Behavioural ecology of the red collared brown lemur (Eulemur collaris): comparison between groups living in well preserved and degraded littoral forest fragments, in South-eastern Madagascar

(1)

i

Università Degli Studi Di Pisa

Facoltà Di Scienze Matematiche Fisiche e Naturali Corso Di Laurea Specialistica In Biodiversità ed Evoluzione

Tesi Di Laurea Specialistica

“Bahavioural ecology of the red collared brown

lemur (Eulemur collaris): comparison between

groups living in well preserved and degraded littoral

forest fragments, in South-eastern Madagascar”

Relatori:

Prof.ssa Silvana Borgognini-Tarli Dott. Giuseppe Donati

Candidata: Valentina Serra

(2)

ii To my family and all my friends, especially Marta.

(3)

iii “Alina anaty ala:

kintana be, volana kely raiky, lanitra akaiky. Biby matory milamina ohatra ny foko.”

“La notte nella foresta:

grandi le stelle, una piccola luna, il cielo vicino. Gli animali dormono sereni, come il mio cuore.”

“The night in the forest: big stars, a small moon, the sky so near. The animals sleep peaceful, like my heart.”

(4)

iv

GENERAL INTRODUCTION

The situation in Madagascar

adagascar, located some 400 km east of Africa, is the world’s fourth largest island, with an area of about 587,000 km2. Biologically it is widely regarded as one of the richest areas on Earth containing nearly 8000 endemic species of flowering plants (Green & Sussman, 1990).

Madagascar is one of the primary countries where the principles of the Biodiversity Convention and the IUCN-ICMM guidelines apply. In fact, this country has been identified as one of the world’s most important biodiversity hotspots (Mittermeier et al., 2004). Biodiversity hotspots are defined as areas with very high species richness and endemism, which have lost more than 75% of their original habitats due to human activities in recent times (Fig 1.1).

M

Fig. 1.1. Map of distribution of rain forest in Eastern Madagascar throughout time (Green & Sussman,

(8)

2

Numerous habitats of Madagascar have been degraded since the arrival of humans some 1500-2000 years ago: over 80% of the island has already been stripped of its native vegetation cover (Dupuy & Moat, 1998) and extinctions of species of large mammals and birds have been severe. At current deforestation rates it is estimated that in 2025 forest will only remain on the steepest slopes and in remote areas and nature reserves (Green & Sussman, 1990). The main cause of forest loss in Madagascar is due to subsistence needs of the population and cutting for fuel. Cultivation areas are obtained by the practice of tavy (shifting slash-and-burn cultivation), which is often accompanied by bushfires. These bushfires are frequently uncontrolled, resulting in further deforestation. Tavy contributes directly to habitat loss, edge effects and soil sterilization. After clearance, grasslands develop in early successional stages and are later replaced by fire-resistant shrublands dominated by invasive heath shrubs Erica spp. (Lewis Environmental Consultants, 1992). Logging is another threat: people cut mainly large, mature trees for fuelwood, constructions, tools and illegal export of valuable woods such as rose wood or ebony.

On the socio-economic side, Madagascar ranges among the poorest countries on earth. In 2008, the population of Madagascar was estimated at around 19 million people (INSTAT/Direction de la Démographie et des Statistiques Sociales, 2010), 85% living on less than two dollars per day. Average annual salary in 2005 was

991,000 Ariary (354.36 Euro or 469.56 Dollars)

(INSTAT/DSM/EPM 2005).

The Institut des Statistiques de Madagascar lists a Gross Domestic Product (GDP) with a growth rate of -3.7% in 2009, and an

(9)

3

(www.instat.mg/MADA/indic.htm; accessed on December, 13 2011). The poverty rate rose from 70% in 1993 to 76.5% in 2010 and the unemployment rate was 3.8% (EPM 2010).

Approximately 80% of the population is engaged in subsistence agriculture and levels of health care and education are low. Only 54% of the population living in urban areas and only 4% in rural areas has access to drinkable water. Average life expectancy is 52 years and the infant mortality rate is 89 per 1000. Literacy is estimated at 46% (CRD 2003).

In this critical socio-economic situation it is very clear how difficult and controversial was to carry on an effective conservation of Malagasy environments, already deeply compromised. The current protected area network, managed by the Madagascar National Parks Association (PNM-ANGAP), covers some 1.7 million hectares or about 3% of the country (Mittermeier et al., 2005). The network includes three types of protected areas: Strict Nature Reserves (IUCN category Ia), National Parks (IUCN category II) and Wildlife Reserves (IUCN category IV).

At the 2003 IUCN World Parks Congress in Durban, the Malagasy President at the time, Marc Ravalomanana, announced a bold initiative to more than triple the area under protection from approximately 17,000 km² to over 60,000 km² (from 3% to 10% of Madagascar's area). But when the military forces backed Andry Rajoelina's ascension in 2009 March 17, and he became “President of the High Transitional Authority of Madagascar”, the pro-environmental policy had an abrupt stop.

At the moment, to save what remains of Malagasy ecosystems would take decisive and sudden action, but the extreme poverty of the country, the lack of education and awareness, as well as of a strong policy makes it extremely difficult to be realized.

(10)

4

The role of QMM

QIT Madagascar Minerals (QMM) is a company jointly owned by Rio Tinto plc (80%), UK and by the Malagasy State (20%) represented by the Office des Mines Nationales et des Industries Stratégiques de Madagascar (OMNIS).

QMM has built a mineral sands mining operation near Fort-Dauphin, in the South-eastern Madagascar, to extract ilmenite and zircon from heavy mineral sands over an area of about 6,000 hectares along the coast. The exploitation operations will be conducted over the next 40 years.

The exploration of the Anosy region by QMM started in the late 1980, togheter with social and environmental studies.

QMM discovered an economically viable ore deposit in the Fort Dauphin (Tolagnaro) area, rich of heavy mineral sands, which are a source of titanium dioxide, principally in the form of ilmenite and rutile. Part of the proposed mining areas consist of heavily degraded ecosystems, although, major sediments are located underneath some of the last remnant littoral forest in southeastern Madagascar, at Mandena, Sainte Luce, and Petriky (Vincelette et al., 2003).

In 1998 a legal and fiscal framework agreement was concluded between QMM and the Government of Madagascar, that was subsequently ratified by the Malagasy National Assembly and promulgated into law by the President of Madagascar.

Within the minining project QMM devoted particular attentions to recognize and to protect the region’s biodiversity, via a formal Social and Environmental Impact Assessment (SEIA), implemented between 1998 and 2001 (the government issued an environmental permit in 2001). In fact, the mining project is located in one of the richest biodiversity sites in the world with a high endemic level and

(11)

5

a unique coastal ecosystem. Moreover, there is a very complex socio-economic condition that causes a strong pressure on natural environment. The population turns to natural resources to meet its needs. So, deforestation for slash-and-burn farming or tavy and making charcoal, is the biggest factor in massive destruction of the littoral forest habitat. Altought the survival coastal forests are fragmented and partially degraded, their biodiversity remains very high. To protect these unique biodiversity sites, QMM agreed to create conservation zones, leaving intact 12% of the deposit area. Three conservation zones were enstabilished (totaling 620 hectares), together with 31,275 hectares of legally protected biodiversity offsets. QMM co-manages the zones with local communities, the Malagasy forest department, in terms of a DINA (a set of rules mutually agreed for managing local natural resources).

Rio Tinto QMM is determined to show that the company’s environmental activities will lead to resulting benefit for this greatly endangered environment. Rio Tinto QMM has therefore committed to conserve this biodiversity on the sites where it operates and even to support conservation on the regional level via an Environmental Management Plan (PGE), attached to the Environmental permit issued by the government (ONE) in 2001. The ultimate goal of QMM is to minimize the mine’s negative impacts on biodiversity and create a model project for investment compatible with biodiversity conservation.

(12)

6

The littoral forests of the Fort Dauphin area

The tropical rain forests of

Madagascar, before human

colonization, are thought to have covered much of the eastern coastal plains and the eastern escarpment of the central plateau that runs along most of the 1600 km length of the island. So the original extent of this habitat was about 11.2 million ha. Estimates of the coverage of the remaining eastern rain forests in 1985 have ranged from 2.5 million to 6.9 million ha, and between 10,000 to 20,000 and 165,000 ha of forest are estimated to have been cleared per year (Green & Sussman, 1990). In this threatened ecosystem, the littoral forests, a peculiar type of tropical rain forest resting on sandy substrates and distributed along the eastern coastal plain, have received the most critical damages because its facilitate accessibility and have been attributed high conservation priority (Ganzhorn et al., 2001; Bollen & Donati, 2006).

The littoral forest ecosystem, which is restricted to unconsolidated sand (originated during the Quaternary by podzols1) in a narrow coastal band extending along Madagascar’s east coast from north of Vohimarina to just southwest of Fort Dauphin (Consiglio et al., 2006), has long been subjected to both natural disturbances, such as

1

Podzols are soils characterized by the presence of a dark brown horizon (spodic horizon) containing an accumulation of organic matter, iron and aluminium within 200 cm of the soil surface. The process of translocation and accumulation is usually shown by the occurrence of an albic horizon (a light coloured subsurface horizon from which clay and free iron oxides have been removed) underlain by the dark horizon.

Fig. 1.2. Anosy region (in red) and Fort Dauphin area.

(13)

7

cyclones (Birkinshaw et al., 2002), and human-induced pressures including fires, deforestation, conversion for agriculture, and urban expansion.

In Fort Dauphin area in 2005, just 3128 ha of littoral forest remained, distributed in three distinct zones, Petriky, Mandena and Ste Luce and consisting of many fragments, ranging in size from 1 to 377 ha (Rabenantoandro et al., 2007).

Establishment of conservation zones has been made by QMM, in order to conserve and manage biodiversity and to ensure sustainable access by the population to renewable forest products (e.g. medicinal plants). The Mandena Conservation Zone (230 ha, M15/M16 fragments), established in 2000, has served as a pilot project, and has been successful in conserving and regenerating portions of the forest. A conservation zone was also established at Ste Luce in 2005 encompassing 747 ha of forest and wetland (S8, S9, S17 fragments), and 125 ha were included in a conservation zone at Petriky in 2008 (part of fragment P1), (Vincelette et al., 2007a and 2007c).

Parcel S9 at Ste Luce shows all the characteristics of nearly intact, low elevation dense humid forest, with about 60% cover among trees that are 12 m or more in height, and a clearly stratified structure (Rabenantoandro, 2001).

In low elevation dense humid forest at non-littoral sites the canopy can reach 18 m in height (Rabenantoandro, 2001), whereas it is typically only about 14 m tall in intact littoral forest, presumably a result of the sandy substrate and the influence of winds. At Mandena, the bioclimatic factors are nearly identical to those at Ste Luce, although rainfall is slightly lower. The observed differences in structure found at Mandena probably indicate that the forest present there today is a degraded form of a vegetation type that is

(14)

8

shared with Ste Luce. Located just 9 km north of the town of Fort Dauphin, Mandena has clearly been heavily impacted by humans (Rabenantoandro et al., 2007).

Even though the floristic composition of the forest in Ste Luce and Mandena is similar, that of Ste Luce has the highest species diversity and is one of the most intact littoral forests remaining in Madagascar (Rabevohitra et al., 1996). In a 377 ha forest fragment at Ste Luce 98% of the 189 plant species are endemic to Madagascar (Rabevohitra et al., 1996).

Petriky, located at the extreme southern end of Madagascar’s east coast, has species characteristic of humid formations, but it differs from the other two sites by also having taxa typical of dry areas. Floristically and faunistically, Petriky could be interpreted as a transition between the dry forest and the humid littoral forest (Rabenantoandro et al., 2007). At least 40 plant species are endemic to the littoral forest of the Fort Dauphin region (Lowry, 2001). So, the Anosy region is one of the ecologically most diverse regions of Madagascar (Ramanamanjato et al., 2002), but also one of the poorest and most isolated of this island nation. The population of the Anosy region is expected to double before 2020, with an average annual increase of 2.9%. The regional GDP is estimated at US$ 61 million, which translates into a per capita GDP of US$ 182 (CRD 2003). The economy is largely agricultural and dominated by rice and sisal cultivation. Given the reduced demand for sisal on the world market, production is dwindling and contributes less to the regional economy. Rice productivity is extremely low, and the region imports 12,000 tons annually. The percentage of the regional population living below the poverty line is 82%, higher than the national average of 74%. Not one regional human development indicator (life expectancy, revenue, literacy, etc.) meets the national

(15)

9

average. Ninety percent of the population does not have access to standard drinkable water, and 74% of children do not attend school. Health centers are relatively few, and are marginally staffed and equipped. The most common diseases affecting the population are malaria, filarial worms, leprosy, polio, and sexually transmitted illnesses.

In this difficult socio-economic status the biggest challenge, for anyone involved in conservation projects in the province of Fort Dauphin, is to reconcile the demands of the population with the urgent need to protect the surviving littoral forest fragments. Not always, for example, the prohibition to take natural resources from conservation areas of Mandena and Ste Luce is respected. In these cases imposing strict prohibitions by force would be counterproductive and unfair. The winning solution is to provide alternative resources to supply the basic needs and, above all, to educate the population. This is what, in recent years, researchers and staff of the QMM are trying to implement, in cooperation with the NGO Azafady and PeaceCore.

Important milestones have been reached, such as the cessation of hunting lemurs in conservation areas of Mandena and Ste Luce, but the way towards a complete resolution of the problem is still long.

The red collared brown lemur (Eulemur collaris)

The red collared brown lemur, Eulemur collaris GEOFFROY, 1817, is a strepsirrhine primate belonging to the family Lemuridae, endemic of the Madagascar eastern forests. In the recent past it was considered as a subspecies of Eulemur fulvus and it has been identified as Eulemur fulvus collaris, but recent DNA studies have

(16)

collaris can be seen in the Luce Private Reserve

The red collared brown lemur humid littoral forests of South

2007b), where it plays a fundamental ecological role as the principal seed disperser for many

2006).

In fact, previous stud

frugivore in these forests able to swallow and thus seeds, up to 16.

2004). Local extinction

of those plant species that are depe E. collaris is as a medium

and maximum length of 1 m) with a brown orange collar around the cheeks of

differences in size between sexes, but present, mostly in the head region

collar around the cheeks of gray as in females

Fig. 1.3. Distribution of collaris (in red).

allowed us redefining different and independent (Groves, 2005).

The distribution of the brown lemur (Fig. 1.3)

to the South-eastern Madagascar, in the Anosy region, in the forest of Kalambatritra, and in Fort Dauphin area. The typical habitat is the mountain rain forest,

live in the littoral forests can be seen in the Mandena Conservation Zone Luce Private Reserve, and in Andohahela National Park

collared brown lemur is the largest lemur species of humid littoral forests of South-East Madagascar (Donati et al.,

where it plays a fundamental ecological role as the principal seed disperser for many plant species (Bollen

In fact, previous studies have shown that E. collaris frugivore in these forests able to swallow and thus to

.5 mm diameter (Ganzhorn et al. 1999, Bollen et al., extinction of E. collaris could lead to the eventual loss of those plant species that are dependent on it for seed dispersal.

as a medium-sized lemur (maximum weight of 3 kg and maximum length of 1 m) with a brown-gray fur and a wide orange collar around the cheeks of the males. There are no

in size between sexes, but a sexual dichromatism is present, mostly in the head region: only males show the typical red around the cheeks and have the top of the head black, instead

females (Fig. 1.4). Distribution of Eulemur

10

ing it as a and independent species

The distribution of the red collared (Fig. 1.3) is restricted Madagascar, in , in the forest of and in Fort Dauphin . The typical habitat is the but they also live in the littoral forests: E. Mandena Conservation Zone, in the Ste

Andohahela National Park.

is the largest lemur species of the East Madagascar (Donati et al., where it plays a fundamental ecological role as the species (Bollen & Donati,

E. collaris is the only to disperse large 5 mm diameter (Ganzhorn et al. 1999, Bollen et al., could lead to the eventual loss ndent on it for seed dispersal. lemur (maximum weight of 3 kg

gray fur and a wide There are no sex a sexual dichromatism is : only males show the typical red and have the top of the head black, instead

(17)

11

The species is mainly frugivorous but it shows a great dietary flexibility, feeding also on leaves, flowers, mushrooms, petioles, gums and invertebrates in variable proportions depending on forest phenology and food availability (Donati et al, 2007; Donati et al, 2011). The species is cathemeral (active both day and night throughout the year), a trait seen in other members of the genus (Curtis & Rasmussen, 2002; Kappeler & Erkert, 2003; Schwitzer et al., 2007a).Previous research has suggested that metabolic dietary-related needs are the leading factor behind this behavior, although the specific hours of this activity pattern can shift in relation to lunar luminosity and seasonal changes in the photoperiod (day length). Previous studies have ruled predation as a factor affecting the expression of this trait, and have pointed instead to fruit availability and fiber intake as more important factors (Colquhoun, 2006). Multimale–multifemale groups of E. collaris have been observed in the wild (Donati, 2002; Donati & Borgognini-Tarli, 2006; Donati et al., 2007b; Donati et al., 2010; Donati et al., 2011), but social structure is still poorly understood and the literature available only concerns the phylogenetically near species of the genus Eulemur (Sussman, 1974; Overdorff, 1998; Pereira &

Fig. 1.4. Female (on the left) and male of Eulemur collaris (Ste Luce forest). Photo by Valentina Serra.

(18)

12

McGlynn, 1997; Wimmer & Kappeler, 2002; Ostner & Kappeler, 2004; Marolf et al., 2007). Females give birth to one or two offspring between October and December. The red collared brown lemur was listed as Vulnerable (VU A2cd) in the 2008 IUCN Red List assessment. Its greatest threat is habitat loss from slash-and-burn agriculture and charcoal production. It is also hunted for food and captured for the local pet trade. Populations of the collared brown lemur have been successfully bred in captivity as a safeguard

against their extinction (ISIS, 2011 -

(19)

13

CHAPTER 2

THE AIM OF THE STUDY

lant-frugivore interactions are important components of the complex forest communities, and seed dispersion by vertebrates is a key process in vegetation dynamics and recovery (Wallace & Painter, 2002). Frugivores play a vital role in the maintenance of biodiversity in tropical forests, where they constitute a large proportion of the vertebrate biomass (Fleming et al., 1987), and zoochorous tree species make up the bulk of tropical plant species (Howe & Smallwood, 1982). Loss of fruit-frugivore interactions can thus have profound consequences (Corlett, 1998). E. collaris is particularly important for dispersal of the seeds of numerous plant species in the littoral forests of South-eastern Madagascar (Bollen et al., 2004). It is probably the last of the remaining large-bodied frugivores in these littoral forests that is capable of ingesting large-sized seeds, thus matching the situation of brown lemurs in some of Madagascar’s dry deciduous forests (Ganzhorn et al., 1999).

P

Fig. 2.1. Male of E. collaris eating a fruit of Voakarepoka (Brexia madagascariensis), at Mandena forest. Photo

(20)

14

In respect to tight co-evolution there are five species of plants that apparently are dispersed exclusively by Eulemur collaris, namely Canarium boivinii, Diospyros sp., Eugenium sp, Hyperacanthus mandenensis, Cinnamosma madagascariensis var. namoronensis (Bollen et al., 2004). These fruits were significantly heavier and longer (up to 30 mm seed length) than the other fruits. Even though these lemurs often drop the large seeds under the parent plant, occasionally seeds are swallowed and defecated or dropped some distance away from the parent plant. Thus, in terms of conservation, these relationships are of crucial importance to preserve the integrity of the littoral forest.

As Malagasy forests become more fragmented the remaining patches become increasingly isolated and inaccessible to arboreal lemur species (Ganzhorn et al., 2001). Large frugivores are often the most vulnerable to habitat fragmentation (Kannan & James, 1999) and this is the case for E. collaris in Ste Luce, where the species is only present in the largest fragments, S9 (377 ha) and S17 (237 ha), two of the last remaining areas of littoral forest in South-eastern Madagascar. This species is also present in the Conservation Zone of Mandena, a series of littoral forest fragments located about 50 km South-West from Ste Luce area (Fig. 2.2). Floristic and structural characteristics of the Mandena forest show a fragmented and seriously compromised habitat (Vincelette et al., 2007c; Donati et al., 2007b). Nevertheless, the degradation of the forest of Mandena provides a model to study and to understand ecological and behavioral mechanisms used by animals to cope with such degraded habitat. Previous studies in degraded and/or fragmented areas have underlined the influence of habitat modification on primate ecology and behaviour (Ganzhorn et al., 2007).

(21)

15

The most significant changes concern time budget, feeding choices, social interactions, group size and structure, and habitat use (Donati, 2002; Bollen et al., 2005; Bollen & Donati, 2005; Ganzhorn et al., 2007; Donati et al., 2007a; Donati et al. 2010, Donati et al., 2011). However, what is not yet clear is how the above behavioral and ecological flexibility may affect animal stress levels. Therefore, the aim of our work is to understand whether the population of red collared brown lemurs in Mandena uses different ecological and social strategies as compared to the population living in the well preserved forest of the Ste Luce area.

To answer this question our work was structured in several different levels of investigation: habitat use, activity pattern, feeding ecology, behavioral thermoregulation, and sociality.

These eco-ethological aspects were studied in both sites of Mandena and Ste Luce. The data collected allowed us to Fig. 2.2. Map of the Ste Luce and Mandena sites, in South-eastern Madagascar (Bollen,

2003).

Ste Luce

(22)

16

integrate knowledge on Eulemur collaris, that in some aspects were scarce, especially as concerns social structure.

The comparison of the data obtained from groups living in well preserved and in degraded habitats allowed us to recognize

ecological and social strategies/behaviours used by

Eulemur collaris to survive in a secondary forest. This knowledge is essential for proper wildlife management which leads to preservation of the species and of the ecosystem. In fact, to protect

Eulemur collaris means to protect the littoral forest and

its biodiversity.

Fig. 2.3. Female of E. collaris at Ste Luce forest. Photo by Valentina Serra.

(23)

17

CHAPTER 3

GENERAL METHODOLOGY

Study Sites

he study was conducted in the littoral forest fragments of Mandena (M) and Sainte Luce (SL) (Fig. 3.1) in the Fort Dauphin region (25°01′ S, 46°58′ E) in South-eastern Madagascar from February to July 2011.

T

Fig. 3.1. Study sites. Map of the Sainte Luce and Mandena areas in South-east

Madagascar, indicating the location of major littoral forest parcels (light green), including those that comprise the established conservation zones at the two sites (dashed red and yellow lines), (Lowry et al., 2008).

(24)

18

The region is characterized by a tropical wet climate (average monthly temperature is 23°C) and by high levels of annual rainfall (ranging 1600-2480 mm), mainly due to the orographic features of the Anosyenne and Vohimena mountains. These climatic conditions were favorable to the formation of a dense humid forest, that in the recent past was probably present across the entire region (Rabenantoandro et al., 2007). There is not a clear annual dry season, but during our study we have identified two climatic segments depending upon temperature variability: a hot season from February to April (average temperature: 24.7°C) and a cool season from May to July (average temperature: 19.7°C) (Fig. 3.2).

The Conservation Zone of Mandena (24°57' S, 47°0' E), 11 km North-West of Fort Dauphin, is located on sandy soils at an altitude 0–20 m above sea level. Our research was done in the fragments M15/M16 (about 148 ha) and M20, (about 40 ha) not included in the Conservation Zone. All these areas are surrounded by a swamp forest (about 82 ha), regularly used by Eulemur collaris for its activities (such as moving, feeding, resting). So we considered these three fragments as a unique study area (about 250 ha in total).

HOT COOL 0 5 10 15 20 25 30 0 50 100 150 200 250 300 350 400

Feb Mar Apr May Jun Jul M

e a n M o n th ly T e m p e ra tu re ( C °) T o ta l M o n th ly P re ci p it a ti o n ( m m ) Rainfall Temperature

Fig. 3.2. Average monthly temperature and rainfall during the study period.

(25)

19

The Mandena littoral forest is defined as a “fairly degraded” (M15) or “heavily degraded” (M16, M20) habitat (Vincelette et al., 2007c) (Fig. 3.3).

The most frequent tree diameter class is 10-15 cm DBH, with an average canopy height of ( ±SD) 8.9±4.4 m (Rabenantoandro et al., 2007). Five lemur species reside in Mandena Conservation

Fig. 3.3. Forest conservation status at the two study sites.

Map of the littoral forest areas remaining in 2000 at Ste Luce and Mandena. Conservation status of the forest fragments is indicated by colors: intact (dark green), well preserved (light green), fairly degraded (yellow), heavily degraded (orange), extremely degraded (red). Conservation Zones are indicated by dashed lines.

(26)

20

Zone: the red collared brown lemur (Eulemur collaris), the gray bamboo lemur (Hapalemur griseus), the southern woolly lemur (Avahi meridionalis), the fat-tailed dwarf lemur (Cheirogaleus medius), and the gray mouse lemur (Microcebus murinus).

The second study site, the Conservation Zone of Sainte Luce (24°46’S, 47°10’E), 50 km North of Fort Dauphin, is located on sandy soils at an altitude 0–20 m a.s.l. The study fragment was S9, 377 ha of littoral forest and swamp. It is defined as a “well preserved” habitat (Vicelette et al., 2007c), probably one of the largest and most intact surviving fragments of Malagasy littoral forest. The most frequent tree diameter class is 10-15 cm DBH, with an average canopy height of ( ±SD) 14.7±4.3 m. In this site large trees are also present measuring at least 20 m in height, totally absent in the Mandena forest (Rabenantoandro et al., 2007). Four lemur species reside in Ste Luce Conservation Zone: the red collared brown lemur (Eulemur collaris), the southern woolly lemur (Avahi meridionalis), the fat-tailed dwarf lemur (Cheirogaleus medius), and the gray mouse lemur (Microcebus murinus). The forests of Mandena and Ste Luce appear to be floristically very similar, but the different preservation status makes them two clearly distinguishable habitats.

Study species and captures

Eulemur collaris is an arboreal, medium-sized lemur (average body mass = 2.15 kg; average body length= 46.16 cm) living in multi-male multi-femulti-male groups (range: 2-17 individuals) (Donati et al., 2007b). Its dietary regime is mainly frugivorous and it shows cathemeral activity.

(27)

21

Individuals of red collared brown lemurs presently living in the Mandena forest fragments have been translocated there in 2000 (Donati et al., 2007b) from other littoral forest fragments (M3 and M4) threatened by charcoal maker, to avoid the extinction of the species in the Mandena area. At the moment of translocation the animals were 28, then, after a slight reduction they were 36 in 2003, suggesting a good response to translocation and a good adaptation to the new environment (Donati et al., 2007b). Unfortunately, in 2004 animals decreased to 25 due to predation by two immigrated subjects of Cryptoprocta ferox, that, subsequently, have been captured and then transferred to the Antananarivo Botanical and Zoological Park of Tsimbazaza.

During our study period the population of red collared brown lemurs in Mandena was about 12 animals. The cause of this marked reduction of population density is still unknown. It may be due to predation, scarcity of food resources, or to animal migration towards other forest fragments out of the Conservation Zone. So, during our observation period, the density of E. collaris in Mandena was about 4,8 individuals per km2. This is well below the density of 38 ind/ km2 found in Saint Luce in 2002 (Donati, 2002), which probably increased until 2011 due to cessation of human hunting. However, this remains an unproved hypothesis since we were not able to make a census.

To ensure a continuous observation of E. collaris, four animals (one for each group) were captured in order to install radio-tracking collars, to be monitored by the use of a SIKA receiver (Biotrack). The adult individuals were captured by caging them, using banana slices as a bait, and rapidly anesthetized with Zoletil 100 (5 mg/kg of tiletamine hydrochloride) to prevent trauma. Morphometric measurements were taken and then thermo-sensitive collars TW-3

(28)

22

(medium mammal tag) were fixed. All animals recovered from anesthesia within 1.5 hours and were not moved from the capture area nor kept in a cage, but were followed until regaining full mobility. There were no injuries as a consequence of the captures.

Behavioural observations

Ethological data were collected on four E. collaris groups of different size, two in Mandena and two in Sainte Luce by Marta Barresi, Valentina Serra (February-July), Murielle Ravaolahy (March), Michela Balestri and Marco Campera (July). The groups in Mandena were named AB and C, those in Ste Luce A and B. Group composition is shown in Table 3.1.

Every group was followed four days per month and one group per site was followed one night per month. Overall, 495 diurnal observation hours in Mandena were compared with 645 hours in Sainte Luce, while the amount of nocturnal observations was 131 hours, 70 in Mandena and 61 in Ste Luce. Diurnal observation hours were distributed as shown in Figure 3.4.

Tab. 3.1 Composition of the four study groups.

STE LUCE MANDENA

Group A Group B Group AB Group C

Adult males 4 4 3-2 2

Adult females 3 4 1 2

Juveniles 2-1 2 0 1

(29)

23

The lower amount of observation hours in Mandena is mainly due to the impossibility to follow group C in the swamp area, where it lived from February to May, when the marsh has a high water level. Moreover, the same adverse swamp conditions and the arrival of a cyclone did not allow us to work in Mandena on group AB during the first month of the study.

The groups were followed from 6:00 to 18:00 during each observation day and from 18:00 to 6:00 during each observation night. A focal animal was chosen among the adult individuals, trying to balance the time spent observing males and females. The focal animal was followed as long as possible, and substituted with the first visible animal of the same sex when it was lost.

Individual recognition was possible observing radio-collars and some characteristics of the animals such as age, sex, size, canine length, tail shape, fur color, and distinctive traits.

We collected ethological data using the instantaneous sampling method, with records every five minutes, the all occurrences method and the ad libitum method (Altmann, 1974).

Behavioural data concerned animal activity: resting, moving, feeding, social interactions and anti-predator strategies, recorded by the use of a specific ethogram, shown in Appendix I (Donati, 2002,

0 20 40 60 80 100 120 140 160

FEB MAR APR MAY JUN JUL

H o u rs A B AB C

Fig. 3.4. Monthly amount of observation hours. A, B: groups of Ste Luce;

(30)

24

modified in this study). We also recorded resting/huddling postures, position in the tree crown (height, distance from the trunk), proximity, food items, and tree species.

During nocturnal observations it was often impossible to recognize a specific animal, we therefore followed the general activity of the group. We used a very basic ethogram, with the only voices of resting, moving and feeding, but we could often only assess if animals were active or not.

Trees were marked after three instantaneous of focal animal resting or during one instantaneous of feeding with a flag signed by a specific code (group name-activity-progressive number), and the diameter at breast height (DBH) was measured. Then the field assistant provided the vernacular name of the species and a small branch sample for identification. By these data a botanist of QMM was able to identify the scientific names of the plants. List of species used as feeding trees by E. collaris is shown in Appendix II. Every half hour the position of the animals was recorded by GPS (Garmin eTrex Legend HCx ).

Statistical analysis

We mostly used non parametric statistical analyses (Mann-Whitney test, Wilcoxon test, Kruskal-Wallis test, Chi-Square test, Spearman correlation index) because the sample size was often small, and the Kolmogorov-Smirnov test indicate significant deviations from normality.

We performed all tests with STATISTICA for Windows, version 8.0 and we considered p<0.05 as the significant level.

(31)

25

CHAPTER 4

HABITAT USE

Introduction

cological factors are critical in determining the spatial distribution of primates, such as forest structure, resource availability and predation. Home range was defined as “the area in which an individual engages in its regular activities” (Burt, 1943). In primates, home range size is often positively related to group size, but food resources availability may also be a good predictor for ranging area dimensions (Clutton-Brock & Harvey, 1979). In fact, a uniform distribution of food resources allows animals to minimize their movements to fulfil energy needs, while a habitat where resources are more scattered, often, requires more travelling. So, the conservation status of the habitat plays a fundamental role in shaping home range dimension, in relation to resource distribution and availability (Merker et al., 2005; Schwitzer et al., 2007b).

Home range varies also in relation to animal size, social systems and activity cycle. The home range of the smallest lemur, the nocturnal and solitary Microcebus murinus is 0.6-4.8 ha, depending on sex and reproductive cycle (Radespiel, 2000), while the ranging area of one of the largest lemurs, the diurnal and group living Propithecus diadema edwardsi is 25-100 ha (Wright, 1995). However, ranging patterns in the family Lemuridae appear to be mostly influenced by forest type than by group size (Sussman, 1974; Curtis & Zaramody, 1998): groups of Lemur catta and Eulemur rufus, which inhabit forests of different type, have

(32)

26

different home range areas in spite of similar group sizes (Sussman, 1974; Overdorff, 1996).

The well preserved forest fragment at Ste Luce and the degraded fragments at Mandena represent ideal observational conditions to make comparisons between strategies used by groups of Eulemur collaris living in different habitats in the wild. In this chapter we will provide a brief characterization of the forest at the two study sites, in order to evaluate the ecological factors that might influence Eulemur collaris ranging patterns. Moreover, the analysis of habitat use will provide further information about the ecological flexibility of E. collaris in a degraded area. In Mandena we expect to find significant differences as compared to Ste Luce, such as increase of ranging areas and of the number of feeding trees, as already shown by Donati et al. (2011).

Materials and methods Plot analysis

We examined forest structure and botanical diversity of animal home ranges, analyzing five plots (20 m x 50 m) within the home range of each study group (Fig. 4.1). We considered every tree with a diameter at breast height (DBH) > 5cm, noting vernacular name, DBH and crown diameter. We also characterized the vertical forest structure via the Gautier method (Gautier et al., 1994): a 50 m transect was established across the middle of each plot, at 1 m intervals, a measuring pole was held vertically and each point of contact with the vegetation was recorded.

(33)

Botanical diversity for each plot was calculate index (H’) formula:

where S is the number of species present, p

abundance of each species in the plot (number of trees of each species i over the total number of trees). The greater the botanical diversity, the greater

a)

c)

Fig. 4.1. Overall home range of the four study groups

White squares repre

lines are the forest pathways at the two study sites, tracked via GPS. performed with Map Source.

a) Ste Luce, home range

(February-July); c) M, home range range of group C (June

Botanical diversity for each plot was calculated with the Shannon index (H’) formula:

where S is the number of species present, pi is the relative

abundance of each species in the plot (number of trees of each species i over the total number of trees). The greater the botanical diversity, the greater is H’.

b)

d)

. Overall home range of the four study groups with plot locations

White squares represent group waypoints; black circles indicate plot locations; lines are the forest pathways at the two study sites, tracked via GPS. performed with Map Source.

, home range of group A (February-July); b) SL, home range July); c) M, home range of group AB (March-July); d) M group C (June-July).

27

d with the Shannon

is the relative abundance of each species in the plot (number of trees of each species i over the total number of trees). The greater the botanical

with plot locations.

black circles indicate plot locations; gray lines are the forest pathways at the two study sites, tracked via GPS. Graphs July); b) SL, home range of group B d) Mandena, home

(34)

28

Habitat use analysis

We collected behavioural data using the instantaneous sampling method, with records every five minutes, the all occurrences method and the ad libitum method (Altmann, 1974). Trees were marked after three instantaneous of focal animal resting or during one instantaneous of feeding with a flag and signed by a specific code (group name-activity-progressive number), and the diameter at breast height (DBH) was measured. Then the field assistant provided vernacular name of the species and a small branch sample for identification. By these data a botanist of QMM was able to identify the scientific names of the plants.

We recorded animal location every 30 minutes by GPS (Garmin eTrex Legend HCx ) during diurnal and nocturnal observations. We conducted analyses of home range size and location via Ranges 7 (Anatrack Ltd.). We estimated the monthly overall home range per group via 100% minimum convex polygons (MCP) using all the location points. The method creates a polygon including all the locations where an individual was observed (Mohr, 1947). Although it is a quick and easy method, it is also well known to overestimate home range size (Brugiere & Fleury, 2000; Burgman & Fox, 2003; Mohr, 1947; Pimley et al., 2005; Boyle et al., 2009a; Grueter et al., 2009). We therefore also used 95% and 50% adaptive core weighted Kernel analyses. The method is based on the actual amount of use of different areas by an individual (the utilization distribution) and therefore produces much more accurate home range estimates, excluding the outliers (Barg et al., 2005; Pimley et al., 2005). Home range graphic representations were performed with Map Source (version 6.13.7, Garmin Software), and with Range 7 (95% Kernel method), using all waypoints per group and the site pathways tracked by GPS.

(35)

29

Statistical analysis

Statistical comparisons between sites or groups were made by the non parametric Mann-Whitney U test and Kruskal-Wallis H test for independent data, while to compare different seasons we used the non parametric Wilcoxon T test for dependent data. We performed all tests with STATISTICA for Windows, version 8.0 and we considered p<0.05 as the significant level.

Results

Forest structure and floristic composition

Plot analysis showed that trees at Ste Luce had, on average, significantly greater dimension than those at Mandena. Mean tree DBH and height were ( ±SD) 11.6±6.7cm, 10.0±3.2 m at Ste Luce, and 9.3±5.2 cm, 7.9±2.7 m at Mandena, respectively (DBH: U10,10=9, p=0.001; height: U10,10=7, p=0.001 ). In fact the tree

classes of 14-16 m, 16-18 m, 18-20 m were totally absent in Mandena, while they were present at Ste Luce (Fig. 4.2).

Average number of trees per plot was higher at Mandena ( ±SD: 196.5±51.0) than at Ste Luce (174.8±54.1), where 76.2% and 73.7% of the trees, respectively, could be used by Eulemur collaris as feeding trees. The Shannon index of botanical diversity in the plots was 4.0±0.4 for those of group A, 3.8±0.4 for group B, at Ste Luce , while it was 3.5±0.1 for group AB, and 3.7±0.1 for group C at Mandena. No significant differences were found between groups (H ( 3, N= 25) =2.724162 p =0.4361).

(36)

30

Habitat use

Home ranges of Eulemur collaris at Ste Luce and Mandena, calculated via 100% minimum convex polygon method (MCP), were on average ( ±SD) 58.0±55.9 ha and 49.4±23.4 ha, respectively; using the 95% Kernel method they were 23.3±13.4 ha in Ste Luce and 26.0±12.3 ha in Mandena. Average core areas calculated via 50% Kernel method were ( ±SD) 8.7±4.0 ha for Ste Luce groups and 7.8±4.5 ha for those at Mandena (Tab. 4.1).

0 5 10 15 20 25 30 35 [0-2[ [2-4[ [4-6[ [6-8[ [8-10[ [10-12[ [12-14[ [14-16[ [16-18[ [18-20[ 20+ % of trees H e ig h t MANDENA 0 5 10 15 20 25 30 35 [0-2[ [2-4[ [4-6[ [6-8[ [8-10[ [10-12[ [12-14[ [14-16[ [16-18[ [18-20[ 20+ % of trees H e ig h t STE LUCE

(37)

31

Comparisons between the groups living at the two study sites gave non significant differences (H (3, N= 19) =7.098947 p =0.068). In fact, the home range of group B (SL) is similar to that of group AB (M) and the home range of group A (SL) is similar to that of group C (M). However, appreciable differences in ranging area were found between groups of the same site. In Ste Luce group A had a home range significantly larger than that of group B (U6,6=3,

p<0.05), and Mandena group C had a larger ranging area in comparison to group AB, but statistical analysis was not performed because of the low sample size.

During the cool season the home range of group A decreased, while that of group AB increased in size (Fig. 4.3).

0 5 10 15 20 25 30

Group A Group B Group AB Group C

H o m e r a n g e ( h a ) HOT COOL

Table 4.1 Home range values of the four study groups Total home range (ha)

100% MCP 95% Kernel 50% Kernel

Group A (SL) 97.49 32.69 11.48

Group B (SL) 18.50 13.80 5.81

Group AB (M) 32.82 17.35 4.65

Group C (M) 65.98 34.69 11.01

SL: Sainte Luce, M: Mandena; MCP: minimum convex polygons.

Fig. 4.3. Seasonal variations of the home range of the four study groups.

Data obtained via 95% Kernel method. Error bars represent standard deviations.

(38)

Also the shape of the home range is similar between group B and AB, round and regular, and between group A and C, irregular and much longer than wider (Fig. 4.4

Lemurs at Ste Luce used

than those living in Mandena. The mean daily number of feeding plants was ( ±SD) 14.7±6.1 in Ste Luce and 9.3±4.4 in Mandena. The number of feeding trees decreased during the

both sites (Fig. 4.

a)

c)

Fig. 4.4. Overall home range of the four study groups.

White squares represent group waypoints; gray lines are the forest pathways at the two study sites, tra

a) Ste Luce, home range group B (February Mandena, home range

Also the shape of the home range is similar between group B and AB, round and regular, and between group A and C, irregular and much longer than wider (Fig. 4.4, Fig 4.5).

Lemurs at Ste Luce used a higher number of feeding trees

than those living in Mandena. The mean daily number of feeding ±SD) 14.7±6.1 in Ste Luce and 9.3±4.4 in Mandena. The number of feeding trees decreased during the cool

both sites (Fig. 4.6).

b)

d)

. Overall home range of the four study groups.

White squares represent group waypoints; gray lines are the forest pathways at the two study sites, tracked via GPS. Graphs performed with Map Source.

, home range of group A (February-July); b) Ste Luce

group B (February-July); c) Mandena, home range of group AB (March , home range of group C (June-July).

32

Also the shape of the home range is similar between group B and AB, round and regular, and between group A and C, irregular and

a higher number of feeding trees per day than those living in Mandena. The mean daily number of feeding ±SD) 14.7±6.1 in Ste Luce and 9.3±4.4 in Mandena. cool season in

White squares represent group waypoints; gray lines are the forest pathways at the Graphs performed with Map Source.

uce, home range of of group AB (March-July); d)

(39)

33

a) b)

c) d)

Fig. 4.5. Overall home range of the four study groups.

Graphic obtained via 95% Kernel method analysis. White squares represent group waypoints (Range 7).

a) Ste Luce, home range of group A (February-July); b) Ste Luce, home range of group B (February-July); c) Mandena, home range of group AB (March-July); d) Mandena, home range of group C (June-July).

(40)

34

A total of 427 feeding plants used by lemur groups in Ste Luce and 208 in Mandena were marked and measured. The analysis of the size of feeding trees showed that Mandena groups fed on smaller plants, in term of DBH, than those used by Ste Luce groups. The mean monthly DBH of the feeding trees used by lemurs was ( ±SD) 18.0±5.3 cm in Ste Luce and 12.5±2.9 cm in Mandena (Fig. 4.7). 0 5 10 15 20 25

STE LUCE MANDENA

N ° o f p la n ts HOT COOL 0 5 10 15 20 25 30 0 5 10 15 20 25 30

M e a n D B H ( cm ) M e a n N ° o f tr e e s STE LUCE 0 5 10 15 20 25 30 0 5 10 15 20 25 30

M e a n D B H ( cm ) M e a n N ° o f tr e e s MANDENA Tree DBH

Fig. 4.6. Mean daily number of feeding trees visited at the two study sites during the hot and the cool season. Error bars represent standard

deviations. deviations.

Fig. 4.7. Mean daily number of feeding trees and their DBH throughout the study period. Error bars represent standard errors.

(41)

35

Discussion

Plot analysis confirmed our expectations about the conservation status of Ste Luce and Mandena forests. Mandena habitat is more degraded than Ste Luce, as shown by the lower height and DBH of trees in the first site. However, the similar Shannon index values and percentages of trees used in the E. collaris diet between the two sites suggested that floristically the two forests were comparable. Within the genus Eulemur the largest home range was found in Eulemur rufus (95-100 ha) living in eastern rainforests, while the smallest was recorded for E. fulvus (0.75-1.0 ha) in Western Madagascar (Sussman, 1974; Overdorff, 1993; Curtis & Zaramody, 1998). This high variability is mainly due to differences in resource distribution and availability between western and eastern forests. Our results for home range size (~13-18 ha) are lower than those found by Donati et. al (2011) for E. collaris in the same sites (~21-28 ha). This might be due to the different duration of the studies (six months vs one year) or to actual changes in ranging behaviour. The most interesting data from the home range analysis was that groups of the same site used different ranging strategies: one group resided in a defined and regular area that changed very little throughout the study period (group B in Ste Luce and group AB in Mandena); while the other group moved opportunistically across the whole study area (group A in Ste Luce and group C in Mandena). Moreover, groups at the two sites had a similar home range area in spite of their different size. This means that Mandena groups had in proportion larger home ranges than those of Ste Luce. This fact might be due to the more abundant and perhaps more uniform distribution of food resources that characterizes Ste Luce as compared to Mandena. In fact, small home ranges often indicate

(42)

36

abundancy of food resources, as in the case of the chimpanzees of Budongo Forest, Uganda (Newton-Fisher, 2003).

Probably, group size reduction in Mandena was a crucial step for E. collaris to minimize intra-group resource competition, that in a degraded habitat is more intense and challenging. Decrease in group size has been reported also for other primate species that inhabit fragmented and/or degraded forests, such as the black-and-white colobus (Onderdonk & Chapman, 2000; Marshall et al., 2005), the red colobus (Marshall et al., 2005), the black spider monkey (Symington, 1988), the crowned lemur (Wilson et al., 1989), and the barbary macaque (Machairas et al., 2003).

The analysis of habitat use confirmed our expectation over the effects of a degraded habitat at Mandena: red collared brown lemurs in this site used smaller feeding trees than those living in the intact forest of Ste Luce. However, our results showed that Mandena groups used a lower number of feeding trees per day than those of Ste Luce, contrary to previous findings (Donati et al., 2011). This result may be due to the fact that groups at Mandena usually visited the same few feeding trees of the largest available fruiting species, more than once per day, moving in a circle (i.e. Brexiella sp. in the hot season and Vitex chrysomalum in the cool season). Also beard saki monkeys (Chiropotes satanas chiropotes) that inhabit forest fragments tended to intensify their use of the forest in a similar way (Boyle et al., 2009b).

Resource use also varied between seasons in both sites: even if DBH size of visited trees did not change, the mean daily number of feeding trees decreased significantly in the cool season. Further data on E. collaris diet are available in Chapter 6.

Although our results are in part different from those of previous studies on the same species (Donati et al., 2007a; Donati et al.,

(43)

37

2011), our data confirm the high behavioural flexibility of E. collaris. This species was able to cope with the degraded habitat of Mandena via behavioural and ecological adaptations, e.g. by reducing group size to minimize food competition and increasing home range area. Moreover, E. collaris used more than one strategy to cope as well as possible with the environment, degraded or not, e.g. different ranging and social behaviours (see Chapters 4 and 9). These behavioural variability and ecological flexibility are at the basis of the resilience of red collared brown lemurs, that made them capable to survive in a challenging habitat, albeit in a delicate balance.

(44)

38

CHAPTER 5

ACTIVITY PATTERN

Introduction

Time budgets and activity patterns of primates are influenced by many factors: group size (Dunbar, 1992), habitat quality (Oates, 1977; Menon & Poirier, 1996), proximity to human settlements (Singh & Vinanthe, 1990), and 24 h activity phase (Dunbar, 1992; Fernandez-Duque, 2003).

Whithin the primate order, many species are known to be adapted to the diurnal or the nocturnal activity phase. In particular, Malagasy lemurs show a wide range of different lifestyles, being either almost completely diurnal or nocturnal, or cathemeral. The term “cathemeral” was introduced and defined by Tattersal (1987) as follows: “the activity pattern of an organism can be regarded as cathemeral when it is about evenly distributed over the 24 h daily cycle, or when significant amounts of activity, particularly feeding and/or travelling, occur within both light and dark portions of the cycle” (p. 201). Cathemerality is present in Eulemur (Donatiet al., 1999, 2001, 2009; Kappeler & Erkert, 2003; Schwitzer et al., 2007a), Hapalemur (Tan, 1999; Mutschler, 2002), and in the New World monkey Aotus (Dahl & Hamingway, 1988; Wright, 1996; Fernandez-Duque & Erkert, 2006).

E. collaris, as other Eulemur species, shows cathemeral behaviour (Curtis & Rasmussen, 2002; Kappeler & Erkert, 2003; Schwitzer et al., 2007a), considered by Engqvist and Richard (1991) as an adaptation to cope with a diet having a high fiber and a low carbohydrate content, especially during lean periods, in order to

(45)

39

minimize the time in which no food is being processed. In support to this hypothesis Eulemur does not show anatomical or physiological adaptations that allow it to retain food in the gut for extended periods of time, thus increasing food energy absorption (Overdorff & Rasmussen, 1995). Other studies supported the “dietary theory” (Wright, 1999; Tarnaud, 2006; Donati et al., 2007a) and demonstrated that cathemerality is also influenced by lunar phases and by light intensity (Donati et al., 2001; Kappeler & Erkert, 2003; Donati & Borgognini-Tarli, 2006). Further hypotheses on the origin of cathemerality involve antipredator strategy (van Schaik & Kappeler, 1996) and interspecific competition (Tattersall & Sussman, 1998; for a review see: Curtis & Rasmussen, 2006; Donati & Borgognini-Tarli, 2006). In the first case cathemerality has been interpreted as a response to predation pressure by diurnal raptors, while for the second theory, it might be an ecological adaptation to minimize competition for food resources with other sympatric primate species.

Previous studies on E. collaris registered a seasonal fluctuation from a cathemeral activity during the dry season (May-September) to a mostly diurnal activity in the wet season (October-April) (Donati & Borgognini-Tarli, 2006; Donati et al., 2007a).

We will examine here diurnal and nocturnal activity of E. collaris, making comparisons between study sites and seasons, in order to find out the contribution of these variables. Given the dramatic resource seasonality of Malagasy rainforests (Ganzhorn et al., 1999; Wright, 1999) we expect to find variations in the activity pattern of E. collaris between seasons, especially in the secondary forest of Mandena, by increasing moving and feeding time overall the 24 h cycle, to cope with a supposed lower quality diet.

(46)

40

Materials and methods Astronomical data

Astronomical data such as sunrise and sunset time, the duration of astronomical twilight (i.e., the period when the center of the Sun is between 12 and 18 degrees below the horizon), and moon phases were obtained from the Moon® program (version 1.0), using Sainte Luce’s geographical coordinates as entries. The time between sunrise and sunset on the 15th day of every month represented the mean monthly day length in the subsequent analyses. Mean monthly night length was obtained in the same way. At the study site, day length (time from sunrise to sunset) varied in our study period between 12.9 and 10.7 h, whereas night length (time from sunset to sunrise) varied between 11.1 and 13.2 h. Astronomical twilights lasted about 73 min, and were considered only in the analysis of the 24 h activity pattern. So the actual amount of nocturnal observation hours used in our analysis was 112.05 h (SL: 50.50 h; M: 61.15 h).

Statistical analysis

To determine time budget we calculated the daily percentages of each activity of the focal animal, using the instantaneous method, grouping them into four principal categories: resting (including also G, MG, LK and SC; see Appendix I), moving, feeding and other (all the behaviours not included in the previous categories). The daily percentages were compared between groups and sites using the non-parametric Mann-Whitney U test. To compare time budget between the hot and the cool season we calculated seasonal percentages of activity for each animal and then we compared them with a Wilcoxon matched pairs test T. We used the Spearman

(47)

41

correlation index to test the relation between nocturnal activity, lunar phase and luminosity. All tests were performed with STATISTICA for Windows, version 8.0, and we considered p<0.05 as the significant level.

Results

Diurnal time budget

Resting is the main activity for both groups of Ste Luce and Mandena (median, range: 61.2%, 17.7-86.5%), followed by moving (23.5%, 8.3-57.1%), feeding (13.7%, 1.4-48.1%) and other behaviours (0.0%, 0.0-4.5%). Details for sites are shown in Table 5.1. Ste Luce groups rested significantly less than Mandena groups (U57,45=536, p<0.001), but moved (U57,45=737.50, p=<0.001) and

fed (U57,45=521.50, p=<0.001) significantly more (Fig 5.1).

No differences in other behaviours were found (U57,45=1169.5,

p=0.448). Some differences were found between groups living in the same site: in Ste Luce forest (Fig. 5.2), group B moved significantly more than group A (U28,29=645.50, p=0.007) and

showed more other behaviours (U28,29=264.50, p=0.023).

Tab 5.1 Median, minimum, maximum values of percentage of records in wich E. collaris

spent time in the various categories of time budget at the two study sites.

STE LUCE MANDENA

N Median Min Max N Median Min Max

RESTING 57 49.62 17.74 82.71 45 67.38 47.76 86.49

MOVING 57 28.47 8.27 57.14 45 20.57 8.51 35.07

FEEDING 57 18.71 1.41 48.12 45 11.64 1.44 21.54

OTHERS 57 0.00 0.00 3.52 45 0.00 0.00 4.55

(48)

42

STE LUCE

RESTING MOVING FEEDING OTHERS

-10 0 10 20 30 40 50 60 70 80 90 % o f re c o rd s Median 25%-75% Min-Max MANDENA

RESTING MOVING FEEDING OTHERS -10 0 10 20 30 40 50 60 70 80 90 % o f re c o rd s Median 25%-75% Min-Max

Comparisons between group AB and group C in Mandena were available only for June and July, and for these months no differences in time budget were found.

Moreover, time budget changed depending on season in the Ste Luce forest. During the hot season animals exhibited less resting (T11=0, p=0.003) and more moving (T11=0, p=0.003) and feeding