Effects of Forest Fragmentation on the Abundance of Colobus angolensis palliatus in Kenya's

SPECIAL ISSUE: BEHAVIOR, ECOLOGY, AND CONSERVATION OF COLOBINE MONKEYS

Effects of Forest Fragmentationon the Abundance of Colobus angolensis palliatusin Kenya’s Coastal Forests

J. Anderson & G. Cowlishaw & J.M. Rowcliffe

Received: 13 January 2005 /Revised: 25 July 2005 /Accepted: 29 September 2005 /Published online: 5 June 2007# Springer Science + Business Media, LLC 2007

Abstract We documented the occurrence and abundance patterns of Angola black-and-white colobus (Colobus angolensis palliatus) in 46 coastal forest fragmentsranging from 1 ha to >1400 ha in the Kwale District, Kenya. In field surveysconducted in 2001, we also recorded forest spatial, structural, resource, anddisturbance characteristics to determine the effects of habitat quality and fragmen-tation and the factors most critical to the continued survival of the little-knownspecies. We tested 13 hypotheses to explain variation in patch occupancy andabundance patterns of Colobus angolensis palliatus in relation to habitat attributes.Minimal adequate models indicated that the occurrence of colobus in forestfragments is positively associated with fragment area and canopy cover, whereasthe density of colobus in occupied fragments is attributable to forest area, theproportion of forest change over the previous 12 yr, and the basal area of 14 majorfood trees. Large-scale illegal extraction of major colobus food trees in the Districtfor human resource use, in both protected and unprotected forests, together withongoing forest clearance and modification, are the major threats to Colobusangolensis palliatus in Kenya.

Keywords abundance .Colobus angolensis . forest fragmentation . habitat quality

Int J Primatol (2007) 28:637–655DOI 10.1007/s10764-007-9143-7

J. Anderson (*) :G. Cowlishaw : J. RowcliffeInstitute of Zoology, Zoological Society of London, Regents Park, London NW1 4RY, UKe-mail: [email protected]

J. AndersonDepartment of Anthropology, University College London, Gower Street, London WC1E 6BT, UK

J. AndersonWakuluzu, Friends of the Colobus Trust, P.O. Box 5380, 80401 Diani Beach, Kenya

mailto:[email protected]

Introduction

Primatologists have long cited habitat loss, fragmentation, and modification asleading threats to the persistence of primate communities in East African tropicalforests (Oates 1996; Rodgers and Homewood 1982; Struhsaker 1981b; Struhsakerand Siex 1998). Massive human population growth in the region has led to highdemand for forest resources, including fuelwood, charcoal, poles, and timber, andwidespread agricultural expansion, primarily in the form of food crops and exotictree plantations, which in turn have had a dramatic impact on East African forests;now only a small fraction of the original forest cover remains (Sayer et al. 1992).The impacts of forest loss alone, independent of the effects of fragmentation andmodification in the remaining forest habitat, threaten between 2 and 5 endemicprimate species (17–39% of all species) with extinction in this region(Cowlishaw 1999).

Colobines may be particularly vulnerable to the threats. They are highly arborealspecies that depend on leaves, seeds, and unripe fruit (Davies 1994), and maytherefore be at high risk of extinction from deforestation (Davies 1994; Marsh et al.1987). However, colobine species in West Africa are mainly threatened by humanhunting for the bushmeat trade (Oates 1996), as they are particularly sought byhunters for their large body size and higher financial returns (Davies 1987; Lahm1993; Martin and Asibey 1979). In contrast, the people of East Africa rarely eatprimate meat (Oates 1996), and so current threats in the region are primarily drivenby the expanding population growth and resultant deforestation.

Authors of previous studies found positive relationships between colobusabundance and forest habitat characteristics such as the basal area of food trees(Mbora 2004), protein-to-fiber ratio of mature leaves (Oates et al. 1990; Wassermanand Chapman 2003), and canopy height and tree cover (Medley 1993). Colobusoccupancy in forest patches is also related to the amount of forest edge and canopytree species composition (Mbora 2004). However, it would be incorrect to infergeneral trends for the colobinae, because at specific, or even subspecific, levelpopulations can respond differently to habitat alteration (Cowlishaw and Dunbar2000). For example, Asian colobine biomass correlates positively with the abundanceof leguminous trees (Davies 1994), whereas similar research has failed to find therelationship within the African colobines (Chapman et al. 2002; Davies et al. 1999).Similarly, sympatric studies at Kibale Forest, Uganda, showed that densities ofColobus guereza increased in light and heavily logged forest, but in contrast,densities of Procolobus badius declined under the same habitat modification(Chapman and Chapman 2002; Plumptre and Reynolds 1994; Skorupa 1986). Thedifficulty of making generalizations across taxa emphasizes the need for detailedstudies of individual species.

Angola black-and-white colobus (Colobus angolensis, Sclater 1860) live in theforests of northeast Angola, Democratic Republic of Congo, Rwanda, Tanzania, andKenya (Kingdon 1997). Relatively little is known about their abundance (Mate et al.1995), or response to habitat fragmentation. However, from the limited dataavailable, it is known that the species can achieve a diverse range of mean groupsizes within differing forest habitats, from 6 individuals in the Diani Forest, Kenya(Kanga 2001) to super-troops comprising ≥300 individuals in the Nyungwe Forest of

638 J. Anderson et al.

southwestern Rwanda (Fashing et al. 2004; Fimbel et al. 2001; Vedder and Fashing2002). Though the species is not currently listed as threatened, conservationists havehighlighted the subspecies Colobus angolensis palliatus Peters 1868, confined toislands of fragmented forest in coastal Kenya, Tanzania, and the Eastern ArcMountains (Kingdon 1997; Rodgers 1981; Tarara 1986), as vulnerable to extinctiondue to deforestation caused by tourist development schemes (Kahumbu 1997;Struhsaker 1981b). Rapid rates of human population growth in the region have alsoled to heightened requirements for local timber resources (Marshall and Jenkins1994; Robertson and Luke 1993), together with an intensification of forest clearancefor animal husbandry and agricultural practices (UNEP 1982), adding further threatto remaining populations of Colobus angolensis palliatus. The IUCN AfricanPrimate Action Plan also recommended that the status of the subspecies needed to beassessed in Kenya, in conjunction with stringent management plans to conserveremaining coastal forest fragments within the region (IUCN 1996).

Given the lack of knowledge concerning the subspecies and the conservationpriorities linked to its future existence in Kenya, we aimed to 1) identify the keyhabitat attributes that determine occupancy of Colobus angolensis palliatus withincoastal forest patches in Kenya and 2) identify the attributes that determine theabundance of populations of Colobus angolensis palliatus in the occupied patches.We test 13 hypotheses that relate colobus patch occupancy and abundance to thequality of the patch, which we define according to its spatial attributes, foodresources, structural attributes, and human disturbance. The hypotheses are based onthe findings of previous research on colobine abundance, summarized in Table I.

Methods

Study Site

Kwale District, in the Coastal Province of Kenya, lies between Mombasa and theborder of North Eastern Tanzania (3°30′, 4°45′ S; 38°31′ and 39°31′ E). Roughly8322 km2 in area, the District is largely an agroecological zone (Muchoki 1990),resulting in a heterogeneous mix of land cover types that include grasslands,woodlands, swamps, shrublands, forestry plantations, and annual and perennialcropland. Average temperature is about 26°C for the District, with highest meantemperatures of 33°C between November and April. The rainfall pattern is bimodal(long rains between March and July, short rains in October to December).Precipitation diminishes from the coastline to the hinterland, with the initial 36 kmfrom the sea: the coastal forest belt (Clarke 2000), receiving 900–1500 mm ofrainfall annually (Jaetzold and Schmidt 1983). Altitude varies from sea level alongthe Indian Ocean shore to 1000 m in the hinterland (HSEDCO 1998). Owing to thealtitudinal and climatic conditions of the coastal belt, the area is also interspersedwith fragmented and largely threatened coastal forest. The forests are remnants ofwhat was once an extensive coverage of lowland rain forest, swamp forest, scrubforest, and undifferentiated forest types (Clarke 2000). Part of the Zanzibar-Inhambane floristic region (White 1976), and recently reclassified as the SwahilianRegional Centre of Endemism (Clarke 1998), the unique forests largely grow on

Abundance of Colobus angolensis palliatus in Kenya’s Coastal Forests 639

Table I Independent habitat variables examined in the GLM statistical analysis and the hypothesizedrelationships with colobus abundance and occupancy

Habitat attributes Proposedrelationship

Hypothesized relationships with colobus abundance

SpatialForest size + A greater coverage of food resources will encourage population

occurrence and support greater numbers of colobus (Connor et al.2000; Estrada and Coates-Estrada 1996). Density responses areunknown.

Habitat edge + Patches with a greater degree of forest edge will support a greaterdensity/occupancy of colobus owing to the species’ exploitation ofsuccessional food resources, i.e., young leaves, lianes and vines that arisefrom edge effects (Coley and Barone 1996; Johns and Skorupa 1987).

Forest isolationdistance

- As a forest becomes increasingly isolated, chance immigrations fromneighboring colobus populations decline. Small populations ofcolobus may risk future declines or extinction owing to a breakdownof rescue effects (Brown and Kodric-Brown 1977; Dunbar 1987).

ResourceTree diversity + A greater variety of food resources, including infrequently used items,

may enhance the resource quality and thus carrying capacity of forestpatches (Chapman and Chapman 1999; Medley 1993). Lesser knowntree species may also provide important fallback foods at times offood scarcity, thus encouraging colobus occupancy/abundance inpatches (Cowlishaw and Dunbar 2000).

Tree, food tree ,and major foodtree density

+ Increased density of trees will provide greater food resources(Chapman and Chapman 1999; Mbora 2004) and canopy coverage(Medley 1993) for colobus. The degree of dietary flexibility withinColobus angolensis palliatus may be revealed through the relativeinfluence of food trees and major food trees on colobus density/occupancy.

StructuralForest canopyheight

+ Higher mean canopy heights greatly improve the canopy structure foran arboreal primate that rarely descends <10 m in the canopy(McGraw 1998a; Medley 1993), facilitating habitat occupancy andhigher colobus densities.

Forest canopycover

+ Greater degree of canopy coverage facilitates arboreal access to foodresources (McGraw 1998a; Medley 1993), also providing greatershelter from avian predators (Struhsaker 2000; Struhsaker andLeakey 1990). Increased density/occupancy predicted.

DisturbanceHistory of forestloss

+ Patches with a lower proportion of 1989 forest cover (<1) will have lowerdensities (or extinctions) of colobus owing to reduction of foodresources and lowered carrying capacity of forest. Conversely,patches that gained forest cover since 1991 (>1) will increase bothcarrying capacity and colobus density/occupancy (Bender et al. 1998;Marsh 1986).

Tree damage - The removal of trees from forest patches ultimately depletes foodresources (Medley 1993) and disrupts the structural characteristicsof the forest (Chapman and Chapman 1999), limiting the locomotionof an arboreal primate. Decreased abundance/occupancy predicted.

Humanutilization(4 variables)

- 1) Villages near forests, 2) hunting, 3) logging and charcoal activitiesmay have negative effects on colobus densities. Each activity cancause noise and structural disturbance to the forest (Cowlishaw andDunbar 2000; Marsh et al. 1987). 4) Paths and roads will also furtherfragment existing patches, encouraging higher levels of human trafficand additional noise disturbance. Decreased abundance/occupancypredicted.


coastal sedimentary rocks (Hawthorne 1993) and provide habitat for the present-dayKenyan distribution of the Angola black-and-white colobus (Colobus angolensispalliatus; Fig. 1). There are 124 coastal forest fragments remaining in KwaleDistrict, ranging from 1 ha to 160 km2. Some forest patches are occupied byColobus angolensis palliatus, while others are not. We focused on both occupiedand empty forest patches.

Data Collection

Colobus incidence and abundance. We randomly chose 46 forest patches fordetailed ecological study from a larger national survey involving all 124 forestpatches (Anderson 2005). Between July and November 2001, we systematicallysurveyed each of the 46 patches via a 1–day sweep sample technique (Whitesides etal. 1988) to obtain an estimate of colobus density for each forest patch. Researchershave also used the technique effectively to survey populations of Tana River redcolobus (Procolobus rufomitratus) in forest patches in the Tana River delta region,Kenya (Butynski and Mwangi 1994; Karere et al. 2004; Mbora 2004; Muoria et al.2003). Four or more survey teams, each comprising 2 trained observers plus 1

KENYA

KWALEDISTRICT

TanzaniaIndianOcean

Mombasa/KilifiDistrict

Mombasa

0 5 10 20 km

N

Fig. 1 Distribution of coastal forest fragments in Kwale District, Kenya.


experienced colobus field researcher (from Wakuluzu, Friends of the Colobus Trust,Diani Beach, Kenya), began surveys between 0630 and 0700 h each day. Teamswalked parallel transects 100 m apart, starting at the same time and traveling at thesame speed, along predetermined routes (Struhsaker 1981a) until an entire forestpatch had effectively been covered. Total survey time was thus a function of patcharea and the number of census teams. Maintenance of compass bearings throughouttransect walks and regrouping of survey teams after each forest transect sweep toresynchronize movements facilitated sweep sample accuracy. We recorded allprimate encounters in relation to species, group size and age-sex composition, time,and location along transect. In all instances when we encountered groups of Colobusangolensis palliatus, we spent ca.=10 min with each group utilizing all 3 membersof each survey team to count and to identify accurately the age-sex composition ofgroup members. We also recorded the time and direction of primate group departureafter the encounter. All teams discussed and enumerated results immediately after thecompletion of each survey and we removed double counts by comparing similaritiesin team observations, i.e., primate sighting times, departure directions, and groupcompositions.

We ranked all primate encounters on a 4–point scale in accordance withobservation quality: 0, primate vocalization detected but no individual observed; 1,primate group detected but group count incomplete; 2, primate group detected andcounted but age-sex composition incomplete; and 3, primate group detected withcomplete count and age-sex composition. We used the codes to guide the subsequentallocation of group size to each group encountered, as follows: 0, data discarded (noconfirmed group encounter), 1, group size based on the mean group size (6) obtainedfrom all group counts of quality 3 in the national survey (n=196 groups), 2 and 3,group size taken from the actual count. Once we established group size for allgroups, we calculated the number of colobus in each patch from the sum ofindividuals in each group plus all solitary individuals.

Habitat attributes. During the same survey period, we mapped all forest patchboundaries on foot by traversing the patch perimeter and recording positional data at10–s intervals via a Garmin 12 XL global positioning systems (GPS). We thendownloaded the GPS data into MapSource software (MapSource 5.3, GarminCorporation 1999), before importing them into an ArcView geographic informationsystem (ArcView GIS 3.2, ESRI Inc. 1999), in which subsequent GIS analysis viaXtools (Version 6.1, 2001) and Nearest Features (Version 3.6e, 2001) extensionsallowed for detailed estimation of forest spatial attributes, including forest area (ha),perimeter (m), habitat edge (area-to-perimeter ratio), and patch isolation distance (m).

Four botanical survey teams, each comprising 2 researchers, used transects tocollect data on forest resource and disturbance for each of the 46 patches. We carriedout 3–26 transects in each patch, depending on the patch area. Transect length alsovaried between patches (we used longer transects in larger patches), so that totaltransect length varied between 100 m and 3000 m per patch. The systematicplacement of transects throughout each forest also ensured an accurate representationof each patch. Transect data can be divided into 3 headings: forest resources, foreststructure, and forest disturbance. In the first case, forest resources, we enumerated alltrees >10 cm in diameter-at-breast height (DBH) ≤4 m of the transect (Grieg-Smith1983). We also recorded tree species, height (m), and DBH (cm), the latter providing


a relative index of canopy cover—and possible resource availability—for thecolobus (Decker 1994). We used the data to calculate tree diversity (number treespecies/ha), absolute tree density (basal area m2/ha), absolute food tree density (basalarea m2/ha), and absolute major food tree density (basal area m2/ha). We identifiedfood trees from ongoing research involving whole-day follows of colobus groups inthe Diani Forest, Kwale, conducted by the Colobus Trust (unpubl. data). The DianiForest is 1 of the 124 patches studied in our national survey of Kenya’s coastal forestsystem, but it was not one of the 46 patches chosen for more detailed study here. Werecorded feeding behavior and plant food species via instantaneous scan sampling ofall group members at 10–min intervals, which provided us with a basic list of foodtrees that we supplemented with feeding observations during our 2001 survey and bylocal field-assistant knowledge of colobus inhabiting other forests within the District.We established major food trees by calculating the proportion of colobus feedingbouts (n=14,445 individual feeding scans) for each tree species field researchersrecorded during August–October 1999 and February–June 2003.

In the second case, we recorded forest structure, the canopy height (m) andcanopy cover (%) every 50 m along the transect. We then calculated mean values foreach patch. In the third case, forest disturbance, we collected 4 types of data: 1) alltree damage ≤4 m of the transect, including stump/stem diameters, tree species, typeof damage (natural death, individual damage, logging, debarking), and age ofdamage (recent, old, very old); 2) perpendicular distances of all visible snares,pitsaws, and charcoal pits from the transect; and 3) all encounters with loggers,hunters, firewood collectors, access paths, and roads along the transect. In addition,at 100–m intervals along the patch perimeter, we estimated the minimum distance ofhuman settlement to the forest. We summarized patch disturbance as the absolutedensity of tree damage (basal area m2/ha), minimum distance (m) to humansettlement, density of snares (snares/km), relative density of human paths and roads(number of paths and roads/km), and incidence of pitsaws and charcoal pits(presence vs. absence).

Historical forest loss. To determine the effect of recent deforestation we digitized8 topographical maps covering the Kwale District. These 1:50,000 maps are basedon 1989 aerial photographs and field survey work by the Japan InternationalCooperation Agency and Survey of Kenya (Edition 4–JICA, 1991). We importedscanned TIFF files of the maps and into ArcMap (ArcInfo 8.3, ESRI Inc. 2002) andgeoreferenced them. We then digitized the forest boundaries. We measured forestchange for each of the patches as the proportional change in forest cover between1989 and 2001 (i.e., values <1 indicated loss in forest cover, 1=no change, and >1indicated gain in forest cover).

Statistics

We used a generalized linear model (GLM) framework (Crawley 1993) for theanalysis of both colobus patch occupancy and density, via the statistical software R,version 1.9.1 http://www.r-project.org). The first model, a stepwise GLM analysiswith binomial error structure, identified the habitat attributes that influenced colobusoccupancy in 46 forest patches. We regarded each forest patch as a unit, withpresence or absence of populations coded as a binary response variable. We entered


http://www.r-project.org

all explanatory variables detailed in Table I into a full model and log (base e)transformed to normalize the distribution of the variables. The only exceptions arecanopy cover, canopy height, and proportion of forest change, because theirdistributions were already normal. We sequentially removed all nonsignificantvariables, least significant first, until we reached a minimal adequate model. Wetested statistical significance via deletion F-tests corrected for overdispersion.

The second model analyzed colobus density in occupied patches via a stepwiseGLM with Poisson error structure. The total number of colobus individuals in eachof the 33 occupied patches was the response variable, with patch area as an offsetparameter. We tested the influence of the habitat variables in Table I by finding theminimal adequate model, using the same approach as for presence-absence data.

During the exploratory phases of both occupancy- and density-model fitting, wetested 3 variations in the estimation of resource availability—basal area of all trees, food

Table II Major food trees of Colobus angolensis palliatus responsible for >75% feeding records in theDiani Forest, Kenya

Family Species Part eaten a Common,Swahili name

Local human useb

Anacardiaceae Lannaewelwitschii

YL, ML,P, Fl, Fr, B

Muyumbu-Maji

Charcoal, furniture, fruit-edible,bark-rope and medicinal

Araliaceae Cussoniazimmermannii

YL, ML,P, Fl, Fr, S

Cabbage Tree,Mbomba Maji

Furniture

Bombacaceae Adansoniadigitata

YL, ML, P,Fl, Fr, B

Baobab, Mbuyu Fruit-edible, bark-ropes, weaving andmedicinal, roots-dye and medicinal

Combretaceae Combretumschumannii

YL, ML, P,Fl, F, S

Mgurure, Mpera-Mwitu

Hardwood timber, building timberand poles, charcoal, fuelwood,woodcarving

Euphorbiaceae Drypetesreticulata

YL, ML,P, Fr

unknown Charcoal, building timber andpoles

Meliaceae Trichiliaemetica

YL, ML,Fr, S

Mnwamaji Fuelwood, building timber,furniture, roots-medicinal andseeds-oil used for soap

Moraceae Milicia excelsa YL, ML, Fl,Fr, S

Iroko, Mvuli Building timber, boat building,furniture, joinery and bark-medicinal

Papilionaceae Millettiausaramensis

YL,ML Mwino, Mtupa Hardwood timber used in building

Rutaceae Zanthoxylumchalybeum

YL, ML, P,Fr

Knobwood,Mjafari

Building timber, bark and leaves-medicinal

Sapindaceae Lecaniodiscusfraxinifolius

YL, ML, P,Fl, Fr, S

Mkunguma Hardwood timber used in building

Lepisanthessenegalensis

YL, ML, S unknown Building timber and furniture

Sapotaceae Sideroxyloninerme

YL, ML, P,Fl, Fr, S

Mkokobara,Mtunda

Fruit-edible

Tiliaceae Grewiavaughanii

YL, ML unknown Building poles, fuelwood, bowsand fruit-edible

Grewiaplagiophylla

YL, ML, P Mkone,Mfukufuku

Building timber, fuelwood, bows,arrows, rungus, fruit-edible,roots-medicinal

a Plant parts eaten by Colobus angolensis. palliatus are: YL=young leaves, ML=mature leaves, P=petioles, Fl=flowers, Fr=fruits, S=seeds, B=bark.b Local human use is also recorded because it may reflect possible human resource conflicts.


trees, and major food trees. The variables correlate closely and were not expected tohave independent effects, so we did not test them jointly in any given model. Instead, wetested the alternatives to determine which, if any, had the strongest effect on density andoccupancy.

Results

Colobus Occurrence and Density

We surveyed 13 empty and 33 (71.7%) occupied forest patches, which ranged from1 ha to 1417 ha. We recorded 769 Angolan black-and-white colobus in 136 social

0

10

20

30

40

50

60

70

80

Individual forest patches

Bas

al a

rea

(m2 /

ha)

Other

Other food

Major food

265

Fig. 2 Basal area coverage of major food trees, other food trees and other trees, highlighting the resourcevariability between forest fragments. Patches plotted in order of increasing forest area (ha). The smallestforest patch has an unusually high major food tree density owing to the prevalence of baobab Adansoniadigitata within the patch (tree species known for <1000 cm dbh).

Table III GLM analysis of habitat variables determining occupancy and density of Colobus angolensispalliatus

Response variable Predictor variable Parametercoefficient

SE F p

Occupancy a Area .84 .297 14.38 .0004Canopy cover .05 .024 5.96 .02

Individual colobus density b Area –.55 .027 52.43 <.0001Proportion of forest change 1.11 .152 9.33 .005Major food tree density (m2/ha) .17 .033 4.95 .03

aGLM Binomial presence/absence model (null deviance=54.78, residual deviance=37.42, df=43,1).bGLM Poisson density model (null deviance=473.99, residual deviance=156.66, df=29,1).


groups. Mean colobus group size is 6 (median=6), ranging between 2 and 13individuals per group. In 2 of the 33 occupied patches we found only solitaryindividuals. Patterns of abundance varied dramatically between forest patches.Colobus numbers varied between 0 and 110 individuals per patch, while densitiesranged from .04 to 1.29 individuals/ha (4.33–128.95 individuals/km2). Residentcolobus groups inhabited small forest fragments (the smallest being 3.1 ha), whichgave rise to unusually high densities of colobus within the fragments.

Habitat Attributes

The GPS mapping and ArcView GIS analysis show that forest areas ranged between 1and 1417 ha (median=35 ha, average=166 ha±35 SE, n=46). Forest isolation distancesalso ranged from .08 to 6.64 km (median=.32 km, average=.96 km±.23 SE, n=46).

0

20

40

60

80

100

0.1 1 10 100 1000 10000

Forest area (ha)

Mea

n c

ano

py

cove

r (%

)

empty

occupied

Fig. 3 Occupancy patterns of Colobus angolensis palliatus within forest fragments.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1 10 100 1000 10000

Forest area (ha)

Co

lob

us

den

sity

(i

nd

ivid

ual

s / h

a)

Fig. 4 Population density of Colobus angolensis palliatus in forest fragments, Kenya.


From the 875,217 m2 of vegetation transects we enumerated 325 tree species in 53families. From the tree list it was possible to establish 116 species as food trees ofColobusangolensis palliatus (Colobus Trust, unpub. data) and thus calculate food treeabundance (basal area of food trees m2/ha) in each patch. Fourteen tree species makeup >75% of the colobus diet in the Diani Forest (Table II).

Overall resource abundance, as measured by basal area m2/ha for all trees, ishighly variable between patches, regardless of forest area (Pearson correlation; n=46, r=–.08, p=.6). We found not only considerable variation in overall tree densitybut also that the availability of colobus food and major-food resources also differedbetween forests (Fig. 2). The severity of forest disturbance also ranged widely from3% to 44% basal area (m2/ha) tree removal between patches (median 10%).Extraction practices included local firewood, building pole and medicinal barkcollection (small-scale disturbance), as well as illegal hardwood timber logging andcharcoal production (large-scale clearance).

Occupancy of forest patches by colobus relates positively to forest area and todegree of canopy cover (Table III). The presence of colobus thus becomesincreasingly rare as patch area diminishes and canopy cover declines (Fig. 3). Thecombination of the 2 factors explained 31.7% of the variance in occupancy ofcolobus populations. None of the remaining 11 habitat attributes (Table I) predictedoccupancy in the model.

In the occupied patches, colobus density is positively associated with both theproportion of forest change and the density of major food trees >10 cm DBH, andnegatively associated with patch area (Table III). Though colobus density increasedin smaller patches (Fig. 4), the effect was evidently a relatively small one, becausethe absolute number of colobus still showed a significant positive relationship withforest area (regression; B=.49, t=6.32, df=31, p<.0001, R2

adjusted ¼ :55 Fig. 5).Overall, our model, encompassing the 3 explanatory variables, accounted for 67.1%of the variance in colobus density. None of the other 10 habitat attributes (Table I)explained any further variance.

1

10

100

1000

1 10 100 1000 10000

Forest area (ha)

To

tal p

op

ula

tio

n s

ize

(No

. in

div

idu

als)

R2= 0.55

Fig. 5 Population size of Colobus angolensis palliatus in forest fragments, Kenya.


Discussion

Our study highlights the wide range of population responses that a colobine speciescan exhibit to ongoing habitat modification, fragmentation, and loss. Importantly, theoccurrence and abundance of Colobus angolensis palliatus in Kenyan coastal forestfragments are significantly influenced by a number of habitat attributes, encompass-ing multiple aspects of forest spatial, resource, structural and disturbance character-istics. We now review the effects in more detail and consider their implications forthe conservation management of the Angola black-and-white colobus in the coastalforests of Kenya.

Forest Spatial Attributes

Forest area is the most significant of all habitat variables, strongly affecting patternsof both patch occupancy and density in Colobus angolensis palliatus. The formereffect might be expected given our initial hypothesis (Table I) and the widespreadpositive species-area relationships that island biogeography theory (MacArthur andWilson 1967) and more recently metapopulation theory (Hanski and Gilpin 1991)have explicitly addressed. In contrast, recent colobine studies ofProcolobus rufomitratusin Tana River, Kenya (Mbora 2004) and Procolobus badius and Colobus guereza inKibale, Uganda (Onderdonk and Chapman 2000) showed no evidence of a relationshipbetween species occupancy and patch area. Mbora (2004) suggested possiblemetapopulation dynamics could be influencing the observation of suitable unoccupiedpatches within the Tana River system, while Onderdonk and Chapman (2000)highlighted a high degree of dietary flexibility within the Kibale Forest colobinesallowing species to exist as remnant populations in small patches. However, rather thaninfer any underlying difference between the species and Colobus angolensis palliatus,the disparity between our findings and the results may simply be the result of a greaternumber and size range of forest patches within our analysis that has allowed us todetect the effect. Studies of another folivorous primate, Aloutta palliata, may highlightthis point, as specific incidence increases in larger patches across 64 forest fragments inSouthern Veracruz, Mexico (Rodriguez-Toledo et al. 2003).

The relationship between forest area and primate density is more complex.Though colobus population size increased with increasing patch area (Fig. 5), theincrease was not proportional, resulting in lower colobus densities in larger forestpatches (Fig. 4). This is very similar to patterns of abundance in Alouatta palliatathat Rodriguez-Toledo et al. (2003) documented. There are 2 possible reasons for theincrease in density in smaller patches. First, it may be the outcome of randomprocesses, and not a real effect because small absolute deviations from the expectednumber of groups counted in the smallest patches can lead to very high densities,purely at random. Owing to the density analysis excluding unoccupied patches, thehigh densities are not then offset by zero counts, and we would therefore expect anegative relationship between patch area and density to arise without any underlyingfunctional pattern. Second, small patches with high colobus densities may be theones that have recently undergone substantial contraction of area, and in whichdensities of colobus have not yet adjusted to a new equilibrium (Cowlishaw and


Dunbar 2000). Our analysis of the effects of proportional forest change does notsupport this possibility. However, we cannot rule this process out entirely as ouranalysis measured change over a 12–yr period, which may be inadequate time fordemographic readjustment to take place where habitat loss has occurred morerecently, e.g., in the last 2–3 yr.

We found no evidence for the expected positive relationship between colobusabundance/occupancy and the extent of edge habitat, i.e., the forest area-to-perimeterratio (Johns 1987; Johns and Skorupa 1987; Mbora 2004). Colobus guereza differfrom C. angolensis palliatus in southern Kenya, in that they principally exploit theunderstory of forests instead of high-canopy species (Oates 1977). They also thrive inmoderately disturbed habitat (Chapman et al. 2004; Johns 1985; Skorupa 1986),feeding on the proliferation of young leaf growth and vines that accompany theincreased light conditions within forest gaps and edges (Coley and Barone 1996;Ganzhorn 1995). Therefore, the absence of habitat edge effects on Colobus angolensispalliatus may highlight different responses to habitat fragmentation between thespecies as a result of differences in feeding ecology. In fact, the relationship betweenproportional forest change and density of Colobus angolensis palliatus suggests thatthe species may be negatively affected by the amount of habitat edge, because asforest is progressively lost a reduction in forest area will increase the amount of edgehabitat within each patch (Bender et al. 1998). Further differences in the feedingecology and habitat use of black-and-white colobine species can also be found in adetailed comparison of niche separation in sympatric Colobus guereza and C.angolensis in the Ituri Forest of northeastern Zaire (Bocian 1997).

Abundance and occupancy patterns revealed no effect of patch isolation, eventhough patch isolation distances varied between .08 and 6.64 km. This suggeststhat dispersal by Colobus angolensis palliatus between patches is not stronglylimited by distance in the system, i.e., the intervening matrix between fragments(Berggren et al. 2001; Gascon et al. 1999; Ricketts 2001) may be sufficientlypermeable to make both recolonization and rescue effects (Brown and Kodric-Brown 1977) equally likely at both short and long isolation distances. In fact, wesighted colobus in various matrix vegetation types, e.g., mangrove, perennialcropland, coastal shrubland, and between forest patches, in the course of the fieldresearch (Anderson et al. 2007), and further analyses will hopefully clarify theextent and relative importance of the additional habitat variable in primateoccupancy patterns in the coastal forests of Kenya.

Forest Resource Attributes

The abundance of major food resources positively influences colobus density. Theinfluence of 1 of the 14 tree species; the baobab Adansonia digitata, with itsunusually large trunk diameter <1000 cm (Beentje 1994), could have biased resultsthrough an overestimation of major food availability, because basal area is used as arelative index of canopy cover and thus food availability (Decker 1994). However,when basal areas of Adansonia digitata were substituted with mean basal areas offood trees, which excluded A. digitata in the calculation, and were patch specific, thesignificance of the variable remained constant.


It is particularly interesting that neither density nor diversity of all trees, food trees,or major food trees predicted occupancy, while only the latter predicted density. It maybe that Colobus angolensis palliatus exhibit a high degree of dietary flexibility, eatinglow-quality, less preferred foods when required, allowing populations to occupypatches when resource quality is relatively low. The large number of indigenous treespecies (n=116) the colobus exploited in the south coast of Kenya, may alreadyhighlight the phenomenon. However, the relationship between major food treedensity and colobus abundance indicates that, once a patch is occupied, the key localfood resources can play an important role in supporting high colobus populationdensities. Researchers have also found the importance of food trees in predictingcolobine abundance in a number of African study sites (Chapman and Chapman1999; Decker 1989; Mbora 2004; Skorupa 1986). In addition, Chapman andChapman (1999) discovered that dietary compositions varied dramatically betweencolobus populations in Kibale forest fragments, Uganda. A similar situation could beinferred from the coastal forest fragments of Kenya, given the differences in foodresource distribution (Fig. 2.) A clearer validation of the result would come fromfurther studies of the feeding ecology of Colobus angolensis palliatus in differingforest fragments, as well as extended research into the nutrient quality (Moreno-Blackand Bent 1982), especially the protein and fiber content (Chapman et al. 2004; Oateset al. 1990; Wasserman and Chapman 2003) of colobus food plants in the region.

Forest Structural Attributes

In light of current theories that focus on the geometry of forest fragments to explainpatterns of occupancy (Hanski and Gilpin 1991), it is interesting to note acharacteristic of habitat quality, i.e., canopy cover, is a significant predictor ofpopulation incidence. Umapathy and Kumar (2000) found a similar effect of foreststructure in south India: 2 arboreal primate species, Trachypithecus johnii andMacaca silenus, were more likely to occur in fragments with high tree density,canopy height, and canopy cover. Our result is likely to reflect the fact that canopycover is extremely important to an arboreal primate that relies heavily on continuouscanopy to gain access to food resources (Arosen 2004; McGraw 1998b) and also toavoid predators. Researchers have documented avian predation on colobines in Eastand West Africa (Struhsaker 2000; Struhsaker and Leakey 1990) and linked highrates of terrestrial predation of Presbytis entellus in India to low canopy cover (Rossand Srivastava 1994). The same may apply to Colobus angolensis palliatus in theKwale District, where sea eagle predation on primates occurs (pers. obs) and brokenforest canopies increase the proportion of time colobus spend on the ground, thusincreasing the frequency of incidental encounters with feral and hunter’s dogs. Theeffects occurred in 1 colobus group that contained no adult as a result of repeateddog attacks (confirmed through local interviews).

Forest Disturbance Attributes

Colobus density was significantly affected by the proportion of forest changebetween 1989 and 2001: colobus density was higher in patches with increasingforest area but, conversely, was lower in patches with declining forest area. Though


some forest patches expanded, probably through the abandonment of agriculturalplots on their boundaries (Ganade 2001) and the implementation of Kenyan ForestryDepartment plantation schemes over the past 12 yr (Marshall and Jenkins 1994), theforest patch gains were in a clear minority. Thirty-eight of the patches suffered 3–96% declines in forest coverage with corresponding reductions in colobus numbers.The combined effects of reduced patch size, increased resource competition, anddecreased connectivity of the landscape (Bender et al. 1998), may be jointlyresponsible for the changes in colobus abundance beyond the ones associated withhabitat area alone.

More direct measures of disturbance, as measured by the density of tree removal,paths, roads, snares, presence of charcoal activities and proximity of humansettlement, do not significantly affect colobus abundance as independent variables,which need not mean that they lack influence in particular cases, but that there was noconsistent effect overall that we could detect in the sample. There can be little doubtthat Colobus angolensis is vulnerable to human disturbances, given that a study in theIturi Forest, Zaire showed that the abundance of colobus was more than halvedbetween mixed and logged forests (Thomas 1991). In fact, our analyses may indicatelonger-term effects of disturbance on colobus populations, because 2 of our 4significant habitat measures—canopy cover and food tree density—may reflecthistorical disturbance that is now detectable only through the structural/compositionalforest attributes.

There is a major degree of resource overlap between colobus food trees andhuman extraction practices. From our analyses of absolute density of tree damage weidentified 216 tree species that were logged for hardwood timber, woodcarving,domestic timber, fuelwood, and charcoal within the forests of the Kwale District.More than 45% of the extraction targeted just 10 species of indigenous tree. Nine ofthe species are food trees of the colobus while 4 are major food trees: Millettiausaramensis, Combretum schumannii, Grewia sp., and Lecaniodiscus fraxinifolius.Given the relative importance of the food trees for colobus abundance and thecontinued extraction by humans, it may only be a matter of time before decliningpopulations of Colobus angolensis palliatus or extinction responses, or both, occurin response to the permanent removal of the major food resources (Decker 1989;Skorupa 1986).

Conclusion

Our study highlights the negative impact of coastal forest habitat destruction onColobus angolensis palliatus in the Kwale District, Kenya. Identification of theprecise mechanisms responsible for the variety of occupancy and abundance patternsobserved in primate studies is extremely complex, particularly with regard toindependent measures of forest disturbance. However, our GLMs were successful inattributing >65% of the variance in colobus density, and 32% of the variance inoccupancy, to habitat attributes. The latter response still has a considerable amountof variance left unexplained: a more comprehensive analysis of 124 patch spatialattributes and Kwale District matrix structure may provide a better understanding ofpatch dynamics, and subsequent colobus occupancy patterns, in a regional


metapopulation context (Gustafson and Gardner 1996; Hanski 1999; Vandermeerand Carvajal 2001).

Forest loss and ongoing tree extraction in the Kwale District is a dynamic andongoing process, even within protected Forest Reserves and Kayas (sacred localforests, gazetted as National Monuments; Robertson and Luke 1993). The high degreeof human and colobus resource overlap, with local human populations showing apropensity for extracting the major food trees of the colobus, has serious implications.As a result of ongoing forest loss, and the extraction of food trees in the remainingforests, affecting both the availability of food resources and the structure of the forestcanopy, it is very likely that we shall witness future declines of Colobus agolensispalliatus, and increased population extinctions, over the coming years. It is essentialto maintain large, closed-canopy forests within the District and to restore degradedhabitat wherever possible. This will require improved law enforcement of illegallogging, forest management, and the promotion of alternative human resources.

Acknowledgments We thank the Government of Kenya for permitting this research (permit MOEST 13/001/31C 58) and the following institutions, organizations, and individuals who made the work possible:The Natural Environment Research Council, Humane Society of the United States, Born Free Foundation,and Alistair Voller Travel Fund for financial support; Wakuluzu, Friends of the Colobus Trust, DianiBeach Kenya, for logistics, research support, data collaboration, local primate/botanical field researchers,volunteers, and their help in the early conception of this research project; the Coastal Forest ConservationUnit (CFCU), National Museums of Kenya and Kenyatta University for botanical survey teams; KenyaWildlife Service, Kenya Forestry Department, CFCU and local Kaya elders for permission to conductresearch and for logistical and field-staff support. We give special thanks to Quentin Luke, Hamisi Pakia,and Bakari Garise for additional botanical assistance.

References

Anderson, J. (2005). Habitat fragmentation and metapopulation dynamics of the Angola black-and-whitecolobus (Colobus angolensis palliatus) in coastal Kenya. PhD. thesis. London: University CollegeLondon.

Anderson, J., Rowcliffe, M., & Cowlishaw, G. (2007). Does the matrix matter? A forest primate in acomplex agricultural landscape. Biological Conservation, 135, 212–222.

Arosen, G. P. (2004). The energetics of positional behaviour and support use in arboreal primates:Conservation implications. Folia Primatologica, 75, 233.

Beentje, H. (1994). Kenya Trees, Shrubs and Lianes. Nairobi, Kenya.Bender, D. J., Contreras, T. A., & Fahrig, L. (1998). Habitat loss and population decline: A meta-analysis

of the patch size effect. Ecology, 79, 517–533.Berggren, A., Carlson, A., & Kindvall, O. (2001). The effect of landscape composition on colonization

success, growth rate and dispersal in introduced bush-crickets (Metrioptera roeseli). Journal ofAnimal Ecology, 70, 663–667.

Bocian, C. M. (1997).Niche Separation of Black-and-White Colobus Monkeys (Colobus angolensis and C.guereza) in the Ituri Forest. Ph.D. program in Biology, Graduate Centre, City University of NewYork.

Brown, J. H., & Kodric-Brown, A. (1977). Turnover rates in insular biogeography: Effect of immigrationon extinction. Ecology, 58, 445–449.

Butynski, T., & Mwangi, G. (1994). Conservation status and distribution of the Tana River Red Colobusand Crested Mangabey. Zoo Atlanta/Kenya Wildlife Service/National Museums of Kenya/Institute ofPrimate Research/East African Wildlife Society, Nairobi.

Chapman, C. A., & Chapman, L. J. (1999). Implications of small scale variation in ecological conditionsfor the diet and density of red colobus monkeys. Primates, 40, 215–231.


Chapman, C. A., & Chapman, L. J. (2002). Foraging challenges of red colobus monkeys: influence ofnutrients and secondary compounds. Comparative Biochemistry and Physiology ¯ Part A: Molecularand Integrative Physiology, 133, 861–875.

Chapman, C. A., Chapman, L. J., Bjorndal, K. A., & Onderdonk, D. A. (2002). Application of protein-to-fiber ratios to predict colobine abundance on different spatial scales. International Journal ofPrimatology, 23, 283–231.

Chapman, C. A., Chapman, L. J., Naughton-Treves, L., Lawes, M. J., & McDowell, L. R. (2004).Predicting folivorous primate abundance: Validation of a nutritional model. American Journal ofPrimatology, 62, 55–69.

Clarke, G. P. (1998). A new regional centre of endemism in Africa. In C. R. Huxley, J. M. Lock, & D. F.Cutler (Eds.), Chorology, Taxonomy ad Ecology of the Floras of Africa and Madagascar (pp. 53–65).Kew: Royal Botanical Gardens.

Clarke, G. P. (2000). Defining the eastern African Coastal Forests. In N. D. Burgess, & G. P. Clarke(Eds.), Coastal Forests of Eastern Africa (pp. 9–27). Gland, Switzerland: IUCN.

Coley, P. D., & Barone, J. A. (1996). Herbivory and plant defenses in tropical forests. Annual Review ofEcology and Systematics, 27, 305–335.

Connor, E. F., Courtney, A. C., & Yoder, J. M. (2000). Individuals-area relationships: The relationshipbetween animal population density and area. Ecology, 81, 734–748.

Cowlishaw, G. (1999). Predicting the pattern of decline of African primate diversity: An extinction debtfrom historical deforestation. Conservation Biology, 13, 1183–1193.

Cowlishaw, G., & Dunbar, R. (2000). Habitat disturbance. In G. Cowlishaw, & R. Dunbar (Eds.), PrimateConservation Biology (pp. 191–241). Chicago: University of Chicago Press.

Crawley, M. J. (1993). GLIM for Ecologists. Oxford: Blackwell.Davies, A. G. (1987). The Gola forest reserves, Sierra Leone: Wildlife conservation and forest

management. Gland, Switzerland: IUCN.Davies, A. G. (1994). Colobine populations. In A. G. Davies, & J. F. Oates (Eds.), Colobine Monkeys: Their

Ecology, Behaviour and Evolution (pp. 285–231). Cambridge, UK: Cambridge University Press.Davies, A. G., Oates, J. F., & Dasilva, G. L. (1999). Patterns of frugivory in three West African colobine

monkeys. International Journal of Primatology, 20, 327–357.Decker, B. S. (1989). Effects of habitat disturbance on the behavioural ecology and demographics of the

Tana River Red Colobus (Colobus badius rufomitratus). Ph.D. Thesis, Emory University, Atlanta,Georgia.

Decker, B. S. (1994). Effects of habitat disturbance on the behavioral ecology and demographics of the Tanariver red colobus (Colobus badius rufomitratus). International Journal of Primatology, 15, 703–737.

Dunbar, R. I. M. (1987). Habitat quality, population-dynamics, and group composition in colobusmonkeys (Colobus-Guereza). International Journal of Primatology, 8, 299–329.

Estrada, A., & Coates-Estrada, R. (1996). Tropical rain forest fragmentation and wild populations ofprimates at Los Tuxtlas, Mexico. International Journal of Primatology, 17, 759–783.

Fashing, P. J., Plumptre, A. J., Mullindahabi, F., Gakima, J. B., Masozera, M., Munanura, I. et al. (2004). Rangingpatterns of an unusually large Colobus angolensis group in Rwanda. Folia Primatologica, 75, 23.

Fimbel, C., Vedder, A., Dierenfeld, E., & Mulindahabi, F. (2001). An ecological basis for large group sizein Colobus angolensis in the Nyungwe Forest, Rwanda. African Journal of Ecology, 39, 83–92.

Ganade, G. (2001). Forest restoration in abandoned pastures of central Amazonia. In R.O. Bierregaard, C.Gascon, T. E. Lovejoy, & R. C. G. Mesquita (Eds.), Lessons from Amazonia (pp. 313–324). NewHaven, CT: Yale University Press.

Ganzhorn, J. U. (1995). Low-level forest disturbance effects on primary production, leaf chemistry, andlemur populations. Ecology, 76, 2048–2096.

Gascon, C., Lovejoy, T. E., Bierregaard, R. O., Malcolm, J. R., Stouffer, P. C., Vasconcelos, H. L. et al. (1999).Matrix habitat and species richness in tropical forest remnants. Biological Conservation, 91, 223–229.

Grieg-Smith, P. (1983). Quantitative Plant Ecology, 3rd ed. Oxford: Blackwell.Gustafson, E. J., & Gardner, R. H. (1996). The effect of landscape heterogeneity on the probability of

patch colonization. Ecology, 77, 94–107.Hanski, I. (1999). Habitat connectivity, habitat continuity, and metapopulations in dynamic landscapes.

Oikos, 87, 209–219.Hanski, I., & Gilpin, M. (1991). Metapopulation dynamics- brief-history and conceptual domain.

Biological Journal of the Linnean Society, 42, 3–16.Hawthorne, W. D. (1993). East African Coastal Forest botany. In J. C. Lovett, & S. K. Wasser (Eds.),

Biogeography and Ecology of the Rain Forests of Eastern Africa (pp. 57–99). Cambridge, UK:Cambridge University Press.


HSEDCO (1998). Environmental baseline data report of Kwale prospecting area and ship loading facility.Human Settlements, Environment and Development Collaborative, Nairobi.

IUCN (1996). African primates. status survey and conservation action plan. Revised edition. Gland,Switzerland: IUCN.

Jaetzold, R., & Schmidt, H. (1983). Farm management handbook of Kenya. Vol II: Natural conditions and farmmanagement information. Part C—East Kenya. Ministry of Agriculture, Kenya and GTZ, Germany.

Johns, A. D. (1985). Differential detectability of primates between primary and selectively logged habitatsand implications for population surveys. American Journal of Primatology, 8, 31–36.

Johns, A. D. (1987). Long-term responses of a west Malaysian primate community to selective logging.International Journal of Primatology, 8, 511–511.

Johns, A. D., & Skorupa, J. P. (1987). Responses of rainforest primates to habitat disturbance-a review.International Journal of Primatology, 8, 157–191.

Kahumbu, P. (1997). The decline of Angolan colobus monkeys Colobus angolensis palliatus in Diani.EANHS Bulletin, 27, 2–4.

Kanga, E. M. (2001). Survey of black and white colobus monkeys (Colobus angolensis palliatus) inShimba Hills National Reserve and Maluganji Sanctuary, Kenya. ASP Bulletin, 25, 8–9.

Karere, G. M., Oguge, N. O., Kirathe, J., Muoria, P. K., Moinde, N. N., & Suleman, M. A. (2004).Population sizes and distribution of primates in the lower Tana River forests, Kenya. InternationalJournal of Primatology, 25, 351–365.

Kingdon, J. (1997). The Kingdon field guide to African Mammals. London: Academic Press.Lahm, S. A. (1993). Ecology and economics of Human/Wildlife interaction in Northeastern Gabon. Ph.D.

thesis, New York University., New York.MacArthur, R. H., & Wilson, E. O. (1967). The theory of Island Biogeography. Princeton, NJ: Princeton

University Press.Marsh, C.W. (1986). A re-survey of the Tana River primates and their habitat. Primate Conservation, 7, 72–81.Marsh, C. W., Johns, A. D., & Ayres, J. M. (1987). Effects of habitat disturbance on rain forest primates.

In A. L. I. Riss (Ed.), Primate Conservation in the Tropical Rain Forest (pp. 83–107).Marshall, N. T., & Jenkins, M. (1994). Hard times for Hardwood: Indigenous Timber and the Timber

Trade in Kenya. Cambridge, United Kingdom: TRAFFIC International.Martin, C., & Asibey, E. O. A. (1979). Effect of timber exploitation on primate population and distribution

in the Bia rainforest area of Ghana. In 7th Congress of the International Primatological Society,Bangalore, India.

Mate, C., Colell, M., & Escobar, M. (1995). Preliminary observations on the ecology of forestcercopithecidae in the Lokofe-Ikomaloki region (Ikela, Zaire). Folia Primatologica, 64, 196–200.

Mbora, D. N. M. (2004). Forest fragmentation and the distribution, abundance and conservation of theTana River Red Colobus (Procolobus rufomitratus). Biological Conservation, 118, 67–77.

McGraw, W. S. (1998a). Comparative locomotion and habitat use of six monkeys in the Tai forest, IvoryCoast. American Journal of Physical Anthropology, 105, 493–451.

McGraw, W. S. (1998b). Posture and support use of old world monkeys (Cercopithecidae): The influenceof foraging strategies, activity patterns, and the spatial distribution of preferred food items. AmericanJournal of Primatology, 46, 229–25.

Medley, K. E. (1993). Primate conservation along the Tana river, Kenya-an examination of the foresthabitat. Conservation Biology, 7, 109–121.

Moreno-Black, G. S., & Bent, E. F. (1982). Secondary compounds in the diet of Colobus angolensis.African Journal of Ecology, 20, 29–36.

Muchoki, C. H. K. (1990). Land use in Kwale district. Department of Resource Surveys and RemoteSensing, Ministry of Planning and National Development, Kenya., Nairobi.

Muoria, P. K., Karere, G. M., Moinde, N. N., & Suleman, M. A. (2003). Primate census and habitatevaluation in the Tana delta region, Kenya. African Journal of Ecology, 41, 157–163.

Oates, J. F. (1977). The guereza and its food. In T. H. Clutton-Brock (Ed.), Primate Ecology (pp. 275–321). New York: Academic Press.

Oates, J. F. (1996). Habitat alteration, hunting and the conservation of folivorous primates in Africanforests. Australian Journal of Ecology, 21, 1–9.

Oates, J. F., Whitesides, G. H., Davies, A. G., Waterman, P. G., Green, S. M., Dasilva, G. L. et al. (1990).Determinants of variation in tropical forest primate biomass-new evidence from West Africa. Ecology,71, 328–343.

Onderdonk, D. A., & Chapman, C. A. (2000). Coping with forest fragmentation: The primates of KibaleNational Park, Uganda. International Journal of Primatology, 21, 587–611.


Plumptre, A. J., & Reynolds, V. (1994). The effect of selective logging on the primate populations in theBudongo Forest Reserve, Uganda. Journal of Applied Ecology, 31, 631–641.

Ricketts, T. H. (2001). The matrix matters: Effective isolation in fragmented landscapes. AmericanNaturalist, 158, 87–99.

Robertson, S. A., & Luke, W. R. Q. (1993). Kenya Coastal Forests: The Report of the NMK/WWF CoastForest Survey, National Museums of Kenya, Nairobi.

Rodgers, W. A. (1981). The distribution and conservation status of colobus monkeys in Tanzania.Primates, 22, 33–45.

Rodgers, W. A., & Homewood, K. M. (1982). Biological values and conservation prospects for the forestsand primate populations of the Uzungwa mountains, Tanzania. Biological Conservation, 24, 285–304.

Rodriguez-Toledo, A. M., Mandujano, S., & Garcia-Orduna, F. (2003). Relationships between forestfragments and hower monkeys (Aloutta palliata mexicana) in Southern Veracruz, Mexico. In L. K.Marsh (Ed.), Primates in Fragments (pp. 79–97). New York: Kluwer Academic/Plenum.

Ross, C., & Srivastava, A. (1994). Factors influencing the population density of the hanuman langur(Presbytis entellus) in Sariska tiger reserve. Primates, 35, 361–367.

Sayer, J. A., Harcourt, C. S., & Collins, N. M. (1992). The conservation Atlas of tropical forests: Africa.London: Macmillan.

Skorupa, J. P. (1986). Responses of rainforest primates to selective logging in Kibale forest, Uganda: Asummary report. In K. Benirshcke (Ed.), Primates: The Road to Self-sustaining Populations (pp. 57–7). Berlin: Springer-Verlag.

Struhsaker, T. T. (1981a). Census methods for estimating densities. In Techniques for the Study of PrimatePopulation Ecology (pp. 36–38). Washington, DC: National Academy Press.

Struhsaker, T. T. (1981b). Forest and primate conservation in East Africa. African Journal of Ecology, 19,99–114.

Struhsaker, T. T. (2000). The effects of predation and habitat quality on the socioecology of Africanmonkeys: Lessons from the islands of Bioko and Zanzibar. In P. F. Whitehead, & C. J. Jolly (Eds.),Old World Monkeys. Cambridge, UK: Cambridge University Press.

Struhsaker, T. T., & Leakey, M. (1990). Prey selectivity by crowned hawk-eagles on monkeys in theKibale Forest, Uganda. Behavioral Ecology and Sociobiology, 26, 435–443.

Struhsaker, T. T., & Siex, K. S. (1998). The Zanzibar red colobus: Conservation status of an endangeredisland endemic. Primate Conservation, 18, 51–58.

Tarara, E. (1986). Results of the Swara readers survey on black and white colobus. Swara, 10, 23–24.Thomas, S. C. (1991). Population densities and patterns of habitat use among anthropoid primates of the

Ituri forest, Zaire. Biotropica, 23, 68–83.Umapathy, G., & Kumar, A. (2000). The occurrence of arboreal mammals in the rain forest fragments in

the Anamalai Hills, south India. Biological Conservation, 92, 311–319.UNEP (1982). Environmental problems of the East African Region, Nairobi, Kenya.Vandermeer, J., & Carvajal, R. (2001). Metapopulation dynamics and the quality of the matrix. American

Naturalist, 158, 211–222.Vedder, A., & Fashing, P. J. (2002). Diet of a 300-member Angolan colobus monkey (Colobus

angolensis) supergroup in the Nyungwe Forest, Rwanda. American Journal of PhysicalAnthropology, 159–16.

Wasserman, M. D., & Chapman, C. A. (2003). Determinants of colobine monkey abundance: theimportance of food energy, protein and fibre content. Journal of Animal Ecology, 72, 650–659.

White, F. (1976). The vegetation map of Africa: The history of a completed project. Boissiera, 24, 659–666.

Whitesides, G. H., Oates, J. F., Green, S. M., & Kluberdanz, R. P. (1988). Estimating primate densitiesfrom transects in a West African rainforest-a comparison of techniques. Journal of Animal Ecology,57, 345–367.


Documents

Effects of Forest Fragmentation on the Abundance of Colobus angolensis palliatus in Kenya's