4 
ai Stream order 5 
t tampled •Uliooi 



Fig. 9-3. Altitudinal distribution of Ephemeroptera: 
Leptophlebiidae in the RNI d'Andringitra on the basis 
of 1 1 sites on the eastern slopes and four sites on the 
western slopes. Key to species: AA = Lepto-AA, N = 
Lepto-N, O = Lepto-O, etc. 



ilar stream order. (6) Lepto-U and Lepto-V oc- 
curred only at station 5. 



Western Slope Leptophlebiidae 

Only one species, Lepto-O, was collected along 
the western slopes of the RNI d'Andringitra, thus 
supporting the idea that this family is largely for- 
est-dwelling. At station 1 2, three Leptophlebiidae 
subimagos were collected. At this development 
stage there are no clear characters for morpho- 
species determination. These three subimagos ap- 
pear to be different from subimagos obtained on 
the eastern slopes and from station 15. On the 
basis of current information, Lepto-O is known 
only from the high plateau ericaceous savannah 
of the RNI d'Andringitra, where it is presumably 
endemic. 



Other Ephemeroptera Species 

Among the other Ephemeroptera species col- 
lected during the 1993 mission to the RNI d'An- 
dringitra and currently under study by specialists 



are: (1) Several Baetidae (> 25 species). (2) One 
Ephemerellidae species. This family lives in cold 
waters with aquatic vegetation, and it is practi- 
cally unknown in Africa. Only three species are 
recorded from South Africa (Demoulin, 1970). In 
Madagascar, only one specimen was previously 
known, a larva identified as belonging to the ge- 
nus Ephemerella (Eurylophella); this identifica- 
tion may be incorrect. The shape of the nymph 
and the genitalia of the imago belong to Ephem- 
erellidae and not Telagodinae. In the RNI d'An- 
dringitra we were able to collect all developmen- 
tal stages of a mayfly that does not correspond to 
Ephemerella. The RNI d'Andringitra material un- 
questionably belongs to an undescribed genus. (3) 
Larvae of Trichorthidae, but no imago. (4) Larvae 
and adults of Prosopistomatidae that are probably 
referable to Prosopistoma variegatum. In view of 
the poor original description of this species, how- 
ever, examination of the holotype is needed to 
confirm this identification. Unfortunately, it ap- 
pears that the holotype is lost. (5) Heptageniidae 
(see Chapter 10). 

Conclusions — Mayflies 

More than 40 species of mayflies were collect- 
ed during the Andringitra mission. The great bulk 
of these species are new to science and have been 
distributed to specialists for study. These descrip- 
tions will be published in due course. The RNI 
d'Andringitra material will be used in the broad 
context of an island-wide survey of the aquatic 
insects of Malagasy rivers, with specific reference 
to speciation and ecological preferences. 

The order Ephemeroptera is a good indicator 
of environmental conditions. However, the ima- 
ginal sampling of this group is seasonal, and im- 
agos have a short existence, in general less than 
1 day. Furthermore, emergence of most mayfly 
families on Madagascar is synchronous. Never- 
theless, the presence of pupae in rivers is constant, 
and their collection can provide species identifi- 
cations in most cases. Few associations between 
larvae and adults, however, are known. Rearing 
of larvae is critical in determining the life history 
stages of each species, and this fieldwork is time- 
consuming. 



Trichoptera 

The families and genera collected during this 
study were Hydropsychidae: Macrostemum, Cheu- 



GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 



115 



Table 9-2. Distribution of Leptophlebiidae in the RNI d'Andringitra (see Fig. 9-3). 



Species 



Eastern stations 



Western stations 



10 11 12 13 14 15 



Lepto-P 


+ 


Lepto-Q 


+ 


Lepto-N 


+ 


Lepto-S 


+ 


Lepto-AA 


+ 


Lepto-Y 




Lepto-U 




Lepto-V 




Lepto-W 




Lepto-X 




Lepto-Z 




Lepto-O 









Rain 






Rain 






Rain 






Rain 






Rain 




+ 


Rain 




+ 


Rain 




+ 


Rain 


+ 


+ 


Rain 




+ 


Rain 


+ 




Rain 
Rain 



matopsyche, and Leptonema; Hydroptilidae: Hy- 
droptila, Orthotrichia, and Oxyethira; Philopo- 
tamidae: Chimarra, Dolophilodes, and Pauliano- 
des; Ecnomidae: Psychomyiellodes; Pisuliidae: 



Western slope 



2000 -- 

"fill ■- 



1500 -r- 
1000 
500 



Eastern slope 



Trichoptera Philopotamidae 




aEJdAdBpEpC| 



aEaFK 



aGpE 
MNJK 




4 3 



JaG " 



HLaG] 



- Stream order I ^^^_ Stream order 4 
_ Stream order 2 ■■■■■■ Stream order 5 
. Stream order 3 N° - sampled stations 



Fig. 9-4. Altitudinal distribution of Trichoptera: 
Philopotamidae in the RNI d'Andringitra on the basis of 
1 1 sites on the eastern slopes and four sites on the west- 
ern slopes. Key to species: I = Chimarra sp. I; D = 
Chimarra sp. AH, sp. D, and sp. E; pD = Paulianodes 
sp. D; pC = Paulianodes sp. C; pE = Paulianodes sp. 
E; aG = Chimarra sp. AG; L = Chimarra sp. L; aE = 
Chimarra sp. AE; M = Chimarra sp. M; N = Chimarra 
sp. N; J = Chimarra sp. J; aF = Chimarra sp. AF; dA 
= Dolophilodes sp. A; dB = Dolophilodes sp. B; d. = 
Dolophilodes sp.; and K = Chimarra sp. K. 



Pisulia; Lepidostomatidae: Goerodes; Polycentro- 
podidae: Paranyctionphylax and Pseudoneur- 
eclipsis; and Leptoceridae: Adicella, Oecetis, Tri- 
anodes, Setodes, Leptocerus, and Athripsodini. 

On the large rivers, the caddisfly populations 
are dominated by the genus Macrostemum and the 
Athripsodini group, the latter remarkable for its 
high species diversity. We present an analysis of 
the family Philopotamidae, because this group 
was captured commonly and its alpha taxonomy 
is relatively straightforward. 



Philopotamidae 

Philopotamidae were caught at almost all sta- 
tions (Fig. 9-4 and Table 9-3). Three genera were 
identified: Paulianodes (three species), Dolophil- 
odes (two species), and Chimarra (eight species 
from the forest of the eastern slope and one spe- 
cies on the high plateau at the foot of Pic Boby). 
Three Chimarra species collected on an earlier 
mission to the region and outside the reserve 
along the Zomandao River (Mangoky Basin) were 
taken into consideration in order to compare the 
eastern and western slopes of the massif. 

All of the Paulianodes, Dolophilodes, and Chi- 
marra species collected are undescribed (Table 
9-4). Most species are being studied at the 
LRSAE, Antananarivo. Some of these taxa are 
represented in the collections of the Museum Na- 
tional d'Histoire Naturelle (MNHN), Paris; these 
species will be described by J. Olah, who kindly 
communicated his results. This taxonomic infor- 
mation allowed us to identify the species collected 
during this study and to summarize the available 
information on their geographical distributions. 



116 



FIELDIANA: ZOOLOGY 



Table 9-3. Distribution of the Philopotamidae in the RNI d'Andringitra. 



Sampling station: 15 12 11 10 9 6 5 4 2 3 1 

(west) (west) (east) (east) (east) (east) (east) (east) (east) (east) (east) 
Altitudinal zone (m): 2000 580 1625 1625 1210 810 810 810 720 720 720 



Chimarra sp. E 
Chimarra sp. AH 
Paulianodes sp. D 
Dolophilodes sp. 
Dolophilodes sp. A 
Dolophilodes sp. B 
Paulianodes sp. C 
Chimarra sp. AE 
Paulianodes sp. E 
Chimarra sp. J 
Chimarra sp. AF 
Chimarra sp. K 
Chimarra sp. M 
Chimarra sp. N 
Chimarra sp. AG 
Chimarra sp. L 





+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 






+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 






+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 






+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 








+ 




+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 


+ 







Table 9-3 lists the species found within the re- 
serve and the collecting stations. 

Conclusions — Caddisflies 

The caddisflies show a considerable faunal dis- 
similarity between the western and eastern slopes 
of the Andringitra Massif, specifically between 
the Zomandao River (Mangoky Basin) and the 
Sahavatoy and Iantara rivers (Manampatrana Ba- 
sin), respectively. There are some exceptions: at 
least one Athripsodini species, one Hydroptila 
species, and three Orthotrichia species occur up- 



stream in both basins. For the Philopotamidae the 
faunal dissimilarity is complete. 

It is interesting to examine this dissimilarity, 
because there are few available data on the geo- 
graphical distribution of aquatic insect species in 
Madagascar. (1) "Open area" species of the 
western slope — Chimarra sp. I, C. sp. D, and C. 
sp. E — also occur upstream of the Mandrare Ba- 
sin (especially on the Mananara River). Chimar- 
ra sp. AH is one of the most common Trichop- 
tera species known in Madagascar, especially in 
the southern and western regions. (2) Forest spe- 
cies on the eastern slope — Chimarra sp. N, C. 
sp. K, Paulianodes sp. D, P. sp. C, P. sp. E, 



Table 9-4. List of Philopotamidae species. 



Species 



Taxonomic status 



Proposed name 


Material 


fabienneae 


LRSAE 


langrandi 


LRSAE 


goodmani 


LRSAE 


andringitra 


LRSAE 


sahavatoyae 


LRSAE 


voromhola 


LRSAE 


sahanivorakyae 


LRSAE 


michaeli 


LRSAE 


erici 


LRSAE 


chatugana 


LRSAE and MNHN 


nadia 


LRSAE and MNHN 


watayana 


LRSAE and MNHN 


wigota 


LRSAE and MNHN 


atana 


LRSAE and MNHN 


elatrasoa 


LRSAE 


anadabolava 


LRSAE 


anka 


LRSAE and MNHN 



Paulianodes sp. C 
Paulianodes sp. D 
Paulianodes sp. E 
Dolophilodes sp. A 
Dolophilodes sp. B 
Chimarra sp. K 
Chimarra sp. N 
Chimarra sp. J 
Chimarra sp. L 
Chimarra sp. AE 
Chimarra sp. AF 
Chimarra sp. AG 
Chimarra sp. M 
Chimarra sp. I 
Chimarra sp. D 
Chimarra sp. E 
Chimarra sp. AH 



Gibon (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Olah (in prep.) 
Olah (in prep.) 
Olah (in prep.) 
Olah (in prep.) 
Olah (in prep.) 
Gibon (in prep.) 
Gibon (in prep.) 
Olah (in prep.) 



GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 



117 



Dolophilodes sp. A, and D. sp. B — are known 
only from the primary forests of the RNI d'An- 
dringitra. Chimarra sp. AF, C. sp. AE, C. sp. J, 
and C. sp. L are also known from the Namorona 
Basin in the primary forests of the Pare National 
de Ranomafana. Chimarra sp. AG occurs from 
the Mandraka River (Mangoro Basin) and the 
upper Mananara River. Chimarra sp. I has been 
collected from the latter locality. Finally, Chi- 
marra sp. M occurs on one tributary of the Man- 
goro River in the forested area south of Mora- 
manga (Anosibe an'ala). 

The "open area" or savannah species are 
known from other open areas (western and 
southern Madagascar), whereas "forest species" 
are known from other forests (eastern Madagas- 
car) or known only from the RNI d'Andringitra. 
From an ecological point of view, two distinct 
faunal assemblages are present on the island: a 
humid forest group and a savannah group. There 
is probably no endemicity with regard to 
streams. However, because parameters account- 
ing for the distribution of Philopotamidae (phys- 
icochemistry of water, substrates, and nutrition 
resources) are dependent on the vegetation pre- 
vailing around the stream, the effects of the bi- 
ome are determinant factors. These biogeograph- 
ical results parallel results from studies conduct- 
ed on west African Cheumatopsyche (Statzner, 
1984) and Chimarra (Gibon, 1985), and they 
confirm that the geographical distributions of 
these lotic insects depend more on vegetation 
zones or biomes than on stream orders. 

A second point associated with the distribution 
of species on the eastern slopes (see Table 9-4) is 
that each of the four elevational zones (710, 820, 
1210, and 1625 m) appears to have a relatively 
distinct fauna. The highest sampled zone on the 
eastern slopes (1625 m) consists only of small 
streams, where Paulianodes sp. D was collected, 
together with a species belonging to Dolophilo- 
des. The torrent at 1210 m is where Paulianodes 
sp. C, Dolophilodes sp. A, and D. sp. B were 
found. The zone around 820 m had the highest 
species diversity, including Chimarra sp. AF, C 
sp. K, C. sp. M, and C sp. N. Chimarra sp. L 
was limited to rivers in the 710 m zone. The re- 
maining species were distributed in two zones: 
between 820 m and 1210 m {Chimarra sp. AE, 
C. sp. J, and Paulianodes sp. E), and between 710 
m and 820 m (C. sp. AG). 

It is premature to provide precise details on 
habitat specificity of these Philopotamid species 
given the period of study and the number of spec- 



imens collected. However, our current information 
shows a clear pattern of altitudinal distribution of 
the various species that can be divided into "fau- 
nal zones" (Table 9-3). 

If no endemicity is demonstrated at the level of 
the lower rivers, then the faunistic changes related 
to elevational variables would support the idea of 
local endemicity in the mountainous streams. This 
hypothesis is supported by the fact that the species 
collected in the 1210 and 1625 m zones are 
known only from the Andringitra Massif, whereas 
the majority of the species collected at lower el- 
evations (710 and 820 m zones) are also known 
from other forests (Ranomafana, Andohahela, or 
Moramanga). 

The three genera of Philopotamidae occurring 
on the Andringitra Massif are distinct clades 
(Ross, 1956). Paulianodes is endemic to Mada- 
gascar and is a primitive element of the fauna; it 
has probably been isolated since the beginning of 
the Cretaceous. On the other hand, the genus Chi- 
marra, which is distributed worldwide, has colo- 
nized Madagascar in more recent geological time 
(Miocene). Dolophilodes may occupy an inter- 
mediary position; its status is still poorly known. 
It is interesting to note that elements that represent 
the highest level of endemicity (the subfamily 
Paulianodinae) and that are among the oldest rep- 
resentatives of the Malagasy fauna are restricted 
to small, high-altitude tributaries in primary for- 
ests. Before this study, the genus Paulianodes was 
only known from one specimen (P. tsaratananae 
Ross, 1956), reported from identical environments 
on the Tsaratanana Massif. Other species belong- 
ing to Paulianodes and Dolophilodes are being 
studied at the LRSAE. All of these specimens 
were caught in intact primary forests (Ranoma- 
fana, Andohahela, and Montagne d'Ambre), and 
each locality has its own species assemblage. This 
confirms the hypothesis of endemism localized in 
high mountain areas. 

The highest species richness of Philopotamidae 
(six species) occurs in the intermediary forest 
streams (station 5 and 9). Small tributaries were 
probably undersampled because of the method 
used (light traps). In contrast, the faunas of the 
larger rivers (Sahavatoy and Iantara) are relativelj 
depauperate. This is due to the combination ol 
steep slopes, frequent torrents, and dramatic in- 
creases in water level that disrupt the benthic en 
vironment and suppress the diversification of tht 
bank ecosystems. 



118 



FIELDIANA: ZOOLOGY 



Other Aquatic Insects 

Among the other aquatic insects of relative im- 
portance collected during the Andringitra mission 
are Megaloptera and Plecoptera. 

Megaloptera — Several pupae belonging to 
two species were collected; one was found in the 
710, 820, and 1210 m zones and the second at 
1625 m. Three Megaloptera species are currently 
known from Madagascar — Protosialis afra Navas 
1936, P. madegasca Navas 1927, and Madachau- 
lioda torrentialis (Paulian, 1951). Protosialis spp. 
are known from the high mountain regions of 
Tsaratanana and Ankaratra (Paulian, 1951). The 
two species recorded in Andringitra seem to be- 
long to the genus Madachaulioda. At least one of 
the species is known to science. Because the pu- 
pae of other Malagasy species have not yet been 
described, we could not identify our specimens. 

Plecoptera — All known Malagasy Plecoptera 
belong to the genera Madanemura and Tsarane- 
mura (Paulian, 1951). Three species had been pre- 
viously described by Paulian (1959): Madane- 
mura andringitrinsis, M. perrieri, and M. descar- 
pentriesi. 

During the sampling of the RNI d' Andringitra, 
we collected only pupae of Plecoptera between 
750 and 1210m (stations 5 and 9). The specimens 
collected in 1993 are probably not representative 
of species described from the massif by Paulian 
(1959), because all of those taxa were taken either 
at the base of Pic Boby (within the Sohanihindra- 
no Plateau) or at the peak. In these areas, ranging 
from 2000 to 2600 m, the vegetation is sclero- 
phyllous forest. Malagasy Plecoptera are known 
from all the main massifs — Andringitra, Ankara- 
tra, Tsaratanana, and Montagne d'Ambre. In all 
known cases the species are endemic and repre- 
sent geographical isolates. 



General Conclusions 

The results presented in this chapter illustrate 
the main problem associated with research on 
tropical invertebrates. The vast majority of species 
collected during the mission are new to science. 
The proper study of these undescribed forms takes 
a considerable amount of time and needs to be 
integrated in a broad systematic review of the spe- 
cific group. Moreover, the importance of these 
data in regard to species conservation can only be 
determined after extensive fieldwork and collect- 



ing in order to gather data on geographical and 
ecological distribution. This type of information 
is generally lacking for most new species. 

On the basis of the few groups of aquatic in- 
sects that we have studied on Madagascar, the fol- 
lowing conclusions can be presented: (1) There 
are clear differences between the faunas found on 
the eastern slopes, in humid forest, and those on 
the western slopes, in open savannah. (2) For sev- 
eral groups there is elevational variation in the 
distribution of various taxa, although the precise 
parameters giving rise to this pattern (e.g., river 
substrates, water temperatures, and the importance 
of the overlapping canopy) are unclear. (3) Open 
area species from the western slopes generally 
have a broader geographical distribution than hu- 
mid forest species. Upstream species have more 
restricted ranges than downstream species. (4) In 
the case of Madagascar, elements presenting the 
greatest level of endemicity (Paulianodes, repre- 
senting a subfamily endemic to the island; Me- 
galoptera; and Plecoptera) are restricted to rivers 
associated with primary and high-altitude forests. 



Acknowledgments 

We are grateful to Mr. Abel Ralaiteferana for 
his help during the Andringitra mission. We ex- 
press our gratitude to the World Wide Fund for 
Nature for inviting the LRSAE group to partici- 
pate in the Andringitra mission, and to Steven 
Goodman for translating the text into English. 



Literature Cited 

Demoulin, G. 1966. Contribution a l'enide des Euthy- 
plociidae IV (un nouveau genre de Madagascar). An- 
nates de la Soctete" Entomologique de France, N.S., 2: 
941-949. 

1970. Ephemeroptera des faunes ethiopiennes 



et malgaches. South African Animal Life, 14: 24-170. 

Elouard, J.-M., and Sartori, M. In press. Probosci- 
doplocia (Ephemeroptera: Euthyplociidae), a singular 
plural. Proceedings of the VHIth International Mayfly 
Conference, 14-20 August 1995, Lausanne, Switzer- 
land. 

Fontaine, J. 1968. Contribution a l'&ude des Ephem- 
eropteres malgaches: La super f ami lie des Epheme- 
roidea. Bulletin Mensuel de la Soctet6 Linneenne de 
Lyon, 20(6): 228-242. 

Fraser, F. C. 1956. Faune de Madagascar. I. Odonates 
Anisopteres. ORSTOM/CNRS, Paris and Antananari- 
vo. 

Gibon, F.-M. 1985. Recherches sur les Trichoperes 



GIBON ET AL.: SPATIAL DISTRIBUTION OF AQUATIC INSECTS 



119 



d'Afrique Occidentale. 3 — Philopotamidae de Cote 
d'lvoire. Revue d'Hydrobiologie Tropicale, 18(1): 23- 
30. 

Kjener, A. 1963. Poissons, peches et pisculture a Mad- 
agascar. Centre Technique Forestier Tropical, 24: 1- 
160. 

Malicky, H., and Chantaramongkol, P. 1993. The al- 
titudinal distribution of Trichoptera in Mae Klang 
catchment on Doi Inthanon, northern Thailand: 
Stream zonation and cool- and warm-adapted groups. 
Revue d'Hydrobiologie Tropicale, 26(4): 279-291. 

Navas, L. 1936. Communicaciones entomologicas. 19 
Insectos de Madagascar (3 dmc serie). Revista de la Ac- 
ademia de Ciencias de Zaragoza, third series, 19: 
100-110. 

Paulian, R. 1949. Decouverte de l'ordre des P16cop- 
teres a Madagascar. M6moires Institut de Recherche 
Scientifique de Madagascar, seYie E, 4: 359-363. 

. 1951. Faune des eaux douces de Madagascar. 

Plecopteres et M6galopteres. M6moires Institut de Re- 
cherche Scientifique de Madagascar, s6rie A, 6(1): 53- 
61. 



. 1959. Nouveaux Plecopteres malgaches. M6- 

moires Institut de Recherche Scientifique de Mada- 
gascar, s6rie E, 23: 9-16. 

Peters, W. L., and Edmunds, G. F. 1964. A revision of 
the generic classification of the Ethiopian Leptophle- 
biidae (Ephemeridae). Transactions of the Royal En- 
tomological Society of London, 116(10): 225-253. 

Ross, H. H. 1956. Evolution and classification of the 
mountain caddisflies. The University of Illinois Press, 
Urbana, 213 pp. 

Schmidt, E. 1951. The Odonata of Madagascar (Zy- 
goptera). Memoires Institut de Recherche Scientifique 
de Madagascar, 6(1): 115-279. 

Statzner, B. 1984. Keys to adult and immature Hy- 
dropsychinae in the Ivory Coast (West Africa) with 
notes on their taxonomy and distribution. Spixiana, 
7(1): 23-50. 

Vayssiere, A. 1895. Description zoologique de VEu- 
thyplocia Sikorai, nouvelle espece d' Ephemeridae de 
Madagascar. Annales de la Societe Entomologique de 
France, 64(26): 297-306. 



120 



FIELDIANA: ZOOLOGY 



Chapter 10 

New Heptageniidae (Insecta: Ephemeroptera) 
from the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 

Michel Sartori and Jean-Marc Elouard 



Abstract 

Two new species of Heptageniidae (Insecta: Ephemeroptera) are described from the nymphal 
stage from the Reserve Naturelle Integrate d'Andringitra. One species belongs to the genus 
Afronurus and the second to the genus Thalerosphyrus. 

Resume 

Deux nouvelles especes d' Heptageniidae (Insecta: Ephemeroptera) sont decrites a partir des 
larves de la Reserve Naturelle Integrate d'Andringitra. Une d'entre elles appartient au genre 
Afronurus, la seconde au genre Thalerosphyrus. 



Introduction 

Malagasy mayflies (Ephemeroptera) belonging 
to the family Heptageniidae are poorly known. 
The first report from the island is a larva identified 
as Notonurus sp. by Demoulin (1973). In the fol- 
lowing years Edmunds (1975, 1979) mentioned 
undescribed species from Madagascar belonging 
to genera Afronurus, Compsoneuriella, and Thal- 
erosphyrus. In these publications he drew a gen- 
eral outline of the biogeography of these taxa. 

Three new taxa from the Reserve Naturelle In- 
tegrate (RNI) d'Andringitra are discussed here. 
The nymphs of two forms are described as new 
species for which the imagos are unknown, and 
the nymph of a third species is briefly discussed 
but not formally named. Genus attribution is pro- 
visional and awaits the availability of adults, es- 
pecially male imagos. For details on collection 
techniques and stations see Chapter 9. 

In order to avoid any confusion, some terms are 
defined, as follows (Hubbard, 1995). Nymph: a 
larva that possesses black or extremely dark wing 



pads and is almost ready to molt to the subima- 
ginal stage; larva: any other immature stage youn- 
ger than the nymph; and pecten(s): "any comb- 
like structure or organ" (Torre-Bueno, 1989); we 
use this term to name the "comb-shaped struc- 
tures" that are found on the outer margin of the 
galea-lacinia of the maxillae (see Fig. 10-25). 



Descriptions 

Afronurus matitensis Sartori & Elouard, new 

species 

(Figs. 10-1 to 10-16) 

Nymph — Total length without caudal filaments: 
male nymph up to 9.5 mm; female nymph up to 
11.0 mm. Overall coloration dark brown, pale 
brown head, dark brown thorax and abdomen; 
pale brown femora with a dark dorsal spot at mid- 
line and apex (Fig. 10-1); yellow-brown ventral 
side with the same spots as on dorsum; yellow- 



SARTORI & ELOUARD: NEW HEPTAGENIIDAE 



121 








^ 



vT 





Figs. 10-1 through 10-4. Afronurus matitensis n. sp. 1: dorsal view of nymph; 2: ventral view of nymph; 3: 
details of head of a female nymph; 4: prothorax (arrows indicate chitinous creases). 



122 



FTELDIANA: ZOOLOGY 



brown tibiae, dark brown tarsi; gray-brown ster- 
nites with paired small pale spots on each sternite; 
gray-brown meso- and metasternite with a dark 
brown transverse stripe (Fig. 10-2). Purple and 
blackish abdominal gills with visible tracheation; 
yellow-brown caudal filaments with dark brown 
rings. 

Head — Very broad head capsule, broadest part 
at level of the insertion of antennae; female head 
with posterolateral margins rectilinear to slightly 
concave (Fig. 10-3); male characterized by a 
slightly rounded head. Trapezoidal labrum, about 
four times wider than long (3.75-4.25); internal 
margin with a well-marked medium notch; round- 
ed lateral projections slightly bent backward (Fig. 
10-5). Mandibles with long sharp-pointed inci- 
sors, bearing six to eight small teeth in the internal 
margin. Asymmetrical prostheca, fine on the left 
mandible, thicker and slightly indented on the 
right one. Hypopharynx with lateral lobes round- 
ed and covered with long bristles up to the inner 
side (Fig. 10-6). Maxillae densely covered with 
scattered bristles (Fig. 10-8). Apex of galea-lacin- 
ia with about 14-16 pectens, fifth pecten with 1 1- 
12 long sharp-pointed teeth. Maxillary palpi 
three-segmented: segment I very long, at least the 
length of galea-lacina, with 18-22 stout bristles 
on the external margin; segment II narrower and 
equal in length to segment I; and segment III 
about 2.5 times shorter than segment II, with the 
apex gradually narrowing to form a sharp-pointed 
tip (Fig. 10-8). Apical part of segment II and seg- 
ment III form a brush made of bristles on their 
outer margins. Rib-shaped glossae, rounded api- 
cally. Paraglossae moderately bent laterally (Fig. 
10-7). Labial palpi three-segmented. Inner margin 
of segment HI markedly sinuous, apex of segment 
II and entire segment III covered with a brush 
structure similar to that of maxillary palpi (Fig. 
10-9). 

Thorax — Pronotum laterally rounded, with two 
chitinous creases along symmetrical line, but with 
no posterolateral projections (Fig. 10-4). Legs 
thick with dilated femora. Tarsal claws with two 
small subapical teeth. Outer margin of femora 
covered with long and fine bristles; dorsal surface 
of all femora covered with thick short heart- 
shaped spines (Fig. 10-10), longest not exceeding 
1.5 times width. 

Abdomen — No lateral projections on segment 
I, small projections on segment II through IV but 
increasing in length towards segment VIII, where 
they reach the middle of segment IX. Posterior 
margin of tergite V, with one to two rows of sharp 



microdenticles and a posterior row of stout spat- 
ulate scales with rounded apex (Fig. 10-11). Sev- 
en pairs of gills, gill VII without tracheal fila- 
ments (Figs. 10-12 to 10-16). Gills I elongated 
with subparallel margins and slightly acute apex; 
gills II-IV more rounded and slightly asymmetri- 
cal; gills VII elongated with rounded apex. Distal 
part of each segment of caudal filaments with a 
whorl of stout spines apically rounded, as well as 
long and fine bristles. 



Examined Material 

Holotype — One female nymph; Madagascar, 
Fianarantsoa Province, approximately 45 km S 
Ambalavao, RNI d'Andringitra, the Matitanana 
Basin, station 1, 717 m, 22°13'S, 47°01'E, 17 No- 
vember 1993. It is deposited in the Mus6e Can- 
tonal de Zoologie, Lausanne, Switzerland. The 
type locality is the original collection sample sta- 
tion number P0165 of the LRSAE field series, as- 
sociated with the "Biodiversity and Biotypology 
of Malagasy Rivers" program. 

Paratypes — One female nymph, one male 
nymph, five larvae, same data as for holotype; 14 
larvae at station 5 (P0167), 19 November 1993; 
seven larvae at station 4 (P0168), 20 November 
1993, RNI d'Andringitra (see Chapter 9); two lar- 
vae, Manampanihy River, 24°41'00"S, 46°53'39"E 
(P0091), 15 April 1992; one male nymph, Na- 
morona River, Ranomafana, Ifanadiana, 21°15'15"S, 
47°27'34"E (P0212), 17 April 1994. All speci- 
mens collected by J.-M. Elouard, F-M. Gibon, M. 
Sartori, and others. Some paratypes deposited in 
the Museum National d'Histoire Naturelle 
(MNHN), Paris, and Centre National sur l'Envi- 
ronnement, Antananarivo. 

Other material not considered as para- 
types — Nine larvae, stream near Ambatolampy, 
near confluence with the Namorona River, 7 Au- 
gust 1958; two larvae, Amborompotsy River, An- 
tsampandrano (forest station), 25 July 1958; five 
larvae, Andranomalona River (forest station An- 
dasibe), along Antananarivo-Toamasina Road, 3 1 
July 1958. Collected by F. Starmuhlner and J. 
Fontaine. 



Affinities 

Considering the morphology of its mouthparts, 
this new species probably belongs to the genus 
Afronurus. Afronurus matitensis resembles some 



SARTORI & ELOUARD: NEW HEPTAGENIIDAE 



123 




Figs. 10-5 through 10-11. Afronurus matitensis n. sp. 5: labrum; 6: hypopharynx; 7: glossae and paraglossae; 8: 
maxilla; 9: labial palp; 10: spines on dorsum of hind femora; 11: posterior margin of tergite V. 



124 



FIELDIANA: ZOOLOGY 




Figs. 10-12 through 10-16. Afronurus matitensis n. sp. 12: gill I; 13: gill III; 14: gill V; 15: gill VI; 16: gill VII. 



African species, such as A. oliffi Schoonbee and 
A. harrisoni Barnard, but it is easily distinguished 
by the shape of the gills (distinctly more asym- 
metrical in the African species), by the shape of 
the maxillary palpi (first segment shorter in A. 
oliffi and A. harrisoni), by the number of pectens 
of the galea-lacinia (more than 20 for the above 
two species), and by the shape of the spines on 
the dorsum of the femora (sharp-pointed for A. 
harrisoni; see Schoonbee, 1968). 



Thalerosphyrus josettae Sartori & Elouard, new 

species 

(Figs. 10-17 to 10-32) 

Nymph — Body length without caudal filaments: 
male nymph up to 9.5 mm, female nymph up to 
10.5 mm. Overall coloration chestnut brown to 
pale brown; evenly colored head and thorax. Ab- 
dominal tergites with distinctively dark brown on 
yellowish brown ground (Fig. 10-17). Sternites 
also distinctive, yellowish beige, with paired elon- 
gated dark brown spots and paired points all along 
the median line, the complete structure lamp- 
shade shaped (Fig. 10-18). Evenly yellowish 



brown colored legs, apex of femora of nymphs 
with dark brown spots. Brown tibia and tarsi dark 
brown. 

Head — Labrum about four times wider than 
long, with nearly rectilinear outer margins and a 
wide median notch. Lateral projections slightly 
bent backward (Fig. 10-21). Hypopharynx with 
lateral lobes rounded and covered with long bris- 
tles that do not extend to the inner side. Mandibles 
with indented incisors. Asymmetrical prostheca. 
Galea-lacinia of maxillae covered with scattered 
long bristles (Fig. 10-24). Apex with 14-16 pec- 
tens, fifth with 10-11 thick sharp-pointed teeth 
(Fig. 10-25). Maxillary palpi three-segmented. 
Segment I shorter than galea-lacina, with 25-28 
stout bristles on its outer margin. Segment II 
about 1 .2 times longer than the first. Segment III 
short, with a rounded apex (Fig. 10-24). Rib- 
shaped glossae, paraglossae moderately stretched 
toward lateral side (Fig. 10-23). Labial palpi 
three-segmented. Subrectilinear internal margin 
for third segment. 

Thorax — Prothorax markedly rounded on both 
sides with a beginning of rounded projections in 
latero-posterior part (Fig. 10-19). Prothoracic 
creases not heavily sclerotized. Well-developed 



SARTORI & ELOUARD: NEW HEPTAGENIIDAE 



125 



--• » 





17 



id 





Figs. 10-17 through 10-20. Thalerosphyrus josettae n. sp. 17: dorsal view of nymph; 18: ventral view of nymph; 
19: prothorax (arrows indicate chitinous creases); 20: details of thoracic margin (arrows indicate supracoxal spurs). 



supracoxal spurs for meso- and metathorax (Fig. 
10-20). Legs with stout tarsal claws generally 
with three sharp-pointed subapical teeth. Dorsum 
of hind femora with spines more than two times 
longer than wide, with diverging rounded to 
slightly opened apexes (Fig. 10-26). 

Abdomen — Lateral projections unmarked on 



segments I-VI, barely visible on segments VII 
and VIII. Posterior margin of tergite V presenting 
two to three rows of microdenticles as well as a 
row with teeth alternating with pyriform scales 
about 1.5 times longer than the teeth (Fig. 10-27). 
Seven pairs of gills, last pair without tracheal fil- 
aments. All gills are markedly asymmetrical and 



126 



FIELDIANA: ZOOLOGY 




Figs. 10-21 through 10-27. Thalerosphyrus josettae n. sp. 21: labrum; 22: hypopharynx; 23: labium; 24: maxilla; 
25: median pectens of galea-lacinia; 26: spines on dorsum of hind femora; 27: posterior margin of tergite V. 



SARTORI & ELOUARD: NEW HEPTAGENIIDAE 



127 








Figs. 10-28 through 10-32. 
VII. 



Thalerosphyrus josettae n. sp. 28: gill I; 29: gill III; 30: gill V; 31: gill VI; 32: gill 



with sharp-pointed apexes, gills V most prominent 
(Figs. 10-28 to 10-32). Caudal filaments with 
whorls of stout spines at articulation of each seg- 
ment. 



Examined Material 

Holotype — One female nymph, Madagascar, 
Amborompotsy stream, Antsampandrano (forest 
station), 25 July 1958. Collected by F. Starmuhl- 
ner and J. Fontaine. Deposited in the Musee Can- 
tonal de Zoologie, Lausanne. 

Paratypes — Six larvae, Andringitra Massif, 
station 14 (P0178; see Chapter 9), 29 November 
1993; two larvae, Fenoevo, on tributary of Man- 
ampanihy River, 24°41'00"S, 46°53'39"E (P0091), 
15 April 1992. Some paratypes deposited in 
MNHN and Centre National sur PEnvironnement, 
Antananarivo. 



Affinities 

Two genera are known whose larvae possess 
well-developed supracoxal spurs: Compsoneuriel- 
la Ulmer (syn. Notonurus Crass fide Gillies, 1984) 
and Thalerosphyrus Eaton. On the basis of the 
shape of the pronotum, the glossae of the labium, 
and the arrangement of scattered bristles on the 
ventral side of the galea-lacinia, this new species 
probably belongs to the genus Thalerosphyrus, 
according to Jensen (1972). The 10 known species 
are all from the Oriental Realm, except one from 
the Afrotropical Realm. Three are known only at 
the imaginal stage. The remaining species are T. 
ethiopicus Soldan, 1977 (Sudan); T. determinatus 
(Walker, 1853), T. sinuosus (Navas, 1933), T. su- 
matranus (Ulmer, 1939) (all from the Sunda Is- 
lands); T. vietnamensis (Dang, 1967) (Vietnam); 
T. bishopi Braasch and Soldan, 1986 (Malaysia); 
and T flowersi Venkataraman and Sivaramakrish- 



128 



FIELDIANA: ZOOLOGY 



nan, 1987 (South India). Thalerosphyrus josettae 
can be distinguished from all of these species ex- 
cept T. ethiopicus and T. sinuosus by the lack of 
well-developed lateral projections on the abdo- 
men, as well as by the shape of the gills. Gill I is 
more slender in T. ethiopicus, T. determinants, 
and T. sumatranus. Gill VII is especially elongat- 
ed in T. sinuosus (see Ulmer, 1939, p. 668), 
whereas in T flowersi it does not possess an acu- 
minate apex. 



Thalerosphyrus species A 

Young larvae similar to T josettae were also 
captured in the RNI d'Andringitra. These larvae 
differ from T. josettae in the uniform whitish col- 
oration of abdominal tergites VI and VII. Apart 
from tergite coloration, the other main differences 
are in the maxillae. In Thalerosphyrus sp. A the 
first segment of the palpi has 15-17 bristles on its 
outer margin, and the galea-lacina crown bears 
12-13 pectens, the fifth of which has 12-13 teeth. 
The remaining characteristics are similar to those 
of T. josettae, including pronotum shape, spines 
on the dorsum of femora, tarsal claws, and den- 
ticulation on posterior margin of tergite V. 

Because only young larvae of Thalerosphyrus 
sp. A were available for this study, it is possible 
that they represent younger stages of T josettae, 
and we refrain from giving them a specific epithet. 



that preferentially lives in the metarhitral zone. 
Thalerosphyrus josettae n.sp. was found at one 
station located in a zone of savannah on the west- 
ern slope of the massif at 1900 m. Water temper- 
ature at the time of sampling was at 19°C. Station 
physiognomy would imply that this species lives 
in small streams and would be characteristic of 
the epirthral zone. There are obvious ecological 
differences between these two species. A. matiten- 
sis lives in rivers of eastern humid forest, whereas 
T josettae appears to be restricted to warmer 
streams running through eastern coast savannah. 
In continental Africa, Afronurus is associated with 
mountain streams and turbulent waters (Schoon- 
bee, 1968; Gillies, 1984). Little is known about 
the ecological requirements of Thalerosphyrus 
larvae. They live in perennial streams or brooks 
where water flow is moderate, with shallow and 
warm water (Soldan, 1977, 1991; Braasch & Sol- 
dan, 1986; Venkataraman & Sivaramakrishnan, 
1987). This information fits what we actually 
know about the ecology of T josettae. Neverthe- 
less, both species have been collected once at the 
same place outside Andringitra (Fenoevo, on trib- 
utary of Manampanihy River), suggesting that 
more data are needed before a precise typology 
can be drawn for these two species. 



Acknowledgments 



Examined Material 

Four larvae from station 14 (P0178; see Chap- 
ter 9), 29 November 1993; two larvae, Amborom- 
potsy stream, Antsampandrano (forest station), 25 
July 1958. Collected by F. Starmuhlner and J. 
Fontaine. Deposited in the Museum Cantonal de 
Zoologie, Lausanne. 



Ecology of Heptageniidae from 
RNI d'Andringitra 

Afronurus matitensis n.sp. was found exclu- 
sively on the eastern slope of the Andringitra 
Massif, in primary and degraded forest at altitudes 
ranging between 715 and 750 m. Water tempera- 
tures at the time of sampling were 16°-18°C. On 
the basis of the stream order of the colonized riv- 
ers (medium to large), A. matitensis is a species 



We are deeply indebted to Prof. W L. Peters 
(Tallahassee, Florida) for his help in making avail- 
able to us some hard to obtain literature, as well 
as for his advice. Thanks are also due to L. Ruf- 
fieux (Lausanne) for her help and fruitful discus- 
sion. 



Literature Cited 

Braasch, D., and T. Soldan. 1986. Die Heptageniidae 

des Gombak River in Malaysia (Ephemeroptera). Rei- 

chenbachia, 24: 41-52. 
Dang, T. N. 1967. Nouveaux genres, nouvelles especes 

de la faune des invert£br£s des eaux douces et sua- 

matres du Nord Vietnam. Tap San Sinh Vat-dia Hoc 

VI, ser. 3,4: 155-165. 
Demoulin, G. 1973. Eph6me>opteres de Madagascar 

III. Bulletin de l'lnstitut Royal des Sciences Naturelles 

de Belgique, 49(7): 1-20. 
Edmunds, G. E, Jr. 1975. Phylogenetic Biogeography 

of Mayflies. Annals of the Missouri Botanical Garden, 

62: 251-263. 
. 1979. Biogeographical relationships of the Ori- 



SARTORI & ELOUARD: NEW HEPTAGENIIDAE 



129 



ental and Ethiopian mayflies, pp. 11-14. In Pasternak, 
K., and R. Sowa, eds. Proceedings of the Second In- 
ternational Conference on Ephemeroptera. Polish 
Academy of Sciences, Krakow, 312 pp. 

Gillies, M. T. 1984. On the synonym of Notonurus 
Crass with Compsoneuriella Ulmer (Heptageniidae), 
pp. 21-25. In Landa, V., et al., eds. Proceedings of 
the IVth International Conference on Ephemeroptera. 
Czechoslovak Academy of Sciences, 345 pp. 

Hubbard, M. D. 1995. Towards a standard methodol- 
ogy for the description of mayflies (Ephemeroptera), 
pp. 361-369. In Corkum, L., and I. Ciborowski, eds. 
Current directions in research on Ephemeroptera. Ca- 
nadian Scholars' Press, Toronto, 478 pp. 

Jensen, S. L. 1972. A generic revision of the Hepta- 
geniidae of the world (Ephemeroptera). Ph.D. thesis, 
University of Utah (unpublished). 

Navas, L. 1993. Insecta Orientalia. XII. Memoire Pon- 
tificia Accademia delle Scienze Nouvi Lincei, 17: 75- 
108. 

Schoonbee, H. J. 1968. A revision of the genus Afro- 
nurus Lestage in South Africa. Memoirs Entomolog- 
ical Society of Southern Africa, 10: 1-46 + 7 plates. 



Soldan, T 1977. Three new species of mayflies 
(Ephemeroptera) from the mist oasis of Erkwit, Su- 
dan. Acta Entomologica Bohemoslovaca, 74: 289- 
294. 

1 99 1 . An annotated list of mayflies (Ephem- 



eroptera) found in the Nam Cat Tien National Park, 
pp. 4-9. In Spitzer, K., et al., eds. Nam Cat Tien, 
Czechoslovak Vietnamese Expedition. November 
1989, Research Report. Institute of Entomology, CAS, 
Ceske Budejovice, 45 pp. 

Torre-Bueno, J. R. de la. 1989. The Torre-Bueno 
glossary of entomology, revised edition. American 
Museum of Natural History, New York. 

Ulmer, G. 1939. Eintagsfliegen (Ephemeroptera) von 
den Sunda-Inseln. Archiv fur Hydrobiologie, Supple- 
ment, 16: 443-692. 

Venkataraman, K., and Sivaramakrishnan, K. G. 
1987. A new species of Thalerosphyrus from South 
India (Ephemeroptera: Heptageniidae). Current Sci- 
ence, 56(21): 1126-1129. 

Walker, F. 1853. Catalogue of the specimens of neu- 
ropteroid insects in the collection of the British Mu- 
seum. London, 658 pp. 



130 



FIELDIANA: ZOOLOGY 



Chapter 11 

Two New Species of Simulium (Diptera: Simuliidae) 
from the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 

Jean-Marc Elouard, Theogene Pilaka, 
and Fabienne Ranaivoharindriaka 



Abstract 

Two new species of Simulium (Diptera: Simuliidae), Simulium metecontae and S. brunhesi, 
are described from the Reserve Naturelle Integrate d'Andringitra based on pupae. 

Resume 

Deux nouvelles especes de Simulies (Diptera: Simuliidae), Simulium metecontae et 5. brun- 
hesi, provenant de la Reserve Naturelle Integrate d'Andringitra sont decrites en se fondant sur 
les nymphes. 



Introduction 

Female Simulium (Diptera: Simuliidae) are he- 
matophagous and generally dependent on specific 
vertebrates for their feeding regime. Collection of 
females in the wild that are not anthropophagous 
is extremely rare. Male Simulium are floricolous, 
thus making capture infrequent. Further, adult Si- 
mulium are generally diurnal; therefore collection 
at night by light traps is rare and generally con- 
fined to cases when they are disturbed in their 
resting places. 

In contrast, collecting aquatic stages (larva and 
pupae) is relatively easy. For pupae, varying 
shapes of gills or respiratory filaments are the eas- 
iest and most frequently used characteristics for 
specific determination. Further, the number and 
shape of dorsal and ventral hooks (thoracic and 
abdominal), as well as the shape and structure of 
the cocoon, provide diagnostic characters. Corre- 
lation between pupae and imagos can generally 
only be established after pupae have been reared. 



Morphological characters used to determine 
the species of larvae are more difficult than those 
for pupae, and they are largely based on man- 
dibles, the post-genal bridge, and overall chae- 
totaxy. Correlation between these two stages is 
established by uncoiling seventh-stage larval 
gills, which often parallel the structures in the 
future pupae. Larvae at their last stages are not 
always found associated with pupae. For these 
reasons, Simulium is a genus for which system- 
atic characters are mainly based on pupal mor- 
phology. 

Two new Simulium species were discovered in 
November 1993 during the inventory of the Re- 
serve Naturelle Int6grale (RNI) d'Andringitra. 
One species was found in the humid forest zone 
on the eastern slope of the massif, whereas the 
other was discovered in streams running through 
the high plateau on the western slope. For infor- 
mation on collection localities, stations, and 
stream order, see Chapter 9. 



ELOUARD ET AL.: TWO NEW SPECIES OF SIMULIUM 



131 






Description 

Simulium metecontae Elouard & Pilaka, new 

species 

(Fig. 11-la-e) 

Holotype — Currently only known from the pu- 
pae. The cocoon is simple, without heel or pro- 
jection (Fig. 11-le). The overall weft of the co- 
coon is quite loose. Gills of pupae are made up 
of two filaments originating from a common stem 
and curving toward one another. Stem and gills 
together are lyre-shaped (Fig. 11- lc). Four finer 
filaments originate from the stem and gill com- 
plex. Two of these filaments make an apical tip, 
and the other two filaments branch between two- 
thirds and three-quarters of their length from the 
apex. The surface of the gill filament creases (Fig. 
11-1 d) . Dorsal fangs are arranged in series of four 
or five (Fig. 11- la). Ventral fangs are thick but 
not numerous (Fig. 11-lb). 

Smete-1 (one tube containing the body of the 
pupa preserved in 70% alcohol and slide Smete- 
1.1, with gills treated with Euparal) is designated 
as the holotype. It was collected in Fianarantsoa 
Province, approximately 45 km S Ambalavao, 
RNI d'Andringitra, station 4, on the Sahanivoraky 
River, at 735 m and 22°13'S, 47°01'E. It is de- 
posited at Museum National d'Histoire Naturelle 
(MNHN), Paris. The original collection sample 
station number is P0168 of the LRSAE collection 
field series (POxxx refers to sampled stations in 
the general CNRE/ORSTOM program— Biodiver- 
sity and Biotypology of Malagasy Waters). 

Paratypes — Station 1 (P0167), one pupa; sta- 
tion 9 (P0171), nine pupae; station 11 (P0175), 
one pupa. All were in the RNI d'Andringitra (see 
Chapter 9 for details of sites). Paratypes are de- 
posited in the Centre National de Recherches sur 
l'Environnement (Antananarivo), ORSTOM col- 
lection, Montpellier, MNHN, and the Field Mu- 
seum of Natural History (FMNH). 

Etymology — This species is dedicated to Mrs. 
Sophie Metecont, entomologist and friend. 

Biological notes — This species was found in 
small freshwater rivers (16°-18°C at time of sam- 
pling) at altitudes ranging from 500 to 1600 m. 
Pupae colonize herbaceous substrates in water. 
Pupae were located as single individuals; this fact 
may emphasize their relative rarity or their wide 
spatial distribution. Pupae of S. metecontae were 
always associated with other species, although 
this varied between stations. Sympatric species in- 
cluded Simulium gyas, S. pentaceros, and 5. 



iphias lOf. See Chapter 9 for further details on 
the ecology of this genus in the RNI d'Andrin- 
gitra. 



Simulium brunhesi Elouard & 
Ranaivoharindriaka, new species 
(Fig. ll-2a-f) 

Holotype — Currently only known from the pu- 
pae. The cocoon has no heel, but a structure pro- 
tecting the upper part of the head (Fig. 1 l-2e and 
1 l-2f). The weft of the cocoon is quite fine. Pupa 
has four subequal gill filaments originating from 
a common stem (Fig. 11 -2c). Each of these 
branches terminates in a fine flagellated extremity. 
Cuticle of branchial arm is reticulate (Fig. 1 l-2d). 
These gill filaments have a discontinuous and ir- 
regular reticulation when magnified. Backside 
fangs are set on first abdominal segments (seg- 
ments 2-4) (Fig. ll-2a). The eighth abdominal 
segment has a pecten composed of fine fangs. In 
contrast, ventral fangs are set on median segments 
(Fig. 11 -2b). 

Sbrun-1 (one tube containing the body of the 
pupa preserved in 70% alcohol and slide Sbrun- 
1.1, with gills treated with Euparal) is designated 
as the holotype. It was collected at station no. 14 
at 1900 m (see Chapter 9) and is deposited at 
MNHN. The original collection sample station 
number is P0179 of the LRSAE field series. 

Paratypes — Zomandao River, effluent of the 
Mangoky Basin on the Andringitra Massif, station 
no. 14 (P0178), seven pupae; station no. 15 
(P0179), two pupae, all collected in the RNI 
d'Andringitra (see Chapter 9 for details on col- 
lection stations). Paratypes deposited in Centre 
National de Recherches sur l'Environnement (An- 
tananarivo), ORSTOM collection, Montpellier, 
MNHN, and FMNH. 

Etymology — This species is dedicated to Dr. 
Jacques Brunhes, a researcher at ORSTOM. 

Taxonomic Notes — Several African Simulium 
species have gills consisting of four filaments. In 
Madagascar, S. tolongoinae has an arrangement 
similar to 5. brunhesi, with four gill filaments and 
a reticulated pattern. However, in S. tolongoinae 
the gill filaments become progressively narrower 
and are not made up of finer filaments, as in 5. 
brunhesi. Two African species, S. nigritarsis and 
S. katangae, have the same number and gill fila- 
ment arrangement, but without the flagellated tips. 

Biological Notes — Simulium brunhesi was 
found above 1800 m on the high plateau of the 



132 



FIELDIANA: ZOOLOGY 




1 mm 




1 mm 




Fig. 11-1. Pupae of Simulium metecontae. a: abdomen in dorsal view; b: abdomen in ventral view; c: gill; d: 
ornamentation of the gill; e: cocoon in lateral view. 



RNI d' Andringitra, near the source of the Zoman- 
dao River, Mangoky Basin. Waters in this area of 
the Andringitra Massif are warmer (19°-19.5°C at 
the time of sampling) at 1900 m compared to wa- 



ters on the eastern slopes (1 1°-16.5°C). At station 
14, 5. brunhesi was associated with S. imerinae 
and with 5. iphias 8f; at station 15 it was asso- 
ciated only with S. iphias 8f. For more informa- 



ELOUARD ET AL.: TWO NEW SPECIES OF SIMULIUM 



133 



1 mm 




0,05 mm 

Fig. 11-2. Pupae of Simulium brunhesi. a: abdomen in dorsal view; b: abdomen in ventral view; c: gill, d: 
ornamentation of the gill; e: cocoon with lateral view, f: cocoon in dorsal view. 



134 



FIELDIANA: ZOOLOGY 



tion on the ecology of RNI d'Andringitra Simu- 
lium see Chapter 9. 

Pupae of three Malagasy Simulium (5. mete- 
contae, S. brunhesi, and S. starmuhlneri) have 
thin gill filaments supported by thicker filaments. 
Simulium species belonging to the continental Af- 
rican Medusiforme group possess parallel fila- 
ment structure. However, in this group the number 
of thin filaments is greater and the overall shape 
of gills is very different from the three Malagasy 
species. It is still premature to discuss the evolu- 
tionary history of the three species, but the shape 
of fine secondary filaments is probably an impor- 
tant character to discern relationships. At this 



stage, it is not clear if these species belong to the 
same group or imago subgenus, or if the filament 
characters are convergent. 



Acknowledgments 

This study is part of the Biodiversity and Bio- 
typology of Malagasy Continental Waters (PEC 7) 
Program, jointly run by CNRE and ORSTOM. 
The program is financed through the Fonds 
d'Aide et de Cooperation (FAC) francais. We 
thank Steven Goodman for translating and cor- 
recting the manuscript. 



ELOUARD ET AL.: TWO NEW SPECIES OF SIMULIUM 



135 



Chapter 12 

Parasitic and Commensal Arthropods of Some 

Birds and Mammals of the Reserve 

Naturelle Integrale d'Andringitra, Madagascar 

Barry M. OConnor 



Abstract 

Examination of 165 specimens representing 21 species of mammals and 25 specimens rep- 
resenting 16 species of birds collected in the Reserve Naturelle Integrale d'Andringitra yielded 
numerous collections of parasitic and commensal arthropods. All mammals and 23 of 25 birds 
harbored arthropod associates. Preliminary identifications of symbionts are presented primarily 
at the family and occasionally generic level for the Acari. Parasitic insects belonging to the 
orders Diptera, Siphonaptera, and Phthiraptera are also reported. The following families of 
mites are reported from native Madagascar vertebrates for the first time: Parasitiformes: Pa- 
chylaelapidae; Trombidiformes: Demodicidae, Psorergatidae, Ereynetidae, and Syringophilidae; 
and Sarcoptiformes: Turbinoptidae and Cytoditidae. 

Resume 

L'examen de 165 specimens appartenant a 21 especes de mammiferes et 25 specimens ap- 
partenant a 16 especes d'oiseaux collectes dans la Reserve Naturelle Integrale d'Andringitra a 
permis la collecte de nombreux arthropodes parasites et commensaux. Tous les mammiferes et 
23 des 25 oiseaux collectes ont recele des arthropodes associes. L' identification preliminaire 
des organismes commensaux est presentee principalement par famille et par genre pour les 
acariens. Les insectes parasites appartenant a l'ordre des Diptera, Siphonaptera et Phthiraptera 
sont aussi repertories. 

Les families suivantes de mites font l'objet d'une premiere mention en qualite d' ectoparasites 
de vertebres endemiques de Madagascar: Parasitiformes: Pachylaelapidae; Trombidiformes: De- 
modicidae, Psorergatidae, Ereynetidae, Syringophilidae; Sarcoptiformes: Turbinoptidae, Cyto- 
ditidae. 



Introduction 

Madagascar harbors a unique assemblage of 
terrestrial vertebrates, which themselves harbor an 
extensive fauna of parasitic and commensal ar- 
thropods. Host groups with limited dispersal abil- 
ity (amphibians, most reptiles, and non-volant 
mammals) typically exhibit a high degree of en- 
demism, whereas many volant taxa (birds and 



bats) are more widely distributed. Study of the 
arthropod communities associated with these host 
groups is in its infancy. The present state of 
knowledge varies considerably among different 
taxa, with groups having known medical or vet- 
erinary importance having received the most in- 
tense study. The basic taxonomy (i.e., naming of 
species) in certain groups such as the fleas, ticks, 
and myobiid and atopomelid fur mites of mam- 



136 



FIELDIANA: ZOOLOGY 



mals has been worked out to a large extent (de 
Meillon, 1950; Lumaret, 1952; Hoogstraal, 1953; 
Fain, 1976, 1978; Uilenberg et al., 1980). Other 
diverse groups, such as the bloodsucking laelapid 
mites, chiggers, and most groups parasitizing 
birds and reptiles are almost completely unstud- 
ied. Uilenberg (1969) provided a listing of the 
parasites of domestic and laboratory animals in 
Madagascar, and Brygoo (1975) compiled a bib- 
liography of the vertebrate parasites of the region. 
Contemporary surveys of Madagascar's diverse 
vertebrate fauna can assist in expanding our 
knowledge of the associated parasite and com- 
mensal arthropod communities if host specimens 
can be preserved in a manner that preserves the 
associated fauna intact and uncontaminated. The 
present report provides a preliminary listing of 
parasitic and commensal arthropods recovered 
from mammal and bird specimens collected dur- 
ing the recent survey of the Reserve Naturelle In- 
tegrate (RNI) d'Andringitra. 



Methods 

A total of 165 mammals and 25 birds were ex- 
amined for arthropod associates. Specimens were 
collected by S. M. Goodman and other investi- 
gators during the survey of the RNI d'Andringitra 
in 1993. Specimens to be examined for parasites 
were kept separated from others during collecting, 
individually wrapped in cheesecloth, and fixed in 
formalin. The wrapped specimens were later 
transferred to 70% ethanol and loaned to my lab- 
oratory at the University of Michigan. There, each 
specimen was unwrapped and examined in detail 
using a dissecting microscope. Ectoparasites and 
commensal arthropods were removed with for- 
ceps and sorted by taxon and location on the host. 
The nasal passages of the hosts were flushed using 
a modification of the technique described by 
Yunker (1961) to remove upper respiratory en- 
doparasites. The process of preparing specimens 
on microscope slides for final identification is 
continuing, with a large number of previously un- 
described species expected. 

The following list is a preliminary report of the 
results of this sampling. Because final identifica- 
tions and descriptions of new taxa for most col- 
lections are still pending, I report here the para- 
site/commensal faunas for each host species, gen- 
erally at the level of family or occasionally genus 
or species. For each host species, the number of 



individuals examined is indicated following the 
species name, and the number of host individuals 
harboring a particular arthropod taxon is listed 
following the taxon name if more than one host 
was examined. Most of the following represent 
the first records of these taxa from the listed host 
species. 



Results 

Class Mammalia 

Order Insectivora 

Family Tenrecidae 

Microgale cowani (N = 16) 
Acari: Parasitiformes 
Ixodidae (9) 
Laelapidae (13) 
Acari: Acariformes 
Trombiculidae (16) 
Myobiidae 

Microgalobia sp. (7) 
Madamyobia sp. (5) 
Glycyphagidae (2) 
Atopomelidae 

Listrophoroides (Alistrophoroides) (12) 
Listrophoroides (Madlistrophoroides) (14) 

Microgale dobsoni (N = 1) 
Acari: Acariformes 
Trombiculidae 

Microgale gracilis (N = 1 ) 
Acari: Parasitiformes 
Ixodidae 
Laelapidae 
Acari: Acariformes 
Trombiculidae 
Glycyphagidae 
Atopomelidae 

Listrophoroides {Alistrophoroides) 
Listrophoroides {Madlistrophoroides) 

Microgale longicaudata (N = 5) 
Acari: Parasitiformes 

Ixodidae (1) 

Laelapidae (5) 
Acari: Acariformes 

Trombiculidae (4) 

Myobiidae (1) 

Atopomelidae (4) 



OCONNOR: PARASITIC AND COMMENSAL ARTHROPODS 



137 



Insecta: Siphonaptera 
undetermined family (1) 

Microgale melanorrhachis (N = 1) 
Acari: Acariformes 
Trombiculidae 
Myobiidae 
Microgalobia 
Madamyobia 
Atopomelidae 

Listrophoroides (Alistrophoroides) 
Listrophoroides (Madlistrophoroides) 

Microgale parvula (N = 7) 
Acari: Parasitiformes 
Ixodidae (6) 
Laelapidae (4) 
Acari: Acariformes 
Trombiculidae (7) 
Glycyphagidae (6) 
Atopomelidae 

Listrophoroides (Alistrophoroides) (6) 
Listrophoroides (Madlistrophoroides) (6) 

Microgale taiva (N = 18) 
Acari: Parasitiformes 
Ixodidae (13) 
Laelapidae (16) 
Acari: Acariformes 
Trombiculidae (18) 
Myobiidae 

Microgalobia (9) 
Madamyobia (2) 
Glycyphagidae (11) 
Atopomelidae 

Listrophoroides (Alistrophoroides) (14) 
Listrophoroides (Madlistrophoroides) (14) 

Microgale gymnorhyncha (N = 3) 
Acari: Parasitiformes 
Ixodidae (3) 
Laelapidae (1) 
Acari: Acariformes 
Trombiculidae (3) 
Myobiidae 

Microgalobia (3) 
Madamyobia (1) 
Glycyphagidae (1) 
Atopomelidae 

Listrophoroides (Alistrophoroides) (3) 
Listrophoroides (Madlistrophoroides) (2) 

Microgale sp. A (N = 2) 
Acari: Parasitiformes 
Ixodidae (2) 
Laelapidae (1) 



Acari: Acariformes 
Trombiculidae (2) 

Setifer setosus (N = 1) 
Acari: Parasitiformes 

Laelapidae 
Acari: Acariformes 
Trombiculidae 
Atopomelidae 

Tenrecobia pauliana (Lawrence, 1955) 

Order Chiroptera 

Family Vespertilionidae 

Miniopterus minor (N = 28) 
Acari: Parasitiformes 
Ixodidae (1) 
Spinturnicidae (24) 
Macronyssidae (28) 
Acari: Acariformes 
Trombiculidae (1) 
Myobiidae (22) 
Chirodiscidae (8) 
Sarcoptidae 

Nycteridocoptes sp 
Notoedres sp. (1) 
Insecta: Diptera 
Nycteribiidae (20) 
Streblidae 
Ascodipteron sp. (1) 

Myotis goudoti (N = 4) 
Acari: Parasitiformes 

Spinturnicidae (4) 

Macronyssidae (4) 
Acari: Acariformes 

Myobiidae (2) 

Demodicidae (1) 

Chirodiscidae (2) 

Sarcoptidae (1) 
Insecta: Diptera 

Nycteribiidae (4) 



(3) 



Order Carnivora 

Family Viverridae 

Galidia elegans (N = 2) 
Acari: Parasitiformes 
Ixodidae 

Ixodes sp. (2) 
Haemaphysalis sp. (2) 
Laelapidae (2) 



138 



FTELDIANA: ZOOLOGY 



Acari: Acariformes 
Trombiculidae (1) 
Atopomelidae (2) 



Order Rodentia 
Family Muridae 

Eliurus majori (N = 12) 
Acari: Parasitiformes 
Ixodidae (3) 
Laelapidae (10) 
Pachylaelapidae (1) 
Acari: Acariformes 
Trombiculidae (12) 
Psorergatidae 

Psorergates (2) 
Sarcoptidae (3) 
Atopomelidae 

Listrophoroides (Eulistrophoroides) (7) 
Listrophoroides (Pallistrophoroides) (7) 
Insecta: Phthiraptera: Anoplura 

unidentified family (12) 
Insecta: Siphonaptera 
unidentified family (1) 

Eliurus minor (N = 10) 
Acari: Parasitiformes 

Laelapidae (5) 
Acari: Acariformes 
Trombiculidae (10) 
Psorergatidae 

Psorergates (1) 
Glycyphagidae (4) 
Atopomelidae 

Listrophoroides (Eulistrophoroides) (4) 
Listrophoroides (Pallistrophoroides) (5) 
Insecta: Phthiraptera: Anoplura 
unidentified family (4) 

Eliurus tanala (N = 16) 
Acari: Parasitiformes 
Ixodidae (5) 
Laelapidae (10) 
Pachylaelapidae (3) 
Acari: Acariformes 
Trombiculidae (16) 
Demodicidae: Demodex (1) 
Ereynetidae (1) 
Glycyphagidae (12) 
Atopomelidae 

Listrophoroides (Eulistrophoroides) (8) 
Listrophoroides (Pallistrophoroides) (16) 



Insecta: Phthiraptera: Anoplura 

unidentified family (4) 
Insecta: Siphonaptera 

unidentified family (4) 

Eliurus webbi (N = 12) 
Acari: Parasitiformes 
Laelapidae (12) 
Pachylaelapidae (1) 
Acari: Acariformes 
Trombiculidae (12) 
Demodicidae 

Demodex (3) 
Glycyphagidae (1) 
Atopomelidae 

Listrophoroides (Eulistrophoroides) (5) 
Listrophoroides (Pallistrophoroides) (8) 
Insecta: Phthiraptera: Anoplura 

unidentified family (2) 
Insecta: Siphonaptera 
unidentified family (1) 

Gymnuromys roberti (N = 2) 
Acari: Parasitiformes 

Ixodidae (1) 

Laelapidae (1) 
Acari: Acariformes 

Trombiculidae (2) 

Glycyphagidae (1) 

Atopomelidae 

Listrophoroides (Pallistrophoroides) (2) 
Insecta: Phthiraptera: Anoplura 

unidentified family (2) 

Monticolomys koopmani (N = 2) 
Acari: Acariformes 
Trombiculidae (2) 
Sarcoptidae (1) 

Nesomys rufus (N = 15) 
Acari: Parasitiformes 

Ixodidae (3) 

Laelapidae (9) 

Pachylaelapidae (2) 
Acari: Acariformes 

Trombiculidae (15) 

Ereynetidae (2) 

Atopomelidae 

Listrophoroides (Eulistrophoroides) (3) 
Listrophoroides (Pallistrophoroides) (14) 
Insecta: Phthiraptera: Anoplura 

unidentified family (13) 
Insecta: Siphonaptera 

unidentified family (1) 



OCONNOR: PARASITIC AND COMMENSAL ARTHROPODS 



139 



Rattus rattus (N = 9) 
Acari: Parasitiformes 
Ixodidae (1) 
Laelapidae 

Laelaps nuttalli Hirst, 1915 (2) 
Echinolaelaps echidninus (Berlese, 1887) 

(1) 

Acari: Acariformes 

Trombiculidae (9) 

Ereynetidae (2) 

Myobiidae 

Radfordia ensifera (Poppe, 1896) (1) 

Glycyphagidae (3) 

Listrophoridae: Afrolistrophorus sp. (8) 
Insecta: Phthiraptera: Anoplura 

Polyplacidae 

Polyplax spinulosa (Burmeister, 1839) (3) 



Analgidae (5) 
Proctophyllodidae (4) 
Avenzoariidae (3) 
Trouessartiidae (5) 
Insecta: Phthiraptera: Mallophaga 
family undetermined (1) 

Phyllastrephus cinereiceps (N = 1) 
Acari: Acariformes 
Analgidae 
Proctophyllodidae 
Avenzoariidae 
Trouessartiidae 
Insecta: Phthiraptera: Mallophaga 
family undetermined 



Family Turdidae 



Class Aves 

Order Passeriformes 

Family Eurylaimidae 

Neodrepanis coruscans (N = 1) 
Acari: Parasitiformes 

Rhinonyssidae 
Acari: Acariformes 

Proctophyllodidae 

Trouessartiidae 

Neodrepanis hypoxantha (N = 2) 
Acari: Parasitiformes 

Rhinonyssidae (1) 
Acari: Acariformes 

Proctophyllodidae (2) 

Trouessartiidae (2) 



Family Pycnonotidae 

Phyllastrephus madagascariensis (N = 1) 
Acari: Acariformes 
Analgidae 
Proctophyllodidae 
Trouessartiidae 
Insecta: Phthiraptera: Mallophaga 
family undetermined 

Phyllastrephus zoster ops (N = 5) 
Acari: Parasitiformes 

Rhinonyssidae (1) 
Acari: Acariformes 

Ereynetidae (1) 



Copsychus albospecularis (N = 2) 
Acari: Parasitiformes 

Rhinonyssidae (1) 
Acari: Acariformes 

Trombiculidae (1) 

Analgidae (2) 

Proctophyllodidae (2) 

Avenzoariidae (1) 

Trouessartiidae (2) 

Epidermoptidae (1) 



Family Sylviidae 

Nesillas typica (N = 1) 
Acari: Parasitiformes 

Ixodidae 
Acari: Acariformes 

Analgidae 

Avenzoariidae 

Trouessartiidae 

Newtonia amphichroa (N = 1) 
Acari: Parasitiformes 

Rhinonyssidae 
Acari: Acariformes 

Avenzoariidae 

Cytoditidae 

Cryptosylvicola randrianosoloi (N = 1) 
no parasites observed 



Family Monarchidae 

Terpsiphone mutata (N = 1) 
Acari: Parasitiformes 



140 



FIELDIANA: ZOOLOGY 



Rhinonyssidae 
Acari: Acariformes 

Analgidae 

Proctophyllodidae 

Trouessartiidae 

Cytoditidae 
Insecta: Phthiraptera: Mallophaga 

family undetermined 



Family Timaliidae 



= 1) 



Oxylabes madagascariensis (N 
Acari: Parasitiformes 

Ixodidae 
Acari: Acariformes 

Trombiculidae 

Analgidae 

Proctophyllodidae 

Avenzoariidae 

Trouessartiidae 

Turbinoptidae 
Insecta: Phthiraptera: Mallophaga 

family undetermined 



Family Nectariniidae 

Nectarinia souimanga (N = 1) 
Acari: Acariformes 
Proctophyllodidae 



Family Zosteropidae 

Zosterops maderaspatana (N = 1) 
Acari: Parasitiformes 

Rhinonyssidae 
Acari: Acariformes 

Ereynetidae 

Syringophilidae 

Analgidae 

Proctophyllodidae 

Avenzoariidae 



Family Vangidae 

Hypositta corallirostris (N = 1) 
no parasites observed 



Family Ploceidae 

Ploceus nelicourvi (N = 1 ) 
Acari: Acariformes 

Analgidae 

Proctophyllodidae 

Trouessartiidae 
Insecta: Phthiraptera: Mallophaga 

family undetermined 

Foudia omissa (N = 4) 
Acari: Parasitiformes 

Ixodidae (1) 
Acari: Acariformes 

Analgidae (3) 

Proctophyllodidae (1) 

Avenzoariidae (3) 

Trouessartiidae (2) 



Literature Cited 

Brygoo, E.-R. 1975. Bibliographic de la Zoologie Par- 
asitaire des vertebras de Madagascar. Archives de 
l'lnstitut Pasteur de Madagascar, 44: 245-288. 

de Meillon, B. 1950. The Madagascar Siphonaptera. 
M6moires de l'lnstitut des Sciences de Madagascar, 
Sene A, 4: 67-73. 

Fain, A. 1976. Arachnides Acariens, Astigmata Listro- 
phoroidea. Faune de Madagascar, 42: 1-131. 

. 1978. Les Myobiidae d'Afrique au sud du Sa- 



hara et de Madagascar Acarina-Prostigmata. Annales 
du Musee Royal de l'Afrique Centrale, Serie 8°, Sci- 
ences Zoologiques, 224: 1-186. 

Hoogstraal, H. 1953. Ticks (Ixodoidea) of the Mala- 
gasy faunal region excepting the Seychelles; their or- 
igin and host-relationships; with description of five 
new Haemaphysalis species. Bulletin of the Museum 
of Comparative Zoology, Harvard University, 111: 
37-113. 

Lumaret, R. 1952. Insectes Siphonapteres. Faune de 
Madagascar, 15: 1-107. 

Uilenberg, G. 1969. Inventaire artropodes, protozoai- 
res et rickettsiales parasites des animaux domestiques 
et des animaux de laboratoire a Madagascar. Archives 
de l'lnstitut Pasteur de Madagascar, 38: 69-105. 

Uilenberg, G., H. Hoogstraal, and J.-M. Klein. 1980. 
Les tiques (Ixodoidea) de Madagascar et ler rdle vec- 
teur. Archives de l'lnstitut Pasteur de Madagascar, nu- 
mero special, 1-153. 

Yunker, C. E. 1961. A sampling technique for intra- 
nasal chiggers (Trombiculidae). Journal of Parasitol- 
ogy, 47: 720. 



OCONNOR: PARASITIC AND COMMENSAL ARTHROPODS 



141 



Chapter 13 

Blood Parasites from Birds in the Reserve 
Naturelle Integrate d'Andringitra, Madagascar 

Ellis C. Greiner, Michael S. Putnam, 
and Steven M. Goodman 



Abstract 

Six of the 10 blood smears collected from birds in the Reserve Naturelle Integrale d'An- 
dringitra showed blood parasite infection. 

Resume 

Sur dix frottis sanguins realises a partir d'echantillons collectes sur des oiseaux de la Reserve 
Naturelle Integrale d'Andringitra, six ont montre des infections sanguines parasitaires. 



Introduction 

There has been only one previous report on avi- 
an hematozoa from birds on Madagascar (Bennett 
& Blancou, 1974). This note reports the results of 
a recent small sample from the Reserve Naturelle 
Integrale (RNI) d'Andringitra, southeastern Mad- 
agascar. 



Methods 

During the netting operation conducted in the 
RNI d'Andringitra (Chapter 18), blood smears 
were obtained from six birds collected as speci- 
mens and from clipped claws of four others that 
were released. Smears were fixed in the field with 
methanol and prepared in the laboratory with 
Giemsa stain. 



Results 

Below we present the results from the RNI 
d'Andringitra survey. Field Museum of Natural 



History (FMNH) and field catalog numbers of 
S.M.G. (SMG) are also given for each bird spec- 
imen. 



Otus rutilus (FMNH 363793, SMG 6373)— 810 
m, 6 October 1993, with Haemoproteus. 

Philepitta castanea (FMNH 363798, SMG 
6368)— 810 m, 6 October 1993, with Leuco- 
cytozoon and microfilariae. 

Phedina borbonica — 720 m, 24 September 1993, 
no parasites found. 

Motacilla madagascariensis (FMNH 363815, 
SMG 6340)— 720 m, 25 September 1993, no 
parasites found. 

Phyllastrephus madagascariensis (FMNH 
363819, SMG 6340)— 720 m, 25 September 
1993, no parasites found. 

Terpsiphone mutata (rufous male) — 720 m, 26 
September 1993, single schizont of Plasmodi- 
um. 

Terpsiphone mutata (rufous male) — 720 m, 26 
September 1993, Haemoproteus or Plasmodi- 
um. 

Vanga curvirostris (FMNH 363831, SMG 



142 



FIELDIANA: ZOOLOGY 



6397)— 810 m, 10 October 1993, Leucocyto- 
zoon and microfilariae. 

Ploceus nelicourvi (male) — 720 m, 26 September 
1993, no parasites found. 

Foudia omissa (FMNH 363869, SMG 6396)— 
810 m, 10 October 1993, Leucocytozoon and 
Plasmodium. 



Discussion 

The following blood parasites were found in six 
of 10 individuals (60%): Haemoproteus, Leuco- 
cytozoon, Plasmodium, and microfilariae nema- 
todes. This prevalence is higher than that found 
by Bennett and Blancou (1974) on Madagascar, 
where they identified parasites in 14 of 64 (22%) 
birds sampled. The present study also contrasts 
with that of Bennett and Blancou (1974) in that 
Haemoproteus, as well as multiple infections, 
were found for the first time on Madagascar. Two 
studies of avian hematozoa on the Mascarene Is- 
lands found Leucocytozoon, Plasmodium, multi- 
ple infections, and combined prevalences of 40% 
on Mauritius, 54% on Reunion, and 29% on Ro- 
drigues. Haemoproteus was found only in the in- 
troduced Rock Dove (Columba livid), whereas 
microfilariae were absent (Peirce et al., 1977; 
Peirce, 1979). Twenty-two percent of the land 
birds sampled on Aldabra were infected with 



Haemoproteus, Plasmodium, or both, with micro- 
filariae found in wading birds (Lowery, 1971). 
The results reviewed above therefore suggest that 
blood parasites are not uncommon in land birds 
of the western Indian Ocean. 

Species of the families Culicidae, Simuliidae, 
and Ceratopogonidae, that elsewhere contain vec- 
tors of avian hematozoa, are known from Mada- 
gascar (Paulian, 1961), although their role in avi- 
an hematozoan transmission is unknown. Nor is 
the pathogenicity of the parasites found in this 
small survey known. 



Literature Cited 

Bennett, G. E, and J. Blancou. 1974. A note on the 
blood parasites of some birds from the Republic of 
Madagascar. Journal of Wildlife Diseases, 10: 239- 
240. 

Lowery, R. S. 1971. Blood parasites of vertebrates on 
Aldabra. Philosophical Transactions of the Royal So- 
ciety London B, 260: 577-580. 

Paulian, R. 1961. Faune de Madagascar. 13. La zo- 
ogeographie de Madagascar et des iles voisines. Office 
de la Recherche Scientifique, Paris, 484 pp. 

Peirce, M. A. 1979. Some additional observations of 
haematozoa of birds in the Mascarene Islands. Bul- 
letin of the British Ornithologists' Club, 99: 68-71. 

Peirce, M. A., A. S. Cheke, and R. A. Cheke. 1977. 
A survey of blood parasites of birds in the Mascarene 
Islands, Indian Ocean, (with descriptions of two new 
species and taxonomic discussion by M. A. Peirce). 
Ibis, 119: 451-461. 



GREINER ET AL.: BLOOD PARASITES FROM BIRDS 



143 



Chapter 14 

Elevational Variation in Soil Macroinvertebrates 
on the Eastern Slopes of the Reserve 
Naturelle Integrate d'Andringitra, Madagascar 

Steven M. Goodman, Philip P. Parrillo, 
Sam James, and Petra Sierwald 



Abstract 

A series of soil plots in the Reserve Naturelle Integrate d'Andringitra were sampled in three 
microhabitats (ridge, slope, and valley) within the 720, 810, 1210, and 1625 m elevational 
zones to assess variation within and between these sites in the soil macroinvertebrates. Of the 
166 morphospecies identified from the samples, 126 (76%) were restricted to a single elevation, 
21 (13%) occurred at two adjacent elevations, and 19 (11%) were identified from three or four 
zones. Within each altitudinal zone there was no clear relationship found between the three 
microhabitats and the density or diversity of soil macroinvertebrates. In general the overall 
species richness and density were lower than those found at other sites in the tropics, presum- 
ably due to sampling technique. Two sites at approximately the same elevation but with dif- 
ferences in the level of human-induced habitat degradation showed no marked differences in 
the soil macroinvertebrates. Total taxonomic diversity peaked at 1210 m, whereas the total 
number of individuals per plot increased steadily between the 720 and 1625 m sites. Thus, 
species richness (as measured by taxonomic diversity) and density (as measured by the number 
of individuals per plot) do not exhibit precisely the same altitudinal pattern. 

Resume 

Une serie de quadrats de sol a ete echantillonnee dans trois microhabitats (crete, versant et 
vallee) au sein des zones altitudinales suivantes: 720, 810, 1210, et 1625 m, arm d'evaluer les 
variations de la macrofaune endogee des trois differents sites. 

En general, la richesse et la densite globale des especes etaient inferieures a celles trouvees 
au sein d'autres sites tropicaux. Quelques taxons presentent des distributions altitudinales lim- 
itees, meme si la majorite des especes echantillonnees de facon satisfaisante presente de grandes 
variations altitudinales dans leur distribution. Au sein de chaque zone d' altitude, la plus haute 
densite de macrofaune endogee se trouve dans les quadrats localises dans les vallees. La ma- 
crofaune endogee ne presente pas de difference nette au sein des sites d' altitude equivalente, 
mais seulement au sein des habitats affectes par des niveaux variables de degradation d'origine 
anthropique. 

La diversite taxinomique globale semble atteindre un maximum a 1210 m, tandis que le 
nombre total d'individus par quadrat augmente graduellement pour les sites localises entre 810 
et 1625 m. Ainsi, la richesse specifique mesuree d'apres la diversite taxinomique, et la densite 
mesuree d'apres le nombre des individus par quadrat, ne montrent pas la meme tendance 
generate. 

144 FIELDIANA: ZOOLOGY 



Introduction 

A significant proportion of the animal biomass 
and species richness in tropical forests is soil and 
litter invertebrates (Stork, 1988). However, little 
is known about the density, diversity, and eleva- 
tional distribution of these invertebrates in the 
tropics. Olson (1994) studied leaf litter ants, spi- 
ders, and beetles along an elevational gradient in 
Panama between 300 and 2020 m and found little 
evidence of fine-grain elevational zonation in the 
insect assemblages. The mean elevational range 
of most invertebrate species in his sample was 
500 m, and about one-half of the species found in 
an elevational zone did not occur at 500 m in 
elevation above or below the zone. There was a 
mid-elevational peak in species richness at about 
800 m elevation. This is in contrast to the study 
of Collins et al. (1984) in the Gunung Mulu Na- 
tional Park, Sarawak, where soil macroinverte- 
brates declined with increasing altitude, from 
2,579 individuals per square meter at 130 m ele- 
vation to 145 individuals per square meter at 
2,376 m elevation. However, total invertebrate 
biomass does not necessarily parallel species rich- 
ness. In the Volcan Barva, Costa Rica, there are 
significant differences in litter invertebrates and 
soil invertebrates occurring in the first 15 cm of 
soil (Atkin & Proctor, 1988). There are also dra- 
matic seasonal differences in the density and spe- 
cies richness of soil invertebrates (Chiba et al., 
1975). In short, although the above comparisons 
are of different measures of two different micro- 
habitats in geographically isolated regions of the 
world, they confirm that few generalities can be 
made about soil and litter macroinvertebrates in 
tropical forests, particularly along an elevational 
transect. Confounding variables such as sampling 
techniques and seasonal variation further compli- 
cate interpretation. Moreover, soil and litter in- 
vertebrate densities and species richness are pre- 
sumably site or region specific, and they probably 
hold the key to understanding the distribution of 
numerous other organisms, such as insects, that 
feed extensively on macroinvertebrates. 

We are unaware of any studies of the soil mac- 
roinvertebrates on Madagascar. In this chapter we 
examine the distribution and species richness of 
the invertebrate macrofauna in three microhabitats 
(ridge, slope, and valley) within and between four 
elevational transects. Two of the transects were at 
approximately the same altitude, but one (720 m) 
was in a forest degraded by human activity, 
whereas the other (810 m) was in relatively intact 



forest. The other two transects (1210 and 1625 m) 
were in pristine forest (see Chapter 2 for a de- 
scription of the sites). In a subsequent chapter 
(Chapter 20), the data on soil macroinvertebrates 
are used to examine ecological correlates with the 
altitudinal distribution and species richness of In- 
sectivora. 



Methods 

In each elevational zone, pitfall buckets were 
set up to capture small mammals, reptiles, and 
amphibians (see Chapters 17 and 20). After the 
pitfall lines had been in place for between 6 and 
8 days, a series of soil plots were set up adjacent 
to each line. Five plots were sampled per pitfall 
line during daylight hours. Each plot measured 2 
m (length) X 2 m (width) X 100 mm (depth). 
Plots within a line were 20 m apart. The leaf litter 
above each plot was discarded, and the underlying 
soil was placed in rice sacks. Within 1 hour the 
soil was examined by hand, and all visible inver- 
tebrates were removed and preserved in alcohol 
for later sorting and determination. Earthworms 
were fixed in 12% formalin and subsequently 
transferred to alcohol. P. P. P. was responsible for 
all of the invertebrate identifications, with the ex- 
ception of Annelida and Arachnida; these were 
done by S.J. and PS., respectively. The level of 
taxonomic determination varied among the mac- 
roinvertebrate groups. Even though some organ- 
isms were identified only to order or family, most 
could be assigned to "morphospecies." Our des- 
ignation of species richness is a tabulation of the 
number of morphospecies. Voucher specimens are 
deposited at the Field Museum of Natural History 
(FMNH). 



Results 

A general summary of the macroinvertebrates 
recovered from each of the plots per pitfall line is 
presented in Table 14-1. 



Nematomorpha 

Two individuals of one morphospecies were re- 
covered from one plot at 1625 m. 



GOODMAN ET AL.: ELEVATIONAL VARIATION IN SOIL MACROINVERTEBRATES 



145 



Table 14-1. Summary of invertebrates recovered from soil plots along pitfall lines. 



Pitfall 




720 m 






810 m 






1210 m 






1625 m 




line 


1 


2 


3 


4 


5 


6 


7 


8 


9 


10 


11 


12 


placement* 


R 


S 


V 


R 


S 


V 


R 


V 


S 


V 


R 


S 


Nematomorpha 1 
























1/2 


Gastropoda 


























Subulinidae 1 


1/1 
























Annelida ~5 


1/3 






2/3 


3/5 


3/6 




2/5 


5/5 


2/6 






Arachnida 


























Aranea 


























Miturgidae 1 


1/1 






1/1 












1/3 


1/1 


1/3 


Amaurobidae 1 


1/1 
























Dipluridae 1 
























1/2 


Mygalomorpha 1 




















1/31 




1/28 


Scorpionidae 


























Buthidae 1 


1/1 
























Chilopoda 


























Geophilomorpha 11 


1/1 




1/1 


1/1 


1/1 




2/3 


1/3 


2/9 


3/19 


4/11 


5/22 


Scolopendromorpha 5 


1/5 


1/1 


1/2 


2/5 


2/5 




3/6 


2/3 


1/5 


1/1 


1/8 


3/14 


Lithobiomorpha 1 


1/2 




1/1 


1/2 


1/1 




1/1 








1/1 


1/1 


Scutigeromorpha 1 






1/1 


















1/1 


Crustacea 


























Isopoda 6 


1/4 


1/1 


2/4 


2/6 


1/2 




2/4 




1/20 


1/1 


2/8 


1/1 


Diplopoda 5 


1/1 






1/1 


1/4 


1/2 


1/2 


3/11 


4/9 


1/4 


2/2 


3/10 


Diplura 


























Diplura 


























Heterojapygidae 1 


1/2 


1/2 


1/1 


1/4 


1/2 


1/2 














Japygidae 1 


1/2 


1/1 


1/3 


1/4 






1/1 






1/3 


1/6 


1/15 


Insecta 


























Blattodea 


























Blaberidae 5 


1/1 




1/1 


2/2 


2/2 


1/1 


1/1 








1/1 


1/1 


Coleoptera 10 


1/1 






1/1 


1/1 




5/5 




1/1 


1/1 






Anthribidae 1 
















1/1 










Carabidae 4 


1/1 












1/4 


3/4 


1/1 








Cerambycidae 1 






1/1 




















Dermestidae 1 














1/1 












Elateridae 21 


2/3 


6/6 




3/3 


2/2 




2/3 


2/5 


2/2 


2/2 


3/4 


3/3 


Scarabaeidae 38 


1/1 


1/1 


4/8 


8/10 


7/7 


4/12 


4/5 


8/35 


4/19 


2/2 


2/2 


1/2 


Staphylinidae 4 




2/2 




2/4 


















Tenebrionidae 12 


2/12 


2/4 


1/2 


4/4 


4/5 




3/4 


4/18 


1/4 


2/4 




1/1 


Dermaptera 1 










1/1 
















Labiidae 2 


1/1 




2/3 


1/1 


















Diptera 12 






2/2 


1/1 


1/1 


1/1 


2/2 


3/3 


1/1 


2/2 


1/1 


1/1 


Hemiptera 


























Cicadidae 1 










1/2 


1/3 




1/2 


1/1 








Cydnidae 1 






1/1 




















Enicocephalidae 1 


1/1 






1/2 






1/1 












Hymenoptera 3 








1/1 






1/1 


1/1 










Formicidae 1 






















1/1 




Isoptera 


























Termitidae 2 


1/8 




1/10 


1/1 


1/1 




2/4 












Lepidoptera 2 








1/1 










1/1 








Orthoptera 


























Gryllacrididae 1 


1/2 








1/1 
















Gryllidae 1 




1/2 




1/1 




1/1 














Gryllotalpidae 1 












1/1 















146 



FIELDIANA: ZOOLCXT' 



Table 14-1. Continued. 



Pitfall 




720 m 






810 m 






1210 m 






1625 m 




line 
placement* 


1 
R 


2 
S 


3 
V 


4 
R 


5 

S 


6 
V 


7 
R 


8 
V 


9 

S 


10 
V 


11 
R 


12 

S 


Phasmatodea 1 










1/1 
















Thysanura 
Nicoletiidae 1 


1/14 


1/3 


1/3 


1/13 


1/3 




1/11 


1/1 


1/2 


1/5 


1/3 


1/7 


Total 
taxonomic 
diversityt 
per line 


8.4 

5-12 

2.97 


4.4 

0-10 

4.39 


5.6 

3-11 

3.21 


10.6 
7-14 
2.51 


8.6 

4-12 

2.97 


2.6 
0-7 
2.70 


8.8 

4-13 

3.70 


7.2 

2-15 

5.06 


8.0 

5-12 

3.31 


7.0 

3-10 

2.92 


6.8 

4-10 

2.39 


8.8 

5-16 

4.55 


per 

elevational 

zone 




6.1 

0-12 

3.74 






7.3 

1-14 

4.33 






8.7 

2-15 

3.98 






7.5 

3-16 

3.29 




Total 
individuals 
per line 


13.6 

6-22 
6.35 


5.0 

0-12 

5.09 


8.6 

3-16 

6.77 


14.2 
8-24 
6.18 


9.4 

4-13 

3.51 


4.8 

0-14 

5.45 


11.6 
5-20 
6.27 


16.8 
2-35 
13.08 


16.0 
7-34 
11.90 


16.8 
3-51 
19.63 


10.4 
5-14 
4.09 


24.8 
5-43 
17.53 


per 

elevational 

zone 




9.1 

0-22 

6.73 






9.5 

0-24 

6.22 






14.9 

2-35 
10.32 






16.7 
3-51 
14.87 





* R = ridge; S = slope; V = valley. Number after name is total number of morphospecies identified in combined 
samples. Data are presented as total number of morphospecies/total number of individuals per line. 

t Descriptive statistics are presented as mean, range (minimum-maximum), and standard deviation per line. In each 
line there are five different plots. 



Gastropoda 

The only record from the soil samples is one 
individual at 720 m. 



er percentage of the biomass, because A. diffrin- 
gens is much larger. 



Arachnida 



Annelida 

Earthworms were recovered from all elevation- 
al zones, but not from each plot. The highest den- 
sity of earthworms was recorded in the 810 and 
1210 m zones. Five was the maximum number of 
morphospecies recorded along any pitfall line 
(1210 m, slope). Only one of three lines at 1625 
m yielded earthworms. In general, at 810, 1210, 
and 1625 m the greatest density of earthworms 
was in valley bottom plots (Table 14-1). 

In the 810 m zone the majority of collected 
earthworms belonged to one introduced species, 
Amynthas diffringens, a native of eastern Asia. 
This species was not recorded in the other three 
transect zones. Amynthas diffringens has been 
widely transported throughout the world and is 
now present in many temperate and tropical areas 
(Gates, 1972). Madagascar native species were 
present in the 810 m transect, but they represented 
only about 10% of the total individuals and a low- 



The only spider specimens represented in the 
soil samples by adults (two females) belonged to 
the Malagasy genus Uduba Simon, 1980 (Family 
Miturgidae). Presently, the genus contains two 
species, U. dahli and U. madagascariensis, and it 
is currently under revision by C. E. Griswold 
(pers. comm.). Both adult females belong to the 
same morphospecies and may be conspecific with 
Uduba spp. as figured by Griswold (1993, figs. 15 
and 16). The subadult Uduba specimens are cer- 
tainly congeneric; whether they are conspecific 
cannot be determined at this point. Members of 
Uduba apparently occur at least over wide por- 
tions of the eastern humid forest and have been 
collected from Pare National de Ranomafana, 
about 120 km northeast of the RNI d'Andringitra 
(Griswold, 1993:7). Uduba are represented in 
samples at 725, 810, and 1625 m, and they occur 
in both pristine forest and regions with human dis- 
turbance. 

The low number of spiders recovered from the 



GOODMAN ET AL.: ELEVATIONAL VARIATION IN SOIL MACROIN VERTEBRATES 



147 



soil samples appears to be a result of the collect- 
ing method, because only soil below the leaf litter 
was examined. According to C. E. Griswold (pers. 
coram.), Uduba live in subterranean burrows. 



Chilopoda 

Eighteen species of Chilopoda were recovered 
from the soil samples. Of these, the Geophilo- 
morpha were the most dominant, representing 1 1 
species or 61% of the centipede fauna. Diversity 
was lowest at the two lower elevations, with sin- 
gle species recorded at each of the 720 and 810 
m sites. Within the Geophilomorpha there were 
positive correlations between elevation and mor- 
phospecies diversity (df = 3, r 2 = 0.99, P - 
0.007) and elevation and density (df = 3, r 2 = 
0.92, P - 0.04). The Scolopendromorpha repre- 
sented 28% of the recovered chilopods, with in- 
creasing density as a function of elevation (df = 
3, r 2 = 0.97, P = 0.01). The Lithobiomorpha and 
Scutigeromorpha (Scutigeridae) were each repre- 
sented by a single species, and both groups had 
broad altitudinal distributions but low densities. 
Overall, there was a strong positive correlation for 
the Chilopoda between elevation and diversity (df 
= 3, r 2 = 0.99, P = 0.001) as well as between 
elevation and density (df = 3, r 2 = 0.92, P = 
0.04). Seven species (39%) were found at a single 
site. On the basis of these samples (Table 14-1), 
Chilopoda showed no significant preference for 
ridges, slopes, or valleys. 



Crustacea-Isopoda 

Six morphospecies were recovered from the 
plots, and density peaked at 1210 m. Isopoda 
showed a gradual replacement of species from 
720 to 1625 m. Three species were found only 
along a single line. No correlation was found be- 
tween diversity or density and microhabitat. 



Diplopoda 

Five morphospecies were identified, three of 
which showed broad altitudinal distributions, oc- 
cupying at least three elevational zones. Two spe- 
cies were recorded only at single plots. Diversity 
and density peaked at 1210 m, with five morpho- 
species and 22 individuals, respectively. No rela- 



tionship was found between microhabitat and den- 
sity. 



Diplura 

Two families were identified in the soil sam- 
ples. The Heterojapygidae were restricted to the 
lowland forest below 810 m elevation, whereas 
the Japygidae were distributed across the com- 
plete elevational range sampled and were distinct- 
ly more common in the 1625 m zone. 



Insecta 

In general, a wide variety of insects, many of 
which were rare (e.g., Hemiptera: Cicadidae, Cyd- 
nidae, and Enicocephalidae; Lepidoptera; Orthop- 
tera; Phasmatodea; and Hymenoptera), were iden- 
tified from the soil samples. 

Thysanura — A single morphospecies was col- 
lected in the soil samples along the complete al- 
titudinal gradient. Densities were relatively con- 
stant at each elevation. Ridge microhabitats 
seemed to support greater populations at 720, 8 1 0, 
and 1210 m than slopes or valleys. 

Blattodea — Five morphospecies were recov- 
ered from the soil samples. Three were found with 
distributions ranging across three elevational sites, 
and two were found only at single sites. Roaches 
occurred in low densities in all transects, reaching 
a peak in both diversity and density at 810 m. No 
recognizable association with microhabitat was 
found. 

Dermaptera — Three morphospecies were iden- 
tified from the transect samples. Two were found 
only at 810 m and the third at 720 m. 

Coleoptera — Beetles represented the majority 
of the species diversity from the soil samples, 
composing 55% of the taxa recovered. Of these, 
77% were Elateridae, Scarabaeidae, and Tenebri- 
onidae. The remaining families consisted of 
Anthribidae (1 morphospecies), Carabidae (4), 
Staphylinidae (4), Dermestidae (1), Cerambycidae 
(1), and 10 specimens that could not be deter- 
mined to family. Nineteen (90%) of these 21 mor- 
phospecies were found only along single lines. No 
clear microhabitat specificity was found for any 
beetle group. Carabids showed maximum diver- 
sity and density at 1210 m. Staphylinidae were 
not present above 810 m. Twenty-one morpho- 
species of Elateridae were identified from the 
samples, 17 (81%) were found at single sites, and 



148 



FIELDIANA: ZOOLOGY 



species diversity and density were evenly distrib- 
uted across the transect. The most abundant beetle 
family was the Scarabaeidae, with 38 morpho- 
species identified from the soil samples. This 
group showed elevational specificity, with 32 
(84%) of the taxa found within a single line. Scar- 
abaeid density peaked at 1210 m. Twelve mor- 
phospecies of Tenebrionidae were identified from 
the samples. Four of these were collected from 
one line, and this group reached maximum density 
at 1210 m. 

Diptera — Within the soil samples, Diptera 
were not common. Of the 1 2 morphospecies iden- 
tified, 1 1 (92%) were found at single sites. 

Discussion 

The maximum density of earthworms recorded 
in the RNI d'Andringitra soil samples was seven 
individuals per square meter at 810 m, substan- 
tially less than at many other tropical forest sites. 
For example, densities (individuals/m 2 ) have been 
reported as high as 401 in Volcdn Barva, Costa 
Rica (Atkin & Proctor, 1988), and 121 at Chajul, 
Mexico (Fragoso & Lavelle, 1987). In general. 
New World tropical forest sites tend to have a 
higher density and biomass of earthworms than 
Old World sites, and islands tend to have a lower 
density and biomass than continental sites (Fra- 
goso & Lavelle, 1992). Furthermore, there is con- 
siderable variation in density and species richness 
along elevational transects. Atkin and Proctor 
(1988) sampled earthworms in Costa Rica every 
500 m between 100 and 2600 m elevation; they 
found the highest density at 2000 m and the high- 
est biomass at 500 m. On Mt. Kosciusko, Austra- 
lia, there is little difference in the species richness 
along an elevational transect between 910 and 
2100 m elevation, although there is some species 
turnover (Wood, 1974). 

In general, the number of insects recovered 
from the soil plots was distinctly less than in other 
tropical forest sites. For example, Olson (1994) 
recorded up to 63 morphospecies of staphylinid 
beetles within an elevational zone in litter samples 
in Panama, whereas in the RNI d'Andringitra a 
maximum of two morphospecies was identified 
from soil samples in a single elevational zone. 
These differences certainly reflect collecting tech- 
niques (Leakey & Proctor, 1987), specifically, that 
leaf litter was not used in the RNI d'Andringitra 
samples. On the other hand, certain groups such 



as Annelida and Chilopoda showed equivalent or 
higher levels of species richness at Andringitra 
compared to Olson's study site in Panama (Olson, 
1994). 



Microhabitat Variation 

Within each elevational zone the soil plots were 
placed in three microhabitats: ridges, slopes, and 
valleys. In contrast to vertebrates captured in the 
pitfall traps adjacent to the soil plots (see Chapter 
20), no clear differences were found between mi- 
crohabitats in the taxonomic diversity or density 
of soil macroinvertebrates (Table 14-1). At the 
two highest elevations, the taxonomic diversity 
and the number of individuals recovered in the 
valley and slope plots were distinctly higher than 
in ridge plots. 



The Effects of Human Disturbance 

The 720 m elevation site was along a major 
forest trail between two villages and in close 
proximity to a small community of swidden ag- 
riculturalists, whereas the nearby 810 m site was 
relatively intact forest. Thus, differences in the 
soil macroinvertebrate communities of these two 
sites may reflect the affects of forest degradation. 
On the basis of presence or absence of various 
taxa, species richness as measured by the number 
of morphospecies, or the numbers of individuals 
recovered from the soil plots, no difference, in 
general, could be ascertained between the soil 
macroinvertebrate communities in the 720 and 
810 m zones. 

Human activity often results in the movement 
of soil and plants, and with them, earthworms 
(e.g., Michaelsen, 1903). Numerous species of 
earthworms, but only a small fraction of the total 
species pool, have become globally distributed by 
human activity. Where cultivation alters natural 
vegetation or other human disturbance changes 
the habitat, exotic earthworms are generally suc- 
cessful colonizers. However, at least in some tem- 
perate zone forests, habitat alteration is not nec- 
essary. The presence of exotic species is likely in 
small remnants of original vegetation, whereas 
native species are usually found in large areas of 
unaltered vegetation (Kalis/ & Dotson, 1989). 
Unfortunately for biodiversity conservation, the 
invasion of exotic species often means local ex- 
tinction or severe reduction of native species pop- 



GOODMAN ET AL.: ELEVATIONAL VARIATION IN SOIL MACROINVERTEBRATES 



149 



ulations (Smith, 1928; Lee, 1961; Ljungstrom, 
1972). 

Something like this may be in progress in the 
810 m zone, where an exotic earthworm is the 
most abundant species. It was probably intro- 
duced accidentally by the movement of plants, 
root stocks, or soil. However, why exotic earth- 
worms were more common at 810 rather than 720 
m (closer to agricultural areas) is unclear. One of 
us (S.W.J.) has observed trails through undis- 
turbed forests in the Caribbean Islands to be high- 
ways of earthworm dispersal. The same unknown 
process by which earthworms are dispersed along 
these trails may operate on the trail by the 810 m 
elevation site. Because earthworm self-dispersal is 
usually quite slow (maximal rates around 10 m 
per year), some human assistance is probably in 
effect. The distance between the nearest village or 
tavy and the pitfall trap sites is more than a few 
hundred meters. 

Another factor contributing to the colonization 
of many of the widely distributed, or peregrine, 
earthworms is uniparental reproduction. Though 
not all peregrins are parthenogenetic or otherwise 
capable of uniparental reproduction (earthworms 
are hermaphroditic), some are. The fully devel- 
oped Amynthas diffringens specimens collected at 
810 m showed some signs of lacking the male 
functions: loss of prostate glands, absence of ma- 
tured sperm on the male funnels, and absence of 
sperm received in spermathecae. 

Collection data on Madagascar earthworms, ei- 
ther native or introduced, are limited and scat- 
tered. Michaelsen (1897) listed 15 introduced spe- 
cies in addition to describing four new native spe- 
cies and giving new collection locations for six 
other native species. Gates (1972) is a more recent 
source of information on the exotics. The native 
species are generally known only from the origi- 
nal descriptions, in which localities are seldom 
given with any degree of precision. In view of the 
extensive habitat modification since the last of 
these descriptions was published in the early 20th 
century, it is unlikely that the native species still 
exist at type localities. 



Elevational Gradients 

Using the 810, 1210, and 1625 m sites as points 
along an elevational gradient, there are some ap- 
parent patterns in the soil macroinvertebrates. The 
taxonomic diversity at 810 m elevation is on av- 
erage 7.3 morphospecies per plot; at 1210 m, 8.7; 



and at 1625 m, 7.5. Even though there is a con- 
siderable amount of variance within each eleva- 
tional zone, the data suggest that the taxonomic 
diversity of the soil macroinvertebrates is highest 
at around 1210 m elevation. The total number of 
individual specimens per plot rises from 9.1 at 
710 m, to 9.5 at 825 m, 14.9 at 1210 m, and 16.7 
at 1625 m (df = 3, r 2 = 0.95, P = 0.03). Once 
again there is a large amount of variance within 
an elevational zone in the number of individuals 
found in each plot, but the density increases as a 
function of increasing altitude. These results sug- 
gest that species richness and species density do 
not show the same elevational trend. 



Acknowledgments 

We are grateful to John Slapcinsky of the Field 
Museum of Natural History for identifying the 
gastropod. Brian Fisher and an anonymous re- 
viewer provided important comments on an ear- 
lier version of this paper. 



Literature Cited 

Atkin, L., and J. Proctor. 1988. Invertebrates in the 
litter and soil on Volcan Barva, Costa Rica. Journal 
of Tropical Ecology, 4: 307-310. 

Chiba, S., T. Abe, J. Aoki, G. Imadate, K. Ishikawa, 
M. Kondoh, M. Shiba, and H. Watanabe. 1975. 
Studies on the productivity of soil animals in Pasoh 
Forest Reserve, West Malaysia. 1 . Seasonal change in 
the density of soil mesofauna: Acari, Collembola and 
others. Science Reports of the Hirosaki University, 22: 
87-124. 

Collins, N. M., J. M. Anderson, and H. W. Vallack. 
1984. 2. Studies on the soil invertebrates of lowland 
and montane rain forests in the Gunung Mulu Nation- 
al Park. Sarawak Museum Journal, 30: 19-33. 

Fragoso, C, and P. Lavelle. 1987. The earthworm 
community of a Mexican tropical rain forest (Chajul, 
Chiapas), pp. 281-295. In Bonvincini Paglai, A. M., 
and P. Omodeo, eds., On earthworms. Mucchi, Mo- 
dena, 562 pp. 

. 1992. Earthworm communities of tropical rain 

forests. Soil Biology and Biochemistry, 24: 1397- 
1408. 

Gates, G. E. 1972. Burmese earthworms. Transaction? 
of the American Philosophical Society, new series. 
62(7): 1-326. 

Griswold, C. E. 1993. Investigations into the phylog- 
eny of the lycosoid spiders and their kin (Arachnida 
Araneae: Lycosoidea). Smithsonian Contributions tc 
Zoology, 539: 1-39. 

Kalisz, P. J., and D. B. Dotson. 1989. Land-use his 
tory and the occurrence of exotic earthworms in th( 



150 



FIELDIANA: ZOOLOGY 



mountains of eastern Kentucky. American Midland 
Naturalist, 122: 288-297. 

Leakey, J. G., and J. Proctor. 1987. Invertebrates in 
the litter and soil at a range of altitudes on Gulum 
Silam, a small ultrabasic mountain in Sabah. Journal 
of Tropical Ecology, 3: 1 19-129. 

Lee, K. E. 1961. Interactions between native and intro- 
duced earthworms. Proceedings of the New Zealand 
Ecological Society, 8: 60-62. 

Ljungstrom, P. O. 1972. Taxonomical and ecological 
notes on the earthworm genus Udeina and a requiem 
for the South African acanthodrilines. Pedobiologia, 
12: 100-110. 

Michaelsen, W. 1897. Die Terricolen des Madagas- 
sischen Inselgebiets. Abhandlungen herausgegeben 
von der Senckenbergischen Naturforschenden Ge- 
sellschaft, 21: 217-252. 

. 1903. Oligochaeten von Peradeniya auf Cey- 



lon, ein Beitrag zur Kenntnis des Einflusses botan- 
ischer Garten auf die Einschleppung peregriner Thi- 
ere. Sitzungsberichte der koniglichen bohmischen Ge- 
sellschaft der Wissenschaften (Prague), 40: 1-16. 

Olson, D. M. 1994. The distribution of leaf litter in- 
vertebrates along a Neotropical altitudinal gradient. 
Journal of Tropical Ecology, 10: 129-150. 

Smith, E 1928. An account of changes in the earth- 
worm fauna of Illinois and a description of one new 
species. Bulletin of the Illinois Natural History Sur- 
vey, 17: 347-362. 

Stork, N. E. 1988. Insect diversity: facts, fiction and 
speculation. Biological Journal of the Linnaean Soci- 
ety, 35: 321-337. 

Wood, T. G. 1974. The distribution of earthworms 
(Megascolecidae) in relation to soils, vegetation and 
altitude on the slopes of Mt Kosciusko, Australia. 
Journal of Animal Ecology, 43: 87-106. 



GOODMAN ET AL.: ELEVATTONAL VARIATION IN SOIL MACROINVERTEBRATES 



151 



Chapter 15 

Shrimps (Crustacea: Decapoda: Atyidae) of the 
Reserve Naturelle Integrate 
d'Andringitra, Madagascar 

Mahefason Richard Andriamihaja 



Abstract 

Four Caridina species (Decapoda: Atyidae: Caridinae) were recorded in the humid forest 
streams of the Reserve Naturelle Integrale d'Andringitra. No Caridinae were found in the 
streams on the western slopes of the reserve. 

Resume 

Quatre especes de Caridina (Decapoda: Atyidae: Caridinae) ont ete trouves dans les cours 
d'eau de la zone de foret humide de la Reserve Naturelle Integrale d'Andringitra. En revanche 
aucun Caridinae n'a ete trouve dans les cours d'eau du versant ouest du massif de l'Andringitra. 



Introduction 

Malagasy freshwater shrimps belong to the 
families Atyidae and Palaemonidae (Crustacea: 
Decapoda). Both families contain genera that are 
consumed for food and have considerable eco- 
nomic importance. During the biological inven- 
tory of the Reserve Naturelle Integrale (RNI) 
d'Andringitra, shrimps were collected, all of 
which belong to the genus Caridina (Atyidae). 



Methods 

Shrimps were captured with a hoop net and pre- 
served in 70% ethanol. Determinations were made 
using the keys published by Holthuis (1965). Spe- 
cific determination of Caridina species can often 
be difficult due to insufficient information on in- 
dividual variation within a species. Thus, it is crit- 
ical to compare and define morphological differ- 
ences in ovigerous females and adult males of the 



same age. In this case specific determination was 
often based on adult males using numerous exter- 
nal morphological characters. 



Results 

Species of Caridina Edwards, 1 837 captured in 
the rivers of the RNI d'Andringitra include Car- 
idina xiphias, C hova, and C isaloensis. A fourth 
species was also collected that is not mentioned 
in Holthuis's (1965) keys and is probably new to 
science. This must be verified because many Car- 
idina species are described from Indo-Pacific wa- 
ters, and it may be known from elsewhere. 

Table 1 5- 1 lists all captures of shrimps made in 
the RNI d'Andringitra. No species of Caridina 
was taken at altitudes higher than 810 m (Fig. 
15-1). Increasing altitude is correlated with de- 
creasing water temperatures (see Table 9- 1 , Chap- 
ter 9). Stream order may also be important in re- 
stricting the ranges of this genus, because Cari- 



152 



FIELDIANA: ZOOLOGY 



Table 15-1. Species, collection sites, and relative abundance of Caridina shrimps collected in the RNI d'An- 
dringitra. For information on stations see Chapter 9. 









Stations 






Relative 
abundance 


Males 




Species 


1 


2 




3 


5 


Females 


Caridina hova 


+ 






+ 




13.3 


5 


3 


Caridina xiphias 


+ 


+ 




+ 


+ 


16.7 


10 





Caridina isaloensis 


















grandidieri 


+ 


+ 




+ 


+ 


55.1 


27 


6 


Caridina nov. sp.? 


+ 


+ 




+ 


+ 


15.0 


7 


2 



The plus sign ( + ) indicates the presence of the listed species at the particular station. 



dina species in Malagasy waters are known only 
at altitudes lower than 1210 m. At about 1210 m 
most waterways contain rapids and large water- 
falls, which presumably form a dispersal barrier 
for Caridinae. 

Three species of Caridina (C. xiphias, C. isal- 
oensis grandidieri, and Caridina sp.) were found 
in local sympatry at four stations (nos. 1, 2, 3, 
and 5) in the 720 and 810 m zones (Table 15-1). 
Caridina hova was found only at stations 1 and 3 
along the main portion of the Iantara River in the 
720 m zone. This is probably related to this spe- 



Western slope 



2000 -- 



1500-- 



1000-- 



SOO -- 



Eastern slope 

CRUSTACEA 

Caridina 



-- 8 o 




Stream order 1 
. Stream order 2 
. Stream order 3 



aa Stream order 4 
■ Stream Older 5 
i aunrj4cd a lal i on i 



Fig. 15-1. Altitudinal distribution of Caridina (Crus- 
tacea) in the RNI d'Andringitra on the basis of 1 1 sites 
on the eastern slopes and four sites on the western 
slopes. Key to species: C. n. sp. — Caridina new spe- 
cies?; C. h. — C. hova; C. x. — C. xiphias; and C. i. g. — 
C. isaloensis grandidieri. 



cies's preference for certain water conditions. 
Generally Caridina species occurring in rivers 
with slow-flowing waters are found either in the 
aquatic vegetation or in clumps of fallen leaves 
deposited on the river bottom. 



Caridina xiphias Bouvier, 1925 

In the RNI d'Andringitra this endemic species 
was found at all stations where shrimps were col- 
lected (Table 15-1). It is apparently relatively 
common near Didy, south of Lac Alaotra, and it 
is known to occur south of the Fianarantsoa re- 
gion (Holthuis, 1965). This species is distin- 
guished by the absence of up to three dorsal teeth 
of the rostrum placed behind the orbit, a long ros- 
trum reaching far beyond antennular peduncle, 
and the toothless ultimate portion of the upper 
margin of the rostrum. Ten adult males of C. xiph- 
ias were collected during the survey; they ranged 
in size from 7 to 26 mm. 



Caridina hova Nobili, 1905 

In the RNI d'Andringitra, C. hova was collect- 
ed only at stations 1 and 3 (Table 15-1). This spe- 
cies was previously known only from southeast- 
ern Madagascar. The type specimens were cap- 
tured in a small stream, called Mandromodromo- 
tra, located 20 km north of Tolagnaro. Caridina 
hova is distinguished from other members of this 
genus by the rostrum not usually reaching beyond 
the base of the third antennular segment and even- 
ly dentate or with a short distal portion lacking 
teeth. Furthermore, the dorsal teeth of the rostrum 
are prominent and strong. The specimens collect- 
ed were five adult males and three ovigerous fe- 
males. Adult males vary in total length from 8 to 



ANDRIAMIHAJA: SHRIMPS 



153 



, and ovigerous females vary from 17 to 



18 mm 
18 mm 



Caridina isaloensis grandidieri Bouvier, 1901 

Caridina isaloensis grandidieri was found at all 
stations where shrimps were collected (Table 
15-1). This subspecies is known from western 
Madagascar. The main area of occurrence of C. i. 
grandidieri is the Fandiamanana River (Fandra- 
manona River). This locality may be the Fandra- 
mana River, 22°25'S, 46°20'E, between Ihosy and 
the Andringitra Massif. 

Caridina isaloensis grandidieri is similar to C. 
hova except the dorsal teeth of the rostrum are 
distinctly smaller. The collection made in the RNI 
d' Andringitra is composed of 27 adult males and 
six ovigerous females. Specimens of the former 
species range in total length from 7 to 24 mm, 
and ovigerous females range from 17 to 24 mm. 
The presence of C. isaloensis grandidieri in the 
RNI d' Andringitra is unexpected. Previously this 
form was known only from warm waters of west- 



ern, southern, and northern Madagascar. Other 
species, such as C. longirostris, C. gracilirostris, 
and C. serratirostris, have broad distributions 
across much of the island. They are found in a 
variety of river types and biomes from savannah 
to forest. 



Caridina sp. 

A fourth species of Caridina taken in the RNI 
d'Andringitra cannot be identified using the keys 
of Holthuis (1965) and may well be new to sci- 
ence. As noted above, this is in need of further 
study because numerous Caridina species are de- 
scribed from Indo-Pacific waters, and the unde- 
termined Andringitra Massif species may be 
known from areas away from the island. 



Literature Cited 

Holthuis, L. B. 1965. The Atyidae of Madagascar. Me- 
moire Museum National d'Histoire Naturelle, Nou- 
velle ser., ser. A, Zool., 23(1): 1-48. 



154 



FIELDIANA: ZOOLOGY 



Chapter 16 

Crayfish (Parastacidae) and 

Crabs (Potamonidae) of the Reserve 

Naturelle Integrate d'Andringitra, Madagascar 

Bako Rabeharisoa 



Abstract 

Of the six described species of Malagasy crayfish, only one, Astacoides granulimanus, was 
recorded during the survey of the R6serve Naturelle Integrate d'Andringitra. They were col- 
lected in small brooks of the eastern humid forest of the massif. This species was not recorded 
on the western slopes. 

Two subspecies of the crab Hydrothelphusa agilis — H. a. agilis and H. a. madagascarien- 
sis — were recorded in watercourses ranging from 4 to 5 in stream order. The sympatric occur- 
rence of these two forms at the same elevations and in the same river system brings into 
question the subspecific arrangement of this species. 

Resume 

Parmi les six especes d'Ecrevisses malgaches, seule 1'espece Astacoides granulimanus a 6l6 
recoltee lors de la prospection de la Reserve Naturelle Integrate (RNI) d'Andringitra. Elle a 
616 trouvee dans de petits ruisseaux de la foret humide du versant Est. En revanche, aucun 
specimen n'a 6te recolte dans les cours d'eau du meme type du versant Ouest. 

Deux sous-especes du crabe Hydrothelphusa agilis: Hydrothelphusa a. agilis et H. a. mad- 
agascariensis ont 6t6 captures sur les cours d'eau d'ordre 4 et 5 de la RNI d'Andringitra. La 
presence simultande des deux sous-especes a la meme altitude et dans le meme bassin pose le 
probleme de la valeur taxinomique des deux sous-especes d' Hydrothelphusa agilis. 



Crayfish 

All named Malagasy crayfish (Crustacea: De- 
capoda) belong to the family Parastacidae and the 
genus Astacoides Gu£rin, 1839, and all species 
are endemic to the island. Madagascar is one of 
the few sites in the intertropical zone where cray- 
fishes are known. In general, the majority of cray- 
fishes are found in temperate areas, but they are 
also known in Honduras, Guatemala, New Guin- 
ea, Aru and Misol Islands, and northern Australia. 
No species is known on the African continent or 
in the Indian subregion. Astacoides has a closer 



affinity with the Tasmanian genus Astacopis than 
with any other genus in the world. Although Mad- 
agascar and Tasmania are geographically distant, 
Astacoides and Astacopis appear to share a com- 
mon recent ancestor. 

Until a few years ago, the genus Astacoides 
was considered monospecific, with Astacoides 
madagascariensis comprising several subspecies. 
Currently six species are recognized: Astacoides 
crosneri Hobbs, 1987; A. petiti Hobbs, 1987; A. 
granulimanus (Monod & Petit, 1929), A. mada- 
gascariensis (Edwards & Audouin, 1839), A. 
caldwelli (Bate, 1865), and A. betsileoensis (Petit, 



RABEHARISOA: CRAYFISH AND CRABS 



155 



1923). Their vernacular Malagasy names vary ac- 
cording to species and local dialects: orana, or- 
ambanonga, orambato, and oramboro. 



Methods 

Collections of crayfish within the Reserve Na- 
turelle Integrate (RNI) d'Andringitra were made 
by hand in small tributaries (stream order 1) of 
larger streams and rivers. Specimens were caught 
under rocks and inside many-galleried holes in the 
shallow streams. Water flow was slow, ranging 
from 1 to 3 liters per second. See Chapter 9 for 
further information on collecting sites and infor- 
mation on stream order. 



Results and Discussion 

Among the six crayfish species recorded in 
Madagascar, only one, Astacoides granulimanus, 
was collected in the RNI d'Andringitra. Speci- 
mens were obtained at altitudes ranging from 750 
(nine specimens) to 1650 m (two specimens) (Fig. 
16-1). The highest previous altitudinal record for 
crayfishes on the island is 1650 m (Hobbs, 1987). 
The particular conditions in the RNI d'Andringi- 
tra region, clear fresh forest streams at high alti- 
tude, presumably account for its local presence. 
The water systems of the upper mountain zone of 
the reserve are generally a series of falls between 
shelves. Presumably crayfishes find their way 
around waterfalls by passing on solid ground. As- 
tacoides granulimanus has the broadest distribu- 
tion of any Malagasy crayfish, from near Anta- 
nanarivo south to the Farafangana region (Hobbs, 
1987). This species is relatively common in the 
Ikongo Forest, northeast of RNI d'Andringitra 
(Hobbs, 1987). 

Astacoides are abundant in shallow freshwater 
with pH ranges of 4-6. Astacoides attain a rela- 
tively large size; adults range from 25 to 30 cm 
in total length and weigh between 100 and 130 g. 
They are nocturnal or crepuscular. The geographic 
range of Astacoides appears to be from Anjozo- 
robe in the north (northeast of Antananarivo) 
south to the Isaka River, a tributary of the Efaho 
River (northeast of Tolagnaro), for a total distance 
of about 700 km along the eastern coast. In the 
east, crayfish occur at altitudes from 500 to 1700 
m, and in westward drainages they are concen- 
trated in small high-mountain streams (1400 to 
1700 m). Compared to other tropical crayfish, the 



Western slope Eastern slope 


,m CRUSTACEA 




Astacoides & Hydrothelphusa 


,«l 2000- 
















11-- 










^\T™ 


1500- 

130 












9 (£&■) 




1000- 

12 o 

500- 






\ 7 

\ L 6 \ 

5J|r*T 3 






H.a.a. | 








h 






r^4 ' 


Lm 


1 Hxun. [ 


i 






Stream order 1 ^^_ Stream order 4 










Stream order 2 mMti Stream order 5 

Stream order 3 9 = sampled stations 














<^> Astacoides 1 1 Hydrothelphusa 



Fig. 16-1. Altitudinal distribution of Astacoides and 
Hydrothelphusa in the RNI d'Andringitra on the basis 
of 11 sites on the eastern slopes and four sites on the 
western slopes. Key to species: A. gr. — Astacoides gran- 
ulimanus; H.a.a. — Hydrothelphusa agilis agilis; and 
H.a.m. — H. a. madagascariensis. 



Malagasy species appear to live in the lowest al- 
titudes and highest temperatures. 



Crabs 

Methods and Results 

Crabs were collected either by searching under 
rocks or using worms at the same collecting sites 
reported for insects (see Chapter 9). Only one spe- 
cies, Hydrothelphusa agilis (Decapoda: Potamon- 
idae: Hydrothelphusinae) was captured in the RNI 
d'Andringitra, at stations 2 and 3 on water of 
stream orders 4 and 5. 

On the basis of keys provided in Bott (1965), 
the specimens are composed of two different sub- 
species, H. a. agilis and H. a. madagascariensis 
(Table 16-1). Hydrothelphusa a. agilis was caught 
using worms in Stillwater zones of streams and H. 
a. madagascariensis was collected by hand under 
rocks in stream riffles. These two subspecies were 
not sympatric at any station, although they were 



156 



FIELDIANA: ZOOLOGY 



Table 16-1. Distribution by station of Hydrothel- present in the same elevational zone. Whether 
phusa crabs captured in the RNI d'Andringitra. For in- m represent distinct taxa or two stages of me 
formation on stations see Chapter 9. ...«.-. . , , 

same species with different morphology and ecol- 

7~ ogy must be determined. 

Station OJ 

no. Subspecies Males Females 

2 Hydrothelphusa agilis agilis 3 3 Literature Cited 

3 Hydrothelphusa agilis agilis 3 Borr, R. von. 1965. Die Susswasserkraben von Mad- 

4 Hydrothelphusa agilis 4 3 agascar (Crustacea, Decapoda). Bulletin Museum Na- 
madagascariensis tional d'Histoire Naturelle, 2nd ser., 37: 335-350. 

Hobbs, H. H. 1987. A review of the crayfish genus 
Astacoides (Decapoda: Parastacidae). Smithsonian 
Contributions to Zoology, 443: 1-49. 



RABEHARISOA. CRAYFISH AND CRABS 157 



Chapter 17 

Amphibians and Reptiles of the Reserve 
Naturelle Integrale d'Andringitra, Madagascar: 
A Study of Kiev ational Distribution 
and Local Pandemicity 

Christopher J. Raxworthy and Ronald A. Nussbaum 



Abstract 

Amphibians and reptiles were surveyed on the eastern side of the Reserve Naturelle Integrale 
(RNI) d'Andringitra, between 650 and 2300 m elevation. This region of the reserve had not 
previously been studied. A total of 57 amphibian and 35 reptile taxa were found. Of these, 
68% of the amphibian and 88% of the reptile species are new records for the reserve. Pitfall 
captures suggest that species diversity and abundance are highest in valley bottom forest and 
lowest in ridge forest. 

No species occurs throughout the complete elevational range in the reserve, and 45% of the 
species are restricted to just one of the five elevational transects surveyed (650-800, 750-860, 
1100-1350, 1550-1700, and 1850-2300 m). Herpetofaunal species diversity declines with 
increasing elevation above 800 m. Above 1300 m, species diversity is 21% or less of the 
maximum diversity recorded at 800 m elevation. 

Twelve taxa are currently considered to be endemic to the Andringitra Massif, which includes 
species in both montane and low-elevation habitats. All other species (not locally endemic) are 
widely distributed on the eastern escarpment or other Central High Plateau massifs. Eight 
montane specialist species occur both at Andringitra and Ankaratra, and they provide evidence 
that montane habitats were once widespread on the Central High Plateau, possibly during 
glacial periods of the Pleistocene. Conservation of the herpetofauna at the RNI d'Andringitra 
will require protection of all habitats found within the reserve, of which the low-elevation rain 
forest and montane sclerophyllous forest appear to be the most vulnerable. 

Resume 

Un inventaire des amphibiens et des reptiles a ete effectue dans la partie est de la Reserve 
Naturelle Integrale (RNI) d'Andringitra, entre 650 et 2300 m d' altitude. Cette region de la 
reserve n'avait jamais ete inventorize auparavant. Un total de 57 amphibiens et de 35 reptiles 
a ete trouve, parmi lesquels 68% des especes d'amphibiens et 88% des especes de reptiles 
constituent de premiers recensements pour la reserve. Les captures realisees au moyen de pieges 
"pitfall" demontrent que la diversite des especes et l'abondance des individus sont maximales 
dans la foret des fonds de vallee et minimales dans la foret de crete. 

Aucune espece n'apparait sur l'ensemble des etages d'altitude presents au sein de la reserve 
et sur les cinq transects altitudinaux inventories (650-800, 750-860, 1100-1350, 1550-1700 
et 1850-2300 m), 45% des especes ont une presence limitee a un seul transect. La diversite 

158 FIELDIANA: ZOOLOGY 



specifique de 1'herpetofaune d'ecrott a partir de 800 m d'altitude. La diversite specifique au 
dessus de 1300 m est de 21% ou moins de la diversite maximale enregistree a 800 m d'altitude. 

Douze taxons sont maintenant considers comme etant endemiques au massif d'Andringitra, 
parmi lesquelles des especes de montagne et de basse altitude. Toutes les autres especes sont 
largement distributes au sein des forets de Test ou dans les massifs forestiers du haut-plateau 
central. Huit especes de montagne ont 6l6 recensees a la fois dans l'Andringitra et dans 
l'Ankaratra et ont fourni l'dvidence que les habitats montagnards 6taient autrefois repandus sur 
le haut-plateau central, probablement au cours des periodes glaciaires du Pleistocene. 

La conservation de 1'herpetofaune de la RNI d'Andringitra necessite la protection de tous 
les habitats que Ton trouve dans la reserve, parmi lesquels les forets humides de basse altitude 
et les forets sclerophylles de montagne qui semblent etre les plus vulneYables. 



Introduction 

The higher elevation forests and montane 
heathlands of Reserve Naturelle Integrate (RNI) 
d'Andringitra have been subject to opportunistic 
collecting of amphibians and reptiles for more 
than 70 years. Some of the earliest collections of 
frogs were made by Petit before 1924 and Millot 
in 1949 (Guibe\ 1978; Blommers-Schlosser & 
Blanc, 1991). In 1937, a small collection of am- 
phibians was made for the Museum National 
d'Histoire Naturelle (MNHN), Paris (Millot & 
Guibe, 1950). The first reptile species described 
from Andringitra was Lygodactylus montanus, 
which was catalogued in 1956 at MNHN (Pasteur, 
1964). The collector, although not named by Pas- 
teur, was probably Millot. 

In 1970-1971 a multidisciplined survey was 
made of RNI d'Andringitra. The survey included 
zoologists who (primarily C. P. Blanc) collected 
amphibians and reptiles (Paulian et al., 1971). 
Like earlier collections made in the reserve, this 
survey focused on the higher elevation habitats of 
the massif, with all specimens taken from 1200 to 
2658 m. Currently, a total of 28 amphibian spe- 
cies are recorded for RNI d'Andringitra (Blom- 
mers-Schlosser & Blanc, 1993). The only reptile 
list for the reserve is based on published and un- 
published observations and includes a total of 13 
species (Nicoll & Langrand, 1989). 

Based on existing collections and literature, it 
was clear that the herpetological diversity of RNI 
d'Andringitra was incompletely known. This was 
especially true for the lower elevation wet forests 
on the eastern slopes that had apparently never 
been surveyed for amphibians and reptiles. The 
elevational distribution of most species recorded 
from RNI d'Andringitra was also largely un- 
known. The specific aims of our survey were to 
determine the full elevational range of habitats 



and species found on the eastern slopes of the 
reserve. 



Study Sites 

Fieldwork was centered on five elevational 
transects: (1) 650-800 m; Iantara River (camp 1, 
720 m, near Ambarongy Village); 22°13'20"S, 
47°01'29"E; 14-21 November and 17 December 
1993; primary to degraded rain forest, tavy clear- 
ings. (2) 750-860 m; Sahavatoy River (camp 2, 
810 m, junction with Sahanivoraky River); 
22°13'40"S, 47°00'13"E; 22-29 November 1993; 
primary to degraded rain forest, tavy clearings. (3) 
1 100-1350 m; Volotsangana River (camp 3, 1210 
m). 22°13'22"S, 46°58'18"E; 30 November-7 De- 
cember 1993; primary rain forest. (4) 1550-1700 
m; Ridge E of Volotsangana River (camp 4, 1625 
m); 22°11'39"S, 46°58'16"E; 8-12 and 15-16 De- 
cember 1993; primary rain forest. (5) 1850-2300 
m; Kimora River (camp 5, 2075 m, 4 km NE of 
Pic Bory); 22°10'10"S, 46°56'40"E; 13-14 De- 
cember 1993; montane heathland. 



Methods 

The three members of the herpetological survey 
team were Nirhy Rabibisoa (DEA student, Univ- 
ersite d' Antananarivo), C.J.R., and Angelin Ra- 
zafimanantsoa (research guide, Montagne d'Am- 
bre). 

The survey was timed for the onset of the rainy 
season, when most species are breeding and ac- 
tivity is at its highest. Field techniques used to 
sample animals were (1) pitfall trapping with drift 
fences; (2) opportunistic day and night searching; 



RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 



159 




amphibians 
reptiles 



720 810 1210 1625 2075 
Transect 

Fig. 17-1. Herpetofaunal diversity at each transect. 



and high plateau of the RNI d'Andringitra. The 
majority of searching was done close to new trails 
made during the study, although ridges and river- 
banks were also used to orient search paths. Night 
searches were made using headlamps. 

The following information was recorded at the 
time of capture for each individual: date, time, 
longitude, latitude, altitude, microhabitat, and cir- 
cumstances of capture. Animals not retained for 
specimens were returned to the site of original 
capture. Voucher specimens were fixed in 10% 
buffered formalin and later transferred to alcohol. 
Liver and muscle were removed from represen- 
tatives of almost all species and frozen in liquid 
nitrogen. Color slides were taken of live individ- 
uals of most species. Frog calls were recorded 
where possible. Collected material was deposited 
at two main centers, the Museum of Zoology, 
University of Michigan, and the Department de 
Biologie Animale, Universite d' Antananarivo. 



(3) refuge examination (under and in fallen logs 
and rotten tree stumps; under bark; under rocks; 
in leaf litter, root-mat, and soil; and in leaf axils 
of Pandanus screw palms and Ravenala travel- 
ler's palm. 

The pitfall traps were buckets (275 mm deep, 
290 mm top internal diameter, 220 mm bottom 
internal diameter), sunk in the ground at 10 m 
intervals below a 100-m-long drift fence (Fig. 
17-1). The handles of the buckets were removed, 
and small holes (2 mm diameter) were punched 
in the bottom to allow water drainage. 

The fence (0.5 m high) was made from plastic 
sheeting stapled in a vertical position to thin 
wooden stakes. The fence bottom was buried 50 
mm deep into the ground using leaf litter and po- 
sitioned to run across the middle of each pitfall 
trap. A pitfall trap was positioned at both ends of 
the drift fence, with the other nine traps at 10 m 
intervals. The trap lines were checked each morn- 
ing and late afternoon. After heavy rain the buck- 
ets were sponged dry. At each camp a line was 
placed in each of the following forest types: ridge 
(along the crest of a ridge), slope (on a gradient, 
intermediate between ridge top and valley bot- 
tom), and valley (within 20 m of a stream in a 
valley bottom). Three lines were used in each of 
the lower four transect zones, and sampling was 
made over 6-8 days. A total of 12 lines were 
erected during the survey. 

Opportunistic searching and refuge examina- 
tion were made throughout the full elevation 
range of habitats available on the eastern slopes 



Results 

A total of 57 amphibian and 35 reptile species 
were recorded during this survey within the RNI 
d'Andringitra, with another three reptile species 
(Furcifer lateralis, Zonosaurus madagascariensis, 
and Z. ornatus) found at Ambalamanenjana Vil- 
lage, 12 km north of the reserve limit (Table 
17-1). Ten amphibian and seven reptile species 
have not been identified and may represent un- 
described taxa. Of the 57 amphibian species re- 
corded, 39 (68%) are new records for the reserve; 
of the 34 reptile species, 30 (88%) are new re- 
cords. 

Amphibian species diversity was greatest at 
810 m and lowest at 2075 m (Table 17-1; Fig. 
17-1). Reptile species diversity was greatest at 
720 m and lowest at 1625 m (Table 17-1; Fig. 
17-1). For both amphibians and reptiles, species 
diversity was much lower at 1550-2350 m than 
at lower elevations, 650-1350 m. 

The characteristics of the pitfall trap lines and 
captures are given in Table 17-2. A total of 124 
vertebrates were captured over 902 pitfall trap 
(bucket) days, giving an average vertebrate cap- 
ture rate of 0.14 per trap day (14% trap success 
per trap day). Pitfall lines captured seven amphib- 
ian, five reptile (all skinks), and 12 small mammal 
species. At the four transects where pitfalls were 
used, species diversity and capture rate were 
greatest in valley forest and lowest in ridge forest, 



160 



FIELDIANA: ZOOLOGY 



with the exception of 810 m, where the lowest 
species diversity and capture rate were in slope 
forest. The 1210 m pitfalls had higher capture 
rates and species diversity than the 720, 810, and 
1625 m transects, with the only exception being 
the 810 m valley forest line, which had a slightly 
greater species diversity. 



Discussion 
Species Not Surveyed 

Although many of the species found during this 
survey are new records for RNI d'Andringitra, 
other species reported for the reserve were not 
found. Table 17-3 is a list of the missing 21 am- 
phibian and 14 reptile species. Considering first 
the amphibians, the record of Plethodontohyla 
tuberata (Nicoll & Langrand, 1989) apparently is 
not supported by a voucher specimen, and Blom- 
mers-Schlosser & Blanc (1993) did not list this 
species for RNI d'Andringitra. The vouchers of 
the amphibian species reported by Blommers- 
Schlosser and Blanc (1991, 1993) and Glaw and 
Vences (1994) need to be examined to determine 
if they actually represent species different from 
those recorded in this study. Some "missing spe- 
cies" (e.g., Mantidactylus pliciferus, M. cornutus, 
M. brevipalmatus, M. spiniferus, Boophis lauren- 
ti, and B. untersteini) are easily confused with 
other species we recorded in the RNI d'Andrin- 
gitra. Three species, Scaphiophryne madagascar- 
iensis, Mantidactylus domerguei, and M. elegans, 
were missed; these species are probably restricted 
to the high-elevation forests and heathlands that 
were less intensively studied during this survey. 

Nicoll and Langrand (1989) recorded seven 
reptile species that are seemingly not represented 
by voucher specimens and probably represent per- 
sonal observations. Two oplurid species, Opiums 
cuvieri and O. cyclurus, are not known from lo- 
calities close to RNI d'Andringitra (Blanc, 1977), 
and they most likely represent misidentified O. 
quadrimaculatus. The most recent reviews of Ma- 
buya and Zonosaurus by Brygoo (1983, 1985) do 
not include RNI d'Andringitra as a locality for 
Mabuya aureopunctata and Zonosaurus laticau- 
datus. Two reptile species for which Andringitra 
vouchers exist were certainly missed by us. These 
are Mabuya boettgeri and Oplurus quadrimacu- 
latus (Blanc, 1977; Brygoo, 1983). Both species 
occur in the heathland on or near rocky outcrops 



at high elevation, and they are probably present 
in the 2075 m transect. They may have been over- 
looked because of the overcast and rainy condi- 
tions during the short period this transect was sur- 
veyed. 

Other reptiles that we missed in the reserve 
(Furcifer lateralis, Mabuya elegans, Zonosaurus 
ornatus, and Mimophis mahfalensis) were seen by 
us in degraded habitats around villages close to 
the reserve. Glaw and Vences (1994) also record- 
ed Lygodactylus pictus (near Sendrisoa) and Ma- 
buya madagascariensis. These identifications re- 
quire confirmation, although it is not clear if 
voucher specimens exist for these records. 



Sampling 

Pitfall traps yielded just one species of amphib- 
ian (.Plethodontohyla alluaudi) that was not ob- 
tained by opportunistic searching. All reptile spe- 
cies captured in pitfalls were also sampled by di- 
rect searching. Although pitfall traps did not sig- 
nificantly contribute to the herpetofaunal species 
inventory for the reserve, pitfalls did capture spe- 
cies otherwise represented by very few specimens 
(e.g., Plethodontohyla sp. 1, P. serratopalpebro- 
sa, Amphiglossus anosy, and A. macrocercus). 
Without pitfalls, several species would not have 
been recorded at some elevations. Pitfall traps 
also made a significant contribution to the mam- 
mal studies, particularly of insectivores (see 
Chapter 20). Overall trap success for vertebrates 
was slightly lower in the RNI d'Andringitra (14%, 
902 trap days) compared to Pare National (PN) 
de la Montagne d'Ambre, in extreme northern 
Madagascar, where identical pitfall trapping tech- 
niques produced an overall trap success of 21% 
over the course of 2,244 trap days (Raxworthy & 
Nussbaum, 1994). 

Few new species were being discovered by the 
end of each sampling period (Fig. 17-2). The spe- 
cies lists for the four lowest transects are consid- 
ered to be nearly complete. Because sampling 
time at the 2075 m transect was much shorter, 
some species are likely to have been missed (see 
above). 

Of the 14 species with an elevational range 
that spanned three or more transects, just three 
species {Plethodontohyla inguinalis, Boophis sp. 
4, and Brookesia nasus) have a gap in their dis- 
tribution (i.e., they were not sampled in an in- 
termediate transect). Although elevation gaps in 
distribution suggest that species were missed at 



RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 



161 



Table 17-1. Amphibian and reptile species recorded at RNI d'Andringitra. 



Taxon 



Elevation of transect 



Range (m) 



720 m 810 m 1210 m 1625 m 2075 m 



Min. 



Max. 



Amphibia 

Microhylidae 
Scaphiophryne marmorata 
Anodonthyla boulengeri 
Anodonthyla montana 
Plethodontohyla bipunctata 
Plethodontohyla inguinalis 
Plethodontohyla serratopalpebrosa 
Plethodontohyla alluaudi 
Plethodontohyla notosticta 
Plethodontohyla sp. 1 
Platypelis turberifera 
Platypelis sp. 1 
Platypelis sp. 2 

Mantellidae 
Mantella madagascariensis 
Mantidactylus aerumnalis 
Mantidactylus aglavei 
Mantidactylus argenteus 
Mantidactylus asper 
Mantidactylus bertini 
Mantidactylus betsileanus 
Mantidactylus bicalcaratus 
Mantidactylus biporus 
Mantidactylus eiselti 
Mantidactylus femoralis 
Mantidactylus flavobrunneus 
Mantidactylus grandidieri 
Mantidactylus grandisonae 
Mantidactylus liber 
Mantidactylus lugubris 
Mantidactylus luteus 
Mantidactylus majori 
Mantidactylus opiparis 
Mantidactylus peraccae 
Mantidactylus pulcher 
Mantidactylus redimitis 
Mantidactylus tornieri 
Mantidactylus ulcerosus 
Mantidactylus sp. 1 

Rhacophoridae 
Aglyptodactylus madagascariensis 
Boophis albilabris 
Boophis albipunctatus 
Boophis boehmei 
Boophis goudoti 
Boophis luteus 
Boophis madagascariensis 
Boophis microtympanum 
Boophis miniatus 
Boophis rappoides 
Boophis reticulatus 
Boophis viridis 
Boophis sp. 1 
Boophis sp. 2 
Boophis sp. 3 



800 


820 


1240 


1700 


1700 


2300 


800 


1280 


740 


1280 


1240 


1300 


800 


900 


700 


800 


800 


1240 


650 


860 


1240 


1240 


1240 


1240 


650 


1280 


1850 


1850 


700 


840 


700 


800 


700 


1250 


1450 


1650 


650 


1220 


740 


780 


650 


1400 


800 


1350 


700 


1250 


700 


800 


700 


820 


820 


820 


800 


1150 


650 


1250 


650 


1700 


650 


820 


650 


1400 


1240 


1260 


780 


800 


650 


1700 


720 


720 


1450 


1700 


720 


820 


650 


850 


700 


850 


820 


820 


700 


820 


1450 


1850 


700 


820 


650 


850 


1850 


2300 


800 


1200 


700 


720 


800 


1650 


700 


720 


800 


820 


830 


830 


1240 


1260 



162 



FIELDIANA: ZOOLOGY 



Table 17-1. Continued. 



Taxon 



Elevation of transect 



Range (m) 



720 in 810 in 1210 m 1625 m 2075 m 



Min. 



Max. 



Boophis sp. 4 
Boophis sp. 5 
Boophis sp. 6 

Hyperoliidae 

Heterixalus betsileo 
Ranidae 

Ptychadena mascareniensis 
Totals — Amphibia 
Reptilia 
Gekkonidae 

Lygodactylus montanus 

Phelsuma barbouri 

Phelsuma I. lineata 

Phelsuma q. quadriocellata 

Uroplatus ebenaui 

Chamaeleonidae 
Brookesia n. nasus 
Brookesia superciliaris 
Calumma b. brevicornis 
Calumma b. hilleniusi 
Calumma fallax 
Calumma g. gastrotaenia 
Calumma g. andringitraensis 
Calumma nasuta 
Calumma oshaughnessyi 
Furcifer campani 

Seine idae 
Amphiglossus anosy 
Amphiglossus macrocercus 
Amphiglossus melanopleura 
Amphiglossus melanurus 
Amphiglossus punctatus 
Amphiglossus sp. 1 
Amphiglossus sp. 2 
Mabuya gravenhorsti 

Cordylidae 

Zonosaurus sp. 1 
Colubridae 

Geodipsas heimi 

Geodipsas infralineata 

Geodipsas sp. 1 

Geodipsas sp. 2 

Geodipsas sp. 3 

Liophidium rhodogaster 

Liophidium sp. 1 

Liopholidophis grandidieri 

Liopholidophis stumpffi 

Liopholidophis thieli 

Liopholidophis sp. 1 
Totals— Reptilia 



1240 


1850 


1240 


1240 


1240 


1280 


720 


720 


650 


720 



29 



36 



26 



2000 


2250 


2000 


2300 


720 


750 


720 


1350 


810 


840 


720 


1630 


650 


840 


1200 


1300 


1550 


1850 


1220 


1260 


1100 


1280 


1550 


1680 


720 


830 


1240 


1300 


1850 


2000 


1240 


1300 


720 


900 


720 


800 


720 


780 


800 


1300 


750 


750 


2000 


2100 


650 


800 



650 



* 



800 



1260 


1260 


700 


820 


720 


720 


1240 


1240 


1630 


1630 


750 


750 


730 


850 


1300 


1300 


720 


720 


750 


840 


800 


800 



17 



14 



11 



RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 



163 



'C 



<£ 



Oq^O\^00 VON i~~ 



$ 



o > — r~ r» 



8* 

ON 00 >/-) 

— CM 






OV-M^ O VOm 

a* r»» — ■ -h 



s m o\r~ en o\ r» 

"^ cs cm r~ — — i 



5£ 



^ 



O ■> NO — NO 

<N "^ — (N NO 



O *r> <d 



£ 



O,/,*- no -^ en NO 



v— ' rr* ^ " ^ 
Tt w — CM NO 



o p^ no *-> no en en 

— * — Cl vD 



$ 



O o 

1 1 

is s 

^ >> 

2 CS c/3 

T3 -Q 



>^ h <- 





Z 


E *>£^ 




— 


s- 


> 

i 


ude 
st ty 
sarr 
sam 
1 tra] 


Ed 

u 


5 


•S ^ B a 1 


a 

J3 


u 
1 


-5 o -a 5i Jo 

< e pu j p 


u 


S 





CM CM — 



E E 



2 K - "-a S 3 




OS 



164 



FIELDIANA: ZOOLOGY 



Table 17-3. Species not found during the 1993 survey, but reported as occurring at RNI d'Andringitra. 



Species 



Source 



Amphibia 

Platypelis grandis 
Plethodontohyla tuberata 
Scaphiophryne madagascariensis 
Mantidactylus brevipalmatus 
Mantidactylus cornutus 
Mantidactylus elegans 
Mantidactylus boulengeri 
Mantidactylus decaryi 
Mantidactylus domerguei 
Mantidactylus madecassus 
Mantidactylus pliciferus 
Mantidactylus spiniferus 
Mantidactylus curtus 
Mantidactylus alutus 
Boophis ankaratra 
Boophis majori 
Boophis laurenti 
Boophis untersteini 
Boophis sp. b 
Heterixalus tricolor 
Heterixalus madagascariensis 



Reptilia 



Lygodactylus pictus 
Furicifer lateralis 
Opiums cuvieri 
Opiums cyclurus 
Opiums quadrimaculatus 
Mabuya aureopunctata 
Mabuya boettgeri 
Mabuya elegans 
Mabuya madagascariensis 
Zonosaurus laticaudatus 
Zonosaurus ornatus 
Leioheterodon madagascariensis 
Dromicodryas bernieri 
Mimophis madagascariensis 



Glaw & Vences (1994) 
Nicoll & Langrand (1989) 
Blommers-Schlosser & Blanc 
Glaw & Vences (1994) 
Glaw & Vences (1994) 
Blommers-Schlosser & Blanc 
Blommers-Schlosser & Blanc 
Blommers-Schlosser & Blanc 
Blommers-Schlosser & Blanc 
Glaw & Vences (1994) 
Blommers-Schlosser & Blanc 
Blommers-Schlosser & Blanc 
Blommers-Schlosser & Blanc 
Blommers-Schlosser & Blanc 
Glaw & Vences (1994) 
Glaw & Vences (1994) 
Blommers-Schlosser & Blanc 
Blommers-Schlosser & Blanc 
Glaw & Vences (1994) 
Blommers-Schlosser & Blanc 
Glaw & Vences (1994) 



(1993) 



(1993) 
(1993) 
(1993) 
(1993) 

(1993) 
(1993) 
(1993) 
(1993) 



(1993) 
(1993) 

(1993) 



Glaw & Vences (1994) 

Brygoo (1978) 

Nicoll & Langrand (1989) 

Nicoll & Langrand (1989) 

Blanc (1977), Nicoll & Langrand (1989) 

Nicoll & Langrand (1989) 

Brygoo (1983) 

Brygoo (1983) 

Glaw & Vences (1994) 

Nicoll & Langrand (1989) 

Glaw & Vences (1994) 

Nicoll & Langrand (1989) 

Nicoll & Langrand (1989) 

Nicoll & Langrand (1989), Glaw & Vences (1994) 



some transects, these gaps may be real due to 
other factors, such as the absence of specific mi- 
crohabitats. 



Influence of Human Disturbance 

Although the 720 and 810 m transects were in 
similar types of forest, the forest at the lower tran- 
sect was more disturbed, through clearing and se- 
lective logging, and the density of human trails 
was much higher (see Chapter 4). Two frog spe- 
cies recorded in the 720 m transect, Ptychadena 
mascareniensis (not endemic to Madagascar) and 
Heterixalus betsileo, are characteristic of second- 
ary habitats or forest edge. Both these species 



were found in degraded habitats along the Iantara 
River and were not found within the 810 m tran- 
sect. Thirteen amphibians and three reptiles found 
in primary rain forest in the 810 m transect were 
not observed in the 720 m transect (Table 17-1). 
Other faunal differences between the 720 and 
810 m transects, however, are clearly not due to 
greater habitat disturbances at the lower elevation. 
Two species of frog (Mantidactylus tornieri and 
Boophis viridis) and four species of reptile (Am- 
phiglossus melanopleura, Geodipsas sp. 1, Lio- 
pholidophis stumpffi, and Liopholidophis sp. 1) 
that are restricted to primary rain forest were 
found only in the 720 m transect and were absent 
from the higher elevation transects. Their absence 
in the 810 m transect is not explained by the more 



RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 



165 




Table 17-4. Herpetofaunal coefficients of similarity 
between 100 m elevation bands. 



012345678 
Time (days) 

Fig. 17-2. Accumulation curves of herpetofaunal di- 
versity (all sampling techniques). 



pristine condition of the primary forest in this 
transect. The faunal differences between the two 
transects probably reflect (at least in part) their 
slight difference in elevation. 



Elevational Distribution 

No species was distributed throughout the com- 
plete elevational range sampled within RNI 
d'Andringitra. The greatest amphibian elevational 
range was 650-1700 m (Mantidactylus luteus), 
and for reptiles, 720-1630 m (Brookesia nasus 
nasus). The herpetological communities changed 
sharply between each transect (Table 17-1). 

A comparison was made of the five most inten- 
sively surveyed 100 m elevation bands, centered 
at 700, 800, 1250, 1650, and 1900 m, using the 
following coefficient of similarity, S: 

where C is the number of species in common and 
N 1+2 is the total number of species found for both 
transects combined. 

The coefficients are given in Table 17-4. The 
highest similarity (0.53) was found between the 
700 and 800 m bands. All other coefficients of 
similarity were less than 0.50, demonstrating that 
the majority of species were not widely distrib- 
uted between the elevation bands. No species was 



166 





ation 




Elevation ( 


m) 




Elev 






1200- 


1600- 


1850- 


(m) 


650-750750-850 


1300 


1700 


1950 


650- 


-750 


1.00 


0.53 


0.20 


0.04 


0.00 


750- 


-850 




1.00 


0.26 


0.03 


0.00 


1200- 


-1300 






1.00 


0.09 


0.02 


1600- 


-1700 








1.00 


0.17 


1850- 


-1950 










1.00 



found at both the 1900 m and either the 700 or 
800 m band. 

Elevational limits are typically proposed to help 
classify rain forest types in Madagascar (e.g., 
White, 1983; Jenkins, 1987), with low-altitude 
rain forest occurring below 800 m, montane forest 
at 800-1300 m, and sclerophyllous rain forest at 
1300-2000 m. The 720 and 810 m transects are 
in the transition between low- and mid-altitude 
rain forest, and modest differences between the 
communities of these two transects may reflect, at 
least in part, their slight difference in elevation. 
A sharp transition in the herpetofaunal community 
at 900 m elevation was reported for rain forest at 
PN de la Montagne d'Ambre (Raxworthy & 
Nussbaum, 1994). The 1210m transect is in mon- 
tane rain forest, the 1625 m transect in a mixture 
of montane and sclerophyllous rain forest, and the 
2075 m transect in montane heathland. The low 
similarity coefficients (^ 0.26) between the her- 
petofaunal communities of each habitat suggest 
that many species of amphibians and reptiles are 
specialized to a subset of vegetation types along 
this altitudinal gradient. 

Figure 17-3 shows the change in species diver- 
sity of amphibians and reptiles as a function of 
altitude. Elevation is divided into 100 m bands, 
and species distributions are inferred to be contin- 
uous between the minimum and maximum ele- 
vation recorded at RNI d'Andringitra. Both am- 
phibian and reptile diversity show an obvious 
trend toward decreasing with increasing elevation. 
Above 1 300 m, total herpetological species diver- 
sity does not exceed 1 1 species, representing at 
most 21% of the maximum species diversity re- 
corded between 800 and 900 m. 

Similar results recently were reported for Broo- 
kesia chameleons in northern Madagascar, where 
maximum species diversity occurs at 700 m, and 
above 1300 m just 20% or less of the maximum 
species diversity is represented (Raxworthy & 

FIELDIANA: ZOOLOGY 



40 -i 




amphibians 
reptiles 



LfcjaJjJlll* 



7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 

Altitude (x 100 m) 
Fig. 17-3. Herpetofaunal diversity as a function of altitude. 



Nussbaum, 1995). Brookesia also exhibits a mid- 
altitude bulge in species diversity, between 500 
and 700 m in altitude, with species diversity be- 
low 500 m representing 50% of the maximum 
species diversity at 700 m. On the eastern slopes 
of the RNI d' Andringitra, the increase in amphib- 
ian diversity between 700 and 800 m may repre- 
sent the maximum of a similar mid-altitude bulge. 
Explanations for a decrease in species diversity 
with altitude have not yet been proposed for hab- 
itats in Madagascar. Elevational diversity gradi- 
ents in general are poorly understood, although 
examples are well known for bird communities in 
the tropics (Brown, 1988). Possible factors are 
habitat productivity and stability. Both probably 
decrease at higher elevation due to environmental 
conditions such as cooler temperatures, extreme 
seasonality, and increased vulnerability to cyclone 
and fire damage. 



Endemicity 

The following 1 1 taxa are known only from 
RNI d' Andringitra and are probably endemic to 
the massif: amphibians — Anodonthyla montana, 
Plethodontohyla sp. 1, Platypelis sp. 2, Boophis 
sp. 1, Boophis sp. 3, and Boophis sp. 6; reptiles — 
Lygodactylus montanus, Calumma gastrotaenia 
andringitraensis, Amphiglossus sp. 1 , Zonosaurus 
sp. 1 , and Liopholidophis sp. 1 . Another endemic 



chameleon subspecies, Brookesia nasus pauliani, 
is restricted to the Andringitra Massif rain forest 
between 1620 and 1650 m elevation and repre- 
sented by only two males (Brygoo, 1978). It is 
diagnosed by the lack of paired vertebral tubercles 
and strong heterogeneity of the skin granules. All 
Brookesia nasus caught during this survey had 
characters of B. n. nasus (paired dorsal tubercles 
and moderate heterogeneity of skin granules). 
However, we found no males at high elevation. A 
female caught at 1630 m had characters identical 
to those at lower elevations. To assess the actual 
status of B. n. pauliani, it will be necessary to 
examine the type material. 

The Andringitra Massif is isolated by the Am- 
balavao Pass to the north, and by the Mananara 
Valley (south of Pic dTvohibe) to the south. Both 
areas do not exceed 1000 m elevation. Lower al- 
titude rain forest (below 1000 m) at RNI d' An- 
dringitra, therefore, is not isolated (except by riv- 
ers) from the same forest type to the north and 
south. By contrast, the montane forests are com- 
pletely isolated from other montane areas by cor- 
ridors of lowland forest. Therefore, based on de- 
gree of isolation, endemicity should be more pro- 
nounced for high-altitude specialist species. 

Seven RNI d' Andringitra endemic species (An- 
odonthyla montana, Platypelis sp. 2, Boophis sp. 
3, Boophis sp. 6, Lygodactylus montanus, Cal- 
umma gastrotaenia andringitraensis, and Broo- 
kesia nasus pauliani) are restricted to habitats 



RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 



167 



above 1000 m altitude, with minimum known el- 
evations of between 1240 and 2000 m. By con- 
trast, there are five locally endemic species re- 
stricted below 1000 m elevation: Plethodontohyla 
sp. 1, Boophis sp. 1, Amphiglossus sp. 1, Zono- 
saurus sp. 1, and Liopholidophis sp. 1. Clearly, 
local endemicity is not confined to just montane 
taxa. The presence of low-altitude endemic spe- 
cies in the RNI d'Andringitra may be due to: (1) 
species actually being more widespread than is 
currently known, and therefore not regionally en- 
demic; (2) other barriers (e.g., major rivers) con- 
tributing to isolation; and (3) local adaptation in 
the absence of geographical barriers. 



Comparisons with Other Sites 

Table 17-5 is a list of species that were record- 
ed in this survey and have also been sampled at 
three other rain forest sites: Ankaratra (Central 
High Plateau), PN de Ranomafana (eastern es- 
carpment), and the Anosy Mountains (eastern es- 
carpment). The Anosy Mountains as here defined 
include both the Anosy and Vohimena Mountains, 
and they actually represent six sites between 50 
and 1200 m elevation (Marovony, Ampamakie- 
siny, Manangotry, Marosohy, Nahampoana, and 
Manantantely) surveyed by Raxworthy and Nuss- 
baum between 1989 and 1991. The Anosy Moun- 
tains represent the southern limit of the rain forest 
on the eastern escarpment, with Marovony, the 
closest site to Andringitra, 200 km to the south. 

Pare National de Ranomafana, rain forest 120 
km north of the Andringitra Massif, was surveyed 
by Raxworthy (unpublished) in 1990 between 900 
and 1050 m elevation. Ankaratra Massif, 300 km 
north of RNI d'Andringitra, was surveyed by 
Raxworthy (unpublished data) in 1993 between 
1600 and 2640 m, which included both montane 
forest and heathland. 

Despite the fact that forest above 1200 m ele- 
vation was not surveyed in the Anosy Mountains, 
a large proportion of the Andringitra herpetofauna 
was recorded in these forests. Twenty-eight am- 
phibians (48%) and 19 reptiles (56%) were found 
in the Anosy Mountains, demonstrating that a sig- 
nificant component of the Andringitra fauna has 
a geographical range that extends to the southern 
limit of the eastern rain forest. The following spe- 
cies appear to drop out somewhere between the 
RNI d'Andringitra and the Anosy Mountains: 
Platypelis tuberifera, Heterixalus betsileo, Boo- 
phis albipunctatus, Boophis reticularis, Phelsuma 



quadriocellata quadriocellata, Uroplatus eben- 
aui, and Geodipsas heimi. 

Only a narrow elevation range of rain forest 
was surveyed at the PN de Ranomafana (900- 
1050 m); nevertheless, many of the Andringitra 
species also occur at Ranomafana: 32 amphibians 
(55%) and 16 reptiles (47%). Undoubtedly, other 
Andringitra Massif species would have been re- 
corded if a greater elevational range of forest had 
been surveyed. Most of the species at RNI 
d'Andringitra, with the exception of the local en- 
demic species, are widely distributed along the 
eastern escarpment. 

Ankaratra, although 300 km north of Andrin- 
gitra Massif, has similar montane forests and 
heathlands at the same altitudes. Five amphibians 
(Mantidactylus aerumnalis, Boophis microtym- 
panum, B. goudoti, Boophis sp. 4, and Ptychadena 
mascareniensis) and four reptiles (Phelsuma bar- 
bouri, Calumma brevicornis hilleniusi, Furcifer 
campani, and Amphiglossus sp. 2) occur at both 
sites, representing 80% of the RNI d'Andringitra 
montane herpetofauna that occurs above 1800 m 
elevation. Two other species, Mantidactylus dom- 
erguei and Mabuya boettgeri, recorded at RNI 
d'Andringitra (see above), also occur at Ankara- 
tra. The similarity in montane herpetofauna is 
striking, especially considering the distance and 
current isolation between the two massifs. 

The lowest known elevation for the montane 
species occurring at both the Andringitra and An- 
karatra massifs (excluding Boophis goudoti, Boo- 
phis sp. 4, and Ptychadena mascareniensis) is be- 
tween 1600 and 1850 m elevation. Dispersal be- 
tween the two massifs is currently impossible for 
these eight montane species because of the exten- 
sive areas of relief between 1000 and 1200 m el- 
evation. The major low-elevation barriers for 
montane endemic species are the Fianarantsoa 
plateau (1000-1300 m) and the Ambalavao Pass 
(800-1000 m). This vicariant distribution pattern 
strongly suggests a period in Madagascar's history 
when the montane habitats of Andringitra and An- 
karatra formed a continuous belt between the 
massifs, most likely during glacial periods of the 
Pleistocene when cooler climatic conditions al- 
lowed montane habitats to expand into lower el- 
evations. 

The two high-altitude RNI d'Andringitra en- 
demics, Lygodactylus montanus and Anodonthyla 
montana, may have been more widespread during 
glacial periods. But, because no other populations 
are known, it appears that either they were not 



168 



FIELDIANA: ZOOLOGY 



Table 17-5. The distribution of Andringitra species 
(1993 survey) at other sites (see text). 



Site 



Table 17-5. Continued. 



Species 



Anka- Rano- Anosy 
ratra mafana Moun- 

1600- 900- tains 
2640 1050 50- 
m m 1200 m 



Amphibia 

Microhylidae 
Scaphiophryne marmorata 
Anodonthyla boulengeri 
Plethodontohyla bipunctata 
Plethodontohyla inguinalis 
Plethodontohyla alluaudi 
Plethodontohyla notosticta 
Platypelis turberifera 
Platypelis sp. 1 

Mantellidae 
Mantella madagascariensis 
Mantidactylus aerumnalis 
Mantidactylus aglavei 
Mantidactylus asper 
Mantidactylus bicalcaratus 
Mantidactylus bertini 
Mantidactylus betsileanus 
Mantidactylus biporus 
Mantidactylus eiselti 
Mantidactylus femoralis 
Mantidactylus grandidieri 
Mantidactylus grandisonae 
Mantidactylus liber 
Mantidactylus lugubris 
Mantidactylus luteus 
Mantidactylus majori 
Mantidactylus opiparis 
Mantidactylus peraccae 
Mantidactylus pulcher 
Mantidactylus redimitis 
Mantidactylus ulcerosus 
Mantidactylus sp. 1 

Rhacophoridae 
Aglyptodactylus 

madagascariensis 
Boophis albilabris 
Boophis albipunctatus 
Boophis boehmei 
Boophis goudoti 
Boophis luteus 
Boophis madagascariensis 
Boophis microtympanum 
Boophis miniatus 
Boophis reticulatus 
Boophis sp. 2 
Boophis sp. 4 

Hyperoliidae 

Heterixalus betsileo 
Ranidae 

Pytchadena mascareniensis 
Totals — Amphibia 



Site 



Species 



Anka- Rano- Anosy 
ratra mafana Moun- 
1600- 900- tains 
2640 1050 50- 
m m 1200 m 



Reptilia 






Gekkonidae 






Phelsuma barbouri * 






Phelsuma I. lineata 




* 


Phelsuma q. quadriocellata 


* 




Uroplatus ebenaui 


* 




Chamaeleonidae 






Brookesia nasus 


* 


* 


Brookesia superciliaris 


* 


* 


Calumma b. brevicornis 


* 


* 


Calumma b. hilleniusi * 






Calumma fallax 


* 




Calumma g. gastrotaenia 


* 


* 


Calumma nasuta 


* 


* 


Calumma oshaughnessyi 


* 


* 


Furcifer campani * 






Scincidae 






Amphiglossus anosy 




* 


Amphiglossus macrocercus 




* 


Amphiglossus melanopleura 


* 


* 


Amphiglossus melanurus 


* 


* 


Amphiglossus punctatus 


* 


* 


Amphiglossus sp. 2 * 






Mabuya gravenhorsti 


* 


* 


Colubridae 






Geodipsas heimi 


* 




Geodipsas infralineata 




* 


Geodipsas sp. 1 




* 


Geodipsas sp. 2 




* 


Liophidium rhodogaster 


* 


* 


Liopholidophis stumpjfi 




* 


Liopholidophis thieli 


* 


* 


Totals— Reptilia 4 


16 


19 



33 



28 



successful at colonizing other massifs, or that oth- 
er populations have gone extinct or unrecorded. 



Conservation 

The most vulnerable elements of the RNI 
d' Andringitra herpetofauna are the six amphibians 
and five reptile species, and the two reptile sub- 
species considered to be endemic to the massif. 
The distribution of the endemics covers virtually 
the full elevational range of the reserve from 650 
to 2350 m and all vegetation types. Conservation 



RAXWORTHY & NUSSBAUM: AMPHIBIANS AND REPTILES 



169 



of these species, therefore, makes it necessary to 
protect all habitats within the reserve. 

The most vulnerable habitat appears to be the 
low-elevation rain forest at the eastern limits of 
the reserve, which is threatened by clearing for 
tavy. By contrast, the high-elevation sclerophyl- 
lous forest (above 2000 m), now confined to small 
isolated patches, is vulnerable to damage from fire 
and possibly grazing from cattle. 

The high herpetofaunal diversity of RNI 
d'Andringitra (at least 95 species) and the degree 
of local endemicity (13% of the herpetofauna sur- 
veyed) demonstrate the importance of the reserve 
within the network of protected areas in Mada- 
gascar. The conservation of the natural habitats 
found within the Andringitra Massif will be crit- 
ical to maintaining Madagascar's existing biodi- 
versity. 



Acknowledgments 

We thank Angelin Razafimanantsoa and Nirhy 
Rabibisoa for their substantial help in the field, 
Steven Goodman for efficiently organizing the 
complicated logistics of the survey, and Olivier 
Langrand and Sheila O'Connor (World Wide 
Fund for Nature) for their strong support for this 
project. We are grateful to the officials of Direc- 
tion des Eaux et Forets, Association Nationale 
pour la Gestion des Aires Protegees, and the 
Universite d' Antananarivo for assistance during 
this survey. This research was supported in part 
by a grant from the National Science Foundation 
(DEB-90-24505). 



Literature Cited 

Blanc, C. P. 1977. Reptiles Sauriens Iguanidae. Faune 
de Madagascar, 45: 1-400. 

Blommers-Schlosser, R. M. A., and C. P. Blanc. 
1991. Amphibiens (premiere partie). Faune de Mad- 
agascar, 75(1): 1-385. 

. 1993. Amphibiens (deuxieme partie). Faune de 

Madagascar, 75(2): 385-530. 



Brown, J. H. 1988. Species diversity, pp. 57-89. In 
Myers, A. A., and P. S. Giller, eds., Analytical bio- 
geography: An integrated approach to the study of an- 
imal and plant distributions. Chapman and Hall, Lon- 
don, 578 pp. 

Brygoo, E. R. 1978. Reptiles Sauriens Chamaeleonti- 
dae. Genre Brookesia et complement pour le genre 
Chamaeleo. Faune de Madagascar, 47: 1-173. 

. 1983. Systematique de lezards scincides de la 

region malgache. XI. Les Mabuya de Madagascar. 
Bulletin du Museum National d'Histoire Naturelle, 
Paris, 4th ser., sect. A., 5(4): 1079-1108. 

. 1985. Les Gerrhosaurinae de Madagascar Sau- 



ria (Cordylidae). M6moires du Museum National 
d'Histoire Naturelle. Ser. A, Zoology, 134: 1-65. 

Glaw, F, and M. Vences. 1994. A fieldguide to the 
amphibians and reptiles of Madagascar. Second edi- 
tion. Moss Druck, Leverkusen, Germany, 480 pp. 

Guibe, J. 1978. Les Batraciens de Madagascar. Bonner 
Zoologische Monographien, 11: 1-140. 

Jenkins, M. 1987. Madagascar, an environmental pro- 
file. International Union for the Conservation of Na- 
ture and Natural Resources, Cambridge, 374 pp. 

Millot, J., and J. Guibe. 1950. Les batraciens du nord 
de l'Andringitra (Madagascar). Memoires de l'lnsti- 
tute Scientifique de Madagascar. Ser. A, Biologie An- 
imate, 4(1): 197-206. 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: 
Revue de la Conservation et des Aires Protegees. 
World Wide Fund for Nature, Gland, 374 pp. 

Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. Etude des ecosystemes mon- 
tagnards dans la region malgache. I. Le Massif de 
l'Andringitra. 1970-1971. Geomorphologie, climato- 
logie et groupement vegetaux. RCP 225. Bulletin de 
la Societe d'Ecologie, 2(2-3): 198-226. 

Pasteur, G. 1964. Notes preliminaires sur les Lygo- 
dactylus (Gekkonides). IV. Diagnoses de quelques for- 
mes africanes et malgaches. Bulletin du Museum Na- 
tional d'Histoire Naturelle, Paris, 2nd sen, 36(3): 311- 
314. 

Raxworthy, C. J., and R. A. Nussbaum. 1994. A rain 
forest survey of amphibians, reptiles and small mam- 
mals at Montagne d'Ambre, Madagascar. Biological 
Conservation, 69: 65-73. 

1995. Systematics, speciation, and biogeogra- 



phy of the dwarf chameleons (Brookesia Gray; Rep- 
tilia; Sauria; Chamaeleontidae) of northern Madagas- 
car. Journal of Zoology, London, 235: 525-558. 
White, F 1983. The vegetation of Africa. A descrip- 
tive memoire to accompany the UNESCO/AETFAT/ 
UNSO vegetation map of Africa. Natural Resources 
Research 20. UNESCO, Paris, 356 pp. 



170 



FIELDIANA: ZOOLOGY 



Chapter 18 

The Birds of the Eastern Slopes of the Reserve 
Naturelle Integrate d'Andringitra, Madagascar 

Steven M. Goodman and Michael S. Putnam 



Abstract 

An elevational transect of the birds occurring along the eastern slopes of the Reserve Na- 
turelle Integrate d'Andringitra was conducted using a variety of techniques: point counts, gen- 
eral observations, and mist-netting. A total of 99 species were recorded; three of these were 
boreal winter migrants to Madagascar. The resident birds included 82 species at 720 m, 66 at 
810 m, 56 at 1210 m, and 46 at 1625 m. Although the ericoid and heathland zone from 1800 
to 2200 m was not systematically surveyed, 20 species of birds were found there. 

Species accumulation curves showed that the majority of species (85%) recorded in each 
elevational zone were identified after 4 days of survey work. After 6 days at each elevation at 
least 93% of species were found. In all elevational zones additional species were added for up 
to 8 days after the start of each elevational transect. Information is also presented on a com- 
parison of mist-netting results between the elevational zones, composition of mixed-species 
foraging flocks, the effects of forest disturbance on species richness, human hunting pressure, 
and general natural history for selected species. 



Resume 

Un transect altitudinal destin6 a r£aliser l'inventaire des oiseaux a €\€ mis en place le long 
du versant est de la R6serve Naturelle Integrale d'Andringitra en utilisant di verses techniques: 
points de comptage, observations aldatoires et capture au moyen de filets. 

Un total de 99 especes a 6t6 recense" dont trois sont des migrateurs boreaux hivernant a 
Madagascar. Les oiseaux residents comptent 82 especes a une altitude de 720 m, 66 a 810 m, 
56 a 1210 m et 46 a 1625 m. Meme si la zone de fourr6 encoi'de trouvee de 1800 m a 2200 
m n'a pas €\€ syst6matiquement inventorize, 20 especes d'oiseux y ont 6t6 trouvees. 

Les courbes d'accumulation des especes montrent que la majority (85%) des especes r^per- 
toriees dans chaque zone altitudinale a pu etre identified apres 4 jours d'inventaire. Apres 6 
jours, a chaque altitude, un minimum de 93% des especes a 6l6 inventoried Des especes sup- 
ptementaires ont 6t6 recensees dans toutes les zones d'altitude jusqu'a 8 jours apres le d£but 
de chaque transect altitudinal. Les informations relatives a la comparaison des r£sultats acquis 
au moyen de filets au sein des diffeientes zones d'altitude, a la composition des groupes 
plurispecifiques, aux effets de la perturbation de la forSt sur la richesse specifique, a la pression 
de chasse exercee par l'homme et a l'histoire naturelle de quelques especes selectionnees sont 
egalement presentees. 

GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 171 



Introduction 

Madagascar is well known for its remarkable 
avifauna; over 50% of the 201 breeding birds are 
endemic (Langrand, 1990). In the past decade the 
amount of ornithological work on Madagascar has 
increased considerably, and much new informa- 
tion is available about the island's birds. Examples 
of new directions in Malagasy ornithology include 
systematic and phylogenetic studies (Prum, 1993; 
Schulenberg, 1993; Schulenberg et al., 1993), 
ecological work on species (Hawkins, 1993; Wat- 
son et al., 1993), and studies of community struc- 
ture (Andrianarimisa, 1993; Langrand & Wilme, 
1993; Razafimahaimodison & Andrianantenaina, 
1993). Furthermore, numerous inventories of pre- 
viously unknown or poorly known forest sites 
have been conducted (e.g., Thompson, 1987; Saf- 
ford & Duckworth, 1990; Thompson & Evans, 
1991). However, little information is currently 
available on the ecological zonation of birds along 
an elevational transect. 

The intent of the present study was to examine 
the pattern of bird species richness along an ele- 
vational gradient on the eastern slopes of the Re- 
serve Naturelle Integrate (RNI) d'Andringitra, as 
well as to gather general information on the nat- 
ural history of the local avifauna. Both authors 
made general observations on the birds of the 
area. S.M.G. was responsible for bird netting (pre- 
sented in this chapter), and M.S.P was responsible 
for point count surveys (M. S. Putnam, unpub- 
lished data). 



Previous Work 

Little has been reported on the birds within the 
RNI d'Andringitra. The only known detailed ob- 
servations were made by Langrand, mostly on the 
west side of the reserve and in the high mountain 
area (Nicoll & Langrand, 1989). No ornithologist 
accompanied the La Recherche Cooperative sur 
Programme (no. 225) expedition to the area (Pau- 
lian et al., 1971). A Cambridge University student 
expedition to the area in 1993 compiled some un- 
published bird records (O'Keeffe & Ashmore, not 
dated). 

The majority of our work in the RNI d'Andrin- 
gitra was on the eastern side of the massif in hu- 
mid and montane forest. The humid forest site 
closest to the reserve that is ornithologically rea- 
sonably well known is the Pare National (PN) de 



Ranomafana, about 120 km to the northeast. In 
between these two sites is the forest of Ikongo 
(Fort Carnot), an important late 19th- and early 
20th-century collecting locality, but for which lit- 
tle recent information is available. 



Methods 

Observations 

The main ornithological studies were conduct- 
ed by both authors during the first field trip to the 
reserve (23 September-1 November 1993), al- 
though general observations were made by 
S.M.G. during the second trip (14 November-18 
December 1993). Daily field notes were kept on 
the species observed or heard in each transect 
zone, elevation range, and surrounding habitat(s). 
Altimeters and known positions on maps were 
used to calculate elevation. Each study zone in- 
cluded an elevational band of ±75 m, centered on 
a transect camp (720, 810, 1210, and 1625 m). 
The only exception was at our first camp, which 
was at 720 m, and which formed the lower ele- 
vational limit of the transect. Thus, there was 
broad elevational overlap between the zones re- 
ported herein as 720 and 810 m. See Chapter 1 
for details concerning the itinerary of the expe- 
dition and Chapter 2 for descriptions of vegeta- 
tional zones. 

Netting 

Mist nets were operated in each transect zone, 
in a variety of microhabitats (ridge, slope, and 
valley bottoms), and within 1 km ground distance 
of camp. All nets used were 2.6 m high, 12 m 
long, and with 36-mm mesh size. A 24-hour pe- 
riod during which a net was left up continuously 
was termed a "net-day." Nets were regularly 
checked during daylight hours, before dawn, and 
after dusk. In each zone, nine nets were run for 5 
days, making a total of 45 net-days per transect, 
and they were left up throughout the night to cap- 
ture bats and nocturnal birds. The bottom panel 
of each net was generally within 10 cm of or 
touching the ground. 

Net Captures and Specimens 

Most netted birds were marked for individual 
recognition and released. All birds caught on the 



172 



FIELDIANA: ZOOLOGY 



same day had the same primary feather marked 
with indelible ink; for example, birds caught on 
day 1 had the outermost primary feather on the 
left wing dyed. All birds were measured and 
weighed. Birds were released in the immediate vi- 
cinity of the net in which they were captured. 

Some birds were collected as voucher speci- 
mens or for systematic studies. These were pre- 
pared as either standard museum skins, fluid-pre- 
served specimens, or full skeletons. This material 
is housed in the Field Museum of Natural History 
(FMNH), and a representative portion of the col- 
lection will be returned to the Departement de 
Biologie Animale, University d' Antananarivo, 
Madagascar. Tissue samples of collected speci- 
mens were saved in liquid nitrogen or a buffered 
solution. Whole carcasses preserved in formalin 
were wrapped in fine cheesecloth before immer- 
sion to prevent mixing of ectoparasites (see Chap- 
ter 12). Blood smears were made from several 
released and collected birds for a study of blood 
parasites (see Chapter 13). 



Systematic Order, Nomenclature, 
and Common Names 

Except as noted, we follow the systematic order 
and English common and scientific names used 
by Langrand (1990). 



Results and Discussion 

General 

A total of 99 species were recorded on the east- 
ern slopes of the RNI d' Andringitra; three of these 
were boreal winter migrants to Madagascar (Table 
18-1). The resident birds included 82 species at 
720 m, 66 at 810 m, 56 at 1210 m, and 46 at 1625 
m. Although the ericoid and heathland zone from 
1800 to 2200 m was not systematically surveyed, 
20 species of birds were found there. Further- 
more, Egretta alba, Egretta dimorpha, and Den- 
drocygna viduata were observed just outside of 
the reserve in the area between Col de Vohipia 
and Ambalamamenjana; these three species have 
not been included within the list. Combining the 
list presented by Nicoll and Langrand (1989) with 
our work, 1 14 species are known to occur in the 
RNI d' Andringitra. This number does not include 
Pseudocossyphus bensoni, reported from the re- 



serve by O'Keeffe & Ashmore (not dated), but 
for which insufficient details of the record are 
available. 



Species Addition Curves 

In advance of any detailed analysis of the or- 
nithological results of the 1993 mission, it was 
critical to determine whether our survey was suf- 
ficient to reflect the species richness for each ele- 
vational zone. By plotting the number of new spe- 
cies of birds added to each elevational zone per 
day of fieldwork, species addition curves could be 
calculated. Here we combined information from 
the point counts, netting, and general observa- 
tions. Each time we shifted camps, the first day 
in each zone was only partially devoted to bird 
surveys. Thus, we combined data for the first par- 
tial and the first complete day in each elevational 
zone. 

An examination of these curves (Fig. 18-1) 
shows that the majority of species (85%) recorded 
in each elevational zone were identified after 4 
days of survey work. After 6 days at each ele- 
vation, at least 93% of the species were found 
(Table 18-2). However, in all cases additional spe- 
cies were added for up to 8 days after the start of 
each elevational transect. Two species were re- 
corded on the second field trip to the same portion 
of the reserve, when no detailed ornithological 
studies were carried out. These species were Fal- 
co zoniventris, in the 810 m elevation zone, and 
Dromaeocercus brunneus, in the 1625 m zone. 
Two points are clear: (1) there was a rapid de- 
crease in the addition of unrecorded species 
through time within each elevational zone; and (2) 
after 6 days of fieldwork by two ornithologists 
experienced with Malagasy birds, over 93% of the 
birds recorded in each zone had already been de- 
tected. On the basis of species addition curves for 
each elevational zone, we conclude that the local 
avifaunas were well documented. 



Species Richness 

There was marked elevational variation in bird 
species richness on the eastern slopes of the RNI 
d' Andringitra. A total of 99 species were recorded 
along an altitudinal gradient from 720 to 2200 m. 
No introduced species was recorded along the 
transect, although one, Acridotheres tristis, is 
known from other parts of the reserve and sur- 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



173 



Table 18-1. Elevational distribution of birds on the eastern slopes of RNI d'Andringitra. 













[1800- 


Species 


720 m 


810 m 


1210 m 


1625 m 


2200 m]* 


Tachybaptus pelzelnii 


+ 










Phalacrocorax africanus'f 


+ 










Bubulcus ibis 


+ 










Ardea purpurea 


+ 










Lophotibis cristatcr\ 


+ 


+ 








Aviceda madagascariensis 






+ 






Milvus migrans 


+ 










Polyboroides radiatus 


+ 


+ 


+ 






Accipiter francesii'f 


+ 


+ 








Accipiter madagascariensis 




+ 








Accipiter henstii 


+ 


+ 


+ 






Buteo brachypterus 


+ 


+ 


+ 


+ 


A 


Falco newtoni 


+ 










Falco zoniventris^ 




+ 








Falco eleonoraeX 


+ 










Falco concolor\% 


+ 










Falco peregrinus 






+ 






Turnix nigricollis 


+ 










Mesitornis unicolorf 


+ 


+ 


+ 


+ 




Sarothrura insularis 


+ 


+ 


+ 


+ 


H 


Canirallus kioloidest 


+ 


+ 








Actitis hypoleucosX 


+ 










Streptopelia picturata 


+ 


+ 


+ 


+ 




Treron australis^ 


+ 










Alectroenas madagascariensis 


+ 


+ 








Coracopsis nigra 


+ 


+ 


+ 


+ 




Coracopsis vasa 


+ 


+ 


+ 






Agapornis cana\ 


+ 










Cuculus audebertif 




+ 








Cuculus rochii 


+ 


+ 


+ 


+ 


F, H 


Coua reynauditf 


+ 


+ 


+ 


+ 




Coua caerulea 


+ 


+ 


+ 


+ 




Centropus toulou 


+ 






+ 




Tyto alba 


+ 










Otus rutilus 


+ 


+ 


+ 


+ 




Asio madagascariensis 




+ 




+ 




Caprimulgus madagascariensis^ 


+ 










Zoonavena grandidiertf 


+ 


+ 


+ 


+ 


A 


Cypsiurus parvus^ 


+ 










Apus barbatus 


+ 






+ 


A 


Apus melba 


+ 


+ 


+ 


+ 


A 


Alcedo vintsioides 


+ 


+ 








Ispidina madagascariensis^ 


+ 


+ 


+ 






Merops superciliosus 


+ 


+ 




+ 




Eurystomus glaucurus 


+ 


+ 








Brachypteracias leptosomusf 


+ 




+ 






Brachypteracias squamigerf 


+ 










Atelornis pittoides 


+ 


+ 


+ 






Atelornis crossleyif 






+ 


+ 




Leptosomus discolor 


+ 


+ 


+ 


+ 


A 


Philepitta castanea~\ 


+ 


+ 


+ 


+ 




Neodrepanis coruscans^ 


+ 


+ 








Neodrepanis hypoxantha 






+ 


+ 


F, H 


Phedina borbonica 


+ 








A 


Motacilla flaviventris 


+ 


+ 








Coracina cinerea 


+ 


+ 


+ 


+ 




Phyllastrephus madagascariensis 


+ 


+ 


+ 






Phyllastrephus zosterops 


+ 


+ 


+ 






Phyllastrephus cinereiceps^ 




+ 


+ 


+ 




Hypsipetes madagascariensis 


+ 


+ 


+ 


+ 


F,H 


Copsychus albospecularis 


+ 


+ 


+ 






Saxicola torquata 


+ 








H 



174 



FIELDIANA: ZOOLOGY 



Table 18-1. Continued. 













[1800-2200 


Species 


720 m 


810 m 


1210 m 


1625 m 


m]* 


Pseudocossyphus sharpei 


+ 


+ 


+ 


+ 




Nesillas typica 


+ 


+ 


+ 


+ 


F, H 


Cisticola cherina 


+ 










Cryptosylvicola randrianasoloif 




+§ 


+ 


+ 


F 


Dromaeocercus brunneus 








+ 




Dromaeocercus seebohmif 










H 


Randia pseudozosteropsi 


+ 


+ 


+ 






Newtonia amphichroai 


+ 


+ 


+ 


+ 


F, H 


Newtonia brunneicauda 


+ 


+ 


+ 


+ 




Neomixis tenella 


+ 


+ 


+ 






Neomixis viridisf 


+ 


+ 


+ 


+ 




Neomixis striatigulai 


+ 


+ 


+ 


+ 




Hartertula flavoviridisX 


+ 


+ 


+ 


+ 




Terpsiphone mutata 


+ 


+ 


+ 


+ 




Pseudobias wardi 


+ 


+ 


+ 






Oxyiabes madagascariensisX 


+ 


+ 


+ 


+ 




Crossleyia xanthophrysX 


+ 


+ 


+ 


+ 




Mystacornis crossleyif 


+ 


+ 




+ 




Nectarinia notatai 


+ 


+ 


+ 


+ 


F, H 


Nectarinia souimanga 


+ 


+ 


+ 


+ 


F, H 


Zosterops maderaspatana 


+ 


+ 


+ 


+ 


F H 


Calicalicus madagascariensis 


+ 


+ 


+ 


+ 




Schetba rufaX 


+ 


+ 


+ 






Vanga curvirostrisX 


+ 


+ 


+ 


+ 




Xenopirostris pollenif 




+ 








Leptopterus viridis 


+ 


+ 


+ 


+ 




Leptopterus chabert 


+ 


+ 


+ 






Cyanolanius madagascarinus 


+ 




+ 






Hypositta corallirostrisX 


+ 


+ 








Tylas eduardi 


+ 


+ 


+ 


+ 




Dicrurus forficatus 


+ 


+ 


+ 


+ 




Corvus albus 


+ 






+ 


H 


Hartlaubius auratus 


+ 










Ploceus nelicourvi 


+ 


+ 


+ 


+ 




Foudia madagascariensis 


+ 








H 


Foudia omissai 




+ 


+ 


+ 




Lonchura nana 


+ 










Total number of species 


82 


66 


56 


46 


20 



* The elevational zone in brackets was not part of the standard transect, and records from these zones were not 
obtained in a systematic manner. The habitats include the following: F = upper montane forest; H = heath; A = 
aerial. 

t Species whose names are followed by a dagger (t) were not listed for the reserve by Nicoll and Langrand (1989). 

X Species whose names are followed by a double dagger (X) are boreal winter migrants to Madagascar. 

§ Recorded at the upper limit of the transect zone. 



rounding areas. The numbers of species recorded 
in each elevational zone (Table 18-1) were as fol- 
lows: 82 in the 720 m zone, 66 in the 810 m zone, 
56 in the 1210 m zone, 46 in the 1625 m zone, 
and 20 in the zone between 1800 and 2200 m 
(Fig. 1 8-2). Thus, species richness decreased as a 
function of altitude (linear regression, P = 0.007, 
r 2 = 0.94, N = 5; or, excluding the 720 m datum 
point, the zone of disturbed forest, P = 0.038, r 2 
= 0.93, N = 4). This is the same basic pattern 
found throughout the tropics, whether on conti- 



nental land masses or islands (e.g., Chapman, 
1917; Terborgh, 1977; Diamond & LeCroy, 1979; 
Goodman & Gonzales, 1990), and in North Amer- 
ica (Able & Noon, 1976; Finch, 1989). 



Segregation of Congeners Along an 
Elevational Gradient 

Of the 76 genera of resident birds known from 
the eastern slopes of RNI d'Andringitra, 16 in- 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



175 



90- 



80- 



CO 

(!) 


70 


*w* 




o 




4> 




Cu 




00 


60 


<*-H 




o 




1- 

1 


50 


a 




3 


40 


Z 





30- 



20' 




10 



11 



12 



Working Days 



Fig. 18-1. Species addition curves for four elevational zones. Data from transect counts, general observations, 
and mist-netting are combined in this analysis. 



eluded more than one species. Very few of these 
genera contain species with non-overlapping ele- 
vational ranges, but in several cases these disjunct 
distributions may be an artifact of the species be- 
ing rare and only recorded on a few occasions. 
However, there are a few examples of apparent 
altitudinal replacement of relatively common spe- 
cies. Neodrepanis coruscans was recorded in the 
720 and 810 m zones and not above 900 m ele- 
vation, whereas N. hypoxantha was found in the 
1210 and 1625 m elevational zones. Thus, the al- 
titudinal ranges of these two species almost abut, 
and these two species appear to replace one an- 
other. 

Two species in each of three different genera 
were found across the complete elevational gra- 
dient of forested habitat from 720 to 1625 m el- 
evation (Coua, Newtonia, and Nectarinia). Within 
these three genera there are behavioral or mor- 
phological features that separate congeners: Coua 



Table 18-2. Percentage of new species added to 
each elevational zone after 4 and 6 days' work in a zone. 

Percentage of new species added 
Elevation (m) After 4 days After 6 days 



720 

810 

1210 

1625 



14 

15 

5 

4 



caerulea is arboreal, whereas C. reynaudii is ter- 
restrial; Newtonia brunneicauda is a mid- to up- 
per-canopy species, whereas N. amphichroa in- 
habits the understory; and Nectarinia notata has 
a distinctly longer bill and is more massive than 
N. souimanga. Other congeneric species pairs 
have complete or nearly complete overlapping 



100 




2500 



Elevation (m) 






Fig. 18-2. Plot of total number of species per ele- 
vational zone. The straight line shows the relationship 
y = -0.04 Ix + 106.7 (r 2 adjusted = 0.94, P = 0.007), 
where y = number of species and x = elevation in 
meters. 



176 



FIELDIANA: ZOOLOGY 



Table 18-3. 


Netting results presented by elevational zone. 








720 m 810 in 1210 m 


1625 m 


Total 


Species 


(45)* (45)* (45)* 


(45)* 


(180)* 


Otus rutilus 


1 1 




2 


Alcedo vintsioides 


4 




4 


Philepitta castanea 


6 7 11 


9 


33 


Neodrepanis coruscans 


3 




3 


N. hypoxantha 


2 


3 


5 


Phedina borbonica 


7 




7 


Motacilla flaviventris 


1 




1 


Coracina cinerea 


1 




1 


Phyllastrephus madagascariensis 


2 8 




10 


P. zosterops 


12 17 6 




35 


P. cinereiceps 


3 


3 


6 


Hypsipetes madagascariensis 


4 5 




9 


Copsychus albospecularis 


13 10 3 




26 


Pseudocossyphus sharpei 




2 


2 


Nesillas typica 


1 1 


6 


8 


Cryptosylvicola randrianasoloi 




1 


1 


Newtonia amphichroa 


1 2 


2 


5 


N. brunneicauda 


1 




1 


Terpsiphone mutant 


12 3 1 




16 


Oxylabes madagascariensis 


5 1 


3 


9 


Crossleyia xanthophrys 


1 1 




2 


Nectarinia souimanga 


2 1 4 


3 


10 


Zosterops maderaspatana 


5 3 


2 


10 


Vanga curvirostris 


2 1 


1 


4 


Cyanolanius madagascarinus 


2 




2 


Hypositta corallirostris 


1 




1 


Dicrurus forficatus 


3 3 




6 


Ploceus nelicourvi 


5 3 


2 


10 


Foudia omissa 


2 5 


3 


10 


Total individuals 


94 62 42 


42 


240 


Total species 


23 13 13 


15 


30 


Average birds/net-day 


2.1 1.4 0.9 


0.9 


1.3 



* The number in parentheses is the total number of net-days accrued for that zone. 



elevational ranges (Coracopsis, Neomixis, and 
Leptopterus). 

In general, there were few records of ground- 
rollers, so it is difficult to compare their eleva- 
tional ranges. Both species of Brachypteracias 
and Atelornis pittoides tend to be lowland or mid- 
montane in distribution, whereas A. crossleyi is 
more common at higher elevations. The distribu- 
tion of Foudia seems to be related to forest cover. 
Foudia madagascariensis was found only in the 
disturbed areas at 720 m and in the open heath- 
land above 1950 m, whereas F. omissa occurred 
in forested areas between 810 and 1625 m ele- 
vation. 

One of the more interesting cases of differences 
in the altitudinal ranges of congeneric species is 
with Phyllastrephus. Phyllastrephus madagascar- 
iensis and P. zosterops were found at the 720, 
810, and 1210 m elevational zones. On the basis 
of netting results, both of these species were more 



common at 810 m than 720 m, with P. zosterops 
being the more common of the two (Table 18-3). 
At 1210 m P. madagascariensis was rare and ob- 
served only a few times; P. zosterops was also 
decreased in relative abundance. The third spe- 
cies, P. cinereiceps, was rare at 810 m, where it 
was only observed, and three individuals were 
netted in both the 1210 m and 1625 m zones. 
Thus, in sympatry there are some differences be- 
tween Phyllastrephus species in their relative 
abundances. In other zones there appears to be 
species replacement. 



Elevational Differences in Species 
Richness Based on Netting 

Mist nets were used to assess species richness 
of lower understory birds within four transect 
zones (Karr, 1981). In total, 180 net-days were 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



177 



Table 18-4. Mass measurements (g) of birds netted during the Andringitra survey. 







Minimum- 




Standard 


Species 


N 


maximum* 


Mean 


deviation 


Otus rutilus 


2 


85.5, 110 






Alcedo vintsioides 


4 


17.5-19.5 


18.38 


0.85 


Atelornis crossleyi 


1 


77.5 






Philepitta castanea 


33 


27.0-42.5 


35.02 


3.64 


Neodrepanis coruscans 


4 


6.2-6.6 


6.43 


0.17 


N. hypoxantha 


5 


6.4-8.1 


7.18 


0.67 


Phedina borbonica 


7 


20.5-25.0 


22.86 


1.60 


Motacilla flaviventris 


1 


24.5 






Coracina cinerea 


1 


46.5 






Phyllastrephus madagascariensis 


11 


20.5-37.0 


30.50 


5.22 


P. zosterops 


29 


14.0-20.0 


17.20 


1.60 


P. cinereiceps 


6 


16.5-24.5 


19.33 


2.77 


Hypsipetes madagascariensis 


8 


41.5-50.0 


45.81 


2.74 


Copsychus albospecularis 


23 


18.5-27.5 


23.48 


2.26 


Pseudocossyphus sharpei 


2 


26.5, 25.0 






Nesillas typica 


8 


16.5-21.5 


18.38 


1.53 


Cryptosylvicola randrianasoloi 


2 


8.2, 8.3 






Newtonia amphichroa 


5 


11.5-14.5 


12.80 


1.20 


N. brunneicauda 


1 


12.5 






Terpsiphone mutata 


17 


12.5-15.0 


13.65 


0.72 


Oxylabes madagascariensis 


8 


19.0-26.0 


22.75 


2.56 


Crossleyia xanthophrys 


2 


21.5, 21.5 






Nectarinia souimanga 


7 


5.5-9.5 


7.47 


1.19 


Zosterops maderaspatana 


11 


9.5-14.5 


11.55 


1.37 


Vanga curvirostris 


1 


67 






Cyanolanius madagascarinus 


2 


22.0, 24.5 






Hypositta corallirostris 


1 


15.5 






Dicrurus forficatus 


7 


41.5-50.5 


47.43 


3.17 


Ploceus nelicourvi 


7 


21.0-25.5 


23.57 


1.69 


Foudia omissa 


10 


18.0-21.4 


19.45 


1.48 



* In cases where three or fewer measurements are available, only the masses themselves are given. 



accrued. The numbers of birds captured and the 
capture rate per transect zone (total captured/total 
number of net-days) were 94 and 2.1 at 720 m, 
62 and 1.4 at 810 m, 42 and 0.9 at 1210 m, and 
42 and 0.9 at 1625 m. The masses of captured 
birds are presented in Table 18-4. 

Within the first three transect zones there was 
a pattern of decreasing bird capture rate as a func- 
tion of increasing elevation. However, the results 
from the 1625 m transect zone were identical to 
those from 1210 m, and on the basis of the current 
data, understory bird density, if indeed it could be 
measured by capture rates, was similar in these 
two zones. The number of species netted within 
each zone varied, from 23 at 720 m, to 13 at 810 
m, 13 at 1210 m, and 15 at 1625 m. On mountains 
in the tropics, there is generally a strong negative 
relationship between bird species richness and el- 
evation (see p. 175), and the patterns emerging 
from mist-netting on the Andringitra Massif do 
not follow this pattern. However, this technique 
introduces several types of biases into interpreta- 



tions of avifaunal patterns (Karr, 1981; Graves et 
al., 1983; Remsen & Parker, 1983), particularly 
by sampling only a small proportion of the species 
at a given elevation and specifically those occu- 
pying a subset of available habitats. However, in- 
traspecific differences in the capture rates between 
elevational zones of certain species parallel 
changes in species richness as measured by point 
counts and general observations. 



Mixed-Species Flocks 

Nineteen mixed-species flocks were recorded in 
the four elevational zones (Table 18-5). Flock 
members were classified as being either canopy 
or understory foragers, or as foraging at both lev- 
els (Table 18-6). Canopy flocks contained species 
that forage in the lower to upper canopy. Five 
flocks contained canopy foragers plus one or two 
species that foraged at both levels and were clas- 
sified as canopy flocks. Understory flocks con- 



178 



FIELDIANA: ZOOLOGY 



Table 18-5. Elevational distribution of different 
types of mixed-species flocks in RNI d'Andringitra. 



Eleva- 
tion 

(m) 



Flock type 



Canopy Combined Understory Total 



720 


1 


2 





3 


810 


1 


1 


2 


4 


1210 


4 


3 


3 


10 


1625 





1 


1 


2 


Totals 


6 


7 


6 


19 



tained species that forage in the understory. Five 
flocks contained one of four understory foragers 
and a bilevel forager (Phyllastrephus cinereiceps); 
these were considered understory flocks. 

Thirty-seven species were recorded in the 
mixed-species flocks (Table 18-6). This total in- 
cludes three species (Coracopsis vasa, Pseudo- 
cossyphus sharpei, and Newtonia amphichroa) 
not previously reported as members of mixed-spe- 
cies flocks (Langrand, 1990; Eguchi et al., 1992, 
1993; Razaflmahaimodison & Andrianantenaina, 
1993; Goodman, pers. obs.). Newtonia amphich- 
roa was thrice found in combined flocks. The 
presence of the other two species may have been 



Table 18-6. Members of mixed-species flocks, the level at which they forage, and their percentage of occurrence. 
The possible number of flocks in which they might occur is based on the number of flock types (Table 18-5) in the 
elevational range of each species (Table 18-1). 





Foraging 


Actual no. 


Possible no. 


Percentage 


Species 


level* 


of flocks 


of flocks 


of flocks 


Coracopsis vasa 


C 


1 


12 


8 


Coua reynaudii 


U 


1 


12 


8 


Coua caerulea 


C 


1 


13 


8 


Philepitta castanea 


B 


1 


19 


5 


Neodrepanis coruscans 


C 


1 


5 


20 


Neodrepanis hypoxantha 


C 


2 


8 


25 


Coracina cinerea 


C 


3 


13 


23 


Phyllastrephus madagascariensis 


U 


2 


12 


17 


Phyllastrephus zosterops 


U 


1 


10 


10 


Phyllastrephus cinereiceps 


B 


7 


16 


44 


Hypsipetes madagascariensis 


C 


5 


13 


38 


Pseudocossyphus sharpei 


U 


1 


12 


8 


Nesillas typica 


U 


1 


12 


8 


Cryptosylvicola randrianasoloi 


c 


2 


10 


20 


Randia pseudozosterops 


c 


4 


12 


33 


Newtonia amphichroa 


u 


3 


12 


25 


Newtonia brunneicauda 


c 


3 


13 


23 


Neomixis tenella 


c 


2 


12 


17 


Neomixis viridis 


c 


2 


13 


15 


Neomixis striatigula 


c 


2 


13 


15 


Hartertula ftavoviridis 


u 


4 


12 


33 


Pseudohias wardi 


c 


4 


12 


33 


Terpsiphone mutata 


c 


3 


13 


23 


Oxylabes madagascariensis 


u 


5 


12 


42 


Nectarinia souimanga 


c 


5 


13 


38 


Nectarinia notata 


c 


1 


13 


8 


Zosterops maderaspatana 


c 


8 


13 


62 


Calicalicus madagascariensis 


c 


6 


13 


46 


Vanga curvirostris 


B 


1 


19 


5 


Leptopterus viridis 


c 


5 


13 


38 


Leptopterus chabert 


c 


1 


12 


8 


Cyanolanius madagascarinus 


c 


4 


12 


33 


Hypositta corallirostris 


c 


1 


5 


20 


Tylas eduardi 


c 


7 


13 


54 


Dicrurus forficatus 


c 


6 


13 


46 


Ploceus nelicourvi 


B 


5 


19 


26 


Foudia omissa 


u 


2 


10 


20 



C = canopy forager, U = understory forager, B = forages at both levels. 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



179 



coincidental (e.g., Coua reynaudii with Pseudo- 
cossyphus sharpei or Coracopsis vasa with a can- 
opy flock). The number of species in each flock 
type varied. Canopy flocks ranged from three to 
12 species (7.8 ± 3.5, mean ± standard deviation, 
N = 6), and combined flocks from two to 16 (7.9 
± 4.9, N = 7), whereas understory flocks always 
had two species (N = 6). Canopy and combined 
flocks each had on average more species than un- 
derstory flocks, but they did not differ from each 
other in this respect (Mann- Whitney test: W = 43, 
P = 0.94). 

Individual species occurred in 5% to 62% of 
the mixed-species flocks (Table 18-6). Razafima- 
haimodison and Andrianantenaina (1993, Table 3) 
classified species into four groups based on the 
frequency with which they occurred in mixed-spe- 
cies flocks. We made too few observations to sim- 
ilarly classify all of the species that were ob- 
served. However, we found two canopy species 
(Zosterops maderaspatana and Tylas eduardi) 
that occurred in over 50% of the canopy and com- 
bined flocks and correspond to their classification 
of a constant or very frequent member ("especes 
constantes"). 



Differences Between Disturbed and 
Undisturbed Forest 

The placement of our elevational zones was de- 
signed in part to investigate the potential effects 
of habitat degradation in forested areas at approx- 
imately the same elevation and botanical com- 
munities, but with differences in the level of hu- 
man disturbance. The 720 m zone was in an area 
adjacent to a small settlement of slash-and-burn 
(tavy) agriculturalists, and close to the Iantara 
River and a trail running along the eastern side of 
the RNI d'Andringitra. The 810 m site showed 
fewer signs of human disturbance (see Chapter 4 
for descriptions of these two botanical commu- 
nities). 

We documented the presence of 82 resident 
bird species in the 720 m zone and 66 resident 
bird species in the 810 m zone (Table 18-1). 
Twenty-four resident species were found at 720 m 
and not at 810 m. Of these species, three were 
associated with relatively large and slow-moving 
rivers {Tachybaptus pelzelnii, Phalacrocorax af- 
ricanus, and Ardea purpurea), a habitat that did 
not occur in the 810 m zone. Fourteen of the 24 
species were associated with open areas or hu- 
man-modified habitats: Bubulcus ibis, Milvus mig- 



rans, Falco newtoni, Turnix nigricollis, Centropus 
toulou, Agapornis cana, Tyto alba, Caprimulgus 
madagascariensis, Cypsiurus parvus, Phedina 
borbonica, Saxicola torquata, Cisticola cherina, 
Corvus albus, and Lonchura nana. The balance 
included aerial foragers that were almost certainly 
missed at 810 m because of the closed canopy 
(Apus spp.), relatively rare species found at ele- 
vations higher than 810 m and so presumably 
overlooked at that site (Brachypteracias leptoso- 
mus), and those whose upper elevational limit in 
the reserve may have been 720 m (Brachyptera- 
cias squamiger). The absence of three species at 
810 m that were found at 720 m (Treron australis, 
Cyanolanius madagascarinus, and Hartlaubius 
auratus) may simply have been due to chance. 

Eight species were found at 810 m and not at 
720 m. All of these were forest-dwelling birds, 
two of which had their lower elevational limits at 
the upper end of the 810 m zone (Phyllastrephus 
cinereiceps and Cryptosylvicola randrianasoloi). 
Four species were very rare along the eastern 
slopes of the RNI d'Andringitra: Accipiter mad- 
agascariensis, Falco zoniventris, Cuculus aude- 
berti, and Xenopirostris polleni, and their appar- 
ent absence at 720 m may again be chance. The 
remaining two species, Asio madagascariensis 
and Foudia omissa, have been recorded below 
810 m at other sites on Madagascar, and their ab- 
sence from the 720 m zone cannot be simply a 
question of altitudinal distribution. 

Now we return to the question of whether there 
were any noticeable additions or deletions to the 
avifaunas of the 720 and 810 m zones that could 
be attributed to human habitat modification. The 
major difference in these zones is the presence of 
about 14 species in the disturbed areas, particu- 
larly in the immediate area of the Ambarongy 
tavy, that were not found at 810 m. These were 
all species associated with heavily modified en- 
vironments. The areas bordering the Iantara River 
below our 720 m camp had been heavily dis- 
turbed, and presumably this disturbance provided 
a corridor from the lowlands and coastal area for 
species to colonize the open areas and forest edge 
near Ambarongy. With the current data it is dif- 
ficult to demonstrate the disappearance of forest 
species found at 810 m from the 720 m zone, even 
though there is some evidence to suggest it. 

Bird Hunting 

The trail between Ambatomboay and Ambala- 
manenjana is regularly traveled by people carry- 



180 



FIELDIANA: ZOOLOGY 



ing supplies between the two villages. Some days 
up to 20-30 people moved along the trail, many 
of whom carried slings to hunt birds and, occa- 
sionally, lemurs. Generally, if a shot bird was still 
alive, the remiges would be removed and the bird 
carried to a resting site along the trail, where it 
would be cooked and eaten. Numerous feather re- 
mains were found in the vicinity of fire pits at the 
Korokoto River crossing. 

As we traveled up and down this trail we re- 
covered the feathers of recently plucked birds. 
Thus we were able to identify species and roughly 
quantify the number of individuals taken. Records 
have been divided into two sections of the trail: 
(1) Ambalamanenjana to Col de Vohipia — Coua 
caerulea, two individuals, and Atelornis pittoides, 
two individuals; (2) Col de Vohipia to 720 m 
camp — Accipiter francesii, one individual; Coua 
caerulea, seven individuals; Coua reynaudii, four 
individuals; Atelornis pittoides, two individuals; 
Brachypteracias squamiger, one individual; and 
Coracina cinerea, one individual. 

Many of the feather remains were found after 
the start of our surveys, and this apparent rise in 
hunting activity was almost certainly related to an 
increase in traffic along the trail associated with 
porters supplying and shifting our camps. Feather 
remains of a plucked Canirallus kioloides and a 
Coua reynaudii were found within the reserve 
along the trail leading from the 810 to 1210 m 
camps. These remains are almost certainly those 
of porters' meals. Although we found no signs of 
it, several local people mentioned that Lophotibis 
cristata is captured for its flesh using snares and 
a variety of different traps. 

With the exception of Coracina, all of the spe- 
cies hunted are relatively large-bodied birds (> 
100 g), and the majority tend to be ground-dwell- 
ing, or at least occur in the lower canopy of the 
forest. Furthermore, many of these species, when 
disturbed, fly to a nearby perch and remain mo- 
tionless; they are then easily shot by a keen-eyed 
hunter. In conclusion, there is hunting pressure on 
couas, ground-rollers, and other relatively large- 
bodied birds along trails that pass through forest- 
; ed areas. 



is already available in the general literature on the 
Malagasy avifauna, we have concentrated on new 
or interesting natural history aspects of selected 
species. 



Tachybaptus pelzelnii 

The Madagascar Little Grebe was found at sev- 
eral sites along the Iantara River from near the 
junction of the Korokoto River to below Amba- 
tomboay. These sites were generally in areas 
where the river was relatively broad and slow- 
moving. On 13 September 1993, near Ambaron- 
gy, at 720 m elevation, two birds in breeding 
plumage were observed performing a courtship 
display, which included head-tossing behavior. 
Although this species was observed along the Ian- 
tara during both periods we occupied the 720 m 
camp, no definite evidence of local breeding was 
found. 



Phalacrocorax africanus 

The Reed Cormorant was observed on a few 
occasions along the Iantara River. On the large 
slow-moving rivers outside of the reserve, such as 
the Manambolo and the Mananantanana, this spe- 
cies was relatively common. 



Lophotibis cristata 

This species, the Madagascar Crested Ibis, was 
not commonly detected in the reserve. It was 
found only in the 720 and 810 m zones, generally 
along riverbanks or in areas with wet ground. 



Aviceda madagascariensis 

The Madagascar Cuckoo-falcon was recorded 
on one occasion at 1210 m elevation. The bird 
was observed flying across the Volotsangana Riv- 
er valley. 



Selected Species Accounts 



Polyboroides radiatus 



Rather than reiterating general information on 
the breeding season, food habits, and distribution 
of birds occurring in the RNI d'Andringitra that 



Between 5 and 13 October 1993, two adult 
Madagascar Harrier Hawks were observed near to 
or resting on a nest at 850 m elevation and about 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



181 



20 m from the Sahavatoy River. We found no ev- 
idence that the pair was breeding during our visit. 



Accipiter henstii 

On 18 November 1993, at 720 m elevation, an 
adult Henst's Goshawk was observed flying over 
the forest behind the Ambarongy tavy. The bird 
called continuously as it flew in circles over the 
canopy. This behavior continued for about 10 
min. 

On 22 October 1993, at 1210 m and 10:00 am, 
near the 1210 m camp, a shadow was observed 
moving over the trail, and a large raptor was 
found perched in a tree 5-8 m away and 6 m off 
the ground. The bird had white extending from 
the throat to the vent. The white underparts were 
barred with dark gray from the chin or throat to 
the vent. The tail was long and alternately barred 
with light and dark gray bars; the latter bars were 
narrower. The feet were yellow and the talons 
dark. The bird was observed for 15 s or less be- 
fore it flew off. The wings were broad and round- 
ed. They appeared dark brown with no noticeable 
patterning. Neither the rump nor the upper side of 
the tail was observed. There was no visible ruff 
of nape feathers, although the bird was seen only 
frontally. The only two species that fit this de- 
scription are Accipiter henstii and Eutriorchis as- 
tur (the Madagascar Serpent Eagle). 

Eutriorchis astur is exceptionally rare and until 
recently was thought to be on the verge of ex- 
tinction. In the past few years it has been found 
at sites in northeastern Madagascar (Sheldon & 
Duckworth, 1990; Raxworthy & Colston, 1992; 
Thorstrom et al., 1995). Whether the 22 October 
1993 observation was of this species remains in- 
conclusive. However, on the basis of forest type 
and size, it is quite possible that E. astur inhabits 
the RNI d'Andringitra. 



er Malagasy rain forests, Accipiter spp. were ex- 
ceptionally uncommon in the RNI d'Andringitra. 
France's Sparrowhawk was recorded in the 720 
and 810 m transect zones. 



Buteo brachypterus 

While being watched by an arboreal group of 
brown lemurs (Eulemur fulvus) giving alarm calls, 
one of us (M.S.R) whistled imitations of the Mad- 
agascar Buzzard's call. Before the imitations were 
given, the lemurs looked only toward the ground. 
During and following imitations of the hawk's call 
the lemurs alternated between looking toward the 
ground and scanning the sky. This observation 
suggests that brown lemurs recognize the calls of 
potential avian predators. 

A recent review (Goodman et al., 1993) re- 
ported a brown lemur group giving alarm calls 
upon seeing a large raptor and a case of an adult 
being killed by an unidentified raptor. Harrington 
(1975) reported brown lemurs looking at flying 
Buteo and sometimes giving soft grunts. Two 
species of lemurs in captivity {Lemur catta and 
Varecia variegata) vocalized in response to the 
calls of North American Buteo spp. (Macedonia, 
1993). 



Falco zoniventris 

The only record we have of the Banded Kestrel 
in the reserve is one observed on 24 November 
1993 at the forest edge near the Sahanivoraky 
River, just below the 810 m camp. 

Falco peregrinus 



Accipiter madagascariensis 

The only record we have of the Madagascar 
Sparrowhawk in the reserve is one observed at 
810 m elevation on 10 October 1993. 



Accipiter francesii 

Accipiter spp. were seldom observed, and none 
was netted. On the basis of our experience in oth- 



This species, the Peregrine Falcon, was ob- 
served only in the 1210 m zone. The numerous 
sheer cliff faces in the reserve might provide nest- 
ing sites for this species. 



Turnix nigricollis 

During the survey, we only found the Mada- 
gascar Buttonquail in or near the tavy at Amba- 
rongy, at 720 m elevation. 



182 



FIELDIANA: ZOOLOGY 






Mesitornis unicolor 



Coracopsis nigra 



The Brown Mesite was recorded between 720 
and 1625 m elevation, although on the basis of 
general observations it appeared to be less com- 
mon above 1210 m. The previous highest reported 
elevation for this species was 900 m (Langrand, 
1990). On occasion the Brown Mesite was ob- 
served on tree limbs or suspended dead branches 
1-5 m off the ground. Several times birds were 
scared up from the forest floor. They would often 
fly 10 or 20 m, land on a tree perch, and remain 
motionless. After a period of a few minutes they 
would walk deftly along branches and then jump 
to the ground. On several occasions birds were ob- 
served with the atypical plumage coloration de- 
scribed by Langrand (1990, plate 15, fig. 77b, p. 
150), as well as with gray rather than pale orange 
or black bills. 



The Lesser Vasa Parrot was observed feeding 
on the following plants: Cinnamosma fragrans 
(Canellaceae) — at 1650 m, they would remove 
exocarp from seed and consume the latter; Sym- 
phonia sp. (Clusiaceae) — at 870 m, they fed on 
unopened blossoms and left behind only the 
flower base and petiole; and Chassalia sp. (Ru- 
biaceae) — at 1210 m, they consumed whole 
fruits. 



Coracopsis vasa 

The Greater Vasa Parrot was distinctly less 
common than C. nigra and was not recorded 
above 1210 m elevation. 



Sarothrura insularis 

On the basis of general observations and calls, 
this species, the Madagascar Flufftail, was record- 
ed in all forested zones between 650 and 1 800 m. 
A few individuals were heard calling in areas of 
tavy near Ambarongy at 720 m and in the heath- 
land at about 2200 m. On 28 October 1993, at 
about 1700 m elevation, one individual was ob- 
served carrying dried grass in its bill. 



Streptopelia picturata 

On 10 October 1993, at 810 m elevation, bro- 
ken eggshells of the Madagascar Turtle Dove 
were found below a nest. The clutch had hatched 
that morning. This species was observed at 710 
m elevation feeding on whole fruits of Ficus re- 
flexa (Moraceae). 



Alectroenas madagascariensis 

Identified Madagascar Blue Pigeon food items 
included Ficus reflexa (Moraceae) — whole fruits, 
at edge of clearing, 710 m; Cryptocarya sp. (Lau- 
raceae) — whole fruits, at 810 m; Pachytrophe sp. 
(Moraceae) — whole fruits, at 730 m; Maesa Ian- 
ceolata (Myrsinaceae) — whole fruits, at 740 m; 
and Oncostemum racemiferum (Myrsinaceae) — 
whole ripe fruits, near forest edge, 730 m (see 
Chapter 4, Appendix 4-2). 



Agapornis cana 

This species, the Gray-headed Lovebird, was 
only recorded in disturbed areas at the lower el- 
evations, including the Ambarongy tavy at 720 m, 
and near Ambatomboay at 650 m elevation. 



Cuculus audeberti 

On 10 October 1993, at 820 m and 8:15 am, a 
Thick-billed Cuckoo was observed perched in the 
top of a 20-m-tall tree about 500 m from the edge 
of the Sahavatoy River. The observer had an 
unobstructed view of the bird and watched it for 
2 or 3 minutes. Notes made at the time describe 
its vocalization as "a single strong note followed 
by a weaker descending note, kind of Green Sand- 
piper [Tringa ochropus] like in quality, and given 
in rapid succession and in couplets of two to 
four." The bird was in adult plumage, with a dis- 
tinctly white and unbarred ventrum, a dark char- 
coal dorsum, and a bicolored bill. It was easily 
distinguishable from Cuculus rochii or other spe- 
cies in Madagascar with which it could be con- 
fused. The only other record of this species in 
Madagascar after 1922 (Langrand, 1990) is from 
the Maromiza Forest near Analamazaotra/Pennet, 
about 400 km to the northeast of the eastern part 
of the RNI d'Andringitra, in late November 1992 
(Langrand & Sinclair, 1994). 

A tape recording of the Thick-billed Cuckoo 
made in South Africa was broadcast at 15 of the 
30 census points in the 720 and 810 m elevational 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



183 



zones. A broadcast was also made within 1 hour 
of the above encounter at the location of the ob- 
servation. None of these tape playbacks elicited a 
vocal response. 



this migratory species was relatively common 
along the Sahavatoy River near 810 m and in the 
vicinity of the 720 m camp. During the second 
trip to the reserve, it was regularly observed or 
heard in the 720 and 810 m zones. 



Cuculus rochii 

This species, the Madagascar Lesser Cuckoo, 
was relatively common in all forested zones be- 
tween 650 and 1800 m. A few individuals were 
heard calling in the heathlands at about 2200 m. 



Tyto alba 

The Barn Owl was recorded only near the 720 
m camp. On several nights this bird was heard 
vocalizing from the direction of the Ambarongy 
tavy. 



Asio madagascariensis 

A roost or hunting perch of this species, the 
Madagascar Long-eared Owl, was found on a 
ridge at about 900 m elevation, just above the 810 
m camp. The bird was seen in the same tree on 
several nights during the course of 1 week. Below 
the perch, two whole and several disintegrated 
pellets were found that contained the remains of 
at least two adult and two subadult Rattus rattus. 
A Madagascar Long-eared Owl was heard calling 
near the village of Ambatomboay during the night 
of 13-14 September 1993. 



Caprimulgus madagascariensis 

On the basis of vocalization, the Madagascar 
Nightjar was relatively uncommon along the Ian- 
tara River in the vicinity of our 720 m camp. This 
species appeared to be more common near Am- 
batomboay, where on the nights of 13 and 14 Sep- 
tember 1993, numerous individuals were heard 
calling. 



Eurystomus glaucurus 

Between 23 September and 13 October 1993, 
the Broad-billed Roller was not recorded in the 
720 and 810 m zones. On 1 November 1993, as 
the first research group was leaving the reserve, 



Brachypteracias leptosomus 

On 21 October 1993, in the 1210 m zone, one 
Short-legged Ground-roller was observed perched 
at a height of 10 m and holding in its bill a 10- 
to 15-cm-long black millipede with yellow legs. 
The bird shook the millipede and seemed to strike 
it against a branch before swallowing the animal 
whole. The bird then flew a short distance away. 
During the time it was dispatching its prey, the 
ground-roller was mobbed by two Neodrepanis 
hypoxantha and several Zosterops maderaspatana 
and Nectarinia souimanga. These birds moved 
about, in positions close to the ground-roller, and 
gave repeated shrill calls. 



Brachypteracias squamiger 

On 28 September 1993, two Scaly Ground-roll- 
ers were observed at 720 m elevation, about 850 
m from our camp and 10 m from the trail that led 
from Ambatomboay to Ambalamanenjana. At 
least one of two birds dropped from a perch to 
the ground and then ran away in short dashes, 
followed by freezing postures. A tape recording 
of this species was played back to the pair. Soon 
one bird, and then the second, began to approach. 
The first bird progressed toward the observer with 
its head lowered to the same level as its back and 
with its tail spread. The bird occasionally changed 
direction as it ran on the ground, keeping the tail 
fanned throughout the sequence. Minutes later the 
bird made harsh grating sounds. Until this point 
the tape recording was being played; soon after 
the tape recording was stopped, the bird began 
calling. A long sequence of vocalizations was giv- 
en, similar to the "coo" calls on the playback 
tape. The bird also performed a wing-flick dis- 
play: with the bird standing in a regular, head-up 
posture and the tail spread, the folded wings were 
rapidly raised above the back and then lowered to 
a normal position. 

While giving a short "coo" call, the bird would I 
begin from a typical upright posture with the head 
up and the bill parallel to the ground. The head 
was lowered to a point where the bill was 45° 



184 



FIELDIANA: ZOOLOGY 



below its starting position. At the bottom of the 
bill's trajectory the "coo" call was given, seem- 
ingly with the bill closed. The head and bill were 
then returned to their starting position. 

This site was in an area of forest along the Ian- 
tara River with gently sloping ground, an open un- 
derstory, and nearly complete leaf litter coverage 
of the ground. On 16 November 1993, within a few 
meters of where the behavioral observations of this 
species were made on 28 September 1993, a bur- 
row was found that contained two Eliurus webbi, 
a nesomyine rodent (see Chapter 22). On 20 No- 
vember 1993, the burrow was excavated, and the 
nest of a Scaly Ground-roller was found within the 
system (Goodman, 1994). On the basis of the lay- 
out of the tunnel, it was clear that the ground-roll- 
ers and rodents were cohabiting the burrow. A sin- 
gle egg was resting in a shallow depression, about 
2 cm deep, and surrounded by a platform of small 
dried leaves. The placement of this nest was similar 
to that reported in another nesting record of this 
species (Benson et al., 1976). 



Atelornis pittoides 

The Pitta-like Ground-roller was regularly 
heard at the edge of the Ambarongy village clear- 
ing. This species appeared to be the least sensitive 
to human disturbance of the four ground-rollers 
recorded in the reserve. 



Atelornis crossleyi 

On 19 October 1993, at 1230 m elevation, B. 
Fisher (Chapter 8) was searching for ants in fallen 
wood. The principal technique was to open rotten 
logs resting on the ground. For about 2.5 h, a 
Rufous-headed Ground-roller followed him. After 
a site was abandoned, the ground-roller would ap- 
proach the log, often within a few meters of B. 
Fisher, and peck at areas of the wood that had 
been opened. Several times the bird was observed 
exposing and eating beetle larvae removed from 
the rotten wood. It deftly climbed and jumped 
over objects, such as large logs and a backpack, 
on the forest floor. Two birds were observed, al- 
though one was more retiring than the other and 
did not approach closely. 

Little is known about the diet of this species. 
Langrand (1990) stated that it chiefly eats insects, 
particularly small Lepidoptera. The stomach con- 
tents of one collected individual contained re- 



mains of Carabidae, Tenebrionidae, and Curcu- 
lionidae beetles, Formicidae, Diplopoda, and Gas- 
tropoda. These prey items are consistent with a 
general terrestrial existence, and this species for- 
ages on the ground and possibly less frequently 
on stumps or downed logs. 

The Rufous-headed Ground-roller was only 
found in the 1210 and 1625 m elevational zones. 
It was observed in areas with relatively wet soil 
and thick forest with large trees. 



Philepitta castanea 

This species, the Velvet Asity, was observed 
feeding on several plants (see also Chapter 4, Ap- 
pendix 4-2): Macaranga oblongifolia (Euphorbi- 
aceae) — at 1650 m, on small orange-colored buds; 
Gravesia caliantha (Melastomataceae) — at 1210 
m, fruits; Medinilla ericarum (Melastomata- 
ceae) — at 1210 m, orangish-red fruits; Pittospo- 
rum sp. (Pittosporaceae) — at 800 m, whole fruits; 
Pauridiantha lyallii (Rubiaceae) — at 810 m and 
1625 m, small cranberry-colored fruits; and Chas- 
salia sp. (Rubiaceae) — at 1210 m, bright red 
fruits. 



Neodrepanis coruscans 

This species, the Sunbird-asity, was observed 
visiting the dull reddish-pink flowers of Bakerella 
clavata (Loranthaceae) at 720 m and the white 
flowers of Liparis sp. (Orchidaceae) at 810 m. 

In late September 1993, three males were col- 
lected. Two of the three were in breeding plumage 
and had well-developed wattles, fully ossified 
skulls, and testes measuring left 4x2 mm and 
right 3X2 mm. The other male was molting into 
breeding plumage and had developed but dull 
wattles, an 85% ossified skull, and testes measur- 
ing left 3x2 mm and right 2X2 mm. 



Neodrepanis hypoxantha 

Little is known about the Yellow-bellied Sun- 
bird-asity. We found it to be relatively common 
between 1210 and 2050 m. It was neither sparse 
nor elusive, as suggested by Benson (1974); nor 
was it "very sparse," as reported for the RNI de 
Marojejy (Safford & Duckworth, 1990). In the 
RNI d'Andringitra, N. hypoxantha was seen dai- 
ly in the 1625 m elevational band. The mean 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



185 



number of detections of N. hypoxantha from 
high counts at census points between 1100 and 
1765 m (x = 0.633, standard deviation [SD] = 
0.850, N = 30) did not differ from that of N. 
coruscans at census points between 720 and 960 
m (jf = 0.667, SD = 0.606, N = 30) (Mann- 
Whitney test, P = 0.54). 

Although we found N. hypoxantha as numer- 
ous within its restricted elevational range as TV. 
coruscans was at lower elevations, the former 
species should remain a concern for conservation 
(Collar & Stuart, 1985; Safford & Duckworth, 
1990). Neodrepanis hypoxantha is known only 
from elevations above 1050 m (Langrand, 1990), 
and so the total area of its range is much smaller 
than that of N. coruscans. Although deforestation 
of lowland forests has been severe (Green & 
Sussman, 1990), high-altitude forests are not im- 
mune to this process, as witnessed by the near 
elimination of the native biota of Ankaratra, one 
of the island's highest peaks (Perrier de la Bath- 
ie, 1927; Nicoll & Langrand, 1989). Many of the 
localities where N. hypoxantha was collected in 
the 19th century are now deforested (Dee, 1986). 

Neodrepanis hypoxantha was observed visiting 
the vermilion-colored flowers of Bakerella cla- 
vata (Loranthaceae) at about 1625 m and the dull 
reddish-pink flowers of Bakerella tandrokensis at 
2050 m. A male N. hypoxantha was observed 
gleaning a small green caterpillar. He passed the 
animal back and forth through his mandibles be- 
fore swallowing it. 

A nest of this species was found at 1600 m and 
was watched on 29 and 30 October 1993 by 
M.S. P. The female made frequent visits to the 
domed nest, lining its interior with dry leaves of 
climbing bamboo. The male perched nearby much 
of the time, displaying when the female was pres- 
ent. He occasionally hung at the nest entrance but 
was not observed entering the structure or engag- 
ing in nest construction. The nest did not have the 
"entrance-roof" typical of the family Eurylaimi- 
dae (Amadon, 1951; Prum, 1993). The construc- 
tion presumably had not been completed. This is 
further supported by the fact that the nest, when 
collected on 30 October 1993, did not contain 
eggs. 

A few specimens of TV. hypoxantha were col- 
lected in late October 1993. These included a fe- 
male, on 20 October 1993, with a slightly en- 
larged oviduct (largest follicle 2 mm) and a fully 
ossified skull. Two males were collected, one with 
testes measuring left 5X3 mm, right 4X3 mm, 
and the other with testes left 7X5 mm, right 6 



186 



X 4 mm. Both had fully ossified skulls and well- 
developed wattles. The coloration of the soft parts 
of the two males differed. In one the portion of 
the caruncle around the eye was brilliant aqua or 
cobalt blue, and the balance was brilliant lime 
green. In the other male the portion surrounding 
the eye was almost black and the remaining sec- 
tion cobalt blue. The coloration of the proximal 
part of the bill in both males paralleled that found 
on the caruncle. Specifically, the area around the 
nares was the same color as the inner portion of 
the caruncle, and the area at the base of the bill 
was the same color as the outer portion of the 
caruncle. In both males the eyelid was bright lime 
green. 



Phedina borbonica 

Flocks of up to 30 Mascarene Martins were oc- 
casionally noted along the Iantara River, resting 
on dead limbs, in banana trees, or on the river- 
banks. 



Hypsipetes madagascariensis 

We observed the Madagascar Bulbul feeding 
on several food plants (see also Chapter 4, Ap- 
pendix 4-2): Oncostemum racemiferum (Myrsi- 
naceae) — at 730 m, ripe fruits; Pauridiantha lyal- 
Hi (Rubiaceae) — at 810 m, small cranberry-col- 
ored fruits; Bakerella sp. (Loranthaceae) — at 120C 
m, fruits; Medinilla ericarum (Melastomata- 
ceae) — at 1210 m, orangish-red fruits; Chassalic 
sp. (Rubiaceae) — at 1210 m, bright red fruits; anc 
Vaccinium secundiflorum (Ericaceae) — at 162* 
m, green-colored fruits. 



Saxicola torquata 

This species, the Stonechat, was relatively com 
mon in disturbed areas, particularly open tav 
fields, near our 720 m camp. It was also observe< 
in the open heathland above 1950 m elevation. 



Cryptosylvicola randrianasoloi 

The type locality of the Cryptic Warbler, a re 
cently described genus and species, is the easter 
slope of the RNI d'Andringitra (Goodman et al 
1996), where its lower elevational limit was 90 



FIELDIANA: ZOOLOG i 



m. Several individuals were observed or heard vo- 
calizing in areas of ericaceous vegetation up to 
2100 m. A nest of this species with three eggs 
was found at 1625 m elevation on 24 October 
1993. 

Dromaeocercus brunneus 



elevation, feeding three fledglings. The adult en- 
tered relatively open areas, gave a contact call, 
and then the fledglings would scurry out of the 
thick herbaceous understory to be fed. Only one 
adult was seen at a time feeding the young birds. 



Nectarinia notata 



We only have one record of the Brown Emutail 
in the reserve. On 11 December 1993, two birds 
were observed at about 1650 m elevation in an 
open area with thick herbaceous vegetation asso- 
ciated with a rock slide. 



Dromaeocercus seebohmi 

The Grey Emutail was only found in the heath- 
land zone above 1900 m. This rather skulking 
species occurred in the low and thick heath veg- 
etation. On the basis of the number of birds call- 
ing, it was relatively common. 



On 28 September 1993, at 810 m elevation, 
one adult male Long-billed Green Sunbird was 
observed chasing a female. This male gave a 
chirping note. At one point he assumed a possi- 
ble display posture; he perched with his neck ex- 
tended upward, the bill slightly agape, and with 
the closed wings lifted slightly away from his 
body. The wings were raised and lowered several 
times. 

At 2050 m elevation, this species, along with 
Neodrepanis hypoxantha and Nectarinia soui- 
manga, was observed visiting the dull reddish- 
pink flowers of Bakerella tandrokensis (Loran- 
thaceae). 



Rand ia pseudozosterops 

The majority of our records of Rand's Warbler 
within the reserve are from tree perches along riv- 
er margins between 720 and 1210 m. It was often 
seen or heard singing from exposed trees. 



Terpsiphone mutata 

On 1 November 1993, at about 850 m eleva- 
tion, eggshells of recently hatched birds were 
found below a nest of the Madagascar Paradise 
Flycatcher. 



Nectarinia souimanga 

We were able to gather some information on 
the food habits of this bird, the Souimanga Sun- 
bird. It was often seen feeding on various species 
of Bakerella (Loranthaceae). At 720, 810, and 
1625 m the local species was Bakerella clavata, 
and at 2050 m it was B. tandrokensis. The Soui- 
manga Sunbird was also observed at 810 m vis- 
iting the white-colored flowers of Liparis sp. (Or- 
chidaceae); at 1210 m the lavender-colored flow- 
ers of Plectranthus sp. (Lamiaceae); and at 1650 
m the white-colored flowers of Medinilla torren- 
tum. It apparently pierced flower buds with its 
bill. 



Oxylabes madagascariensis 

Schulenberg (1991) described the duet song of 
this species, the White-browed Oxylabes. On sev- 
eral occasions in the RNI d'Andringitra we heard 
and observed pairs of birds giving similar types 
of vocalizations. 



Crossleyia xanthophrys 

An adult Yellow-browed Oxylabes was ob- 
served on 3 and 4 December 1993, at 1220 m 



Zosterops maderaspatana 

We have a few records on the food habits of 
the Madagascar White-eye. At 720, 810, and 1625 
m it was seen feeding on the vermilion-colored 
flowers of Bakerella clavata (Loranthaceae). It 
was attracted to the fruits of Oncostemum racem- 
iferum (Myrsinaceae) at 730 m, but it was not 
clear whether it was actually eating them or in- 
sects. At 810 m it was seen feeding on the small 
cranberry-colored fruits of Pauridiantha lyallii 
(Rubiaceae). It was seen picking fruits of Vernon- 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



187 



ia alleizettei (Asteraceae) at about 1300 m, and, 
at 1625 m it was plucking and eating flower buds 
of Oncostemum ankifiense (Myrsinaceae). 



Schetba rufa 

On 28 November 1993, at 810 m elevation, an 
adult Rufous Vanga was observed feeding a fledg- 
ling. 



Xenopirostris polleni 

The only record we have of Pollen's Vanga in 
the reserve is one on 8 October 1993 at 810 m 
elevation. 



Foudia madagascariensis 

All of our records of the Madagascar Red Fody 
are from the tavy of Ambarongy at 720 m and in 
the heathland between 1950 and 2150 m. We 
found no evidence of this species in the forest 
proper. 



Foudia omissa 

The Forest Fody was only recorded in forested 
areas. We found no evidence of this species co- 
existing with the Madagascar Red Fody, nor was 
there any clear evidence of hybridization between 
them. At 1200 m this species was observed feed- 
ing at the tubular red flowers of Bakerella sp. 
(Loranthaceae). 






Hypositta corallirostris 

On 13 October 1993, at 810 m elevation, a 
male and female Nuthatch Vanga were observed 
foraging on the trunks of larger trees. The birds 
characteristically spiraled upward, very creeper- 
like (Family Certhiidae), along the bole and larger 
branches. The birds dropped down to the trunk of 
a nearby tree and then began to work upward. 
Once the male sallied after a flying insect that was 
presumably disturbed from the bark of the tree; 
he then returned to the trunk. Both the male and 
female gave a high-pitched "tsee" call just as 
they flew from one tree to another. 



Acridotheres tristis 

This introduced species, the Common Myna, 
was not recorded in the eastern portion of the RNI 
d'Andringitra. On the northern and western sides 
of the reserve, particularly in rice paddies and 
near villages, it was seen often. 



Ploceus nelicourvi 

In September and October 1993, male Neli- 
courvi Weavers were observed at several sites in 
the process of weaving nests. In virtually all 
cases, the nests were suspended from branches or 
limbs of trees over rivers and streams. This may 
be an adaptation against snake or lemur predation 
on the eggs and young. 



Lonchura nana 

The Madagascar Mannikin was only found at 
720 m in the tavy of Ambarongy. 



Acknowledgments 

We are grateful to members of the expedition, 
particularly Brian Fisher and Eleanor Sterling, for 
sharing their bird observations with us. Philip Par- 
rillo identified invertebrate remains found in bird 
stomachs. Frank Hawkins, Olivier Langrand, Tim 
Moermond, Tom Schulenberg, Steve Zack, and an 
anonymous reviewer kindly provided comments 
on an earlier draft of the manuscript. 



Literature Cited 

Able, K. P., and B. R. Noon. 1976. Avian community 
structure along elevational gradients in the northern 
United States. Oecologia, 26: 275-294. 

Amadon, D. 1951. Le pseudo-souimanga de Madagas- 
car. Oiseau et R. F. O., 21: 59-63. 

Andrianarimisa, A. 1993. Variation structurale de la 
communaute avienne d'une foret dense seche semi- 
caducifoliee dans l'Ouest Malgache. Annales Musee 
Royal de l'Afrique Centrale (Zoologie), 268: 187— 
191. 

Benson, C. W. 1974. Another specimen of Neodrepanis 
hypoxantha. Bulletin of the British Ornithologists' 
Club, 94: 141-143. 

, J. F. R. COLEBROOK-ROBJENT, AND A. WILLIAMS. 

1976. Contribution a l'ornithologie de Madagascar. 



188 



FIELDIANA: ZOOLOGY 



L'Oiseau et Revue Franchise d'Ornithologie, 46: 209- 
242. 

Chapman, F. M. 1917. The distribution of bird-life in 
Colombia; a contribution to a biological survey of 
South America. Bulletin of the American Museum of 
Natural History, 36: 1-729. 

Collar, N. J., and S. N. Stuart. 1985. Threatened 
birds of Africa and related islands. The ICBP/IUCN 
Red Data Book, part 1, 3rd edition. International 
Council for Bird Preservation and International Union 
for Conservation of Nature and Natural Resources, 
Cambridge, 761 pp. 

Dee, T. J. 1986. The endemic birds of Madagascar. In- 
ternational Council for Bird Preservation, Cambridge, 
173 pp. 

Diamond, J. M., and M. LeCroy. 1979. Birds of Kar- 
kar and Bagabag islands, New Guinea. Bulletin of the 
American Museum of Natural History, 164: 467-531. 

Eguchi, K., S. Yamagishi, H. Nagata, M. Nakura, and 
V. Randrjanasolo. 1992. The mixed-species flocks 
of forest-living birds in Madagascar, pp. 28-38. In Ya- 
magishi, S., ed., Social structure of Madagascar higher 
vertebrates in relation to their adaptive radiation. Osa- 
ka City University, Osaka, 92 pp. 

Eguchi, K., S. Yamagishi, and V. Randrianasolo. 
1993. The composition and foraging behaviour of 
mixed-species flocks of forest-living birds in Mada- 
gascar. Ibis, 135: 91-96. 

Finch, D. M. 1989. Habitat use and habitat overlap of 
riparian birds in three elevational zones. Ecology, 70: 
866-880. 

Goodman, S. M. 1994. A description of the ground 
burrow of Eliurus webbi (Nesomyinae) and case of 
cohabitation with an endemic bird (Brachypteraciidae, 
Brachypteracias). Mammalia, 58: 670-672. 

, and P. C. Gonzales. 1990. The birds of Mt. 

Isarog National Park, southern Luzon, Philippines, 
with particular reference to altitudinal distribution. 
Fieldiana: Zoology, new series, 60: 1-39. 

-, O. Langrand, and B. M. Whitney. 1996. A 



new genus and species of passerine from the eastern 
rain forest of Madagascar. Ibis, 138: 153-159. 

-, S. O'Connor, and O. Langrand. 1993. A re- 



view of predation on lemurs: implications for the evo- 
lution of social behavior in small, nocturnal primates, 
pp. 51-66. In Kappeler, P. M., and J. Ganzhorn, eds., 
Lemur social systems and their ecological basis. Ple- 
num Press, New York, 274 pp. 

Graves, G. R., M. B. Robbins, and J. V. Remsen, Jr. 
1983. Age and sexual difference in spatial distribution 
and mobility in manakins (Pipridae): inferences from 
mist-netting. Journal of Field Ornithology, 54: 407- 
412. 

Green, G. M., and R. W. Sussman. 1990. Deforestation 
history of the eastern rain forests of Madagascar from 
satellite images. Science, 248: 212-215. 

Harrington, J. E. 1975. Field observations of social 
behavior of Lemur fulvus fulvus E. Geoffroy 1812, pp. 
259-279. In Tattersall, I., and R. W. Sussman, eds., 
Lemur biology. Plenum Press, New York, 365 pp. 

Hawkins, F 1993. Relationships among vegetation 
structure and White-breasted Mesite Mesitornis var- 
iegata presence and foraging site choice. Annales Mu- 



see Royal de TAfrique Centrale (Zoologie), 268: 155— 
164. 

Karr, J. R. 1981. Surveying birds with mist nets. Stud- 
ies in Avian Biology, 6: 62-67. 

Langrand, O. 1990. Guide to the birds of Madagascar. 
Yale University Press, New Haven, 364 pp. 

, and J. C. Sinclair. 1994. Additions and sup- 
plements to the Madagascar avifauna. Ostrich, 65: 
302-310. 

, and L. Wilme. 1993. Protection des zones 



humides et conservation des especes d'oiseaux end6- 
miques de Madagascar. Annales Musee Royal de 
TAfrique Centrale (Zoologie), 268: 201-208. 

Macedonia, J. M. 1993. Adaptation and phylogenetic 
constraints in the antipredator behavior of ringtailed 
and ruffed lemurs, pp. 67-84. In Kappeler, P. M., and 
J. Ganzhorn, eds.. Lemur social systems and their eco- 
logical basis. Plenum Press, New York, 274 pp. 

Nicoix, M. E., and O. Langrand. 1989. Madagascar: 
revue de la conservation et des aires protegees. Fonds 
Mondial pour la Nature, Gland, xvii + 374 pp. 

O'Keeffe, J., and L. Ashmore. Not dated. Madagascar 
1993 small mammal project. Unpublished Report, 
Emmanuel College, Cambridge University [available 
from S. M. Goodman]. 

Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. RCP 225. Etudes des eco- 
systemes montagnards dans le region malgache. I. Le 
massif de l'Andringitra. 1970-1971. Geomorphologie, 
climatologie et groupements vlgltaux. Bulletin de la 
Soci&e" d'Ecologie, 11(2-3): 198-226. 

Perrier de la Bathie, H. 1927. Le Tsaratanana, l'An- 
karatra et l'Andringitra. M6moires de I'Acad^mie 
Malgache, 3: 1-71. 

Prum, R. O. 1993. Phylogeny, biogeography, and evo- 
lution of the broadbills (Eurylaimidae) and asities 
(Philepittidae) based on morphology. Auk, 110: 304- 
324. 

Raxworthy, C. J., and P. R. Colston. 1992. Conclu- 
sive evidence for the continuing existence of the Mad- 
agascar Serpent-eagle Eutriorchis astur. Bulletin of 
the British Ornithologists' Union, 112: 108-111. 

Razafimahaimodison, J.-C, and R. Andrianantenaina. 
1993. Ecologie et comportement des groupes d'ois- 
eaux dans une foret dense humide de Madagascar. An- 
nales Musee Royal de TAfrique Centrale (Zoologie), 
268: 197-200. 

Remsen, J. V., Jr., and T A. Parker. 1983. Contribu- 
tion of river-created habitats to bird species richness 
in Amazonia. Biotropica, 15: 223-231. 

Safford, R., and W Duckworth. 1990. A wildlife sur- 
vey of Marojejy Nature Reserve, Madagascar. Inter- 
national Council for Bird Preservation, Study Report 
no. 40. 

Schulenberg, T. 1991 . On the songs of Oxylabes mad- 
agascariensis. Working Group on Birds of the Mad- 
agascar Region Newsletter, 1:8. 

. 1993. Phylogeny of the Vangidae: inferences 

from mitochondrial DNA. Annales Musee Royal de 
TAfrique Centrale (Zoologie), 268: 23-28. 

-, S. M. Goodman, and J.-C. Razafimahaimodi- 



son. 1993. Genetic variation in two subspecies of 



GOODMAN & PUTNAM: BIRDS OF THE EASTERN SLOPES 



189 



Nesillas typica (Sylviinae) in south-east Madagascar. 
Annales Musee Royal de l'Afrique Centrale (Zoolo- 
gie), 268: 173-177. 

Sheldon, B. C, and J. W. Duckworth. 1990. Redis- 
covery of the Madagascar Serpent-eagle Eutriorchis 
astur. Bulletin of the British Ornithologists' Club, 
110: 126-130. 

Terborgh, J. 1977. Bird species diversity on an Andean 
elevational gradient. Ecology, 58: 1007-1019. 

Thompson, P. M., ed. 1987. Zahamena Forest (Mada- 
gascar) expedition report 1985. International Council 
for Bird Preservation, Study Report no. 20. 

, and M. I. Evans. 1991. The birds of Amba- 



tovaky, pp. 4-1 to 4-29. In Thompson, P. M., and M. 
I. Evans, eds., A survey of Ambatovaky Special Re- 
serve, Madagascar. Madagascar Environmental Re- 
search Group, London. 

Thorstrom, R., R. T. Watson, B. Damary, F. Toto, M. 
Baba, and V. Baba. 1995. Repeated sightings and 
first capture of a live Madagascar Serpent-eagle Eu- 
triorchis astur. Bulletin of the British Ornithologists' 
Club, 115: 40-44. 

Watson, R. T, J. Berkelman, R. Lewis, and S. Raza- 
findramanana. 1993. Conservation of the Madagas- 
car Fish Eagle Haliaeetus vociferoides. Annales Mu- 
see Royal de l'Afrique Centrale (Zoologie), 268: 192— 
196. 



190 



FIELDIANA: ZOOLOGY 



Chapter 19 



The Shrew Tenrecs (Microgale) 
(Insectivora: Tenrecidae) of the 
Reserve Naturelle Integrate 
d'Andringitra, Madagascar 

Paulina D. Jenkins, Steven M. Goodman, and 
Christopher J. Raxworthy 



Abstract 

A collection of Microgale species made in humid forest between 720 and 1625 m elevation 
on the eastern slopes of the Reserve Naturelle Integrate d'Andringitra in late 1993 is reviewed. 
The material contains a total of 10 taxa, including two previously unknown species, one of 
which is described; two species that were known only from the type locality (M. parvula and 
M. soricoides); and three taxa (M. cowani, M. taiva, and M. melanorrhachis) for which the 
new material allows redefinition of diagnostic characters. We formally remove M. taiva and 
M. melanorrhachis from synonymy with M. cowani but provisionally synonymize M. pulla 
with M. parvula. Information is presented on external and craniodental morphology, measure- 
ments, variation, and reproduction. 

Resume 

La collection d'especes de Microgale effectuee dans la foret humide entre 720 m et 1625 m 
d' altitude sur le versant est de la Reserve Naturelle Integrate d'Andringitra effectuee a la fin 
de 1'annee 1993 fait 1'objet d'une discussion. 

Le materiel echantil tonne" comprend un total de dix taxons, y compris: deux especes encore 
inconnues auparavant, dont une est decrite ci-dessous et deux especes qui ne sont connues que 
localement (Af. parvula et M. soricoides) et enfin trois especes (M. cowani, M. taiva, et M. 
melanorrhachis), pour lesquelles les nouveaux echantillons permettant une redefinition des 
caracteres utilises pour 1' identification des especes. 

Nous avons abandon n£ la synonymie de M. taiva et M. melanorrhachis avec M. cowani, 
mais avons donnd M. pulla et M. parvula comme synonymes. Des informations relatives a la 
morphologie externe et craniodentaire, aux mesures biomemques, a leurs variations et a la 
reproduction sont presentees. 

Introduction monly occurring as a commensal with humans 
The insectivores of Madagascar are composed and almost certainly introduced (Hutterer & Tra- 
of two families: (1) the Soricidae, including two nier, 1990), and Suncus madagascariensis (Co- 
species, Suncus murinus (Linnaeus, 1766), com- querel, 1848); and (2) the Tenrecidae. With the 

JENKINS ET AL.: SHREW TENRECS 191 



exception of the subfamily Potamogalinae, the 
other three subfamilies of Tenrecidae are endemic 
to Madagascar. The degree of morphological vari- 
ation within the Tenrecidae is absolutely remark- 
able, including the spiny "hedgehog"-like genera 
Echinops Martin, 1838, Hemicentetes Mivart, 
1871, Setifer Froriep, 1806, and Tenrec Lacepede, 
1799; the aquatic genus Limnogale Major, 1896a; 
the semi-fossorial Oryzorictes A. Grandidier, 
1870; and the shrew-like Microgale Thomas, 
1882, and Geogale Milne Edwards & A. Gran- 
didier, 1872. Although there is no explicit phylog- 
eny for the group, it is assumed to be monophy- 
letic and represents one of the more spectacular 
adaptive radiations of mammals in the world. 

The largest genus of Malagasy Tenrecidae is 
Microgale (shrew tenrecs), which has been re- 
vised by MacPhee (1987). He listed 22 validly 
published and available names but drastically re- 
duced the number of recognized species to 10; 
furthermore, he grouped these species into phe- 
netic clusters as follows: 

cowani cluster: M. cowani, M. thomasi, M. par- 

vula 
gracilis cluster: M. gracilis 
longicaudata cluster: M. longicaudata, M. prin- 

cipula 
pusilla cluster: M. pusilla 
brevicaudata cluster: M. brevicaudata 
dobsoni cluster: M. dobsoni, M. talazaci 

More recently, three new species have been de- 
scribed, one of which, Microgale pulla Jenkins, 
1988, is here provisionally synonymized with M. 
parvula G. Grandidier, 1934. The other two spe- 
cies present greater problems: M. dryas Jenkins, 
1992, groups with both the cowani and gracilis 
clusters, whereas M. soricoides Jenkins, 1993, ap- 
pears to warrant a distinct cluster. 

MacPhee's (1987) major revision was based 
primarily on dental morphology providing the 
framework for ontogenetic studies associated with 
tooth eruption patterns and measures of morpho- 
logical variation within and between various spe- 
cies groups. Metric data on the cranium and body, 
and external features, were accorded somewhat 
less consideration. In a few cases, the limits of 
several "species" are poorly defined, and new 
specimen material is necessary to work out these 
taxonomic problems. This bias away from exter- 
nal features has in some cases caused subsequent 
field-workers, often dealing with live animals, to 
continue to use the nomenclature of Eisenberg 
and Gould (1970), employing a combination of 



ratios of external measurements to divide the ge- 
nus into four behavioral classes, and that of Ge- 
nest and Petter (1975), whose key primarily em- 
ployed external features. In particular, several of 
the nominal species synonymized under Micro- 
gale cowani by MacPhee have continued to ap- 
pear as distinct species in the subsequent litera- 
ture, including M. taiva Major 1896b and M. me- 
lanorrhachis Morrison-Scott, 1948 (see Nicoll & 
Rathbun, 1990; Stephenson, 1995). 

In the past few years, several reports (Nicoll & 
Rathbun, 1990; Rax worthy & Nussbaum, 1994; 
Stephenson, 1995) have been made of specimens 
that were identified in the field, on the basis of 
external appearance and variations in general be- 
havior, as "M cowani" , "M. taiva", and "M. me- 
lanorrhachis" , all synonyms of M. cowani sensu 
MacPhee (1987). MacPhee maintained that dental 
evidence supported the view that the holotypes of 
M. melanorrhachis and M. taiva are juveniles of 
M. cowani. Among recent collections made for 
the University of Michigan Museum of Zoology 
by C.J.R. at Montagne d'Ambre, Mantady, and 
Ambatovaky, and for the Field Museum by 
S.M.G. and C.J.R., all three of the field-identified 
taxa include adult specimens. On the basis of this 
new material, some redefinition of M. cowani sen- 
su MacPhee is now feasible. 

First, all of these species appear to be closely 
related and show only slight, if any, dental vari- 
ation. Second, in most cases, diagnosis is depen- 
dent on a combination of a few, mainly external, 
characters. Finally, several of the taxa redefined 
here occur sympatrically, with no evidence of in- 
termediate specimens. Additional collections are 
needed to clarify several points. Also, biochemi- 
cal analysis of tissues already collected may pro- 
vide additional insight into the relationships of 
this complicated genus. 

Herein we report on one of these recent collec- 
tions, made during a small mammal survey un- 
dertaken by two of us (S.M.G. and C.J.R.) on the 
eastern slopes of the Reserve Naturelle Integrate 
(RNI) d'Andringitra during November and De- 
cember 1993. The regions surveyed proved highly 
speciose for shrew tenrecs of the genus Micro- 
gale. Some of these are common and their pres- 
ence is predictable from their widespread occur- 
rence in other areas of Madagascar; a few others 
are believed to be rare; and two are apparently 
undescribed. Many of the shrew tenrecs in the 
sample were identified from external features, but 
a considerable proportion were not immediately 
identifiable and were eventually sent to the Nat- 



192 



FIELDIANA: ZOOLOGY 



ural History Museum (BM(NH)) for comparative 
identification by P.D.J. 

In this paper we review the species limits of 
the Microgale obtained in this survey. In Chapter 
20 the ecology and elevational distribution of Mi- 
crogale and other insectivores found within the 
reserve are discussed. 



Previous Work 

The only previous work conducted on the in- 
sectivores of the RNI d'Andringitra was in late 
1970 and early 1971 during an expedition to the 
area (Paulian et al., 1971; see Chapter I, pp. 4-5 
for details). Some information on this collection, 
made by R. Albignac and A. Peyrieras, was pre- 
sented by Genest and Petter (1975). MacPhee 
(1987) made reference to these species in his sys- 
tematic revision of the genus Microgale, and in a 
few cases he provided new specific identifications. 
The expedition participants concentrated their ef- 
forts in the higher elevation zones, and most if 
not all of their trapping effort was with rodent 
traps (R. Albignac, pers. comm.). The collection 
of Albignac and Peyrieras is in the Museum Na- 
tional d'Histoire Naturelle (MNHN), Paris. A list 
of the known mammal fauna of the RNA d'An- 
dringitra was compiled by Nicoll and Langrand 
(1989), based on a review of the literature and 
their own unpublished records. The only other 
known collections of Microgale made in the re- 
serve were in 1993 during a Cambridge Univer- 
sity student expedition (O'Keefe & Ashmore, 
1994) for eventual return to Pare Botanique et 
Zoologique de Tsimbazaza, Madagascar. 



Materials and Methods 

Trap Lines 

The principal means of capturing insectivores 
was with pitfall buckets and drift fence; a few 
were obtained with standard mammal live traps. 
See Chapters 17 and 20 for more details on the 
trapping techniques. 



Specimens and Measurements 

Captured animals were prepared as standard 
museum skins with associated skulls and skele- 



tons, fluid-preserved carcasses, or full skeletons. 
Many tissue samples were frozen in liquid nitro- 
gen for biochemical studies, and the viscera were 
preserved in alcohol for endoparasite research. 
Whole carcasses preserved in formalin were 
wrapped in fine cheesecloth before immersion to 
prevent mixing of ecotoparasites (see Chapter 12). 
A representative collection of the taxa described 
herein will be returned to the D^partement de 
Biologie Animale, University d' Antananarivo. 

Cranial measurements, in millimeters, were 
taken using dial calipers and a microscope mea- 
suring stage. The dental nomenclature follows that 
of Mills (1966), Swindler (1976), Butler and 
Greenwood (1979), and MacPhee (1987). Dental 
notations are given in parentheses in the text; pre- 
maxillary and maxillary teeth are noted by upper 
case letters and mandibular teeth by lower case 
letters. The following measurements were made 
from specimens in the flesh or prepared crania. 
Abbreviations and definitions for these measure- 
ments (all in millimeters, with the exception of 
weight [WT], in grams) follow. 

BB (breadth of braincase): the greatest distance 
measured across the squamosals. 

BL (braincase length): from the superior artic- 
ular facet to the occipital condyle, parallel to 
the long axis of the skull. 

CIL (condyloincisive length): cranial length 
from first upper incisor to occipital condyle. 

EL (ear length): measured from the notch at the 
base of the ear to the distalmost edge of the 
pinna. 

HB (head and body length): measured from the 
tip of the nose to the distalmost point of the 
body (at base of tail). 

HF (hind foot length): measured from the back 
edge of the heel to the tip of the longest toe 
(not including claw). 

I1-P3 (length of anterior upper teeth): from an- 
terior of first upper incisor to anterior of sec- 
ond upper premolar. 

TL (tail length): measured from the base of the 
tail (at right angles to the body) to the end 
of the distalmost vertebra. Does not include 
terminal hair tufts. 

TOTL (total length of body and tail): measured 
from the tip of the nose to the end of the 
distalmost tail vertebra (does not include any 
terminal tail hair tufts). Animal is positioned 
on its back straight with vertebrae parallel to 
rule, but not stretched out. 

UTL (upper toothrow length): from anterior of 



JENKINS ET AL.: SHREW TENRECS 



193 



first upper incisor to posterior of third upper 
molar, parallel to the long axis of the skull. 
WT (weight): measured with Pesola spring 
scales. Animals weighing less than 10 g were 
weighed within 0.2 g; those weighing between 
10 and 100 g were weighed within 0.5 g. 

Reproductive condition was recorded for males 
as length X width of the testes and degree of con- 
volution of the epididymis. Females were noted 
as nonperforate or perforate, nonparous or parous, 
and the numbers and locations of any embryos 
and placental scars were recorded. The mammary 
formula is presented as the number of paired ax- 
ial, abdominal, or inguinal teats. 

The following age classes are recognized: 

Infant: individuals in which the deciduous an- 
temolar dentition and the molars are not fully 
erupted; premaxillary, parietal, and basioc- 
cipital sutures are unfused. 

Juvenile: individuals in which the molars are 
fully erupted and the deciduous antemolar 
dentition is erupted and in the process of re- 
placement by the permanent teeth; cranial su- 
tures are in the process of fusing. The erup- 
tion sequence of the permanent teeth has 
been subdivided into four stages by MacPhee 
(1987); these stages have been accepted in 
this text, unless otherwise stated. 

Adult: individuals with a fully erupted perma- 
nent dentition; cranial sutures generally 
fused, although their position is more or less 
clearly marked. 



Other Abbreviations 

BM(NH): The Natural History Museum, Lon- 
don (formerly British Museum [Natural His- 
tory]). 

FMNH: The Field Museum, Chicago. 

MCZ: Museum of Comparative Zoology, Har- 
vard. 

MNHN: Museum National d'Histoire Naturelle, 
Paris. 

USNM: National Museum of Natural History, 
Washington, D.C. (formerly United States 
National Museum). 

C/c: canine. 

d: deciduous. 

I/i: incisor. 

M/m: molar. 

P/p: premolar. 



Results 

Ten species of Microgale were collected in the 
RNI d'Andringitra; they may be distinguished by 
the characters given in the key (Appendix 19-1). 
Unless otherwise indicated the data given below 
are confined to specimens collected at RNI 
d'Andringitra. 



Systematic Section 
Microgale cowani Thomas, 1882 

Holotype— BM(NH) 82.3.1.25: adult female 
body in alcohol, skull extracted. Collected mid- 
March to mid-February 1 880 by the Reverend W. 
Deans Cowan. 

Type Locality — Ankafana Forest, eastern Bet- 
si leo (Ankafana = Ankafina, Fianarantsoa, Fian- 
arantsoa Province, 21°12'S 47°12'E; see Mac- 
Phee, 1987; Carleton & Schmidt, 1990). 

Referred Material— FMNH 151758, 151759: 
43 km S Ambalavao, junction of Sahanivoraky 
and Sahavatoy rivers, RNI d'Andringitra, 810 m, 
22°13'S 47°00'E; FMNH 151653, 151654, 
151655, 151767, 151770, 151772, 151776, 
151777, 151779, 151782, 151783, 151784, 
151786, 151788, 151789, and 151796: 40 km S 
Ambalavao, along Volotsangana River, RNI 
d'Andringitra, 1210 m, 22°13'S 46°58'E; FMNH 
151652, 151656, 151798, 151810, and 151811: 
38 km S Ambalavao, along Volotsangana River, 
RNI d'Andringitra, 1625 m, 22°11'S 46°58'E. 
Specimens D, E, and F collected by the Cam- 
bridge University Expedition from RNI d'Andrin- 
gitra, 1400-1800 m. 

Description — The following description is 
based on the holotype and the RNI d'Andringitra 
specimens. Medium-sized, tail moderately short, 
shorter or subequal to HB. General appearance of 
the pelage is dark brown, especially on the rump 
where lighter speckling is reduced; color dorsally 
speckled brown, hairs with dark gray bases and a 
mixture of buff and red-brown tips; ventrally gray 
with a buff wash, hair bases gray with buff tips; 
tail bicolored, dark brown dorsally, sharply de- 
marcated from the paler venter, which is reddish 
buff in most specimens, especially proximally; tail 
well-clothed with long scale hairs that partially 
obscure the scales, those on the mid-dorsal basal 
third of the tail overlying 2.5-3 scales; hind feet 
brown above, dark gray below. Skull (Fig. 19-1) 



194 



FIELDIANA: ZOOLOGY 




Fig. 19-1. Crania from left to right of Microgale cowani FMNH 151783 and M. taiva FMNH 151643. Above, 
lateral view; below, dorsal view left and ventral view right. 



medium in size, rostrum elongated, nasals extend 
posteriorly into the interorbital region; frontals 
large, relative to the parietals, which are reduced 
in length in the mid-dorsal region; braincase rel- 
atively broad and short, junction of supraoccipital 
and parietal subrectangular, high on dorsal surface 
of the braincase, occipital large; pronounced dia- 
stemata separate teeth of the upper anterior den- 
tition from the first upper incisor (II) to the sec- 
ond upper premolar (P3), and diastemata are also 
present on either side of the lower canine and p2. 
All elements of the talonid of the third lower mo- 
lar (m3) are present, including the hypoconid, en- 
toconid ridge, talonid basin, and entoconid. For 
distinctive features of the morphology of the per- 
manent and deciduous dentition, see Figures 19-2 
and 19-3. 

Measurements — External and cranial measure- 
ments are presented in Tables 19-1 and 19-2. 

Variation — There is no obvious indication of 
sexual dimorphism in size of adults, although the 



sample is too small for this hypothesis to be test- 
ed. The ratio of females to males in the sample is 
1:1; that of juveniles to adults is 1.8:1. All of the 
juveniles are classed as Stage 1 (MacPhee, 1987, 
p. 13) or younger; the anterior dentition is com- 
pletely deciduous in most specimens but the mo- 
lars are usually fully erupted, although the third 
upper molar (M3) is incompletely erupted in two 
specimens (FMNH 151796 and 151798), whereas 
the third upper incisor (13) is the only permanent 
antemolar tooth erupting or erupted in the (Stage 
1) individuals. 

Juveniles were compared with adults in four 
linear dimensions (HB, CIL, UTL, and BB) and 
WT. The mean of each measurement is slightly 
smaller in juveniles, with the exception of HB, in 
which the mean size of juveniles is greater than 
that of adults (see Table 19-1). These data on 
body size confirm the observation of Leche 
(1907) that juveniles may attain adult body size 
before replacing any deciduous teeth, and that of 



JENKINS ET AL.: SHREW TENRECS 



195 





Fig. 19-2. Above, buccal view of permanent left anterior dentition of M. cowani FMNH 15 1783; below, M. taiva 
FMNH 151635 (upper toothrow) and FMNH 151638 (lower toothrow). Scale 1 mm. 



MacPhee (1987) that subadults may exceed mean 
adult size in some parameters, including head and 
body length. 

Reproduction — An adult male (FMNH 151652) 
had abdominal testes measuring 7X5 mm with 
a convoluted epididymis. Four adult females for 



196 



which details on reproductive condition were 
available had enlarged mammae, perforated va- 
ginas, and in one case two placental scars on the 
left oviduct and three on the right. Mammary for- 
mula: 2-0-4 (N = 5). There is little information 
on the reproductive age of the juvenile specimens; 

FIELDIANA: ZOOLOGY 



one female (FMNH 151655) with completely de- 
ciduous antemolar dentition was nulliparous, 
whereas a second female (FMNH 151653) with a 
deciduous antemolar dentition apart from 13 was 
perforate but lacked embryos. This limited infor- 
mation suggests that females in this species may 
become reproductively mature while still dentally 
immature. 

Remarks — Formerly considered to be the most 
widespread and commonly occurring species 
throughout northern, eastern, and southeastern 
Madagascar, this species is here restricted to the 
holotype from Ankafina and the specimens listed 
above from the RNI d'Andringitra. It may be 
much rarer in collections than previous accounts 
have indicated. Microgale cowani occurs sympat- 
rically with M. melanorrhachis and M. taiva at 
RNI d'Andringitra; those taxa were considered to 
be synonyms of M. cowani by MacPhee (1987). 
Characters used to discriminate this species from 
others occurring in the RNI d'Andringitra are giv- 
en in the key (Appendix 19-1) and Table 19-1. 
Those that separate M. cowani and M. taiva are 
given in Table 19-2. 



Microgale taiva Major, 1896b 

Microgale cowani Thomas: MacPhee, 
1987, in part 

Holotype— BM(NH) 97.9.1.112: juvenile fe- 
male, skin and skull; BM(NH) 1975.2233: partial 
skeleton. Collected January 19, 1895 by C. I. For- 
syth Major. 

Type Locality — Ambohimitombo forest, Tana- 
la Country (Ambohimitombo town, Fianarantsoa, 
Fianarantsoa Province, 20°43'S 47°26'E; see 
MacPhee, 1987). 

Referred Material— FMNH 151633, 151634, 
151642, 151643, and 151645: 45 km S Ambala- 
vao, E bank of Iantara River, along Ambalama- 
nenjana — Ambatamboay Trail, edge of RNI 
d'Andringitra, 720 m, 22°13'S 47°01'E; FMNH 
151635, 151636, 151637, 151755, 151757, 
151760: 43 km S Ambalavao, junction of Sahan- 
ivoraky and Sahavatoy rivers, RNI d'Andringitra, 
810 m, 22°13'S 47°00'E; FMNH 151638, 151639, 
151761, 151762, 151763, 151765, 151769, 
151771, 151774, 151775, 151780, and 151781: 
40 km S Ambalavao, along Volotsangana River, 
RNI d'Andringitra, 1210 m, 22°13'S 46°58'E; 
FMNH 151640, 151641, 151657, 151658, 
151724, 151725, 151790, 151791, 151792, 



151797, 151802, 151809, and 151813: 38 km S 
Ambalavao, along Volotsangana River, RNI 
d'Andringitra, 1625 m, 22°11'S 46°58'E. Cam- 
bridge University Expedition specimens B and C, 
RNI d'Andringitra, 1400-1800 m. 

Description — Based on the holotype and spec- 
imens collected in the RNI d'Andringitra. Medi- 
um-sized, tail moderately long, subequal or longer 
than head and body length (see Tables 19-1 and 
19-2). Dorsal pelage dark brown with buffy 
brown speckling, ventral coloration gray-brown 
with buffy brown wash; tail not obviously bicol- 
ored, dark gray-brown above, slightly paler gray 
below; tail scale hairs short, overlying 1.5-2 
scales on the dorsal proximal third of the tail, so 
that the scales are visible. Skull medium in size 
(see Fig. 19-1), with a moderately elongated ros- 
trum, nasals extend to level of zygomatic plate, 
barely into interorbital region; frontals posteriorly 
inflated so that the dorsal surface of the skull ap- 
pears slightly concave in profile; braincase broad 
and long, parietals moderately large, supraoccip- 
ital-parietal suture suboval. Short diastemata sep- 
arate the anterior upper dentition from II to P3 
and are present between the lower canine and p2. 
Talonid of m3 slightly reduced, entoconid lacking. 
Distinctive features of the morphology of the per- 
manent and deciduous dentition are shown in Fig- 
ures 19-2 and 19-3. 

Measurements — External and cranial measure- 
ments are presented in Table 19-1 and 19-2. 

Variation — Although the mean of HB, WT, 
and three cranial dimensions in adult female M. 
taiva is consistently higher than that of males, this 
difference was found to be statistically insignifi- 
cant, and it is concluded that sexual dimorphism 
is not present in the small sample available. The 
sex ratio of females to males in the adult sample 
is 1:2.3, and the ratio of juveniles to adults is 1: 
1.2. Two specimens (FMNH 151641 and 151813) 
are dentally immature — the antemolar teeth are 
completely deciduous and the third upper and 
lower molars (M3 and m3) are in the process of 
eruption; five specimens show deciduous ante- 
molar dentition but fully erupted molars; 13 and/or 
the third lower incisor (i3) are erupting in three 
other specimens. In one specimen (FMNH 
151636) only i3 has erupted and 13 and the first 
lower premolar (p2) are beginning to erupt; it is 
classed as Stage 1-2 (following MacPhee, 1987, 
p. 13), yet it has fairly large testes (4 X 2) and a 
convoluted epididymis. Another specimen 
(FMNH 151642) is difficult to fit into a particular 
category, because II, 13, i3, p2, and p4 are all 



JENKINS ET AL.: SHREW TENRECS 



197 






Fig. 19-3. Top, buccal view of deciduous left anterior upper dentition of M. cowani FMNH 151798; middle, M. 
taiva FMNH 151725; below, M. melanorrhachis FMNH 151626. Scale 1 mm. Figure continues on page 199. 



starting to erupt, but in contrast it has small testes 
(2 X 1 ), and the epididymis is not convoluted. In 
HB, CIL, UTL, and BB, the mean of these juve- 
niles is slightly less than that of the adults, al- 
though the mean WT of juveniles is considerably 
less (9.0 ± 1.8 g, N = 15) than that of adults 



(12.4 ± 1.2 g, N = 18; see Table 19-1). The den- 
tally most mature juvenile and many of the other 
juveniles are well within the adult range in all of 
these dimensions, supporting the view that im- 
mature Microgale achieve large size while decid- 
uous teeth are still present and that the deciduous 



198 



FIELDIANA: ZOOLOGY 









Fig. 19-3 (cont.). Top: above, lingual view of deciduous left anterior lower dentition of M. cowani FMNH 
151776; middle, M. taiva FMNH 151725; below, M. melanorrhachis FMNH 151626. Scale 1 mm. Bottom: Lingual 
view of right m3 from left to right of M. cowani FMNH 151776, M. taiva FMNH 151725, and M. melanorrhachis 
FMNH 151626. Scale 1 mm. 



dentition may be maintained into adulthood (Le- 
che, 1907; MacPhee, 1987; Jenkins 1992). 

Reproduction — Two of the adult females col- 
lected were carrying embryos, both with single 
embryos in the left and right oviducts, measuring 
6 and 7 mm crown-rump length. Mammary for- 



mula: 0-2-4 (N = 1), 0-2-2 (N = 1), 2-0-4 (N = 
2). The testes of three adult males in the sample 
(FMNH 151633, 151643, and 151547) measure 6 
X 3, 7 X 3, and 8X4 mm, respectively, all with 
convoluted epididymides. Information is lacking 
for females of this species, but it seems likely 



JENKINS ET AL.: SHREW TENRECS 



199 



K^ 



— 3 



< 



a 



3C 



rj 



CO 

+ 1 I +1 



q 
I +1 



oc 



; CO 



O 



J 5 5 i?id J *i P 1 ^ 

ci >/"} ^^ ^ ^ -~ ^ ro oo oo 

1 © J ^ - ^ <n ^ - 2 - 2 <* ~ 2 2 <* o 2 in n 
+l © 2 +l o +l cs +l .A +l si < +l -A « => ^ 



iCN^iu-l^rt ^ — H 00 ^ O - 

I VD ■* m — i-c C4 — ON 



00 — 



cm rn 



^c 



^V^^^VO^ ON — — — < Tfr 

+i i +i t +i 7 +i 7 !£ +i 7 +i 7 +i 7 +i 7 to +i 7 " 

oo c ^ n -* c~> in m • — 

t"» 00 ir> — O — On — © — On — -<fr — ;j>/"> — 

oo r-' ■«* ro -"t ""t «n ^ ^* 



on oo 



SO — 
— (N 



„ S ^ Jj Tt 



d ™ d © — NOd<n""©oodoo©ON — ' On i/->" d >n >n m 



— _ — <N 



+i i +i i +i i +i i 2 +i 

ON ON -* «n ~" 

O 00 00 — O — i — 



I +1 1 +1 1 +1 I .+11 
n no no >n oo ■<*■ 

- -3- — on — ■ — — — 00 — 



/-v ^ Tfr NO ^-v-^-00lO --v 

NO Tt ' — ' ' — ' 00 — — — • t — '"3" 

-. w 9^9» p iNft1 iS, » w « w '* w *P w ir> 

vo <o — — r~~ r~~ on o\ ^ no no no ^ 

+1 I +1 I +1 I +1 I in +1 I +1 l +1 I +1 I ^ +1 [ « "* rf< 3 ° £ 

©inoqnoq — oq — ©NONqNOCNoooqNopf>ir> 

O •/"> © ~™ no "~ vi in oo in d 

NO n <N — NONO0000 NO 



O 



m 



3 



-on Con 



on _. — _ — 



— 00 

--I ~~: --; wnoio 

ONmCNrorrr-fN'OTtOOr^ONr'-iON^'O'O 



ON S 



^N 



oc 



in 



I +1 T +l 
tj- in 

coinininciNOooin 
on no r^ — 
>n >n no no 



^ . On 

I +1 I — +1 1 +1 I +1 I +1 I 

On oo NO <n — t> v « 

r^r^inr^ONr^ONNO^moN 

O — — Tfr" T^ 

oo oo oo r- On 



»n n© 

^ S 5 " 8 «• . 



QN 



" O 



nC 



© oo <o ON ""> 
r^ <N co — oo 

+i 7 +i 7 



m oo ,-^ 

— < CO Tf 

— _„, _ _ >-•,«>»• o in in 

S^MNOMvo^^^r^^r- .o^, oo oo — 

r^oo — TtinTtjOTf-'r-NONO^-^ingo^:--'^^^ 

+i "7 +i 7 ss +i 7 +i 7 +i 7 +i 7 ~ +i 7 — ~ s - 



o <n cm r^^ in o o * ^* en 

r-. — in — ooooono-" — -^-(NTffNNONDTt^ioin 
od" - Tf-""r-~-'" - ON" - r^"^iri"~ , ON" -, ON'^™_Tf : ' — ' 



oc 



rn 



— (N(N 



— — 00 



"3 <U "3 



*- C w w 



<U "3 *3 





c 




c 


C - 


e c — 


= _- _ 












^li 




3 3 




3 




> 3 


> 3 3 


T3 T3 

CO ca 


s 




_3 


.2,-S 


.2,.S"S 





c^ 



b «* 



I! 

1 



§§ 



.8 § 



^' ^ 



200 



FIELDIANA: ZOOLOGY 



I 



^ — 00 



nSmSL^- *-" •* o> ^ oo w — '—'— w r»» 

— ^^o— <^> — — 



n «r — 

©2d 



(N CI - O 



+ l 4 +, ^ +l i +l oc +l ^ +l oo +l oo +l ^ +l 4 



NO^NO^oo^r-^ON^ON^ON^ON'^ON 

nOnOOOOOOnOOnOnOn 



»2 2-:d22 = «; 



•-v ^ ^ ^ pn — oo — o\ po Tt/-^'* ^^ 
w JinJ 3Lo3L""> v ]!ro^irr~ w oo w p>") w (N w ''i- w r ,| oo w <n<n 

r- r~ o\ oo «-< »-> — — — — __rn— X^ <f> 

+l ^i +l ^. +l J> +l ' +l 4, +l J. +l ' +| J, +| J, +| J. +i i - +i J. ® — 

(N (S NO — O T} - — PO <N O ifl . O ^ • > 



fi t~: OP on -* 



; r- "! oo 

oo 



X 



X 



k NO -J 00 



. O ^ O 



s0 



^0P^rOTt^-(Np 



O — O — © — O — O — O — PO — PO rfr — 



(N >n 



— < oo 



— ON 



ON 2 w ^ T *' 

,,*d6*od»dodd--dwd(Ndm6w 

+ — « "T — — (NCNfN(NC4 

~ I +1 ' +i i +ii+ii +1 _L +' I +1 I +1 J. 

po — o _ <d po _ r~ ^ 



— ■ — en 






NO NO H NO Q 00 

— — .NO — ON— '00— « — <N<N<NCMCS<NCSC4<N 

- - - N (N| 



C2,o\ w cj,<n~ G 

^p^m^r-^^^ 

^ONO^d^CC 00 . 

<N cm >o CM _ — 

™ +1 1^+1 I ""! *°. f*> 

« NO On 00 tJ- 

-NON^od(N O 
fN ri <N <S PO 



> 3 > 3 3 
3 "O g T3 "O 



,3 
a 



I 
I 

Q 

C 

S 



5 5 



5" 5 



3'3 



£ 



JENKINS ET AL.: SHREW TENRECS 



201 



Table 19-2. Dimensions (in mm) distinguishing hii- 
crogale cowani from hi. taiva in the RNI d'Andringitra. 



Parameter 


hi. cowani 


hi. taiva 


Tail length 


60.6-71.0 


71.0-89.0 




65.6 ± 3.25 


88.4 ± 4.60 


Ratio of tail length to 

condyloincisive 

length 
Length of anterior denti 

tion I-P3 


(10) 

2.7-3.1 

3.0 ± 0.13 

(10) 

5.5-6.2 

5.8 ± 0.2 


(14) 

3.5-4.2 

3.9 ± 0.19 

(14) 

5.2-5.8 

5.5 ± 0.21 


Ratio of anterior denti- 


(ID 
52.8-56.4 


(14) 
49.1-52.8 


tion (I-P3) to upper 
toothrow length 
Braincase length 


53.9 ± 1.08 

(11) 

7.3-8.2 

7.8 ± 0.24 


50.7 ± 0.88 

(14) 

7.8-8.4 

8.1 ± 0.17 




(ID 


(14) 



from the well-developed testes in the dentally im- 
mature (Stage 1-2) male (FMNH 151636) that 
males may become sexually mature before they 
acquire their complete permanent dentition. 

Remarks — The holotype of this species is a ju- 
venile. It was considered a synonym of Microgale 
cowani by MacPhee (1987) on the basis of its 
dentition. Recently collected adults confirm that 
although the permanent dentition is indeed simi- 
lar, it does differ slightly from that of M. cowani 
in that the talonid of m3 is slightly reduced and 
the entoconid is lacking. The deciduous antemolar 
dentition also differs from that of M. cowani (and 
M. melanorrhachis). The two taxa do occur sym- 
patrically in the RNI d'Andringitra, where they 
are readily distinguished from each other by dif- 
ferences in their external appearance, including 
those of absolute and relative tail length, tail col- 
oration and pilosity, by a combination of cranial 
characters, and by minor differences in the den- 
tition (see Tables 19-1 and 19-2, and Figs. 19-1, 
19-2, and 19-3). 



Microgale melanorrhachis Morrison- 
Scott, 1948 

Microgale cowani Thomas: MacPhee, 
1987, in part 

Holotype— BM(NH) 48.88: juvenile female, 
skin and skull, collected November 22, 1939 by 
C. S. Webb. 

Type Locality — Perinet (= Andasibe), near 
Moramanga, eastern Madagascar, 19°00'S 



48°30'E, altitude 3000 feet (= approximately 900 
m). 

Referred Material — FMNH 151626: 45 km 
S Ambalavao, E bank of Iantara River, along Am- 
balamanenjana — Ambatamboay Trail, edge of 
RNI d'Andringitra, 720 m, 22°13'S 47°01'E; 
FMNH 151627 and 151756: 43 km S Ambalavao, 
junction of Sahanivoraky and Sahavatoy rivers, 
RNI d'Andringitra, 810 m, 22°13'S 47°00'E. 

Description — This description is based solely 
on juvenile specimens from RNI d'Andringitra. 
Similar in size to M. cowani and M. taiva juve- 
niles of equivalent dental age (HB 68-77, mean 
73.7 ± 4.03, N = 3); tail length (67-73, mean 
69.0 ± 2.83, N = 3) similar to that of M. cowani 
but much shorter than that of M. taiva (see Table 
19-1). Dorsal pelage gray-brown with some yel- 
lowish speckling and with a distinct dark brown 
mid-dorsal stripe extending from the head at the 
level of the ears to the base of the tail. Dorsal 
hairs with light gray bases, buffy red terminally 
with short dark brown tips; venter gray with a buff 
wash, hairs with light gray bases and long white 
tips. Tail bicolored, dark gray dorsally, buff ven- 
trally; sparse covering of scale hairs, each over- 
lapping 1.5-2 scales in the dorsal basal third of 
the tail. Hind feet buff above, dark gray-brown 
laterally and below. Skull similar in size to those 
of juvenile M. cowani and M. taiva (CIL 21.2, 
22.5 mm); rostrum moderately elongated and with 
short diastemata between the anterior upper teeth, 
as in M. taiva but unlike the longer diastemata of 
M. cowani; braincase narrower and shallower than 
that of M. cowani and M. taiva, shorter than that 
of M. taiva. The deciduous antemolar dentition 
differs from those of M. cowani and M. taiva (see 
Fig. 19-3a,b). The talonid of m3 is slightly re- 
duced, the talonid basin is narrow, and the ento- 
conid is lacking (see Fig. 19-3c). 

Measurements — External and cranial measure- 
ments are presented in Table 19-1. 

Variation — The pelage of the RNI d'Andrin- 
gitra specimens differs in coloration from that of 
the holotype, which is more rufous brown dorsal- 
ly, and the dorsal surface of the hind feet and 
ventral surface of the tail are more reddish buff 
in color than that of the holotype. 

Reproduction — All of the specimens obtained 
were juveniles, with deciduous antemolar denti- 
tions and with fully erupted molars. In one of the 
two males for which reproductive condition was 
recorded (FMNH 151626), the immature testes 
measured 2X2 mm and the epididymis was not 
convoluted. The single female (FMNH 151627) 



202 



FIELDIANA: ZOOLOGY 



in which i3 was beginning to erupt had a possible 
perforated vagina but showed no other signs of 
sexual maturity. 

Remarks — MacPhee (1987) considered M. me~ 
lanorrhachis to be a synonym of M. cowani on 
the basis of its dentition, arguing that the mid- 
dorsal stripe used to characterize the taxon is vari- 
ably present or absent in M. cowani. Although 
Grandidier (1934) suggested that striping is a ju- 
venile characteristic, MacPhee (1987) pointed out 
that loss or diminution of the stripe, perhaps fol- 
lowing molting, was not in phase with replace- 
ment of the deciduous dentition and also occurred 
in dental adults. The contention that M. mela- 
norrhachis is a synonym of M. cowani has never 
gained wide acceptance. Nicoll and Rathbun 
(1990) states that adults of M. cowani and M. me- 
lanorrhachis occur with overlapping ranges at 
Andasibe, whereas Rax worthy and Nussbaum 
(1994) and Stephenson (1995) also treat M. me- 
lanorrhachis as a distinct species. All of the spec- 
imens from the RNI d'Andringitra are juveniles 
and are clearly distinguishable from juveniles of 
M. cowani and M. taiva by craniodental features 
in addition to the distinctive pelage. 



Microgale longicaudata Thomas, 1882 

Microgale majori Thomas, 1918 

Holotype— BM(NH) 82.3.1.15: adult female, 
body in alcohol, skull extracted, collected mid- 
March to mid-February 1 880 by the Reverend W. 
Deans Cowan. 

Type Locality — Ankafana Forest, eastern Bet- 
sileo (Ankafana = Ankafina, Fianarantsoa, Fian- 
arantsoa Province, 21°12'S 47°12'E; see Mac- 
Phee, 1987; Carleton & Schmidt, 1990). 

Referred Material— FMNH 151628, 151630, 
and 1 5 1 63 1 : 45 km S Ambalavao, E bank of Ian- 
tara River, along Ambalamanenjana-Ambatam- 
boay Trail, edge of RNI d'Andringitra, 720 m, 
22°13'S 47°01'E; FMNH 151632, 151795, 
151799, 151800, 151803, and 151804: 38 km S 
Ambalavao, along Volotsangana River, RNI 
d'Andringitra, 1625 m, 22° ITS 46°58'E; Cam- 
bridge University Expedition specimen A: RNI 
d'Andringitra, 1400-1800 m. 

Description — Body small in size, with a very 
long tail, considerably longer than that of the head 
and body. Dorsal pelage dark brown with a red- 
dish brown wash, ventrally dark gray with a red- 
dish buff wash; tail gray-brown above, sharply 



distinguished from the reddish buff ventral col- 
oration; hind feet brown but reddish buff laterally. 
Skull small, dorsal profile slightly concave (Fig. 
19-4), rostrum moderately short, nasals extend 
posteriorly into the interorbital region; braincase 
long, parietals and occipital large, suboval su- 
praoccipital suture on dorsal surface of the brain- 
case. Diastemata are present between II and 12 
and on both sides of C and P2; well-developed 
anterior and posterior accessory cusps are present 
on 12, C, and P2. Lower p2 caniniform. Talonid 
of m3 with well-developed hypoconid and hypo- 
conulid, arrow talonid basin, and reduced ento- 
conid ridge and entoconid. 

Measurements — External and cranial measure- 
ments are presented in Table 19-1. 

Variation — In the sex ratio of the small sam- 
ple available, males outnumber females (2.5:1); 
the juvenile to adult ratio is 1.5:1. The juveniles 
collected in the RNI d'Andringitra exhibit a com- 
pletely deciduous antemolar dentition and in most 
M3 is incompletely erupted. There are only slight 
differences between the morphology of the decid- 
uous incisors and first premolar (dp2) and that of 
the permanent teeth. 

Reproduction — Only two of the specimens 
showed signs of reproductive activity — a male 
(FMNH 151799) with testes measuring 5x3 mm 
and a convoluted epididymis; and a lactating fe- 
male (FMNH 151630), collected on November 
20, 1993, with single placental scars on the left 
and right oviducts. Mammary formula: 0-2-2 (N 
= 1), 2-0-4 (N = 1). 

Remarks — Microgale longicaudata has been 
reported in eastern Madagascar from Antsiranana 
in the extreme north (12°16'S 49°18'E) to Anka- 
fina (21°12'S 47°13'E) (MacPhee, 1987; Jenkins, 
1993). The presence of this species in the RNI 
d'Andringitra marks a southern extension of the 
recorded range. Thomas (1882) described a long- 
tailed species (M. longicaudata) from a series of 
specimens from Ankafina, smaller specimens of 
which he later transferred to a new species, M. 
majori Thomas, 1918. MacPhee (1987) consid- 
ered the two taxa synonymous. The specimens 
from the RNI d'Andringitra fall in the lower part 
of the size range given by MacPhee (1987) for M. 
longicaudata. This species shows several morpho- 
logical features that suggest it is adapted to an 
arboreal lifestyle (see Thomas, 1918). These fea- 
tures include the extremely long tail, the bare re- 
gion near the tip with transverse scales, the elon- 
gated fifth digit on the hind foot, and the elon- 
gated cheiridia. The elongated tail distinguishes 



JENKINS ET AL.: SHREW TENRECS 



203 




Fig. 19-4. Crania from left to right of M. longicaudata FMNH 151799 and M. parvula FMNH 151623. Above, 
lateral view; below, dorsal view left and ventral view right. 



this species from all others occurring in the RNI 
d'Andringitra. Other distinguishing features are 
given in the key (Appendix 19-1) and Table 19-1. 



Microgale parvula Grandidier, 1934 

IMicrogale pulla Jenkins, 1988 

Holotype — MCZ 45465: juvenile male, body 
in alcohol, skull extracted. Collected by M. Drou- 
hard. 

Type Locality — Environs of Diego Suarez 
(Antsiranana, ca. 12°16'S 49°18'E— see MacPhee, 
1987). 

Referred Material — FMNH 151621: 45 km 
S Ambalavao, E bank of Iantara River, along Am- 
balamanenjana-Ambatamboay Trail, edge of RNI 
d'Andringitra, 720 m, 22°13'S 47°01'E; FMNH 
151622: 43 km S Ambalavao, junction of Sahan- 



ivoraky and Sahavatoy rivers, RNI d'Andringitra, 
810 m, 22°13'S 47°00'E; FMNH 151722, 151764, 
and 151766: 40 km S Ambalavao, along Volot- 
sangana River, RNI d'Andringitra, 1210 m, 
22°13'S 46°58'E; FMNH 151623, 151723, 
151793, 151794, 151801, 151805, and 151806: 
38 km S Ambalavao, along Volotsangana River, 
RNI d'Andringitra, 1625 m, 22°11'S 46°58'E. 

Description — Very small, tail subequal in 
length to that of head and body. Dorsal pelage 
dark gray-brown, ventral pelage dark gray, tail 
uniform dark gray. Skull very small, delicate, and 
elongated in appearance (see Fig. 19-4), rostrum 
slender, moderately short; braincase shallow and 
long, frontals and occipital large relative to pari- 
etals, occipital condyles posterodorsally orientat- 
ed. Diastemata present between II and 12 and on 
either side of C and P2; anterior and posterior 
accessory cusps present on 12, 13, and P2. Diaste- 
ma between c and p2. Talonid of m3 with well- 



204 



FIELDIANA: ZOOLOGY 






Fig. 19-5. Left anterior dentition of M. parvula. 
Above, buccal view of permanent dentition of FMNH 
151623; middle, buccal view of deciduous upper den- 
tition of FMNH 151806; below, lingual view of decid- 
uous lower dentition of FMNH 151622. Scale 1 mm. 



developed hypoconulid but reduced hypoconid, 
entoconid, and entoconid ridge, and narrow, shal- 
low talonid basin. See Figure 19-5 for illustrations 
of the permanent and deciduous dentition. 

Measurements — External and cranial measure- 
ments are presented in Table 19-1. 

Variation — The sex ratio of males to females 
in the small sample is 1.3:1, and that of juveniles 



to adults is 1.5:1. The antemolar dentition is de- 
ciduous, and all molars are fully erupted in all 
four juveniles. 

Reproduction — Few of the specimens collect- 
ed showed active reproductive organs. At least 
two of the males that were adult based on denti- 
tion had small testes without convoluted epidid- 
ymides; the testes of FMNH 1 5 1 623 measured 2 
X 1 mm, and those of FMNH 151723 measured 
3X2 mm. One adult female (FMNH 151621) 
obtained on November 15, 1993, had two embry- 
os in each oviduct measuring 4 mm crown-rump 
length. Mammary formula: 0-0-4 (N = 1), 0-2-4 
(N = 1). Information is lacking on the reproduc- 
tive condition of juvenile males with deciduous 
dentition, but a female (FMNH 151766) with de- 
ciduous antemolar dentition was perforate. 

Remarks — The specimens collected in the RNI 
d'Andringitra include both adults and juveniles 
that are clearly attributable to the same species. 
Microgale parvula is known only from the orig- 
inal description of the single juvenile specimen, 
the dentition of which was illustrated by MacPhee 
(1987), who also corrected the measurements giv- 
en in the original description. The dentition of the 
juveniles from RNI d'Andringitra agrees well 
with the illustrations of the dentition of M. par- 
vula in MacPhee (1987), and the corrected upper 
toothrow length given by MacPhee is similar. 
There are, however, some differences in size; the 
corrected skull length of the holotype (15.5 mm) 
is less than that in the three juvenile specimens 
from RNI d'Andringitra with deciduous antemolar 
dentitions (i.e., of equivalent dental age), in which 
CIL is 16.2-16.7 mm. The tail of the holotype is 
longer than its head and body length (TL:HB 
1.21), whereas the tail is subequal to or slightly 
larger than head and body length in these speci- 
mens (TL:HB 0.96-1.07). Comparisons have also 
been made between the adult specimens from RNI 
d'Andringitra and the adult holotype specimen of 
M. pulla Jenkins, 1988. They agree in most fea- 
tures of size, external characters, and craniodental 
morphology, except for the TL to HB ratio, which 
is greater than that in the holotype of M. pulla 
(0.83). When M. pulla was described (Jenkins, 
1988), the possibility was mentioned that it might 
simply represent the adult of M. parvula, and the 
conclusion from the study of these specimens 
from RNI d'Andringitra is that this is probably 
correct, although direct comparison of specimens 
is required for confirmation. 



JENKINS ET AL.: SHREW TENRECS 



205 



Microgale dobsoni Thomas, 1884 

Nesogale dobsoni Thomas, 1918 

Holoytpe— BM(NH) 84.10.20.1: immature 
male, in alcohol, skull extracted. Collected Feb- 
ruary or March 1884 by W. Waters. 

Type Locality — Nandesen forest, Central Bet- 
sileo (Nandihizana, 10 miles S Ambusitra — 
manuscript note in Thomas's private copy of orig- 
inal description. Nandihizana, ca. 20 miles [30 
km] SSW Ambositra, see MacPhee, 1987. Esti- 
mated as 20°50'S 47°10'E). 

Referred Material— FMNH 151624, 151625, 
and 151785: 40 km S Ambalavao, along Volot- 
sangana River, RNI d'Andringitra, 1210 m, 
22°13'S 46°58'E. 

Description — Head and body large, tail sub- 
equal to or longer than head and body. Pelage 
long, dorsally gray-brown with a buff or reddish 
buff wash, venter gray with a buff or reddish buff 
wash. Tail gray above, buff below; hind feet buff. 
Skull and mandible very large and robust, sutures 
fused and obscure; rostrum broad, interorbital re- 
gion long, slightly concave; braincase angular, su- 
perior articular facets very prominent, occipital 
region reduced, supraoccipital crests well devel- 
oped. Diastemata between II and 12 and between 
13 and C. Lower i2 considerably larger than ca- 
nine. The teeth of both individuals are too worn 
for detailed observations. 

Measurements — External and cranial measure- 
ments are presented in Table 19-1. 

Variation — The two adults collected in 1993 
differ in external appearance; one specimen, with 
TL 93% of HB, is more rufous in coloration than 
the other, in which the tail is relatively much lon- 
ger (TL 1 12% of HB). A specimen in the MNHN 
collected from Anjavidilava is slightly lighter in 
coloration than either of the FMNH specimens, 
with TL 96% of HB. All three specimens are 
within the range of color variation exhibited by 
other specimens of this taxon in the BM(NH) col- 
lection. The range of TL:HB is 82-111% (mean 
94.1 ± 7.67%, N = 11) for M. dobsoni in the 
BM(NH) collection, so even the long-tailed spec- 
imen (FMNH 151624) is only just outside the 
known range on this feature. There is no indica- 
tion in the RNI d'Andringitra specimens of the 
incrassation of the tail found commonly in this 
species. 

Reproduction — The two female specimens 
taken in 1993 were adults, on the basis of denti- 
tion, and both were lactating. One individual 
(FMNH 151624) had two placental scars on both 




-i 




Fig. 19-6. Above, right lateral view of the rhinarium 
of M. gymnorhyncha FMNH 151808; below, M. gracilis 
FMNH 151773. Scale 1 mm. 



the right and left oviducts. Mammary formula: 
2-0-4 (N = 1). One infant (FMNH 151785), with 
unopened eyes, was found at night along a trail. 
Remarks — The few specimens collected sug- 
gest that this species is uncommon in the area or 
that population numbers were low at the time of 
the trapping program. Four specimens of M. dob- 
soni in the MNHN were collected during the 
1970-1971 expedition to the Andringitra Massif, 
in the vicinity of Anjavidilava, and at approxi- 
mately 1995 m. Two of the specimen labels bear 
the note "foret mousse." Thus, the elevation 
range of this species within the reserve is pre- 
sumed to be between about 1200 and 2000 m. 



Microgale soricoides Jenkins, 1993 

Holotype — BM(NH) 91.565: adult male in al- 
cohol, skull extracted. Collected April 13, 1991, 
by C.J.R. 

Type Locality — Mantady National Park, ca. 
15 km north of Perinet (Andasibe), 18°51'S 
48°27'E, in primary rain forest, altitude 1 100— 
1150 m. 

Referred Material — FMNH 151768 and 
151778: 40 km S Ambalavao, along Volotsangana 
River, RNI d'Andringitra, 1210 m, 22°13'S 
46°58'E. 






206 



FIELDIANA: ZOOLOGY 




Fig. 19-7. Crania from left to right of M. gymnorhyncha FMNH 151807 and M. gracilis FMNH 151773. Above, 
lateral view; below, dorsal view left and ventral view right. 



Description — Size large, tail subequal to or 
longer than head and body. Pelage light gray- 
brown dorsally, gray-brown ventrally with a buff 
wash. Tail brown above, paler buffy brown below. 
Skull and mandible moderately large and robust, 
rostrum and interorbital region broad, braincase 
short and broad; parietals large, occipital small, 
supraoccipital ridge present. First upper II mark- 
edly robust and proodont. First lower il and i2 
robust and procumbent, i2 smaller than il but 
larger than c. The first upper and lower premolars 
are very small and have a single root. The teeth 
are too worn for determination of the features of 
m3. 

Measurements — External and cranial measure- 
ments are presented in Table 19-1. 

Reproduction — One of the specimens (FMNH 
151768) taken on December 1, 1993 was an adult 
female, based on dentition, and was lactating. 
Mammary formula: 2-0-4 (N = 1). 



Variation — The two specimens from RNI 
d'Andringitra are slightly greater in body size 
than those from Mantady but fall within the cra- 
nial size range (dimensions of the type series, with 
the RNI d'Andringitra specimens in square brack- 
ets: HB 77.0-85.5 mm [86, 100 mm], ratio of TL 
to HB 0.95-1.13 [1.04, 1.11], CIL 25.1-26.7 mm 
[24.6 mm], UTL 12.1-13.0 mm [ca. 12.2, 12.3 
mm]). 

Remarks — This is the second locality record 
for the species. Whereas the type series was col- 
lected in primary rain forest at an altitude be- 
tween 1 100 and 1 150 m, the RNI d'Andringitra 
specimens were collected at a slightly higher el- 
evation of 1210 m. One of the RNI d'Andrin- 
gitra specimens was collected 2 m above the 
ground on a tree limb 3 cm in diameter that led 
from the ground to a tangle of vines, so this 
species is obviously capable of climbing in low 
vegetation. 



JENKINS ET AL.: SHREW TENRECS 



207 





Fig. 19-8. Top, buccal view of permanent left anterior dentition of M. gracilis FMNH 151773. Scale 1 mm. 
Bottom, buccal view of permanent left anterior dentition of M. gymnorhyncha BM(NH) record number 1995.R258. 
Scale 1 mm. Figure continues on page 209. 



208 



FIELDIANA: ZOOLOGY 







Fig. 19-8 (cont.). Above, lingual view of left P4-M3 of M. gymnorhyncha BM(NH) record no. 1995.R258; 
middle, M. gracilis FMNH 151773; below, lingual view from left to right of right m3 of M. gymnorhyncha FMNH 
151651 and M. gracilis FMNH 151649. Scale 1 mm. 



Microgale gracilis (Major, 1896a) 

Oryzoryctes [sic] gracilis Major, 1896a 
Leptogale gracilis Thomas, 1918 

Holotype— BM(NH) 97.9.1.78: adult of un- 
determined sex; skin and skull. Collected Novem- 
ber 1 894 by C. I. Forsyth Major. 

Type Locality — Ambohimitombo forest (Am- 



bohimitombo town, 43 km [by road] SE Ambo- 
sitra, 10 km into eastern forest; Fianarantsoa; 
20°43'S 47°26'E— see MacPhee [1987]). Mac- 
Phee gives the altitude for this locality variously 
as 1300 m [1987, p. 6] and 1200 m [1987, table 
5] but, as pointed out by Carleton & Schmidt 
[1990], the altitude recorded for this locality by 
Major [1897] is 1400-1500 m.). 



JENKINS ET AL.: SHREW TENRECS 



209 






Referred Material — FMNH 151648, 
151649, and 151773: 40 km S Ambalavao, along 
Volotsangana River, RNI d'Andringitra, 1210 m, 
22°13'S 46°58'E; MNHN 1972-606, 1972-607: 
Foret Agauria (Marositry), Andringitra (2000 
m). 

Description — Size large, tail shorter than head 
and body. Pelage dark brown dorsally with buff 
speckling, ventrally dark gray with a buff wash, 
tail dark brown above, light brown below; the ju- 
veniles are less speckled on the rump than the 
adults. Muzzle very long; rhinarium large, naked 
region extends posterodorsally for 4-5 mm, an- 
terior portion reticulated, striae on the posterior 
region incomplete (Fig. 19-6). Eyes very small; 
ears small, partially concealed by the pelage, just 
reaching the posterior corner of the eye if pressed 
forward. Forefeet broad with enlarged claws, no- 
ticeably longer than those of the hind feet. Skull 
very elongated and gracile (Fig. 19-7), with a 
slender, elongated, markedly attenuated rostrum; 
the premaxillae meet the nasals above P2; the na- 
sals extend posteriorly into the elongated inter- 
orbital region. The braincase is rounded, moder- 
ately broad and long; the parietals are small rel- 
ative to the long frontals; the occipital is relatively 
shallow. The pterygoid foramina are partially or 
completely bridged by the palatine lip, which 
forms an incomplete or complete transverse bar. 
The mandible is elongated, tapers anteriorly, and 
the mental foramen lies below p2 or anterior to 
p3. The dentition is reduced (Fig. 19-8a,b); upper 
incisors subequal in height, incisors and canine 
very slender; extensive diastemata between all of 
the anterior teeth from II to P3, diastema partic- 
ularly long between P2 and P3; lingual cingulum 
sometimes present on P2, posterolingual acces- 
sory cusp absent; talon on P4 narrow; talons on 
molars very reduced, resembling cingula. The 
lower canine lacks an accessory cusp; the talonid 
of m3 is slightly reduced and the entoconid is 
lacking. 

Measurements — External and cranial measure- 
ments are presented in Tables 19-1 and 19-3. 

Variation — The small sample includes an 
adult male and female and a juvenile male. The 
dentition of the juvenile (Fig. 19-9) is at Stage 1 
of MacPhee (1987, p. 13), at which most of the 
antemolar teeth are deciduous, except for 13 and 
i3, whereas II has just started to erupt, and the 
upper and lower molars are fully erupted, as usu- 
al in the genus. Features of the deciduous den- 
tition that are common to Microgale (small size 
of the anterior teeth and the resemblance of the 



210 



Table 19-3. Table of comparative measurements (in 
mm) of Microgale gracilis and M. gymnorhyncha from 
the RNI d'Andringitra. Measurements of the holotype of 
M. gracilis are included in square brackets. 





M. gymno- 




Parameter 


rhyncha 


M. gracilis 


Head and body 


83-95 


91-105 [ca.93] 


length 


91.5 ± 4.92 


96.3 ± 5.26 




(4) 


(4) 


Tail length 


59-63.3 


75-84 [81] 




60.8 ± 1.48 


79.3 ±3.19 




(4) 


(4) 


Ratio of tail length 


2.3-2.4 


2.7-3.0 [2.8] 


to condyloincisive 


2.3 ± 0.05 


2.8 ± 0.13 


length 


(3) 


(4) 


Condyloincisive 


25.6-26.5 


27.9-29.6 [29.0] 


length 


26.0 ± 0.39 


28.8 ± 0.72 




(3) 


(4) 


Upper toothrow 


13.5-14.1 


14.0-14.9 [14.4] 


length 


13.8 ± 0.24 


14.4 ± 0.38 




(4) 


(4) 


Length of anterior 


7.0-7.6 


8.1-9.0 [8.4] 


dentition I-P3 


7.4 ± 0.23 


8.5 ± 0.34 




(4) 


(4) 


Ratio of anterior 


51.9-54.7 


57.8-60.4 [58.3] 


dentition to upper 


53.6 ± 1.11 


58.8 ± 1.05 


toothrow length 


(4) 


(4) 


Rostral breadth at 


2.8-3.0 


2.4-2.7 [2.5] 


level of P2 


2.9 ± 0.08 


2.6 ± 0.11 




(4) 


(4) 


Braincase length 


7.4-7.9 


8.4-8.8 [9.0] 




7.6 ± 0.22 


8.7 ± 0.15 




(3) 


(4) 



buccal cusps on the deciduous third upper pre- 
molar [dP4] to those of Ml and M2) also occur 
in this specimen. This juvenile is, however, un- 
usual in the presence of an additional cuspid be- 
tween the principal and posterior cusp on the first 
lower deciduous premolar (dp2) and between the 
metaconid and the posterior cusp on the third 
lower deciduous premolar (dp3); these cuspids 
are lacking in adult M. gracilis and have not 
been observed in other juvenile Microgale. The 
RNI d'Andringitra specimens closely resemble 
the holotype. In the small sample from the re- 
serve, the pelage of the Marositry specimens is 
marginally longer than that of those the Volot- 
sangana River. 

Reproduction — An adult male (FMNH 
151648), based on dentition, had abdominal testes 
measuring 7X3 mm, with a convoluted epidid- 
ymis. There are no data on the reproductive con- 
dition of the other two individuals. 

Remarks — This species is considered to be rare 
in museum collections (MacPhee, 1987) and is 
confirmed only from the holotype and the speci- 

FIELDIANA: ZOOLOGY 



mens recorded above. Unconfirmed records in- 
clude a specimen from Ankeramadinika (see Ma- 
jor, 1896a) and an illustration in Leche (1907) of 
the "milk" dentition of a specimen that is prob- 
ably correctly attributed to this species. The three 
specimens from the 1993 mission to RNI d'An- 
dringitra are therefore important to increase 
knowledge of the taxon. The two MNHN speci- 
mens (MNHN 1972-606 and 1972-607) are from 
the 1970-1971 Andringitra expedition and were 
taken in "foret Agauria." Based on the date of 
collection, this would have been at their camp 4 
(Marositry), at about 2000 m (Paulian et al., 
1971). Thus, the elevational range of this species 
on the Andringitra Massif appears to be between 
about 1200 and 2000 m. 



Microgale gymnorhyncha, new species 

Microgale gracilis (Major): MacPhee, 
1987, in part 

Holotype— FMNH 151807: adult female in al- 
cohol, skull extracted (field number SMG 6697). 
Collected December 13, 1993, by S.M.G. and 
C.J.R. 

Type Locality — 38 km S Ambalavao, RNI 
d' Andringitra, on ridge E of Volotsangana River, 
Fianarantsoa Province, 22° 1 1 '39"S 46°58' 16"E, al- 
titude 1625 m. 

Paratypes— FMNH 151808: old adult female, 
intact body in alcohol, (field number SMG 6698) 
collected December 13, 1993; FMNH 151726: 
adult female, skull, and complete skeleton (field 
number SMG 6712), collected December 14, 
1993; FMNH 151812: infant, in alcohol, skull ex- 
tracted (field number SMG 6724) collected De- 
cember 15, 1993, all with the same collection data 
as the holotype. FMNH 151650: juvenile male, 
skin and skull (field number SMG 6579) collected 
December 1, 1993; FMNH 151651: juvenile male, 
skin and skull (field number SMG 6614) Decem- 
ber 4, 1993, both collected by S.M.G. and C.J.R., 
40 km S Ambalavao, RNI d' Andringitra, along 
Volotsangana River, Fianarantsoa Province, 
22°13'22"S 46°48'18"E, altitude 1210 m. BM(NH) 
record number 1995. R258: adult, in alcohol, skull 
extracted, collected 1993 from unspecified local- 
ity in the RNI d' Andringitra, Fianarantsoa Prov- 
ince. 

Referred Material— MNHN 1961-204: 5 km 
from Fanovana (18°55'S 48°34'E, altitude ca. 



600-800 m [see Carleton & Schmidt, 1990]), near 
to Perinet (Andasibe). 

Diagnosis — Rhinarium very large, naked pos- 
terodorsal extension with transverse striae. Ears 
small, virtually concealed by the pelage. Skull 
pyriform, rostrum elongated, blunt anteriorly, 
braincase short. 

Description — Moderately large (see Tables 
19-1 and 19-3), tail shorter than head and body. 
Pelage soft, lustrous gray-brown dorsally, gray 
ventrally; dorsal hairs are three-banded, with light 
gray bases, light buff terminally, with bright 
brown tips; guard hairs dark brown or black; ven- 
tral hairs have light gray bases and brown tips. 
The feet are light-colored and the tail darker dor- 
sally, grading into the paler ventral surface. Muz- 
zle very long, forming a proboscis protruding an- 
teriorly well beyond mouth; rhinarium very large, 
transversely striated naked region extends postero- 
dorsally for ca. 6-7 mm (see Fig. 19-6). Eyes very 
small. Ears small, virtually concealed in the pel- 
age, anterior border lies far behind eye when 
pressed forward against head. Forefeet broad, 
claws enlarged. Skull long, moderately gracile 
and pyriform in shape; the anterior part of the 
skull is elongated in appearance but the posterior 
interorbital and braincase region appear somewhat 
anteroposteriorly shortened (see Fig. 19-7). The 
rostrum is slender and elongated but not notice- 
ably tapered, and the anterior is blunt and the na- 
sals slightly flared anteriorly; the nasals extend 
posteriorly into the interorbital region; the pre- 
maxillae meet the nasals at a level between C and 
P2. The braincase is short and broad; the superior 
articular facets are rectangular and visible in dor- 
sal view; the frontals and parietals are moderately 
small and the occipital moderately deep; the su- 
praoccipital suture is dorsally positioned on the 
braincase. The palatine lip forms the anterior bor- 
der of the pterygoid foramina. The mandible is 
sinuous, moderately elongated, and the mental fo- 
ramen lies below p3. The dentition is moderately 
reduced, and long diastemata are present between 
all of the anterior teeth from II to P3 (see Fig. 
19-8b). The first upper incisor is slightly pro- 
odont, with a well-developed distostyle approxi- 
mately one-third the height of the principal cusp; 
II is larger than 12 and 13, and subequal to C; 12 
is subequal in crown height to 13 but slightly less 
than C. The upper canine has a prominent disto- 
style and a small anterior accessory cusp. The first 
upper premolar (P2) is tricuspid; P3 is premolar- 
iform and resembles P4, with a small distinct me- 
sostyle, an anterior ectostyle, distostyle, and a 



JENKINS ET AL.: SHREW TENRECS 



211 



posterolingual accessory cusp; the mesostyle is 
present on P4, the anterior ectostyle is distinct, the 
distostyle and posterior ectostyle are present, and 
the talon is well marked and bicuspid. The molars 
have narrow but well-marked talons (see Fig. 19- 
8c). The first lower incisor (il) is procumbent 
with a distinct hypoconulid and a well-developed 
lingual cingulum. The lower canine has a promi- 
nent accessory cusp. The first lower premolar (p2) 
is tricuspid, taller than the canine, and with a ro- 
bust principal cusp. Diastemata are present be- 
tween all teeth from i2 to p3. The second lower 
premolar (p3) is similar to but larger than p2. The 
talonid of the third lower molar (m3) is slightly 
reduced; a talonid basin, hypoconid, hypoconulid, 
and entoconid ridge are present with a slight trace 
of an entoconid (see Fig. 19-8c). 

Deciduous Dentition — Of the three immature 
specimens belonging to this species, one is an in- 
fant (FMNH 151812) in which M2 and m2 are 
partially erupted and M3 and m3 are just begin- 
ning to erupt; the other two are juveniles (FMNH 
151650 and 151651) with fully erupted molars but 
a completely deciduous antemolar dentition. The 
deciduous dentition of this species (Fig. 19-9) 
shows similar traits to that of other species of Mi- 
crogale (see MacPhee, 1987), except that the de- 
ciduous incisors and canines are only slightly 
smaller than the permanent teeth. The lower sec- 
ond deciduous incisor (di2) has a prominent lin- 
gual cingulum unlike i2, and the lower deciduous 
canine (dc) has a small anterior accessory cusp 
unlike the permanent lower canine. 

Measurements — External and cranial measure- 
ments are presented in Tables 19-3 and 19-4. 

Variation — The pelage of the single specimen 
from Fanovana is harsher in texture than that of 
the RNI d'Andringitra specimens. 

Reproduction — All three of the adult females 
collected during the first half of December had 
large mammae; two specimens were lactating and 
one had three placental scars (one on left and two 
on right oviducts). Mammae formula: 2-0-4 (N = 
2). 

Etymology — The name of this species is de- 
rived from the Greek gymno — naked, rhynch — 
snout or nose, and refers to the prominent, naked 
rhinarium. 

Comparison with Other Species — With the ex- 
ception of M. gracilis, M. gymnorhyncha is readi- 
ly distinguished from all other species of Micro- 
gale recorded from the RNI d'Andringitra by fea- 
tures including the long muzzle and elongated rhi- 
narium, the small ears partially concealed in the 



pelage, the broad forefeet with enlarged claws, 
and the elongated rostrum and short braincase 
characteristic of the skull. 

Microgale gymnorhyncha is remarkably similar 
morphologically to M. gracilis and there are few 
differences in external appearance between the 
two species, which presumably share many adap- 
tive features. Microgale gracilis is slightly larger 
than M. gymnorhyncha (see table 19-3), with a 
longer tail relative to the head and body, In both 
species the muzzle is prolonged into a flexible 
proboscis; both have a large rhinarium, but that 
of M. gracilis is smaller ventrally and the pos- 
terodorsal extension is shorter (4-5 mm long), 
with a reticulated anterior portion and incomplete 
striae on the posterior portion. The eyes of M. 
gracilis are similarly nearly concealed but are 
slightly larger than those of M. gymnorhyncha, 
and the ears are also slightly larger, only partially 
concealed by the pelage, just reaching the poste- 
rior border of the eye if pressed forward. Both 
have broad forefeet with enlarged claws, but the 
hind foot is shorter in M. gymnorhyncha than in 
M. gracilis. In contrast to M. gymnorhyncha, the 
skull of M. gracilis is even more gracile and elon- 
gated, with a slender, elongated, markedly atten- 
uated rostrum and a longer interorbital region and 
braincase; the premaxillae meet the nasals above 
P2; the braincase is rounded and longer; the pter- 
ygoid foramina are partially or completely 
bridged by the palatine lip, which forms an in- 
complete or complete transverse bar. The mandi- 
ble of M. gracilis is elongated and anteriorly ta- 
pering, and the mental foramen lies below p2 or 
anterior to p3. The dentition of M. gracilis is grac- 
ile and more reduced than that of M. gymnorhyn- 
cha, with extensive diastemata between all of the 
anterior teeth, so that the anterior portion is more 
elongated relative to the whole toothrow than in 
M. gymnorhyncha (see Table 19-3). In contrast to 
M. gymnorhyncha, the upper incisors in M. grac- 
ilis are subequal in height, the incisors and canine 
are very slender, and the diastema between P2 and 
P3 is much longer; a lingual cingulum may be 
present on P3, but there is no posterolingual ac- 
cessory cusp; the talon on P4 is narrow; the talons 
on the molars are very reduced and resemble cin- 
gula; the lower canine lacks an accessory cusp; 
the talonid of m3 is more reduced and lacks an 
entoconid. There are few differences in the decid- 
uous dentitions of the two species, except that a 
metaconid is present on the first deciduous lower 
premolar (dp2) and an accessory cuspid is present 
posterior to the metaconid on the second lower 



212 



FIELDIANA: ZOOLOGY 






Fig. 19-9. Top: left anterior deciduous dentition of M. gymnorhyncha FMNH 151651 (above) and M. gracilis 
FMNH 151649 {below). Bottom: buccal view of upper dentition (above); lingual view of lower dentition (below). 
Scale 1 mm. 



deciduous premolar (dp3) of M. gracilis, in con- 
trast to M. gymnorhyncha, which lacks a meta- 
conid on dp2 and shows no trace of an accessory 
cuspid on dp4 (although a metaconid is present 
on dp3, contra MacPhee [1987, p. 18], who stated 
that a metaconid on dp3 is usually absent in all 
species other than M. parvula). 



Microgale gymnorhyncha is somewhat similar 
in external appearance to M. thomasi Major, 
1896a, a species that has not been recorded from 
the RNI d'Andringitra but is known to occur in 
eastern Madagascar from Ivohimanitra (ca. 
20°42'S 47°35'E) to Vondrozo (ca. 22°49'S 
47°20'E) (see MacPhee, 1987). Microgale tho- 



JENKINS ET AL.: SHREW TENRECS 



213 



masi differs from M. gymnorhyncha in the follow- 
ing features: the muzzle is shorter, and although 
the rhinarium is moderately large, it does not ex- 
tend posteriorly; the ears are obvious and the fore- 
feet are not broadened, nor are the foreclaws en- 
larged. The skulls of both species are similar in 
size but differ in their proportions — that of M. 
thomasi is more robust. In contrast to that of M. 
gymnorhyncha, the rostrum of M. thomasi is mod- 
erately broad and short; the braincase is rounded, 
broader, longer, and deeper; the parietals are larg- 
er relative to the frontals and occipital; the pter- 
ygoid foramina open buccally into the posterior 
part of the palate; the mandible is robust and not 
sinuous, and the mental foramen lies below the 
posterior root of p2. The two species differ den- 
tally: in M. thomasi the upper incisors and canine 
are not reduced, the diastemata between II and P3 
are much smaller, P2 is larger, the lingual shelves 
on the upper molars are larger; p2 is very large, 
the talonid of m3 is similar, although the talonid 
basin is slightly broader. 

Remarks — Currently M. gymnorhyncha is 
known only from the localities listed above in the 
RNI d'Andringitra, Fianarantsoa Province, south- 
ern Madagascar and from Fanovana, central east- 
ern Madagascar. All of the specimens from RNI 
d'Andringitra were trapped on slopes or ridges in 
montane forests at an altitude of 1210 m or in 
upper montane forest at 1625 m. No altitude was 
recorded for the specimen from Fanovana; it is 
estimated as 600-800 m based on the relief of the 
area. 

No observations have been made on the behav- 
ior of this species. Although it is inadvisable to 
interpret morphological features in behavioral 
terms without direct observations, the combina- 
tion of such features as the long muzzle with a 
large, naked rhinarium, the small eyes and par- 
tially concealed ears, and the broad forefeet with 
enlarged foreclaws suggests that this animal may 
use its forefeet and snout to rootle in the humus 
layer. The elongated rostrum and well-spaced 
teeth also suggest dietary specializations, al- 
though the identifiable gut contents of a single 
specimen included only arthropod fragments. 



Microgale sp. A 

Referred Material — FMNH 151646 and 
151647: 40 km S Ambalavao, RNI d'Andringitra, 
along Volotsangana River, Fianarantsoa Province, 
22°13'22"S 46°58'18"E, altitude 1210 m; BM(NH) 



95.257: Basecamp 1, Maitso, RNI d'Andringitra, 
ca. 22°10'S, 46°50'E. 

Description — In preparation (RD.J. et al.). 

Measurements — External and cranial measure- 
ments of the specimens from RNI d'Andringitra 
are given in Table 19-1. 

Variation — This species has been collected at 
several widely separated localities. Specimens 
from the RNI d'Andringitra population are larger 
on average than any of the other populations in 
external and cranial dimensions. In particular, the 
braincase of the RNI d'Andringitra specimens is 
broader and deeper than that of Pare National 
(PN) de la Montagne d'Ambre specimens. The 
two specimens from the PN de Ranomafana, al- 
though geographically closer to the RNI d'An- 
dringitra specimens in southeastern Madagascar, 
are, however, more similar to the PN de la Mon- 
tagne d'Ambre specimens in size, as is the single 
juvenile specimen from the Reserve Speciale (RS) 
d'Ambatovaky. 

Reproduction — Both of the RNI d'Andringitra 
specimens collected in early December were preg- 
nant females with permanent dentition. One in- 
dividual (FMNH 1 5 1 646) had three embryos (two 
left and one right) measuring 15 mm crown-rump 
length, and the other (FMNH 151647) had four 
embryos (two left and two right) measuring 17 
mm crown-rump length. Mammary formula: 
2-0-4 (N = 2). 



Discussion 

Although there is still doubt concerning the 
number of distinct species of Microgale occurring 
in Madagascar, the 10 species here recorded for 
RNI d'Andringitra form a high proportion of the 
17 species currently recognized by us. Sympatric 
association between several species is a common 
phenomenon in the genus Microgale, and two to 
five species were recorded from half of the col- 
lecting localities mapped by MacPhee (1987). In 
areas that have been studied intensively, even 
greater numbers of species have been recorded. 
For example, six species were recorded from PN 
de la Montagne d'Ambre (Raxworthy & Nuss- 
baum, 1994) and seven from RS d'Analamazaotra 
(Nicoll & Langrand, 1989). This high level of 
sympatry of RNI d'Andringitra is greater than that 
observed to date at any other locality, and al- 
though it certainly reveals the high diversity of 
the area, it may also reflect the thoroughness of 



214 



FIELDIANA: ZOOLOGY 



the inventory. The ecological implications of the 
speciosity of the area are discussed in Chapter 20. 
Although MacPhee (1987) emphasized that the 
clusters he used to group species were a purely 
phenetic device, these clusters continue to be a 
very helpful method of associating species that 
share morphological features. The species occur- 
ring in the RNI d'Andringitra may be placed in 
these clusters, which are recorded below with a 
summary of the features shared by the included 
species: 

cowani cluster — Microgale cowani, M. taiva, 
M. melanorrhachis (M. parvula less certainly 
associated) — craniodental similarities, foot 
proportions unmodified. 

gracilis cluster — M. gracilis, M. gymnorhyn- 
cha — craniodental similarities, forefeet 
broad, foreclaws enlarged, rhinarium modi- 
fied. 

longicaudata cluster — M. longicaudata — cra- 
niodental characters, elongated hind feet, 
elongated tail. 

dobsoni cluster — M. dobsoni — craniodental 
characters, foot proportions unmodified. 

soricoides cluster — M. soricoides — cranio-den- 
tal characters. 

pusilla cluster — not represented in the collec- 
tions from RNI d'Andringitra. 

brevicaudata cluster — not represented in the 
collections from RNI d'Andringitra. 

Limited data on the annual reproductive cycle 
were provided for M. dobsoni and M. talazaci by 
Eisenberg and Gould (1970), both from trapping 
programs and from captive animals. More recent- 
ly, Stephenson and Racey (1993) provided infor- 
mation on reproductive energetics in M. talazaci 
and M. dobsoni, as well as limited data on M. 
cowani and M. melanorrhachis. The data on re- 
productive condition for all 10 species collected 
at RNI d'Andringitra are presented in the preced- 
ing pages, and, although confined to the months 
of November and December when the trapping 
program was in operation, they provide a consid- 
erable increase of information over that previous- 
ly available for the genus. 

Some trends are evident within most species of 
Microgale found at the RNI d'Andringitra. On the 
basis of dentition, both adults and juveniles were 
found in most species, with the exception of M. 
melanorrhachis (juveniles only) and M. dobsoni, 
M. soricoides, and Microgale sp. A (adults only). 
On the basis of reproductive condition of the go- 
nads, sexually adult animals of both sexes were 



evident in M. cowani, M. taiva, and M. longicau- 
data only, but sexually adult animals of one sex 
were found in the other species. Species in which 
dental juveniles showed some signs of sexual ma- 
turity included M. cowani, M. parvula, and pos- 
sibly also M. melanorrhachis, where individual 
females were perforate but showed no other evi- 
dence of sexual maturity, and M. taiva, in which 
an individual male had well-developed testes. Few 
conclusions may be drawn from these results, al- 
though there is evidence to suggest that dental ju- 
veniles may occasionally be recruited into the 
sexually adult population. Whether this indicates 
sexual precocity of juveniles or retardation of 
dental development into adulthood is impossible 
to decide from these limited data. 

Dental development among juveniles of all spe- 
cies appeared to be at a fairly early stage, and 
there was no evidence of any development be- 
yond Stage 1 to 2. This observation may coincide 
with the months of collection, November and De- 
cember; these months are also at the beginning of 
the annual rainy season in eastern humid forest. 
The breeding season for this genus presumably 
starts earlier in the year, and it is probable that 
the presence of juveniles of this age is positively 
correlated with the start of the rainy season when 
food supplies should begin to increase. Individu- 
als in which the molars were only partially erupt- 
ed occurred in M. cowani, M. taiva, M. longicau- 
data, and M. gymnorhyncha. Individuals with ful- 
ly erupted molars but completely deciduous an- 
temolar teeth were found in M. cowani, M. taiva, 
M. melanorrhachis, M. longicaudata, M. parvula, 
and M. gymnorhyncha. Individuals showing Stage 
1 to 2 states of development were found in M. 
cowani, M. taiva, and M. gracilis. It is probably 
significant that those species most commonly cap- 
tured, M. cowani and M. taiva, also demonstrated 
the greatest range of variation in dental develop- 
ment of juveniles. 



Acknowledgments 

For the loan of specimens in their care, we 
thank Laurent Granjon, Michel Tranier, and 
Jacques Cuisin, Mammiferes et Oiseaux, MNHN, 
Paris, and Michael Carleton, Marc Frank, and He- 
len Kafka, Department of Mammalogy, USNM, 
Washington, D.C. Maria Rutzmoser, MCZ, Har- 
vard, kindly provided information on the holotype 
of M. parvula. Members of the 1993 Cambridge 



JENKINS ET AL.: SHREW TENRECS 



215 



University expedition, Louise Ashmore, Juliet 
O'Keefe, Matthew Thomas, and Oliver Tunstall 
Pedoe, made a small collection of specimens. For 
the opportunity to examine these species we thank 
them and, particularly, Nasolo Rakotoarison, Pare 
Botanique et Zoologique de Tsimbazaza for his 
cooperation in the loan of material. During this 
study the salary of C.J.R. was provided by a grant 
(BSR 90-24505) from the National Science Foun- 
dation. The photographs were prepared by Phillip 
Crabbe, Photographic Unit, BM(NH). 



Literature Cited 

Butler, P. M., and M. Greenwood. 1979. Soricidae 
(Mammalia) from the Olduvai Gorge, Tanzania. Zoo- 
logical Journal of the Linnean Society, 67: 329-379. 

Carleton, M., and D. F. Schmidt. 1990. Systematic 
studies of Madagascar's endemic rodents (Muroidea: 
Nesomyinae); an annotated gazetteer of collecting lo- 
calities of known forms. American Museum Novita- 
tes, 2987: 1-36. 

Coquerel, C. 1848. Note sur une espece nouvelle de 
musaraigne trouvee a Madagascar. Annales des Sci- 
ences Naturelles, Zoologie, 9: 193-198. 

Eisenberg, J. E, and E. Gould. 1970. The tenrecs: A 
study in mammalian behavior and evolution. Smith- 
sonian Contributions to Zoology, 27: 1-138. 

Froriep, L. F. 1806. C. Dumeril's Analytische Zoologie 
aus dem Franzosischen, mit Zusatzen. Weimar, 345 
pp. 

Genest, H., and F Petter. 1975. Part 1.1. Family Ten- 
recidae, pp. 1-7. In Meester, J., and H. W. Setzer, eds., 
The mammals of Africa: An identification manual. 
Smithsonian Institution Press, Washington, D.C. 

Grandidier, A. 1870. Description de quelques animaux 
nouveaux, decouverts a Madagascar. Revue et Maga- 
sin de Zoologie, 22: 49-54. 

Grandidier, G. 1934. Deux nouveaux mammiferes in- 
sectivores de Madagascar. Bulletin Museum National 
d'Histoire Naturelle, ser. 2, 6: 474-477. 

Heim de Balsac, H. 1972. Insectivores, pp. 629-660. 
In Battistini, R., and G. Richard- Vindard, eds., Bio- 
geography and ecology in Madagascar. W Junk, the 
Hague, 765 pp. 

Hutterer, R. 1993. Order Insectivora, pp. 69-130. In 
Wilson, D. E., and D. M. Reeder, eds., Mammal spe- 
cies of the world: A taxonomic and geographic ref- 
erence. Smithsonian Institution Press, Washington, 
D.C, 1206 pp. 

, and M. tranier. 1990. The immigration of 

Asian house shrew (Suncus murinus) into Africa and 
Madagascar, pp. 309-320. In Peters, G, and R. Hut- 
terer, eds. Vertebrates in the tropics. Museum Alex- 
ander Koenig, Bonn, 585 pp. 

Jenkins, P. D. 1988. A new species of Microgale (In- 
sectivora: Tenrecidae) from northeastern Madagascar. 
American Museum Novitates, 2910: 1-7. 

. 1992. Description of a new species of Micro- 
gale (Insectivora: Tenrecidae) from eastern Madagas- 



car. Bulletin of the British Museum Natural History 
(Zoology), 58: 53-59. 

1993. A new species of Microgale (Insectiv- 



ora: Tenrecidae) from Eastern Madagascar, with an 
unusual dentition. American Museum Novitates, 
3067: 1-11. 

Lacepede, B. G. E. de 1799. Tableau des divisions, 
sous-divisions, ordres et genres des mammiferes. Par- 
is. 

Leche, W 1907. Zur Entwicklungsgeschichte des Zahn- 
systems der Saugetiere, zugleich ein Beitrag zur Stam- 
mesgeschichte dieser Tiergruppe. Zweiter Teil: Phy- 
logenie. Zweites Heft: Die Familien der Centetidae, 
Solenodontidae und Chrysochloridae. E. Schweizer- 
bartsche (E. Nagele), Stuttgart, 157 pp. 

Linnaeus, C. 1766. Systema naturae per regna tria na- 
turae secundum classes, ordines, genera, species, cum 
characteribus, differentiis, synonymis, locis. 12th edi- 
tion. Holmiae. 

MacPhee, R. D. E. 1987. The shrew tenreces of Mad- 
agascar: Systematic revision and Holocene distribu- 
tion of Microgale (Tenrecidae: Insectivora). American 
Museum Novitates, 2889: 1-45. 

Major, C. I. Forsyth 1896a. Diagnoses of new mam- 
mals from Madagascar. Annals and Magazine of Nat- 
ural History, (6) 18: 318-325. 

. 1896b. Descriptions of four additional new 

mammals from Madagascar. Annals and Magazine of 
Natural History, (6) 18: 461-463. 

1897. On the general results of a zoological 



expedition to Madagascar in 1894-96. Proceedings of 
the Zoological Society of London, for 1896: 971-981. 
Martin, W 1838. On a new genus of insectivorous 
Mammalia. Proceedings of the Zoological Society of 
London, for 1838: 17-19. 

Mills, J. R. E. 1966. The functional occlusion of the 
teeth of Insectivora. Journal of the Linnean Society 
(Zoology), 47: 1-125. 

Milne Edwards, A., and A. Grandidier. 1872. De- 
scription d'un nouveau mammifere insectivore de 
Madagascar (Geogale aurita). Annales des Sciences 
Naturelles, Zoologie, 15, Article 19: 1-5. 

Mivart, St. George. 1871. On Hemicentetes, a new 
genus of Insectivora, with some additional remarks on 
the osteology of that Order. Proceedings of the Zoo- 
logical Society of London, for 1871: 58-79. 

Morrison-Scott, T. C. S. 1948. The insectivorous gen- 
era Microgale and Nesogale (Madagascar). Proceed- 
ings of the Zoological Society of London, 118: 817- 
822. 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: 
Revue de la conservation et des Aires Protegees. 
World Wide Fund for Nature, Gland, Switzerland, 
XVII+374 pp. 

, and G. B. Rathbun. 1990. African Insectivora 

and Elephant-shrews: An action plan for their conser- 
vation. IUCN/SSC Insectivore, Tree-shrew and Ele- 
phant-shrew Specialist Group, Gland, Switzerland, 
iv+53 pp. 

O'Keefe, J., and L. Ashmore. [1994]. Madagascar 
1993 small mammal project. Unpublished report. Uni- 
versity of Cambridge. 



216 



FIELDIANA: ZOOLOGY 



Paulian, R., J.-M. Betsch, J.-L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. RCP 225. Etude des ecosys- 
temes montagnards dans le region malgache. I. Le 
massif de l'Andringitra. 1970-1971. G6omorphologie, 
climatologie et groupements v6g&aux. Bulletin de la 
Soci6te" d'Ecologie, 11(2-3): 189-266. 

Raxworthy, C. J., and R. A. Nussbaum. 1994. A rain- 
forest survey of amphibians, reptiles and small mam- 
mals at Montagne d'Ambre, Madagascar. Biological 
Conservation, 69: 65-73. 

Stephenson, P. J. 1995. Taxonomy of shrew-tenrecs 
(Microgale spp.) from eastern and central Madagascar. 
Journal of Zoology, 235: 339-350. 

, and P. A. Racey. 1993. Reproductive energet- 
ics of the Tenrecidae (Mammalia: Insectivora). 2. The 



shrew-tenrecs, Microgale spp. Physiological Zoology, 
66: 664-685. 

Swindler, D. R. 1976. Dentition of Living Primates. 
Academic Press, London, 308 pp. 

Thomas, [M. R.] Oldfield. 1882. Description of a new 
genus and two new species of Insectivora from Mad- 
agascar. Journal of the Linnean Society (Zoology), 16: 
319-322. 

1 884. Description of a new species of Micro- 



gale. Annals and Magazine of Natural History, (5) 14: 
337-338. 

1918. On the arrangement of the small Ten- 



recidae hitherto referred to Oryzorictes and Microga- 
le. Annals and Magazine of Natural History, (9) 1: 
302-307. 



Appendix 19-1. 

Key to the Species of Microgale Occurring in RNI d'Andringitra 

1- Size very small: HB < 64, CIL < 16.6; TL subequal to HB 0.97-1.03; dark brown dorsal and ventral 
pelage M. parvula 

Size larger: HB > 63, CIL > 17.3; pelage not dark brown dorsally and ventrally 2 

2- Ratio of TL:HB 1.7-1.9; dorsal pelage reddish M. longicaudata 

Ratio of TL:HB < 1.4 3 

3- Digits and tail tip contrastingly paler than body, tail, and feet Microgale sp. A. 

Tail tip and digits not obviously paler than rest of body 4 

4- Dorsal pelage with a distinct, dark mid-dorsal stripe extending from the head to the base of the tail; 
tail bicolored, dark brown above, lighter below M. melanorrhachis 

No distinct mid-dorsal stripe 5 

5 - Size very large: HB > 108, CIL > 30.0; i2 » c M. dobsoni 

Size smaller: HB < 105, CIL < 30.0; i2 subequal or > c 6 

6- Proboscis long, large rhinarium extends posterodorsally onto muzzle; forefeet broad, foreclaws en- 
larged 7 

Small rhinarium confined to anterior of short proboscis; forefeet slender without lengthened foreclaws 
8 

7- Posterior region of rhinarium with transverse striae; ratio of I1-P3:UTL < 55.0; BL < 8.0 

M. gymnorhyncha 

Posterior region of rhinarium reticulated; ratio of I1-P3:UTL > 57; BL > 8.4 M. gracilis 

°- II robust, markedly proodont; il » i2 > c; P2 and p2 very small with single roots M. soricoides 

II neither robust nor markedly proodont; il < or subequal to i2; P2 and p2 with two roots 9 

9- Ratio of TL:CIL 2.7-3.1; ratio of I1-P3:UTL 52.8-56.4 M. cowani 

Ratio of TL:CIL 3.5-4.2; ratio of I1-P3:UTL 49. 1-52.8 M. taiva 



JENKINS ET AL.: SHREW TENRECS 



217 






Chapter 20 

Insectivore Ecology in the Reserve Naturelle Integrate 
d'Andringitra, Madagascar 

Steven M. Goodman, Christopher J. Raxworthy, 
and Paulina D. Jenkins 



Abstract 

Insectivores were surveyed on the eastern side of the Reserve Naturelle Integrate d'Andrin- 
gitra, at four elevational zones between 720 and 1625 m. Little previous information was 
available on the small mammals of the reserve. A total of 12 species of insectivores (Setifer, 
Tenrec, and 10 species of Microgale) were found in the reserve during the 1993 survey, and 
with earlier records and specimens, a total of 14 species are known from the Andringitra Massif. 
Microgale taiva and M. principula occurred across the complete transect, M. melanorrhachis 
was restricted to the lowland forest, and Microgale sp. A, M. dobsoni, M. soricoides, M. 
gracilis, and M. gymnorhyncha were only captured in montane forest. On the basis of trap 
results (principally pitfall buckets with drift fence), the highest diversity and density of Insec- 
tivora was at 1210 m elevation, with eight species. Pitfall buckets placed in valley bottoms 
had a higher capture rate and species diversity than those on slopes or ridges, although there 
was no clear pattern of species habitat segregation. Collections of soil invertebrates were made 
adjacent to the pitfall lines, and the general trends of invertebrate density and Microgale density 
and abundance paralleled each other. 

Resume 

L'inventaire des insectivores du versant est de la Reserve Naturelle Integrate d' Andringitra 
a ete effectue dans quatre zones d' altitude comprises entre 720 m et 1625 m. Peu d' informations 
relatives aux petits mammiferes de la reserve etaient jusqu'alors disponibles. Au cours de 
l'inventaire de 1993, un total de 12 especes d' insectivores {Setifer, Tenrec et 10 especes de 
Microgale) a ete trouve dans la reserve. Avec les specimens recenses auparavant, un total de 
14 especes est connu du massif d' Andringitra. Microgale taiva et M. principula sont presents 
sur l'ensemble du transect. M. melanorrhachis est limitee a la foret de basse altitude. Microgale 
sp. A, M. dobsoni, M. soricoides, M. gracilis et M. gymnorhyncha ont ete capturees dans la 
foret de montagne. D'apres les resultats des pieges (principalement les "pitfall"), la plus grande 
diversite et densite d' insectivores est constatee a une altitude de 1210 m avec huit especes. 
Les pieges "pitfall" places dans les vallees recelent un taux de capture et une diversite spe- 
cifique plus eleves que ceux places sur les pentes et les cretes, bien qu'il n'y ait aucune evidence 
de segregation d'habitat par les especes. La collecte d'invertebres du sol a ete faite parallele- 
ment aux lignes "pitfall" et la densite et l'abondance des invertebres et des Microgale montrent 
une tendance generate parallele. 



218 FIELDIANA: ZOOLOGY 



Introduction 

Little is known about the natural history of the 
tenrecs (Order Insectivora, Family Tenrecidae), all 
of which, with the exception of the African otter- 
shrews (Subfamily Potamogalinae), are endemic 
to Madagascar. The tenrecs are speciose, broadly 
distributed, and occur in a variety of biomes on 
the island. They represent a group that shows a 
remarkable degree of morphological and ecolog- 
ical diversification. 

The most in-depth study of the behavioral ecol- 
ogy of the tenrecs to date is by Eisenberg and 
Gould (1970). It was conducted at sites in the 
eastern humid forest, particularly in the Reserve 
Speciale (RS) d'Analamazaotra (Perinet), 400 km 
NE of the Reserve Naturelle Integrate (RNI) 
d'Andringitra. The tenrecs have also been the fo- 
cus of reproductive and physiology studies (Ni- 
coll, 1982; Stephenson, 1991; Stephenson & Ra- 
cey, 1993). 

The purpose of this study was to document in- 
sectivore distribution along an elevational gradi- 
ent on the eastern slopes of the RNI d'Andringitra 
and to present basic natural history information 
on these animals. Abiotic variables (climate) and 
biotic variables (macrofauna soil invertebrates) 
were compared to our trapping data to explore 
possible correlations and explanations of the al- 
titudinal distribution, density, and species richness 
of Microgale along the slopes of the Andringitra 
Massif. 



Table 20-1. The known elevational distribution of 
Insectivora on the eastern slopes of the RNI d'Andrin- 
gitra. Includes sight records, live traps, and pitfall traps 
from 1993 survey. 



Materials and Methods 
Trap Lines 

The principal means of capturing insectivores 
was with pitfall buckets and a drift fence. In the 
four transect zones (720, 810, 1210, and 1625 m), 
three separate pitfall lines (one in valley bottom, 
one on a slope, and one on a ridge crest), each 
100 m long and consisting of 11 buckets, 10 m 
apart, were in operation for a minimum of 6 
nights. More details on the technique are given in 
Chapter 17. A few insectivores were also captured 
with standard Sherman live traps. The trap types, 
placement, baits, etc. for these lines are described 
in Chapter 22. 

Traps and pitfalls were visited at least twice per 
day, once at dawn and again in the late afternoon. 
A "trap-day" and "bucket-day" are defined as a 





Elevational 


range (m) 


Species 


720 


810 


1210 


1625 


Microgale cowani 


- 


+ 


+ 


+ 


Microgale taiva 


+ 


+ 


+ 


+ 


Microgale melanorrhachis 


+ 


+ 


- 


- 


Microgale longicaudata 


+ 


- 


- 


+ 


Microgale parvula 


+ 


+ 


+ 


+ 


Microgale dobsoni 


- 


- 


+ 


- 


Microgale soricoides 


- 


- 


+ 


+ 


Microgale gracilis 


- 


- 


+ 


- 


Microgale gymnorhyncha 


- 


- 


+ 


+ 


Microgale sp. A 


- 


- 


+ 


- 


Setifer setosus 


- 


+ 


- 


- 


Tenrec ecaudatus 


+ 


+ 


- 


- 


Total number of 










Microgale spp. 


4 


4 


8 


6 


Total number of 










Insectivora 


5 


6 


8 


6 



24-hour period (dawn to dawn) of one of these 
devices in use. After rains the buckets were 
sponged dry. Our inventory of the RNI d'Andrin- 
gitra was conducted during a period (November 
14 to December 18, 1993) when all tenrecs were 
expected to be active (Stephenson, 1994). 



The Known Insectivoran Community 
of the RNI d'Andringitra 

A taxonomic analysis of the Microgale ob- 
tained during the 1993 survey of the eastern 
slopes of the RNI d'Andringitra is presented in 
Chapter 19. A total of 10 species were collected 
(Table 20-1). Two other genera and species of 
Tenrecidae were also found in the RNI d'Andrin- 
gitra in 1993; others have been reported from the 
reserve or surrounding areas. Fourteen species 
have been documented in the reserve (Table 
20-2). Below we present a summary of Insectiv- 
ora reported from the reserve, excepting the Mi- 
crogale discussed in Chapter 19. Most of the ad- 
ditional specimen records, housed in the Mus6um 
National d'Histoire Naturelle (MNHN), Paris, are 
from the 1970-1971 Recherche Cooperative sur 
Programme no. 225 du Centre national de la Re- 
cherche Scientifique expedition to the Andringitra 
Massif (see Paulian et al., 1971; and Chapter 19). 
Taxa presented in brackets represent those whose 



GOODMAN ET AL.: INSECTIVORE ECOLOGY 



219 






Table 20-2. The known Insectivora community of 
the RNI d'Andringitra, based on museum specimens and 
photographs. 

Microgale cowani 
Microgale taiva 
Microgale melanorrhachis 
Microgale longicaudata 
Microgale parvula 
Microgale dobsoni 
Microgale soricoides 
Microgale gracilis 
Microgale gymnorhyncha 
Microgale sp. A 
Oryzorictes tetradactylus 
Hemicentetes nigriceps 
Setifer setosus 
Tenrec ecaudatus 



occurrence in the reserve are in need of further 
documentation. 



Subfamily Oryzorictinae 

[Limnogale mergulus Major, 1896a 

This semiaquatic tenrec is known to dismantle 
crustacean and invertebrate prey on rocks along 
river margins. Nicoll and Rathbun (1990) reported 
possible feeding signs of this species along a 
small river 15 km N of Antanifotsy. No evidence 
of this species was found in 1993 in the RNI 
d'Andringitra. Villagers living in and around the 
eastern edge of the reserve were clearly unfamil- 
iar with a precise description of this animal, par- 
ticularly the webbed feet and laterally compressed 
tail. Nevertheless, they noted that a water rat, 
"voalavo rano," occasionally was recovered in 
their eel traps.] 



identification as a juvenile M. cowani. The ab- 
sence of M. talazaci in the RNI d'Andringitra is 
commented on below. 



Oryzorictes tetradactylus 

Milne Edwards & Grandidier, 1882 

The only known records of this species on the 
Andringitra Massif are four specimens taken in 
1970 during the RCP mission. These include sin- 
gle specimens obtained along the Plateau d'An- 
dohariana (2030 m) and near the Cuvette Boby 
(2470 m), just below Pic Boby (Paulian et al., 
1971). Two other specimens taken on the same 
expedition are simply labeled "Andringitra." The 
known elevational range of this species in the RNI 
d'Andringitra is above the forested zone and be- 
tween 2000 and 2500 m. The labels of these spec- 
imens bear no information concerning the repro- 
ductive condition of the animals. 



Subfamily Tenrecinae 

Hemicentetes nigriceps Gunther, 1875 

Nicoll and Langrand (1989) reported this spe- 
cies, under the name H. semispinosus, as occur- 
ring in the RNI d'Andringitra. A photograph 
taken by O. Langrand of an animal in Antanifotsy 
is clearly referable to H. nigriceps. Furthermore, 
Glaw and Vences (1994) photographed H. nigri- 
ceps within the reserve. We found no evidence of 
this species during the 1993 expedition. We fol- 
low Eisenberg and Gould (1970) in considering 
H. nigriceps specifically distinct from H. semis- 
pinosus. 



Microgale Species 

Two species of Microgale that had been pre- 
viously reported from the Andringitra Massif on 
the basis of specimens obtained during the RCP 
campaign were not collected during the 1993 ex- 
pedition: M. talazaci Major, 1896a and M. drou- 
hardi Grandidier, 1934. However, P.D.J, reidenti- 
fied MNHN 1972-609 (originally identified as M. 
talazaci; MacPhee, 1987) as a juvenile M. dob- 
soni Thomas, 1884. MNHN 1972-611 was pre- 
viously identified as M. drouhardi and was sub- 
sequently listed by MacPhee (1987) under the 
synonymy of M. cowani; P.D.J, confirmed its 



Setifer setosus (Schreber, 1777) 

Our only record of this species from the 1993 
expedition is a male captured in a pitfall trap at 
810 m. Nicoll and Langrand (1989) also reported 
Setifer setosus from the reserve. 



Tenrec ecaudatus (Schreber, 1777) 

This species was observed on two occasions at 
720 m and thrice at 810 m during the 1993 ex- 
pedition. No specimen was collected. Nicoll and 



220 



FIELDIANA: ZOOLOGY 



Langrand (1989) also reported this species from 
the reserve. 



[Family Soricidae 

Subfamily Crocidurinae 

Suncus madagascariensis (Coquerel, 1848) 

Nicoll and Langrand (1989) noted that this spe- 
cies occurs in the reserve, although it was not col- 
lected during the 1993 survey. We are unaware of 
any documentation of this species in the reserve.] 



Analysis and Discussion 

General 

A total of 902 pitfall bucket-days were accrued 
during the small mammal survey of the eastern 
slopes of the RNI d' Andringitra, between Novem- 
ber 14 and December 17, 1993, comprising 198 
bucket-days at 720 m, 231 at 810 m, 220 at 1210 
m, and 253 at 1625 m (Table 20-3). Ninety-one 
small mammals were captured, including 88 Mi- 
crogale, one Setifer, and two rodents. Further, 
2,475 trap-nights using small mammal traps were 
also accrued (Chapter 22). Within the Insectivora, 
11 species (10 species of Microgale and the 
monotypic Setifer) were captured. Seven of these 
species had not been previously recorded in the 
reserve (MacPHee, 1987; Nicoll & Langrand, 
1989), and two are new to science (Chapter 19). 
Only one species, Microgale dobsoni, was cap- 
tured in the small mammal traps but not in the 
pitfall buckets, whereas five species of Microgale 
were trapped only in the pitfall buckets. Before 
proceeding with the analysis of the trapping re- 
sults, it is critical to determine if the sampling 
effort was sufficient to reflect some measure of 
completeness for the survey and the actual ten- 
recid species richness within each elevation zone. 



Species Accumulation Curves 

By plotting the total number of species known 
from each pitfall line ( 1 1 pitfall buckets per 
24-hour period) or elevational zone (33 pitfall 
buckets per 24-hour period), species accumulation 
curves may be drawn. An examination of these 



curves (Fig. 20-1) shows that no additional spe- 
cies of Microgale was added in the 720 m zone 
after 66 pitfall bucket-days (total four species in 
198 pitfall bucket-days), in the 180 m zone after 
198 pitfall bucket-days (total four species in 231 
pitfall bucket-days), in the 1210 m zone after 132 
pitfall bucket-days (total seven species in 23 1 pit- 
fall bucket-days), and in the 1625 m zone after 
198 pitfall bucket-days (total six species in 253 
pitfall bucket-days). The decrease in the addition 
of previously unrecorded species with additional 
trapping effort did not coincide with a general de- 
cline in overall pitfall trap success (Fig. 20-1). 

In general, virtually all of the Microgale spe- 
cies expected to occur within the reserve on the 
basis of their known distribution in the southern 
portion of the eastern humid forests were found 
during the 1993 survey. The major exception was 
Microgale talazaci, the largest extant species, 
known from areas north and south of the Andrin- 
gitra Massif, as well as within the reserve during 
the 1970-1971 RCP expedition (Paulian et al., 
1971). Of the specimens from the reserve, how- 
ever, MacPhee (1987) listed two, MNHN 1972- 
608 and MNHN 1972-612 (a skin only), taken in 
the upper elevational zone, outside of the area 
sampled in 1993. Neither of these specimens was 
examined for this study, but a third specimen 
(MNHN 1972-609) bearing the same label name 
was found to be a juvenile M. dobsoni. At other 
sites in the eastern humid forest, M. talazaci is 
known from below 1000 m (MacPhee, 1987; Rax- 
worthy & Nussbaum, 1994). Microgale talazaci 
shows seasonal variation in activity (Stephenson, 
1994); however, this in itself is insufficient to ex- 
plain the absence of this species during the 1993 
survey, because November and December are 
months in which it has been trapped in consider- 
able numbers with pitfalls at other sites (Raxwor- 
thy & Nussbaum, 1994). Further work is neces- 
sary to determine if M. talazaci occurs within the 
RNI d' Andringitra. 

Within the endemic Insectivora there are a few 
other trapping anomalies. Two other Microgale 
are known from the general region, yet they were 
not captured in the RNI d' Andringitra. M. thomasi 
Major, 1896a, a species rare in collections, was 
obtained near Vondrozo, a site about 70 km SE 
of the reserve, and M. pusilla Major, 1 896b, was 
taken from Vinanitelo, a similar distance to the 
NE of the reserve (MacPhee, 1987). Although not 
recorded in such close proximity, M. principula 
Thomas, 1918, is known from several localities in 
the eastern forest from the Reserve Speciale (RS) 



GOODMAN ET AL.: INSECTIVORE ECOLOGY 



221 



rJ ON NO 00 
•* — 00 



o > on no oo 

m — 00 

NO 





O <Z> (N r- NO 




ON NO 


1) 


n 


e 




a 




e 







tN (S 


E 




o > ->' r-^ t-~ 




tF r- 



2a.-<^£ 



> en On t-» 



O > NO >-i vO 
fNl — <N NO 



O £>/ NO — NO 

— •* — rsi no 



6 •£ 
S e 
2 £ 
E E 

S3 

« 5» >> 

g-o-o >, 

3 E « «•§ 

£, <u a- q. _ 

<-> c c & 

(O cd c c ca 

•gag I B 

•3 g e a s 
•5 .a .a co o 



00 Tf O 00 Tt d 



oo r~- 

— no <n no <r> — vS 



fOI~~ NO — 



— <N — ' 



<~o r» ON <N NO 00 



8« 8« 

ON ON 

en *t no m Tt no 



oo >o O oo nn o 



Tf m >/"> -<t tn </S 



■<t «N iri tj- <N iri 



tj- (N >o m — ro 



NO NO 

r- (N O O NO 



m o) ^ en <N ^t 



q q 

MMr<i CS cN en 





•«• a a J3 -2 .59 b 5 s" 






Oh 
E 


alia 

e cowan 
e taiva 
e melan* 
e longic 
e parvul 
e sorico 
e gracili 
e sp. A 
e gymnc 
tosus 
omys ra 
omys ko< 


EM 


E"3"Q"3"a"3"Q"o'a"3 2. 3 "3 




8 




'u 


RCls.kk.l>,ls-.kk<£o7 


u 


c 


§ §§ § § § j| i§ § 3 £ § 



Oh U 

cS O- 
O to 



s 5 



fe W 8 



-J <u 
-O X) 

E E 



5 c g c a c 

S- a- 

t. S p, s a f» 

H H U P H 



O. <j 

50 SI 



222 



FIELDIANA: ZOOLOGY 



•?S 7 



8 
o 

CD 
Q. 
V) 8 

O 
i- 5 

CD 

.O 

E 4 

C 

Q 3 

> 

!• 

E 

3 1 





















a 




A 








■ 


720m 
810 m 
1210 m 
1625 m 












■ 








W 


- 














■ 




i 



33 



66 99 132 165 

trap nights 



198 



231 



264 



297 




trap nights 

Fig. 20-1. Species accumulation curves (top) and pitfall trap success (bottom) plotted for each elevational zone 
against the total number of trap (bucket) days. Information from the three lines at each zone is combined. 



d'Ambatovaky southward into southern Madagas- 
car at Andrahomana (MacPhee, 1987; Jenkins, 
1992). Nicoll and Langrand (1989) reported Hem- 
icentetes (semispinosus) nigriceps and Tenrec 
ecaudatus from the RNI d'Andringitra. The latter 
species was observed several times at the 720 and 
810 m zones but was not captured. During the 
1993 survey we did not find evidence of Hemi- 
centetes in the reserve, although this species has 
been captured using the same methodology else- 
where in the eastern humid forest (Raxworthy, un- 
publ.). We did not search for Hemicentetes bur- 
rows; this is apparently an effective means of lo- 
cating this genus (Eisenberg & Gould, 1970). 

On the basis of the species accumulation curves 
for each elevational zone, and extrapolation of 



what should presumably occur in the reserve and 
what was found, we conclude that the Insectivora 
fauna, particularly species of Microgale, on the 
eastern slopes of the RNI d'Andringitra has been 
thoroughly, although not completely, surveyed. 



Trapping Success and Abundance 

In Table 20-3 the results of the pitfall trapping 
are presented by species, pitfall line, and eleva- 
tion. There was considerable variation in capture 
rates of Microgale between lines at the same el- 
evation and at different zones on the mountain: 
6.0% (range 3.0-10.6%) at 720 m, 4.8% (range 
3.9-5.2%) at 810 m, 15.2% (range 10.4-18.2%) 



GOODMAN ET AL.: INSECTIVORE ECOLOGY 



223 



at 1210 m, and 12.6% (range 5.7-21.6%) at 1625 
m. Using standard small mammal traps, insecti- 
vores were captured only at 1210 (N = 10) and 
1625 m (N = 1). Because the pitfall lines were 
randomly placed in the microhabitats within each 
elevation zone, the same protocol was used for 
each line, species accumulation curves reached a 
plateau, and there is no evidence of preferential 
capture of any species, we feel that it is reason- 
able to equate the capture rates of Microgale spe- 
cies with their relative abundance in each zone 
during the period of the survey. However, climatic 
conditions, particularly rainfall patterns, may have 
affected trap success between the zones. Further- 
more, animal activity may have changed over the 
course of the 2-month survey. 

The only other published study of Microgale 
using large pitfall buckets in humid forest was by 
Raxworthy and Nussbaum (1994) in the Pare Na- 
tional (PN) de la Montagne d'Ambre, in the ex- 
treme north. The pitfall system they used was 
identical to that employed in the RNI d'Andrin- 
gitra, but the lines were in place for between 1 1 
and 16 days. In some cases they had remarkable 
capture rates, with up to 61 Microgale individu- 
als, representing four species, in a single line dur- 
ing 121 accrued pitfall bucket-days (50.4%). The 
use of an identical pitfall system, however, in 
transitional dry deciduous/sub-arid thorn scrub 
forest yielded a different result (Goodman & Ganz- 
horn, 1994; Raxworthy et al., 1994). In 528 ac- 
crued pitfall bucket-days a total of 1 1 individuals 
(2.1%), representing three species (Geogale au- 
rita Milne Edwards & Grandidier, 1872; Setifer 
setosus; and Suncus madagascariensis), were cap- 
tured. The species composition in such a forest is 
not comparable, therefore, to that in humid forest. 

Stephenson conducted small mammal invento- 
ries using Sherman traps and randomly placed pit- 
fall traps (plastic plant propagators), measuring 20 
cm in depth and 12 cm in diameter, without a drift 
fence. In the RS d'Analamazaotra, Stephenson 
(1993) captured seven species of Insectivora, in- 
cluding four Microgale spp. Except for single in- 
dividuals of Oryzorictes hova A. Grandidier, 
1870, and Microgale cowani Thomas, 1882, that 
were trapped during 811 pitfall bucket-days 
(0.2%), all other individuals and species were ob- 
tained in Sherman traps. In the RS d'Ambohitan- 
tely (Stephenson et al., 1994), 345 pitfall bucket- 
days were accrued and no Microgale sp. was cap- 
tured, although four species are known to occur 
in the reserve (Raxworthy, unpubl.). Thus, it 
would appear that the pitfall technique used by 



Stephenson (1993) and Stephenson et al. (1994) 
is less effective than the much larger buckets used 
with a drift fence. On the basis of our results from 
the RNI d'Andringitra, large pitfall buckets are 
much more effective in capturing certain species 
than are conventional traps. 



The Relationship of Abiotic 
Factors and Capture Rates 

Information on daily minimum and maximum 
temperatures and precipitation was recorded at 
each site (Chapter 3). In order to determine if Mi- 
crogale capture rates were influenced by these cli- 
matic variables, a series of nonparametric tests 
were run for each elevational zone using Ken- 
dall's rank correlation, with the Y variable being 
the number of animals captured and the X vari- 
ables as minimum temperature the morning the 
pitfall buckets were checked, the maximum tem- 
perature of the previous afternoon, and the 
amount of precipitation during the previous 
24-hour and 48-hour period. The only significant 
result was at 810 m, between the number of ani- 
mals captured and the amount of rain in the pre- 
vious 24-hour period (r = 0.714, N = 7, P = 
0.05). 

In order to allow us to discern general patterns 
of the relationship between these abiotic variables 
and the number of animals captured, a series of 
regression analyses were conducted. In virtually 
all cases there was a positive relationship between 
these variables and the number of Microgale cap- 
tured. For the 720, 1210, and 1625 m zones, how- 
ever, a few points were outliers of the general pat- 
tern, and the relationship was not statistically sig- 
nificant. At 810 m, all of the climatological vari- 
ables were positively and significantly correlated 
with the number of animals captured: minimum 
temperature (r 2 = 0.98), maximum temperature (r 2 
= 0.99), 24-hour rain (r 2 = 0.97), and 48-hour 
rain (r 2 = 0.99). 

If we assume that the general activity rate of 
Microgale species is correlated with pitfall trap 
capture rate, then it appears that during periods of 
higher minimum and maximum temperature, as 
well as periods of precipitation, there are in- 
creased movements of these animals. For temper- 
ature, this relationship is presumably related to 
metabolic rate, whereas for precipitation it may 
be a function of increased moisture and flooding 
of underground burrow systems. 



224 



FIELDIANA: ZOOLOGY 



Species Diversity 

There is elevational variation in species diver- 
sity of Insectivora on the eastern slopes of the 
RNI d'Andringitra. Eleven Insectivora, all of 
which are endemic to the island, were captured in 
the reserve during the 1993 mission using stan- 
dard small mammal traps and pitfall buckets. No 
introduced species (i.e., Suncus murinus Lin- 
neaus, 1766) was recorded. The number of Insec- 
tivora known in each elevational level was: 720 
m (lines 1-3) — five species, 810 m (lines 4-6) — 
six species, 1210 m (lines 7-9) — eight species, 
and 1625 m (lines 10-12) — six species (Table 
20-1); for Microgale the species composition was 
four species at 720 m, four species at 810 m, eight 
species at 1210 m, and six species at 1625 m. 
Thus, in general for Insectivora and specifically 
for Microgale, there appears to have been a mid- 
elevational peak in species richness at 1210 m. 

Two species, Microgale taiva Major, 1896b, 
and the very small M. parvula G. Grandidier, 
1934, occur at all elevations, and it is probable 
that M. longicaudata Thomas, 1882 is also ubiq- 
uitous but not collected at the intermediate sites. 
The only species apparently confined to lowland 
forest in and around the reserve is M. melanor- 
rhachis Morrison-Scott, 1948. In contrast, five 
species, M. dobsoni, M. soricoides Jenkins, 1993, 
M. gracilis Major 1896a, M. gymnorhyncha, new 
species (Chapter 19), and Microgale sp. A (Chap- 
ter 19; Jenkins et al., unpubl.), were found only 
in montane forest. Microgale cowani may also be 
more common in montane than lowland regions; 
in the RNI d'Andringitra it was collected at var- 
ious localities from 1200 to 1800 m, but it was 
found only rarely in lowland forest at 810 m. 

Those species that were captured at all alti- 
tudes, namely M. taiva and M. parvula, deviate 
from the general pattern of the highest abundance 
at 1210 m; for both, the greatest number of indi- 
viduals were collected at 1625 m, slightly higher 
than the number collected at 1210 m, but mark- 
edly greater than either of the lowland localities. 
M. cowani follows the general pattern, with most 
individuals being trapped at 1210 m, next at 1625 
m, fewest at 810 m, and none at 720 m. 



tive forms. Microgale longicaudata, with its very 
long tail, which has been hypothesized as being 
prehensile (Thomas, 1918; MacPhee, 1987), elon- 
gated hind foot with lengthened fifth digit, and 
elongated cheiridia, was assumed by Eisenberg 
and Gould (1970) possibly to climb and ricochet 
among branches. When handled by S.M.G. and 
C.J.R., one M. longicaudata captured in a pitfall 
trap during the 1993 survey wrapped the tip of its 
tail around a finger and was able to hang sus- 
pended for about 20 seconds; thus, at least the tip 
of the tail is prehensile. Although no M. longi- 
caudata was obtained in Sherman traps placed off 
the ground, it is clear that this species is at least 
scansorial, if not occasionally arboreal. 

Microgale sp. A shows similar but less extreme 
scansorial adaptations, with a slightly elongated 
tail, hind feet, fifth hind digit, and cheiridia. Mi- 
crogale longicaudata and M. sp. A have not been 
found in the same elevational zones, but they may 
well occur in direct sympatry; they differ in body 
size as well as degree of morphological adaptation 
and may be assumed to occupy slightly different 
ecological niches. Microgale gracilis and M. gym- 
norhyncha both show morphological features that 
suggest a semi-fossorial habit, such as the broad- 
ened forefeet with enlarged claws, and the re- 
duced size of the eyes and ears; both are found in 
direct sympatry in the reserve. 

The remaining six species have a fairly gener- 
alized body form, although M. parvula by virtue 
of its very small size, M. dobsoni because of its 
large size, and M. soricoides by its unusual den- 
tition may be assumed to occupy different eco- 
logical niches. Indeed, the capture of one speci- 
men of M. soricoides 2 m above the ground implies 
some climbing ability. The real problems remain 
with M. cowani, M. taiva, and M. melanorrhachis, 
all of which show only slight morphological vari- 
ation, no obvious adaptive specializations, and yet 
are clearly distinct species (Chapter 19). Micro- 
gale cowani and M. taiva, and M. taiva and M. 
melanorrhachis, occur in direct sympatry in the 
RNI d'Andringitra. 



Multiple Captures in a Single 
Pitfall Bucket 



Ecological Separation 

Some of the Microgale species occurring in 
sympatry along the eastern slopes of the RNI 
d'Andringitra may be divided into distinct adap- 



Multiple captures include different animals 
taken in a pitfall bucket during the time that a line 
was in place and multiple animals found in a pit- 
fall bucket on a single occasion. In the latter case, 
referred to below as "simultaneous" captures, the 



GOODMAN ET AL.: INSECTIVORE ECOLOGY 



225 



results might be confounded because larger Mi- 
crogale species will readily eat smaller species 
when confined to the same bucket. In both cases, 
these multiple captures reflect animals moving 
across the same swath of ground 20 m or less in 
width. We interpret such movement as direct syn- 
topy in the utilization of areas of the forest. Fur- 
thermore, this information allows some inference 
to be made into the social organization of the var- 
ious species. 

Multiple Captures During the Tenure of 
Line Placement — At 720 m there were only two 
cases of multiple captures, both in line 3 in a val- 
ley bottom. Two M. taiva were taken in the same 
bucket, as were a M. taiva and two M. longicau- 
data in another bucket. Multiple captures at 810 
m included two M. taiva; a M. cowani and M. 
taiva; and a M. parvula and M. cowani in three 
different buckets. At 1210 and 1625 m the num- 
ber of multiple captures in a single bucket in- 
creased substantially. In numerous cases single 
buckets in a pitfall line collected three to five 
small mammals during the tenure of the line place- 
ment. Exceptional cases included five M. longicau- 
data taken in a single bucket over 3 days; two M. 
cowani, two M. taiva, and a M. longicaudata in 2 
days; and a M. cowani, two M. taiva, and an un- 
described genus and species of rodent (see Chapter 
21) in 2 days. This demonstrates that in some cases 
species that are morphologically similar (e.g., M. 
cowani and M. taiva) occur in microsympatry. Oth- 
ers, such as M. gracilis, M. gymnorhyncha, and M. 
soricoides, however, were never taken in any pair- 
wise combination in the same pitfall bucket. Fur- 
thermore, M. melanorrhachis was not captured in 
the same pitfall bucket with M. cowani or M. taiva. 
In several of these cases, the numbers of animals 
captured for a given species were few, and these 
patterns may simply be chance. 

Simultaneous Multiple Captures — A total of 
902 bucket-days accrued during the survey. Of 
these days there were 823 during which no Mi- 
crogale was captured, 71 bucket-days when single 
individuals were captured, 6 bucket-days when 
two individuals were captured, and 2 bucket-days 
during which three individuals were captured. 
These observed frequencies do not deviate from 
the expected frequencies based on a Poisson dis- 
tribution. 

The only case at 720 m of two animals in the 
same pitfall bucket was a pair of M. longicaudata 
(adult female and juvenile female). No simulta- 
neous multiple capture was recorded at 810 m. At 
1210 m, three M. taiva were found in a bucket 



the first morning it was installed; these included 
two juvenile males and a juvenile female. Also at 
this elevation there was one case of a M. cowani 
and a M. gymnorhyncha in the same bucket. At 
1625 m, a single bucket contained two M. longi- 
caudata on two occasions: on December 10, 
1993, an adult female and juvenile were caught, 
and on December 12, 1993, two adult males were 
caught (On December 11, 1993, this bucket yield- 
ed a juvenile M. longicaudata). The next bucket 
in the same line yielded an adult female M. cow- 
ani and an adult male M. taiva on December 9, 
1993, and on December 10, 1993, a juvenile male 
M. cowani, an adult male M. taiva, and an adult 
male M. longicaudata were captured. Thus, over 
the course of four consecutive days, 10 individual 
Microgale of three species were captured in a sec- 
tion of a pitfall line no more than 30 m long. The 
last example was two juvenile male M. taiva and 
an undescribed genus and species of rodent cap- 
tured on the same night in a single bucket. 

Little is known about the breeding systems or 
social organization of Microgale species. Whether 
the three juvenile M. taiva captured on the same 
night were moving together or were attracted to the 
bucket by movements or calls of captured individ- 
uals is unknown. Female M. taiva have also been 
reported with two embryos and up to six mammae, 
and thus it is possible that the young M. taiva were 
siblings. In the eight cases of "simultaneous" mul- 
tiple captures in the same bucket, there was not a 
single example of a conspecific adult male and fe- 
male, and there was only one case of conspecific 
adult males being trapped together. 



Relationship Between Macrofauna 
Soil Invertebrates and the 
Elevational Distribution of Microgale 

Five soil plots, each 2 (length) X 2 (width) X 
0.1 (depth) m and separated by 20 m, were sam- 
pled during daylight hours along each pitfall line 
to document the variation in density and species 
richness of macrofauna soil invertebrates (Chapter 
14). Because the vast majority of the tenrecs in 
the RNI d'Andringitra are terrestrial and presum- 
ably feed extensively on soil invertebrates, the re- 
sults of this analysis have potential implications 
for understanding the distribution of insectivores 
along the eastern slopes of the massif. Both Mi- 
crogale gracilis and M. gymnorhyncha have well- 
developed claws, and these species, as well as 
others, presumably dig into the ground. 



226 



FIELDIANA: ZOOLOGY 



Table 20-4. Variation in macrofauna soil invertebrates and Microgale spp. between different pitfall lines. 







720 m 






810 m 




1210 m 




1625 m 


Pitfall line: 


1 


2 


3 


4 


5 


6 


7 8 


9 


10 11 12 


Placement: 


R 


S 


V 


R 


S 


V 


R V 


S 


V R S 


Invertebrate taxonomic diversity 
















Average per 


8.4 


4.4 


5.6 


10.6 


8.6 


2.6 


8.8 7.2 


8.0 


7.0 6.8 8.8 


line 


(5-12) (0-10) (3-11) 


(7-14) (4-12) 


(0-7) 


(4-13) (2-15) (4-12) 


(3-10) (4-10) (5-16) 




2.97 


4.39 


3.21 


2.51 


2.97 


2.70 


3.70 5.06 


3.31 


2.92 2.39 4.55 


Average per 




6.1 






7.3 




8.7 




7.5 


line per zone 




(0-12) 
3.74 






(1-14) 
4.33 




(2-15) 
3.98 




(3-16) 
3.29 


Total per 




















zone 




46 






68 




76 




46 


Invertebrate density 




















Average individuals 


13.6 


5.0 


8.6 


14.2 


9.4 


4.8 


11.6 16.8 


16.0 


16.8 10.4 24.8 


per line 


(6-22) (0-12) (3-16) 


(8-24) (4-13) (0-14) 


(5-20) (2-35) (7-34) 


(3-51) (5-14) (5-43) 




6.35 


5.09 


6.77 


6.18 


3.51 


5.45 


6.27 13.08 


11.90 


19.63 4.09 17.53 


Average per line per 




9.1 






9.5 




14.9 




16.7 


zone 




(0-22) 
6.73 






(0-24) 
6.22 




(2-35) 
10.32 




(3-51) 
14.87 


Microgale spp. 




















% captured per line 


3.0 


4.5 


10.6 


3.9 


5.2 


5.2 


10.4 16.9 


18.2 


21.6 5.7 10.4 


% captured per ele- 




















vational zone 




6.0 






4.8 




15.2 




12.6 


Species per 




















line 


2 


2 


2 


1 


2 


3 


5 4 


6 


5 1 4 


Species per eleva- 




















tional zone 




4 






4 




8 




6 



R = ridge, S = slope, V = valley. Averages are presented as mean, range (minimum-maximum), and standard 
deviation. 



In Table 20-4 we present a summary of the tax- 
onomic diversity and number of individual mac- 
rofauna soil invertebrates along each line and at 
each elevational zone. The taxonomic diversity of 
invertebrates sampled within each plot per pitfall 
line showed considerable variation in the mini- 
mum and maximum values and relatively high 
standard deviations. However, average diversity 
between elevational zones varied from a low of 
6.1 morphospecies on average per plot at 720 m 
to a high of 8.7 morphospecies per plot at 1210 
m. The average total number of individual inver- 
tebrates found per plot in each elevational zone 
(N = 15) increased with altitude (r 2 = 0.95, P = 
0.03), from an average total of 9.1 (range 5.0- 
13.6) at 720 m, 9.5 (range 4.8-14.2) at 810 m, 
14.9 (range 1 1.6-16.8) at 1210 m, to 16.7 (range 
10.4-24.8) at 1625 m. No information is available 
on invertebrate biomass in the plots. 

The general trends of invertebrate taxonomic 
diversity and Microgale species density and abun- 
dance (as measured by pitfall trap success) par- 
allel one another relative to elevation, the peak 



for both groups being 1210 m. The density of in- 
vertebrates, however, does not peak at 1210 m, 
but rather shows a linear increase as a function of 
elevation. Because we know little about the spe- 
cific food habits of these various species of Mi- 
crogale, it is impossible to interpret on a fine scale 
what types of invertebrates or specific taxa they 
might consume, although it appears that the den- 
sity and distribution of shrew tenrecs on the 
slopes of the Andringitra Massif is related more 
directly to macrofauna soil invertebrate diversity 
than to density. One species, M. melanorrhachis, 
is only found at low elevations, and its elevational 
distribution does not parallel this general pattern. 



Microhabitat Segregation Between 
Species and Its Relationship to 
Macrofauna Soil Invertebrates 

At each elevational zone, pitfall traps were set 
in three distinct positions: valley bottom, slope, 
and ridge crest. The trapping results for each type 



GOODMAN ET AL.: INSECT! VORE ECOLOGY 



227 



Table 20-5. Variation in macrofauna soil invertebrates and Microgale spp. in relationship to microhabitat and 
elevation. 



Ridge 



Slope 



Valley 



Parameter measured 



720 810 1210 1625 720 810 1210 1625 720 810 1210 1625 

in m in in m in m m in in in m 

Invertebrate taxonomic diversity 

Average number of species 

per line 8.4 10.6 8.8 6.8 4.4 8.6 8.0 8.8 5.6 2.6 7.2 7.0 

Average total number of species 

7.5 5.5 



per microhabitat 




8.7 


Invertebrate density 






Average number 
per line 


13.6 


14.2 11 


Average number per line per 
microhabitat 




12.5 


Microgale spp. 






% capture per 
line 


3.0 


3.9 10 


Average % capture per micro- 
habitat 




5.8 


Species per line 


2 


1 5 


Species per habitat 




8 


Total individuals per 






habitat 




18 



13.6 14.2 11.6 10.4 5.0 9.4 16.0 24.8 8.6 4.8 16.8 16.? 

13.8 11.8 



3.0 3.9 10.4 5.7 4.5 5.2 18.2 10.4 10.6 5.2 16.9 21.6 



9.6 13.6 

1 2264 2345 

6 7 



27 



43 



of microhabitat are given in Tables 20-4 and 20-5. 
At both lowland forest localities (720 and 810 m), 
few specimens of Insectivora were captured on 
slopes and on ridges, whereas a considerable por- 
tion of the individuals in the 720 m zone were 
collected in valley bottoms; at 810 m, capture 
rates in valley bottoms were similar to those of 
slope and ridge. At 720 m, two species of Micro- 
gale were obtained in each of these microhabitats, 
whereas at 810 m, the number of species captured 
varied, from three species in the valley to two on 
the slope and one on the ridge (Table 20-3). The 
pattern was slightly different in montane forest at 
1210 m, with nearly equal numbers of individual 
Microgale captured in the valley and slope lines 
(13 and 12, respectively; Table 20-3) and only 
eight individuals taken in the ridge line. At this 
altitude, six species of Microgale were obtained 
in the slope line, five in the ridge line, and four in 
the valley line. The capture success rate of Micro- 
gale in the 1625 m upper montane/sclerophyllous 
zone showed considerable variation between mi- 
crohabitats. The capture rate in the valley was 
nearly four times greater than that in the ridge line 
and two times greater than that in the slope line. 
Taxonomic diversity at this elevation in micro- 
habitats paralleled overall capture success. 



There is no convincing evidence from this 
study to suggest that any species occurs prefer- 
entially in one of the three habitats. Microgale 
soricoides was found only on a slope and M. sp. 
A only on a ridge, whereas M. gymnorhyncha and 
M. dobsoni both occurred on slope and ridge for- 
est. The sample size for all four species, however, 
is too small for conclusions to be drawn about 
habitat preference. There is slight evidence that 
M. cowani occurs preferentially in slope forest, as 
may M. taiva, at least at high elevations. 

The combined data for the three habitat types 
are given in Table 20-5. The highest capture rate 
was in valley bottoms (N = 43), followed by 
slopes (N = 27) and then ridges (N = 18), where- 
as the highest species richness was on the ridge 
(eight species), then the valley (seven species) and 
then slope (six species). 

When the macrofauna soil invertebrate samples 
were segregated by microhabitat, the ridge lines 
had the highest taxonomic diversity per plot (8.7 
species, range 6.8-10.6), followed by slope (7.5 
species, range 4.4-8.8), and valley (5.7 species, 
range 2.6-7.2). The average number of individual 
invertebrates recovered per plot was remarkably 
constant between microhabitats, although the val- 
ley and slope sites showed considerable differ- 



228 



FIELDIANA: ZOOLOGY 



ences in the range: ridge, 12.5 individuals (range 
10.4-14.2); slope, 13.8 individuals (range 5.0- 
24.8); and valley, 11.8 individuals (range 8.6- 
16.8). In general there seems to be little correla- 
tion in the average values per habitat type be- 
tween the taxonomic diversity and density of 
macrofauna soil invertebrates and density and 
specificity of Microgale. When examined on a fin- 
er scale, there appears to be some relationship be- 
tween these factors. For example, in slope sites 
the highest invertebrate density was at 1210 (16.0 
individuals/plot) and 1625 m (24.8 individu- 
als/plot). These two lines also had the highest per- 
centage capture rate (18.2% and 10.4%, respec- 
tively) and species diversity (six and four, respec- 
tively) of the four lines in this habitat. An almost 
parallel situation occurred in the valley lines at 
1210 and 1625 m. However, in the ridge sets this 
pattern did not exist; at 810 m, the elevation with 
the highest taxonomic diversity and density of soil 
invertebrates, only one species of Microgale was 
collected, and the overall trap success was nearly 
the lowest for this microhabitat. Thus, in short, 
the measurements of soil invertebrates along the 
pitfall lines in three different types of microhab- 
itats were not closely correlated with density or 
Microgale specificity. 

Differences Between Disturbed 
and Undisturbed Lowland Forest 

The 720 m zone was classified as disturbed for- 
est and the 810 m zone as relatively pristine for- 
est. A comparison of the Microgale species rich- 
ness (five vs. four species, respectively) and den- 
sity (seven animals in 198 pitfall bucket-days [= 
3.5%] vs. 12 animals in 231 pitfall bucket-days 
[= 5.2%], respectively) indicates that there is no 
clear difference in the Microgale species com- 
munities between these localities. 

In the RS d'Analamazaotra, Stephenson (1993) 
found that in some cases differences in the level 
of human activity between sites at the same ele- 
vation and in approximately the same contiguous 
forest were related to species diversity. The great- 
est species diversity of Microgale was in the site 
with no or at least little human disturbance. How- 
ever, two of the four species captured at this site 
were exceptionally rare, and their absence from 
the other sites may be a question of chance sam- 
pling rather than actual differences in species di- 
versity. Furthermore, several forest-dwelling Mi- 
crogale do not seem particularly sensitive to gen- 



eral patterns of habitat disturbance (MacPhee, 
1987; Goodman et al., in press). 



Elevational Variation of Species 
at Other Sites 

Microgale melanorrhachis may be a lowland 
forest species in the RNI d'Andringitra, but it 
seems to be more wide ranging in elevation else- 
where. Higher elevational sites include the type 
locality of Andasibe at 3000 ft (= 920 m) (Mor- 
rison-Scott, 1948) and Didy at 1000 m (Eisenberg 
& Gould, 1970). Positively identified specimens 
have also been collected from valley bottom for- 
est at Mantady (1100-1200 m) (Rax worthy, un- 
publ.) and Montagne d'Ambre (1100-1150 m) 
(Raxworthy & Nussbaum, 1994). At the latter lo- 
cality, M. melanorrhachis was common in valley 
bottoms, on ridges, and on slopes between 1125 
and 1 250 m, but in a comparable number of buck- 
et trap-days it was not recorded along the same 
slope between 650 and 670 m. Microgale melan- 
orrhachis has also been collected at 5000 ft (= 
1525 m) at Pic dTvohibe (Morrison-Scott, 1948), 
just south of the RNI d'Andringitra. Limited in- 
formation in the literature suggests that M. grac- 
ilis may be more common in montane regions; it 
has been collected at Ambohimitombo (1200 m), 
Ankeramadinika (1400 m), and Anjavidilava, RNI 
d'Andringitra (1800-2100 m). The supposed low 
elevational record of this species at Fanovana, Be- 
forona (500-1000 m) (MacPhee, 1987), actually 
refers to a specimen of M. gymnorhyncha, which 
in the RNI d'Andringitra appears to be contrast- 
ingly confined to montane forest. Microgale cow- 
ani sensu stricto may also be more common in 
montane regions. The type locality is Ankafina 
(summit at 1570 m, currently most forest confined 
to 1200-1500 m), and at RNI d'Andringitra it was 
collected at various localities from 1200 to 1800 
m, but only rarely in lowland forest at 810 m. 



Acknowledgments 

We are grateful to the officials of DEF and AN- 
GAP for permission to conduct this survey. Crit- 
ical comments were provided on an earlier ver- 
sion of this manuscript by B. D. Patterson and an 
anonymous reviewer. The participation of C.J.R. 
in this project was partially supported by NSF 
grant DEB 90-24505. 



GOODMAN ET AL.: INSECTIVORE ECOLOGY 



229 






Literature Cited 

Coquerel, C. 1848. Note sur une espece nouvelle de 
musaraigne trouv^e a Madagascar. Annales des Sci- 
ences Naturelles, Zoologie, 9: 193-198. 

Eisenberg, J. E, and E. Gould. 1970. The tenrecs: A 
study in mammalian behavior and evolution. Smith- 
sonian Contributions to Zoology, 27: 1-138. 

Glaw, E, and M. Vences. 1994. A Fieldguide to the 
Amphibians and Reptiles of Madagascar, 2nd edition. 
Zoologisches Forschungsinstitut und Museum Alex- 
ander Koenig, Bonn, 480 pp. 

Goodman, S. M., and J. Ganzhorn. 1994. Les petits 
mammiferes, pp. 58-63. In Goodman, S. M., and O. 
Langrand, eds., Inventaire biologique foret de Zom- 
bitse. Recherches pour le Developpement, Serie Sci- 
ences biologiques, No. Special, Centre dTnformation 
et de Documentation Scientifique et Technique, An- 
tananarivo, 106 pp. 

Goodman, S. M., R D. Jenkins, and O. Langrand. In 
press. Exceptional records of Microgale species (In- 
sectivora: Tenrecidae) in vertebrate food remains. 
Bonner Zoologische Beitrage. 

Grandidier, A. 1870. Description de quelques animaux 
nouveaux, decouverts a Madagascar, en novembre 
1869. Revue et Magasin de Zoologie, 22: 49-54. 

Grandidier, G. 1934. Deux nouveaux mammiferes in- 
sectivores de Madagascar. Bulletin du Museum Na- 
tional d'Histoire Naturelle, ser. 2, 6: 474-477. 

Jenkins, P. D. 1992. Description of a new species of 
Microgale (Insectivora: Tenrecidae) from eastern 
Madagascar. Bulletin of the British Museum Natural 
History (Zoology), 58: 53-59. 

. 1993. A new species of Microgale (Insectiv- 
ora: Tenrecidae) from Eastern Madagascar with an un- 
usual dentition. American Museum Novitates, 3067: 
1-11. 

Linnaeus, C. 1766. Systema naturae per regna tria na- 
turae secundum classes, ordines, genera, species, cum 
characteribus, differentiis, synonymis, locis. 12th edi- 
tion. Holmiae. 

MacPhee, R. D. E. 1987. The shrew tenrecs of Mada- 
gascar: Systematic revision and Holocene distribution 
of Microgale (Tenrecidae, Insectivora). American Mu- 
seum Novitates, 2889: 1-45. 

Major, C. I. Forsyth. 1896a. Diagnoses of new mam- 
mals from Madagascar. Annals and Magazine of Nat- 
ural History, (6) 18: 318-325. 

. 1896b. Descriptions of four additional new 

mammals from Madagascar. Annals and Magazine of 
Natural History, (6) 18: 461-463. 

Milne Edwards, A., and A. Grandidier. 1872. De- 
scription d'un nouveau mammifere insectivore de 
Madagascar (Geogale aurita). Annales des Sciences 
Naturelles, Zoologie, 15, Article 19: 1-5. 

. 1882. Description d'une nouvelle espece 

d' insectivore de Madagascar. Le Naturaliste, 2: 55. 

Morrison-Scott, T. C. S. 1948. The insectivorous gen- 
era Microgale and Nesogale (Madagascar). Proceed- 
ings of the Zoological Society of London, 118: 817— 
822. 

Nicoll, M. E. 1982. Reproductive ecology of Tenrec 



ecaudatus (Insectivora: Tenrecidae) in the Seychelles. 
Ph.D. thesis, University of Aberdeen. 

-, and O. Langrand. 1989. Madagascar: Revue 



de la conservation et des aires protegees. World Wide 
Fund for Nature, Gland, Switzerland, XVII+374 pp. 

Nicoll, M. E., and G. B. Rathbun. 1990. African In- 
sectivora and Elephant- shrews: an action plan for their 
conservation. IUCN/SSC Insectivore, Tree-shrew and 
Elephant-shrew Specialist Group, Gland, Switzerland, 
iv+53 pp. 

Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. RCP 225. Etudes des eco- 
systemes montagnards dans le region malgache. I. Le 
massif de l'Andringitra. 1970-1971. Geomorphologie, 
climatologie et groupements vegetaux. Bulletin de la 
Societe d'Ecologie, 11(2-3): 198-226. 

Raxworthy, C. J., and R. A. Nussbaum. 1994. A rain- 
forest survey of amphibians, reptiles and small mam- 
mals at Montagne d'Ambre, Madagascar. Biological 
Conservation, 69: 65-73. 

Raxworthy, C. J., J.-B. Ramanamanjato, and A. Ra- 
selimanana. 1994. Les reptiles et les amphibians, pp. 
41-57. In Goodman, S. M., and O. Langrand, eds., 
Inventaire biologique foret de Zombitse. Recherches 
pour le Developpement, Serie Sciences biologiques, 
No. Special, Centre dTnformation et de Documenta- 
tion Scientifique et Technique, Antananarivo, 106 pp. 

Schreber, J. C. D. von. 1777. Die Saiigethiere in Ab- 
bildungen nach der Natur mit Beschreibungen. Vol- 
ume 3. Leipzig. 

Stephenson, P. J. 1991. Reproductive energetics of the 
Tenrecidae (Mammalia: Insectivora). Ph.D. thesis, 
University of Aberdeen. 

. 1993. The small mammal fauna of Reserve 

Speciale d'Analamazaotra, Madagascar: The effects of 
human disturbance on endemic species diversity. Bio- 
diversity and Conservation, 2: 603-615. 

-. 1994. Seasonality effects on small mammal 



trap success in Madagascar. Journal of Tropical Ecol- 
ogy, 10: 439-444. 

-, and P. A. Racey. 1993. Reproductive energet- 



ics of the Tenrecidae (Mammalia: Insectivora). II. The 
shrew-tenrecs, Microgale spp. Physiological Zoology, 
66: 664-685. 

Stephenson, P. J., H. Randriamahazo, N. Rakotoari- 
son, and P. A. Racey. 1994. Conservation of mam- 
malian species diversity in Ambohitantely Special Re- 
serve, Madagascar. Biological Conservation, 69: 213— 
218. 

Thomas, [M. R.] Oldfield. 1882. Description of a new 
genus and two new species of Insectivora from Mad- 
agascar. Journal of the Linnean Society (Zoology), 16: 
319-322. 

. 1884. Description of a new species of Micro- 
gale. Annals and Magazine of Natural History, (5) 14: 
337-338. 



. 1918. On the arrangement of the small Ten- 
recidae hitherto referred to Oryzorictes and Microga- 
le. Annals and Magazine of Natural History, (9) 1: 
302-307. 



230 



FIELDIANA: ZOOLOGY 



Chapter 21 

Systematic Studies of Madagascar's Endemic Rodents 
(Muroidea: Nesomyinae): A New Genus and Species 
from the Central Highlands 

Michael D. Carleton and Steven M. Goodman 



Abstract 

We report a new genus and species of nesomyine rodent from Madagascar's Central High- 
lands, type locality Antananarivo Province, Manjakatompo, at about 1800 m elevation on the 
Ankaratra Massif. The form also occurs in upper montane forest within the Reserve Naturelle 
Integrate d'Andringitra at 1625 m. The new genus is specifically contrasted to species of 
Macrotarsomys, a taxon that occurs in western Madagascar and is hypothesized to be its closest 
relative among the seven previously known genera of Nesomyinae. The many generalized 
features of the new genus provide a basis for renewed examination of the Miocene cricetodontid 
Protarsomys (Kenya, Rusinga Faunal Assemblage); we discount its proposed synonymy with 
Macrotarsomys and question its progenitive significance in regard to the origin of Nesomyinae. 
The new genus has disjunct populations on the Ankaratra and Andringitra massifs, a distribution 
that resembles those of numerous other Malagasy organisms restricted to upper montane- 
sclerophyllous vegetation. This geographical pattern conforms to Humbert's proposed High 
Mountain Domain. 

Resume 

Nous rapportons un nouveau genre et une nouvelle espece de rongeur nesomyine du haut 
plateau central de Madagascar, locality du specimen type Province d' Antananarivo, Manjaka- 
tompo, vers 1800 m dans le massif de 1' Ankaratra. Dans la Reserve Naturelle Integrate 
d'Andringitra, cette forme apparatt aussi dans la foret de montagne a une altitude sup6rieure a 
1625 m. Le nouveau genre se d6marque sp6cifiquement des especes de Macrotarsomys, un 
taxon qui est distribu6 a l'ouest de Madagascar et qui est consider comme le plus proche 
parmi les sept genres de Nesomyinae connus jusqu'a maintenant. Les nombreuses caracteYis- 
tiques g6n£rales de ce nouveau genre fournit une base d'examen pour r6examiner Protarsomys 
(Kenya, Assemblage faunistique de Rusinga) cricetodontid du Miocene; nous ne tenons pas 
compte de la synonymie proposed avec Macrotarsomys et mettons en doute sa valeur compte 
tenu de l'origine des Nesomyinae. Le nouveau genre pr£sente des populations isotees dans les 
massifs de 1'Ankaratra et de 1'Andringitra, une distribution qui ressemble a celle d'autres or- 
ganismes a la distribution limited a la v6g6tation scterophylle des hautes montagnes. Cette 
distribution g6ographique est conforme a celle proposee par Humbert pour le Domaine des 
Hautes Montagnes. 



CARLETON & GOODMAN: ENDEMIC RODENTS 231 



Introduction 

The Mission Zoologique Franco-Anglo-Amer- 
icaine (MZFAA) was a cooperative biological sur- 
vey of Madagascar conducted from April 1929 to 
May 1931 (Rand, 1932, 1936). The expedition 
concentrated its collecting activities on the is- 
land's avifauna and, to a lesser extent, on larger 
mammals such as lemurs (Archbold, 1932; Rand, 
1935), and all specimens were approximately 
evenly divided and deposited among the three 
participating museums — the Museum National 
d'Histoire Naturelle in Paris (MNHN), the British 
Museum of Natural History in London (BM[NH]) 
(now called The Natural History Museum), and 
the American Museum of Natural History in New 
York City (AMNH). Because of their selective 
field efforts, the MZFAA teams obtained relative- 
ly few examples of Madagascar's endemic ro- 
dents, the Nesomyinae (about 70, compared to 
some 470 lemurs — see Carleton & Schmidt, 
1990), but, for the era, the specimens collected 
represented important additions to our knowledge 
of the indigenous small mammals. Regrettably, 
systematists never attempted a critical taxonomic 
synthesis of the mammalian collections made by 
the MZFAA, such as was undertaken for the birds 
(Delacour, 1932; Rand, 1936), and long over- 
looked was one nondescript, mouse-like rodent 
obtained by Austin L. Rand near Manjakatompo, 
a forestry station on the upper slopes of the Anka- 
ratra Massif in the Central Highlands (synony- 
mous with the Central High Plateau). 

Recognition of the significance of Rand's cap- 
ture was further hindered by an innocent process- 
ing error, which resulted in receipt of the skin at 
the AMNH and accession of the skull into the 
MNHN. For decades, the separated parts of the 
enigmatic specimen went unnoticed and unappre- 
ciated until the early 1970s, when Karl F. Koop- 
man, then Associate Curator in the Department of 
Mammalogy, AMNH, began an attempt to resolve 
his inability to classify the lone museum skin ac- 
cording to available taxonomic standards on ne- 
somyine rodents. Subsequent discussions about 
muroid diagnostic characters with his colleague 
Guy G. Musser prompted inquiries and led to the 
eventual rediscovery of the corresponding crani- 
um and mandibles in the MNHN. Through the 
gracious and understanding cooperation of Dr. 
Francis Petter, Laboratoire des Mammiferes et 
Oiseaux, MNHN, the errant skull was reunited 
with its skin in the late 1970s, and the now com- 
plete museum specimen (AMNH 100727) only 



confirmed Koopman's suspicions of its morpho- 
logical uniqueness and newness to the rodent fau- 
na of Madagascar. Other research commitments 
and priorities intervened to divert the attention of 
Koopman and Musser from jointly describing the 
new taxon. 

The impetus to report Rand's Manjakatompo 
specimen at this time and in the present context 
issues from renewed interest in the native rodents 
of Madagascar and from the resurgence of field- 
work on the island. Ongoing revisionary work by 
Carleton (1994) and Carleton and Schmidt (1990), 
and the 1993 rediscovery by Goodman of Rand's 
unremarkable mouse on the Andringitra Massif, 
over 300 km to the south of Ankaratra, make for- 
mal identification of this form both propitious and 
necessary. The fresh material from the Reserve 
Naturelle Integrate (RNI) d' Andringitra encour- 
ages fuller elaboration of the form's defining 
traits, which further attest to the distinctiveness of 
the taxon and convince us that it represents a ge- 
nus and species new to the fauna of Madagascar. 



Materials and Methods 

The material described herein and the compar- 
ative series of Macrotarsomys that are referenced 
are housed in the AMNH, New York City; the 
Field Museum of Natural History (FMNH), Chi- 
cago; the MNHN, Paris; the National Museum of 
Kenya (KNM), Nairobi; and the National Muse- 
um of Natural History (USNM), Smithsonian In- 
stitution, Washington, D.C. Nesomyine holdings 
in other museums have been studied by M.D.C., 
and one or both authors have examined the ho- 
lotypes of all described forms of Nesomyinae, 
except for Peters's (1870) Nesomys rufus. The fol- 
lowing samples, consisting of conventional round 
skins with skulls, as well as complete skeletons 
and fluid-preserved whole carcasses, were con- 
sulted for the various tabular summaries and an- 
atomical comparisons. 

Macrotarsomys bastardi bastardi — Fianaran- 
tsoa Province: 35 mi (= 56 km) N Ihosy, 3000 ft 
(= 915 m) (AMNH 119710); 5 km E Route Na- 
tionale 7, along road to Ivohibe, 750 m (USNM 
328793-328807). 

Macrotarsomys bastardi occidentalis — Maha- 
janga Province: 40 km S Marovoay, Ampijoroa 
(USNM 341817-24). Toliara Province: 40 km N 
Morondava, 7 km from sea, Beroboka (AMNH 



232 



FIELDIANA: ZOOLOGY 



119708-9); Petriky Forest, 5-7 km SE Manam- 
baro, 20 m (USNM 578715-9 and 578822-3). 

Macrotarsomys ingens — Mahajanga Province: 
Ankarafantsika Reserve, Ampijoroa (MNHN 
1958.636 and 1961.214-5; USNM 328831 and 
576753). 

Protarsomys macinnesi — Lower Miocene, Ru- 
singa Faunal Assemblage, Kenya (cast of holo- 
type, KNM 2350, cranium with associated man- 
dibles); Lower Miocene, Legetet, Kenya (cast of 
a referred lower jaw with ml -3, KNM 2188). 

Six measurements, in millimeters (mm) or 
grams (g), were taken by S.M.G. for each FMNH 
specimen in the flesh. Their abbreviations and def- 
initions are given below. 

TOTL, total length of body and tail: from the 
tip of the nose to the end of the caudal ver- 
tebra (not including terminal hair tuft). 

HBL, head and body length: from the tip of the 
nose to the distalmost point of the body (at 
base of tail). 

TL, tail length: from the base of the tail (held 
at right angles to the body) to the end of the 
last caudal vertebra (not including terminal 
hair tuft). 

HFL, hind food length: from the heel to the tip 
of the longest toe (not including claw). 

EL, ear length: from the basal notch to the dis- 
talmost rim of the pinna. 

WT, weight in grams: measured with Pesola 
spring scales to ± 0.5 g. 

Sixteen cranial and two dental dimensions were 
measured by M.D.C. to the nearest 0. 1 mm using 
handheld digital calipers accurate to 0.03 mm. 
These measurements, and their abbreviations, fol- 
low the anatomical landmarks defined and illus- 
trated in Carleton (1994): BBC, breadth of the 
braincase; BIF, breadth of incisive foramina; 
BM 1 s, breadth of the bony palate across the first 
upper molars; BOC, breadth across the occipital 
condyles; BR, breadth of rostrum; BZP, breadth 
of the zygomatic plate; DAB, depth of the audi- 
tory bulla; IOB, interorbital breadth; LBP, length 
of bony palate; LD, length of diastema; LIF, 
length of the incisive foramina; LM1-3, coronal 
length of maxillary toothrow; LR, length of ros- 
trum; ONL, occipitonasal length; PPB, posterior 
breadth of the bony palate; PPL, postpalatal 
length; WM1, width of the first upper molar; and 
ZB, zygomatic breadth. 

Analytical routines for the standard descriptive 
statistics were carried out using the Systat (ver- 
sion 5.05, 1994) computer program. Morpholog- 



ical terms generally follow Carleton (1980) and 
Voss (1988); names of dental structures follow 
Reig (1977), as illustrated by Carleton and Musser 
(1989). Field methods for the three FMNH spec- 
imens originating from the RNI d'Andringitra are 
described in Chapter 22. 



Monticolomys, new genus 

Type Species — Monticolomys koopmani, de- 
scribed below as new. 

Diagnosis — A form of the murid rodent sub- 
family Nesomyinae (sensu Carleton & Musser, 
1984; Musser & Carleton, 1993) characterized by: 
small size (TOTL = 205-240 mm; HBL = 89- 
101 mm) (in other nesomyines except Eliurus mi- 
nor and Macrotarsomys bastardi, HBL ^ 130 
mm); short (15-18 mm), densely furred, and 
rounded pinnae (pinnae long, ^ 21 mm, and 
sparsely haired in Macrotarsomys); tail apprecia- 
bly longer than head and body together, about 
135-140% of HBL (TL < HBL in Brachyuromys, 
Gymnuromys, Hypogeomys, and Nesomys), with- 
out noticeable elongation of caudal hairs over dis- 
tal half (distal tip penicillate in Macrotarsomys, a 
brushy tuft in Eliurus); hind foot relatively long 
and broad, outer digits I and V comparatively 
long (hind foot narrow, outer digits relatively 
short in Macrotarsomys); plantar surface with six 
fleshy pads, thenar and hypothenar as large as in- 
terdigital pads (thenar and hypothenar diminutive 
in Macrotarsomys); ungual tuft well developed, 
surpassing tip of claw (tuft hairs shorter than claw 
in Macrotarsomys). 

Cranium small (ONL = 26-28 mm, LM1-3 = 
3.5-3.7 mm) and delicately built with slender, 
parallel-sided zygomatic arches and narrow, hour- 
glass-shaped interorbit; vascular groove on squa- 
mosal-alisphenoid, stapedial and sphenofrontal 
foramina present (groove and foramina lacking in 
Brachytarsomys, Brachyuromys, Eliurus, Gym- 
nuromys, and Hypogeomys); alisphenoid strut ab- 
sent (strut present in Brachyuromys, most Eliurus, 
Hypogeomys, and Nesomys); ectotympanic bullae 
moderately inflated, narrow ventromedial wedge 
of periotic exposed (bullae smaller, wide expanse 
of periotic visible in Brachytarsomys, Eliurus, and 
Gymnuromys; bullae notably larger, periotic most- 
ly obscured in Brachyuromys, Hypogeomys, Mac- 
rotarsomys, and Nesomys); tegmen tympani re- 
duced, not contacting squamosal (articulation with 
squamosal in Brachytarsomys, Eliurus, and Gym- 
nuromys). 



CARLETON & GOODMAN: ENDEMIC RODENTS 



233 






234 



FIELDIANA: ZOOLOGY 



Upper toothrows divergent anteriorly (conver- 
gent anteriorly in Brachyuromys, more or less par- 
allel in Brachytarsomys, Eliurus, Gymnuromys, 
Hypogeomys, and Nesomys); molars cuspidate and 
brachyodont (cheek teeth moderately to extremely 
hypsodont, cuspation lost or indistinct, and occlu- 
sal surface planar in Brachyuromys, Brachytar- 
somys, Eliurus, Gymnuromys, and Hypogeomys); 
mesolophs(ids) absent, posterolophs rudimentary, 
anteroconids imperfectly developed (meso- 
lophs[ids] present in Gymnuromys and Nesomys); 
lower third molar reduced, notably smaller than 
second molar (size of third molar approximately 
equal to second in Brachyuromys, Eliurus, Hy- 
pogeomys, and Nesomys; conspicuously larger 
than second molar in Gymnuromys); upper molars 
with three roots and lower molars with two (upper 
molars four-rooted in Hypogeomys). 

Morphological Description — As for the sin- 
gle known species, described below. 



Monticolomys koopmani, new species 
(Figs. 21-1 to 21-5; Tables 21-1 and 21-2) 

Holotype— AMNH 100727: skin and skull of 
young adult male; original no. 62, collected May 
24, 1929, by Austin L. Rand as part of the 
MZFAA (see Rand, 1932, 1936). 

Standard measurements (in mm) from the spec- 
imen tag attached to the skin of the type include: 
total length, 205; tail length, 116; hind foot 
length, without claw, 24 (dry hind foot length, as 
measured by M.D.C. with claw, is 25.5); and ear 
length, 15. 

The condition of the skin is good. The skull is 
damaged, with its right bullae detached and pres- 
ent in the vial, both zygomatic arches incomplete, 
and the orbitosphenoid walls on both sides bro- 
ken; the mandibles are separated, and their an- 
gular and coronoid processes are incomplete. 

Type Locality — Madagascar, Antananarivo 
Province, Manjakatompo, ca. 19°20'S 47°26'E. 

Rand identified the locality only as "Monjak- 
atompo" (spelling in the United States Board on 
Geographic Names, 1989 gazetteer, as Manjaka- 
tompo) on his original tag and later (1936) sup- 



plied the collection elevation as 1800 m. This 
height probably refers to an area of fragmented 
natural forest intermixed with secondary grass- 
land in the vicinity of the forestry station of Man- 
jakatompo. Elevational records mentioned on oth- 
er specimen tags or in Richard Archbold's journal 
(AMNH Mammalogy Department) indicate that 
the team worked the slopes of Ankaratra at least 
from 1650 m to 1950 m (Rand's own field notes 
for the MZFAA were apparently lost or de- 
stroyed). The little natural forest remaining on the 
slopes of Ankaratra above Manjakatompo is es- 
timated to be about 650 ha (Nicoll & Langrand, 
1989). 

Referred Specimens — FMNH 151727 (male, 
skull with postcranial skeleton), 151899 (female, 
skull with whole carcass in fluid), and 151900 
(male, skull with whole carcass in fluid); all from 
Fianarantsoa Province, 38 km S Ambalavao, RNI 
d'Andringitra, ridge E of Volotsangana River, 
1625 m; 22° ITS 46°58'E; collected by S.M.G., 
December 12-14, 1993. 

Distribution — At present known only from 
two localities in the Central Highlands, about 
1800 m on the Ankaratra Massif and 1625 m on 
the Andringitra Massif. 

Diagnosis — As for the genus, above. 

Morphological Description — Fur soft, rela- 
tively thick and fine (based on the holotype, 
AMNH 100727). Cover hairs of dorsum usually 
tricolored — basal two-thirds plumbeous gray, 
middle band a deep ochraceous, and short tip 
dark brown to black; guard hairs entirely black, 
longer, and more heavily concentrated toward the 
middorsum and on the rump; general appearance 
of upperparts a somber or muted dark brown. Fur 
over inguinum, abdomen, and chest with plum- 
beous bases and pale buffy to whitish tips, most- 
ly white to bases on chin and throat; combined 
effect a dark gray ventrum not sharply demar- 
cated from dorsum. Mystacial vibrissae medium 
in length, the longest whiskers reaching the top 
of the pinnae when appressed to the skin. Pinnae 
short and rounded (Fig. 21-1; Table 21-1), dense- 
ly clothed internally and externally with slate 
hairs. Tail very long relative to head-and-body 
length (TL about 138% of HBL) and appearing 



Fig. 21-1. Dorsal and ventral views of round museum skins: left pair, the holotype of Monticolomys koopmani 
(AMNH 100727, TOTL = 205 mm), a young adult male from Antananarivo Province, Manjakatompo; right pair, 
an original topotype of Macrotarsomys bastardi occidentalis (AMNH 119709, TOTL = 228 mm), an adult male 
from Toliara Province, Beroboka. 



CARLETON & GOODMAN: ENDEMIC RODENTS 



235 









Fig. 21-2. Plantar view of the left hind foot and lateral view of the distal phalanges of the fourth digit: A-A', 
Monticolomys koopmani (FMNH 151900, HFL = 25 mm), an adult female from Fianarantsoa Province, RNI d'An- 
dringitra; B-B', Macrotarsomys bastardi (USNM 578822, HFL = 27 mm), an adult male from Toliara Province, 
Petriky Forest. Abbreviations: 1-4, interdigital pads one through four; hy, hypothenar pad; th, thenar pad. Note the 
contrasts in size and position of the plantar pads between Monticolomys (A) and Macrotarsomys (B), greater pilosity 
on the undersurface of the phalanges in Macrotarsomys (B'), and stronger development of the ungual tuft in Mon- 
ticolomys (A')- 



monocolored; caudal hairs a pale brown, medium 
in length over entire tail without notable peni- 
cillation or development of distal tuft; scutella- 
tion fine, partially obscured by tail hairs. Mam- 
mae number six (based on FMNH 151899), dis- 
tributed as postaxial, abdominal, and inguinal 
pairs. 

Tops of metapodials and phalanges covered 
with fine white hairs; ungual tufts present, the 
white hairs extending beyond end of claw (Fig. 
21-2); palmar and plantar surfaces and underneath 
of phalanges naked. Forefoot with stubby pollex 
bearing a nail, other manual digits with claws. 
Hind foot relatively long (Table 21-1), about 25% 
of HBL, and wide, with claws on all digits. Cen- 
tral digits (II-IV) of hind foot almost as long as 
corresponding metatarsals; digit V relatively long, 
its claw reaching to end of second phalanx of digit 
IV; digit I more reduced, its claw extending to 
middle of first phalanx of digit II (Fig. 21-3). 



Plantar surface bearing six moderately sized and 
cushion-like pads (Fig. 21-2); interdigital pads 2- 
4 set close together at base of digits, interdigital 
1 situated somewhat apart at base of hallux; hy- 
pothenar pad about as large as interdigitals, po- 
sitioned near to but slightly posterior of interdig- 
ital 1 ; thenar ovate, situated near middle of tarsus- 
metatarsus, and subequal in size to distal pads 
(Fig. 21-2). 

Cranium lightly constructed, delicate in ap- 
pearance (Fig. 21-4). Rostrum narrow and mod- 
erately long, about 35% of ONL, taper forward 
from nasolacrimal capsules to end of nasals very 
gradual; anterior tips of nasals rounded, posterior 
margins blunt and coextensive with posteriormost 
limit of rostral processes of premaxillae. Zygo- 
matic plate narrow, its anterior edge slightly con- 
vex and set well behind nasolacrimal capsule; 
posterior border of plate positioned appreciably in 
front of anterior root of M 1 ; dorsal notch distinct 



236 



FIELDIANA: ZOOLOGY 







Fig. 21-3. Dorsal view of the right hind foot skele- 
ton: A, Monticolomys koopmani (FMNH 151727, HFL 
= 24 mm), an adult male from Fianarantsoa Province, 
RNI d'Andringitra; B, Macrotarsomys bastardi (USNM 
578719, HFL = 28 mm), an adult male from Toliara 
Province, Petriky Forest. Abbreviations: I-V, pedal 
digits one through five. Note the relatively greater 
lengths of digits I and V in Monticolomys as compared 
to those of Macrotarsomys, as well as the tighter coal- 
ignment of the metatarsals in Macrotarsomys. 



but shallow. Zygomatic arches thin and parallel- 
sided to slightly bowed over midportion, project- 
ing laterad just beyond margins of braincase and 
weakly converging anteriorly; jugal slender but 
relatively long, forming most of midspan of arch 
and distinctly separating zygomatic processes of 
squamosal and maxillary. 

Interorbital region narrow, exposing floor of or- 
bits in dorsal view; hourglass-shaped, lacking su- 
praorbital shelving and ridging; orbital projection 
of lacrimal inconspicuous. Braincase similarly 
free of ridges and crests, smooth and rounded in 
dorsal appearance (Fig. 21-4); frontoparietal su- 
ture evenly curved, not defining a sharp angle at 
midsagittal juncture; interparietal moderately 
large and rhomboidal in outline, its lateral apices 
not contacting squamosals; dorsal profile of skull 
gently arched throughout, its highest point formed 
just anterior to the frontoparietal junction. 



Incisive foramina medium in length, spanning 
about 66% of diastema and terminating at level 
of anterior root of first molars. Palatal bridge rel- 
atively broad and smooth, devoid of corrugations 
and excrescences; posterior palatine foramina oc- 
cur as simple round holes in maxillopalatine su- 
ture, about level with abutment of Ml-m2; pos- 
terior border of palate U-shaped, situated even 
with end of third molars, and lacking posterolat- 
eral palatal pits. Lateral borders (pterygoid pro- 
cesses) of mesopterygoid fossa more or less 
straight sided and enclosing spacious sphenopal- 
atine vacuities. Parapterygoid fossae broad com- 
pared to width of mesopterygoid fossa, relatively 
short and triangular in shape, recessed little rela- 
tive to the plane of the hard palate; roof of par- 
apterygoid fossae almost flat and mostly osseus, 
pierced only by posterior opening to alisphenoid 
canal near their posterolateral corner and there 
scored by a groove for passage of the infraorbital 
branch of the stapedial artery. 

Ectotympanic bullae moderately sized, with 
narrow posteromedial wedge of the periotic visi- 
ble in ventral aspect; anterodorsal rim of the bulla 
abuts the ventrolateral margin of the squamosal 
bone but not obscuring the slit-like middle lacer- 
ate foramen; anterior flange of tegmen tympani 
inconspicuous, not contacting ventrolateral mar- 
gin of squamosal; eustacian tube short, overlap- 
ping but not enveloping tip of pterygoid process; 
malleus of the parallel type, orbicularis apophysis 
knob-like without definitive neck, deflected ven- 
tromedially. Postglenoid foramen large, semioval 
in outline, circumscribing an area about three to 
four times that of the smaller subsquamosal fe- 
nestra; hamular process of squamosal well defined 
and slender. Mastoid capsule small and bulbous, 
perforated by posterodorsal fontanelle. 

Alisphenoid strut absent, masticatory-buccina- 
tor and accessory oval foramina conjoined as one 
expansive opening. Sphenofrontal and stapedial 
foramina present, squamosal-alisphenoid groove 
well defined, posterior opening of alisphenoid ca- 
nal large, and vascular groove crossing the pos- 
terolateral corner of the parapterygoid plate — an- 
atomical landmarks suggesting a complete ce- 
phalic circulatory pattern with retention of the su- 
praorbital and infraorbital branches of the 
stapedial artery. 

Coronoid process of dentary falcate, extending 
dorsad slightly above the condylar process; sig- 
moid and angular notches clearly incised but un- 
remarkable in size and shape; alveolus of lower 
incisor terminates posteriorly at level of coronoid 



CARLETON & GOODMAN: ENDEMIC RODENTS 



237 






.## 




*«*&£/ 





Fig. 21-4. Dorsal, ventral, and lateral views (X2) of the cranium and mandible: left, Monticolomys koopmani 
(FMNH 151899, ONL = 28.1 mm), an adult female from Fianarantsoa Province, RNI d'Andringitra; right, Macro- 
tarsomys bastardi (USNM 578715, ONL = 29.3 mm), an adult female from Toliara Province, Petriky Forest. 



process, well below ventral rim of sigmoid notch; 
capsular process evident as low mound. 

Axial skeleton (per one specimen, FMNH 
151727) with 13 thoracic, seven lumbar, four sa- 
cral (two pseudosacral), and 38 caudal vertebrae; 
tuberculum of first rib articulates only with trans- 
verse process of first thoracic vertebra, not touch- 
ing transverse process of seventh cervical; neural 
spine of second thoracic vertebra rises conspicu- 
ously above the spines of other thoracics. Ent- 
epicondylar foramen present on humerus. 

Upper incisors asulcate and opisthodont; enam- 
el medium orange. Upper molar rows divergent 
anteriorly; upper and lower second molars about 



three-quarters the size of the first molars but sim- 
ilar in occlusal design; upper and lower third mo- 
lars notably smaller, circular to oval, about one- 
half to two-thirds the length and area of the con- 
tiguous second molars, and their serial homologies 
more obscure. Molars decidedly brachyodont, 
cuspidate (Fig. 21-5); principal cusps positioned 
nearly opposite in the upper molars, more alter- 
nate in the lowers. Anterocone of Ml well de- 
fined, broad and undivided (weakly bilobate in 
one specimen); anteroconid of ml hardly distin- 
guishable from metaconid and anterior cingulum; 
entoconid poorly defined on m3; low median 
mure(id) interconnects anterior and posterior pairs 



238 



FIELDIANA: ZOOLOGY 




Fig. 21-5. Occlusal views (X20) of the upper (left member) and lower (right member) right molar rows: left 
pair, Monticolomys koopmani (FMNH 151899, LM1-3 = 3.42 mm), an adult female from Fianarantsoa Province, 
RNI d'Andringitra; right pair, Macrotarsomys bastardi (USNM 578715, LM1-3 = 3.95 mm), an adult female from 
Toliara Province, Petriky Forest. 



of cusps on each first and second molar. Principal 
lingual and labial enamel folds broad and meet 
near the midline of the tooth, not interpenetrating; 
posteroloph (postcingulum) of Ml -2 and corre- 
sponding posteroflexus indistinct, obliterated after 
slight wear; posterolophid and broad posteroflexid 
present and persistent with wear on ml -2, indis- 
tinct or absent on m3; protoflexus of M2 absent, 
anteroloph and paraflexus short; metaflexid of ml 
represented by faint indentation or indiscernible, 
opposite protoflexid well defined; anterolabial cin- 
gulum present but small on m2, variable on m3; 
mesolophs(ids), as well as other accessory crests 
and styles, absent. Three roots on each upper mo- 
lar and two on each lower. 

Notes on Natural History — The habitat 
where Rand collected the type is characterized on 
the skin specimen tag as "grass slope with scat- 
tered bushes." The type specimen, even at the 
time of the MZFAA in 1929, likely originated 
from despoiled habitat. In his later summary of 
the itinerary, Rand (1936: pp. 161 and 163) ex- 
panded upon the vegetation of Ankaratra: 

The slopes of the greater part of the mountain 



were grass-covered and supported a growth of 
health and bracken . . . just above Monjakatom- 
po, was an area of humid forest, an isolated 
remnant of the forest that once covered the cen- 
tral portion of Madagascar. This wooded area 
was of the humid forest type, with large trees 
hung with lianas and mosses. Tree ferns were 
common and in some places there was consid- 
erable undergrowth. 

Populations of Monticolomys koopmani may 
co-occur with a variety of other nesomyine spe- 
cies. On Ankaratra, Rand also collected three 
specimens of Brachyuromys betsileoensis 
(AMNH 100802, BMNH 35.1.8.341, and MNHN 
1932.3521) near Manjakatompo, at 1700 m and 
1950 m. The spare habitat annotations on their 
skin specimen tags ("grass and brush at edge of 
pond"; "grass and brush ridge") leave open the 
question of whether the Brachyuromys were syn- 
topic with Monticolomys. The grasslands on the 
mountain are not natural vegetation formations 
but have resulted from human-set fires (Perrier de 
la Bathie, 1927; Nicoll & Langrand, 1989). To 
judge from the personal field notes of Richard 



CARLETON & GOODMAN: ENDEMIC RODENTS 



239 



Table 21-1. Selected external and craniodental 
measurements (in mm) of the holotype (AMNH 100727) 
and referred specimens (FMNH 151727, 151899, and 
151900) of Monticolomys koopmani. 





Holotype 






(Ankaratra 


Referred specimens 


Variable 


Massif) 


(Andringitra Massif) 


TOTL 


205.0 


236.3 (234-240) 


HBL 


89.0 


98.0(94-101) 


TL 


116.0 


138.0(134-143) 


HFL 


24.0 


24.3 (24-25) 


EL 


15.0 


18.3(18-19) 


WT 




26.3 (25-27) 


ONL 


26.3 


27.8(27.5-28.1) 


ZB 


... 


13.6(13.1-14.1) 


BBC 


12.5 


12.7(12.4-13.2) 


IOB 


3.9 


4.0 (3.9-4.0) 


LR 


8.5 


9.9 (9.6-10.2) 


BR 


4.2 


4.7 (4.5-4.9) 


PPL 


8.7 


9.5 (9.3-9.7) 


LBP 


3.7 


4.2 (4.0-4.4) 


LIF 


5.3 


5.2(5.1-5.3) 


BIF 


2.0 


2.0(1.9-2.0) 


LD 


7.2 


7.9 (7.7-8.0) 


BMls 


5.7 


5.9 (5.8-6.0) 


DAB 


4.4 


4.8 (4.7-4.9) 


BZP 


2.2 


2.2(2.1-2.2) 


BOC 


6.4 


6.6 (6.5-6.8) 


LM1-3 


3.7 


3.5 (3.4-3.7) 


WM1 


1.1 


1.1(1.0-1.1) 



Refer to the Materials and Methods section for an ex- 
planation of the abbreviations. 



Archbold, who accompanied Rand to Ankaratra 
and also collected around Manjakatompo, popu- 
lations of Rattus rattus were well established in 
the area at the time. 

The more thorough survey conducted within 
the RNI d' Andringitra documents sympatry of 
Monticolomys koopmani with seven other rodent 
species — Rattus rattus, Brachyuromys ramirohi- 
tra, Eliurus majori, E. minor, E. tanala, Gymnu- 
romys roberti, and Nesomys rufus (voucher num- 
bers are listed in Chapter 22). Several of these 
Nesomyinae are themselves restricted to montane 
and upper montane vegetational zones {Brachy- 
uromys ramirohitra, Eliurus majori, E. tanala, 
and Nesomys rufus) or are widely ranging in al- 
titudinal occurrence (Eliurus minor and Gymnu- 
romys roberti). 

The specimens from Andringitra were collected 
in upper montane forest (ridge-top sclerophyllous 
forest) that is regularly shrouded in mist and cloud 
cover. The area (Fig. 21-6) is dominated by dense 
bamboo stands of the genera Arundinaria and 
Nastus and by trees of the families Podocarpa- 



ceae, Cunoniaceae, and Pandanaceae (Chapter 2). 
For woody plants over 10 cm diameter at breast 
height (dbh), trees in the 1625 m zone averaged 
17.0 cm dbh and 7.9 m tall (Chapter 4). Further- 
more, at this elevation, there was a marked in- 
crease in epiphytes, mosses, and lichens in tree 
crowns and on trunks in comparison to forests at 
lower elevations. 

The relatively long tail, the conformation of the 
hind foot, and the large plantar pads suggest that 
individuals of Monticolomys are adept climbers. 
In the RNI d' Andringitra, both live-caught spec- 
imens (FMNH 151899 and 151900) were cap- 
tured on successive nights in the same Sherman 
trap placed on a nearly horizontal segment of a 
liana that was less than 5 cm in circumference and 
2 m above the ground. The vine originated at the 
ground and wound its way to the canopy. The 
third specimen (FMNH 151727) was found the 
next morning in a pitfall bucket along with two 
Microgale taiva (FMNH 151724 and 151725). 
With just three records, little can be said about the 
species' focus of activity, whether predominantly 
arboreal, scansorial, or terrestrial. 

The two males from Andringitra, collected in 
December 1993, possessed partially or fully de- 
scended testes; the one female had small teats and 
an imperforate vagina. 

Remarks — The natural environments on the 
Ankaratra and Andringitra massifs inhabited by 
Monticolomys are isolated, their unique montane 
populations wholly allopatric to one another (see 
discussion on biogeography below and Fig. 
21-11). The meager sample sizes available from 
the type locality (N = 1) and from the RNI 
d' Andringitra (N = 3) hinder any meaningful 
evaluation of the level of differentiation between 
the two distantly separated localities. Bias in age 
representation further complicates appreciation of 
intra- and interlocality differences — the holotype 
is a young adult male, his molars little worn but 
in mature pelage, whereas the three paratypes 
(two males, one female) are fully adult with mod- 
erately to heavily worn toothrows. 

Age-related, post-weaning growth may plausi- 
bly account for the generally smaller size of the 
holotype as compared to the three Andringitra 
specimens (Table 21-1). Certain mensural vari- 
ables—such as HBL, TL, ONL, LR, LD, PPL, all 
of which are decidedly shortest in the holotype — 
may exhibit substantial age effects among the 
samples typically available to the systematist, 
even with attempts to control for age biases by 
eliminating juveniles or restricting analyses to 



240 



FIELDIANA: ZOOLOGY 






Table 21-2. Comparison of selected external and craniodental measurements (in mm) of Monticolomys koopmani 
and species of Macrotarsomys. Statistics for the samples include the mean ± standard deviation, and the range. 





Monticolomys 


Macrotarsomys 


Macrotarsomys 


Macrotarsomys 




koopmani 


b. bastardi 


b. occidentalis 


ingens 


Variable 


(N = 4) 


(N = 9) 


(N = 4) 


(N = 4) 


TOTL 


228.5 ± 15.9 


213.0 ± 11.7 


225.5 ± 10.8 


336.5 ± 36.6 




205.0-240.0 


195.0-232.0 


213.0-235.0 


310.0-390.0 


HBL 


95.7 ± 5.4 


91.6 ± 3.2 


96.7 ± 4.2 


127.0 ± 15.7 




89.0-101.0 


86.0-95.0 


92.0-102.0 


115.0-150.0 


TL 


132.5 ± 1.16 


121.5 ± 12.1 


128.7 ± 6.9 


209.5 ±21.7 




116.0-143.0 


100.0-142.0 


121.0-136.0 


193.0-240.0 


HFL 


24.3 ± 0.5 


24.6 ± 0.9 


26.7 ± 1.0 


35.3 ± 2.5 




24.0-25.0 


23.0-26.0 


26.0-28.0 


32.0-38.0 


EL 


17.5 ± 1.7 


21.3 ± 1.2 


23.0 ± 1.6 


24.3 ± 1.5 




15.0-19.0 


20.0-23.0 


21.0-25.0 


23.0-26.0 


WT 


26.3 ± 1.2 


24.5 ± 2.7 


31.0 ± 5.6 


57.0 




25.0-27.0 


21.0-28.0 


26.0-38.0 


54.0, 60.0 


ONL 


27.4 ± 0.8 


28.1 ± 0.7 


28.6 ± 0.5 


38.9 ± 2.5 




26.3-28.1 


26.7-28.9 


28.2-29.3 


35.4-41.3 


ZB 


13.6 ± 0.5 


14.0 ± 0.5 


14.5 ± 0.4 


18.9 ± 1.5 




13.1-14.1 


13.0-14.6 


14.1-14.8 


16.7-20.2 


BBC 


12.7 ± 0.4 


11.3 ± 0.2 


11.7 ± 0.3 


14.3 ± 0.5 




12.4-13.2 


10.9-11.5 


11.3-12.1 


13.5-14.3 


IOB 


3.9 ± 0.1 


4.6 ± 0.1 


4.8 ± 0.1 


5.6 ± 0.4 




3.9-4.0 


4.3-4.7 


4.6-4.9 


5.1-6.0 


LR 


9.5 ± 0.7 


10.2 ± 0.5 


10.1 ± 0.3 


14.4 ± 1.2 




8.5-10.2 


9.5-11.1 


9.9-10.5 


13.0-16.0 


BR 


4.5 ± 0.3 


4.7 ± 0.2 


4.9 ± 0.2 


6.8 ± 0.3 




4.2-4.9 


4.4-5.1 


4.7-5.2 


6.6-7.1 


PPL 


9.3 ± 0.4 


9.5 ± 0.4 


9.8 ± 0.6 


13.3 ± 1.1 




8.7-9.7 


9.1-10.0 


9.1-10.3 


11.9-14.6 


LBP 


4.1 ± 0.3 


4.7 ± 0.2 


4.6 ± 0.3 


6.3 ± 0.5 




3.7-4.4 


4.4-4.9 


4.2-4.9 


5.6-6.9 


LIF 


5.2 ± 0.1 


5.1 ± 0.2 


5.0 ± 0.3 


7.1 ± 0.4 




5.1-5.3 


4.9-5.5 


4.7-5.4 


6.7-7.5 


BIF 


2.0 ± 0.1 


2.0 ± 0.1 


2.1 ± 0.2 


2.6 ± 0.2 




1.9-2.1 


1.8-2.1 


1.9-2.3 


2.4-2.8 


LD 


7.7 ± 0.3 


7.4 ± 0.4 


7.7 ± 0.2 


11.6 ± 0.9 




7.2-8.0 


6.8-8.0 


7.5-7.9 


10.3-12.5 


BMls 


5.9 ± 0.1 


5.9 ± 0.2 


5.9 ± 0.1 


7.7 ± 0.5 




5.7-6.0 


5.5-6.2 


5.7-5.9 


7.0-8.0 


DAB 


4.7 ± 0.3 


5.5 ± 0.2 


5.4 ± 0.1 


7.0 ± 0.6 




4.4-4.9 


5.1-5.7 


5.2-5.5 


6.3-7.5 


BZP 


2.2 ± 0.05 


2.9 ± 0.2 


2.8 ± 0.1 


3.8 ± 0.2 




2.1-2.2 


2.6-3.1 


2.7-2.9 


3.5-4.0 


BOC 


6.5 ± 0.2 


6.1 ± 0.3 


6.5 ± 0.2 


8.4 ± 0.6 




6.4-6.8 


5.6-6.3 


6.3-6.7 


7.7-8.9 


LM1-3 


3.57 ±0.12 


4.08 ±0.17 


3.91 ± 0.05 


4.85 ±0.16 




3.42-3.70 


3.79-4.31 


3.85-3.95 


4.71-5.04 


WM1 


1.09 ± 0.02 


1.28 ± 0.06 


1.19 ± 0.02 


1.64 ± 0.04 




1.07-1.12 


1.21-1.39 


1.17-1.21 


1.59-1.68 



specimens assigned to crudely defined age-tooth- 
wear categories (Voss & Marcus, 1992). We note 
that several cranial dimensions measured across 
the brain (BBC, IOB, BOC) or on the molars 
LM1-3, WM1), regions that change little or not 
at all with age following weaning, are basically 
alike in the Ankaratra and Andringitra examples 
(Table 21-1). 



In view of such insufficiencies in sample size 
and imbalanced age representation, and lacking 
other evidence of substantial divergence, we treat 
the two allopatric samples as one species but 
stress the need for additional voucher material. 
Broader age series are especially critical to verify 
the typical condition of traits such as the presence 
of a posteroloph and formation of the anteroconid. 



CARLETON & GOODMAN: ENDEMIC RODENTS 



241 




Fig. 21-6. View through the understory of upper montane forest at 1625 m on the eastern slopes of the RNI 
d'Andringitra. Portions of this forest, where three specimens of Monticolomys koopmani were taken during the 1993 
inventory, are dominated by dense stands of bamboo (Arundinaria and Nastus), intermixed with Podocarpus trees. 
The groundcover is largely covered by Pandanus (Photo by C. J. Raxworthy). 



These dental characters are obliterated after slight lifelong systematic contributions of Karl F. Koop- 



trituration and are best observed in juvenile spec- 
imens. 

Etymology — The generic name, mountain- 
dwelling mouse, evokes the montane occurrence 
of the form. The species is named to honor the 



man, Curator Emeritus in the Department ol 
Mammalogy, AMNH, and one-time Assistant Cu- 
rator of Mammals, FMNH. His curiosity, piquec 
by the orphaned skin that would not key out prop 
erly according to the standard classification, lee 



242 



FIELDIANA: ZOOLOGY 



him to the perception that it indeed represented a 
fundamentally different kind of Malagasy rodent. 



Morphological Comparisons 

Individuals of Monticolomys koopmani are im- 
mediately distinguishable from other kinds of Ne- 
somyinae on the basis of their diminutive size and 
generalized murine appearance. The mouse-like 
form offers sharp visual contrast to the distinctive 
and highly divergent physiognomies of the much 
larger Brachytarsomys, Brachyuromys, Gymnu- 
romys, Hypogeomys, and Nesomys (for example, 
see descriptions, measurements, and/or keys in 
Ellerman, 1941, 1949; Petter, 1972, 1975; and 
Chapter 22, this work). Among the tufted-tailed 
rats, genus Eliurus, only specimens of E. minor 
approach those of Monticolomys in overall size. 
Although comparable in average total length, in- 
dividuals of Monticolomys possess a still lighter 
build, shorter head-and-body length, relatively 
longer tail, and longer hind feet (see Carleton, 
1994; and Chapter 22, Table 22-2). Furthermore, 
they lack the conspicuous "bottle-brush" tail that 
characterizes E. minor and other species of Eli- 
urus. Crania and dentitions of E. minor and Mon- 
ticolomys differ even more strikingly than exter- 
nal form and permit no confusion of the two (see 
Carleton, 1994). 

Although discrimination of Monticolomys from 
most nesomyines is straightforward, two genera, 
one living and one extinct, present less-pro- 
nounced contrasts and raise broader questions of 
phylogenetic relationship. One, Macrotarsomys, 
whose gerbil-like physique may suggest very dis- 
tant phyletic association, in fact resembles Mon- 
ticolomys quite closely in certain cranial and den- 
tal details that require critical elaboration. The 
other, the Miocene form Protarsomys, merits at- 
tention because some authors have viewed it as 
close to the ancestry of Madagascar's indigenous 
rodents (Lavocat, 1973, 1978), if not actually syn- 
onymous with extant Macrotarsomys (Chaline et 
al.. 1977; Petter, 1990). 



Comparison with Macrotarsomys 
Milne Edwards and G. Grandidier 

The following morphological descriptions and 
comparative accounts are based on the four ex- 
amples of Monticolomys koopmani and specimens 

CARLETON & GOODMAN: ENDEMIC RODENTS 



representing both Macrotarsomys bastardi (in- 
cluding M. occidentalis Ellerman, 1949) and M. 
ingens (see Materials and Methods section). The 
type species of Macrotarsomys, M. bastardi 
Milne Edwards and G. Grandidier (1898), differs 
in subtle aspects from the later named and phys- 
ically much larger congener M. ingens (Petter, 
1959); however, unless noted otherwise, contrasts 
to Monticolomys apply to both species of Macro- 
tarsomys. 

Pelage and External Form — The dark brown 
dorsum and dull gray ventrum of Monticolomys 
present a somber impression compared to the light 
sandy to clay brown upperparts and wholly white 
underparts of Macrotarsomys. Lateral demarca- 
tion of the dorsal-ventral pelage is thus conspic- 
uous in the latter genus. Pelage texture is similarly 
fine in both genera, but appears longer and denser, 
for the size of the animal, in the sample of Mon- 
ticolomys. 

The generalized murine appearance of Monti- 
colomys offers several points of contrast to the 
gerbilline form that characterizes Macrotarsomys 
(Fig. 21-1). The pinnae of Monticolomys are 
small, thick, and rounded, protruding only slightly 
above the body fur, and densely cloaked with 
black hairs. Those of Macrotarsomys, however, 
are much longer and wider (Table 21-2), appear- 
ing almost naked (clothed with fine white hairs 
internally and pale brown externally) and more 
pliant. The tails of both Monticolomys and Mac- 
rotarsomys bastardi are relatively long (TL 133- 
138% of HBL), that of M. ingens exceptionally 
so (TL 160% of HBL). Examples of Monticolo- 
mys, however, lack the distal elongation of caudal 
hairs forming a modest pencil; instead the hairs 
are roughly even in length, spanning about two 
caudal annuli, over the entire tail. In specimens 
of Macrotarsomys, hairs covering the proximal 
two-thirds of the tail are sparse and short, about 
one annulus in length, but they lengthen some- 
what abruptly over the distal one-third to form a 
noticeably penicillate tip (not like the conspicuous 
terminal brush, however, found in species of Eli- 
urus). Notwithstanding the wide range in absolute 
lengths (Table 21-2), the hind feet of Monticolo- 
mys and Macrotarsomys have similar proportions 
(HFL about 26-28% of HBL). 

Hind Foot — The genera differ markedly in 
hind foot conformation, size and apportionment of 
the plantar pads, and distribution of pedal hairs. 
The pes of Macrotarsomys appears more svelte 
and absolutely narrower in view of the greater 
length of the proximal tarsal-metatarsal portion 
relative to the distal phalanges (Fig. 21-2). Not 



243 



only are the digits of Macrotarsomys generally 
shorter compared to those of Monticolomys, the 
outer members (I and V) are more reduced rela- 
tive to the central three (II-IV), a proportional 
configuration evident in both the intact hind foot 
and the foot skeleton (Figs. 21-2 and 21-3). Thus, 
in Macrotarsomys, the claw of digit V extends to 
the junction of phalanges 1-2 of digit IV (junction 
of 2-3 in Monticolomys), and that of digit I reach- 
es the end of the second metatarsal (middle of the 
first phalanx of digit II in Monticolomys). In ad- 
dition, the five metatarsal bones in Macrotarso- 
mys are closely appressed, although not fused, 
along their full length. Those of Monticolomys, 
however, splay noticeably at their distal ends and 
appear in looser contact along their proximal sec- 
tions (Fig. 21-3). 

Individuals of both Monticolomys and Macro- 
tarsomys possess six plantar pads, but their size, 
shape, and spatial arrangement contrast apprecia- 
bly. The six footpads in Monticolomys are fleshy, 
bulbous mounds, more of less equal to one anoth- 
er in circumference. In Macrotarsomys, the pads 
are not only smaller and lower in relief but differ 
in size; that is, the posterior three (thenar, hypo- 
thenar, and interdigital 1) are about half the di- 
ameter and protuberance of the anterior interdi- 
gitals 2-4 (Fig. 21-2). The hypothenar and inter- 
digital 1 are coaligned in Macrotarsomys (more 
alternate in Monticolomys), and the thenar appears 
to be positioned more distally from the heel. Per- 
haps the most noticeable difference when exam- 
ining the plantar surfaces of the two is the devel- 
opment of the thenar pad — an ovate, pillowy 
mound in Monticolomys compared to an insub- 
stantial, circular papilla in species of Macrotar- 
somys. 

In contrast to the naked plantar surface of Mon- 
ticolomys, scattered hyaline hairs are found on the 
sole in examples of Macrotarsomys bastardi, as 
well as along the ventral surface of the distal pha- 
langes (Fig. 21-2). In particular, a small cluster of 
hairs typically marks the midsole, between inter- 
digital pads 2-4 and the paired hypothenar and 
interdigital 1 . The single fluid specimen of Mac- 
rotarsomys ingens examined (USNM 328831) 
lacks such isolated hairs on the midsole but pos- 
sesses them on the undersides of the toes. Unlike 
the hairs covering the dorsum of the foot, these 
plantar hairs seem stiffer, but neither they nor 
hairs on the phalangeal undersurfaces were de- 
tected in the two fluid specimens of Monticolo- 
mys. On the other hand, the ungual tufts, tussocks 
of hairs that emerge at the base of the claws, ap- 



244 



pear more profuse in individuals of Monticolomys 
(Fig. 21-2), arching over and beyond the tip of 
the claw (in both Macrotarsomys, reaching about 
one-half to three-quarters of claw length but not 
surpassing the tip). 

Cranium and Mandible — The crania of Mon- 
ticolomys and Macrotarsomys share many quali- 
tative traits that convey a fundamental resem- 
blance. Notable among these are moderately elon- 
gate incisive foramina (LIF 62-68% of LD); a 
flat, short, and wide palate with simple posterior 
palatine foramina; broad, shallow parapterygoid 
fossae and narrow mesopterygoid fossa with large 
sphenopalatine vacuities; absence of alisphenoid 
struts; size and form of temporal openings and 
well-defined hamular process; and presence of 
vascular foramina and osseous grooves indicative 
of a primitive carotid blood supply. They princi- 
pally depart from one another in features of the 
interorbital region, configuration of the maxillary 
portion of the zygoma, and anatomy of the audi- 
tory bullae and surrounding area. 

The interorbital region and braincase of both 
genera are smooth and gently contoured, devoid 
of pronounced temporal ridges or crests. Speci- 
mens of Macrotarsomys bastardi, however, pos- 
sess an absolutely and relatively wider interorbit 
such that little of the orbital floor (top of maxil- 
lary) is visible from above. Also, they typically 
display a weak supraorbital shelf at the rear of the 
frontals that imparts a sharper constriction to the 
interorbit and longer appearance to the braincase 
(Fig. 21-4). The interorbit in examples of Macro- 
tarsomys ingens has a relatively narrow, hour- 
glass-shaped (amphoral) interorbit, more closely 
resembling that of Monticolomys. 

The zygomatic plate in examples of Montico- 
lomys is absolutely and relatively (BZP 8% of 
ONL) narrower than that observed in both species 
of Macrotarsomys (BZP 10% of ONL). Not only 
do specimens of Macrotarsomys have broader zy- 
gomatic plates, forming deeper zygomatic notch- 
es, but the zygomatic process of the maxilla wid- 
ens appreciably at its union with the plate and has 
a more rugose surface (Fig. 21-7). A bony ledge 
from the lacrimal interrupts the anterodorsal pe- 
rimeter of the orbit in samples of Macrotarsomys 
bastardi but not those of Monticolomys (Figs. 
21-4 and 21-7). This tabular projection, reminis- 
cent of the condition observed in certain Gerbil- 
linae (Carleton, 1980), is intermediate in size in 
M. ingens. 

Bullar inflation is pronounced in both species 
of Macrotarsomys (DAB about 48% of BBC) 

FIELDIANA: ZOOLOGY 





Fig. 21-7. Left lateral-oblique view of the rostrum 
and anterior zygomatic region: top, Monticolomys koop- 
mani (FMNH 151899; Fianarantsoa Province, RNI 
d'Andringitra); bottom, Macrotarsomys bastardi 
(USNM 578715; Toliara Province, Petriky Forest). Ab- 
breviations: lop, orbital projection of the lacrimal bone; 
pm, premaxilla; zn, zygomatic notch; zp, zygomatic 
plate. Note the wider zygomatic plate and deeper notch 
in Macrotarsomys, as well as the tabular extension of 
the lacrimal bone at the anterodorsal rim of the orbit. 



less so in Monticolomys (DAB 37% of BBC), and 
perhaps accounts for certain other cranial dissim- 
ilarities. In Macrotarsomys, hypertrophy of the 
ectotympanic (Figs. 21-8 and 21-9), particularly 
noticeable in the expanse of translucent bone en- 
closing the middle ear cavity, plausibly relates to 
reduction of the middle lacerate foramen (present 
as a slit in Monticolomys), almost complete in- 
vestiture of the periotic in ventral view (visible as 
posteromedial wedge in Monticolomys), and for- 
mation of a deep trough on the medial bullar wall 
for passage of the internal carotid artery (round 
foramen in Monticolomys). In both genera, the 
tegmen tympani, an anterior flange of the periotic 



that roofs the epitympanic recess, lacks articula- 
tion with the squamosal, presumably an ancestral 
connection found in many New World muroids 
(see Voss, 1993) and in some nesomyines (e.g., 
Eliurus; see Carleton, 1994, Fig. 18). The tegmen 
tympani is inconspicuous though visible within 
the postglenoid foramen of Monticolomys, but it 
is smaller yet in Macrotarsomys and wholly ob- 
scured by the thickened, rugose dorsal rim of the 
external auditory meatus (Fig. 21-8). 

Mandibular conformation of the two genera dif- 
fers in several details. In Monticolomys, the cor- 
onoid process is more substantial and oriented 
more vertically; in contrast, the process in species 
of Macrotarsomys tapers to a thin spine that is 
angled more posterodorsally, in parallel to the ori- 
entation of the condylar process. As a conse- 
quence, the sigmoid notch of Macrotarsomys ap- 
pears narrower and shallower (Fig. 21-4); that of 
Monticolomys forms a broader half-oval typical of 
many muroids. The angular notch in specimens of 
Macrotarsomys is deeply excavated, imparting a 
broader and longer definition to the angular and 
condylar processes. The concavity of the angular 
notch is shallower in Monticolomys, and the an- 
gular and condylar processes are normally pro- 
portioned. 

Dentition — The upper incisors of Macrotar- 
somys, like those of Monticolomys, lack enamel 
grooves but differ slightly in color; they are yel- 
low-orange in the former and medium orange in 
the latter. 

The molars of Monticolomys and Macrotarso- 
mys are so remarkably alike that correct segre- 
gation of the genera is doubtful, except on the 
basis of size (Fig. 21-5). A trace of a posteroloph 
and minute posteroflexus is discernible on the 
M 1 -2s of most Monticolomys and Macrotarsomys 
ingens, but they seem to be lacking on those of 
M. bastardi, even for individuals with little tooth 
wear. The anteroconid is weakly defined in both 
genera, appearing as little more than an undiffer- 
entiated procingulum that originates from the an- 
terolingual shoulder of the metaconid, forms the 
front rim of the tooth, and merges imperceptibly 
with the anterolabial cingulum near the proto- 
conid. In some examples of Macrotarsomys, how- 
ever, especially M. ingens (USNM 576753), a 
shallow indentation (vestige of the metaflexid?) 
on the anterolingual face of ml imparts some sep- 
aration of the metaconid from the presumptive an- 
teroconid. Such evident differences must be 
viewed as qualified until verified by larger series 
of each species, including juveniles. 



CARLETON & GOODMAN: ENDEMIC RODENTS 



245 




Fig. 21-8. Left lateral view of otic region and associated foramina: left, Monticolomys koopmani (FMNH 15 1900; 
Fianarantsoa Province, RNI d'Andringitra); right, Macrotarsomys bastardi (USNM 578717; Toliara Province, Petriky 
Forest). Abbreviations: ab, auditory bulla (ectotympanic); ex, exoccipital; hp, hamular process of the squamosal; ms, 
mastoid capsule of periotic; pgf, postglenoid foramen; sq, squamosal; ssf, subsquamosal fenestra; tt, tegmen tympani. 
The tegmen tympani fails to contact the squamosal in either genus, but it remains visible within the postglenoid 
foramen of Monticolomys and is obscured by the rugose dorsal rim of the ectotympanic in Macrotarsomys. 



The Miocene Fossil Protarsomys 
Lavocat 

The time of arrival of the first nesomyine, or 
nesomyines, into Madagascar remains inferential, 
given the absence of terrestrial Tertiary records on 
the island and the rarity of appropriately aged fos- 
sil muroids from sites in eastern Africa, the pre- 
sumed area of origin. Lavocat (1978) postulated 
the entrance of ancestral nesomyines by over-wa- 
ter dispersal in the lower Miocene, in part because 
this epoch marks the first appearance of archaic 
cricetodontids in Africa. A period later than the 
lower to middle Miocene would prove difficult to 
reconcile with the lack of murines in Madagascar; 
they appear in the late Miocene of northern Africa 
and become commonplace in Pliocene beds of 
sub-Saharan Africa (see, e.g., Jacobs, 1985; De- 
nys, 1987). 

A more pivotal consideration, however, con- 
cerns the afrocricetodontine Protarsomys macin- 
nesi (Muroidea: Cricetodontidae). Lavocat (1973) 
described this fossil rodent from the Rusinga Fau- 
nal Assemblage, lower Miocene of Kenya, and 
viewed (1978: p. 80) it as close to the ancestry of 
the Malagasy Nesomyinae, in particular Macro- 
tarsomys: "The genus Protarsomys, approximate- 
ly the size of recent Mus musculus, is so close to 
the Recent Macrotarsomys [his comparisons in- 
volved only M. bastardi, not the larger M. ingens] 
of Madagascar that one must evidently find in this 
Protarsomys, or in very close relatives of this ge- 



nus, the ancestors of the Malagasy rodents." Cha- 
line et al. (1977) had earlier formalized this in- 
terpretation by synonymizing Protarsomys under 
Macrotarsomys, an action and affinity reaffirmed 
by Petter (1990). 

Description of Monticolomys, which itself ex- 
hibits traits suggesting relationship to Macrotar- 
somys, necessarily requires some consideration of 
Lavocat 's (1973) Protarsomys. The remarkable 
holotype of Protarsomys (KNM 2350) consists of 
a mostly intact skull with associated lower jaws 
embedded in matrix. Partial removal of the en- 
casing matrix has revealed the general dimensions 
and morphology of the cranium and mandibles in 
dorsal and lateral view, but features of the palate, 
basicranium, and otic capsules, as well as occlusal 
surfaces of the upper and lower molars, are still 
obscured on the type specimen. Accordingly, size 
comparisons of Protarsomys to specimens of 
Monticolomys and Macrotarsomys are limited to 
those variables accessible to measurement on the 
borrowed cast of the holotype. The configuration 
of other cranial and dental features has been 
gleaned from Lavocat 's (1973) excellent descrip- 
tion, from the cast of a referred lower right man- 
dible with ml -3 (KNM 2188), and from inspec- 
tion of his Plate 10, which portrays another attrib- 
uted specimen (KNM 2353) in which the ventral 
surface of the skull is exposed. 

As noted by Lavocat (1973, 1978), Protarso- 
mys has a small skull, about the size of a house 
mouse, and as such, it is considerably smaller than 



246 



FIELDIANA: ZOOLOGY 





Fig. 21-9. Ventral view of left otic region and parapterygoid fossa: left, Monticolomys koopmani (FMNH 151900; 
Fianarantsoa Province, RNI d'Andringitra); right, Macrotarsomys bastardi (USNM 578717; Toliara Province, Petriky 
Forest). Abbreviations: ab, auditory bulla (ectotympanic); cc, carotid canal for passage of the internal carotid artery; 
iag, groove marking the course of the infraorbital branch of the stapedial artery; mf, mesopterygoid fossa; mlf, middle 
lacerate foramen; pac, posterior opening to the alisphenoid canal; ppf, parapterygoid fossa; sag, squamosal-alisphenoid 
groove, which indicates the presence of the supraorbital branch of the stapedial artery; sf, stapedial foramen. The 
middle lacerate foramen remains patent in Monticolomys but is ventrally obscured by the inflated ectotympanic bulla 
in Macrotarsomys. 



individuals of either Monticolomys or Macrotar- 
somys bastardi (Table 21-3). Disparity in size, of 
course, is not relevant to generic membership, but 
the condition of other characters does bring into 
question the closeness of the purported phyletic 
affinity. 

Lavocat (1973) stressed similarities in overall 
skull shape and curve of the cranial vault of his 
new genus Protarsomys and Macrotarsomys. 
They both agree, in a general way, in having a 
somewhat delicate rostrum, narrow zygomatic 
span, smoothly contoured braincase, and gently 
arched skull. Although not expanded laterally, the 
arches of Protarsomys over their midsection are 
notably stout for the size of the skull, heavier than 
those observed in the larger Macrotarsomys. The 



comparatively broad interorbit and posteriorly di- 
vergent frontals in Protarsomys convey additional 
resemblance to the cranium of Macrotarsomys 
bastardi; however, possession of wide frontals 
that slightly overhang the rear of the orbit does 
not characterize the other living member of the 
genus, M. ingens. 

Temporal fenestration and definition of a ham- 
ular process appear comparable in the fossil and 
the Recent genera, but the otic capsules of Mac- 
rotarsomys are substantially larger than those of 
Protarsomys. The inflation of the bullae of Pro- 
tarsomys with respect to Monticolomys is difficult 
to ascertain because of the distorted preservation 
in this region, but they seem still smaller, to judge 
from the figured specimen (Lavocat, 1973: Plate 



CARLETON & GOODMAN: ENDEMIC RODENTS 



247 



Table 21-3. Comparison of selected cranial dimen- 
sions (in mm) of the Miocene fossil Protarsomys (cast 
of holotype, KNM 2350) with recent Monticolomys (N 
= 4) and Macrotarsomys (N = 9). For the latter two 
taxa, the mean and range are given. 





Protar- 








somys 






Vari- 


mac- 


Monticolomys 


Macrotarsomys 


ables 


innesi 


koopmani 


b. bastardi 


ONL 


21.6 


27.4(26.3-28.1) 


28.1 (26.7-28.9) 


ZB 


10.8 


13.6(13.1-14.1) 


14.0(13.0-14.6) 


BBC 


9.3 


12.7(12.4-13.2) 


11.3(10.9-11.5) 


IOB 


3.7 


3.9 (3.9-4.0) 


4.6 (4.3-4.7) 


LR 


7.3 


9.5 (8.5-10.2) 


10.2(9.5-11.1) 


BR 


4.1 


4.5 (4.2-4.9) 


4.7(4.4-5.1) 


LD 


5.6 


7.7 (7.2-8.0) 


7.4 (6.8-8.0) 


BMls 


4.4 


5.9 (5.7-6.0) 


5.9 (5.5-6.2) 


BZP 


2.0 


2.2(2.1-2.2) 


2.9(2.6-3.1) 



10). Lavocat identified a stapedial foramen and 
carotid canal in Protarsomys; the presence of oth- 
er cranial foramina and placement of vascular 
tracings indicative of the carotid circulatory plan 
are unknown. The incisive foramina are moder- 
ately long and wide, as in Macrotarsomys and 
Monticolomys, and they terminate posteriorly near 
the level of the anterior roots of the Mis. 

Protarsomys contrasts strikingly with Macro- 
tarsomys and Monticolomys in the architecture of 
its infraorbital canal and accompanying zygomat- 
ic plate. The infraorbital foramen of Protarsomys 
is broadly elliptical or nearly circular. That ob- 
served in Monticolomys and Macrotarsomys re- 
calls the oval "keyhole shape" typical of more 
derived muroids; that is, the ventral portion is nar- 
rowed as a neurovascular fissure relative to the 
spacious dorsal section, which is rounded for pas- 
sage of the anterior body of the masseter medialis. 

Presumably correlated with the different design 
of the infraorbital foramen is the construction of 
the zygomatic plate. In Protarsomys, the plate is 
small, about as long as wide, and marginally in- 
clined above the horizontal; the front edge of the 
plate extends little beyond the superior ramus of 
the anterior zygomatic arch and defines only a 
shallow zygomatic notch. The zygomatic plate in 
examples of Monticolomys and Macrotarsomys 
occupies a greater area, clearly longer (taller dor- 
sad) than wider, and angles noticeably oblique 
(dorsolaterad) to the midsagittal plane of the skull; 
the leading margin of the plate is produced farther 
forward, more so in Macrotarsomys than in Mon- 
ticolomys, creating a deeper zygomatic notch. 
Relative to the molar rows, the zygomatic plate 



of Protarsomys appears displaced posteriad, its 
rear edge coaligned with the anterior root of M 1 , 
a condition that Lavocat (1973) interpreted as ple- 
siomorphic. In contrast, the plates in Monticolo- 
mys and Macrotarsomys are positioned further 
forward on the rostrum, well in advance of the 
front of the Mis. 

We emphasize that the infraorbital foramen of 
Monticolomys and Macrotarsomys, compared to 
many murids and some nesomyines, forms a fairly 
broad oval that lacks pronounced compression of 
the ventral space as a narrow slit distinct from a 
much wider dorsal opening. Nevertheless, the in- 
fraorbital-zygomatic construction of these two 
genera more strongly suggests the typical murid 
condition. The round infraorbital canal and rudi- 
mentary, almost horizontal zygomatic plate of 
Protarsomys, on the other hand, resemble the de- 
velopment found in other contemporaneous Mio- 
cene Afrocricetodontinae, such as Afrocricetodon 
and Notocricetodon (Lavocat, 1973). 

Lavocat (1973) noted two principal differences 
between the mandibles of Protarsomys and Mac- 
rotarsomys. No lateral protuberance marks the 
posterior limit of the alveolus of the lower incisor 
in Protarsomys, whereas a small but distinct cap- 
sular process occurs on the ascending ramus of 
Macrotarsomys (including M. ingens and Monti- 
colomys). Furthermore, the apex of the masseteric 
crest in Protarsomys reaches only as far forward 
as below the middle of ml; in Macrotarsomys and 
Monticolomys, the crest passes beyond the ante- 
rior root of ml to end just behind the mental fo- 
ramen. 

All three genera possess opisthodont upper in- 
cisors and very low-crowned, cuspate molars that 
retain longitudinal enamel connections (mures 
and murids). Their dentitions contrast substantial- 
ly in other particulars. The molar tubercles in ex- 
amples of Macrotarsomys and Monticolomys are 
relatively thickly enameled, their borders round- 
ed, and their occlusal topography bunodont. Cor- 
responding to the robust cusp development, the 
enamel folds and valleys appear narrower and 
more crevice-like. In Protarsomys, the principal 
cusps are less massive, forming tapered conical 
peaks, the enamel seemingly thinner (Lavocai 
[1973] characterized their form as pinched ante 
riorly-posteriorly or somewhat selenodont); tht 
valleys between major cones are more open anc 
their bottoms U-shaped. The anterocone of Mi 
forms a broad, occasionally bilobed cusp in Ma 
crotarsomys and Monticolomys; that of Protar 
somys is narrower, without suggestion of an an 



248 



FIELDIANA: ZOOLOG" 



teromedian sulcus or bilobation. An anteroconid 
is weakly expressed in all three taxa, but that of 
Protarsomys is physically discrete, a distinct me- 
taflexid, equal in size to the contralateral proto- 
flexid, incising its posteromedial margin from the 
nearby metaconid. Posterolophs and companion 
posterofiexi are eminently discernible on the 
M 1 -2s of Protarsomys; these enamel features are 
imperfectly expressed or vestigial in specimens of 
Macrotarsomys and Monticolomys. The upper and 
lower third molars of Protarsomys are larger rel- 
ative to the length of the toothrow. In particular, 
the m3 of the Miocene fossil exhibits a small but 
distinct entoconid, posterolophid, and posteroflex- 
id — features that are indistinct or lost on the re- 
duced m3 talonids of Macrotarsomys and Monti- 
colomys. Finally, paramount among the dental 
structures mentioned by Lavocat (1973) Protar- 
somys displays low and thin mesolophs(ids) on its 
first two molars of the upper and lower toothrows. 
Such transverse enamel ridges do not occur in in- 
dividuals of the two living genera. 



Discussion 

Comments on genealogical relations and his- 
torical biogeography of Monticolomys are neces- 
sarily preliminary in nature, a caveat that under- 
scores the meager number of specimens and field 
data as yet available for fuller appreciation of its 
morphology, ecology, and geographical distribu- 
tion. Small sample sizes, for instance, hinder de- 
finitive characterization of dental structures and 
recognition of qualitative differences, if they ex- 
ist, between Monticolomys and species of Macro- 
tarsomys. Treatment of their relationship within a 
rigorous context is hampered by the lack of res- 
olution surrounding our understanding of higher- 
level phylogeny within Muroidea and, most im- 
portant for our topic, the corroboration of Neso- 
myinae as monophyletic and a working hypothe- 
sis of generic interrelationships. In view of these 
uncertainties, the formal diagnosis of Monticolo- 
mys was broadly stated with reference to all seven 
other nesomyine genera. Those diagnostic char- 
acters, as well as others mentioned below, and the 
rationale for their polarity are elaborated by 
Hershkovitz (1962), Bugge (1970), Carleton 
(1980, 1994), Voss (1988, 1993), and Carleton 
and Musser (1989). 



Relationships 

The extant species of Neosomyinae have long 
been arranged in only seven genera, each one 
highly distinctive and all remarkable for the ex- 
treme scope of intergeneric morphological diver- 
gence (Ellerman, 1941; Petter, 1972; Carleton & 
Musser, 1984). The striking variety of Madagas- 
car's native rodents is attested by the various com- 
parisons drawn, albeit not always fittingly, to their 
outward appearances — the genera being likened 
to rabbits (Hypogeomys), gerbils (Macrotarso- 
mys), voles (Brachyuromys), Old World murines 
(Gymnuromys), or generalized cricetines (Neso- 
mys). Just as remarkably, the seven genera were 
thought, until recently, to contain among them just 
ten species (Petter, 1975; Honacki et al., 1982; 
Rakotondravony, 1987; Corbet & Hill, 1991). 
However, the portrayal of Madagascar's endemic 
rodents as depauperate is, at least in part, a spe- 
cious conclusion based on inadequate biological 
survey and uncritical taxonomic study (see Carle- 
ton & Schmidt, 1990). The number of species ac- 
knowledged based on recent taxonomic revisions 
totals 16 (Musser & Carleton, 1993; Carleton, 
1994), and now 17, with the diagnosis of Monti- 
colomys koopmani. Macrotarsomys, named by 
Milne Edwards and Grandidier (1898) almost a 
century ago, was the last nesomyine genus here- 
tofore described. 

The existence of a few strongly differentiated 
genera (seven) embracing nearly as few species 
(ten), as the Malagasy rodents were so long un- 
derstood, certainly influenced Ellerman 's (1941, 
1949) systematic treatment of the Nesomyinae. 
Indeed, Ellerman dismissed the evolutionary re- 
ality of the subfamily and redistributed the genera 
among five family-group taxa, several of them de- 
scribed as new within an inclusive Family Muri- 
dae: Arvicolinae, Brachytarsomyes (Brachytar- 
somys); Cricetinae (Hypogeomys, Macrotarsomys, 
and Nesomys); Gymnuromyinae (Gymnuromys); 
Murinae, Eliuri (Eliurus); and Tachyoryctinae, 
Brachyuromyes (Brachyuromys). His taxonomy 
controverted the conventional viewpoint on ne- 
somyines — that is, a monophyletic group arising 
from a single immigrant stock and subsequently 
radiating in isolation (e.g., as classically narrated 
by Simpson, 1945) — but in doing so, he did cite 
characters and provide diagnoses for his reclas- 
sifications. Ellerman 's (1941, 1949) ideas have yet 
to be seriously challenged as an alternative to the 
traditional arrangement. 

The characterization of Monticolomys, howev- 



CARLETON & GOODMAN: ENDEMIC RODENTS 



249 



er, departs from the prevalent pattern of extreme 
morphological differentiation and obscure kinship 
alliances so far apparent among nesomyine gen- 
era. That is to say, several cranial and dental char- 
acters clearly implicate a sister-group relationship 
between the new genus and a previously de- 
scribed form, namely Macrotarsomys, suggesting 
that the two share a more recent common ancestor 
relative to the cladogenesis of the six other ne- 
somyine genera. Possible synapomorphies include 
reduction of the tegmen tympani and subsequent 
loss of contact with the squamosal, loss of the 
alisphenoid struts, marginal definition of the pos- 
terolophs, absence of mesolophs(ids), and a com- 
parable degree of reduction of the upper and low- 
er third molars. 

To be sure, many more resemblances of Mon- 
ticolomys and Macrotarsomys involve traits that 
are plausibly considered as primitive or those 
whose evolutionary polarity is equivocal. Among 
these are: size and extent of the incisive foramina; 
ovate infraorbital canal; forward position and dor- 
solateral orientation of the zygomatic plate; short, 
simple palate unmarked by corrugations or pos- 
terolateral palatal pits; wide, shallow paraptery- 
goid fossae; mesopterygoid fossa comparatively 
narrow, perforated by spacious sphenopalatine va- 
cuities; postglenoid and subsquamosal foramina 
present, defining a slender hamular process; 
smooth braincase without beading and ridging; a 
primitive orbitofacial blood supply (presence of 
vascular grooves across squamosal-alisphenoid 
and pterygoid; sphenofrontal and stapedial foram- 
ina patent); and development of a capsular pro- 
cess on the dentary. Dental similarities between 
the two genera are equally finely molded: tooth- 
rows anteriorly divergent; low molar crowns and 
bunodont form of their cusps; cusp position and 
retention of interconnecting, low-relief mures(ids); 
undivided anterocone; indistinct formation of an 
anteroconid and the virtual absence of a metaflex- 
id contrasted to the deep protoflexid; and the num- 
ber of roots on the molars. 

Considered by themselves, the few derived 
character states may seem insufficient to posit sis- 
ter-group stature between Monticolomys and Mac- 
rotarsomys. Nor, we acknowledge, are most or 
any of these character-state transformations, as ar- 
ticulated, unique within Muroidea. At the same 
time, and notwithstanding the weakness of infer- 
ring kinship on the basis of shared plesiomor- 
phies, one cannot help but be impressed by the 
utter exactitude and number of their presumably 
ancestral resemblances. In weighing the five syn- 



apomorphic features against the otherwise fun- 
damental likeness of the genera in so many as- 
pects of their craniodental morphology, we be- 
lieve that a hypothesis of cognate affinity warrants 
attention at this formative stage of phylogenetic 
understanding among nesomyines (Fig. 21-10). 

Sharply countershaded and bright pelage col- 
ors, silky fur texture, large pinnae, elongate pes, 
penicillate tail, and moderately inflated bullae are 
some of the superficial resemblances that have 
earned Macrotarsomys the sobriquet of the Mad- 
agascar gerbil (for instance, Webb, 1954). Among 
nesomyines, both M. bastardi Milne Edwards and 
Grandidier (1898) and M. ingens Petter (1959) 
possess these characteristics and other derived 
features that suggest their close kinship and vin- 
dicate their union in the genus Macrotarsomys 
(Fig. 21-10). 

The muted pelage tones with agouti-banded 
ventral hairs, broader hind foot and relatively long 
lateral digits, well-developed ungual tufts, and 
conformation of the plantar pads in Monitcolomys 
suggest the physique of a scansorial forest-dwell- 
er. Such characteristics provide substantial exter- 
nal contrast to species of Macrotarsomys, in ad- 
dition to cranial differences in the anterior xygo- 
matic region and auditory bullae (Fig. 21-10). Ac- 
cordingly, the diagnostic traits we have advanced 
for Monticolomys are unique in combination and 
make its recognition unambiguous. Nevertheless, 
among the characters surveyed, the identification 
of wholly autapomorphic features that define 
Monticolomys is less clear-cut. 

The apparent lack of uniquely derived character 
states poses the question of whether the construc- 
tion of Monticolomys is paraphyletic with regard 
to Macrotarsomys. Other than acknowledging the 
issue, a firm answer cannot be immediately or 
simply mustered without undertaking a detailed 
phylogenetic review of Nesomyinae. We are per- 
suaded, nonetheless, that generic isolation of 
Monticolomys koopmani is both a reasonable tax- 
onomic hypothesis at this stage of investigation 
and a defensible one, as based on the number and 
kinds of morphological differences so far uncov- 
ered that set it apart from Macrotarsomys (Fig. 
21-10). In this regard, the amount of anagenetic 
change that distinguishes M. koopmani from spe- 
cies of Macrotarsomys far surpasses that observed 
within other genera (such as Eliurus and Neso- 
mys) that contain different species inhabiting dry 
western versus humid eastern biomes of Mada- 
gascar (Carleton, 1994, and unpubl. data). Mor- 
phologically and ecologically, Monticolomys ex- 



250 



FIELDIANA: ZOOLOGY 



M° n 



tic 



o\° 






Macrotarsomys 



i(i9 



e" 1 ' 



1. Large, pliant pinnae 

2. Tail moderately penicillate 

3. Dorsal-ventral pelage contrast 

4. Ungual tufts reduced 

5. Plantar pads papillate 

6. Hairs on plantar surface 

7. Lateral pedal digits short 

8. Metatarsals long, appressed 

9. Zygomatic plate broad, notch deep 

10. Lacrimal tabular process large 

11. Incipient supraorbital shelf 

12. Sigmoid notch compressed 

13. Ectotympanic bullae inflated 
Middle lacerate for. obscured 
Tegmen tympani reduced 
Loss of alisphenoid strut 
Mesolophs absent 

18. Posteroloph vestigial 

19. Reduction of 3rd molars 




Fig. 21-10. Distribution of presumptive synapomorphies (solid rectangles) and hypothesized cladistic relationship 
between Monticolomys koopmoni and species of Macrotarsomys. Half-filled rectangles denote intermediate character 
states. 



hibits many of the characteristics that one might 
postulate for an ancestor to Macrotarsomys. The 
generic validity of Monticolomys and its proposed 
relationship to Macrotarsomys should be subject- 
ed to expanded character analyses and cladistic 
scrutiny, drawing upon representatives of all ne- 
somyine genera as well as other intelligently se- 



lected outgroups within Muroidea, and assimilat- 
ing information from additional anatomical sys- 
tems, karyology, and genetics. 

Whatever the outcome of such future phylo- 
genetic investigation, characteristics of the Mio- 
cene form Protarsomys, as reviewed herein, sub- 
vert neither the definition of Monticolomys nor the 



CARLETON & GOODMAN: ENDEMIC RODENTS 



251 



recognition of Macrotarsomys as a genus apart. 
Many resemblances between Protarsomys and 
Macrotarsomys that so impressed Lavocat (1973) 
are aspects of shape and form, features that do not 
simply lend themselves to formulation of char- 
acter states and argument of polarity. Other gen- 
eral likenesses — for instance, a brachyodont den- 
tition, presence of mures(ids), and a rudimentary 
anteroconid — are plausibly interpreted as shared 
primitive conditions that do not necessarily pro- 
vide support for their close relationship. More- 
over, the strict anatomical correspondence of sim- 
ilarities so generally stated is less persuasive when 
examined in detail (such as anteroconid formation 
and cuspate topography of the brachyodont 
crown). When comparison is restricted to quali- 
tative characters, one is struck more by the many 
and ample dissimilarities: configuration of the in- 
frarorbital canal; size, position, and orientation of 
the zygomatic plate; inflation of the auditory bul- 
lae; formation of the anteroconid and definition of 
the metafiexid; occurrence of mesolophs and me- 
solophids, posterolophs and posteroflexi; and de- 
gree of reduction of the third molars. The union 
of the fossil species Protarsomys macinnesi La- 
vocat (1973) with recent Macrotarsomys species 
cannot be construed on such profound differences 
and the want of any contravening synapomorphic 
traits. 

Lavocat (1973, 1978), it should be underscored, 
did not himself advocate the synonymy of Pro- 
tarsomys under Macrotarsomys. That action was 
promulgated by Chaline et al. (1977) and reiter- 
ated by Petter (1990), without discussion of char- 
acter data justifying the new combination. Instead, 
Lavocat (1973, 1978) drew attention to their like- 
nesses and proposed the descent of Nesomyinae 
from a Miocene, eastern African form such as 
Protarsomys. He retained the latter as a genus of 
Afrocricetodontinae, Family Cricetodontidae, set 
apart from a broadly defined Family Nesomyidae, 
in which he included Nesomyinae and several 
other archaic African groups (see comments in 
Carleton & Musser, 1984). In arguing against the 
generic equivalency of Protarsomys and Macro- 
tarsomys, we do not necessarily discount Lavo- 
cat's (1973, 1978) basic thesis with respect to the 
taxonomic and geographical origin of nesomyi- 
nes, yet neither can we, at this stage, identify per- 
suasive character information that would support 
it. 

Like other early Miocene muroids, Protarso- 
mys displays a somewhat generalized morpholo- 
gy, but from such evidence, one can argue no 



more strongly for its phyletic association with ne- 
somyines than with another African group such as 
the petromyscines. Continued exploration of Ter- 
tiary fossil sites in eastern Africa may disclose 
new cricetodontids whose relationship to neso- 
myines is incontestable. The discovery of relevant 
Miocene-Pliocene fossils in Madagascar remains 
a possibility. Fresh examination of the original 
type series of Protarsomys could feasibly reveal 
new characters that would bear critically on the 
question of its relationship to nesomyines. The 
problem is addressable and amenable to improved 
resolution. 



Biogeography 

In a phytogeographic classification of the floral 
associations of Madagascar, Humbert (1955) rec- 
ognized a unique High Mountain Domain that oc- 
curs patchily on a few isolated peaks between An- 
dohohela in the south and Tsaratanana in the 
north. Two of the mountaintop communities fall- 
ing within this domain are found on the massifs 
of Ankaratra and Andringitra. Upper montane for- 
ests (above approximately 1500 m) on these two 
massifs, particularly in the zone just below the 
tree line, share many endemic sclerophyllous 
plants, principally members of the family Erica- 
ceae (Koechlin, 1972; Koechlin et al., 1974). Fur- 
thermore, the fragmented distributions of numer- 
ous animal species — including, for example, ter- 
restrial insects (Paulian, 1961), aquatic insects 
(Chapter 9, this work), and several amphibians 
and reptiles (Chapter 17) — are restricted to the 
upper reaches of these central and south-central 
highlands. So far as is currently known, the dis- 
tributional pattern of the nesomyine rodent Mon- 
ticolomys koopmani also conforms to Humbert's 
(1955) High Mountain Domain (Fig. 21-11). 

The present-day discontinuity in the ranges of 
all these organisms and the reiterative pattern of 
vicariance strongly indicate that a corridor of up- 
per montane vegetation once connected the sum- 
mits of Ankaratra and Andringitra. On the basis 
of modern topography, feasible links between 
these two massifs may have occurred along the 
Fianarantsoa Plateau, between 1000 and 1300 m, 
and through the Ambalavao Pass, between 800 
and 1000 m (Raxworthy & Nussbaum, in press). 
Thus, depression of upper montane zones to as 
low as 1000 m would have been necessary in or- 
der for animals such as Monticolomys to become 



252 



FIELDIANA: ZOOLOGY 




Fig. 21-11. Southeastern sector of Madagascar's Central High Plateau with reference to the two known geographic 
occurrences of Monticolomys koopmani: 1) Manjakatompo, about 19°20'S 47°26'E; 2) 38 km S Ambalavao, RNI 
d'Andringitra, 22°1 1'S 46°58'E. Stippled areas indicate uplands above 1500 m; this elevation approximates the pres- 
ent-day lower boundary of upper montane vegetation. The dot-and-dashed line corresponds to the 1000-m contour, 
the hypothesized lowest depression of the upper montane-sclerophyllous zone during the Holocene (see the Discussion 
section). 



distributed continuously over these mountains 
(Fig. 21-11). 

Evidence for cooler periods and consequent 
lowering of vegetation zones is available for the 
Quaternary, when Madagascar experienced cycli- 
cal climatic shifts (Burney, in press). Pollen cores 



from Lake Tritrivakely, near Antsirabe (80 km S 
Ankaratra and 250 km N Andringitra) on the Cen- 
tral Plateau (Fig. 21-11), reveal considerable vari- 
ation in the natural vegetational communities of 
the area over the past 36,000 years (Burney, 1987; 
Gasse et al., 1994). During both the Pleistocene 



CARLETON & GOODMAN: ENDEMIC RODENTS 



253 



and early Holocene, ericaceous pollen generally 
accounted for over 50% of the taxa deposited in 
the lacustrine sediments, but after the middle Ho- 
locene this proportion dropped to between and 
30%. During portions of the Quaternary, high 
mountain floral communities likely descended to 
approximately 1000 m (Burney, in press), an el- 
evation sufficiently low for upper montane forest 
and sclerophyllous vegetation to have extended 
between the regions of Ankaratra and Andringitra. 
Some time in the middle Holocene, the upper 
montane and sclerophyllous floras retreated to the 
uppermost slopes of the mountains and severed 
the geographic ranges of animals dependent on 
such habitats. 

In summary, information derived from paly- 
nology and from the congruent distributions of 
other plants and animals convincingly explains 
the present-day vicariant occurrence of Montico- 
lomys koopmani on the Andringitra and Ankaratra 
massifs. Specimen-based documentation of neso- 
myine distributions, in general poor over much of 
Madagascar, is almost nonexistent for the island's 
high mountain zones, particularly those on such 
huge mountain systems as Tsaratanana and Ma- 
rojejy in the north and Andohohela in the south. 
Whether the distributions of other nesomyine ro- 
dents repeat a similar geographic pattern is there- 
fore uncertain, but Eliurus majori, as its broken 
range is currently understood, is a possible can- 
didate (see Carleton, 1994; and Chapter 22, this 
work). 



Note added in proof 

Monticolomys koopmani still occurs on the An- 
karatra Massif. During a biological inventory of 
the Foret de Nosiarivo (2000 m) in February 
1996, conducted by S. M. Goodman and D. Rak- 
otondravony, a specimen of Monticolomys koop- 
mani was captured (FMNH 156211). The animal 
was trapped on the ground and in degraded habitat 
close to the forest edge. 



Acknowledgments 

We thank Louis L. Jacobs, Southern Methodist 
University, for loaning the casts of Protarsomys 
from his personal research collection. Access to 
comparative material and/or holotypes of Macro- 
tarsomys was provided by Guy G. Musser 



(AMNH) and Francis Petter and Michel Tranier 
(MNHN). The morphological descriptions and 
comparisons related in our paper have benefited 
substantially from the hind foot drawings and cra- 
niodental photographs carefully prepared by Da- 
vid F Schmidt. Peter Goldberg photographed the 
museum skins. We are grateful to Guy G. Musser 
and Bruce D. Patterson for their critical comments 
on the penultimate version of the manuscript. 
Last, the first author acknowledges his deep grat- 
itude to Karl F. Koopman and Guy G. Musser, 
AMNH, for their unselfish encouragement to pur- 
sue the description of Rand's Manjakatompo 
specimen, particularly because each had already 
devoted sufficient time and effort to the problem 
to appreciate its systematic contribution. 



Literature Cited 

Archbold, R. 1932. A new lemur from Madagascar. 

American Museum Novitates, 518: 1. 
Bugge, J. 1970. The contribution of the stapedial artery 

to the cephalic arterial supply in muroid rodents. Acta 

Anatomica, 76: 313-336. 
Burney, D. A. 1987. Pre-settlement vegetation changes 

at Lake Tritrivakely, Madagascar. Palaeoecology of 

Africa, 18: 357-381. 

-. In press. Theories and facts regarding Holo- 



cene environmental change before and after human 
colonization. In Goodman, S. M., and B. D. Patterson, 
eds., Natural and Human-Induced Change in Mada- 
gascar. Smithsonian Institution Press, Washington, 
DC. 

Carleton, M. D. 1980. Phylogenetic relationships in 
neotomine-peromyscine rodents (Muroidea) and a re- 
appraisal of the dichotomy within New World Cricet- 
inae. Miscellaneous Publications of the Museum of 
Zoology, University of Michigan, 157: 1-146. 

. 1994. Systematic studies of Madagascar's en- 
demic rodents (Muroidea: Nesomyinae): Revision of 
the genus Eliurus. American Museum Novitates, 
3087: 1-55. 

Carleton, M. D., and G. G. Musser. 1984. Muroid 
rodents, pp. 289-379. In Anderson, S., and J. K. 
Jones, Jr., eds., Orders and Families of Recent Mam- 
mals of the World. John Wiley and Sons, New York, 
686 pp. 

. 1989. Systematic studies of oryzomyine ro- 
dents (Muridae, Sigmodontinae): A synopsis of Mi- 
croryzomys. Bulletin of the American Museum of Nat- 
ural History, 191: 1-83. 

Carleton, M. D., and D. F. Schmidt. 1990. Systematic 
studies of Madagascar's endemic rodents (Muroidea: 
Nesomyinae): An annotated gazetteer of collecting lo- 
calities of known forms. American Museum Novita- 
tes, 2987: 1-36. 

Ch aline, J., P. Mein, and F. Petter. 1977. Les grandes 
lignes d'une classification evolutive des Muroidea. 
Mammalia, 41: 245-252. 



254 



FIELDIANA: ZOOLOGY 



Corbet, G. B., and J. E. Hill. 1991. A World List of 
Mammalian Species, 3rd edition. Natural History Mu- 
seum Publications, London, vii + 243 pp. 

Delacour, M. J. 1932. Les oiseaux de la Mission Zoo- 
logique Franco-Anglo-Americaine a Madagascar. 
L'Oiseau et la Revue de Francaise d'Ornithologie, 
nouvelle serie, 2: 1-96. 

Denys, C. 1987. Rodentia and Lagomorpha, pp. 118- 
170. In Leakey, M. D., and J. H. Harris, eds.. Laetoli, 
a Pliocene site in Tanzania. University of Oxford 
Press, London. 

Ellerman, J. R. 1941. The Families and Genera of Liv- 
ing Rodents. Volume 2, Family Muridae. London: 
British Museum (Natural History), xii+690 pp. 

. 1949. The Families and Genera of Living Ro- 
dents. Volume 3, Appendix II [Notes on the rodents 
from Madagascar in the British Museum, and on a 
collection from the island obtained by M. C. S. 
Webb]. London: British Museum (Natural History), 
v+210 pp. 

Gasse, E, E. Cortijo, J.-R. Disnar, L. Ferry, E. Gil- 
bert, C. Kissel, F. Laggoun-Defarge, E. Lallier- 
Verges, J.-C. Miskovsky, B. Ratsimbazafy, F. Ran- 

AIVO, L. ROBISON, P. TUCHOLKA, J.-L. SAOS, A. SlFFED- 

ine, M. Taieb, E. Van Campo, and D. Williamson. 
1994. A 36 ka environmental record in the southern 
tropics: Lake Tritrivakely (Madagascar). Comptes ren- 
dus de l'Academie des Sciences, seYie II, 318: 15 13— 
1519. 

Hershkovitz, P. 1962. Evolution of Neotropical crice- 
tine rodents (Muridae) with special reference to the 
phyllotine group. Fieldiana, Zoology, 46: 1-524. 

Honacki, J. H., K. E. Kinman, and J. W Koeppl. 1982. 
Mammal Species of the World. Allen Press and the 
Association of Systematics Collections, Kansas, 
ix+694 pp. 

Humbert, H. 1955. Les temtores phytogeographiques 
de Madagascar. Leur cartogarphie. Annee Biologique, 
serie III, 31(5-6): 195-204. 

Jacobs, L. L. 1985. The beginning of the age of murids 
in Africa. Acta Zoologica Fennica, 170: 149-151. 

Koechlin, J. 1972. Flora and vegetation of Madagascar, 
pp. 145-190. In Battistini, R., and G. Richard- Vin- 
dard, eds., Biogeography and Ecology in Madagascar. 
W. Junk, The Hague, xv + 765 pp. 

Koechlin, J., J.-L. Guillaumet, and P. Morat. 1974. 
Flore et veg&ation de Madagascar. J. Kramer, Vaduz, 
Germany. 

Lavocat, R. 1973. Les rongeurs du Miocene d'Afrique 
Orientale. I. Miocene infeYieur. M6moires et Travaux 
de l'Ecole Pratique des Hautes Etudes, Institute de 
Montpellier, 1: 1-284. 

. 1978. Rodentia and Lagomorpha, pp. 69-89. 

In Maglio, J. V, and H. B. S. Cooke, eds.. Evolution 
of African Mammals. Harvard University Press, Cam- 
bridge, ix+641 pp. 

Milne-Edwards, A., and G. Grandidier. 1898. De- 
scription d'une espece nouvelle de Murid6 provenant 
de Madagascar. Bulletin de Museum National d'His- 
torie Naturelle, serie 1, 4: 179-181. 

Musser, G. G., and M. D. Carleton. 1993. Family 
Muridae. pp. 501-755. In Wilson, D. E., and D. M. 
Reeder, eds.. Mammal Species of the World: A Tax- 



onomic and Geographic Reference, 2nd edition. 
Smithsonian Institution Press, Washington, D.C., 
xviii+1207 pp. 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: 
Revue de la conservation et des Aires Protegees. 
World Wide Fund for Nature, Gland, Switzerland, 
xvii + 374 pp. 

Paulian, R. 1961. La zoogeographie de Madagascar et 
des iles voisines. Faune de Madagascar, 13: 1-484. 

Perrier de la Bathie, H. 1927. Le Tsaratanana, l'An- 
karatra, et l'Andringitra. M6moire de l'Academie 
Malgache. 3: 1-68. 

Peters, W 1870. Uber Nesomys rufus, eine neue gat- 
tung und art madagascarischer nager. Sitzungs-Beri- 
chte der Gesellschaft naturforsender Freunde, Berlin, 
1870: 54-55. 

Petter, F. 1959. Un nouveau rongeur de Madagascar 
(Nesomyinae): Macrotarsomys ingens nov. sp. Mam- 
malia, 23: 139-148. 

. 1972. The rodents of Madagascar: The seven 

genera of Malagasy rodents, pp. 661-665. In Battis- 
tini, R., and G. Richard- Vindard, eds., Biogeography 
and Ecology in Madagascar. W. Junk, The Hague, 
xv+765 pp. 

1975. Family Cricetidae: Subfamily Nesomyi- 



nae. Part 6.2, pp. 1-4. In Meester, J., and H. W Setzer, 
eds.. The Mammals of Africa: An Identification Man- 
ual. Smithsonian Institution Press, Washington, D.C. 
1990. Relations de parente" des rongeurs de 



Madagascar. Accademie Nazionale dei Lincei, Atti dei 
Convegni Lincei, 85: 829-837. 

Rakotondravony, D. A. 1987. Les rongeurs a Mada- 
gascar, pp. 93-94. In Mittermeier, R. A., L. H. Rak- 
otovao, V. Randrianasolo, E. J. Sterling, and D. De- 
vitre, eds., Priorit^s en matiere de conservation des 
especes a Madagascar. IUCN, Gland, Switzerland, 167 
pp. 

Rand, A. L. 1932. Mission Franco-Anglo- Am^ricaine 
a Madagascar. Notes de voyage. L'Oiseau et la Revue 
de Francaise d'Ornithologie, nouvelle serie, 2: 227- 
282. 

1935. On the habits of some Madagascar mam- 



mals. Journal of Mammalogy, 16: 89-104. 

1936. The distribution and habits of Madagas- 



car birds. A summary of the field notes of the Mission 
Zoologique Franco-Anglo-Am6ricaine a Madagascar. 
Bulletin of the American Museum of Natural History, 
72: 143-499. 
Raxworthy, C. J., and R. A. Nussbaum. In press. Zoo- 
geographic patterns of reptile endemicity and vicari- 
ance in eastern Madagascar. In Goodman, S. M., and 
B. D. Patterson, eds.. Natural and Human-Induced 
Change in Madagascar. Smithsonian Institution Press, 
Washington, D.C. 

Reig, O. A. 1977. A proposed unified nomenclature for 
the enamelled components of the molar teeth of the 
Cricetidae (Rodentia). Journal of Zoology, London, 
181: 227-241. 

Simpson, G. G. 1945. The principles of classification 
and a classification of mammals. Bulletin of the 
American Museum of Natural History, 85: 1-350. 

United States Board on Geographic Names. 1989. 



CARLETON & GOODMAN: ENDEMIC RODENTS 



255 



Gazetteer of Madagascar, 2nd edition. Defense Map- ine" characters. American Museum Novitates, 3073: 

ping Agency, Washington, D.C., xvii+826 pp. 1-44. 

Voss, R. S. 1988. Systematics and ecology of ichthy- Voss ', R : S -. AND L '.J Marcus 1992. Morphological 

omyine rodents (Muroidea): Patterns of morphological ^ volutlon in muroid S?*™* "/ Cnmiometnc factor 

\ .. „ .. ., n „*\. %„. divergence in seven Neotropical genera, with expen- 

^^ i ^ s ^ l ^^^! a ^-^^ Q of A ^ mental results from Zygodontomys. Evolution, 46: 

American Museum of Natural History, 188: 259-493. 1918-1934 

1993. A revision of the Brazilian muroid ro- Webb, C. S. 1954. The odyssey of an animal collector. 



dent genus Delomys with remarks on "thomasomy- Longmans, London, xv+368 pp. 



256 FIELDIANA: ZOOLOGY 



Chapter 22 

The Rodents of the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 

Steven M. Goodman and Michael D. Carleton 



Abstract 

Rodents were surveyed on the previously unstudied eastern side of the Reserve Naturelle 
Intdgrale (RNI) d'Andringitra, between 720 and 1625 m elevation. Ten species were trapped, 
and one more is known from earlier collections made in the reserve. One introduced rodent, 
Rattus rattus, has colonized the reserve, and the balance of species belong to the endemic 
Subfamily Nesomyinae. We present information on identification and distribution, measure- 
ments, and ecology and reproduction for each species. Trap results showed the greatest rodent 
species diversity and numbers to be at 1625 m. Two species, Eliurus minor and Gymnuromys 
roberti, occur across the complete elevational range sampled. Nesomys audeberti and Eliurus 
webbi occur in forests below 810 and Brachyuromys ramirohitra and Monticolomys koopmani 
in forests above 1625 m. The balance of species have relatively broad altitudinal ranges. No 
rodent taxon is known to be endemic to the RNI d'Andringitra, although the recently discovered 
Monticolomys koopmani is apparently restricted to the upper portions of the Andringitra and 
Ankaratra massifs. 

Resume 

Dans la partie est de la Reserve Naturelle Integrate (RNI) d'Andringitra, ou aucun inventaire 
n'avait encore 6\€ effectue\ on a r6alis6 l'inventaire des rongeurs entre 720 m et 1625 m 
d' altitude. Dix especes ont 6t6 capturees et une supptementaire 6tait deja connue de collectes 
pr6c6dentes effectuees dans la reserve. Rattus rattus, espece introduite, a colonist la reserve 
et le reste des especes appartient a la sous-famille end^mique des Nesomyinae. Nous pr^sentons 
des informations sur l'identification, la distribution, les mensurations, 1'ecologie et la repro- 
duction de chaque espece. La plus grande diversity sp6cifique et le plus grand nombre d'in- 
dividus ont €t€ documented a 1625 m d'altitude a 1'aide de pieges. Deux especes, Eliurus minor 
et Gymnuromys roberti, apparaissent sur 1'ensemble de la ligne altitudinale inventorize. On a 
recensd Nesomys audeberti et Eliurus webbi dans les forets localises en-dessous de 810 m 
d'altitude et Brachyuromys ramirohitra et Monticolomys koopmani dans les forets localisees 
au dessus de 1625 m. Le reste des especes a une distribution altitudinale relativement large. 
Aucun taxon de rongeurs n'est end6mique de la RNI d'Andringitra, mais il faut noter la r^cente 
d6couverte de Monticolomys koopmani, qui est apparemment limitee aux parties sup^rieures 
des massifs de 1' Andringitra et 1' Ankaratra. 



GOODMAN & CARLETON: RODENTS 257 



Introduction 

The native rodents of Madagascar, as currently 
understood, comprise a group of 15 species that 
are placed within the endemic Subfamily Neso- 
myinae of a broadly defined family Muridae 
(Carleton & Musser, 1984; Musser & Carleton, 
1993). Nesomyinae contains a remarkable array 
of morphological types yet embraces relatively 
few species, and, if indeed monophyletic (see dis- 
cussions in Petter, 1972, 1990, and Carleton & 
Musser, 1984), the subfamily represents one of the 
more unusual radiations compared to other mu- 
roid groups that have diversified within an insular 
setting (see Carleton & Schmidt, 1990). Species 
diversity, relationships, and geographic limits re- 
main poorly known (Carleton & Schmidt, 1990), 
and revisionary systematic studies have recently 
begun (Carleton, 1994). Petter (1972) and Rako- 
tondravony (1987) offered general reviews of the 
rodents of Madagascar, and Petter (1975) provid- 
ed keys to the genera and species then recognized. 

Although the endemic carnivores and primates 
of Madagascar have long inspired attention from 
biologists, interest in its indigenous rodents has 
emerged only in the past decade. This interest is 
manifested in general faunal inventories (Duck- 
worth & Rakotondraparany, 1990; Barden et al., 
1991; Goodman & Ganzhorn, 1994), elevational 
transects (Ryan et al., unpubl.), review of Holo- 
cene subfossils (Goodman & Rakotondravony, in 
press), and ecological or behavioral studies (Ni- 
coll et al., 1988; Ryan et al., 1993; Stephenson, 
1993). Still, this surge of activity has done little 
to advance basic information about the ecology, 
reproductive biology, and populations of the is- 
land's rodents, even for those forms inhabiting 
more frequently visited areas of the eastern humid 
forest. 

The present study documents nesomyine rodent 
diversity in the Reserve Naturelle Integrate (RNI) 
d'Andringitra, examines species richness and 
numbers along an elevational gradient on the east- 
ern slopes of the reserve, and summarizes basic 
natural history information on these animals. The 
broad range of plants and animals studied during 
this survey affords a unique opportunity to ex- 
amine the biotic interactions between rodents and 
other organisms along the slopes of the Andrin- 
gitra Massif (see Chapter 6). 

Previous Work 

Although there were earlier biological surveys, 
little has been published on the rodents of the RNI 



d'Andringitra per se. From November 1970 to 
January 1971, a group of scientists visited the An- 
dringitra Massif to study the local geomorpholo- 
gy, climate, and vegetational structure (Paulian et 
al., 1971). Participants in this campaign, under the 
name of Recherche Cooperative sur Programme 
(RCP) no. 225, included the mammalogist Roland 
Albignac, then of the ORSTOM laboratory in An- 
tananarivo, and Andre Peyrieras, who assisted in 
the collection of small mammals (see pp. 4-5 for 
a listing of the RCP no. 225 itinerary and locali- 
ties). The expedition members concentrated their 
efforts on the higher elevational zones, and all 
small mammals collected were deposited in the 
Museum National d'Histoire Naturelle (MNHN), 
Paris. 

A complete analysis of the collection of Albig- 
nac and Peyrieras has never been published in its 
entirety, but selected specimens and related com- 
munications have been cited in several works 
(Petter, 1975; MacPhee, 1987; Carleton & 
Schmidt, 1990; Carleton, 1994). Nicoll and Lan- 
grand (1989) also compiled the known mammal 
fauna of the RNI d'Andringitra based on review 
of the literature and their own unpublished rec- 
ords. They listed four rodent species, all of them 
nesomyines: Brachyuromys betsileoensis, B. ra- 
mirohitra, Gymnuromys roberti, and Eliurus 
"myoxinus" (probably E. majori, according to 
Carleton, 1994). 

In 1993, a group of Cambridge University stu- 
dents conducted a small mammal survey on the 
north-central slopes of the reserve in the Foret 
Imaitso and near Anjavidilava (O'Keeffe & Ash- 
more, no date). They listed four rodent species 
from the reserve: two nesomyines (Gymnuromys 
roberti and Eliurus "myoxinus") and two mu- 
rines (Rattus rattus and R. norvegicus). 



Materials and Methods 



This investigation is based on fieldwork con- 
ducted between November 14 and December 18, 
1993 by the senior author (S.M.G.), who is re- 
sponsible for general synthesis of the systematic 
and ecological data. The junior author (M.D.C.) 
assisted with taxonomic determinations and sys- 
tematic comparisons. 



258 



FIELDIANA: ZOOLOGY 



Table 22-1. Summary of trap lines. 



Elevation No. of traps 


Length 
of line 

(m) 


Mean distance 

between traps 

(m) 


Mean height 
above ground 

(m) 


720 m (November 15-21) 








Line 1 50 
Line 2 50 
Total 100 


280 
340 
620 


5.5 ± 3.68 (0-19) 
7.2 ±5.51 (0-29) 
6.4 ± 4.79 (0-29) 


1.6 ±0.69 (0.50-3.0), N = 20 
1.3 ± 0.69 (0.25-2.0), N = 17 
1.5 ± 0.73 (0.25-3.0), N = 37 


810 m (November 22-29) 








Line 3 50 
Line 4 50 
Total 100 


370 
410 
780 


7.6 ± 5.67 (0-24) 
8.4 ±7.03 (1-32) 
8.0 ± 6.43 (0-32) 


1.4 ±0.78 (0.2-3.0), N = 25 

1.5 ± 0.63 (0.5-2.5), N = 23 
1.4 ±0.72 (0.2-3.0), N = 48 


1210 m (December 1-7) 








Line 5 50 
Line 6 50 
Total 100 


340 
375 
715 


7.0 ± 4.47 (2-19) 
7.7 ±5.28 (1-26) 
7.4 ±4.94 (1-26) 


1.8 ±0.70 (0.75-3.0), N = 21 
2.1 ±0.93(1.0-4.0), N = 21 

1.9 ±0.85 (0.75-4.0), N = 42 


1625 m (December 8-15) 








Line 7 50 
Line 8 50 
Total 100 


315 
275 
590 


6.4 ±3.80 (1-18) 
5.6 ±4.45 (1-22) 
6.0 ±4.18 (1-22) 


1.7 ±0.62 (0.5-3.0), N = 15 
2.6 ±2.99 (1.0-15), N = 20 
2.2 ±2.37 (0.5-15), N = 35 



Note: Descriptive statistics are presented as mean ± SD (range). Each line consisted of 10 National and 40 Sherman 
live-traps (see page 259). 



Field Methods and Trapping 
Protocol 

At each of four elevational levels (720, 810, 
1210, and 1625 m), two separate trap lines were 
in operation for a minimum of 5 nights (Table 
22-1). Each trap line, numbered sequentially start- 
ing with the 720 m zone, consisted of 50 stations: 
40 Sherman live traps (9 X 3.5 X 3 in.) and 10 
National live traps (16 X 5 X 5 in.). Traps were 
baited daily, generally between 1500 and 1700 
hours, with a fresh mixture of finely ground pea- 
nut butter and oat grain mixed in proportions to 
make a paste. Pitfall traps were installed in each 
elevational zone (see Chapter 17); their use re- 
sulted in the capture of numerous insectivores 
(see Chapters 19 and 20), but few rodents. 

Traps were visited at least twice per day, once 
at dawn and in the late afternoon. A "trap-day" 
is defined as one trap in use for a 24-hour period 
(dawn to dawn). Captured animals were prepared 
as either standard museum skins with associated 
skulls and skeletons, fluid-preserved carcasses, or 
full skeletons. Many tissue samples were frozen 
in liquid nitrogen for biochemical studies and the 
viscera preserved in alcohol for endoparasite re- 
search. Whole carcasses were wrapped in a fine 
cheesecloth, before immersion in formalin, to pre- 
vent loss or mixing of ectoparasites (see Chapter 
12). 



In order to quantify differences in the spatial 
distribution of small mammal captures, several 
trapping variables were systematically recorded: 
(1) type of trap; (2) total length of trap line; (3) 
distance between traps; and (4) specific placement 
of each trap, including its substrate, surrounding 
forest structure, and position on or height above 
the ground. 

The precise location of each trap (variable 4) 
was categorized by microhabitat site, whether trap 
placement was on or above the ground (angles are 
relative to the ground), as follows: 

On Ground — 1', in leaf litter; 2', in leaf litter 
by hole; 3', under downed tree (not decomposed); 
4', under downed tree (at least partially decom- 
posed); 5', by tree root; 6', by tree root with cav- 
ity or hole; 7', by at least partially decomposed 
tree trunk or root with hole or cavity; 8', on 
ground (not fitting other categories); 9', stream 
edge (at or less than 0.5 m from edge). 

Above Ground — 1", on liana, limb, or trunk 
less than 5 cm circumference in horizontal to 15° 
position; 2", on liana, limb, or trunk between 5 
and 10 cm circumference in horizontal to 15° po- 
sition; 3", on liana, limb, or trunk greater than 10 
cm circumference in horizontal to 15° position; 4", 
on liana, limb, or trunk less than 5 cm on circum- 
ference at 15° to 45° angle; 5", on liana, limb, or 
trunk between 5 and 10 cm circumference at 15° 
to 45° angle; 6", on liana, limb, or trunk greater 



GOODMAN & CARLETON: RODENTS 



259 



than 10 cm circumference at 15° to 45° angle; 7", 
on liana, limb, or trunk less than 5 cm circumfer- 
ence at 45° to 90° angle; 8", on liana, limb, or 
trunk between 5 and 10 cm circumference at 45° 
to 90° angle; 9", on liana, limb, or trunk greater 
than 10 cm circumference at 45° to 90° angle; 10", 
on downed tree (not decomposed); 11", on 
downed tree (at least partially decomposed); 12", 
on exposed rock; 13", in tree hole. 



Specimens and Measurements 

A total of 148 rodent specimens were preserved 
as vouchers during the field census of the RNI 
d'Andringitra conducted by S.M.G. This material 
is housed in the Field Museum of Natural History 
(FMNH), Chicago, and a representative series will 
be returned to the Departement de Biologie Ani- 
mate, Universite d' Antananarivo. To amplify 
specimen-based documentation for the RNI 
d'Andringitra, the rodents collected under the Re- 
cherche Cooperative sur Programme (RCP) and 
now stored in the MNHN, Paris are also listed. 
To confirm taxonomic identifications, nesomyine 
holdings in other museums (see Appendix) were 
also consulted, and one or both authors have ex- 
amined the holotypes of all described forms of 
Nesomyinae, except for Peters' (1870) Nesomys 
rufus. 

Six measurements, in millimeters (mm) or 
grams (g), were taken by S.M.G. for each speci- 
men in the flesh; their abbreviations and defini- 
tions are given below. 

TOTL, total length of body and tail: from the 
tip of the nose to the end of the last caudal 
vertebra (not including terminal hair tuft). 

HBL, head and body length: from the tip of the 
nose to the distalmost point of the body (at 
base of tail). 

TL, tail length: from the base of the tail (held 
at right angles to the body) to the end of the 
last caudal vertebra (not including terminal 
hair tuft). 

HFL, hind foot length: from the heel to the tip 
of the longest toe (not including claw). 

EL, ear length: from the basal notch to the dis- 
tal tip of the pinna. 

WT, weight in grams: measured with Pesola 
spring scales, to ± 0.5 g for animals less than 
100 g and to ± 1.0 g for those between 101 
and 300 g. 



Sixteen cranial and two dental dimensions were 
measured by M.D.C. to the nearest 0.1 mm using 
handheld digital calipers accurate to 0.03 mm. 
These measurements, and their abbreviations, fol- 
low the anatomical landmarks defined and illus- 
trated by Carleton (1994): BBC, breadth of the 
braincase; BIF, breadth of incisive foramina; 
BMls, breadth of the bony palate across the first 
upper molars; BOC, breadth across the occipital 
condyles; BR, breadth of rostrum; BZP, breadth 
of the zygomatic plate; DAB, depth of the audi- 
tory bulla; IOB, interorbital breadth; LBP, length 
of bony palate; LD, length of diastema; LIF, 
length of the incisive foramina; LM1-3, coronal 
length of maxillary toothrow; LR, length of ros- 
trum; ONL, occipitonasal length; PPB, posterior 
breadth of the bony palate; PPL, postpalatal 
length; WM1, width of the first upper molar; and 
ZB, zygomatic breadth. 

Standard descriptive statistics (mean, range, 
standard deviation) were derived for adult speci- 
mens in each species sample. We define "adult" 
as the age cohort consisting of animals that lack 
the finer, juvenile pelage and that possess fully 
erupted, though sometimes unworn, third molars. 
Where sample sizes permitted, two-sample t tests 
and one-way analyses of variance (ANOVA) were 
applied to the mensural variables, with gender as 
the categorical variable. For some taxa, patterns 
of morphometric differentiation were summarized 
using principal component analysis, extracted 
from the variance-covariance matrix and based on 
natural log transformations of the 1 8 craniodental 
variables. Analytical routines were carried out us- 
ing Systat (version 5.05, 1994). Mammae formu- 
lae are presented as the number of paired post- 
axial, abdominal, or inguinal teats. 



Accounts of Species 

Family Muridae: Subfamily Murinae 

Rattus rattus (Linnaeus, 1758) 

Identification — In general, all external mea- 
surements (Table 22-2) fall within the known 
range of Malagasy populations of Rattus rattus 
(Rakotondravony, 1992). No sexual dimorphism 
is discernible in our limited series, although other 
studies have found males to be larger than females 
in body measurements (e.g., King, 1990). 

Ecology and Reproduction — During the 1993 



260 



FIELDIANA: ZOOLOGY 



Table 22-2. External measurements and sample statistics 


of adult rodents from RNI d'Andringitra. 


Species 


N 


TOTL 


HBL 


TL 


HFL 


EL 


WT 


Rattus rattus 


10 


364.4 


167.0 


194.8 


32.8 


24.7 


105.7 






17.2 


8.0 


11.2 


1.0 


0.7 


13.1 






340-388 


151-178 


176-212 


31-34 


24-26 


86-134 


Brachyuromys betsileoensis 


14 


222.5 


147.1 


76.4 


30.6 


19.3 








16.6 


11.0 


7.5 


1.3 


1.0 








200-252 


125-165 


70-92 


29-34 


17-21 




B. ramirohitra 


3 


252.0 


155.0 


95.3 


32.3 


22.7 


90.5 






222-282 


140-165 


84-110 


30-34 


21-24 


64-117 


Eliurus majori 


23 


353.6 


157.9 


190.4 


28.8 


19.6 


97.2 






15.6 


6.6 


10.4 


1.1 


0.7 


12.9 






315-376 


145-171 


170-207 


27-31 


18-21 


77-122 


E. minor 


25 


239.6 


112.8 


128.9 


21.4 


17.6 


35.1 






9.5 


6.1 


5.1 


1.0 


0.7 


5.8 






220-262 


101-124 


119-137 


20-23 


17-19 


21.5-49.5 


E. tanala 


24 


333.1 


152.0 


178.3 


29.9 


23.6 


81.7 






12.0 


4.7 


10.1 


1.3 


0.9 


7.9 






307-350 


140-159 


152-194 


27-33 


22-25 


66-97.5 


E. webbi 


19 


306.0 


138.3 


165.7 


28.6 


22.7 


70.1 






14.5 


13.9 


10.1 


1.1 


0.9 


9.7 






286-331 


105-161 


150-183 


27-31 


21-24 


54-90 


Gymnuromys roberti 


4 


335.0 


161.5 


168.0 


34.7 


22.3 


107.7 






310-353 


145-168 


155-176 


34-36 


21-23 


73-124 


Monticolomys koopmani 


3 


236.3 


98.0 


138.0 


24.3 


18.3 


26.1 






234-240 


94-101 


134-143 


24-25 


18-19 


25-27 


Nesomys audeberti 


1 


381 


185 


180 


50 


30 


182 


N. rufus 


31 


343.5 


179.4 


158.6 


43.7 


26.1 


158.7 






9.2 


9.0 


6.1 


1.0 


0.9 


16.6 






323-360 


160-196 


148-171 


42-46 


24-28 


126-208 


Note: Values are means ± 


SD (range). 















survey, this murine species, which was introduced 
to Madagascar, was captured only in forested 
zones between 810 and 1625 m. Three specimens 
(MNHN) were earlier obtained by R. Albignac 
during the RCP survey at Ambalamarovandana 
(1530 m), Anjavidilava (1995 m), and Ivango- 
mena (2400-2500 m). Thus, Rattus rattus has an 
elevational distribution across nearly the complete 
altitudinal range of the massif. 

This species was not trapped within the 720 m 
zone, which was near a small human settlement 
and partially disturbed forest. The number of Rat- 
tus rattus trapped increased as a function of alti- 
tude; there were no captures at 720 m, one at 810 
m, four at 1210 m, and six at 1625 m (regression, 
P = 0.009, r = 0.99, N = 4). This positive cor- 
relation also reflects increased distances from the 
nearest cleared forest. Cranial remains of this rat 
were found in Cryptoprocta ferox scats collected 
in the heathland zone between 2000 and 2100 m 
elevation. 

Individuals of Rattus rattus were captured in a 



variety of trap placements. Seven of 11 (64%) 
Rattus were captured on the ground — in leaf litter, 
by downed dead wood, and by root tangles. An- 
imals taken from above the ground (36%) had 
been crossing lianas, limbs, or trunks of various 
sizes and at a variety of angles. 

Of the 13 R. rattus taken in 1993, nine were 
scrotal males, two were adult males with abdom- 
inal testes, one was a female with enlarged mam- 
mae, and one was a female with small mammae. 

Comments — The history of the black rat's col- 
onization of Madagascar, in particular the eastern 
humid forest, is unknown. On the basis of RCP 
specimens collected in 1970-1971, it is clear that 
Rattus rattus has been on the mountain for over 
20 years. The western side of the Andringitra 
Massif has been settled and cleared for a consid- 
erable period and contains numerous agricultural 
areas, particularly rice paddies. The eastern side, 
however, is relatively intact and lacks dense hu- 
man population. It is conceivable that R. rattus 
colonized the eastern slopes from the western 



GOODMAN & CARLETON: RODENTS 



261 



side via the central crystalline ridge. This sup- 
position is supported by the observation that 
male R. rattus are known to disperse greater dis- 
tances than females (Spencer & Davis, 1950), 
and 85% of the animals collected in 1993 in the 
RNI d'Andringitra were males. This biased trap 
success is interpreted as an expanding population 
front. Further, at the time of the 1993 survey, R. 
rattus was not found in the 720 m zone, which 
was close to a small village. At other sites on the 
island R. rattus tends to be common in similar 
circumstances. The pattern of greater trap suc- 
cess with increasing elevation, combined with 
the other points outlined above, would support 
the idea that this species is invading the eastern 
portion of the massif from the central ridge. 

On other tropical islands where R. rattus has 
been introduced, considerable variation is appar- 
ent in its ability to colonize montane forest. For 
example, on some islands in the Philippines and 
Sulawesi, the species tends to be restricted to dis- 
turbed lowland forest or agricultural areas (Mus- 
ser, 1987; Rickart et al., 1993), whereas on others 
it occurs broadly across an ecological and altitu- 
dinal gradient from lowland to mossy forest (Hea- 
ney et al., 1989). What factors lead to such vari- 
ation in the ability of R. rattus to colonize differ- 
ent natural forests are not clear, but at least in the 
Philippines and Sulawesi, they may be related to 
the presence/absence of sympatric, forest-dwell- 
ing native congeners. 

On the basis of feeding trials with captive an- 
imals in the RNI d'Andringitra, Rattus rattus and 
nesomyine rodents feed on similar types of forest 
fruits and nuts (see Chapter 6). Thus, the potential 
for direct competition between these rodents ex- 
ists (Goodman, 1995). Whether this competition 
is sufficient for R. rattus to displace nesomyine 
rodents in the RNI d'Andringitra is unknown. 

No evidence of Rattus norvegicus was found 
within the reserve, but this species was reported 
to have been captured in 1993 at 1400 m in the 
Foret Imaitso, in the north-central portion of the 
reserve (O'Keeffe & Ashmore, no date). 

Specimens Examined — Anjavidilava (MNHN 
1972.521); Ambalamarovandana (MNHN 
1972.613); Ivangomena (MNHN 1972.614); 38 
km S Ambalavao, ridge E of Volotsangana River, 
1625 m (FMNH 151921 through 151924); 40 km 
S Ambalavao, along Volotsangana River, 1210 m 
(FMNH 151729 and 151918 through 151920); 43 
km S Ambalavao, junction of Sahanivoraky and 
Sahavatoy rivers, 810 m (FMNH 151728, 151916, 
and 151917). 



Family Muridae: Subfamily Nesomyinae 
Brachyuromys Major, 1896a 

Recognition — A broad head and compact tor- 
so, small and rounded pinna, short legs and rela- 
tively narrow hind feet, and tail vertebrae con- 
spicuously shorter than the length of head-and- 
body (TL about 50-60% of HBL) identify mem- 
bers of this nesomyine genus (Table 22-2). The 
body fur is thick and fine, its texture soft and lax. 
The dorsum is colored a rich brown to reddish 
brown, highlighted by buff or ochraceous tips, 
and the venter is approximately the same in color 
and tone, without any lateral border defined be- 
tween them. The dark monocolored tail is evenly 
covered with short hairs, which do not lengthen 
perceptibly toward the tip. Plantar pads number 
six, with the thenar and hypothenar unreduced and 
set close to the interdigitals. The mammae count 
is also six: one pair each postaxial, abdominal, 
and inguinal. 

The skull appears stoutly constructed, its length 
short relative to width, particularly as observed 
across the braincase and lateral reaches of the zy- 
goma (Table 22-3). The narrowness of the inter- 
orbital constriction is accentuated by the heavy 
zygomatic arches that converge abruptly toward 
broad zygomatic plates. The rostrum is deep and 
short, the ectotympanic bullae moderately inflat- 
ed. The hypsodont molars possess planar occlusal 
surfaces, oval to cylindriform in shape. The pat- 
tern of each molar is configured as three broad 
lamina or enamel loops that are oriented obliquely 
to the longitudinal axis of the toothrow. Maxillary 
toothrows are convergent anteriorly, but the man- 
dibular rows are more or less parallel to one an- 
other. 

As remarked by each discoverer of the two 
know species (Bartlett, 1879; Major, 1896a), these 
rodents resemble members of the Holarctic sub- 
family Arvicolinae externally. Both species of 
Brachyuromys, B. betsileoensis and B. ramirohi- 
tra, occur in the RNI d'Andringitra. 



Brachyuromys betsileoensis (Bartlett, 1879) 

Identification and Distribution — Bartlett's 
(1879) B. betsileoensis is the noticeably smaller 
of the two species, as reflected in almost every 
univariate comparison of the skin and skull (Ta- 
bles 22-2 and 22-3). Although the cranium of B. 
betsileoensis is more diminutive than that of B. 



262 



FIELDIANA: ZOOLOGY 



Table 22-3. Comparison of selected craniodental measurements of adult Brachyuromys betsileoensis and B. 
ramirohitra from Andringitra and Ampitambe. 





B. 


betsileoensis 


B. 


ramirohitra 




Andringitra 


Ampitambe 


Andringitra 


Ampitambe 


Variable 


(N = 16) 


N = 10 


(N = l) 


(N = 27) 


ONL 


34.6 ± 0.8 


33.6 ± 1.5 


40.7 


39.1 ± 1.3 




33.5-36.0 


31.5-36.2 




36.0-41.5 


ZB 


20.8 ± 0.6 


20.1 ± 1.0 


23.0 


22.9 ± 0.9 




20.1-21.7 


18.5-21.4 




20.5-24.4 


BBC 


14.4 ± 0.2 


13.9 ± 0.5 


15.1 


14.8 ± 0.4 




14.1-14.6 


13.3-14.7 




14.1-15.7 


IOB 


4.9 ± 0.2 


5.0 ± 0.2 


5.4 


5.1 ± 0.2 




4.5-5.2 


4.7-5.3 




4.5-5.5 


LR 


10.5 ± 0.3 


10.2 ± 0.5 


13.6 


13.0 ± 0.6 




9.9-11.1 


9.6-11.0 




11.6-13.9 


PPL 


13.4 ± 0.6 


13.0 ± 0.8 


15.2 


14.3 ± 0.8 




13.1-14.7 


12.0-14.3 




12.4-16.0 


LIF 


7.2 ± 0.3 


6.8 ± 0.5 


7.9 


7.1 ±0.4 




6.5-7.7 


5.9-7.5 




6.1-7.7 


LD 


9.3 ± 0.7 


8.9 ± 0.6 


10.2 


9.5 ± 0.7 




8.3-10.3 


8.0-9.7 




8.3-10.7 


BM1S 


8.7 ± 0.4 


8.6 ± 0.5 


9.5 


9.2 ± 0.4 




8.3-9.2 


8.1-9.7 




8.1-10.2 


BZP 


4.2 ± 0.2 


4.1 ±0.3 


5.0 


5.0 ± 0.3 




3.8-4.6 


3.6-4.7 




3.9-5.4 


LM1-3 


6.9 ± 0.3 


6.9 ± 0.2 


9.0 


8.7 ± 0.2 




6.5-7.4 


6.6-7.2 




8.1-9.1 


WM1 


2.2 ±0.13 


2.3 ± 0.2 


2.9 


2.7 ±0.1 




1.9-2.5 


2.0-2.6 




2.4-3.0 



Note: Sample parameters are mean ± SD (range). 



ramirohitra, observed ranges of most variables do 
overlap, both in the Andringitra samples and in 
additional series of the two species from Ampi- 
tambe (Table 22-3). Only the coronal length of 
their upper molars (LM1-3) does not overlap, a 
difference that offers a key means of species sep- 
aration: LM1-3 always < 7.5 mm (range = 6.5- 
7.4) in samples of B. betsileoensis and LM 1 -3 al- 
ways > 8.0 mm (range = 8.1-9.1) in B. ramiro- 
hitra. Other cranial variables — such as occipito- 
nasal length (ONL), zygomatic breadth (ZB), and 
width of the first molar (WM1) — are almost as 
reliable in providing clear discrimination of the 
two kinds (Table 22-3; see also Petter, 1975). 

Other than size, certain qualitative traits may 
distinguish the species of Brachyuromuys. The 
pelage of specimens of B. betsileoensis is shorter 
and less luxuriant than that found on B. ramiro- 
hitra, and the color is a plainer brown, lacking the 
deep reddish tones typical of B. ramirohitra, es- 
pecially on the sides and ventrum. A vestige of 
the anterior mure connects the two anterior enam- 
el loops on the first and second upper molars in 
specimens of B. betsileoensis; the molars of B. 
ramirohitra generally lack this connection. 



The voucher specimens preserved by the 1970 
RCP expedition constitute the southernmost dis- 
tributional point for B. betsileoensis, a species that 
ranges over the central highlands as far north as 
the vicinity of Didy (1000 m), just southeast of 
Lac Alaotra (Carleton & Schmidt, 1990). The el- 
evations for the several Andringitra localities (see 
below), as later supplied by Paulian et al. (1971), 
extend from approximately 1 800 to over 2400 m. 
Based on museum records, Carleton and Schmidt 
(1990) had reported the altitudinal distribution for 
this species from about 900 to 2000 m. 

Ecology and Reproduction — On the Andrin- 
gitra Massif, the material collected by the RCP 
team suggests that B. betsileoensis occurs in 
heathland areas above the tree line down to the 
upper portion of montane and sclerophyllous for- 
est (1700-1900 m). In late 1993, cranial remains 
of B. betsileoensis were recovered from Crypto- 
procta ferox scats collected in heathland at 2000 
and 2100 m. The lower range of the elevational 
transect performed (720 through 1625 m) presum- 
ably accounts for the absence of the species in the 
1993 survey. 

No reproductive information is available. 



GOODMAN & CARLETON: RODENTS 



263 



Specimens Examined — Foret Aguaria (MNHN 
1972.586, 1972.587, 1972.600, and 1972.601); 
Plateau Andohariana (MNHN 1972.594 through 
1972.596); Cuvette de Boby (MNHN 1972.588 
and 1972.589); Ivangomena, S Varavarana 
(MNHN 1972.592, 1972.593, 1972.598, and 
1972.599); Gue de la Riembavy (MNHN 
1972.590); Varavarana (MNHN 1972.591 and 
1972.597). 



Brachyuromys ramirohitra Major, 1896a 

Identification and Distribution — Discrimi- 
nation of this species from its congener B. betsi- 
leoensis is covered in the previous account (also 
see Tables 22-2 and 22-3). 

The meager provenance for B. ramirohitra sug- 
gests that its geographical and elevational distri- 
bution is more confined. As with B. betsileoensis, 
the records of B. ramirohitra on the Andringitra 
Massif represent its southernmost limit as cur- 
rently understood. The two specimens obtained in 
1993, together with the one collected by Albignac 
in 1970, document its presence from 1210 to 
about 2000 m in the RNI d' Andringitra. Carleton 
and Schmidt ( 1 990) reported B. ramirohitra from 
two other localities, Ampitambe and Amboasary, 
at about 900 and 1300 m, respectively. In both 
places, which lie within the southern sector of the 
island's Central High Plateau, the two species of 
Brachyuromys occur in close proximity if not in 
actual sympatry. 

Ecology and Reproduction — Brachyuromys 
ramirohitra apparently inhabits slightly lower 
vegetational zones and penetrates primary forest. 
The single RCP specimen was collected in Agua- 
ria forest, probably near camp 4 at Marositry, at 
approximately 2000 m, to judge from its date of 
collection (Paulian et al., 1971). This record and 
those reported above confirm the syntopic exis- 
tence of B. betsileoensis and B. ramirohitra at 
least in the Aguaria zone at the fringe of upper 
montane (sclerophyllous) forest; in fact, both spe- 
cies were taken in Aguaria habitat on the same 
day. The two specimens taken in 1993, at 1210 
and 1625 m, were trapped in primary montane 
and sclerophyllous forest, respectively. The trap 
at 1625 m was placed at the opening of a natural 
tunnel system associated with a massive root en- 
tanglement covered by soil and matted leaf litter. 

The male from 1210 m has partially erupted 
third molars and is reproductively immature, its 
testes small and within the abdomen. The adult 



female trapped at 1625 m has slightly enlarged 
mammae and two embryos (3 mm crown-rump 
length). 

Specimens Examined — Foret Aguaria (MNHN 
1972.585); 38 km S Ambalavao, ridge east of 
Volotsangana River, 1625 m (FMNH 151660); 40 
km S Ambalavao, along Volotsangana River, 1210 
m (FMNH 151659). 



Eliurus Milne Edwards (1885) 

Recognition — Among Madagascar's endemic 
rodents, members of this genus are small to me- 
dium in body size (Table 22-2) and are immedi- 
ately recognizable by their densely haired tail that 
bears a conspicuous terminal pencil or tuft. In all 
species of tufted-tailed rats, the tail length is 
greater than the head-and-body (TL about 1 15— 
120% of HBL), their fur is soft and fine, and the 
hind foot is relatively short and broad, with six 
fleshy plantar pads and an elongate fifth digit. The 
mammae number six, distributed as one postaxial, 
one abdominal, and one inguinal pair. Diagnostic 
traits of the skull include the hourglass-shaped in- 
terorbit devoid of supraorbital shelves or ridges; 
short incisive foramina, which terminate well an- 
terior to the molar rows; and the moderately hyp- 
sodont molars, each with a uniquely trilaminate 
occlusal configuration (see Carleton, 1994). No 
regular pattern of sexual dimorphism in cranio- 
dental size is evident among the RNI d'Andrin- 
gitra samples of Eliurus, a finding consistent with 
the analyses by Carleton (1994). 

Four of the eight known species of Eliurus have 
been documented in the RNI d' Andringitra. 



Eliurus majori Thomas, 1895 

Identification and Distribution — Prior to this 
survey, E. majori was known only from five spec- 
imens from three widely separated localities: two 
from Montagne d'Ambre, an isolated peak at the 
northernmost end of Madagascar; two from Am- 
bohimitombo, the type locality on the Central 
High Plateau; and one from Anjavidilava in the 
RNI d' Andringitra (Carleton, 1994). The fine se- 
ries now available from the Andringitra Massif 
permits descriptive elaboration and additional tax- 
onomic comments on the species. 

Specimens of E. majori rank with those of E. 
tanala as nearly the largest-bodied in the genus 
(Table 22-2), a comparison not appreciated by 



264 



FIELDIANA: ZOOLOGY 



Carleton (1994) with the limited specimen data 
then at hand. Relative to its head-and-body size, 
however, the hind feet of E. majori appear shorter 
than those observed in examples of E. tanala and 
E. webbi, and its pinnae are both absolutely and 
relatively smaller than those of either of those 
species and scarcely surpass the average ear 
length recorded for the diminutive E. minor (Table 
22-2). The dorsum is consistently somber in tone, 
colored blackish gray to blackish brown. In con- 
trast to other Eliurus, chromatic demarcation be- 
tween the dorsal and ventral pelage is typically 
indistinct, although the ventrum appears generally 
lighter; in some individuals, a suffusion of cream- 
tipped hairs does lend more dorsal-ventral con- 
trast. The top of the metatarsum is uniformly 
dusky and contrasts with the dull whitish phalan- 
ges. 

Although the tail is well furred and noticeably 
penicillate in E. majori, the terminal hairs form a 
less pronounced tuft than in other Eliurus. The 
caudal hairs gradually become longer toward the 
tip of the tail, unlike the abrupt elongation into a 
terminal brush over the distal one-third to one- 
fourth of the tail, as found in other species, in- 
cluding the smaller, shorter-tailed E. minor. The 
caudal hairs are uniformly dark in all 18 speci- 
mens with an intact tip. The constancy of this 
trait, as can now be substantiated in a large pop- 
ulation sample of E. majori, elevates our confi- 
dence in the diagnostic value of tail color with 
regard to its specific distinction from E. penicil- 
latus, a form described from Ampitambe (Thom- 
as, 1908). Pelage color and cranial morphology of 
the two species are remarkably alike, except for 
the white-tipped tail exhibited by all individuals 
of E. penicillatus from Ampitambe, the one col- 
lecting site so far known (Carleton, 1994). 

Aside from the distinctiveness of its pelage and 
external form, many proportional features of the 
cranium and dentition set E. majori (and E. pen- 
icillatus) apart from other species of Eliurus (El- 
lerman, 1949; Carleton, 1994). These shape con- 
trasts retain their discriminatory power for the 
RNI d' Andringitra series of E. majori and are best 
reflected in their robust molar rows, long and wide 
incisive foramina relative to the short diastema, 
and laterally bowed zygomatic arches (compare 
Tables 22-4 and 22-7). 

Although the disparate sample sizes discourage 
meaningful comparisons, substantial morphologi- 
cal variation is evident among the three widely 
separated samples of E. majori. In most cranio- 
dental measurements, the animals from the An- 



Table 22-4. Comparison of selected craniodental 
measurements of adult Eliurus majori from Montagne 
d'Ambre (N = 2), Ambohimitombo (type locality, N = 
2), and Andringitra (N = 17). 





Mt. 


Ambohimi- 




Variable 


d'Ambre* 


tombof 


Andringitra 


ONL 


36.5, 37.5 


35.2, 35.8 


37.6 ± 1.0 
(35.5-39.4) 


ZB 


19.3, 19.3 


18.3, 18.6 


19.1 ±0.5 
(18.1-19.8) 


BBC 


14.1, 14.2 


13.7, 14.1 


14.3 ± 0.4 
(13.4-14.8) 


IOB 


5.2, 5.5 


5.0, 5.2 


5.3 ± 0.2 
(4.9-5.7) 


LR 


12.0, 13.1 


12.1, 12.2 


13.1 ±0.5 
(11.6-14.0) 


PPL 


13.0, 13.2 


11.9, 12.7 


13.4 ± 0.5 
(12.4-14.6) 


LIF 


5.5,6.1 


5.7, 6.0 


6.3 ± 0.3 

(5.6-6.7) 


LD 


10.8, 11.1 


9.8, 10.2 


10.8 ± 0.4 
(10.0-11.8) 


BM1S 


7.9, 8.3 


7.2, 7.8 


8.1 ±0.3 
(7.1-8.6) 


BZP 


3.1, 3.2 


2.9, 3.1 


3.5 ± 0.2 
(2.9-3.9) 


LM1-3 


6.1, 6.3 


6.3, 6.8 


6.6 ± 0.2 
(6.1-7.1) 


WM1 


1.7, 1.8 


1.5, 1.8 


1.8 ±0.1 
(1.7-1.9) 



Note: Values for Andringitra specimens are means 
SD (range). 
* AMNH 100854 and 100867. 
t MCZ 45929 and BMNH 97.9.1.147 (holotype). 



dringitra Massif average larger and attain greater 
maxima than recorded for either the two speci- 
mens from Montagne d'Ambre or the two from 
Ambohimitombo, the type locality (Table 22-4). 
However, observed ranges of the RNI d' Andrin- 
gitra sample generally circumscribe individual 
values from the latter two sites. Biased age rep- 
resentation further complicates evaluation of in- 
terlocality differences and their taxonomic signif- 
icance. According to evidence from tooth wear, 
the two specimens from Ambohimitombo are a 
young animal and a full adult (the type); the 17 
from the RNI d' Andringitra represent a mixture 
of young to old adult animals. The two AMNH 
records from Montagne d'Ambre are extremely 
old adults, to judge by their flatly worn molars, 
yet they appear generally smaller when placed 
side by side with comparably aged individuals 
within the Andringitra series. Considering both 
relative age and sample size, the population from 
Montagne d'Ambre appears most divergent; this 
may be anticipated in view of its greater isolation 



GOODMAN & CARLETON: RODENTS 



265 



Table 22-5. Microhabitat occurrences of rodent species. 



















Above-ground location 


















Vine, 
















( 


.round locatior 


i 


limb, 
or 


Limbs. 










Trap position 




Under 


By 




trunk trunks 


Sus- 






No. 


On 


Above 


Leaf 


rotten 


roots, 




<10 


>10 


pended 




Elevation and species 


taken 


ground ground 


litter 


wood 


trunks 


Misc. 


cm 


cm 


trunks 


Misc. 


720 m 
























Trap distribution 




63 


37 


33 


7 


18 


5 


32 


2 


3 





Eliurus minor 


5 


3 


2 


2 


1 








2 











Eliurus webbi 


10 


4 


6 


4 











2 





2 


2 


Gymnuromys roberti 


1 


1 




1 



















810 m 
























Trap distribution 




52 


48 


27 


9 


16 





39 


1 


7 


1 


Rattus rattus 


1 





1 










1 











Eliurus minor 


13 


2 


11 


1 





1 





11 











Eliurus tanala 


6 





6 










4 





2 





Eliurus webbi 


13 


2 


11 


1 





1 





11 











Gymnuromys roberti 


1 


1 





1 



















Nesomys audeberti 


1 


1 





1 



















Nesomys rufus 


5 


5 





5 



















1210 m 
























Trap distribution 




58 


42 


33 


6 


16 


3 


22 


3 


14 


3 


Rattus rattus 


4 


3 


1 








3 











1 





Eliurus majori 


10 





10 










4 


1 


4 


1 


Eliurus minor 


7 


3 


4 








3 





1 





1 


2 


Eliurus tanala 


9 


5 


4 


3 


1 





1 


2 





2 





Gymnuromys roberti 


3 


3 








1 


2 













Nesomys rufus 


11 


11 





6 





4 


1 










1625 m 
























Trap distribution 




65 


35 


25 


8 


31 


1 


23 


9 


2 


1 


Rattus rattus 


6 


4 


2 


2 


2 








1 


1 








Brachyuromys ramirohitra 


1 


1 











1 













Eliurus majori 


12 


4 


8 


1 





2 


1 


1 


6 


1 





Eliurus minor 


3 


1 


2 








1 








2 








Eliurus tanala 


9 


6 


3 





1 


5 





1 





2 





Gymnuromys roberti 


1 


1 








1 
















Nesomys rufus 


14 


14 





9 


5 
















Monticolomys koopmani 


2 





2 










2 











Totals: 720-1625 m 
























Trap distribution 




238 


162 


118 


30 


81 


9 


116 


15 


26 


5 


Rattus rattus 


11 


7 


4 


2 


2 


3 





2 


1 


1 





Brachyuromys ramirohitra 


1 


1 











1 













Eliurus majori 


22 


4 


18 


1 





2 


1 


5 


7 


5 


1 


Eliurus minor 


28 


9 


19 


3 


1 


5 





14 


2 


1 


2 


Eliurus tanala 


24 


11 


13 


3 


2 


5 


1 


7 





6 





Eliurus webbi 


23 


6 


17 


5 





1 





13 





2 


2 


Gymnuromys roberti 


6 


6 





2 


2 


2 













Nesomys audeberti 


1 


1 





1 



















Nesomys rufus 


30 


30 





20 


5 


4 


1 










Monticolomys koopmani 


2 





2 










2 












from the central and southern highlands of Mad- 
agascar. Additional samples from Montagne 
d'Ambre and from intermediate localities are re- 
quired to validate this supposition and to deter- 
mine whether more than one species is represent- 



ed among the populations currently classified as 
E. majori. 

Specimens taken during the 1993 expedition 
were restricted to the 1210 and 1625 m zones. In 
1971, Albignac collected the sole previous ex- 



266 



FIELDIANA: ZOOLOGY 



Table 22-6. Comparison of selected craniodental 
measurements of adult Eliurus minor from Ampitambe 
(type locality; N = 2) and Andringitra (N = 17). Sample 
parameters are mean ± standard deviation, and range. 



Table 22-7. Comparison of selected craniodental 
measurements of adult Eliurus tanala (N = 14) and E. 
webbi (N = 19) from Andringitra. Sample parameters 
are mean ± standard deviation, and range. 



Variable 


Ampitambe* 


Andringitra 


Variable 


/■.". tanala 


E. webbi 


ONL 


28.9, 30.8 


30.1 ± 1.0 


ONL 


40.5 ± 1.0 


37.5 ± 1.1 






27.7-31.3 




38.8-42.3 


35.4-39.0 


ZB 


15.3, 15.9 


15.3 ± 0.6 


ZB 


19.5 ± 0.6 


17.7 ± 0.7 






14.1-16.3 




18.5-20.7 


16.4-19.0 


BBC 


12.2, 12.4 


12.0 ± 0.3 


BBC 


14.6 ± 0.3 


14.0 ± 0.5 






11.4-12.5 




14.1-15.3 


13.1-15.0 


IOB 


4.8,5.1 


5.0 ± 0.2 


IOB 


5.9 ± 0.3 


5.8 ± 0.2 






4.8-5.3 




5.4-6.4 


5.3-6.2 


LR 


9.8, 10.1 


10.3 ± 0.4 


LR 


14.7 ± 0.6 


13.3 ± 0.5 






9.4-10.9 




13.7-15.5 


12.2-14.4 


PPL 


10.0, 11.7 


10.8 ± 0.5 


PPL 


14.1 ±0.5 


13.2 ± 0.5 






9.8-11.4 




13.6-14.7 


12.0-13.9 


LIF 


4.2, 4.4 


4.2 ± 0.3 


LIF 


5.4 ± 0.3 


5.1 ± 0.3 






3.7-4.7 




4.8-5.9 


4.6-5.8 


LD 


8.3, 8.7 


8.8 ± 0.4 


LD 


12.3 ± 0.5 


11.2 ±0.5 






8.1-9.3 




11.4-13.3 


10.2-12.5 


BM1S 


6.0, 6.3 


6.1 ± 0.2 


BM1S 


7.9 ± 0.2 


7.5 ± 0.3 






5.7-6.6 




7.4-8.0 


6.9-7.9 


BZP 


2.3, 2.5 


2.6 ± 0.2 


BZP 


3.8 ± 0.2 


3.3 ± 0.4 






2.2-3.0 




3.5-4.1 


2.9-3.6 


LM1-3 


4.0, 4.5 


4.1 ±0.1 


LM1-3 


5.7 ± 0.2 


5.3 ± 0.2 






3.9-4.4 




5.3-6.0 


4.9-5.6 


WM1 


1.1, 1.2 


1.1 ± 0.05 


WM1 


1.5 ±0.05 


1.4 ± 0.07 






1.0-1.2 




1.4-1.6 


1.2-1.5 


* BMNH 97.9.1.153 (holotype) 


and FMNH 5629. 









ample of this species on the Andringitra Massif 
near Anjavidilava, around 2000 m (elevation fide 
Paulian et al„ 1971). These records collectively 
document the local altitudinal range of E. majori 
from 1210 to 2000 m, a zone coinciding with mid- 
dle to upper montane (sclerophyllous) forest, and 
they constitute the first vouchered evidence of 
sympatry with any other species of Eliurus, name- 
ly E. minor and E. tanala. Elsewhere, the species 
is known from elevations of approximately 1000 
(Montagne d'Ambre) and 1200 m (Ambohimi- 
tombo). 

Ecology and Reproduction — Trapping evi- 
dence portrays E. majori as principally arboreal 
in habits, with occasional forays to the ground 
(Table 22-5). All examples taken at 1210 m were 
in elevated traps placed on lianas, limbs, and 
trunks of various sizes and different inclinations. 
At 1625 m, four of 12 animals trapped were on 
the ground, specifically in leaf litter and by roots 
and trunks of trees, whereas eight were collected 
in a variety of trap placements above the ground. 

The 22 E. majori taken during the 1993 season 
included nine males with scrotal testes, two males 
with abdominal testes, nine females with enlarged 



mammae, and one adult female with small mam- 
mae. One female had four 11 -cm (crown-rump 
length) embryos, and three other females each had 
three discernible placental scars. 

Specimens Examined — 38 km S Ambalavao, 
ridge east of Volotsangana River, 1625 m (FMNH 
151666, 151667, 151732, 151752, 151851 
through 151853, and 151854 through 151858); 40 
km S Ambalavao, along Volotsangana River, 1210 
m (FMNH 151661 through 151665, 151730, 
151731, and 151847 through 151849); Anjavidi- 
lava (MNHN 1972.602). 



Eliurus minor Major, 1896b 

Identification and Distribution — Compared 
to other Eliurus, individuals of this species are 
readily recognized by their smaller size in most 
dimensions of the body and skull (Tables 22-2 and 
22-6) and by the well-developed, dark pencil on 
the distal half of the tail. Within Nesomyinae, 
only specimens of Monticolomys koopmani ap- 
proach those of E. minor on the basis of overall 
size. Individuals of Monticolomys, however, pos- 
sess a lighter build, shorter head-and-body length, 



GOODMAN & CARLETON: RODENTS 



267 



Table 22-8. Comparison of selected craniodental 
measurements of adult Gymnuromys roberti from An- 
dringitra (N = 4) and from Ampitambe (N = 20), the 
type locality. Sample parameters are mean ± standard 
deviation, and range. 



Table 22-9. Comparison of selected craniodental 
measurements of adult Nesomys audeberti (N = 1) and 
N. rufus (N = 21) from Andringitra. Sample parameters 
are mean ± standard deviation, and range. 



Variable 



Variable 



N. audeberti 



Andringitra 



Ampitambe 



N. rufus 









ONL 


47.5 


44.5 ± 0.9 


ONL 


39.9 ± 1.9 


39.4 ± 1.0 






42.3-46.3 




37.9-41.7 


36.7-41.0 


ZB 


24.7 


23.3 ± 0.6 


ZB 


19.5 ± 1.1 


19.6 ± 0.6 






22.5-24.3 




18.6-21.0 


18.3-20.9 


BBC 


17.3 


16.3 ± 0.3 


BBC 


14.6 ± 0.03 


14.5 ± 0.3 






15.7-16.7 




14.6-14.7 


13.9-15.0 


IOB 


8.5 


8.1 ±0.2 


IOB 


5.9 ± 0.3 


5.9 ± 0.3 






7.7-8.4 




5.6-6.3 


5.4-6.4 


LR 


18.8 


16.9 ± 0.6 


LR 


14.4 ± 0.7 


14.2 ± 0.5 






15.6-17.9 




13.6-15.0 


13.3-15.6 


PPL 


16.0 


14.6 ± 0.4 


PPL 


13.8 ± 1.1 


14.0 ± 0.7 






13.6-15.4 




12.9-15.0 


12.4-15.1 


LIF 


9.6 


8.3 ± 0.4 


LIF 


4.7 ± 0.4 


4.5 ± 0.3 






7.4-8.9 




4.2-5.1 


4.1-5.2 


LD 


13.0 


12.1 ± 0.3 


LD 


11.6 ±0.5 


11.6 ±0.6 






11.5-12.6 




11.2-12.1 


10.5-12.7 


BM1S 


9.8 


9.9 ± 0.4 


BM1S 


8.2 ± 0.4 


8.2 ± 0.2 






9.2-10.8 




7.8-8.6 


7.8-8.7 


BZP 


4.3 


4.5 ± 0.3 


BZP 


2.9 ± 0.3 


2.9 ± 0.2 






4.0-5.3 




2.5-3.3 


2.5-3.3 


DAB 


7.3 


6.7 ± 0.1 


LM1-3 


5.8 ± 0.1 


5.6 ± 0.2 






6.4-7.1 




5.7-5.9 


5.3-6.0 


LM1-3 


7.1 


6.9 ± 0.2 


WM1 


1.6 ±0.1 


1.6 ±0.1 






6.6-7.4 




1.5-1.8 


1.3-1.8 


WM1 


2.2 


2.2 ±0.1 












2.1-2.4 



relatively longer tail (140% of HBL), and longer 
hind feet (Table 22-2); furthermore, they lack the 
conspicuous terminal brush on the tail that is so 
characteristic of all species of Eliurus. Crania and 
dentitions of E. minor and Monticolomys differ 
strikingly and allow no confusion (see Chapter 
21). 

Eliurus minor is found throughout eastern for- 
est, in lowland and montane settings and from 
near sea level to over 1500 m. In view of its broad 
geographic range, the degree of differentiation of 
certain population samples, particularly those 
around Antongil Bay, and their taxonomic status, 
require further clarification. The series from the 
RNI d' Andringitra, however, agrees closely with 
the type material from Ampitambe (Table 22-6) 
and conforms closely in mensural and pelage vari- 
ation to other E. minor from southern Madagascar, 
such as specimens collected in Pare National (PN) 
de Ranomafana and vicinity (see Carleton, 1994: 
appendix 2). 

The elevational distribution of E. minor on the 
eastern slopes of the RNI d' Andringitra (720 to 
1625 m) falls within its previously recorded range 



elsewhere on the island, from near sea level to 
1800 m (Carleton & Schmidt, 1990). 

Ecology and Reproduction — Eliurus minor 
occurs in all four elevational zones sampled, but 
relative trapping success suggests that it is most 
abundant at middle elevations, 720 to 1210 m (Ta- 
ble 22-10). 

Data pooled from the four elevations indicate 
that Eliurus minor frequents a wide variety of ter- 
restrial and arboreal microhabitats (Table 22-5). 
Although 50% (14 of 28) of all individual cap- 
tures occurred on lianas, vines, or trunks less than 
10 cm in circumference, the overall proportion of 
traps placed above ground was just 29% (116 of 
400 traps). Eliurus minor appears to move across 
relatively thin vines and branches. In three of the 
four elevational zones, E. minor was captured on 
or off the ground with nearly equal frequency; the 
exception was at 810 m, where 11 of 13 (85%) 
animals were in arboreal sets. 

The reproductive state of the 28 specimens col- 
lected in 1993 included nine scrotal males, five 
males with abdominal testes, 10 females with 
large mammae (four of which were lactating), and 



268 



FIELDIANA: ZOOLOGY 



Table 22-10. Number of individuals trapped for each species of small mammal at all four elevations surveyed 
(total trap-days accrued is shown in parentheses). 





720 m 


810 m 


1210 m 


1625 m 


Species 


(550) 


(650) 


(625) 


(650) 


Rutins rutins 




1 


4 


6 


Brachyuromys ramirohitra 




... 




1 


Eliurus majori 






10 


12 


Eliurus minor 


5 


13 


7 


3 


Eliurus tanala 




6 


9 


9 


Eliurus webbi 


10 


13 




... 


Gymnuromys roberti 


1 


1 


3 


1 


Nesomys audeberti 




1 




... 


Nesomys rufus 




5 


11 


14 


Monticolomys koopmani 








2 


Microgale spp. 






10 


1 


Total individuals 


16 


40 


54 


49 


% trap success 


2.9 


6.2 


8.6 


7.8 


Total rodents 


16 


40 


44 


48 


% rodent trap success 


2.9 


6.2 


7.0 


7.7 


Total nesomyines 


16 


39 


40 


42 


% nesomyine trap success 


2.9 


6.0 


7.0 


7.4 



three females with small mammae and imperfor- 
ated vaginas. Two females with large mammae 
had placental scars, numbering three and four, and 
no embryos. One female had three embryos mea- 
suring 10 mm in crown-rump length. Nine fe- 
males exhibited the mammae apportionment typ- 
ical of Eliurus, but one appeared to lack the post- 
axial pair and instead possessed an extra abdom- 
inal pair (0-4-2). 

Specimens Examined — 38 km S Ambalavao, 
ridge E of Volotsangana River, 1625 m (FMNH 
151679, 151867, and 151868); 40 km S Amba- 
lavao, along Volotsangana River, 1210 m (FMNH 
151678, 151736 through 151738, and 151864 
through 151866); 43 km S Ambalavao, junction 
of Sahanivoraky and Sahavatoy rivers, 810 m 
(FMNH 151672 through 151677, 151734 through 
151737, and 151859 through 151863); 45 km S 
Ambalavao, E bank of Iantara River, along Am- 
balamanenjana-Ambatomboay trail, 720 m 
(FMNH 151668 through 151671, and 151733). 



Eliurus tanala Major 1896b 

Identification and Distribution — Eliurus ta- 
nala is a distinctive species characterized by mod- 
erately large size and a long tail with feathery 
white tip. The white caudal plume provides defin- 
itive separation from specimens of E. majori and 
E. webbi, which overlap those of E. tanala in cer- 
tain external dimensions (such as TOTL, HBL, 
TL, and WT— Table 22-2). Eliurus tanala and E. 



webbi are otherwise very similar in craniodental 
size and shape, but the former can be reliably dis- 
criminated by its generally longer, broader skull 
and more robust molar rows (Table 22-7). The 
morphometric differences between the two spe- 
cies in the RNI d'Andringitra resemble those 
demonstrated for other places where they co-oc- 
cur, for instance the PN de Ranomafana area 
(Carleton, 1994). Other anatomical traits that al- 
low for discrimination between the two species 
are discussed and illustrated by Carleton (1994). 

Samples of Eliurus tanala are morphologically 
homogeneous over its range in middle elevation 
forests along the eastern flanks of the central high- 
lands. The RNI d'Andringitra records extend the 
known distribution slightly farther south of its for- 
mer limit at Vinanitelo, the type locality. Eliurus 
tanala was found in the 810, 1210, and 1625 m 
zones in the RNI d'Andringitra; Carleton (1994) 
had summarized its elevational occurrence as 455 
to 1300 m. 

Ecology and Reproduction — Trap success of 
this species increased slightly as a function of 
higher altitude (Table 22-10). At 810 m, the low- 
est elevational occurrence, E. tanala was captured 
only in arboreal sets, largely on nearly horizontal 
small vines and limbs. At higher elevations, ap- 
proximately half of the additional 1 8 individuals 
captured were on the ground. In both situations, 
the species was taken in a variety of sets (Table 
22-5). At 1625 m, four individuals were collected 
over the course of 7 trap-days in the same Na- 



GOODMAN & CARLETON: RODENTS 



269 



tional trap placed at the opening of an extensive 
natural tunnel system under root tangles and mat- 
ted leaf litter. 

During November and December 1993, the 
sample of 23 E. tanala exhibited the following 
reproductive conditions: four males with scrotal 
testes, seven males with abdominal testes, six fe- 
males with large teats, and four females with 
small teats. One female was pregnant with three 
embryos, each measuring 15 mm crown-rump 
length, and another female with small mammae 
had three placental scars. The one juvenile E. ta- 
nala (molars partially erupted) was trapped on 
December 1, 1993 at 1210 m. 

Specimens Examined — 38 km S Ambalavao, 
ridge E of Volotsangana River, 1625 m (FMNH 
151691, 151692, 151744, and 151878 through 
151883); 40 km S Ambalavao, along Volotsan- 
gana River, 1210 m (FMNH 151690, 151743, 
151872 through 151877, and 151897); 43 km S 
Ambalavao, junction of Sahanivoraky and Sahav- 
atoy rivers, 810 m (FMNH 151687 through 
151689 and 151869 through 151871). 



Table 22- 1 1 . Comparison of nesomyine rodents 
documented to date for adequately censused regions in 
eastern humid forests. 







Rano- 


Anala- 




Andrin- 


ma- 


ma- 




gitra* 


fanat 


/aotra 




720- 


575- 


500- 


Species 


2400 m 


1225 m 


1300 m 


Brachytarsomys albicauda 




X 


X 


Brachyuromys betsileoensis 


X 


X 


X 


Brachyuromys ramirohitra 


X 






Eliurus majori 


X 






Eliurus minor 


X 


X 


X 


Eliurus petteri 






X 


Eliurus tanala 


X 


X 


X 


Eliurus webbi 


X 


X 


X 


Gymnuromys roberti 


X 


X 


X 


Nesomys audeberti 


X 


X 


X 


Nesomys rufus 


X 


X 


X 


Monticolomys koopmani 


X 







* This study. 

t Carleton and Schmidt (1990); Ryan et al. (1993). 
X Compiled from Carleton and Schmidt (1990); Carle- 
ton (1994). 



Eliurus webbi Ellerman, 1949 

Identification and Distribution — Eliurus 
webbi suggests a more lightly built version of the 
larger E. tanala, but with a browner (rather than 
grayish) cast to its dorsal pelage and with a lush 
brown (rather than airy white) tail tuft. Other con- 
trasts are given in the previous account (and see 
Tables 22-2 and 22-7). 

As currently known, the species has the greatest 
range among Eliurus, occurring along the eastern 
versant of Madagascar, from its northernmost end 
(Montagne d'Ambre) to the southeastern lowlands 
(near Vondrozo and south of Farafangana, the 
type locality). Although much chromatic varia- 
tion, especially of the ventral pelage, is apparent 
among locality samples of E. webbi, the close 
agreement in size and proportions of the skin and 
skull, and in color and development of the tail 
pencil, reinforces the notion that only one species 
is represented. In general, the dimensions of E. 
webbi specimens taken in the RNI d'Andringitra 
(Tables 22-2 and 22-7) approximate those of an- 
imals collected in the PN de Ranomafana region 
(Carleton, 1994: appendix 2). 

In the RNI d'Andringitra, E. webbi was con- 
fined to the 720 and 810 m zones, at which latter 
elevation it is sympatric with E. tanala. Else- 
where, specimens of this broadly ranging species 



have been taken at sites from near sea level to 
about 1500 m (Carleton & Schmidt, 1990; Carle- 
ton, 1994). 

Ecology and Reproduction — Populations of 
E. webbi appeared approximately equal in abun- 
dance at 720 and 810 m (Tables 22-10 and 22-1 1). 

Four of 10 E. webbi obtained at 720 m were 
on and six off the ground, whereas 810 m the 
numbers in 13 captures were two and 11, respec- 
tively (Table 22-5). The only nesting site of Eli- 
urus found during the 1993 Andringitra survey 
was a ground burrow of E. webbi, a species for 
which only six of the 23 individuals (26%) cap- 
tured were taken with ground sets. Whether this 
burrow placement is typical of E. webbi is un- 
known, but this information combined with trap- 
ping results may indicate that this species nests 
and forages in different strata of the forest. 

In forest with open understory near the Iantara 
River (720 m), a pair of E. webbi were found 
living in the same underground burrow with nest- 
ing Scaly-breasted Ground-Rollers (Brachyptera- 
cias squamiger) (Goodman, 1994). The birds had 
built a nest in the principal tunnel of the burrow 
system, anterior to the rodents' food storage 
cache, the distalmost portion of the tunnel. 
Chewed seeds of Cryptocarya (Family Lauraceae) 
were found in the storage cache (see Chapter 4, 
Appendix 4-2). 

In contrast to E. tanala, remarkably few of the 



270 



FIELDIANA: ZOOLOGY 



23 E. webbi collected in November and December 
1993 showed signs of active reproduction. One of 
16 adult males had scrotal testes, and two of seven 
females had large mammae. No female dissected 
possessed embryos. 

Specimens Examined — 43 km S Ambalavao, 
junction of Sahanivoraky and Sahavatoy rivers, 
810 m (FMNH 151684 through 151686, 151741, 
151742, 151888 through 151889, and 151891 
through 151895); 45 km S Ambalavao, E bank of 
Iantara River, along Ambalamanenjana-Amba- 
tomboay trail, 720 m (FMNH 151680 through 
151683, 151739, 151740, and 151884 through 
151887). 



Gymnuromys Major, 1896a 

Recognition — This sleek-furred, dark grey rat 
is medium-sized but has a stout body, an impres- 
sion perhaps heightened by its relatively short tail 
and comparatively robust hind feet (Table 22-2). 
Cream to white-tipped hairs over slate bases, or 
sometimes wholly white hairs, give the ventral fur 
a silvery sheen. The tops of the metapodials and 
phalanges, both fore and rear, are light colored, 
covered with fine whitish to hyaline hairs. The 
hind foot is long but relatively broad, with the 
outer toes (I and V) somewhat shorter than the 
central three (II-IV). It bears six plantar pads. The 
tail is scantily haired, appearing naked to the 
unaided eye, and only slightly longer than head- 
and-body length (TL about 104% of HBL); it is 
countershaded, dusky above and white below, for 
much of its length, but in most individuals, the 
distal one-third to one-fourth of the caudal epi- 
dermis is white over the entire circumference. The 
six mammae are distributed as one postaxial, one 
abdominal, and one inguinal pair. 

The dorsal colors of young Rattus rattus or 
some Eliurus, such as E. majori, have a superfi- 
cial resemblance to the grayish-black dorsum of 
Gymnuromys. Closer inspection discloses that 
melanistic individuals of Rattus possess noticea- 
bly longer and monocolored tails (TL about 1 17% 
HBL). The fur of Eliurus majori is more woolly 
than sleek, and, like others in the genus, the tail 
is conspicuously longer than the length of the 
head-and-body and the hind feet are short and 
broad compared to those of Gymnuromys (Table 
22-2). In Eliurus, the dense caudal hairs, ending 
in a well-developed distal tuft, provide absolute 
separation from the naked tail of Gymnuromys. 

The smoothly contoured braincase, hourglass- 



shaped interorbital shape, short incisive foramina, 
and tiny auditory bullae of Gymnuromys recall 
features of the cranium in members of Eliurus. 
Specimens of Gymnuromys, however, lack a sub- 
squamosal foramen, which is typical of most Eli- 
urus, and their moderately hypsodont molars with 
convoluted enamel loops and compressed laminae 
are singularly distinctive (Major, 1 896a; Ellerman, 
1941). Size and proportions of the individual mo- 
lars that constitute each toothrow are unique with- 
in Nesomyinae, if not within Muroidea. The third 
molars, both uppers and lowers, are markedly 
wider and longer than the second molars, which 
are slightly wider and longer than the first molars, 
a reversal of the typical muroid pattern, in which 
the Mis > M2s > M3s. This unusual combination 
of occlusal configuration and intermolar propor- 
tions led Ellerman (1940) to create the monotypic 
subfamily, Gymnuromyinae, within a broadly de- 
fined Muridae. 



Gymnuromys roberti Major, 1896a 

Identification and Distribution — The small 
series from Andringitra is consistent in average 
size and range compared to the larger sample of 
G. roberti from Ampitambe, the type locality (Ta- 
ble 22-8). Indeed, examples of G. roberti through- 
out its range appear fairly homogeneous in mor- 
phology and pelage, although sample sizes, with 
the exception of Ampitambe, for critically assess- 
ing geographic variation, are paltry. 

The specimens from RNI d' Andringitra fall 
within the known geographic range of Gymnuro- 
mys, which inhabits eastern forest formations bor- 
dering the Central High Plateau. Carleton and 
Schmidt (1990) noted an elevational distribution 
based on museum records from 500 to 950 m. The 
elevational expanse of G. roberti is actually great- 
er, because it occurs at every site sampled, 720 to 
1625 m, in the 1993 Andringitra survey, and 
O'Keeffe and Ashmore (no date) reported it ear- 
lier, near Anjavidilava, at about 1800 m. 

Ecology and Reproduction — Although few 
individuals of G. roberti were captured in 1993, 
the species is one of the more broadly distributed 
on the slopes of the RNI d' Andringitra (Table 
22-10). Only Eliurus minor and Nesomys rufus 
are also found in all four elevational zones. 

Gymnuromys roberti appears to be an exclu- 
sively terrestrial rat. All six examples were caught 
on the ground, generally in sparse ground cover 
and equally split between traps placed in leaf lit- 



GOODMAN & CARLETON: RODENTS 



271 



ter, near dead wood, or by roots and trunks (Table 
22-5). O'Keeffe and Ashmore (no date) trapped a 
subadult in a pitfall line set across the ecotone 
between dry ericaceous habitat and boggy vege- 
tation at the edge of a small stream. 

A burrow of the species was discovered on the 
crest (810 m) of a slight ridge and in an area with 
relatively open understory. The entrance opened 
under a fallen log and continued for about 1 m as 
a relatively straight tunnel at about 15-20° to a 
small chamber. Twenty-two Canarium (Family 
Burseraceae) fruits were found in the chamber, 2 1 
of which had been gnawed open and the contents 
eaten (see Chapter 6). A partial cranium of an 
adult G. roberti was found in a National trap. A 
carnivore had dragged the trap from its initial 
placement site and eaten most of the rodent. 

Among the six Gymnuromys roberti collected, 
two adult males possessed abdominal testes, and 
two adult females had enlarged teats, one of 
which had two placental scars. A subadult male 
was taken on December 2, 1993 at 1210 m. 

Specimens Examined — 38 km S Ambalavao, 
ridge E of Volotsangana River, 1625 m (FMNH 
151695); 40 km S Ambalavao, along Volotsan- 
gana River, 1210 m (FMNH 151896, 151898, and 
151927); 43 km S Ambalavao, junction of Sahan- 
ivoraky and Sahavatoy rivers, 810 m (FMNH 
151694); 45 km S Ambalavao, E bank Iantara 
River, along Ambalamanenjana-Ambatomboay 
trail, 720 m (FMNH 151693). 



Monticolomys Carleton & Goodman, 1996 

Recognition — Diagnostic characters, differen- 
tiating comparisons, measurements, and biogeo- 
graphic comments are presented in Chapter 21. 



Monticolomys koopmani Carleton & 
Goodman, 1996 

Identification — The small size of this form 
renders its segregation from most Nesomyinae un- 
problematic (Table 22-2). Only specimens of Eli- 
urus minor are similarly diminutive, but they can 
be distinguished by numerous qualitative features 
of the skin and skull (see Chapter 21). 

Ecology and Reproduction — This species 
was found only in the 1625-m zone (Table 22-5). 
Both live-trapped individuals were captured in the 
same set placed on a nearly horizontal liana less 
than 10 cm in circumference. The third specimen 



was obtained in a pitfall bucket. With just three 
records, little can be said about its focus of activ- 
ity, whether predominantly arboreal or scansorial. 

The two males obtained possessed partially or 
fully descended testes; the one female had small 
teats and an imperforated vagina. 

Specimens Examined — 38 km S Ambalavao, 
ridge E of Volotsangana River, 1625 m (FMNH 
151727, 151899, and 151900). 



Nesomys Peters, 1870 

Recognition — These are comparatively big, 
thickset rats with long and narrow hind feet, large 
pinnae, and a tail not quite as long as the head- 
and-body (Table 22-2). Their soft fur is generally 
a reddish brown on the dorsum, often with a dense 
intermixture of black hairs mid-dorsally and a 
pronounced suffusion of chestnut to rufous col- 
oration on the cheeks, flanks, and rump. Through- 
out the distribution of the genus, the color of the 
ventrum varies, from a uniform rufous to pure 
white and all stages of intermediacy between 
them. The tail pilosity (in eastern Nesomys) is 
moderately developed, although it is not penicil- 
late, nor fully obscuring the scutellation; most 
caudal hairs are blackish, but the terminal seg- 
ment (about 10-20 mm) may be tipped with 
white; the length of the caudal vertebrae is either 
slightly less than or subequal to the head-and- 
body (TL about 88-97% of HBL). The confor- 
mation of the foot suggests cursorial locomotion 
and terrestrial habits: elongate and narrow, the lat- 
eral digits (I and V) markedly shorter than the 
central three, and the six plantar pads widely 
spaced. Mammary glands may usually consist of 
one pair abdominal and one pair inguinal (N = 
4), but in one individual a postaxial pair was ob- 
served (N = 6). 

As with external form, the cranium and denti- 
tion in examples of Nesomys are robust and per- 
mit ready identification (Table 22-9). The skull 
has a long rostrum, smoothly contoured braincase, 
stout and laterally flaring zygomatic arches, and a 
relatively wide, hourglass-shaped interorbit. The 
molars are incipiently high-crowned yet retain a 
cuspidate topography and distinct mesolophs and 
mesolophids. 

Carleton and Schmidt (1990) and Musser and 
Carleton (1993) cautioned that the taxonomy of 
Nesomys is more complex than had been con- 
veyed by the recognition of the one species, N. 
rufus (see, e.g., Petter, 1972, 1975, and Honacki 



272 



FTELDIANA: ZOOLOGY 



et al., 1982). At least three species — N audeberti 
(1879), N. lambertoni Grandidier (1928), and N. 
rufus ( 1 870) — merit recognition, but the status of 
other populations remains uncertain pending 
completion of a generic revision (Carleton, un- 
publ.). The association of names and morpholo- 
gies to the two species of Nesomys in the RNI 
d'Andringitra must therefore by considered ten- 
tative. 



Nesomys audeberti (Jentink, 1879) 

Identification and Distribution — Specimens 
of N. audeberti average greater than those of N 
rufus in most cranial and external dimensions, but 
individual variation and consequent range overlap 
are substantial. More reliable discrimination is 
provided by certain length (ONL, LR) and breadth 
(ZB, BBC, IOB) dimensions of the skull and glo- 
bosity of the auditory bullae (DAB), less so the 
size of the molars (Table 22-9). Length of the hind 
foot is a helpful yardstick (Ryan et al., 1993): typ- 
ically HFL (with claw) 2: 52 mm in samples of 
N. audeberti and HFL ^ 50 mm in those of N. 
rufus. In addition, most specimens of N. audeberti 
exhibit a large expanse of pure white hairs on 
their underparts, in contrast to the rufous ventral 
coloration characteristic of N rufus, but this pel- 
age distinction is not an absolute criterion of spe- 
cies assignment. So it is with the one specimen 
that we provisionally refer to N audeberti — it 
matches the large size of skin and skull typical of 
the form (Tables 22-2 and 22-9), yet it has an 
entirely rufescent ventrum that suggests an indi- 
vidual of N. rufus. 

The 810 m record from RNI d'Andringitra falls 
within the known altitudinal and geographic dis- 
tribution of N audeberti, which has an extensive 
north-south range in eastern lowland rain forest 
(Carleton & Schmidt, 1990). Elsewhere, N au- 
deberti and N. rufus have been demonstrated to 
occur in sympatry or actual syntopy in the regions 
of Analamazaotra and Ranomafana (Carleton & 
Schmidt, 1990; Ryan et al., 1993). Most localities 
where A/, audeberti has been documented are un- 
der 800 m, but potential overlap with the montane 
N. rufus appears likely within the 800-1000 m 
elevational zone, or approximately coincident 
with the floristic transition from lowland rain for- 
est to montane associations. 

Ecology and Reproduction — The lone male 
was captured on a ridge in pristine lowland rain 



forest; the live-trap was placed on leaf litter and 
by a tunnel entrance. He appeared reproductively 
immature or inactive (testes abdominal). 

Specimens Examined — 43 km S Ambalavao, 
junction of Sahanivoraky and Sahavatoy rivers, 
810 m (FMNH 151696). 



Nesomys rufus Peters, 1870 

Identification and Distribution — Separation 
of A/, rufus from N. audeberti on the basis of men- 
sural characters and pelage color is discussed in 
the preceding account (and see Tables 22-2 and 
22-9). 

The mean weight reported by Ryan et al. 
(1993) for male and female A/, rufus from the 
Ranomafana region generally falls within the 
range of the RNI d'Andringitra series. Naturally, 
pregnant females are exceptions; the extreme 
weight (208 g) obtained within the RNI d'An- 
dringitra sample was that of a female with two 
near-term embryos (crown-rump length = 47 mm 
each). Ryan et al. (1993) reported that adult males 
of N. rufus are significantly heavier than adult fe- 
males, but this pattern is not manifest in the RNI 
d'Andringitra sample, even when pregnant fe- 
males are removed from the comparison. None of 
the other five external measurements and only one 
(LR) of the 18 craniodental variables exhibited 
significant sexual dimorphism in size for the An- 
dringitra sample (11 males, 10 females). 

The RNI d'Andringitra marks the southernmost 
distributional point of N. rufus reported to date; it 
was previously documented only as far south as 
the PN de Ranomafana region, about 115 km to 
the northeast (Carleton & Schmidt, 1990). The 
presence of this richly colored Nesomys on the 
upper forested slopes of the Andringitra Massif 
(810-1625 m) concords with its montane habitus 
as recorded elsewhere on the island (900-2300 
m). 

Ecology and Reproduction — Nesomys rufus 
was captured in the 810, 1210, and 1625 m zones 
(Table 22-10). The number of individuals cap- 
tured increased with altitude (regression P = 0.07, 
r = 0.93, N = 4). Other Nesomys, species inde- 
terminate, were observed during daylight hours at 
720 m, but they seemed rare at this elevation and 
were never trapped. 

Nesomys rufus is strictly terrestrial and seems 
to prefer forests with an open understory and rel- 
atively thick mat of leaf litter (Table 22-5). Of the 
30 individuals trapped, two-thirds were in sets 



GOODMAN & CARLETON: RODENTS 



273 



placed in leaf litter on the forest floor; others were 
in a variety of situations under fallen tree trunks 
or among root entanglements, but always on the 
ground. In a radio-tracking study of N. audeberti 
and N. rufus at the PN de Ranomafana, Ryan et 
al. (1993) found both species to be exclusively 
diurnal and terrestrial. 

Several N. rufus were trapped at the openings 
of burrows that descended below fallen and rotten 
tree trunks or large limbs. These burrow systems 
often contained caches of fruits and nuts (e.g., 
Cryptocarya, Family Lauraceae, and Canarium, 
Family Burseraceae). Other food plants this di- 
urnal species was observed eating were Artabo- 
trys mabifolius (Family Annonaceae) and Sloanea 
rhodanta (Family Elaeocarpaceae) (see Chapter 4, 
Appendix 4-2). 

Signs of active reproduction were evident 
among the 31 N. rufus collected during the No- 
vember and December 1993 survey. Twelve of 18 
males possessed fully scrotal testes, and nine of 
12 females displayed prominently visible mam- 
mae. Three females with large teats had fresh pla- 
cental scars (mean = 3; range = 2-4), and four 
had one or two embryos in utero, varying in 
crown-rump length from 12 to 47 mm. 

Specimens Examined — 38 km S Ambalavao, 
ridge E of Volotsangana River, 1625 m (FMNH 
151702, 151703, 151747 through 151751, 151754, 
and 151908 through 151915); 40 km S Ambala- 
vao, along Volotsangana River, 1210 m (FMNH 
151699 through 151701, 151704, 151745, 
151746, and 151904 through 151907); 43 km S 
Ambalavao, junction of Sahanivoraky and Sahav- 
atoy rivers, 810 m (FMNH 151697, 151698, 
151753, 151901, and 151902). 



Discussion 

Trapping Effort and Sampling 
Confidence 

The 1993 small mammal survey of the eastern 
slopes of RNI d'Andringitra collected nine kinds 
of native rodents, as well as the introduced human 
commensal Rattus rattus. Another species, Brachy- 
uromys betsileoensis, had been collected previ- 
ously by the RCP survey in 1970-1971, and those 
voucher specimens are also reported above. The 
1 1 rodent species documented herein surpass by 
seven the number that had been previously re- 



corded in the reserve (Nicoll & Langrand, 1989; 
Carleton & Schmidt, 1990). 

During 5 weeks of fieldwork, 2,475 trap-days 
were approximately evenly divided among the 
four elevation sites (Tables 22-1 and 22-10) and 
produced 159 animals, 148 rodents, and 11 insec- 
tivores (see Chapter 20). No lemurs were live- 
trapped, although at least two species, Microcebus 
rufus and Cheirogaleus major, known to occur in 
the eastern reserve (see Chapter 25), fall within 
the size range that can enter the traps used (Na- 
tional and Sherman traps). 

Overall trap success of small mammals, includ- 
ing rodents and insectivores, varied from about 
3% at 720 m to approximately 8% at the two up- 
per elevation localities (Table 22-10). Insectivores 
(Microgale spp.) were captured in live-traps only 
at 1210 and 1625 m; trap success for rodents only 
at these two elevations was 7.0% and 7.7%, re- 
spectively. And when only the native nesomyine 
rodents are considered, the rate of capture de- 
clines slightly more, to 6.0% at 810 m, 7.0% at 
1210 m, and 7.4% at 1625 m. In general, these 
rates of trap return agree favorably with the re- 
sults of other surveys in Madagascar (Stephenson, 
1993; Goodman & Ganzhorn, 1994; Stephenson 
et al., 1994) and on other large tropical islands 
(Heaney et al., 1989; Rickart et al., 1991, 1993). 
Moreover, the consistently lower trap success in 
lowland rain forest at 720 m concords with the 
lower number of species, and estimates of relative 
density and relative biomass captured there (Ta- 
bles 22-10, 22-12, and 22-13). 

The rodents captured are either terrestrial or 
scansorial, although probably none is exclusively 
arboreal (Table 22-5). In several cases, especially 
for members of Eliurus, there appeared to be al- 
titudinal variation within species in the proportion 
taken on and off the ground. At least four, and 
probably five, rodent species on the Andringitra 
Massif are strictly ground-dwelling, namely 
Brachyuromys ramirohitra (and presumably B. 
betsileoensis), Gymnuromys roberti, Nesomys au- 
deberti, and N. rufus. Because the vast majority 
of traps were placed no more than 4 m above the 
ground (Table 22-1), canopy specialists, if they 
indeed exist in Malagasy rain forest, would likely 
have been overlooked. 

Two lines of evidence, derived from species ac- 
cumulation curves and analogous results from 
other small mammal inventories, bolster the no- 
tion that the trapping effort was suitable to estab- 
lish basic rodent diversity with the reserve. Plots 
of the cumulative number of species documented 



274 



FIELDIANA: ZOOLOGY 






Table 22-12. Relative numbers of rodents (given as the number of individuals trapped per 100 m of trap line) 
in the four elevational zones.* The number in parentheses is total length (in m) of trap lines for each zone. 





720 m 


810 m 


1210 m 


1625 m 




Species 


(620) 


(780) 


(715) 


(590) 


Grand mean 


Rattus rattus 




0.13 


0.56 


1.02 


0.53 


Brachyuromys ramirohitra 


... 






0.17 


0.17 


Eliurus majori 






1.40 


2.03 


1.68 


Eliurus minor 


0.81 


1.67 


0.98 


0.51 


1.04 


Eliurus tanala 




0.77 


1.26 


1.53 


1.15 


Eliurus webbi 


1.61 


1.67 






1.64 


Gymnuromys roberti 


0.16 


0.13 


0.42 


0.17 


0.22 


Nesomys audeberti 




0.13 






0.13 


Nesomys rufus 




0.64 


1.54 


2.37 


1.44 


Monticolomys koopmani 








0.34 


0.34 


All rodents 


2.58 


5.00 


5.60 


7.12 





Averaged only for elevational zones in which a species was trapped. 



at each elevational zone over the duration of trap- 
ping disclose distinct plateaus before the end of 
the survey (Fig. 22- la). That is, no additional ro- 
dent species were encountered after 200 to 500 
trap-days of census, depending upon the elevation 
and total number of species eventually recorded 
at each. Nevertheless, the decrease in species add- 
ed through time within each altitudinal zone does 
not correspond to any general decline in overall 
trap success during the course of the survey pe- 
riod (Fig. 22- lb). 

The number and composition of rodent species 
thus far documented for other places in the eastern 
humid forest at intermediate elevations provide 
circumstantial evidence about those that might be 
predicted to occur in the RNI d'Andringitra. Two 
general localities — PN de Ranomafana and Re- 
serve Speciale (RS) d'Analamazaotra (Perinet) — 
are informative in this regard, because consider- 
able small mammal inventories have been con- 



ducted in and around each, the former within the 
past few years and the latter since the early ex- 
plorations by naturalists. At their closest points, 
the PN de Ranomafana is 1 1 5 km northeast of the 
Andringitra Massif and RS d'Analamazaotra 
some 380 km northeast, but both are similarly sit- 
uated in moist forest along the eastern flank of the 
central highlands. Eight and nine species of Ne- 
somyinae have been reported from Ranomafana 
and Analamazaotra, respectively, or one less than 
the number now actually known from the RNI 
d'Andringitra (Table 22-11). Furthermore, the 
composition of indigenous rodents is similar 
among the three areas, because of the extensive 
geographic ranges of most species — such as Eli- 
urus minor, E. tanala, E. webbi, Gymnuromys 
roberti, and Nesomys audeberti — along the mid- 
dle elevation, forested eastern slopes of the Cen- 
tral High Plateau and the eastern escarpment. 
Of the several widely distributed nesomyine 



Table 22-13. Estimated biomass (g) of rodents trapped along an elevational transect. Summated from average 
weight of adults captured over the first 550 trap-days within each zone. 



Species 


720 m 


810 m 


1210 m 


1625 m 


Rattus rattus 




103 


414 


620 


Brachyuromys ramirohitra 


... 






117 


Eliurus majori 


... 




580 


869 


Eliurus minor 


148 


443 


258 


74 


Eliurus tanala 


... 


408 


735 


654 


Eliurus webbi 


506 


940 




... 


Gymnuromys roberti 


119 


119 


358 


119 


Nesomys audeberti 


... 


181 






Nesomys rufus 


... 


793 


1,427 


1,427 


Monticolomys koopmani 


... 






52 


Number of species 


3 


7 


6 


8 


Total biomass 


773 


2,964 


3,772 


3,932 


Average total biomass/species 


258 


494 


629 


492 



GOODMAN & CARLETON: RODENTS 



275 




100 200 300 400 500 

trap nights 



600 



700 



800 




100 



200 



300 



400 



500 



600 



700 



800 



trap nights 



Fig. 22- 1 . Species accumulation curves (top) and trap success (bottom) plotted for each elevational zone against 
the number of trap-nights. 



species, only one, Brachytarsomys albicauda, is 
curiously absent, thus far, from the rodent fauna 
in the RNI d'Andringitra (Table 22-11). This ar- 
boreal species is nocturnal, dens in tree holes, 
and, as a result, is relatively difficult to trap using 
standard techniques. However, it has the agreeable 
habit of chattering at the entrance of tree holes if 
disturbed during the day. On numerous occasions, 
in all elevational zones, S.M.G. located seemingly 
appropriate cavities for this species, but in no case 
did a Brachytarsomys pop its head out after the 
tree was thumped with a hard object, a technique 



that has worked elsewhere in Madagascar. Nor 
were local people who lived near the forest fa- 
miliar with any large arboreal rodent with a white- 
tipped tail. 

Other deficiencies in our knowledge of Mala- 
gasy rodents may affect the completeness of our 
rodent list. Of immediate relevance here, and as 
stressed by Carleton and Schmidt (1990), the em- 
pirical basis underlying comprehension of neso- 
myine distribution and taxonomic diversity is ex- 
tremely poor for an island-continent so climati 
cally, topographically, and floristically diverse 



276 



FIELDIANA: ZOOLOGY 



Also relevant is the want of critical systematic 
study of the limited museum material available. 
The description of Monticolomys (Chapter 2 1 ) is 
a telling case in point and underscores the possi- 
bility that still other undescribed rodents inhabit 
remote areas of the reserve and certainly of Mad- 
agascar. 

Excepting these caveats, which can be ad- 
dressed only by renewed field and taxonomic 
studies, extrapolation from the species accumu- 
lation curves and from the diversity uncovered in 
comparable survey efforts in eastern forest per- 
suades us that the rodent fauna on the eastern 
slopes of RNI d'Andringitra has been largely, 
though perhaps not exhaustively, documented. 



Elevation and Rodent Associations 

Over the limits of the transect (720-1625 m), 
species diversity becomes greater at higher ele- 
vations on the eastern slopes of the RNI d'An- 
dringitra. Of the nine nesomyines collected, the 
fewest species coexist at the 720 m locality (three) 
in lowland rain forest, whereas six or seven spe- 
cies were found at the three higher elevations, 
from the upper limit of lowland forest to upper 
montane (sclerophyllous) forest (Table 22-10), 
where the black rat also occurs. The bulge in spe- 
cies richness within montane forest associations 
mirrors the basic pattern found in undisturbed for- 
ested zones on mountains throughout the Old 
World tropics, whether on continental land masses 
or on large islands (see, e.g., Rupp, 1980; Yalden, 
1988; Heaney & Rickart, 1990). 

Differences in Relative Densities and Rela- 
tive Biomass — Relative density is herein estimat- 
ed as the number of individuals of a given species 
per 100 m of trap line, including all eight lines. 
Because species captures seldom differed notably 
between the two trap lines within the same ele- 
vational zone (the exception at 1625 m, where 
line 7 yielded 4.36 rodents per 100 m and line 8 
yielded 9.52 per 100 m), the results are averaged 
for each elevation of the survey. 

Overall rodent captures increase as a function 
of elevation, from an average 2.58 individuals per 
100 m at the 720 m site to 7.12 at 1625 m ele- 
vation (Table 22-12). Particularly marked shifts 
occur between the 720 and 810 m zones and again 
between 1210 and 1625 m. These trends parallel 
results obtained from other rodent studies on trop- 
ical islands (see, e.g., Heaney et al., 1989; Rickart 
et al., 1991, 1993). 



On the basis of total numbers captured, Rattus 
rattus, Eliurus majori, E. tanala, and Nesomys ru- 
fus are more common in the 1625 m zone than at 
lower elevations (Table 22-12). The relative abun- 
dance of Eliurus minor was lowest at the extremes 
of its altitudinal distribution (at 720 m, 0.81 in- 
dividuals per 100 m; at 1625 m, 0.51 individuals 
per 100 m), but it was greater at the middle ele- 
vations, particularly 810 m, where the relative 
density measured 1.67 individuals per 100 m. The 
captures of Gymnuromys roberti were relatively 
constant (0.13-0.42 individuals per 100 m) and, 
on average, the lowest of any rodent species oc- 
curring in more than one elevational zone. Eliurus 
webbi showed similar measures of relative density 
at the two lowest elevations; there were 1.61 in- 
dividuals per 100 m at 720 m and 1.67 individuals 
per 100 m at 810 m. 

Collective altitudinal comparison of numbers of 
the various rodent species reveals an interesting 
relationship between relative density and ubiquity 
of elevational occurrence. Species found in low 
densities (less than one individual per 100 m) — 
Rattus rattus, Brachyuromys ramirohitra, Gym- 
nuromys roberti, and Monticolomys koopmani — 
are either altitudinally limited or broadly distrib- 
uted on the eastern slopes of the reserve. Those 
of intermediate abundance (1-1.5 individuals per 
100 m) — Eliurus minor, E. tanala, and Nesomys 
rufus — are animals found at least across the band 
of montane-upper montane (sclerophyllous) for- 
est, from 810 to 1625 m. The last class (relative 
density > 1.5 individuals per 100 m) consists of 
two species, Eliurus majori and E. webbi, that re- 
place one another on the mountain, the latter oc- 
curring from 720 to 810 m and the former from 
1210 to 1625 m. 

Some tenets of interspecific competition theory 
hold that when morphologically similar congeners 
distributed along elevational gradients co-occur, 
particularly those species that utilize the same as- 
pects of the environment, their relative abun- 
dances will shift over the zone of overlap. A pos- 
sible case involves Eliurus webbi and E. tanala, 
both of which occur at 810 m elevation. In the 
former, relative densities were similar at 720 and 
810 m, but the latter demonstrated a clinal in- 
crease from the 810 to 1625 m localities. Al- 
though no reciprocal change in relative abundance 
of these two species was evident at the elevation 
of sympatry (810 m), there appeared to be a 
change in the portion of the forest strata used by 
E. tanala (Table 22-5). At 810 m, none of six E. 
tanala was taken on the ground, at 1210 m five 



GOODMAN & CARLETON: RODENTS 



277 



of nine (56%), and at 1625 m six of nine (67%). 
This increase in arboreality of E. tanala when in 
contact with E. webbi may be spurious, because 
the terrestrial incidence of E. webbi does not alter 
predictably in the presence of E. tanala. There is 
no persuasive evidence of competitive interaction 
between these two animals, or between any oth- 
ers, as measured by changes in relative numbers 
at zones of sympatry. 

Because the rodent species trapped in the re- 
serve range in average body weight from about 
25 g (Monticolomys koopmani) to over 180 g (Ne- 
somys audeberti) (Table 22-2), relative abun- 
dances of rodents on the eastern slopes of the 
massif may differ when measured by relative bio- 
mass compared to relative density. The standing 
relative biomass represented for each species per 
elevational zone is based on the total number of 
individuals captured during the first 550 trap-days 
and summarized by total and average relative ro- 
dent biomass as well (Table 22-13). 

There is a stepwise augmentation in relative 
biomass of rodents related to elevation, from 773 
g at 720 m, through 2,964 g at 810 m and 3,772 
g at 1210 m, to 3,932 g at 1625 m (Table 22-13). 
This relationship parallels the increase in total 
number of rodents captured (Table 22-10) and em- 
phasizes the prevalence of more heavy-bodied ro- 
dents, particularly Nesomys rufus, in higher num- 
bers at the upper elevations. The increase in rel- 
ative biomass is not simply related to species di- 
versity but also reflects the total number of 
rodents captured per elevational zone. For in- 
stance, the 1625 m zone, with eight species, sup- 
ported a greater total rodent relative biomass than 
did the 1210 m sector, with six species; neverthe- 
less, the average biomass per species is actually 
less in the former zone as compared to the latter 
(Table 22-13). In terms of both rodent species 
richness and total relative biomass, the forest at 
the 1625 m zone is the most productive. 

Multiple Captures in Single Trap Sets — In 
order to infer aspects of microhabitat utilization 
and, to a lesser degree, the social organization of 
rodent species, we tabulated cases when conspe- 
cific animals were captured in the same trap set. 
The principal concern was to determine whether 
there were any clear patterns in the species or age 
and sex of conspecific animals taken in the same 
trap, and to use this information to infer aspects 
of syntopy between species and social systems. 

A relatively small percentage of the traps used 
in each elevational zone captured at least two con- 
specific animals (Table 22-14). No species 



Table 22-14. Multiple captures of rodents in a sin- 
gle trap within each elevational zone. 



No. of 
traps 
with 



Traps with 
multiple captures 



Elevation 

720 m 

810 m 

1210 m 

1625 m 



Total with Same Different 
species captures species species 



12 
27 
37 
28 



4 
11 

8 
12 



showed a clear pattern in the sex ratio of conspe- 
cific animals taken in the same set, and thus noth- 
ing can be inferred about their social organization. 
In a few cases three or more individuals of E. 
majori, E. minor, and Nesomys rufus were taken 
in the same trap position. 

A second aspect of multiple trap captures in- 
volves cases of interspecific overlap in microhab- 
itats. When different species were obtained in the 
same trap set over the course of the survey, a 
microhabitat overlap between the species is sug- 
gested. At 720 m all trap sets that captured more 
than one individual consisted of conspecific ani- 
mals (Table 22-14). At 810 m, five of the 11 trap 
sets that caught more than two individuals result- 
ed in the capture of non-conspecific animals. 
These included: two traps in leaf litter and by 
holes in the ground, one of which captured a Ne- 
somys and Gymnuromys and the other a Nesomys 
and E. webbi; one set in a dense vine tangle and 
1 .5 m off the ground that captured E. minor and 
E. tanala; one trap placed on a horizontal liana, 
3 cm in circumference and 2.5 m off the ground, 
that trapped one E. minor and one E. webbi; and 
a trap on a dead 6-cm-circumference limb at 20° 
with one end on the ground and the other on a 
liana tangle that captured two E. minor and one 
E. webbi. 

At 1210 m elevation all eight traps that cap- 
tured more than one animal included at least two 
non-conspecific mammals (Table 22-14). Ground 
sets included: on a wet talus slope and covered 
with bracken fern, one Nesomys and Microgale 
cowani (FMNH 151777); at the base of a large 
tree with stilted roots, three Nesomys and one Rat- 
tus; on a downed, rotten, and moss-covered tree, 
one Nesomys and E. tanala; at the base of massive 
root tangle, one Gymnuromys and E. minor; and 
at the base of a tree with a large cavity, one E. 
minor and two Microgale dobsoni (FMNH 
151624, 151625). Elevated sets included traps 



278 



FIELDIANA: ZOOLOGY 



placed on a large downed tree spanning the banks 
across a 10-m-wide river and 3 m above the water, 
that captured three E. majori and one Rattus; on 
a horizontal portion of a 3-cm-circumference 
arched tree trunk that was connected to the ground 
and a massive vine tangle about 5 m off the 
ground, that trapped one E. majori and one Mi- 
crogale soricoides (FMNH 151778); and at the 
junction of an exposed rock outcrop and the hor- 
izontal tree roots, about 2-m high, covered with 
moss and epiphytes, that captured two E. minor 
and one E. majori. 

In the 1625 m zone, half of the 14 trap sets that 
captured at least two mammals included different 
species (Table 22-14). Traps placed on the ground 
consisted of two next to downed and rotten hol- 
low logs, that trapped a Gymnuromys and Rattus, 
and the other a Nesomys, E. tanala, and Rattus; 
at the opening of a natural underground passage- 
way associated with a massive root tangle covered 
by soil and leaf debris, that captured four E. ta- 
nala and one Brachyuromys; and under a root 
tangle, that captured E. tanala and Microgale tai- 
va (FMNH 151658). Arboreal settings included 
two on approximately 30-cm-circumference tree 
trunks at 25-30° and 1.5 m off the ground, one 
of which linked the ground to an extensive bam- 
boo thicket and captured two E. tanala and one 
E. majori, and the other that spanned the distance 
between the ground and another large tree and 
trapped two E. minor and one Rattus; and a trap 
placed on a horizontal tree trunk 6 cm in circum- 
ference and 1 .5 m off the ground that caught one 
E. tanala and one E. majori. 

At the two highest elevational zones, 1210 and 
1625 m, the number of trap sets that resulted in 
the capture of more than one mammal species was 
higher than at lower elevations (Table 22-14). In 
these two zones, Rattus, E. majori, E. minor, and 
E. tanala were often captured in the same arboreal 
trap sets, and there seems to have been consid- 
erable overlap between these species in substrate 
usage. For ground sets, Rattus, Gymnuromys, and 
Nesomys often occurred in the same microhabitat. 

Intra- and Interspecific Differences in Re- 
production — Little is known about the reproduc- 
tive biology of Malagasay rodents in general, nor 
has there been any information published on in- 
traspecific and interspecific variation along an ele- 
vational gradient to date. Information on breeding 
condition of the Andringitra rodents is summa- 
rized here (Table 22-15), but inferences about the 
seasonality of reproduction are necessarily limited 
by the brevity and timing of the field survey (5 



weeks; November 14, at 720 m, through Decem- 
ber 18, 1993 at 1625 m). 

When records for all rodent populations are 
combined per elevational zone, a clear pattern of 
increasing reproductive activity as a function of 
altitude emerges, ranging from only 25% of the 
rodents sampled at 720 m to over 80% at 1 625 m 
(Table 22-15). Examples of Eliurus tanala and 
Nesomys rufus display this elevational trend most 
sharply, with levels of reproductive activity in- 
creasing from 17 individuals and 0% at 810 m, 
34 animals and 73% at 1210 m, to 66 animals and 
86% at 1625 m, respectively. Eliurus minor is the 
only species found in three or more elevational 
zones that lacked clear evidence of heightened re- 
productive activity as a function of altitude. Rat- 
tus rattus may show a parallel pattern, but the 
numbers trapped were too few to draw conclu- 
sions. In this species there was a disproportionate 
number of males captured. Similarly, the small 
numbers of Gymnuromys roberti and Nesomys au- 
deberti captured are insufficient for comment. 

Among species of Eliurus there appears to be 
considerable variation in the extent and concur- 
rence of reproduction, at least during the months 
of November and December (Table 22-15). Of the 
23 E. webbi captured, two were subadults, and 
only three of 21 adults (13%) exhibited signs of 
active reproduction. The breeding status of this 
species' population contrasts with that of E. ma- 
jori, in which 19 of 22 individuals (86%) gave an 
indication of ongoing reproduction (males with 
scrotal testes and convoluted epididymides, and 
females lactating or carrying embryos). The num- 
bers of sexually active E. minor and E. tanala fall 
in between those of E. webbi and E. majori. 
Among all samples of Eliurus, few subadults were 
encountered; this may indicate that little breeding 
activity had transpired in the months preceding 
the survey. 

Although the temporal glimpse is brief, the 
breeding regime of rats of the genus Eliurus, and 
perhaps for most rodent species in the reserve, 
appears to be seasonal, and the inception of re- 
production in populations at higher elevations 
may precede the onset at lower elevations. 



The Rodent Faunas in Disturbed 
and Undisturbed Forest 

The two lower sites on the eastern slopes of the 
RNI d' Andringitra were chosen to enable a com- 
parison of levels of human activity and the pos- 



GOODMAN & CARLETON: RODENTS 



279 



4 



O — On — ' ■* fS N ^ M O 



O — -- — — (M--PPQ 



in O ^ ' » ' M^n 






NOirim- : : m : no 



■<fr O On r» Ov fS 



Tt — — o o o 



(N vO CN — — Vi 



e<s 



On — 



H *- *■■ 

a S 5 5 

C 3 

3 <j b 






■3 






a i s 



3 

OS OQ 1*3 t*3 



5 



-2 3 3 



5 "8 



•3 -3 .3 5 ^ ^ o 2 •-" <u 



1*1 S3 &£ =£ 5g £ # 



I 



sible effect on habitat degradation. The 720 and 
810 m elevations once harbored similar natural 
botanical communities, but the 720 m area is now 
adjacent to a heavily traveled trail between two 
villages and a small enclave of slash-and-burn ag- 
riculturalists, and the 810 m zone lacks noticeable 
signs of human disturbance (see Chapter 4 for bo- 
tanical descriptions). 

Differences in species composition were appar- 
ent between the rodent communities at these two 
elevations. Only three species (Gymnuromys rob- 
erti, Eliurus minor, and E. webbi) were trapped in 
the 720 m zone, whereas over twice as many spe- 
cies were evident at 810 m (Table 22-10). Three 
of the species that augment the 810 m list are the 
cosmopolitan Rattus rattus, Nesomys audeberti, 
and N. rufus — the latter two both diurnal species; 
one or both were infrequently observed at 720 m 
but regrettably not trapped. The only demonstra- 
ble contrast in native rodents between these zones 
is the presence of E. tanala at 810 m. 

The low number of Rattus rattus captured in 
the 720 and 810 m altitudinal zones is unexpected 
and counterintuitive, because this species is a hu- 
man commensal that prospers around dwellings 
and agricultural clearings such as those found 
near the 720 m site. However, no Rattus was re- 
corded in disturbed forest at 720 m, only one was 
trapped in the 810 m zone, and most individuals 
were collected at the two higher elevations in pris- 
tine montane and sclerophyllous forest. Thus, Rat- 
tus has successfully invaded some of the deepest 
portions of the reserve's forests. 

At other sites on Madagascar, Eliurus tanala 
has been recorded at elevations below 500 m 
(Carleton & Schmidt, 1990). Whether the absence 
of this species in the 720 m zone is part of the 
natural altitudinal stratification of rodents on the 
Andringitra Massif and/or the effect of habitat 
degradation is impossible to determine. 

As noted above, the number of rodent species 
at 810 m was over twice that observed in the par- 
tially disturbed forest at 720 m (Table 22-10). 
This contrast, however, is part of a larger picture 
of increasing species richness as a function of el- 
evation (see, e.g., Heaney & Rickart, 1990) and 
may have little to do with environmental degra- 
dation. The same point can be made for relative 
density and relative biomass — the 810 m locale 
sustains approximately two to four times the ro- 
dent populations compared to the 720 m com- 
munity (Tables 22-12 and 22-13) — but again these 
differences can be as convincingly attributed to 



280 



FIELDIANA: ZOOLOGY 



biological changes along an elevational gradient 
rather than to direct effects of habitat disturbance. 
Stephenson (1993) investigated the effects of 
human disturbance on small mammals inhabiting 
the RS d'Analamazaotra. He sampled four sites 
with levels of human disturbance estimated from 
to 100% and found a significant reduction in 
insectivore and rodent diversity and abundance 
relative to increased human activity. However, the 
extent of disturbance at his sites was significantly 
greater than that in the 720 and 810 m forests 
censused in the RNI d'Andringitra. It is interest- 
ing to note that Stephenson's (1993) trap success 
for Run us nut us in RS d'Analamazaotra was 
markedly higher in disturbed compared to undis- 
turbed areas. This was not the case along the east- 
ern slopes of the RNI d'Andringitra, where, as 
mentioned above, evidence suggests that Rattus 
has expanded its commensal role to successfully 
colonize primary forest. In summary, although 
there are clear differences in rodent species rich- 
ness at the 720 and 810 m transects, the differ- 
ences cannot be segregated from large-scale ele- 
vational trends of the biota on the eastern slopes 
of the reserve. 



Acknowledgments 

We are grateful to the Direction des Eaux et 
Forets and 1' Association Nationale pour la Ges- 
tion des Aires Protegees for authorization to con- 
duct this study, and in particular to M. Georges 
Rakotonarivo, Mme. C61estine Ravaoarinoroman- 
ga, and M. Louis Andriamahaly Rasolofo. Other 
participants in the Andringitra survey, in particu- 
lar Chris Raxworthy, provided help in numerous 
ways. We thank the various museum stewards 
who allowed us to examine specimens under their 
care (see Appendix 22-1 for definitions of acro- 
nyms): Guy G. Musser (AMNH); Paula Jenkins 
and Jean Ingles (BMNH); Bruce Patterson and 
Lawrence R. Heaney (FMNH); Malcolm J. Lar- 
gen (LMCM); Maria Rutzmoser (MCZ); Francis 
Petter and Michel Tranier (MNHN); Chris 
Smeenk (RMNH); and Hans Baggoe and Mogens 
Andersen (UZMC). Louise Emmons, Daniel Rak- 
otondravony, and Robert S. Voss provided critical 
comments on an earlier version of this manu- 
script. 



Literature Cited 

Barden, T. L., M. I. Evans, C. J. Raxworthy, J.-C. 
Razafimahaimodison, and A. Wilson. 1991. The 
mammals of Ambatovaky Special Reserve, pp. 5- 1 to 
5-20. In Thompson, P. M., and M. I. Evans, eds., A 
survey of Ambatovaky Special Reserve, Madagascar. 
Madagascar Environmental Research Group, London. 

Bartlett, E. 1879. Second list of mammals and birds 
collected by Mr. Thomas Waters in Madagascar. Pro- 
ceedings of the Zoological Society of London, 1879: 
767-770. 

Carleton, M. D. 1994. Systematic studies of Mada- 
gascar's endemic rodents (Muroidea: Nesomyinae): 
Revision of the genus Eliurus. American Museum 
Novitates, 3087: 1-55. 

, and G. G. Musser. 1984. Muroid rodents, pp. 

289-379. In Anderson, S., and J. K. Jones, Jr., eds.. 
Orders and Families of Recent Mammals of the 
World. John Wiley and Sons, New York, 686 pp. 

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. 

Cook, J. M., R. Trevelyan, S. S. Walls, M. Hatcher, 
and F. Rakotondraparany. 1991. The ecology of 
Hypogeomys antimena, an endemic Madagascan ro- 
dent. Journal of Zoology, 224: 191-200. 

Duckworthy, J. W, and F. Rakotondraparany. 1990. 
The mammals of Marojejy, pp. 54-60. In Safford, R., 
and J. W Duckworth, eds., A Wildlife Survey of the 
Marojejy Nature Reserve, Madagascar. International 
Council for Bird Preservation, Cambridge, 172 pp. 

Ellerman, J. R. 1940. The Families and Genera of Liv- 
ing Rodents. Volume 1, Rodents Other Than Muridae. 
London: British Museum (Natural History), xxvi+689 
pp. 

. 1941. The Families and Genera of Living Ro- 
dents. Volume 2, Family Muridae. London: British 
Museum (Natural History), xii+690 pp. 

-. 1949. The Families and Genera of Living Ro- 



dents. Volume 3, Appendix II [Notes on the rodents 
from Madagascar in the British Museum, and on a 
collection from the island obtained by Mr. C. S. 
Webb]. London: British Museum (Natural History), 
v+210 pp. 

Goodman, S. M. 1994. A description of the ground 
burrow of Eliurus webbi (Nesomyinae) and case of 
cohabitation with an endemic bird (Brachypteraciidae, 
Brachypteracias). Mammalia, 58: 670-672. 

. 1995. Rattus on Madagascar and the dilemma 

of protecting the endemic rodent fauna. Conservation 
Biology, 9: 450-453. 

-, and J. Ganzhorn. 1994. Les petits mammifer- 



es, pp. 58-63. In Goodman, S. M., and O. Langrand, 
eds., Inventaire biologique ForSt de Zombitse. Re- 
cherches pour le Developpement, S6rie Sciences bio- 
logiques, No. Special. Centre d' Information et de Do- 
cumentation Scientifique et Technique, Antananarivo, 
106 pp. 
Goodman, S. M., and D. Rakotondravony. In press. 
The Holocene distribution of Hypogeomys (Rodentia: 



GOODMAN & CARLETON: RODENTS 



281 



Muridae: Nesomyinae) on Madagascar. In Lourenco, 
W. R., ed., Biogeographie de Madagascar. Editions 
ORSTOM, Paris. 

Grandidier, G. 1928. Description d'une nouvelle es- 
pece de Nesomys, le N. lambertoni, G. Grand. Bulletin 
de l'Academie Malgache, Nouvelle S6rie, 11: 95-99. 

Heaney, L. R., P. D. Heideman, E. A. Rickart, R. B. 
Utzurrum, and J. S. H. Klompen. 1989. Elevational 
zonation of mammals in the central Philippines. Jour- 
nal of Tropical Ecology, 5: 259-280. 

Heaney, L. R., and E. A. Rickart. 1990. Correlations 
of clades and clines: Geographic, elevational, and 
phylogenetic distribution patterns among Philippine 
mammals, pp. 321-332. In Peters, G, and R. Hutterer, 
eds., Vertebrates in the Tropics. Museum Alexander 
Koenig, Bonn, 585 pp. 

Honacki, J. H., K. E. Kinman, and J. W. Koeppl. 1982. 
Mammal Species of the World. Lawrence, Kansas: Al- 
len Press and the Association of Systematics Collec- 
tions, ix+694 pp. 

Jentink, F. A. 1879. On a new genus and species of 
Mus from Madagascar. Notes from the Leyden Mu- 
seum, 1879: 107-109. 

King, C. M. 1990. The Handbook of New Zealand 
Mammals. Oxford University Press, Oxford, 600 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. 1896a. Diagnoses of new mam- 
mals from Madagascar. Annals & Magazine of Natu- 
ral History, series 6, 18: 318-325. 

. 1896b. Descriptions of four additional new 

mammals from Madagascar. Annals & Magazine of 
Natural History, series 6, 18: 461-463. 

Milne Edwards, A. 1885. Description d'une nouvelle 
espece de rongeur provenant de Madagascar. Annales 
des Sciences naturelles, Zoologie et Paleontologie 
(Paris), 20: Article 1 bis. 

Musser, G. G. 1987. The mammals of Sulawesi, pp. 
73-93. In Whitmore, T C, ed., Biogeographical Evo- 
lution of the Malay Archipelago. Clarendon Press, 
Oxford. 

, and M. D. Carleton. 1993. Family Muridae, 

pp. 501-755. In Wilson, D. E., and D. M. Reeder, eds., 
Mammal Species of the World: A Taxonomic and 
Geographic Reference, 2nd edition. Smithsonian In- 
stitution Press, Washington, D.C., 1,026 pp. 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: 
Revue de la conservation et des Aires Protegees. 
World Wide Fund for Nature, Gland, Switzerland, 
xvii+374 pp. 

Nicoll, M. E., F Rakotondraparany, and V. Randri- 
anasolo. 1988. Diversity des petits mammiferes en 
foret tropicale humide de Madagascar, analyse preli- 
minaire, pp. 241-252. In Rakotovao, L., V. Barre, and 
J. Sayer, eds., L'Equilibre des ecosystemes forestiers 
a Madagascar: Actes d'un seminaire international. 
IUCN, Gland, Switzerland, 338 pp. 

O'Keeffe, J., and L. Ashmore. No date. Madagascar 
1993 small mammal project. Emmanuel College, 
Cambridge University. 



Paulian, R., J. M. Betsch, J. L. Guillaumet, C. Blanc, 
and P. Griveaud. 1971. RCP 225. Etudes des eco- 
systemes montagnards dans le region malgache. I. Le 
massif de l'Andringitra, 1970-1971. Geomorphologie, 
climatologie et groupements veg&aux. Bulletin de la 
Societe d'Ecologie, 11(2-3): 198-226. 

Peters, W 1870. Uber Nesomys rufus, eine neue gat- 
tung und art madagascarischer nager. Sitzungs-Beri- 
chte der Gesellschaft naturforsender Freunde, Berlin, 
1870: 54-55. 

Petter, E 1972. The rodents of Madagascar: The seven 
genera of Malagasy rodents, pp. 661-665. In Battis- 
tini, R., and G. Richard- Vinard, eds., Biogeography 
and Ecology in Madagascar. The Hague, W Junk 
B.V., xv+765 pp. 

. 1975. Family Cricetidae: Subfamily Nesomyi- 
nae. Part 6.2: 1-4. In Meester, J., and H. W Setzer, 
eds., The Mammals of Africa: An Identification Man- 
ual. Smithsonian Institution, Washington, D.C. 

1990. Relations de parente des rongeurs de 



Madagascar. Atti dei Convegni Lincei, 85: 829-837. 

Rakotondravony, D. A. 1987. Les rongeurs a Mada- 
gascar, pp. 93-94. In Mittermeier, R. A., L. H. Rak- 
otovao, V. Randrianasolo, E. J. Sterling, and D. De- 
vitre, eds., Priori tes en Matiere de Conservation des 
Especes a Madagascar. IUCN, Gland, Switzerland, 
167 pp. 

. 1992. Etude comparee de trois rongeurs des 

milieux malgaches: Rattus norvegicus Berkenhout, 
(1769), Rattus rattus Linne, (1757) et Eliurus sp. 
(biologie et dynamique des populations). These Doct. 
3eme cycle, Sciences Biologiques Appliquees, Eco- 
logie Animale, Universite d'Antananarivo, Antana- 
narivo, 281 pp. 

Rickart, E. A., L. R. Heaney, and R. C. B. Utzurrum. 
1991. Distribution and ecology of small mammals 
along an elevational transect in southeastern Luzon, 
Philippines. Journal of Mammalogy, 72: 458-469. 

Rickart, E. A., L. R. Heaney, P. D. Heideman, and R. 
C. B. Utzurrum. 1993. The distribution and ecology 
of mammals on Leyte, Biliran, and Maripipi Islands, 
Philippines. Fieldiana: Zoology, new series, 72: 1-62. 

Rupp, H. 1980. Beitrage zur Systematik, Verbreitung 
und Okologie athiopischer Ergebnisse mehrerer For- 
schungsreisen. Saugetierkundliche Mitteilungen, 28: 
81-123. 

Ryan, J. M., G. K. Creighton, and L. H. Emmons. 
1993. Activity patterns of two species of Nesomys 
(Muridae: Nesomyinae) in a Madagascar rain forest. 
Journal of Tropical Ecology, 9: 101-107. 

Spencer, H. J., and D. E. Davis. 1950. Movements and 
survival of rats in Hawaii. Journal of Mammalogy, 31: 
154-157. 

Stephenson, P. J. 1993. The small mammal fauna of 
Reserve Speciale d'Analamazaotra, Madagascar: The 
effects of human disturbance on endemic species di- 
versity. Biodiversity and Conservation, 2: 603-615. 

, H. Randriamahazo, N. Rakotoarison, and P. 

A. Racey. 1994. Conservation of mammalian species 
diversity in Ambohitantely Special Reserve, Mada- 
gascar. Biological Conservation, 69: 213-218. 

Thomas, O. 1895. On a new species of Eliurus. Annals 



282 



FIELDIANA: ZOOLOGY 



& Magazine of Natural History, series 6, 16: 164- 

165. 
. 1908. A new species of the Mascarene genus 

Eliurus. Annals & Magazine of Natural History, series 

8, 2: 453-454. 
Yalden, D. W. 1988. Small mammals of the Bale 

Mountains, Ethiopia. African Journal of Ecology, 26: 

281-294. 



Appendix 22-1. 

Comparative Material 

Enumerated below are the non-Andringitra 
specimens that formed the basis for the sample 
statistics and tabular comparisons presented in the 
various accounts of species. They are contained 
in the following museums: American Museum of 
Natural History (AMNH), New York City; The 
British Museum (Natural History) (BM[NH]), 
London; The Field Museum of Natural History 
(FMNH), Chicago; Merseyside County Museums 
(LMCM), Liverpool; Museum of Comparative 
Zoology (MCZ), Harvard University, Cambridge; 
Museum National d'Histoire Naturelle (MNHN), 
Paris; Rijksmuseum van Natuurlijke Histoire 
(RMNH), Leiden; National Museum of Natural 



History (USNM), Smithsonian Institution, Wash- 
ington, D.C.; and Universitets Zoologisk Museum 
(UZMZ), Copenhagen. 

Brachyuromys betsileoensis — Fianarantsoa Prov- 
ince: Ampitambe (BMNH 97.9.1.126, 98.3.8.11- 
13, 48.290, 48.292, 79.139; LMCM A19.4.98.23; 
UZMC 1823, 7943). 

Brachyuromys ramirohitra — Fianarantsoa Prov- 
ince: Ampitambe (BM[NH] 97.9.1.133-7, 
97.9.1.139, 98.3.8.14-16, 48.282-9, 74.764, 
79.140, 1987.111; LMCM A19.4.98.24; MCZ 
12433, 45935, 46262; RMNH 26525; UZMC 
1223, 7944). 

Eliurus majori — Antsiranana Province: Mon- 
tagne d'Ambre (AMNH 100687, 100854). Fia- 
narantsoa Province: Ambohimitombo (BM[NH] 
97.9.1.147; MCZ 45929). 

Eliurus minor — Fianarantsoa Province: Ampi- 
tambe (BM[NH] 97.9.1.153; FMNH 5629). 

Gymnuromys roberti — Fianarantsoa Province: 
Ampitambe (BM[NH] 97.9.1.140-3, 97.9.1.146, 
98.3.8.8-10, 1939.1893, 48.296; FMNH 5632; 
LMCM A19.4.98.27; MCZ 45930-1; MNHN 
1897.537; RMNH 26523-4; UZMC 1220, 1824, 
7942). 

Monticolomys koopmani — Antananarivo Prov- 
ince: Manjakatompo (AMNH 100727). 



GOODMAN & CARLETON: RODENTS 



283 



Chapter 23 

Results of a Bat Survey of the Eastern 
Slopes of the Reserve Naturelle Integrate 
d'Andringitra, Madagascar 

Steven M. Goodman 



Abstract 

Four bat species were recorded in the Reserve Naturelle Integrate d'Andringitra: Rousettus 
madagascariensis, Hipposideros commersoni, Myotis goudoti, and Miniopterus minor. The 
greatest species richness was at 720 and 810 m. Evidence was found that Myotis and Miniop- 
terus occupy the same day-roosts. There was also the indication that Miniopterus minor has 
synchronized breeding and delayed embryonic development. 

Resume 

Quatre especes de chauve-souris ont ete recensees dans la Reserve Naturelle Integrale 
d'Andringitra: Rousettus madagascariensis, Hipposideros commersoni, Myotis goudoti et Min- 
iopterus minor. La plus grande richesse specifique a ete constatee entre 720 m et 810 m 
d' altitude. La preuve que les especes Myotis et Miniopterus occupent des memes reposoirs 
diurnes a ete etablie. On a aussi releve l'indication que Miniopterus minor a une reproduction 
synchronised et un developpement embryonnaire retarde. 



Introduction 

Information on the bats of Madagascar is mea- 
ger. Dorst (1947a,b; 1948) reviewed the species 
diversity and biogeography of Malagasy bats. 
Over the intervening years, species occurring on 
the island have been included in numerous taxo- 
nomic reviews, but there is still little known about 
their distribution and natural history. The late R. 
L. Peterson of the Royal Ontario Museum (ROM), 
Toronto, was working on systematic revisions of 
the Malagasy bat fauna at the time of his death in 
1989, on the basis of material he collected and 
specimens housed in other museums. J. Eger and 
L. Mitchell, of the same institution, are helping to 
see this work to fruition. 

Our knowledge of the bat faunas of most re- 
serves is rudimentary or nonexistent. The major 



exception is the Reserve Naturelle Integrale (RNI) 
de Marojejy, where a group from Aberdeen Uni- 
versity conducted a survey of bats in 1989 (Pont 
& Armstrong, 1990). In their review of the land 
vertebrates of the RNI d'Andringitra, Nicoll and 
Langrand (1989) did not list any bat species. 



Methods 

During the first trip into the eastern portion of the 
reserve (September 23 to November 1, 1993), nets 
set up for birds in the 720, 810, 1210, and 1625 
m transect zones were checked at dusk and dawn 
for bats. In a few cases nets were erected specif- 
ically to capture bats. Therefore the bat species 



284 



FIELDIANA: ZOOLOGY 



list resulting from this survey should not be con- 
sidered complete. 

The nets used were 12 m long and 2.6 m high, 
with four shelves and 36-mm mesh. All nets were 
erected with the bottom panel touching or within 
30 cm of the ground. Specimens are deposited at 
the Field Museum of Natural History (FMNH), 
Chicago, and a portion will be returned to the De- 
partement de Biologie Animale, University 
d' Antananarivo. The nomenclature of Koopman 
(1993) is followed. 



WC, width across canines: measured across the 
exteriormost alveolar base of the upper ca- 
nines. 

WT, weight: measured using Pesola spring 
scales. Animals weighing up to 10 g were 
weighed to the nearest 0.1 g; those between 
11 and 100 g were weighed to within 0.5 g. 

ZB, zygomatic breadth: the greatest distance 
between the lateral surfaces of the zygomatic 
arches. 



Measurements 

Measurements were made of animals in the 
flesh and from prepared crania. The abbreviations 
and definitions for measurements (all in mm, with 
the exception of WT) follow. 

BBC, breadth of braincase: the distance mea- 
sured across the hamular processes of the 
squamosal at the point where they border the 
mastoid bullae. 

CM 3 , canine-molar length: measured from the 
anterior alveolar border of canine to posterior 
alveolar boarder of M 3 . 

EL, ear length: measured from the notch at the 
base of the ear to the distalmost edge of the 
pinna. 

FA, forearm length: measured from the outside 
edge of the wrist to the outside edge of the 
elbow (with wing folded). 

HF, hind foot length: measured from the heel to 
the end of the longest toe (not including 
claw). 

IOB, interorbital breadth: the minimum dis- 
tance across the frontal bones between the 
orbits. 

ML, mandible length: measured from the mid- 
point of mandibular condyle to the anterior- 
most point of dentary. 

ONL, occipitonasal length: the distance be- 
tween the tip of the nasals and the posterior- 
most edge of the occiput, just above foramen 
magnum. 

TL, tail length: measured from the base of the 
tail (at right angles to the body) to the end 
of the distalmost vertebra. 

TOTL, total length of body and tail: measured 
from the nose tip to the end of the distalmost 
tail vertebra. 

TR, tragus length: measured from the base of 
tragus to the distalmost tip. 



Species Accounts 
Family Pteropodidae 
Rousettus madagascariensis Grandidier, 1929 

This species was captured once. A subadult fe- 
male was netted on September 27, 1993 in the 
forest along the edge of the Iantara River at 720 
m. The highest recorded altitude for this species 
is 990 m (Bergmans, 1994). Measurements of this 
specimen generally fall within the ranges given 
by Bergmans (1994) for this species (Table 23-1). 

Reproduction — The single captured individual 
had small mammae and an unperforated vagina. 

Family Rhinolophidae 

Hipposideros commersoni commersoni 
(E. Geoffroy, 1813) 

This species was observed or netted in the 720, 
810, and 1210 m transect zones. It was often ob- 
served at night, flying along trails, hawking bee- 
tles and other large insects. After subduing its 
prey, it would generally carry the food to a fre- 
quently used perch for consumption. The single 
H. commersoni netted was at 810 m in an area of 
open forest and at the edge of a ravine. This spe- 
cies was less common at 1210 m than at lower 
elevations. Measurements are presented in Table 
23-1. 

Reproduction — The single netted individual 
was a female with a perforated vagina, small 
mammae, and no embryos or placental scars. 

Family Vespertilionidae 
Myotis goudoti (A. Smith, 1834) 

This species was only recorded in the 810 m 
transect zone. All seven individuals were captured 






GOODMAN: BAT SURVEY 



285 



Table 23-1. Selected measurements of bats collected during the survey. 



Species 


Age 


TOTL 




TL 




HF 


EL 


Rousettus 


Subadult 


121 




14 




15 


18 


madagascariensis 
















Hipposideros 


Adult 


139 




42 




16 


30 


commersoni 
















Myotis goudoti 


Adult 


94.7 ± 2.36 




44.7 ± 3.15 




6.9 ± 0.69 


14.7 ± 0.76 






92-97, N = 


7 


40-49, N = 


7 


6-8, N = 7 


14-16, N = 7 


Miniopterus minor 


Adult 


97.5 ± 5.99 




46.3 ±4.12 




6.4 ± 0.50 


10.6 ± 0.50 






89-109, N = 


= 40 


39-55, N - 


40 


6-7, N = 40 


10-11, N = 40 



Note: Descriptive statistics are presented as means ± SD, minimum-maximum, number of specimens. 



at the same spot and were netted together with 
Miniopterus minor. On the basis of their dusk 
flight pattern and the synchronized timing with 
which these bats hit the net, it is possible that the 
two species were occupying the same roost. My- 
otis goudoti were distinctly less common than the 
Miniopterus minor, in a ratio of about 1:10. Mea- 
surements of the captured Myotis are presented in 
Table 23-1. 

Reproduction — Three specimens were dissect- 
ed: a female with large mammae and a single em- 
bryo measuring 6 mm crown-rump length, a fe- 
male with small mammae and a perforated vagina, 
and a male with a slightly convoluted epididymis 
and abdominal testes measuring 3X2 mm. 



Miniopterus minor Peters, 1867 

This species was the most commonly captured 
bat in the reserve; it was recorded in the 720, 810, 
and 1210 m zones. The single M. minor collected 
at 720 m was netted at dusk over the Iantara Riv- 
er. Thirty-one of the 40 M. minor collected during 
the survey came from one net set at 810 m in an 
area within the forest near rock outcrops and just 
above the Sahanivoraky River. Individuals from 
1210m were captured in two different nets placed 
in the forest along a ravine with exposed rocks. 
All individuals are referable to Miniopterus minor 
manavi (Hill, 1993). At other localities on the is- 
land, Miniopterus roosts often contain several 
species of this genus (Pont & Armstrong, 1990; 
Hill, 1993), but only one species was found in the 
RNI d'Andringitra. Measurements of collected in- 
dividuals are presented in Table 23-1. 

Reproduction — between mid-August and late 
September 1993, 26 M. minor were collected, of 



which 13 were adult males with small abdominal 
testes, never exceeding 2X2 mm, and without 
convoluted epididymides. Thirteen were adult fe- 
males (all of which had small mammae), eight of 
which had one embryo measuring on average 10.8 
mm (range 8-12 mm); the remainder had no em- 
bryos. In late November 1993 an additional seven 
females were collected, three of which had single 
embryos varying in size from 18 to 19 mm and 
large mammae; two had recently given birth and 
were lactating, and two showed no signs of recent 
reproduction. None of the seven males obtained 
in late November 1993 was in reproductive con- 
dition. On the basis of these specimens, the fol- 
lowing observations can be extrapolated about the 
reproductive biology of this species: (1) females 
tend to be highly synchronized in their reproduc- 
tive regime; (2) either only a few males are sex- 
ually active throughout the year (and none was 
mist-netted) or they tend to be in reproductive 
condition for only restricted portions of the year 
(as known for other species in this genus; Med- 
way, 1971); and (3) this species may have delayed 
embryonic development, such as has been report- 
ed in several species of Miniopterus (Racey, 
1982). 



Summary 

A total of four bat species were recorded in the 
RNI d'Andringitra (Table 23-2). The greatest spe- 
cies richness was in the lowland area, below 810 
m, with three taxa. Only two species, Hipposide- 
ros commersoni and Miniopterus minor, were re- 
corded in the 1210 m zone. No bat was found at 
1625 m or above. 



286 



FIELDIANA: ZOOLOGY 



Table 23-1. Extended. 



TR 



FA 



WT 



ONL 



ZB 



IOB 



8.0 ± 0.58 
7-9, N = 7 
5.9 ± 0.27 



69 
90 

39.0 ± 1.29 

38-41, N = 7 
40.3 ± 1.63 



49.0 

5.8 ± 0.41 
5.3-6.4, N 
6.3 ± 0.98 



34.9 


19.4 


6.5 


30.0 


16.2 


3.1 


14.7, 14.9, 15.7 


9.0,9.1,9.4 


3.1, 3.1, 3.2 


13.9 ±0.13 


7.5 ±0.14 


3.3 ±0.14 



5-6, N = 40 38-43, N = 40 4.6-9.1, N = 40 13.7-14.1, N = 11 7.4-7.9, N = 1 1 3.0-3.5, N = 12 



Table 23-1. Extended. Continued. 



Species 



BBC 



WC 



CM 3 



ML 



Rousettus 


14.8 


6.0 


12.6 


26.3 


madagascariensis 










Hipposideros 


13.8 


8.4 


10.7 


20.6 


commersom 










Myotis goudoti 


7.5, 7.7, 7.8 


4.0, 4.0, 4.0 


5.5, 5.6, 6.0 


10.6, 10.9, 11.6 


Miniopterus minor 


7.5 ± 0.22 


3.9 ±0.15 


5.2 ± 0.07 


10.2 ± 0.28 




7.3-7.9, N = 1 1 


3.7-4.2, N = 1 1 


5.1-5.3, N= 12 


9.7-10.6, N = 12 



Table 23-2. Altitudinal distribution of bats on the eastern slopes of the RNI d'Andringitra. 



720 m 



810 m 



1210 m 



1625 m 



Chiroptera 








Family Pteropodidae 








Rousettus madagascariensis 


+ 






Family Rhinolophidae 








Hipposideros commersoni 


+ 


+ 


+ 


Family Vespertilionidae 








Myotis goudoti 




+ 




Miniopterus minor 


+ 


+ 


+ 


Total number of species 


3 


3 


2 



Acknowledgments 

For comments on an earlier draft of this paper 
I am grateful to J. Eger and L. Heaney. 



Literature Cited 

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, s6r. 2, 19: 306-313. 



. 1947b. Essai d'une clef de determination des 

chauves-souris Malgaches. M6moires de lTnstitut 
Scientifique de Madagascar, s6r. A, 1: 81-88. 

. 1948. Biogeographie des Chiropteres malgach- 
es. M6moires de lTnstitut Scientifique de Madagascar, 
s6r. A, 1: 193-198. 

Hill, J. E. 1993. Long-fingered bats of the genus Min- 
iopterus (Chiroptera: Vespertilionidae) from Madagas- 
car. Mammalia, 57: 401-405. 

Koopman, K. E 1993. Order Chiroptera, pp. 137-241. 
In Wilson, D. E., and D. M. Reeder, eds., Mammal 
Species of the World: A Taxonomic and Geographic 
Reference, 2nd edition. Smithsonian Institution Press, 
Washington, D.C., 1,206 pp. 

Medway, Lord. 1971. Observations of social and re- 



GOODMAN: BAT SURVEY 



287 



productive biology of the bent-winged bat Miniopte- the bat fauna of the Reserve Naturelle Integrate de 

rus australis in northern Borneo. Journal of Zoology, Marojejy in north-east Madagascar. Report of the Ab- 

165: 261-273. erdeen University expedition to Madagascar 1989. 

.. „ ~ T ,„„,. w , Department of Zoology, University of Aberdeen, Ab- 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: erdeen 57 pp 

Revue de la conservation etdes aires Protegees. World Racey> £ A m2 Eco , of ^ duction 

Wide Fund for Nature, Gland, xvn+374 pp. 57 _ 104 Jn Kunz> T H ed Ecology of Bats plenum 

Pont, S. M., and J. D. Armstrong. 1990. A study of Press, New York, 425 pp. 



288 FIELDIANA: ZOOLOGY 



Chapter 24 

The Carnivores of the Reserve Naturelle 
Integrate d'Andringitra, Madagascar 

Steven M. Goodman 



Abstract 

The eastern slopes of the Reserve Naturelle Integrate d'Andringitra hold five native {Galidia 
elegans, Galidictis fasciata, Cryptoprocta ferox, Eupleres goudotii, and Fossa fossana) and 
three introduced carnivores {Canis lupus, Felis silvestris, and Viverricula indica). The greatest 
diversity was at 820 m, with four native species. The reserve has one of the richest carnivore 
communities known from any site on the island. 

Resume 

On a recense" cinq especes indigenes de cranivores {Galidia elegans, Galidictis fasciata, 
Cryptoprocta ferox, Eupleres goudotti et Fossa fossana) et trois especes introduites {Canis 
lupus, Felis silvestris et Viverricula indica) sur le versant est de la Reserve Naturelle Integrate 
d'Andringitra. A une altitude de 820 m, quatre especes indigenes ont 6l6 inventorizes ce qui 
reprZsente la plus grande diversite specifique. La reserve recele la plus riche communaute" de 
carnivore connue parmi tous les sites de l'tle. 



Introduction 

During the 1993 survey of the R6serve Natu- 
relle Integrate (RNI) d'Andringitra no systematic 
work was conducted on carnivores, but some data 
were obtained on the locally occurring species 
and their elevational ranges by incidental obser- 
vations and a few trap captures. The only known 
published information on the carnivores of the re- 
gion was presented by Albignac (1973) and Nicoll 
and Langrand (1989). Wozencraft (1993) is fol- 
lowed for the order and taxonomy of the carni- 
vores. 

Species Accounts 

Family Canidae 

Canis lupus Linnaeus, 1758 

Dogs were relatively common in the village of 
Ambarongy, near our 720 m camp, and they were 



seen with some regularity in the company of peo- 
ple traveling along the trail between Ambalama- 
nenjana and Ambatomboay. A dog was also ob- 
served in the tavy below our 810 m camp and 
along the Sahavatoy River. This species was in- 
troduced to Madagascar by humans. 



Family Felidae 

Felis silvestris Schreber, 1775 

Feral domestic cats were found near human set- 
tlements and in lowland forest. Cats were regu- 
larly noted within 500 m of the village of Am- 
barongy, on both sides of the Iantara River, and 
near our 720 m camp. Cats were also heard and 
observed in the vicinity of our 810 m camp. This 
species is not native to Madagascar. 



GOODMAN: CARNIVORES 



289 



Family Herpestidae 



Galidictis fasciata (Gmelin, 1788) 



Subfamily Galidiinae 

Galidia elegans I. Geoffroy Saint-Hilaire, 1873 

This diurnal animal was the most frequently en- 
countered carnivore on the eastern side of the RNI 
d' Andringitra. At least four individuals frequented 
our 810 m camp, where they were remarkably 
tame, allowing approach to within a few meters. 
They often raided the camp for any food left out 
in the open. At 720 m this species was distinctly 
more retiring and apparently less common. The 
inhabitants of Ambarongy mentioned their belief 
that local carnivores, including G. elegans, are a 
significant predator on chickens. Local people set 
traps at the ecotone between the forest and agri- 
cultural fields to trap and kill carnivores. The dif- 
ference in the commonness of this species be- 
tween the 720 and 810 m zones may be a result 
of human persecution near settlements. It was dis- 
tinctly less common at 1210 m elevation and not 
recorded in the 1625 m zone or above. 

Feces of Galidia elegans collected along a trail 
at 810 m contained beetle keratin and frog bones. 
At 720 m elevation a G. elegans fled upon my 
approach from the vicinity of a mist-net placed in 
the forest, where it had killed captured birds: two 
Phyllastrephus zosterops and two Terpsiphone 
mutata. One of the Terpsiphone was still entire; 
it had been dispatched by a bite to the neck and 
the vertebral column was broken. On two occa- 
sions single G. elegans were observed climbing 
gracefully on vines and limbs 5-10 m above the 
ground. Such arboreal behavior was previously 
reported by Rand (1935). 

On October 8, 1993 at 810 m a pair of G. ele- 
gans was observed copulating. Uninterrupted co- 
itus lasted about 15 minutes. Two individuals col- 
lected in this elevational zone included a male 
with partially descended testes, measuring 17 X 
12 mm, and a slightly convoluted epididymis; and 
a female with one set of large inguinal mammae. 
The pelage coloration of these two individuals 
falls within the range of G. e. elegans, which had 
been previously reported from the reserve (Albig- 
nac, 1973) 

Measurements (6\ 9) — Total length: 673, 638 
mm; tail: 291, 287 mm; hind foot (excluding 
claw): 72, 70 mm; ear: 29, 29 mm; and weight: 
965, 715 g. 



This nocturnal animal was observed and 
trapped only in the 720 m transect zone. At about 
2230 hours, a G fasciata was observed patrolling 
the area around the camp. The following morning 
a rice sack that contained provisions was found; 
it had been torn open and a quantity of food re- 
moved. An adult female was captured in a Na- 
tional live trap set on the ground at the base of a 
large exposed rock in slightly disturbed forest 
(Fig. 24-1). The subspecies G. f. striata has been 
reported from just east of the reserve (Albignac, 
1973: fig. 5). 

Measurement — Weight: 520 g ($). 



Family Viverridae 

Subfamily Cryptoproctinae 

Cryptoprocta ferox Bennett, 1833 

Inhabitants of the village of Ambarongy, near 
our 720 m camp, reported that C. ferox is a seri- 
ous menace to domestic animals, particularly 
chickens. None of the expedition participants ob- 
served this species in the RNI d' Andringitra, but 
within the 810, 1210, and 1625 m transect zones, 
its tracks were found in soft ground and mud. In 
the open heathland above 2000 m, feces of this 
carnivore were collected; they contained the re- 
mains of rodents Brachyuromys betsileoensis and 
Rattus rattus, frogs, and large insects. Albignac 
(1973) noted that C. ferox occurs at 2000 m on 
the Andringitra Massif. 



Subfamily Euplerinae 

Eupleres goudotii Doyere, 1835 

This species was observed once during our field 
trips to the area. At dusk a single adult was ob- 
served at 810 m moving through an area of rela- 
tively thick bamboo just above the Sahanivoraky 
River. This is the first report of this species in the 
reserve. 

Fossa fossana (Mtiller, 1776) 

This nocturnal carnivore was recorded in the 
720 and 810 m transect zones. In both cases it 



290 



FIELDIANA: ZOOLOGY 




Fig. 24-1. Female Galidictis fasciata captured and released in the 720 m zone. 



was observed at the edge of the camp or along a 
trail. Albignac (1973: fig. 5) reported F. fossana 
south of the RNI d'Andringitra in the Pic d'lvo- 
hibe area. 



Subfamily Viverrinae 

Viverricula indica (Desmarest, 1804) 

This introduced species was relatively common 
at lower elevations, particularly at 720 m. The in- 
habitants of Ambarongy reported that this animal 
also attacks chickens and that, with some regular- 
ity, it is captured in carnivore traps set around the 
village and nearby tavy clearings. In a house with- 
in the reserve and at the edge of a tavy, skins of 
two Viverricula, presumably trapped by local peo- 
ple, were found. Viverricula indica was observed 
numerous times soon after dusk, moving around 
the perimeter of our camps or on nearby trails. It 
was previously reported in the reserve (Nicoll & 
Langrand, 1989). 



Summary 

The eastern slopes of the RNI d'Andringitra 
hold a total of five native and three introduced 
carnivores (Table 24-1). Two of the introduced 
carnivores, Felis and Canis, are associated with 
human habitation, whereas the third introduced 
species, Viverricula indica, is not a domestic an- 
imal and lives away from human settlements. This 
carnivore diversity is one of the highest known 
for any rain forest area on the island (Nicoll & 
Langrand, 1989). The Reserve de Biosphere de 
Mananara, 225 km north of Toamasina, and be- 
tween sea level and 570 m in elevation, holds the 
same suite of native carnivores as the Andringitra 
area, plus Salanoia concolor. The Masoala Pen- 
ninsula, in the extreme northeastern portion of the 
island, has a number of carnivore species equal to 
that of the RNI d'Andringitra; the only difference 
between the two sites is that the former lacks Gal- 
idictis fasciata, and includes Salanoia concolor. 
The Pare National de Ranomafana, about 1 20 km 
NE of our 720 m camp, holds basically the same 
carnivore assemblage as the RNI d'Andringitra. 



GOODMAN: CARNIVORES 



291 



Table 24- 1 . Altitudinal distribution of carnivores on 
the eastern slopes of the RNI d'Andringitra. 





720 


810 


1210 


1625 




m 


m 


in 


m 


Family Canidae 










Canis lupus* 


+ 








Family Felidae 










Felis silvestris* 


+ 


+ 






Family Herpestidae 










Galidia elegans 


+ 


+ 


+ 




Galidictis fasciata 


+ 








Family Viverridae 










Cryptoprocta ferox 




+ 


+ 


+ 


Eupleres goudotii 




+ 






Fossa fossana 


+ 


+ 






Viverricula indica* 


+ 


+ 






Total number of 










native species 


3 


4 


2 


1 



* Species introduced on Madagascar. 



The most reasonable explanation for the rela- 
tively large carnivore lists is related to the amount 
of time researchers have spent exploring the sites. 
Perhaps with the exception of Eupleres, all of the 
carnivores found in the RNI d'Andringitra are rel- 
atively tolerant of human disturbance. 



Literature Cited 

Albignac, R. 1973. Faune de Madagascar. 36. Mam- 
miferes: Carnivores. Office de la Recherche Scienti- 
fique et Technique Outre-Mer et Centre National de 
la Recherche Scientifique, Paris, 210 pp.+plates. 

Nicoll, M. E., and O. Langrand. 1989. Madagascar: 
Revue de la Conservation et des Aires Protegees. 
World Wide Fund for Nature, Gland, xvii+374 pp. 

Rand, A. L. 1935. On the habits of some Madagascar 
mammals. Journal of Mammalogy, 16: 89-104. 

Wozencraft, W C. 1993. Order Carnivora, pp. 279- 
348. In Wilson, D. E., and Reeder, D. M., eds., Mam- 
mal Species of the World: A Taxonomic and Geo- 
graphic Reference, 2nd edition. Smithsonian Institu- 
tion Press, Washington, D.C., 1,206 pp. 



292 



FIELDIANA: ZOOLOGY 



Chapter 25 

Rapid Assessment of the Primate Fauna of the 
Eastern Slopes of the Reserve Natureile 
Integrate d'Andringitra, Madagascar 

Eleanor J. Sterling and Marie Gisele Ramaroson 



Abstract 

The ongoing deforestation of Madagascar's eastern forests threatens a rich array of primate 
species. Accurate assessments of the remaining species' geographical distribution, population 
sizes, and population dynamics directly influence conservation management decisions such as 
location and size of potential reserves as well as management of existing reserves. Intensive, 
long-term survey work to decipher the patterns of species distribution and of population abun- 
dance is ideal but currently impracticable on a large scale. Resource limitations and time 
constraints dictate that conservation managers use quicker methods, such as rapid biodiversity 
assessment techniques, to approximate these patterns over the short term. This paper discusses 
the results of a 6-week rapid survey of the primate fauna in the Reserve Natureile Integrate 
d'Andringitra. 

Resume 

Le d6boisement auquel les forets de Test de Madagascar sont confrontees constitue une 
menace pour une grande varied de primates. Des inventaires precis sur la distribution g6o- 
graphique, la tai lie de la population et la dynamique de la population des especes influencent 
directement les decisions de gestion de la biodiversity, comme par exemple pour ce qui con- 
cerne la localisation et la taille des reserves futures ainsi que la gestion des reserves existantes. 
Un inventaire intensif et a long terme destine" a permettre 1' interpretation des caract^ristiques 
de la distribution des especes et 1'abondance des populations constituerait l'outil id6al, mais 
cela n'est pas actuellement r6alisable sur une grande 6chelle. La limitation des ressources 
humaines et financieres, ainsi que les contraintes de temps forcent les responsables de la con- 
servation a utiliser des techniques moins lourdes, telles que les techniques d' inventaire rapide 
de la biodiversity, afin d'obtenir a court terme des informations. Ce document pr6sente les 
r£sultats d'un inventaire rapide de la faune de primates de la Reserve Natureile Integrate 
d'Andringitra. 



Introduction slopes (Perrier de la Bathie, 1927; Paulian et al., 

1971). These zones support markedly different 

The Reserve Natureile Integrate (RNI) d'An- faunal complements, according to the few existing 

dringitra encompasses a variety of ecological survey reports (Perrier de la Bathie, 1927; Nicoll 

zones due to the major vegetational and climatic & Langrand, 1989). For instance, the list of pri- 

differences between eastern and western-facing mates found in the reserve includes those widely 

STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 293 



Table 25-1. The 15 primate species and subspecies of the RNI d'Andringitra. Data include lemurs previously 
reported from the reserve's eastern and western slopes (Perrier de la Bathie, 1927; Nicoll & Langrand, 1989) and 
those observed during the 1993 inventory on the eastern slopes. 



Taxon 



Eastern 

and western 

slopes 



Eastern slopes — 1993 inventory 



720 m 



810 m 



1210 m 



1625 m 



Lemur catta 
Avahi laniger 

Varecia variegata variegata 
Eulemur fulvus fulvus 
Microcebus rufus 
Eulemur rubriventer 
Lepilemur mustelinus 
Eulemur fulvus albocollaris 
Eulemur fulvus rufus 
Hapalemur aureus 
Hapalemur griseus griseus 
Hapalemur simus 
Propithecus diadema edwarsi 
Cheirogaleus major 
Daubentonia madagascariensis 
Total numbers of species 
and subspecies 



distributed in eastern rain forests, such as Varecia 
variegata variegata, as well as those found in dri- 
er forests of the central plateau and the southwest, 
such as Lemur catta (Table 25-1). Yet, despite its 
breadth, we believed the list to be incomplete. 
Many species, such as Hapalemur griseus, Pro- 
pithecus diadema, and Cheirogaleus major, were 
not listed, although they are found in nearby rain 
forests to the north, (Pare National [PN] de Ran- 
omafana), or to the south (Reserve Speciale [RS] 
du Pic d'lvohibe), or to the southeast (RS de Man- 
ombo) (Table 25-2). Nor were the rare forms, 
such as Hapalemur aureus, listed in the existing 
surveys. 

It is not known to what extent altitude or forest 
type affects the distribution and density of primate 
populations, although several researchers have be- 
gun to address these issues (Skorupa, 1986, 1987, 
1988; Johns and Skorupa, 1987; O'Connor et al., 
1987; Oates et al., 1990; Thomas, 1991; Ganz- 
horn, 1994b). In this study, we attempted to com- 
pare species composition and densities between 
forests at different elevations and between dis- 
turbed and undisturbed forests at approximately 
the same elevation. We were working as part of a 
team of scientists looking at these issues across 
taxa in the RNI d'Andringitra. Rapid survey tech- 
niques for primate populations, however, are in 
their infancy. We were not sure that they would 
be robust enough to use for reliable estimates of 
density and diversity in primates. A variety of 



techniques have been employed to estimate pri- 
mate densities over longer periods of time (see 
National Research Council, 1981). The line tran- 
sect method is the most frequently used (Edwards, 
1992), but we have yet to determine the number 
of transects necessary for accurate density esti- 
mates in a region. 

Results of this current study were therefore 
used to clarify information on primate distribu- 
tional boundaries, to provide a basis for evaluat- 
ing potential ecological correlates in species dis- 
tributions, to evaluate rapid assessment techniques 
for primate populations, and to provide recom- 
mendations for future short-term survey work. 



Methods 

Transects were conducted at four elevations 
(720, 810, 1210, and 1625 m) between September 
23 and October 31, 1993. Two of these zones 
were in some of the lowest remaining forest on 
the eastern slopes of RNI d'Andringitra, one in a 
partially degraded forest (720 m) and the other in 
a relatively intact forest (810 m) located about 2.5 
km (by direct line) to the east. The other two tran- 
sect zones were in undisturbed forest. 

The line transect method (National Research 
Council, 1981) was used to estimate density at 
each site. At the outset, we laid out linear tran- 



294 



FIELDIANA: ZOOLOGY 



Table 25-2. List of primate species observed in re- 
serves within a 100 km radius of RNI d'Andringitra. 

Site 

RS 
RS Pic PN 

Manombo d'lvohibe Ranomafana 
(altitudinal (aititudinal (altitudinal 
range range 775- range 
Species 1-137 mi 2060 m) 800-1200 m) 



Efa 


X 


Eff 


... 


Efr 


... 


Er 


... 


Ha 


... 


Hgg 


X 


Hs 


... 


Lc 


... 


Mr 


X 


Pde 


... 


Vvv 


... 


Al 


X 


Cm 


X 


Dm 


X 


Lm 


... 


Total 


6 



X 
X 
X 
X 

X 
X 
X 
X 
X 
X 
X 

12 



Efa = Eulemur fulvus albocollaris; Eff = Eulemur 
fulvus fulvus; Efr = Eulemur fulvus rufus; Er = Eulemur 
rubriventer. Ha = Hapalemur aureus; Hgg = Hapale- 
mur griseus griseus; Hs = Hapalemur simus; Lc = Le- 
mur catta; Mr = Microcebus rufus; Pde = Propithecus 
diadema edwarsi; Al = Avahi laniger; Cm = Cheiro- 
galeus major. Dm = Daubentonia madagascariensis; 
Lm = Lepilemur mustelinus. Data from: Buettner-Jan- 
usch and Tattersall (1985), Jenkins (1987), Nicoll and 
Langrand (1989), Mittermeier et al. (1992), and Good- 
man and Schulenberg (unpubl.). 



sects along compass bearings regardless of the 
terrain, but this was too time consuming and fur- 
thermore the transects often traversed impassable 
terrain. We therefore followed existing trails left 
by bush pigs, cows, and humans, selecting trails 
to cover equal areas of ridge, slope, and valley 
habitats. Trails were marked every 10 m with flag- 
ging for reference during the surveys. Two re- 
searchers experienced in observing Malagasy pri- 
mates walked either singly or in tandem along 
transects of varying lengths (600-3000 m) for ap- 
proximately 5 hours. Diurnal samples started at 
0530, whereas nocturnal surveys started at 1800, 
2100, or 0100 hours. We walked approximately 1 
km per hour during the day, but we had to slow 
to 750 m per hour at night due to the difficult 
terrain. Diurnal and nocturnal animals were de- 
tected by sound, sight, or movement of branches. 
In addition to being able to distinguish animals at 
night by their shape/color differences from the 



background, one can also locate them by the re- 
flection of their tapetum lucidum. 

We paused every 100 m or so to watch and 
listen for primates, but we never deviated from 
the path in search of them. Each transect was re- 
peated a minimum of six times, three times each 
for diurnal and nocturnal transects. Whenever 
possible during sequential samples along a tran- 
sect, we tried to walk in the opposite direction 
from the previous sample to reduce potential bi- 
ases (Brockelman & Ali, 1987). Similarly, repeat 
samples were spaced 48 hours apart when feasi- 
ble. Due to time constraints, we were forced to 
undertake sampling under conditions not normally 
ideal for census work (light rain or fog). We did 
not conduct samples when our viewing distance 
was restricted to less than 15 m. 

For each animal or group seen, we noted spe- 
cies, mode of detection (sight, vocalization, or 
sound of movements), number and composition of 
group, time of contact, position on transect, height 
from ground, habitat type, behavior, distance from 
observer to first animal seen, and perpendicular 
distance from trail to first animal seen. Diurnal 
animals were defined as members of a "group" 
if they were observed within 50 m of a conspe- 
cific animal. Because nocturnal animals are cryp- 
tic, and therefore more difficult to locate than di- 
urnal species, we always counted them as individ- 
uals. We spent no more than 10 minutes with any 
single cluster of animals trying to determine spe- 
cies, group composition, and pelage characters. 

A number of different models exist for calcu- 
lating density from raw data with the transect 
method (Charles-Dominique, 1977; Burnham et 
al., 1980; Whitesides et al., 1988), all of which 
have their drawbacks (see Whitesides et al., 
1988). We used three density estimate methods: 
two absolute and one relative. The two absolute 
density models were used in this study because 
they had been employed previously by primate 
researchers in Madagascar (Ausilio & Raveloari- 
noro, 1993; Ganzhorn, 1994a). They involve de- 
riving a "detection function" from the distribu- 
tion of observed perpendicular trail-to-animal dis- 
tances (calculated for each species separately) that 
is used to estimate the area censused. 



Absolute Density Estimate 
Method I (AMI) 

First, we derived transect width by grouping 
detection distances for each species into eight fre- 



STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 



295 



quency classes between and 40 m from the trail 
and selecting the distance beyond which obser- 
vations dropped by at least half (Charles-Domi- 
nique, 1971; Charles-Dominique & Bearder, 
1979; Ausilio & Raveloarinoro, 1993). We mul- 
tiplied twice this transect width by the average 
length of a transect to derive total area. The av- 
erage number of groups (for diurnal species) or 
individuals (for nocturnal species) observed with- 
in the derived transect width was divided by total 
area to calculate density. In principle, as distance 
from the trail increases, frequency of detection 
should level off, then drop (Cant, 1978). However, 
this depends upon the acquisition of an adequate 
sample of detection distances; it might not be pos- 
sible to collect such a sample in rapid assessment 
surveys. 



these assumptions are violated when working with 
unhabituated, cryptic primate populations, re- 
searchers have proposed another index of animal 
density that is less accurate, but also less exacting 
(see Cant, 1978). This index, called relative den- 
sity, is calculated as the number of animals ob- 
served per unit time or distance, and it is used to 
compare populations over time or in different ar- 
eas. In this study, we chose to determine relative 
density as animals per km surveyed. 



Results 



General 



Absolute Density Estimate 
Method II (Willi 

Next, following the methodology of Ganzhorn 
(1992, 1994a), we took twice the average of the 
trail-to-animal distances for each species observed 
on three or more transects and multiplied it by the 
distance covered to give total surface area sur- 
veyed per transect. Density estimates were then 
made by multiplying group size estimates by the 
number of groups sighted on a transect and divid- 
ing by the total area for each transect. All obser- 
vations made on the transect were included, not 
just those falling within the averaged transect 
width. 

It is difficult to estimate distances and number 
of animals by vocalizations, and there is a greater 
possibility that animals could be counted twice 
(through counting both vocal and visual contact). 
One assumption underlying the transect method is 
that animals are not counted twice (Burnham et 
al., 1980; Whitesides et al., 1988; Buckland et al., 
1993). Therefore, we did not include in our cal- 
culations (for either diurnal or nocturnal surveys) 
animals that were heard but not seen. 



Relative Density Method 

The transect method includes other critical as- 
sumptions: that animals located directly over the 
transect are always detected; that they are detected 
at their original location, prior to movement in 
response to the observer; and that distances are 
accurately recorded. Due to the fact that some of 



In total, seven diurnal and five nocturnal spe- 
cies were observed in the four transect zones (Ta- 
ble 25-1). Only one diurnal (Eulemur rubriventer) 
and two nocturnal species (Cheirogaleus major 
and Microcebus rufus) were seen in all four 
zones. Overall species and subspecies richness for 
the RNI d'Andringitra was the highest yet record- 
ed in any Malagasy forest, if one includes species 
from the eastern and western slopes (Tables 25-1 
and 25-2). 



Detection Distances and Density 

Species-specific detection distances were cal- 
culated by elevation rather than transect because 
there were far too few observations per transect 
(Table 25-3). Even so, they did not reach a plateau 
and drop off as expected. Nevertheless, we cal- 
culated density on the few species for which we 
could reasonably estimate transect width (Table 
25-4). 

In comparing density estimates derived from 
AMI with those from AMII, one can see that the 
two methods produce similar results for the ele- 
vational zones, although the relative trends be- 
tween sites differ (Table 25-4). However, AMI 
and AMII are correlated significantly only among 
estimates for nocturnal species (r = 0.83, P = 
0.02, N = 7). No other correlation between the 
different estimates approaches significance, nei- 
ther for the pooled data set nor when species are 
subdivided into group-living versus solitary spe- 
cies. 



296 



FIELDIANA: ZOOLOGY 



Table 25-3. Frequency of detection by species for each elevation, by distance from the trail. 

Distance class (m) 



Transects 



1-5 



6-10 



11-15 



16-20 



21-25 



26-30 



31-35 



36-^0 



720 m zone 




Species 
Efa 


2 


Er 




H? 




Al 


4 


Cm 


5* 


Lm 


1 


Mr 


9 


810 mi zone 




Species 
Efa 


1 


Er 


1 


Ha 


1 


Hgg 
Hs 


2 

1 


H? 




Cm 


13* 


Mr 


14 


1210 m zone 




Species 
Efa 


3 


Efr 




Er 




Ha 


1 


Hgg 
H? 


1 


Al 


1 


Cm 


8 


Lm 




Mr 


12* 


1625 m zone 




Species 
Er 




Ha 


1 


Hgg 
Al 


2 

1 


Cm 


3 


Lm 


1 


Mr 


20* 



1 
1 

2* 

2 

2* 

1 



3* 

1 
2* 

1 



Note: Data for diurnal species represent number of groups observed; for nocturnal species they represent the number 
of individuals. Species abbreviations are as in Table 25-2. H? = unidentified Hapalemur. 
* Detection distance used to calculate transect widths in Absolute Method I. 



Effects of Habitat Disturbance 

The total number of species observed at the 720 
m disturbed and the 810 m undisturbed sites re- 
mained relatively constant, although the composi- 
tion of observed species changed (Table 25-1). 
Group size varied little between these sites for the 
two species that we observed in both areas (Table 
25-5). The average density of Eulemur fulvus al- 
bocollaris was lower in disturbed than undisturbed 



habitat using AMI, but AMD estimates showed little 
difference for this species. AMI and AMD density 
estimates showed opposite trends for Cheirogaleus 
major, there was lower density in the disturbed hab- 
itat according to AMI and lower density in the un- 
disturbed habitat according to AMII. The density of 
Microcebus rufus was higher in the undisturbed 
habitat according to AMD, but it remained relatively 
similar using AMI. Relative density was higher in 
the undisturbed that the disturbed sites for all species 



STERLING & RAMAROSON: RAPID ASSESSMENT OF PRIMATE FAUNA 



297 



Table 25-4. Average density estimates and standard deviations for primate fauna in the RNI d' Andringitra during 
September and October 1993, calculated for transects following