The status and ecology of the Angonoka tortoise (Geochelone yniphora) of Western Madagascar


Material Information

The status and ecology of the Angonoka tortoise (Geochelone yniphora) of Western Madagascar
Physical Description:
vii, 180 leaves : ill. ; 29 cm.
Smith, Lora L., 1960-
Publication Date:


Subjects / Keywords:
Turtles -- Habitat -- Madagascar   ( lcsh )
Wildlife conservation -- Madagascar   ( lcsh )
Wildlife Ecology and Conservation thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Wildlife Ecology and Conservation -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1999.
Includes bibliographical references (leaves 160-179).
Statement of Responsibility:
by Lora L. Smith.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 021577657
oclc - 43707755
System ID:

This item is only available as the following downloads:

Full Text








Funding for this study was provided by the Jersey Wildlife Preservation Trust,

and I thank Lee Durrell, the Honorary Director of JWPT, for her support of the project. I

also would like to thank the Malagasy Government, particularly the Direction des Eaux et

For6ts for permission to conduct this study. I particularly thank the members of my

academic committee, C. Kenneth Dodd, Jr. (chair), Dick Franz, Mel Sunquist, George

Tanner, and Elliott Jacobson. I am honored to have had the opportunity to know and

work with each of them.

I am particularly grateful to the three Malagasy students Bourou Robert, Mahatoly

Joby, and Sibo Clement who worked with me in the field. They taught me Malagasy

survival skills including how to cook rice on an open fire and how to cut transects

through bamboo thickets. Don Reid, the Director of the captive breeding program for the

angonoka, helped me get the field project started and made me laugh when life in the

bush got difficult or uncomfortable. I thank the rest of the Project Angonoka team,

including Joanna Durbin, Lala Jean Rakotoniana, Hasina Randriamanampisoa, and

Daurette Razandrizanakanirina, who provided invaluable assistance with project logistics.

I am grateful to Frank Hawkins and Mike and Liz Howe who participated in the 1994-95

field expedition to western Baly Bay. I am indebted to the guides Adany Lolo, Martin,

Mossimo, Fiankina, Norbert, Jakoba, and Koera-be and the people of Soalala, Antsira,

Marotia, Antranolava, Antanandava, Maroboaly, and Maroaleo who graciously provided

food, lodging, and friendship during our visits. Djamaldine Said Aly and family, Ali

Hassanaly and family, Njaribe, Felix and Njarikely were particularly generous in opening

their homes to me.

I am grateful to my friends at Water and Air Research, Inc. (W&AR), who

provided support and employment on and off throughout my graduate career. Charlie

Fellows and others at W&AR donated a solar battery charger that was used in the field.

Pat and Ray Ashton donated a camera for use in the field and kindly provided a place for

me to stay when I returned to Gainesville. I also thank Jay Harrison and Jack Bishko of

the Institute of Food and Agricultural Sciences at the University of Florida for their

statistical consultations. Kenney Krysko, Margaret Cheaney, A. Ross Keister, and S.

Mark Meyers assisted in creating the distribution maps.

Finally, I thank my family for their support and encouragement. My sister Jackie

and brother Rick kept my home life together while I was out of the country. My brother

Rob sent much-appreciated photographs of my nephew and niece while I was in the field,

and also wrote a program that helped me analyze the home range data. My grandmother,

Eleanor Smith, wrote letters faithfully while I was in the field, although she firmly

believed that I inherited my wanderlust and strange affinity for turtles from the other side

of the family. I thank my very good friends Chan Clarkson, Lynn Mosura-Bliss, Charlie

Fellows, Elizabeth Knizley, and Linda LaClaire. I also thank the students in the

Department of Wildlife Ecology and Conservation for the many discussions on

conservation issues and for the good times spent in search of reptiles and amphibians.


A C K N O W LED G M EN T S ...................................... ............................ ........ ............. ii

A B ST R A C T .................................................................................. ................. ..... vi


1 IN TRODUCTION ......................... ............................. ....... ... 1

Madagascar- An Island Continent........................................ ................ 1
Geochelone yniphora- The Angonoka Tortoise............................... ... 3
Project A ngonoka................................. ............ .......................... 6
D description of Study A rea ........................................ ........................ 7

2 STATUS AND DISTRIBUTION... ........ ......................... ....... 12
Introduction .............. ... .................................................................. 12
M methods ................................. ............. .... .... ............. .. 14
Results ............................................................... 16
D discussion ................................... ............ ....................................... 25

3 HOME RANGE AND MICROHABITAT USE ................. ................... 38

Introduction .............. ................................. 38
M methods .... ..................................................... ............. ......... 43
R results ...... ........... .............. ......................... ... ....... 45
D discussion ...................................... .................. ....................... 48

4 BEHAVIOR AND ACTIVITY PATTERNS ................ ...... ............... 63

Introduction..................................................................... ......... .. 63
M eth o d s ....................................... .......... ... ...................... 6 4
Results ........................................ 66
D iscu ssio n ...................................................................... 7 2


5 M O R P H O L O G Y ......................................................................................... 93

Introduction.......... ................ .......... ................ .............. 93
M methods .................................... ............... ...................... 97
R esu lts ........................................................ 99
D iscu ssion ................................. .. .... .. ....... .... ....... ...... 103

RESEAR CH N EED S ................................................... ........... ............. ... 119

A PPE N D IC E S ............ ...... .............. ...... .. ..................................... 129

K ERN A L A REA ............ ..................... ............................. ...... 129

REGION OF WESTERN MADAGASCAR. ................................. 155

R E FE R E N C E S ..................................................... ................... 160

BIOGRAPHICAL SKETCH .................................................. .............. 180

Abstract of Dissertation Presented to the Graduate School
Of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Lora L. Smith

December 1999

Chairman: C. Kenneth Dodd, Jr.
Major Department: Wildlife Ecology and Conservation

An investigation of the status and ecology of the rare angonoka tortoise

(Geochelone yniphora) was conducted in western Madagascar from October 1993

through June 1995. Ten tortoise localities were identified based on regional field

surveys, and 145 tortoises were observed. The 10 localities represent at least five

separate populations, all located within a 30-km radius of Baly Bay (lat. 160 2'S, long.

450 20'E). Juvenile tortoises were observed in all populations indicating that

reproduction is occurring throughout the range. The five populations occur on patches of

bamboo-scrub habitat that range in size from <50 ha to 4,000-6,000 ha. The habitat

patches are isolated from one another by extensive savanna and Baly Bay. No collection

of tortoises was observed in this study and exploitation probably is not a problem at this

time. However, the bamboo-scrub habitat at all areas visited appeared to have been

degraded by frequent brush fires.

Ninety-nine angonoka (14 adult males, 27 adult females, and 58 juveniles) were

marked on Cape Sada where monthly surveys were conducted. The population density

(0.66 tortoises per ha) was low compared to other tortoise species. Nearly 40 % of

tortoises on Cape Sada were juveniles < 85 mm carapace length. A lack of intermediate

sized tortoises and the observation of dead juveniles suggests that juvenile mortality in

this population is high.

Tortoises were most active (e.g., walking, mating, nesting, and feeding) in the wet

season (November April) when food and water were abundant, whereas in dry season

(May October) tortoises usually were resting under vegetative cover. Patches of grass

offered important cover sites for angonoka in the dry season, whereas open rocky areas

were important feeding sites during the wet season.

The angonoka is threatened with extinction in the wild, primarily because of its

small population size and extremely limited geographic distribution. A catastrophic

event such as disease or severe weather could cause extinction of some or all populations.

Removal of adult tortoises from remaining populations also could have a severe impact

on the survival of this species. Long-term survivorship data are needed to determine

whether populations are increasing. However, recovery of the species will be limited by

the amount of available habitat. Therefore, conservation efforts should focus not only on

physically protecting angonoka, but also on protecting and restoring the bamboo-scrub



Madagascar- An Island Continent

The island of Madagascar has been called a microcontinent because of its large

size, long history of isolation, and diverse environment (Paulian 1984). It is the fourth

largest island in the world, encompassing an area nearly 600,000 km2 in size (Battistini

1972). The island rifted from Africa, near present day Somalia, in the mid-to-late Jurassic

(ca. 165 million years ago). It achieved its current position, more than 400 km east of

Mozambique, in the early Cretaceous (ca. 121 mya) (Rabinowitz et al. 1983; Coffin and

Rabinowitz 1987; Reeves et al. 1987; Krause et al. 1997). Madagascar and the Indian

subcontinent separated in the late Cretaceous, ca. 88 mya (Storey et al. 1995). Most of

the extant fauna has clear affinities with Africa, and the founding stock of these species are

thought to have rafted across the Mozambique Channel after the island attained its current

position (Jolly et al. 1984). Other species (e.g., the mollusks, some birds, frogs, and bats)

have Indo-Malayan affinities (Benson 1984; Paulian 1984; Langrand 1990) and a few

species (e.g., iguanids and boids) are most closely allied with groups in South America

(Paulian 1984). More than 90 % of Madagascar's forest species are unique to the island

(Jolly et al. 1984). Of the more than 300 species of reptiles in Madagascar, 95 % are

endemic to the island or to the region (Blanc 1972; Blanc 1984; Raxworthy and Nussbaum

1997). In addition to a high degree of endemism, Madagascar contains a major portion of

the world's biological diversity (Myers 1986; Mittermeier 1988).

Humans arrived on Madagascar approximately 1,500-2,500 years ago (Battistini

and Verin 1972; MacPhee and Burney 1991). The Malagasy people share an African and

Indonesian origin, and at least 20 regional ethnic groups occur on the island (Rakotoarisoa

1986). Today more than 14 million people inhabit Madagascar and the population growth

rate is somewhere between 2-3 % annually (Bos et al. 1994; United Nations 1997). Since

the island was settled, at least 23 species of animal have become extinct, including the

island's megafauna. Two giant tortoises (Dipsochelys abrupta and D. grandidieri),

flightless elephantbirds (Aepyornis spp.), and giant lemurs (Megaladapsis spp.) all have

been lost (Mah6 1972; Burney and MacPhee 1988; Burney 1993; Bour 1994). The issue

of whether or not the recent extinctions were caused by human or climatic changes has

been debated heavily (Burney 1993; Dewar 1997; MacPhee and Marx 1997). In addition

to the recent extinctions, much of the island's original forest has been replaced with

manmade grasslands. Early estimates were that humans had cleared as much 80 % of

Madagascar's forest cover (Jolly 1980; Jenkins 1987). However, grasslands in western

Madagascar may actually pre-date the arrival of humans on the island (MacPhee et al.

1985; Burney and MacPhee 1988; Burney 1997). Nonetheless, today the forests and

other natural communities of Madagascar are declining and much of the remaining wildlife

is critically endangered (Jenkins 1987).

Geochelone yniphora- The Angonoka Tortoise

The reptile fauna of Madagascar includes four endemic tortoises, Geochelone

radiata, G. yniphora, Pyxis arachnoides, and P. planicauda. Humans introduced a fifth

species, Kinxys belliana, to the island from mainland Africa 1,000-1,500 years ago

(Kuchling 1986; Glaw and Vences 1994). None of the endemic species are well known,

and all four species probably are threatened. Geochelone radiata, the radiated tortoise,

and P. arachnoides, the spider tortoise, are the most widely distributed of the four species.

These two species occur in the dry thorn scrub of southern Madagascar. Radiated

tortoises still occur in reasonably high densities in a few areas but are harvested for food,

exported illegally for the pet trade (Durrell et al. 1989a; J. Behler, pers. com.), and killed

on roads (Goodman et al. 1994). Spider tortoises typically are not exploited for food

because of traditional taboos against eating them (Juvik 1975); however, local taboos are

apparently breaking down and this species may become increasingly exploited in the future

(J. Behler, pers. com.). Pyxisplanicauda, the flat-tailed tortoise, occurs only in dry

deciduous lowland forests near Morondava in west central Madagascar (Behler et al.

1993). Deciduous dry forests are extremely fragmented and degraded and only about

15,000 ha of suitable flat-tailed tortoise habitat remains (Kuchling and Bloxam 1988).

Geochelone yniphora, the "angonoka" or ploughshare tortoise, is the largest of

Madagascar's extant tortoises. This species received its English common name for the

distinctive long, plow-like gular projection of adult males. The angonoka is arguably one

of the rarest tortoises in the world (Angel 1931; Juvik et al. 1981; Groombridge 1982;

Curl 1986a; Durrell et al. 1989b). It was declared a protected species by the Malagasy

government in 1931, and is listed as Endangered in the IUCN Red List of Threatened

Animals (IUCN 1996). All historic localities of G. yniphora occur within a very restricted

area in the vicinity of Baly Bay in western Madagascar (Figure 1-1) (Blanc 1974; Juvik et

al. 1981; Curl et al. 1985).

Geochelone yniphora and G. radiata were placed in the subgenus A.stercheliys

based on similarities in osteology and shell morphology (Loveridge and Williams 1957).

Crumly (1982) later described a possible sister relationship between G. yniphora and

Geochelone pardalis, an African species. He also found similarities between Dipsochelys

dussumieri (a species from Aldabra and the Seychelles formerly referred to as Geochelone

gigantea or D. elephantina) and G. radiata. This classification was based only on cranial

osteology and is not widely accepted. A recent examination of mitochondrial DNA of the

endemic Malagasy tortoises revealed that the four species may share a monophyletic origin

(Caccone 1999). This suggests that there was a single colonization event by an ancestral

Testudinid. This ancestral tortoise probably rafted from Africa approximately 22-14 mya.

The Pyxis clade is thought to have diverged from the ancestral form ca. 12-8 mya and the

G. yniphora and G. radiata clade diverged in the Pliocene (ca. 5 mya). Juvik et al. (1981)

speculated that the divergence of G. yniphora and G. radiata occurred during the

Pleistocene (ca. 1.8 mya), and that the ancestral tortoise persisted in southern and western

refugia during mesic conditions.

The present distribution of the angonoka was not discovered until the early 20"

Century. This species was first described from specimens obtained from Arab sailors in

the Comoro Islands in 1884 (Vaillant 1889). The specimens were thought to have

originated from an island north-northeast of Comoro, in the vicinity of Aldabra. A later

specimen was thought to have come from Tulear in southwest Madagascar (Vaillant

1889). The actual distribution of the species was discovered when Siebenrock (1903)

described a wild specimen collected by Voeltzkow at Cape Sada in northwestern


Little was known of the status of wild populations of G. yniphora until the 1970s

when a series of field surveys in the Baly Bay region were conducted (Juvik and Blanc

1974; Juvik et al. 1981). Very few angonoka were discovered in the early surveys,

suggesting that the species was critically endangered. The reasons for the apparent rarity

of the angonoka are not fully understood. However, past commercial exploitation may

have been a contributing factor. Remains of G. yniphora have been recovered from

archaeological sites on Maore, the easternmost island in the Comoro archipelago,

indicating that this species was exploited for food as early as 900 AD (Allibert 1989).

Historic records also indicate that during the 17th Century, large numbers of G. yniphora

were exported to the Comoro Islands where they were sold as food to traveling ships

(Vaillant and Grandidier 1910). Today, local people generally do not eat angonoka;

however, tortoises are sometimes kept as pets (Curl et al. 1984; Durbin et al. 1996).

Anthropogenic brush fires also may have contributed to the endangered status of

the angonoka (Juvik et al. 1981; Groombridge 1982; Curl et al. 1985; Durrell et al.

1989b). Frequent, high intensity fires occasionally kill angonoka and are thought to

convert the bamboo-scrub habitat of the angonoka to savanna (Curl 1986a). Brush fires

are set to promote growth of grasses for free ranging cattle, to drive cattle from the forest

and to maintain clearings around vegetable gardens (Durbin et al. 1996). The clearings

around gardens are created to keep the non-native African bush pig (Potamochoerus

larvatus) from destroying crops. Bush pigs are widespread in western Madagascar

(Grubb 1993) and are thought to prey on eggs and young of the angonoka (Juvik et al.


Project Angonoka

In 1986, Jersey Wildlife Preservation Trust (JWPT) and the Malagasy Direction

des Eaux et Forets initiated a multifaceted conservation program for the angonoka,

entitled "Project Angonoka". One of the first objectives of Project Angonoka was to

establish a captive-breeding program for the angonoka in Madagascar, with the intent of

augmenting wild populations with captive born offspring (Curl 1986b). The breeding

program was begun with a founder stock of 20 adult tortoises that were collected from the

National Zoo at Antananarivo and from villagers in the Baly Bay region who had kept

tortoises as pets. As of 1998, a total of 180 offspring had been produced at the breeding

center (D. Reid, pers. com.). In 1991, JWPT and the World Wide Fund for Nature

(WWF) implemented a social research project to promote the protection and sustainable

use of resources in the region (Durbin and Ralambo 1994, Durbin et al. 1996). In

addition, from 1989 and 1993 JWPT field teams conducted a series of field surveys to

assess the status of wild angonoka populations (Reid 1989, 1990a, 1990b, 1991a, 1991b,

1993). Survey results supported the previous impression that angonoka were

exceptionally rare.

The limited knowledge of the status and ecology of the angonoka made

development and implementation of a conservation plan for this species difficult.

Therefore, this study of wild angonoka was initiated in 1993. My primary objective was

to visit historic tortoise localities and attempt to identify new localities. Results of these

surveys were used to assess the overall status of the species. The second objective was to

describe aspects of the life history and ecology of the angonoka. In October 1993, a field

research station was established at Cape Sada. The research station was located at the

east edge of the Cape Sada peninsula, less than 0.5 km from tortoise habitat. The station

was used as a base camp from which regional tortoise surveys were conducted in 1994

and 1995. From 1993 through 1995, aspects of the demography, movements, and activity

patterns of tortoises in the Cape Sada population were studied.

Description of Study Area

The study area, hereafter referred to as the Baly Bay region, is located on the west

coast of Madagascar approximately 90-km south of the city of Mahajanga (lat. 16 2' S,

long. 45 20' E). The geology of the region is characterized by recent (Eocene) alluvial

rock (Brenon 1972). Coastal areas consist of Miocene marine sediments overlain with a

ferruginous lateritic crust. The average elevation in the region ranges from 10-30 m with

maximum elevations from 50-80 m above mean sea level.

The climate in the Baly Bay region is tropical, with distinct wet and dry seasons.

Nearly all precipitation occurs in the summer or wet season (November through April)

(Donque 1972). Little or no rainfall occurs in the winter or dry season (May through

November). Total annual rainfall typically is more than 1,000 mm and the highest monthly

total (ca. 400 mm) usually occurs in January (Donque 1972). The mean annual

temperature at the town of Soalala on Baly Bay is 26.6 C, with monthly means ranging

from 260 C in July to 280 C in April. Tropical cyclones, originating in both the Indian

Ocean and Mozambique Channel, frequently cross Madagascar in summer months

(November-March). In western Madagascar these storms most often occur in January and


The original vegetation in the Baly Bay region was Western Domain dry deciduous

forest (Koechlin 1972). The coastal area contains extensive mangroves (Rhizophora

spp.). Today, the dry deciduous forest in the region is heavily fragmented and fire-

maintained savanna is a major component of the area. Angonoka inhabit bamboo-scrub

vegetation, a secondary stage of the dry deciduous forest (Curl et al. 1985). Bamboo-

scrub habitat consists of a mosaic of shrubs, bamboo, savanna grasses, and open,

unvegetated areas. The shrubs are generally less than 2-m in height, and the most

common species are Bauhinia sp. and Terminalia spp. Bamboo (Perrierbambos

madagascariensis) occurs in dense thickets within the habitat. A few shrubs and satrana

palms (Bismarkia nobilis) are interspersed in the bamboo thickets, but ground cover

vegetation is largely absent. Small savanna-like patches and unvegetated rocky areas

occur within the scrub-shrub habitat. Grasses in the savanna-like patches include Aristida

sp., Eragrostis sp., and Heteropogon contortus.

The largest town in the region is Soalala, located on the southeastern side of Baly

Bay (Figure 1-1). Soalala has about 1,000 inhabitants, most of whom are Muslim. People

in surrounding coastal villages belong to the Sakalava ethnic group and are primarily small

scale fishermen. Agriculture in the coastal region is limited to subsistence gardening.

Historically, the Sakalava people raised large numbers of cattle (Vrin 1986). Fewer

cattle are kept today than in the past; however, they are still important culturally. Cattle

are used to trample fields to break up soil prior to planting rice, for religious rituals, and

are traded for brides (Durbin 1993). Cattle typically range free in the savannas

surrounding villages but may be hidden from cattle thieves in forest patches (Durbin


Cape Sada, where the research station was established, is a 150-ha peninsula

located on the east side of Baly Bay. The maximum elevation is 50 m above mean sea

level. There are no human settlements on the Cape itself, but the beaches and coastline

are used seasonally by local fishermen. Cape Sada is covered in bamboo-scrub habitat,

except for the northern part, which contains a dense komanga (Erythrophleum couminga)

forest. The region east of the Cape contains extensive savanna and salt flats. Additional


information on the physiography, climate, and vegetation of the study area is presented in

Chapters 2 and 3.

Figure 1-1. Historic localities of the angonoka tortoise (Geochelone yniphora) in the
Baly Bay region of western Madagascar.



The earliest records of wild angonoka suggested that the species was exceedingly

rare (Angel 1931; Siebenrock 1903). However, the first attempts to assess the status of

wild angonoka populations did not take place until the 1970s. Juvik et al. (1981) visited

the Baly Bay region intermittently over a 5-year period and were only able to locate five

tortoises. Four tortoises were found at Cape Sada and one at Ankoro, west of Baly Bay.

They also found fresh tortoise droppings at Beheta and Ambatojoby. Based on the

amount of search time invested (375 person-hours), Juvik et al. (1981) estimated that the

density ofG. yniphora in the wild was probably less than five tortoises per km2. They

further estimated that less than 100-km2 of habitat remained, and that the total population

size was likely no more than a few hundred tortoises. In 1983, Curl et al. (1985)

conducted 6 weeks of field surveys in the region. Tortoise localities described by Juvik

et al. (1981) were re-visited, and verbal interviews with local people were conducted in

an attempt to confirm the location of tortoise populations. Although they were not

specific as to the number of tortoises observed in their surveys, they cautiously estimated

that as few as 100-400 angonoka existed in the wild and that only 40 to 80-km2 of

bamboo-scrub habitat remained (Curl et al. 1985).

Later surveys were focused at Cape Sada, the most accessible tortoise population;

however, surveys also were conducted at Beheta and at localities west of Baly Bay (Reid

1991a, 1993; Juvik et al. 1997; Hawkins and de Valois, 1993). Results of these surveys

indicated that the Cape Sada population probably contained less than 30 individuals. A

single tortoise was found at Beheta and another at Antsahavaky in the vicinity of Ankoro.

Several tortoises at Cape Sada were radio-instrumented to monitor tortoise movements.

These historic surveys yielded basic information about the distribution of

remaining angonoka populations as well as preliminary information on seasonal tortoise

movements. In order to supplement the information from these early studies, this 2-year

field project was initiated in 1993.

A field research station was established at Cape Sada on the eastern side ofBaly

Bay in October 1993. A continuous study of the ecology of the Cape Sada angonoka

population was conducted from October 1993 through June 1995. The Cape Sada

research station also was used as a base camp from which regional tortoise surveys were

conducted in 1994. This chapter summarizes the results of the regional surveys and

includes demographic data collected from the Cape Sada population. The specific goals

of this portion of the study were as follows:

1. To visit all historic and potential angonoka localities in the Baly Bay region

and to assess the status of the populations and the bamboo-scrub habitat.

2. To map all confirmed angonoka localities and the extent of the habitat.

3. To describe aspects of the demography of the Cape Sada angonoka



Field surveys were conducted from October 1994 through April 1995 at 10

localities in the Baly Bay region. Areas surveyed were chosen based on both historic

records and on interviews with local villagers. Surveys were concentrated in the wet

season (from November through April), when angonoka are most active (Juvik et al.

1981; Juvik et al. 1997). The localities surveyed and field trip dates are listed in Table 2-

1. Monthly surveys were conducted at Cape Sada (from October 1993 to March 1995) in

an attempt to mark all individuals in the population. All surveys generally were

conducted between 0600-1100 hrs and 1500-1800 hrs.

The bamboo-scrub habitat of the angonoka contains impenetrable patches of

shrubs and bamboo that precluded the use of formal transect surveys. Rather than clear

transects through the habitat, surveys in this study consisted of timed searches for

tortoises. During these searches, participants walked roughly parallel transects looking

for tortoises, tracks, or feces. Transects were approximately 10-20 m apart, depending

upon the density of the vegetation. In heavily vegetated areas, where visibility was low,

transects were as close as necessary to search each area thoroughly. The number of

survey participants ranged from three to seven individuals including trained field

assistants and guides. The observation rate for each area was calculated as the number of

tortoises observed per survey-hour.

When a tortoise was encountered, the date and time of capture were recorded.

Daily rainfall totals and minimum and maximum temperatures were recorded at Cape

Sada. Tortoise localities were determined with an Ensign XL Global Positioning System

(Trimble Navigation, Sunnyvale, CA), which was accurate to within 10-100 m. The

extent of tortoise habitat in different regions was estimated using 1990 Landsat Thematic

mapper digital data (Juvik et al. 1997).

Adult and subadult tortoises were marked by notching a series of marginal scutes

(Cagle 1939). Hatchling and small juvenile tortoises were marked with enamel paint on

the marginal scutes. Trovan Passive Integrated Transponders (RS Biotech) were

implanted subcutaneously above the forelimb of tortoises greater than 100 mm carapace

length. Straight-line carapace length (CL), shell height, and plastron length of tortoises

greater than 100 mm were measured to the nearest 1.0 mm using HaglofMantax

aluminum calipers (Forestry Suppliers, Inc., Jackson, MS). A metric dial caliper

(accurate to the nearest 0.01 mm) was used to measure CL of hatchlings and juveniles

less than 100 mm in length. Scute growth rings were counted using the first or second

costal plate of all tortoises, following methods outlined in Zug (1991). Growth rings

were difficult to distinguish in very large tortoises; therefore, a minimum number of rings

was recorded in these individuals.

The sex of adult tortoises was determined based on differences in shell

morphology. Adult male angonoka have an elongated gular and a concave plastron.

Also, the width of the anal fork of adult males is nearly twice that of the anal notch.

These characteristics generally could be distinguished in individuals greater than 300 mm

CL or with approximately 13 to 16 scute growth rings (see Chapter 5).


Tortoise presence was confirmed at 10 out of 11 areas surveyed (Figure 2-1), and

145 tortoises were marked (Table 2-1). For comparative purposes, observation rates for

the areas surveyed are presented in Table 2-2. Three populations east of Baly Bay were

confirmed (Cape Sada, Ankasakabe, and Beheta), and at least two populations west of the

Bay were identified (Betainalika and the region including Ambatomainty, Anjaha,

Andrafiafaly, Antsahamena-South, Andranolava, and Antsahavaky). The distance

between the closest east and west population is approximately 24-km over land, and

extensive savanna and the Andranomavo River separate the two areas. Although most of

the tortoises (68 %) were found on Cape Sada, where monthly surveys were conducted,

the most extensive tract of contiguous angonoka habitat occurs west of Baly Bay in the

region including Ambatomainty, Anjaha, Andrafiafaly, Antsahamena-South,

Andranolava, and Antsahavaky. This area extends more or less unbroken from

Antsahamena in the north and Ambarindranahary in the east to Andranolava in the south.

Survey results for the 11 localities are presented below.

East Baly Bay

Cape Sada. Ninety-six tortoises were encountered on Cape Sada between 1993

and 1995. This total includes 21 of 24 tortoises that were marked by JWPT field teams

prior to 1993 (three tortoises that had been marked in previous surveys were not

recaptured in this study). A total of 1,297.25 survey hours were conducted over the

period. Of the 99 individuals marked on Cape Sada, 14 were adult male, 27 were adult

female and 58 were juveniles less than 300 mm CL. The sex ratio among adult tortoises

was 1:1.93 and differed significantly from 1:1 (X2= 4.05, df= 1, P= 0.04).

The mean CL of adult male tortoises on Cape Sada was 409.6 mm (n-=14, range

282.0-456.0, SD = 45.0) as compared to 361.8 mm in adult females (n= 26, range =

285.0- 405.0, SD = 29.0) (Table 2-3). Tortoises classified as juveniles ranged in size

from 43.5 mm CL to 310.0 mm CL. Sixteen of the 58 juveniles were hatchlings that

were measured as they emerged from nests. The mean CL of newly emerged hatchlings

was 47.2 mm (n=16, range = 43.5 52.0, SD = 3.0).

Nearly 40 % of the tortoises captured on Cape Sada were juveniles with less than

85 mm CL (Figure 2-2), indicating that successful reproduction is occurring in this

population. However, only 22 % of the tortoises were in intermediate size classes (84.8-

322 mm CL). Although it is possible that intermediate sized tortoises were simply more

difficult to find than adults or small juveniles, I suspect that these tortoises were

underrepresented in surveys because survivorship of hatchlings and juveniles is very low.

The partial remains of 22 dead juvenile angonoka were found during monthly surveys on

Cape Sada. The remains of 15 of the 22 dead juvenile tortoises consisted of only a few

scutes and the cause of mortality could not be determined. However, five of the dead

juveniles with 2-3 scute growth rings appeared to have been killed by birds of prey or

small mammals. Potential predators of small juvenile angonoka include the Madagascar

buzzard (Buteo brachypterus), common tenrec (Tenrec ecaudatus), fosa (Cryptoprocta

ferox), and introduced species such as civet cat (Viverricula indica), and black rat (Rattus

rattus). Local people claim that boa (Boa "Acrantophis" madagascariensis) and yellow-

billed kite (Milvus migrans) also are predators of young angonoka (Juvik et al. 1981).

Two juveniles with 5 and 7 scute growth rings (approximately 70-100 mm CL) may have

been killed by a large predator. The shells of these tortoises had been broken between

bone sutures and large puncture wounds were present. The fosa is the largest native

carnivore in Madagascar (Garbutt 1999); however, adults of this species weigh less than

10 kg, and it is unlikely that they could kill and consume large juvenile angonoka.

African bush pigs have a head length of approximately 350 mm and probably are capable

of killing large juvenile angonoka (Oliver 1993). However, the tortoises may have died

of other causes and simply been scavenged by African bush pigs.

Tracks, feces, and digging of African bush pigs often were observed on Cape

Sada, particularly during the wet season. Twelve angonoka nests were monitored for

disturbance by predators during this study (see Chapter 4); however, no direct evidence

of egg predation was observed. In a subsequent study at Cape Sada, a nest containing

three eggs was destroyed by bush pigs (Pedrono 1996).

During monthly surveys, tortoises were most frequently observed in January,

when, on average, a tortoise was encountered every 4 survey-hours. High observation

rates generally corresponded with months of high rainfall (Figure 2-3). A comparatively

high observation rate among adult tortoises in October probably reflected breeding

activity (see Chapter 4). A large proportion of adult female tortoises was observed in

May during peak nesting season. Many juvenile angonoka (with 1-3 shell growth rings)

were observed in open areas in February and March. Very few tortoises were

encountered during June, July, and August, the height of the dry season. The radio-

telemetry study revealed that tortoises were inactive during these months, typically

resting beneath vegetation (see Chapter 3).

Although fire scars were evident on some of the large trees on Cape Sada, the area

probably had not burned since the earliest JWPT field surveys in 1988 (Reid 1989).

However, a brush fire on Cape Sada after our study resulted in the death of at least one

adult tortoise (L. Durrell, pers. com.).

Juvik et al. (1997) described three populations of angonoka at Cape Sada that

inhabited "open scrub forest habitat" in the north, central and southern portion of the

peninsula. However, during monthly surveys I found that tortoises often moved between

the open scrub areas delineated by Juvik et al. (1997) and also used the dense bamboo

thickets dividing them. Therefore, I feel that Cape Sada hosts a single population of

angonoka. The number of unmarked tortoises observed on Cape Sada declined steadily

from February through March 1995 (Figure 2-4) and no unmarked tortoises were found

during the last three months of the study. It is likely that nearly all adult tortoises in the

population have been marked, but some juvenile tortoises may have been missed. The

Cape Sada peninsula is roughly 150-ha in size and the estimated tortoise density is 0.66

tortoises per ha.

A small patch of bamboo-scrub habitat occurs east of the Cape Sada peninsula.

Juvik et al. (1997) reported finding old tortoise droppings in this area during a 1992

survey. However, nearly 80 survey-hours were spent in this area during the 1993-94 wet

season and no tortoises or tortoise sign was observed.

Ankasakabe. Two adult male tortoises were observed at Ankasakabe in 32.5

survey hours in April 1995, yielding an observation rate of 0.06 tortoises per survey-hour

(Table 2-2). This value is slightly less than that of Cape Sada during April surveys.

Tracks of juvenile tortoises were seen on two occasions at Ankasakabe. The presence of

juvenile tortoises indicates that a breeding population still exists at Ankasakabe.

However, I suspect that the population is very small because the tortoise habitat is less

than 50 ha in size and contains only a few stands of bamboo and Terminalia sp.

surrounded by extensive savanna. The area appears to burn frequently and is

comparatively accessible to humans because an ox-cart path passes directly through the


Both male tortoises discovered at Ankasakabe had small holes in a rear marginal

scute and the gular of one had been cut off. Local villagers had probably kept these

tortoises as pets in the past. Captive tortoises often are kept tethered and the gular is

removed because it is believed to prevent the tortoise from eating.

Beheta. Beheta was the easternmost locality where tortoises were found and the

habitat was roughly 200 ha in size (Figure 2-1). Eighteen tortoises, nine males (298.0-

410.0 mm CL), six females (282.0-378.0 mm CL), and three juveniles (116.0-218.0 mm

CL) were marked during nearly 164 survey-hours in October and December 1994, and

March 1995. The highest observation rate at Beheta occurred during December (0.14

tortoises per survey-hour); this observation rate was similar to that recorded at Cape Sada

(0.13 tortoises per survey-hour) over the same period. Although the observation rates at

Beheta and Sada in this study were similar, further surveys are needed to determine the

population size at Beheta.

The tortoise habitat at Beheta appeared similar to Cape Sada in that it contained

bamboo, Terminalia spp. and Bauhinia sp. However, unlike Sada, stands of shrubs and

bamboo were comparatively sparse. Although vegetative differences were not quantified

in this study, brush fires appeared to have degraded the tortoise habitat at Beheta.

Villagers mentioned that dry season brush fires are widespread in the region. During the

October 1994 visit, a brush fire swept through the area where an adult tortoise had

recently been observed. The area was searched during December 1994 surveys and no

evidence was found to suggest that the fire killed the tortoise. However, the animal may

have been killed, and the carcass scavenged prior to our return. An adult male fire-

scarred tortoise and the partial remains of two other adult tortoises were observed in areas

that had burned in the past. Although the cause of mortality in these tortoises could not

be determined, due to their large size, adult angonoka probably have no natural predators

suggesting that fire may have caused the death of these individuals.

One adult female tortoise discovered at Beheta had a hole in a rear marginal scute

and had presumably been kept as a pet in the past. Like Ankasakabe, Beheta is

comparatively accessible and villagers at Antanandava claim that tortoises are taken from

the area on occasion. During a 1992 survey, Juvik et al. (1997) found considerable

evidence of bush pig rooting at Beheta. Tracks, rooting, and feces of bush pigs were

observed during this study, but no direct evidence of pig predation on eggs or young

tortoises was documented.

West Baly Bay

Ambatomainty, Anjaha. Andrafiafaly. Antsahavaky, Antsahamena-south,

Andranolava. and Antsokotsoko. Bamboo-scrub habitat in these seven areas appeared to

be contiguous and probably supports the largest remaining angonoka population. Cloud

cover present in the Thematic mapper images of this region did not allow an accurate

estimate of the extent of bamboo-scrub habitat in this region. However, it appears to

include an area between 4,000-6,000 ha in size. The region contains extensive tracts of

bamboo and is relatively inaccessible to humans. The closest human settlements are the

villages of Ankoro to the south and Ambarindranahary to the north. A seasonally

occupied village exists on the coast at Antsahamena (Figure 2-1).

Twenty-five tortoises were observed in the December and January surveys; ten

were adult male (262.0-481.0 mm CL), five were adult female (353.0-395.0 mm CL), and

ten were juveniles (53.1-277.0 mm CL). The largest angonoka reported, a male with 481

mm CL, was found at Ambatomainty in the northern portion of the region. The highest

observation rate also occurred at Ambatomainty (0.17 tortoises per survey-hour), but

comparatively high observation rates also were recorded at Anjaha (0.14 tortoises per

survey-hour) and Andrafiafaly (0.11 tortoises per survey-hour). These numbers are

similar to those recorded at Cape Sada over the same period (0.11 and 0.24 tortoises per

survey-hour in December and January, respectively).

The southern-most localities in this region occurred at Andranolava. Although no

tortoises were observed here, tortoise feces were found in three separate locations in the

area. Tortoise presence was not confirmed at Antsokotsoko, a large tract of bamboo-

scrub habitat that lies between Andranolava and Andrafiafaly. The habitat at

Antsokotsoko appears to be contiguous with Andranolava and Andrafiafaly and is similar

other sites in the region. Further surveys in this area are needed.

Evidence of brush fires was common throughout the West Baly Bay region.

Angonoka were observed using recently burned bamboo-scrub habitat. Two of the 25

tortoises (juveniles with 80 and 140-mm CL, respectively) found during surveys had burn

scars. The remains of an adult angonoka were discovered on the beach near the seasonal

settlement at Antsahamena. Since local people apparently do not eat angonoka (Durbin

et al. 1996), this animal may have been killed and consumed by transient fishermen.

Bush pig sign was observed throughout the region, but insufficient time was spent in the

area to detect any direct impact of pigs on the tortoise population.

An undescribed species of tick (Amblyomma sp.) was observed on five tortoises

from Andrafiafaly, Ambatomainty, and Anjaha (L. Durden, pers. com.). Two of the

tortoises were adult female, two were adult males, and one was a juvenile with six scute

growth rings. Each infected tortoise had from 2-5 ticks attached near the tail and hind

legs. This observation represents the first record of an ectoparasite from free ranging G.

yniphora and a description of this new tick is in preparation at Georgia Southern

University (L. Durden, pers. com).

Betainalika. Betainalika lies approximately 6-km southeast of Andrafiafaly and is

separated from the West Baly Bay tortoise habitat by dense deciduous forest and savanna.

The bamboo-scrub at Betainalika encompasses approximately 340 ha. The habitat

contained only scattered bamboo stalks and appeared to bur frequently. Although the

assessment of the habitat in this study was qualitative, Betainalika appeared to have the

most extensive fire damage of all the areas visited.

Local villagers repeatedly mentioned that Betainalika had many angonoka.

However, during 41 survey-hours in April 1995, only one tortoise was encountered (0.02

tortoise per survey-hour) and neither tracks nor feces were observed. The tortoise

observed was a juvenile with 51.9-mm CL and probably was a 1995 hatchling. Despite

the low observation rate in this study, the presence of a juvenile tortoise indicated that

breeding adults were present in the area. Further surveys in this region are needed to

accurately assess the size and status of this population.


The tortoise localities identified in this study probably represent at least five

separate populations, all of which occur within a 30-km radius of Baly Bay. The "east"

and "west" populations are effectively isolated from one another by Baly Bay, the

Andranomavo River, and extensive savanna. The most extensive tract of angonoka

habitat occurs west of Baly Bay (including the regions of Ambatomainty, Anjaha,

Andrafiafaly, Antsahavaky, Antsahamena-south, Andranolava, and Antsokotsoko). This

area is approximately 4,000-6,000 ha in size and probably contains the largest angonoka

population. More intensive tortoise surveys are needed in this region to determine the

population size, and the area should be evaluated to determine the precise extent and

quality of the tortoise habitat. The Betainalika region also occurs west of Baly Bay and

probably represents a separate population. Betainalika is much smaller than the west

Baly Bay region and appeared to contain low tortoise densities. However, further

surveys in this region also are warranted.

The three angonoka populations east of Baly Bay, Cape Sada, Beheta, and

Ankasakabe, occur on small tracts of habitat (<200 ha in size) and are separated by

extensive savanna. Cape Sada is only about 150 ha in size and hosts a population of 99

tortoises. Beheta is roughly the same size as Cape Sada and although the habitat appears

somewhat degraded by fire, similar observation rates suggest that the populations may be

similar in size. Tortoise habitat at Ankasakabe is extremely restricted and this angonoka

population is undoubtedly very small. The observation of tracks of a juvenile tortoise at

Ankasakabe indicates that angonoka are able to reproduce successfully in small habitat

fragments. However, it is not known whether recruitment into the breeding population is

occurring in this limited habitat.

The tortoise density on Cape Sada (0.66 tortoises per ha) appears low when

compared to other dry scrub-forest tortoises. For example, the density of Hermann's

tortoise (Testudo hermanni) at a site in northeastern Spain was nearly 11 tortoises/ha

(Mascort 1997) and densities of 3-4 tortoises/ha have been reported for the Chaco tortoise

(Geochelone chilensis) in Argentina (Waller and Micucci 1997). Radiated tortoises

(Geochelone radiata) in southern Madagascar inhabit a much drier climate than

angonoka; however, estimates of 7-15 tortoises/ha have been reported for this species (R.

Lewis, pers. cor.). The apparent low tortoise density at Cape Sada may reflect the

effects of past harvest, poor habitat quality, or low survivorship. Comparative studies at

other angonoka populations are needed to determine whether densities are similar in other

patches of bamboo-scrub habitat.

Juvenile tortoises were found in all areas surveyed, indicating that successful

reproduction is occurring throughout the range of the angonoka. Although African bush

pigs are known to have destroyed one tortoise nest after this study (Pedrono 1996), some

nests are clearly surviving to produce hatchlings. However, a large mammal such as an

African bush pig may have killed at least two large juvenile tortoises on Cape Sada.

Bush pig predation may have a significant effect on recruitment, particularly if pigs prey

heavily on juvenile tortoises that are too large to be consumed by native predators.

Additional dead juveniles observed on Cape Sada in this study, and an apparent lack of

intermediate sized tortoises, may indicate low juvenile survivorship in this population.

Brush fires probably pose the most serious threat to remaining angonoka

populations. The 1995 fire on Cape Sada confirmed that brush fires kill angonoka and it

seems likely that frequent brush fires also affect the structure of the bamboo-scrub

habitat. Although no quantitative data were collected, the bamboo-scrub at Beheta,

Ankasakabe, and Betainalika appeared badly degraded by brush fires. Bamboo thickets

at these sites were structurally quite different than at other areas. Bamboo was notably

sparse and dense vines had encroached in some areas.

In order to assess the impact of anthropogenic fires on the bamboo-scrub

ecosystem, the role of natural fire in the region must be determined. Paleoecological

investigations have shown that brush fires occurred in western Madagascar prior to

human settlement (Burney 1997). However, the frequency of anthropogenic dry season

fires is probably quite different from that of natural fires. Given the highly seasonal

rainfall pattern in the region, it is unlikely that lightning fires would occur in the dry

season (when most anthropogenic fires occur). Furthermore, lightning induced wet

season fires would probably be less catastrophic than dry season fires because the fuels

would be more moist. Research is needed to determine the direct effects of fire (both

anthropogenic and natural) on the bamboo-scrub ecosystem.

In addition to evaluating the effects of brush fires on the angonoka, it also will be

important to determine the effects of cattle and African bush pigs on the bamboo-scrub

habitat. In a desert tortoise (Gopherus agassizii) population, tortoises were in direct

competition with cattle for food (Avery and Neibergs 1997). Cattle were responsible for

degrading Chaco tortoise (Geochelone chilensis) habitat by grazing new shoots of shrubs

following anthropogenic fires (Waller and Micucci 1997). Overlap in diet between feral

pigs (Sus scrofa) and giant tortoises (Geochelone nigra) was documented on Isla

Santiago in the Galapagos, although direct competition between the two species was not

confirmed (Coblentz and Baber 1987). Considerable pig rooting activity was observed in

all regions surveyed in this study. Additional information is needed to evaluate the

impacts of cattle and bush pigs on bamboo-scrub habitat.

Most of the remaining angonoka populations are very remote and collection of

tortoises does not appear to be a threat at this time. On a local level, tortoises may

occasionally be kept as pets, but community education efforts have resulted in the

donation of captive animals to the breeding program (Curl 1986b; Reid et al. 1989), and

others have apparently escaped or been released back into the wild (e.g., Beheta and

Ankasakabe). However, the theft of 75 angonoka from the captive-breeding center at

Ampijoroa in 1996 demonstrates a demand for this rare species in the international pet

trade (Webster 1997). Declaration of the Baly Bay region as a National Park by the

Malagasy government is imminent; however, wild populations still may be vulnerable to

illegal collection in the future. The removal of adult tortoises from any of the wild

populations could have catastrophic effects. The apparent low survivorship of juvenile

angonoka coupled with a slow rate of development to sexual maturity could severely

limit the capacity of a population to recover from the loss of adult tortoises (see Congdon

et al. 1993). It may be necessary to use guards to protect the most accessible angonoka

populations (e.g., Cape Sada and Beheta).

The angonoka is vulnerable to extinction in the wild primarily because of its

extremely limited geographic distribution. The remaining populations are small, isolated,

and all occur within only a 30 km radius of Baly Bay. Although juveniles were found in

all populations, the population structure at Cape Sada indicates that recruitment into the

breeding population may be low. A catastrophic event such as disease or severe weather

could cause extinction of some or all populations. Augmentation of the Cape Sada

angonoka population with captive born juveniles is not warranted at this time. However,

repatriation (release of individuals of a species into an area formerly occupied by that

species) may be worthwhile in other areas. It is of critical importance to protect and

manage all wild angonoka populations in order to preserve the species.

Table 2-1. Number of wild angonoka (Geochelone yniphora) observed during regional surveys around Baly Bay in western
Madagascar from 1993 through 1995. *= Although no tortoises were seen, tracks and/or feces were observed.


East Baly Bay
Cape Sada



West Baly Bay





Antsahamena- South*


6 October 1993 24 June 1995

11-12 October 1994; 11-15 December 1994; 17-19
March 1995; 22 May 1995

20 March 1995; 8-9 April 1995

3, 4 and 5 January 1995

21, 22, and 27 December 1994

26 December 1994

23 and 31 December, 1994: 2, 4, and 7, January 1995
24 December 1994

26 and 28 December 1994; 21 April 1995

6 January 1995

17-21 April 1995

2 0 0 2
25 33 61 119

2 2 0 4

3 1 3 7

0 0 0 0

4 2 6 12
1 0 1 2

0 0 0 0

0 0 0 0

0 0 1 1
10 5 11 26

Table 2-2. Observation rates of angonoka (Geochelone yniphora) during 1994-1995 field
surveys in western Madagascar. Rates were calculated as the number of tortoises
observed per person/hour of survey. tOnly December and January survey results are
presented here for comparative purposes. *Tortoise tracks and/or feces observed.




East Baly Bay

Cape Sadat



West Baly Bay









































Table 2-3. Mean carapace length of wild caught angonoka (Geochelone yniphora) in
western Madagascar. Measurements are in millimeters. *Includes Ambatomainty,
Anjaha, Andrafiafaly, and Antsahavaky.


East Baly Bay

Cape Sada



West Baly Bay

Western areas*





























282.0 456.0

285.0 -405.0


43.5 52.0

298.0- 410.0

282.0 378.0


396.0 432.0

262.0 481.0

353.0 395.0

53.1 -277.0












Figure 2-1. Angonoka tortoise (Geochelone yniphora) localities in the Baly Bay, region
of western Madagascar. Locality information is based on observations of tortoises
and/or the presence of tortoise tracks or feces during 1993-94 field surveys.

Juveniles H Females 0 Males

> 25
c 20
f 10
5 I
o 0m ....

Carapace Length (mm)

Figure 2-2. Size class distribution of the Cape Sada angonoka tortoise (Geochelone
yniphora) population (n=98).

Males lC Females rI Juveniles -X- Rainfall





200 C




Figure 2-3. Mean monthly observation rates for angonoka tortoises (Geochelone
yniphora) as related to monthly rainfall totals at Cape Sada, Madagascar. Observation
rates were calculated as the number of tortoises observed per person-hour of survey.








S 80
U 50
o 30

* Marked 1 Unmarked


Figure 2-4. The ratio of marked to unmarked angonoka tortoises (Geochelone yniphora)
observed at Cape Sada, Madagascar. Data were collected from October 1993 through
March 1995.

---------- - m- -



Little is known of the habitat requirements of the angonoka. All records for this

species indicate that it is restricted to bamboo-scrub habitat in the Baly Bay region of

western Madagascar (Siebenrock 1903; Juvik et al. 1981; Curl et al. 1985; Reid 1989,

1990a, 1990b). Today there are less than 7,000 ha of bamboo-scrub remaining and the

habitat is fragmented and surrounded by extensive savanna (Chapter 2). In order to

protect and manage existing angonoka populations given the limited habitat, it is critical

to identify components of the bamboo-scrub ecosystem that are important to these

tortoises. It also is critical to quantify the spatial needs and dispersal behavior of

individual tortoises. Home range is the area used by an animal during its normal

activities of food gathering, mating, and caring for young (Seton 1909; Burt 1943; Stickel

1954) and is an important measure of the spatial needs. The most common methods of

estimating home range size are reviewed below.

The objectives of this part of the study were to determine home range size and

patterns of microhabitat use in adult and juvenile angonoka. I examined (1) variation in

home range size among males, females, and juveniles; (2) differences in home range size

across seasons; (3) differences in microhabitat use between male, female, and immature

tortoises; and (4) seasonal differences in microhabitat use.

Review of Home Range Methods

Most early methods for estimating home range were developed for small mammals

and used observations based on sight records or mark/recapture (Blair 1940; Burt 1943;

Dalke 1942; Hayne 1949; Mohr 1947). Sight records often were opportunistic and these

studies generally had very low sample sizes. Trapping methods used in mark/recapture

studies relied on the unlikely assumption that traps within a home range were equally

successful and that animals did not move beyond the trapping grid. The development of

radio-telemetry in the late 1960s allowed researchers to monitor animal movements more

closely and to directly observe animal behavior (Adams 1965).

Numerous methods for calculating home range have been described since the

1940s (see Worton 1987). Nonstatistical methods, such as polygon estimates, give the

extent of an animal's home range. The three most commonly used polygon methods are

minimum area (MA), minimum convex polygon (MCP), and modified minimum area

(MMA). In the MA method, straight lines connect observation points and the area

enclosed by the polygon is measured (Dalke 1942; Stickel 1954; Southwood 1966). This

method has not been clearly defined and the shape of the polygon depends on the order in

which the points are connected (Jennrich and Turner 1969). MCP is simply the smallest

possible convex polygon containing all of the observation points. This method is very

sensitive to movements on the periphery of the animal's home range, regardless of the

frequency with which that area is visited by the animal. A further limitation to this

method is that large areas of land that are not visited by the animal can be included in the

polygon. The MMA method is a modification of the MCP method that was designed to

eliminate areas within the polygon where an animal does not go. Any points that are

further than one-fourth the range length (the distance between the two farthest

observation points) from any other point are considered forays outside the animal's home

range and are excluded from the polygon.

Polygon methods give no information about the internal anatomy of the home

range. Most animals do not use their entire range with equal intensity (Hayne 1949); they

tend to occupy particular areas within their home range (e.g., dens, nest sites, and limited

resources) with greater frequency than other areas. Areas of concentrated use are called

centers of activity or core areas (Samuel et al. 1985). Statistical home range models

attempt to measure the intensity of use within the home range. These models define

home range as the area of the smallest subregion, which accounts for some proportion

(usually 95 %), p, of its total utilization.

The earliest statistical models include recapture radius/standard circle and the

covariance matrix method (Harrison 1958; Dice and Clark 1953). The recapture radius

method defines a "standard diameter" as twice the square root of the mean square of all

distances between observation points and the geometric center of activity. A circle based

on this diameter is called a "standard circle" and contains 68.26 % of all observations.

The covariance matrix method is a modification of the recapture radius method that is

designed to measure non-circular home ranges (Calhoun and Casby 1958; Jennrich and

Turner 1969). This model assumes that the utilization distribution is bivariate normal.

The most heavily used areas are bounded by concentric density ellipses that account for a

proportion of the animal's total use of its home range. Circular and the covariance matrix

methods have little biological significance because home ranges rarely are circular or


Recent utilization distribution (UD) models calculate home range as a discrete or

continuous function that represents the intensity with which an animal uses points in its

habitat (Siniffand Tester 1965; Adams and Davis 1967; Voight and Tinline 1980;

Anderson 1982; Don and Rennolls 1983; Worton 1987). These methods, unlike circular

or elliptical models, carry no assumption about the form of the home range. UDs are

presented graphically as contour lines, each of which is a set of points where the

probability of occurrence is constant. For example, the kernel area (KA) utilization

distribution method (Worton 1989), uses a probability density function, called a kernel.

A kernel is placed over each observation point and an estimator is constructed by adding

the n components (random sample ofn independent points). Where there is a

concentration of points, the kernel estimate has a higher density than where there are few

points. A smoothing parameter, h, is used to control the amount of variation in each

component of the estimate. A large h obscures all but the most prominent feature,

whereas a small h retains the fine detail of the data.

As yet there is no ideal measure of home range. MCP has historical significance

(Stickel 1954; Rose 1982) and has been found to be more comparable between species

and individuals than other measures (Mohr 1947). The method is simple to use and is

based on actual observations of an animal. Sample size bias can be removed with

adequate sampling; however, the problem of incorporating areas not used by the animal is

more difficult to resolve. MMA does not completely resolve the problem because the use

of one-fourth of the range length to exclude outlying points is arbitrary.

Recapture radius/standard circle and bivariate normal methods are unreliable

because they assume a normal distribution of the data. This assumption rarely is met;

home ranges often are linear or irregularly shaped. Animal movements are not random

and weighted measures of centers of activity probably present a more realistic measure of

spatial patterns. The most robust estimators of UD are methods such as harmonic mean

and kernel area, which do not make assumptions about the underlying distribution of the


In this study, I used both MCP and kernal area (KA) method for estimating home

range size. MCP method was used so that the results would be comparable to those of

other studies. Also, I felt I could resolve the problem of sample bias in this method by

obtaining a large number of sample points for each individual. KA method was used to

obtain a more accurate representation of the functional home range of the angonoka and

to depict areas of concentrated use within the home range.


Thirteen tortoises (3 males, 5 females, and 5 juveniles with 8-10 scute growth

rings) were radio-instrumented and tracked for periods ranging from 121 to 630 days

(Table 3-1). Adult tortoises were fitted with SB-2 module transmitters and juveniles with

smaller SM-1H modules that weighed 30 g and 7 g, respectively (AVM Instrument

Company, Ltd., Livermore, CA). Transmitters (including batteries) weighed less than 1

% of the tortoises body weight and were attached to the posterior costal scutes of the

tortoise with PC-11 Epoxy Paste (Protective Coating Co., Allentown, PA).

The tortoises were located by direct observation using an LA-12 receiver (AVM

Instrument Company, Ltd., Livermore, CA). Each animal was located 7 mornings

(between 0600-1030 hrs) and 7 afternoons (between 1430-1800 hrs) per month, weather

permitting. At each location, the time, weather, air and substrate temperature, relative

humidity, and microhabitat type (scrub-shrub, bamboo, grass, or open) were recorded.

The scrub-shrub microhabitat contained shrubs less than 2 m in height with

scattered grasses and sedges in the ground cover. The most common shrub species were

Bauhinia sp. and Terminalia spp. The bamboo microhabitat consisted of dense thickets

of bamboo (Perrierbambos madagascariensis) with a few shrubs and satrana palms

(Bismarkia nobilis), but ground cover vegetation was rarely present. Small grassy areas,

generally less than 5 m2 in area, were classified as "grass" microhabitat. These areas

contained savanna grasses including Aristida sp., Eragrostis sp., and Heteropogon

contortus and a variety of sedges, although they did not fit the traditional definition of

savanna (Koechlin 1972). Open unvegetated patches (typically <3 m2 in area) with an

igneous rock substrate were considered a fourth microhabitat. These small areas

typically were found within the scrub-shrub and grass microhabitats.

A 50-m2 grid was created across most of Cape Sada. Tortoise locations were

marked with colored flagging, numbered consecutively, and the distance and compass

bearing to the nearest grid point taken. If no transect or grid point was available,

localities were determined with an Ensign XL Global Positioning System (Trimble

Navigation, Sunnyvale, CA).

Minimum convex polygon (MCP) (Mohr 1947) and kernel area (KA) (Worton

1989) methods were used to estimate home range size. Analyses were carried out using

the computer program Ranges V (Kenward and Hodder 1996). MCPs were based on 100

% of the fixes and KAs were calculated using 95 % contours. Least-squares cross-

validation was used to choose the optimum smoothing factors in KA analysis (Silverman

1986; Kenward and Hodder 1996).

Sampling periods were grouped into two seasons for analysis (November April

= wet season; May October = dry season). A Wilcoxon paired-difference test was used

to examine wet and dry season effects on home range size and a Kruskal-Wallis test was

used to test for differences in home range size between males, females, and juveniles in

both wet and dry season.

A 1990 thematic mapper satellite image (Juvik et al. 1997) was used to quantify

the extent of the various microhabitat types on Cape Sada. Areas were estimated using a

polar planimeter. A Chi-squared goodness-of-fit test was used to determine whether

there was a difference between observed and expected microhabitat use, based on habitat


Analysis of categorical data (percent of observations by sex, season, and

microhabitat) was carried out using analysis of covariance (PROC GLM) (SAS Institute,

Inc. 1992). If significant treatment effects were detected, Tukey's studentized range

(HSD) test was used to determine where the differences occurred (Kushner and De Maio

1980). Scrub-shrub was excluded from the microhabitat analysis because it accounted

for most of the available habitat and interactions would be difficult to distinguish.


Home Range. Home range maps are presented in Appendix A. There was

considerable variation in home range size among individual tortoises (Table 3-2).

However, several general patterns were evident. Wet season home ranges were

significantly larger than dry season home ranges with both MCP and KA method (MCP:

t= 2.70; df- 12; P= 0.01; KA: t= 2.88; df= 12; P= 0.02). MCP values were greater than

KA values in both wet and dry season because KA estimates were based on only 95 % of

observations and excluded areas not visited by the tortoise. Because KA probably best

reflected actual home range size, only comparisons among KA values are discussed


Wet season home range areas were significantly different among the sex/life

stages (H= 9.33; P<0.01). The average size of juveniles tracked in this study was 167

mm CL as compared to 427 and 362 mm CL in adult males and females, respectively. In

general, body size is related to home range size, because larger animals often need to

forage farther to meet their energy requirements than do small animals (McNab 1963),

although other factors such as metabolism, feeding strategy, and probability of predation

(Weatherby 1995) also may influence home range size. These factors may explain the

difference in home range size between adult and juvenile angonoka. However, immature

tortoises of other species occasionally use larger home ranges than adults because they

conduct long distance dispersal movements (Auffenberg and Iverson 1979; Aguirre et al.

1984; Diemer 1992). The juveniles followed in this study did not exhibit this behavior.

In adult angonoka, behavior rather than body size probably best explains the variation in

home range size between the sexes (Figure 3-1). Males in this study were more nomadic

than females and traveled over large areas, possibly in search of mates.

No significant difference in dry season home ranges between the sex/life stages

was detected (H= 0.54; P>0.10). These results may be an artifact of the small sample

size in this study because the mean juvenile home range size (0.6 ha) was much smaller

than that of adult males (6.6 ha) and females (3.6 ha).

There was little overlap among home ranges of adult males and females in the dry

season (Figure 3-2). During the wet season, the home range of adult males expanded to

encompass all or portions of that of females, but male ranges only intersected at the

periphery (Figure 3-3). The apparent lack of overlap among adult males may indicate

that angonoka exhibit a dominance hierarchy among males (Harless 1979). However,

more data are needed in order to get a clear picture of the spatial distribution of all

tortoises in this population. The apparent low tortoise density on Cape Sada may

influence spatial distribution. Home range areas of 5 juveniles followed in this study did

not overlap in either the dry or the wet season (Figure 3-4 and 3-5). Again, this may be

an artifact of the low tortoise density on Cape Sada. However, the tortoises radio-tracked

in this study represent only small fraction of those present on Cape Sada. The home

ranges of the individuals in this study may have overlapped with tortoises that were not


Microhabitat Use. Scrub-shrub is the predominant microhabitat type on Cape

Sada and occupies approximately 59% of the peninsula. The bamboo microhabitat

occurs primarily on the highest elevations of the Cape and covers approximately 28% of

the study area. Patches of grass occupy approximately 4%. Open microhabitat was

difficult to distinguish on the satellite image because it typically occurs in patches only a

few square meters in size within the scrub-shrub or grass. However, open areas probably

occupy about 4% of the site. Komanga forest covers approximately 5% of the peninsula.

Tortoises were most often observed in scrub-shrub, (2070 of 3149 observations;

65.7 %) followed by bamboo (819; 26.0 %), grass (164; 5.2 %) and open habitat (96; 3.1

%). Radio-instrumented angonoka were never encountered in the komanga forest.

Angonoka on Cape Sada used microhabitats relative to their availability (X2= 6.26; df= 4;

P= 0.20); however, season appeared to have an effect on microhabitat selection (F= 3.21,

df- 2, P= 0.0447) (Table 3-3). All angonoka used the open microhabitat significantly

more often in wet season than in dry season, while angonoka were observed significantly

more often in the grass microhabitat in dry season than in wet season (Table 3-4).

Although no significant difference in microhabitat use between male, female and

juvenile tortoises was detected (F=0.82; df- 4; P=0.1390), when the data were examined

by month (Figures 3-6, 3-7, and 3-8), it appeared that adults of both sexes used bamboo

more often in the late dry season (September and October) and early wet season

(November). Juvenile tortoises appeared to use the bamboo microhabitat more frequently

during wet season months than during the dry season months (Figure 3-8).


Despite a great deal of individual variability, seasonal differences in adult home

range size were observed in this study. Home range areas were larger in the wet season,

when food and water were plentiful, than in the dry season when these resources were

limited. Tortoises were most often observed in scrub-shrub (SS) and bamboo (BA), the

most common microhabitat types on Cape Sada. However, the use of other microhabitats

differed among seasons. For example, angonoka were encountered more frequently in

the open microhabitat during the wet season, because in general tortoises were more

vagrant in wet season when environmental temperatures were high and moisture was not

limited. Tortoises were observed feeding on the few herbs and grasses available in open

areas during the wet season and often were observed passing through these areas. In the

dry season, angonoka used the patches of grass more often than in the wet season. This

difference may relate to the tortoise's ability to maintain an optimal body temperature by

sheltering under grasses during the cool, but sunny, dry season. Giant tortoises

(Dipsochelys dussumieri) and box turtles (Terrapene carolina) also exhibit seasonal

shifts in habitat use related to thermoregulation (Stickel 1950; Reagan 1974; Swingland

and Lessells 1978; Gibson and Hamilton 1983; Dodd et al. 1994).

Although no statistical difference in microhabitat use was detected between adult

male, adult female, and juvenile angonoka in this study, adults appeared to use bamboo

more often in October and November (late dry season). This coincides with breeding

activity (Chapter 4) and may indicate that the bamboo habitat is important as a staging

area for courtship. Differences in habitat use may exist between different angonoka

populations. For example, female angonoka in a population west of Baly Bay used grass

more often in the wet season than in the dry season. Females in the west Baly Bay

population often nested in grass habitat (M. Pedrono, pers. com.). The grass habitat used

by female angonoka in the western population was more similar to true savanna. Female

tortoises on Cape Sada tended to nest in the scrub-shrub or bamboo microhabitat, and

only rarely in open, unvegetated areas (Chapter 4).

The findings of this study regarding seasonal shifts in microhabitat use by the

angonoka underscore the importance of maintaining the diversity of vegetation types

within the bamboo-scrub ecosystem. Furthermore, grasses and vines in frequently

burned angonoka habitat replace bamboo and shrubs, the two primary components of this

ecosystem. The reduction of these components of the ecosystem may severely reduce the

suitability of the habitat for tortoises. The loss of bamboo and shrubs probably results in

an increase in air and substrate temperatures and a decrease in relative humidity.

The role, if any, of natural bush fires in the maintenance of the bamboo-scrub

habitat of the angonoka is not understood. Only a few studies have documented the

effects of fire on tortoise species and their habitat. Natural fires ignited by lightning

promote species diversity in the sandhill habitat of the gopher tortoise (Gopherus

polyphemus) in the southeastern United States (Laessle 1958). Gopher tortoises have

adapted to a fire-maintained ecosystem by using subterranean burrows for refuge. The

evergreen shrub/grassland (renosterveld) habitat of the South African geometric tortoise

(Psammobates geometricus) also is fire-maintained (Baard 1995). However, frequent

wildfires degrade the renosterveld ecosystem and unseasonable fires can kill adult and

juvenile tortoises (Baard 1993). Catastrophic effects of anthropogenic fires also have

been documented in Hermann's tortoise (Testudo hermanni) in Greece (Hailey and

Goutner 1991).

In western Madagascar, natural fires would be expected to take place in the wet

season, when lightning storms occur. However, anthropogenic fires typically are set in

the dry season and probably function quite differently than natural fires. Natural fires

occur sporadically and undoubtedly with much lower frequency than manmade brush

fires. Furthermore, dry season fires are probably more intense than wet season fires

because fuel loads are greater. Therefore, intense dry season fires could be expected to

have a great impact on tortoises. It seems likely that manmade fires, by altering the

bamboo-scrub habitat or killing tortoises outright, are a major threat to remaining

angonoka populations.

Home ranges of adult angonoka overlapped considerably, particularly during the

wet season. However, the ranges of the three adult males only overlapped at the

periphery, and it appears that male angonoka may exhibit a dominance hierarchy. More

information is needed regarding the social structure of angonoka populations. Questions

concerning spatial needs of the angonoka are particularly important with only 4,000-

6,000 ha of bamboo-scrub habitat remaining. Long-term monitoring of the Cape Sada

population is needed to determine whether the population is increasing (despite high

juvenile mortality), decreasing, or stable (at equilibrium density).

As mentioned previously, angonoka are being bred in captivity with the intent of

augmenting wild populations. I believe that habitat protection and restoration, rather than

augmentation, should be a conservation priority for this species at this time. However,

unoccupied tracts of bamboo-scrub may be suitable for reintroduction of captive-born

tortoises in the future. In the meantime, future research should focus on identifying

differences in the microclimate of scrub-shrub, bamboo thickets, grass and open areas

that may be important to the angonoka.

Table 3-1. Radiotracking information for 13 angonoka (Geochelone yniphora) on Cape
Sada, Madagascar. Data were collected from October 1993 through June 1995.






2/95 6/95
11/93 6/95
10/93 -6/95
10/94 6/95
10/94 -6/95

7 430 10/93 6/95 278
13 398 10/93 6/95 288
54 454 12/93 6/95 229
Mean: 427
1 376 10/93 -6/95 303
3 329 10/93 6/95 301
9 344 10/93 6/95 285
12 361 10/93 6/95 292
19 399 11/93 6/95 264
Mean: 362




Table 3-2. Mean home range size (in hectares) of 13 radio-instrumented angonoka
(Geochelone yniphora) on Cape Sada, Madagascar. Home range size was estimated
using minimum convex polygon method (MCP) (Mohr, 1947) and kernel area (KA)
(Worton, 1989). Data were grouped into two seasons for analysis (November April =
wet season; May October = dry season). For each category the mean is followed by 1
SD and the range.



Juveniles 1.4 1.1 0.6 0.5 2.2 1.5 1.8 1.8
(N=5) (0.1- 2.5) (0.0- 1.3) (1.0-4.8) (0.6-5.0)

Males 9.2 4.6 6.6 4.5 38.5 7.0 21.1 6.4
(N=3) (4.6 -13.9) (1.4 -9.8) (31.6 -45.5) (6.4 28.4)

Females 8.7 9.8 3.6 5.0 17.7 18.2 12.2 9.4
(N=5) (1.0- 25.6) (0.5 12.5) (10.6-42.0) (4.7-28.5)

Table 3-3. Analysis of Covariance of the effects of season and habitat where a tortoise
was observed, and sex/life stage (males, females, and juveniles) on percent of
observations of angonoka tortoises on Cape Sada, Madagascar. *P<0.05 indicates values
that were not uniformly distributed.


Habitat 2 38.42 0.0001*
Season 1 0.03 0.8684
Sex 2 0.82 0.4436
Habitat*Season 2 3.21 0.0447*
Habitat*Sex 4 1.78 0.1390

Table 3-4. Results of Tukey's studentized range (HSD) test comparing mean percent of
observations ofangonoka tortoises by season and habitat type (N=18) on Cape Sada,
Madagascar. Habitat types included bamboo (BA), open (OP) and grass (GS). MSD=
minimum significant difference, *= significantly different (P< 0.05).








GS Wet







2.81 1.89

* Females U Males A Juveniles





Carapace Length (mm)

Figure 3-1. Wet season (November April) home range size as related to carapace length
(in mm) for 13 angonoka tortoises (Geochelone yniphora) at Cape Sada, Madagascar.
Home range was calculated using Kernal Area method (Worton, 1989).

N 30

^' 25

, 20

b 15


S 5



,, F19

M 13

200 m

Figure 3-2. The spatial distribution of dry season home ranges for eight adult angonoka
(Geochelone yniphora) at Cape Sada, Madagascar. Dry season extended from May
through October. Home ranges were calculated using the kernel area method (Worton


M 54



N t
F 19 )

F 9
M 13
M 54

\ M M1F3

M *3

200 m

Figure 3-3. Spatial distribution of wet season home ranges for eight adult angonoka
(Geochelone yniphora) at Cape Sada, Madagascar. Wet season extended from November
through April. Home ranges were calculated using the kernel area method (Worton

-oJ 44A



J 83



100 m

Figure 3-4. Spatial distribution of dry season home ranges for five juvenile angonoka
(Geochelone yniphora) at Cape Sada, Madagascar. Dry season extended from May
through October. Home ranges were calculated using the kernel area method (Worton

J 44



J 88




100 m

Figure 3-5. Spatial distribution of wet season home ranges for five juvenile angonoka
(Geochelone yniphora) at Cape Sada, Madagascar. Wet season extended from November
through April. Home ranges were calculated using the kernel area method (Worton







Figure 3-6. Monthly patterns in microhabitat use in three adult male angonoka tortoises
(Geochelone yniphora) at Cape Sada, Madagascar. Numbers represent the percent of
observations of radio-instrumented tortoises in scrub-shrub (SS), bamboo (BA), grass
(GS), and open (OP) microhabitats. Data were collected from October 1993 through
June 1995.




00 100
" 90
II 80
t 40
8 30
o 20
C- 10



Figure 3-7. Monthly patterns in microhabitat use in five adult female angonoka tortoises
(Geochelone yniphora) at Cape Sada, Madagascar. Numbers represent the percent of
observations of radio-instrumented tortoises in scrub-shrub (SS), bamboo (BA), grass
(GS), and open (OP) microhabitats. Data were collected from October 1993 through
June 1995.


40 -
- 530
0 20


J A S 0


Figure 3-8. Monthly patterns in microhabitat use in five juvenile angonoka tortoises
(Geochelone yniphora) at Cape Sada, Madagascar. Numbers represent the percent of
observations of radio-instrumented tortoises in scrub-shrub (SS), bamboo (BA), grass
(GS), and open (OP) microhabitats. Data were collected from October 1993 through
June 1995.




Activity patterns of terrestrial turtles are directly related to environmental

temperature and moisture (Gourley 1979). Temperature exerts a strong influence on

activity in a temperate climate (Oliver 1955; Carpenter 1957; Sexton 1959; Gibbons

1970; Douglass and Layne 1978; Lovich 1988; Lovich 1990; Lovich et al. 1992; Gibbons

et al. 1990; Claussen et al. 1991; Mushinsky and Wilson 1992; Dodd et al. 1994),

whereas in a tropical climate, moisture often has a strong influence on activity

(Woodbury and Hardy 1948; Rose and Judd 1982; Swingland and Lessells 1979). In the

tropical climate of western Madagascar, activity of the angonoka appeared to be closely

tied to seasonal rainfall (Juvik et al. 1981). More than 90 % of all precipitation in the

region occurs in the summer, from November through March (Donque 1972). The small

number of observations of angonoka during winter surveys (May through October)

supported the assumption that this species was dormant during the dry season (Juvik et al.

1981; Juvik et al. 1997). However, seasonal activity patterns for this species had not

been examined in detail.

Past descriptions of the behavior of angonoka were based on anecdotal

observations of wild tortoises (Juvik et al. 1981; Juvik et al. 1997) and on observations of

captive individuals (McKeown et al. 1982; Reid 1995). Mating and nesting had never

been observed in the wild, and it was thought that population levels were depleted to a

level where little or no reproduction was taking place (Juvik et al. 1981; Burke 1990). In

captivity, male angonoka exhibited considerable agonistic behavior during courtship

(McKeown et al. 1982; D. Reid, pers. com.). It was not known whether this behavior was

critical to successful reproduction in the angonoka and might be affected by the low

population densities that exist today. Predation of eggs and young tortoises by the non-

native African bush pig was considered one of the primary threats to the survival of the

angonoka (Juvik et al. 1981; Curl et al. 1985; Durrell et al. 1989b). Feral pigs are known

to destroy nests of birds and reptiles in other parts of the world (MacFarland and Reeder

1975; Coblentz and Baber 1987).

The objectives of this portion of my study were as follows:

1. To identify and describe seasonal activity patterns in the angonoka.

2. To determine whether angonoka were mating and producing young in the

Cape Sada population, and if so, to describe mating behavior in the wild.

3. To assess the impact of African bush pigs on angonoka nests.


Thirteen tortoises (3 males, 5 females, and 5 juveniles with 8-10 shell growth

rings) were radio-tagged and followed for periods ranging from 121 to 630 days (see

methods in Chapter 3). Tortoises were located by direct observation on seven mornings

(between 0600-1200 hrs) and seven afternoons (between 1400-1800 hrs) per month,

weather permitting. Tortoises also were found during monthly visual surveys (Chapter

2). Each time a tortoise was located the date, time, weather (i.e. clear, partly cloudy,

cloudy, rain), air temperature, substrate temperature, relative humidity, and behavior

were recorded. Minimum and maximum air temperature and rainfall were recorded daily

at the Cape Sada camp.

Tortoise behaviors were categorized as resting with the head retracted (R), resting

with head out (R/HO), walking (W), feeding (F), courtship (male-female encounters)

(CS), combat (male-male interactions)(CO), and nesting (N). Among these behaviors, R

was considered inactive, whereas R/HO, W, F, CS, CO and N were considered active.

Data were grouped into two seasons for analysis (November April = wet season; May -

October = dry season). Correspondence analysis was used to graphically depict

associations among the variables sex, behavior, year, and season (SAS 1992).

In the 1994 nesting season, thread trailers (Breder 1927; Stickel 1950) were used

to follow the five radio-tagged females. The trailers consisted of spools of hip-chain

replacement thread (Forestry Suppliers, Inc., Jackson, MS) that were encapsulated in 125

ml Nalgene bottles. The bottles were mounted with duct tape on the posterior of the

tortoise's carapace, and the end of the thread was tied to a nearby tree or shrub. After 24

hours, the thread trail was searched for evidence of fresh digging that might indicate that

the female had nested. The spent thread was collected and measured to determine the

actual distance traveled by the tortoise in a 24-hour period. Nests were left undisturbed

but checked periodically for signs of disturbance by predators.

Beginning in February 1995, I radio radio-located females daily in order to

observe nesting behavior. Nesting females were observed from a distance until a nest

had been completed. Nests were then excavated and eggs were counted. Care was taken

not to turn or rotate eggs, and once the eggs had been counted the nest was re-buried. In

mid-October, near the end of incubation, chicken wire cages (3-cm mesh) were placed

above the nests so that emerging hatchlings could be caught and marked. Cages were

anchored with tent stakes and a corner of each cage was shaded with a white cloth to

prevent hatchlings from overheating. Cages were checked daily for emerging hatchlings

from mid-October onward. Hatchlings were measured (see methods in Chapter 5),

marked by painting a unique series of marginal scutes, and released at the nest site.


Seasonal Activity Patterns. Monthly rainfall and air temperatures at the study site

are presented in Figure 4-1. More than 70 % of the annual rainfall occurred in January

and February for both years of the study, whereas less than 20 mm of rainfall (< 2 % of

total) was recorded during the dry season. During the wet season (November April),

the mean minimum temperature was 24.2 C as compared to 20.5 C in the dry season

(May October). In the wet season, the average maximum temperature was 34.0 C,

whereas in dry season the average maximum was 32.8 C

In most observations, radio-instrumented angonoka were inactive, resting with

their head-in (R/HI) under vegetation (Figure 4-2, 4-3, and 4-4). When behaviors were

categorized as either "inactive" or "active", seasonal differences in activity levels were

evident (Figure 4-5). These differences were statistically significant for all sex/life stages

(Table 4-1). Tortoises often stopped walking or feeding when they were approached.

While this probably resulted in an under-representation of feeding and walking

observations, tortoises rarely retracted their head when disturbed, and were still

categorized as resting-head-out, an "active" behavior.

Active tortoises were observed throughout the day (0600 hrs 1700 hrs);

however, observations were most frequent from 1100 hrs to 1200 hrs (Figure 4-6). No

nocturnal activity was observed in this study. Tortoises were active at a wide range of air

temperatures (22-450C) and substrate temperatures (21-44C). Tortoises were active

when the relative humidity was 33-98%.

Results of the correspondence analysis revealed that variables such as walking

and resting/head-out (active behaviors) were closely associated with the wet season,

whereas, resting head-in (inactive) was more closely associated with the dry season.

Correspondence analysis represents each season, year, sex/life stage, and behavior on a

plot (Figure 4-7), where the relative positions of the points indicate similarities and

differences. On the plot, the cosine of the angle between two points indicates the affinity

of those two categories for one another. A large, cosine (small angle) indicates high


In addition to observations of the behavior of radio-instrumented tortoises,

tortoises were incidentally encountered during the course of timed visual surveys and

other field tasks (322 observations). Most of the tortoises encountered during visual

surveys were active. Despite the experience of surveyors, tortoises were very difficult to

find when they were inactive and resting under vegetation. Overall, incidental

observations of tortoises reflected activity patterns similar to those of radio-instrumented

tortoises. Most observations occurred during the wet season (76 %) when tortoises were

most often out walking, feeding, mating, or nesting.

Mating Behavior. Male angonoka were most active from November through

January, which corresponded roughly with the reproductive season (Figure 4-2).

Courtship was observed on 8 occasions over the two-year period. All observations

occurred in November and December (Table 4-2). Male-male interactions consisted of

one male circling, ramming, and chasing the other tortoise, followed by the retreat of one

of the males. Male-male interactions lasted from 5 to 25 minutes. Male-female

encounters were observed five times and observations ranged from 35 to 58 minutes in

length. The general pattern of behaviors recorded between male and female tortoises

included the male pursuing the female, bumping the rear of the carapace with the gular,

and circling the female repeatedly before mounting her. On two occasions the male

tortoise actually overturned the female using the gular. One of the overturned females

was able to right herself, whereas the male righted the other as he continued ramming her

with his gular. During copulation, males emitted a wheezing vocalization. The smallest

male and female observed mating were 398 and 338 mm CL, respectively. The largest

male marked in the Cape Sada population and the largest female were observed mating.

Nesting Behavior. Female angonoka nested from February through May and one

female attempted to nest in June (Table 4-3). In October 1993, a nest containing 5 eggs

was found during visual surveys. The nest was very shallow and one egg was visible at

the soil surface. Three nests were located using thread trailers during the 1994 nesting

season. Nests were well hidden and difficult to locate unless the tortoise had buried part

of the thread in the egg chamber. By increasing the frequency of radio-location during

the 1995 nesting season, 8 nests were discovered. On two occasions females produced

multiple clutches within a single season. The maximum number of clutches per season

was three. The interesting interval ranged from 22 to 30 days (n=4).

Tortoises often attempted to nest for several days before successfully depositing

eggs. During a typical failed nest attempt, the female excavated an egg chamber but

abandoned it when she encountered obstructions in the soil such as rocks or roots. One

female attempted to nest on 11 occasions over a 10 day period before depositing eggs

(See Table 4-3).

Most nesting activity occurred in scrub-shrub microhabitat (71.7 %) followed by

bamboo (23.9 %), open (2.2 %) and grass habitat (2.2 %). Females nested in the four

microhabitats relative to their availability (X2 = 9.57, df= 4, P>0.05). They typically

nested at the base of a shrub or bamboo plant; however, one nest was constructed in an

open sandy area with no vegetation. Another nest was deposited in an area within the

scrub-shrub microhabitat that had been heavily disturbed by pigs.

Peak nesting activity occurred in April through June in 1994, and in March and

April in 1995 (Figure 4-8). Nesting activity was recorded from 0736-1645 hrs, but most

activity took place in the morning between 0800 and 1100 hrs (Figure 4-9). The thread

trailer method provided information on the length of nesting forays of individual females.

The mean distance traveled by females in the 24-hour period before nesting was 195.08

m (n= 12; range = 25.0 626.5) as compared to only 31.71 m in the 24-hour period

following nesting (n= 5; range = 0.0 57.3). The maximum distance traveled by a

female on a nesting foray was 626.5 m in a 12-hour period.

Clutch size for the 12 nests ranged from 2-5 eggs with a mean of 3.360.92

(Table 4-3). The smallest female known to have nested was 345 mm CL. Clutch size

varied positively with carapace length, although the relationship was not statistically

significant (F= 3.17; df= 10; P= 0.11). Incubation ranged from 220 to 244 days and

hatchlings emerged between 5 November and 18 December. A hatchling was discovered

within a nest in late October 1994, suggesting that angonoka may exhibit delayed

emergence from the nest. Hatchling emergence appeared to coincide with the onset of

seasonal rains and if undisturbed, this hatchling may have remained in the nest chamber.

Forty-one percent of all eggs hatched successfully (Table 4-3), but hatching success

varied a great deal among nests. Only one nest had 100 % hatching success and 4 of the

12 nests (33 %) produced no hatchlings. The overall hatching success at the captive

breeding center was 42 % and hatching success among individual nests was quite

variable (Reid 1995). No nest predation was observed in this study. However, one nest

was destroyed by African bush pigs following this study (M. Pedrono, pers. com.).

Hatchling and Juvenile Behavior. Twenty hatchling angonoka were observed

walking and feeding from January through April (Figure 4-10). Incidental observations

of active small juvenile tortoises (<85 mm CL) peaked in February through March.

Hatchlings and small juveniles were most often observed walking or feeding in the open

microhabitat (80% of observations) between 0735 and 1727 hrs.

Feeding Behavior. Tortoises were observed feeding on 78 occasions (2.5 % of all

observations). All feeding behavior was observed from October through May, with most

observations (n= 15) occurring in April. Adults were observed feeding in the open

microhabitat more often than would be expected based on habitat availability (X2=

102.91, df- 4, P< 0.0001) (Figure 4-11). They consumed small herbs and forbs typical

of these areas. Juveniles fed in the bamboo microhabitat more often than expected.

Although food plants were not identified to species in this study, in most observations (44

%), tortoises fed upon herbs, forbs and shrubs (Bauhinia spp. and Terminalia spp.) rather

than grasses (15.6 %). Tortoises were never observed feeding on live bamboo; however,

on several occasions they consumed leaf litter that included dead bamboo leaves (15.6

%). Angonoka were observed feeding on dried carnivore feces (12.5 % of observations)

and African bush pig droppings (12.5 %). Of the three life stages, juveniles most often

were observed consuming feces (87 % of observations).


As expected, given the highly seasonal environment in the Baly Bay region,

angonoka exhibited seasonal differences in activity patterns. Tortoises were significantly

more active during the wet season than during the dry season. Seasonal differences in

activity patterns also were evident in tortoises observed during visual surveys (see

Chapter 2). These results suggest that whenever possible, tortoise surveys should be

concentrated in December, January and February. Activities such as mating and nesting

occurred only in wet season months (Figure 4-12). Hatchlings emerged with the onset of

seasonal rains. Small juveniles were active and highly visible from February through

March. These tortoises were undoubtedly vulnerable to diurnal predators at this time.

The primary diurnal predators in the region are the Madagascar buzzard and yellow-

billed kite.

Although mating was observed on only a few occasions, the behavior patterns

were similar to those described for captive angonoka and radiated tortoises (Geochelone

radiata) (Auffenberg 1978; McKeown et al. 1982; Reid 1989). Male angonoka exhibited

considerable agonistic behavior in competition for females. In addition, males made long

forays in the early wet season, presumably in search of mates (Chapter 3). Further

description of the mating system of the angonoka is needed. Home ranges of adult males

and females overlapped considerably during the wet season (Chapter 3). However, the

ranges of adult males overlapped very little. It will be important to examine the degree of

overlap among all males in the population in order to determine if male angonoka exhibit

a dominance hierarchy (Harless 1979).

The smallest reproductively active male and female observed in this study were

398 and 329 mm CL, respectively. However, it is likely that angonoka are sexually

mature at a smaller size than was detected based solely on observations of mating.

Secondary sexual characteristics could generally be distinguished in tortoises that were

greater than 300 mm CL or with greater than 15 shell growth rings (Chapter 5).

Demographic variables such as size (and age) at first reproduction in the angonoka

should be determined in the future.

The maximum number of clutches produced by a female in a single season was

three, which was considerably less than that recorded in captive angonoka. At the

breeding center in Madagascar, females nested up to 7 times in a season and the nesting

period extended from January through July (Reid 1995). This difference could be an

artifact of the field sampling methodology, because it is likely that some nesting activity

of radio-instrumented females was not detected. However, captive females receive a

different diet than wild tortoises, and are given supplemental calcium during nesting

season (Reid 1995). Differences in the nutritional status between wild and captive

tortoises could explain the apparent difference in reproductive output.

Reproductive output in chelonians is a product of clutch size and frequency and

egg size. Two general reproductive patterns have been described for aquatic or semi-

terrestrial chelonians (Moll 1979). The first pattern, typical of sea turtles, is that of

producing multiple clutches of many small eggs in a discreet nesting season. The second

pattern, typified in tropical mud turtles and box turtles, consists of small clutches of large

eggs with acyclic or continuous nesting. In testudinids, larger species (e.g., Geochelone)

tend to lay larger clutches, whereas smaller species tend to be more specialized, and often

produce very small clutches. The angonoka is a large species that produces small

clutches and appears to be an exception to this pattern. The largest reported clutch size in

the angonoka was 6 eggs (Pedrono 1997; Reid 1995) and the mean clutch size in this

study was 3.2+0.92 eggs. In contrast, G. radiata produces much larger clutches with 3-

12 eggs (Zovickian 1973). It will be important to examine reproductive output in wild

angonoka more closely in the future, and to look for differences among the 5 populations.

Incubation length in this study (220 days) was similar to that reported for nests of

captive angonoka (215 days)(Reid 1995). No information on incubation of wild radiated

tortoises is available, however, in the average incubation period for captives of this

species ranged from 155-230 days depending on incubation temperature (Zovickian

1973; Burchfield 1975).

Although hatching success of the wild nests in this study (41 %) was similar to

that reported for in-situ nests at the captive breeding center (42 %) (Reid 1995), it was

low compared to other species of tortoise (Landers et al. 1980; Smith 1995). It is not

known whether the unhatched eggs were infertile or failed to hatch because of a

developmental problem. Eggs seldom were removed from the nest as they were counted,

and when it was necessary to handle the eggs, care was taken not to turn or roll them.

Therefore I feel it is unlikely that handling had an effect on hatching success in this


If female angonoka produce 3.36 eggs per clutch and three clutches per year (a

conservative estimate based on this study and captive angonoka), the 27 adult female

angonoka on Cape Sada would be expected to produce a total of 272.2 eggs per year.

This calculation is based on the unproven assumption that all females nest annually. If

the 41 % hatching success observed in this study is typical for this species, then the adult

population ofangonoka on Sada would be expected to produce 112 offspring per year.

Although nest predation was not observed in this study, the predation rate on small

juveniles may be high (see Chapter 2). High mortality in the early life stages of

terrestrial turtles is not uncommon (Ehrenfeld 1979; Diemer 1986). Long-term data are

needed to evaluate survivorship of juvenile angonoka in order to determine whether

recruitment to adulthood is occurring in this population. In a relatively stable

environment (particularly with no removal of adults from the population), a long-lived

species such as the angonoka may be capable of recovering from past exploitation

without augmentation with captive born juveniles. However, loss of breeding adults

could have catastrophic consequences.

Although the number of nests monitored in this study was limited, the lack of nest

predation was somewhat unexpected. High nest predation rates have been reported in

other chelonians (Wilhoft et al. 1979; Landers et al. 1980; Seigel 1980; Congdon et al.

1987; Hailey and Loumbourdis 1990). Despite a great deal of bush pig activity on Cape

Sada, many tortoise nests clearly are surviving to produce hatchlings. Information is

needed regarding nest predation at other angonoka populations. I suspect that pigs may

exert a more subtle effect on the angonoka by altering the habitat and soils. Bush pig

rooting was extensive on Cape Sada throughout the course of this study. Cattle also are

present on Cape Sada and may be altering the habitat and soil structure. In heavily

grazed areas in Argentina, cattle compact the soil above tortoise nests and injure young

tortoises (Waller and Micucci 1997). The potential impacts of bush pigs and cattle on the

angonoka should be investigated. Cattle are an important part of the Sakalava culture

(Durbin et al. 1996) and it will be difficult to resolve problems involving interactions

between cattle and tortoises. If bush pigs are proven to negatively impact tortoises,

control of this species will present a different problem. Local people are reluctant to

handle bush pigs for religious reasons. If necessary, it may be possible to hire outside

contractors to remove pigs.

Table 4-1. Percent of observations where tortoises were active (walking, feeding, combat,
courtship, or nesting). indicates P< 0.0001.



Sex/Life Stage




All Tortoises

% Active Observations

;eason Dry Season

2.3 17.7







Chi-square Value





Table 4-2. Information on breeding behavior in the angonoka (Geochelone yniphora) on
Cape Sada. Data were collected from October 1993 through June 1995. *nd= no data.





0825 5.18

47 (402)









52 (406)


58 (438)


59 (456)

47 (402)

9 (344)

54 (454)


13 (398)

12 (361)



59 (456)

43 (405)








Table 4-3. Nesting activity in the angonoka (Geochelone yniphora) during the 1994 and
1995 nesting season at Cape Sada, Madagascar.

1994 Feb 22 0742 SS 3 Attempt only

Mar 2 0938 BA 3 Attempt only

Mar 4 0736 SS 6 Attempt only

Apr 8 AM SS 12 Attempt only

Apr 9 AM SS 12 4 eggs

Apr 25 0855 BA 19 Attempt only

Apr 25 1540 BA 19 3 eggs

Apr 26-30 ND SS 9 Several attempts

May 12 0829 SS 12 Attempt only

May 12 0937 SS 1 Attempt only

May 13 0908 OP 1 3 eggs

Jun 7 0911 BA 9 Attempt only

Jun 15-25 ND SS 9 11 attempts

1995 Feb 24 1005 BA 19 4 eggs

Mar 18 1200 SS 19 4 eggs

Mar 20 0921 BA 12 3 eggs

Apr 13 0859 SS 19 4 eggs

Apr 19 0845 BA 12 2 eggs

Table 4-4. Angonoka (Geochelone yniphora) nest data for the 1994 and
seasons on Cape Sada, in western Madagascar.

1995 nesting


00 nd 12/11/94

12 4/9/94 10/26/94;

19 4/25/94 12/16/94:
















220: 244

235; 237

264; 270





































-- Min Temp Max Temp Rainfall










Figure 4-1. Monthly rainfall and temperature data collected at Cape Sada in western
Madagascar, form October 1993 through May 1995.



I 15






8 80

J 60

O 40





Figure 4-2. Monthly behavior patterns for three radio-instrumented adult male angonoka
(Geochelone yniphora) on Cape Sada, Madagascar. R= resting head-in; R/HO= resting
head-out; W= walking; F= feeding; CS= courtship; and CO= male-male combat.





Figure 4-3. Monthly behavior patterns for five radio-instrumented adult female angonoka
(Geochelone yniphora) on Cape Sada, Madagascar. R= resting head-in; R/HO= resting
head-out; W= walking; F= feeding; CS= courtship; and N= nesting.


S50 -
O 40
0 30
10 t



Figure 4-4. Monthly behavior patterns for five radio-instrumented juvenile angonoka
(Geocheloneyniphora) on Cape Sada, Madagascar. R= resting head-in; R/HO= resting
head-out; W= walking; F= feeding.

* Males 1 Females E0 Juveniles

40 -




'93-94 Wet

'94-95 Dry

'94-95 Wet

'95-96 Dry


Figure 4-5. Seasonal activity patterns of 13 radio-instrumented angonoka (Geochelone
yniphora) on Cape Sada, Madagascar. Wet season extended from November through
April and dry season from May through October.

2 40-
, 30
I 20-



Time of Day

Figure 4-6. Percent of observations of active angonoka (Geochelone yniphora) by time of
day on Cape Sada, Madagascar.





*4 Fe






Figure 4-7. Results of correspondence analysis depicting the relationship among
behavior, season, sex/life stage, and year in angonoka (Geochelone yniphora). R= resting
head-in; R/HO= resting head-out; W= walking; F= feeding; CS= courtship; and N=
nesting; WS= wet season; DS= dry season; M= male, Fe= female, J= juvenile.





* Nest Attempts
9 Nests



Figure 4-8. Nesting activity by month in the angonoka (Geochelone yniphora) on Cape
Sada, Madagascar.


_ 25

0 15



Time of Day

Figure 4-9. Nesting activity by time of day in the angonoka (Geochelone yniphora) on
Cape Sada, Madagascar.

IS Hatchlings E Juveniles

- 10



1- 4

Z 2




Figure 4-10. Monthly observations of hatchling and juvenile angonoka (Geochelone
yniphora) on Cape Sada, Madagascar.

* Males 15 Females l Juveniles

Microhabitat Type

Figure 4-11. Feeding observations by microhabitat type for angonoka (Geochelone
yniphora) on Cape Sada, Madagascar. SS= scrub-shrub; BA= bamboo; GS= grasses;
OP= open.




- 1994-95




Figure 4-12. Behaviors of the angonoka (Geochelone yniphora) as related to rainfall on
Cape Sada, Madagascar.




Early descriptions of Geochelone yniphora were based on very few specimens

(Vaillant 1885; Vaillant 1889; Siebenrock 1903; Angel 1931). Although later reports

were more comprehensive (Juvik et al. 1981), small sample sizes still precluded a

detailed description of the species. Similarities between G. yniphora and the closely

related radiated tortoise (G. radiata) were noted. However, the angonoka is the larger of

the two species, is somewhat less colorful, and has an elongate, undivided gular

projection. The angonoka received its English common name, the ploughshare tortoise,

from the long, recurved gular.

Based on early descriptions, adult angonoka appeared to exhibit sexual

dimorphism. This phenomenon is well documented in the general literature (Carr 1952;

Graham 1979; Pritchard 1979; Ernst and Barbour 1989). Divergent characters such as

body size, shell morphology, and tail size, often are used in determining the sex of turtles.

The posterior portion of the plastron in males of some species is concave, which aids in

stabilizing the male as he mounts a female prior to copulation. This phenomenon is most

pronounced in highly domed, terrestrial species such as the red-footed tortoise

(Geochelone carbonaria), yellow-footed tortoise (Geochelone denticulata), radiated

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd