designing the masoala national park in madagascar based on

1055

Conservation Biology, Pages 1055–1068Volume 13, No. 5, October 1999

Designing the Masoala National Park in Madagascar Based on Biological and Socioeconomic Data

CLAIRE KREMEN,*‡ VINCENT RAZAFIMAHATRATRA,† R. PHILIP GUILLERY,‡ JOCELYN RAKOTOMALALA,§ ANDREW WEISS,** ANDJEAN-SOLO RATSISOMPATRARIVO§

*Wildlife Conservation Society and Center for Conservation Biology, Department of Biological Sciences, Stanford University, Stanford, CA 94305, U.S.A., email [email protected]†Département de Biologie Animale, Faculté des Sciences, Université de Madagascar, Antananarivo, Madagascar‡Wildlife Conservation Society, 185th Street and Southern Boulevard, Bronx, NY 10460, U.S.A.§CARE International Madagascar, B.P. 1677, Antananarivo, Madagascar**Center for Conservation Biology, Stanford University, Stanford, CA 94305, U.S.A.

Abstract:

Conservation biologists have developed powerful tools for reserve selection and design over thepast two decades, yet seldom are protected areas actually designed on scientific grounds. Using fundamentalbiological and socioeconomic principles of conservation science, we designed a new protected area and itsmultiple-use zone on the Masoala Peninsula in the humid forest zone of Madagascar. The explicit design cri-teria determined the data gathered for the work, which included (1) spatial distribution and quality of habi-tat, (2) the areas and species at greatest risk, (3) the relationship between environmental gradients and spe-cies distributions, (4) current and predicted human settlement and land and resource use, and (5) theeconomic potential of natural forest management as an alternative to deforestation. We used a geographicinformation system to integrate these data layers and applied the design criteria to develop a park proposalthat balanced human and wildlife needs. The proposal won the approval of local residents, and a national de-

cree in 1997 designated 2100 km

2

of rainforest and three satellite marine reserves as the Masoala NationalPark, with a surrounding multiple-use zone of approximately 1000 km

2

. The new park is Madagascar’s larg-

est protected area and protects more lowland (

,

400 m) humid forest habitat than the entire reserve systemcombined, a significant step forward in conserving a globally important ecoregion. Consideration of localneeds and the national economy was a key element in gaining approval for the Masoala Park. Such an ap-proach toward reserve design could be applied elsewhere to improve chances of establishing and maintainingprotected areas over the long term.

Diseño del Parque Nacional Masola en Madagascar en Base Datos Biológicos y Socioeconómicos

Resumen:

Durante las últimas dos décadas, los biólogos conservacionistas han desarrollado herramientaspoderosas para la selección y diseño de reservas, sin embargo, aún son pocas las áreas protegidas que se dis-eñan realmente con bases cientíificas. Utilizando los principios biológicos y socioeconómicos fundamentalesde la ciencia de la conservación, diseñamos una nueva área protegida y su zona de usos múltiples en la Pen-insula Masola en la zona de bosque húmedo de Madagascar. Los criterios explícitos del diseño determinaronlos datos recogidos para el trabajo, los cuales incluyeron: (1) distribución espacial y calidad del hábitat, (2)áreas y especies en mayor riesgo, (3) relaciones entre gradientes ambientales y distribución de especies, (4)establecimiento actual y predecido de humanos y uso del suelo y los recursos y (5) potencial económico delmanejo del bosque natural como una alternativa a la deforestación. Utilizamos SIG para integrar estas ca-pas de datos y aplicamos el criterio del diseño para desarrollar una propuesta del parque que equilibrienecesidades humanas y de la vida silvestre. La propuesta ganó la aprobación de los residentes locales y un de-

Paper submitted July 17, 1998; revised manuscript accepted February 17, 1999.

1056

Designing a National Park Kremen et al.

Conservation BiologyVolume 13, No. 5, October 1999

creto nacional en 1997, mismo que designó 2100 Km

2

de bosque lluvioso y tres reservas marinas satélitecomo el Parque Nacional Masola, con una zona aledaña de usos múltiples de aproximadamente 1000 Km

2

.El nuevo parque es el área protegida mas grande de Madagascar y protege mas tierras bajas (

,

400m) dehábitat de bosque húmedo que la totalidad del sistema de reservas combinado. Un paso significativo en laconservación de una ecoregión globalmente importante. La consideración de las necesidades locales y de laeconomía nacional fueron elementos clave para ganar la aprobación del Parque Masola. Este tipo de aproxi-mación para el diseño de reservas puede ser aplicado en cualquier otra parte para mejorar las posibilidades

de establecer y mantener áreas protegidas a largo plazo.

Introduction

We count on protected areas to conserve the world’sbiodiversity, yet few reserves were originally designedwith conservation of biodiversity in mind (Terborgh &Winter 1983; Scott et al. 1987; Pressey et al. 1996). In-stead, many reserves were established because of spec-tacular landscape features or the lack of competing landuses. Consequently, many reserves have proved toosmall to maintain viable populations of species requiringlarge areas (Newmark 1985, 1996; Wallis De Vries 1995;Noss et al. 1996) or are poorly configured for optimumprotection of biodiversity (Scott et al. 1987; Margules &Stein 1989). Typically, local interests have been ignored,resulting in continued boundary conflicts and reservedegradation (Western & Gichohi 1993; Smith et al.1997

a

; van Schaik et al. 1997). Despite published scien-tific studies on reserve design (e.g., Soulé 1986; Usher1986; Margules et al. 1988), few have influenced prac-tice (Ehrlich 1997).

Recently, we had the opportunity to design a new na-tional park in Madagascar. Our first step was to con-solidate existing principles of conservation science(here defined as including biological and socioeconomicdisciplines) into a set of design criteria to promote thesustainability of the future reserve. We then used thesecriteria to determine the data-gathering strategy and todevelop and assess the reserve proposal.

Madagascar and the Masoala Peninsula

Madagascar is a global conservation priority (Myers1988; Hannah et al. 1998) because of its rich biota andits exceptionally high levels of endemism (Nicoll & Lan-grand 1989; Jenkins 1990; Goodman 1997) and defores-tation (Green & Sussman 1990; Nelson & Horning1993). Madagascar’s government is striving to protect itsbiodiversity by establishing new protected areas, inte-grated conservation and development programs (ICDPs),and other community-based approaches (World Bank etal. 1988). These conservation strategies aim to protectnatural areas by providing local human populations with

economic incentives to employ sustainable land-usepractices (Wells et al. 1992; Alpert 1996).

The Masoala Peninsula, a land mass of 4265 km

2

innortheastern Madagascar (Fig. 1), is one of the largestforest blocks in Madagascar and a top conservation pri-ority (World Bank et al. 1988) due to its lowland humidforests, now rare elsewhere in Madagascar (Green & Sus-sman 1990; Du Puy & Moat 1996). Our work began in1993, following efforts since 1990 to establish an ICDPproject in the area. The ICDP context dictated that aportion of the forests on the Masoala Peninsula (3000km

2

) would be designated as park, and the rest wouldbe reserved for community-based management. TheICDP strategy consisted of (1) encouraging local peopleto switch from shifting cultivation to permanent agricul-ture in already deforested lands and (2) promoting forestconservation through multiple use. The park designproblem was therefore to demarcate core forest areas to

Figure 1. The location of the Masoala Peninsula in Madagascar, its remaining rain and littoral forest cover, and elevational gradient. Table 2 provides addi-tional site information.


Kremen et al. Designing a National Park

1057

conserve biodiversity and multiple-use zones to providesufficient resources for local human needs.

For ecological viability, design criteria emphasized apark of sufficient size to (1) buffer against natural or an-thropogenic disturbances, (2) represent all habitats andspecies, (3) maintain connectivity between natural ar-eas, and (4) protect viable populations of rare, threat-ened, endangered, or geographically restricted species(Table 1). Criteria to promote socioeconomic sustain-ability aimed to prevent conflicts between local commu-nities and the park through provision of ecosystemgoods and services in the multiple-use zone. Design prin-ciples defined the information needs for park design (Ta-ble 1).

Methods

Site Description

The Masoala peninsula rises from sea level to

.

1200 m(Fig. 1), spanning a gradient from lowland and mid-ele-vation humid evergreen forest (“humid forest”) to mon-tane thicket and cloud forest. The majority of the penin-sula is granitic bedrock, but patches of other soil types

exist (Hottin & Liandrat 1964). These soil types and aneast-to-west rainfall gradient (Andriamampianina 1995)likely influence the vegetation on the peninsula (Gentry1993; Abraham et al. 1996; Du Puy & Moat 1996).

In 1994 the peninsula’s population was approximately44,500 (Banques de Données de l’Etat and MasoalaProject, unpublished data). The principal land use is riceagriculture, carried out by shifting cultivation on rain-fedslopes or in irrigated paddies in valley bottoms. Tradition-ally, land is claimed by clearing and cultivating, often atsubstantial distances from the main village, requiring theestablishment of temporary settlements. Although villag-ers respect one another’s claims, the traditional mecha-nism of claiming land is illegal.

GIS Data Layers and Ground-Truthing

We used geographic information systems (GIS) in work-station Arc/Info (version 7) to integrate the data layers(Table 1). Map-derived layers (topography at 100-m con-tour intervals, hydrology, and geology) were scannedfrom the most recent maps of the Service Géologiqueet Mine (1964) and the 1:100,000 topographic series[(Foiben Tasosarintanin-I Madagsikara (FTM)] from 1957aerial imagery). We developed a digital terrain model

Table 1. Criteria and information needs for designing the Masoala National Park.

Design criteria Information needs

a

Ecological sustainability

b

(1) Largest continuous area consisting of natural habitats must be protected. satellite image analysis and ground-truthing(2) Area to be protected should consist largely of primary or relatively undisturbed

habitats.(3) Area should contain several representative examples (at a minimum) of the

existing habitat types, including the spectrum of environmental gradients.(4) Corridors that link natural habitats should be protected. Corridors must be wide

enough to encourage animal movement or include a zone where forest regeneration can be actively encouraged. Habitat mosaics and transitional zones should be protected.

(5) Special consideration must be given to rare and threatened habitats or endangered habitats and species and to locally endemic species, particularly those not protected elsewhere in Madagascar.

biodiversity surveys; population studies; threats analyses

(6) If there is a choice between two areas, choose the one that will contribute the maximum number of species of interest not already found within the park.

(7) Park limits should be as simple as possible to minimize the edge-to-area ratio. mapping of park limitsSocioeconomic sustainability

c

(8) Human settlements must not exist within the park (Malagasy law). mapping of human settlements(9) Wherever possible, the park boundaries should be placed outside of currently

cultivated lands and traditional harvest zones, including room for expansion.mapping of land-use and tenure; projection

of future land use(10) Forest buffer outside of the park must be large enough to meet the human

population’s commercial and subsistence needs for forest products.forest inventories; socioeconomic surveys;

economic feasibility analyses and projections of resource use

(11) Headwaters of rivers should be included in the park. topographic maps(12) Limits should be easy to see and respect, following geographic features when

possible.(13) There should be points of access for ecotourism and patrolling.

a

Each information need is listed once; blanks indicate a previous listing.

b

Ecological sustainability criteria devised from Pickett and Thompson (1978), Shaffer (1981), Soulé and Simberloff (1986), Usher (1986), Noss(1987), Pernetta et al. (1994), and Polis et al. (1997).

c

Socioeconomic sustainability criteria derived from Hough and Sherpa (1989) and Western and Pearl (1989).

1058



(DTM) from topography and then derived the slope, as-pect, and topographic position layers. Forest cover layerswere developed from a SPOT panchromatic satellite im-age (1991) by manual interpretation, assisted by ground-truthing and aerial overflight data. The Missouri Botani-cal Garden provided 1957 forest-cover data in digitalformat based on a map from the Direction des Eaux etForêts. Settlements, village territories, access (trails, nav-igable reaches of rivers, ports), and biodiversity and for-est inventory sites were digitized from global positioningsystem (GPS) field data collected by means of MagellanNAV 5000 DX (

6

100 m accuracy). We derived a physio-graphic inventory (Rich et al. 1992) using the DTM, hy-drology, geology, 1991 and 1957 forest cover, and topo-graphic position layers.

Panchromatic satellite imagery allows distinction be-tween forested and deforested areas (Wilkie & Finn1996). Ground-truthing of the satellite image was con-ducted by foot in 1994 along the entire border betweenthe main forest block and the deforested zone (Fig. 1)and along each watershed penetrating the main forestblock (until no further signs of deforestation were de-tected). Field workers collected GPS points at approxi-mately 0.5- to 1-km intervals, mapped these points ontoFTM topographic maps enlarged to 1:50,000, and tracedthe intervening forest border continuously, with noteson the condition of the forest (primary, secondary, spe-cial forest type). Fragments of littoral forest–mangrove–

marsh mosaics were identified by their distinctive grayscale and texture on the satellite image and wereground-truthed by the same method.

Biological Inventory

We conducted inventories of selected taxa along five en-vironmental gradients likely to influence species dis-tributions: elevation, rainfall, peninsular effect (Ricklefs1973), parent rock type, and edge effect. The indicatortaxa and references for sampling methods were avifauna(Thorstrom & Watson 1997), primates (Sterling & Rako-toarison 1998), small mammals (nonvolant; Razafindra-koto 1995), butterflies (Kremen et al. 1999

a

), and ci-cindelid and scarab beetles (Andriamampianina 1995;Razafimahatratra & Andriamampianina 1995).

At nine localities (Fig. 1), 5–14 sample sites ortransects were established. Each gradient was dividedinto several zones. Comparisons between zones weremade for each gradient independently so as not to com-pound the potential effects of different gradients (Table2; see also Kremen et al. 1999

a

). Complementarity (ameasure of

b

diversity) was estimated with the Marcze-wski-Steinhaus distance measure (the sum of the num-ber of species unique to each zone, divided by the totalnumber of species found across zones; Pielou 1984).

Human Settlement, Population, Land, and Resource Use

Field agents collected GPS points at the center of eachvillage, and the name, date of settlement, number ofhouseholds and/or population size, and village status(temporary, permanent, or abandoned). Village territo-ries were mapped (by the ground-truthing procedure) asdefined by village participants rather than by legal ten-ure and included both farmlands and areas for collectingforest products. Socioeconomic and agricultural datawere obtained through focus-group interviews in 25 vil-lages (Lance et al. 1995). Data on land tenure were ob-tained from the Service de Domaines and on permits forslash-and-burn farming from the Service Provincielle desEaux et Forêts. Threats to species or habitats wereranked by field staff based on land and resource prac-tices (Lance et al. 1995; Masoala Project, unpublisheddata). Projections of future deforestation were made byGIS, with the physiographic characteristics of previouslydeforested areas used as a predictor.

We estimated the number of hectares of forest neededfor sustainable production of the most important non-commercial uses of wood by volume (fuel, building ca-noes, and house construction and maintenance; Rako-toarisoa 1997; Kremen et al. 1998). The daily volume offirewood was measured on 216 household-days by themethod of Kremen et al. (1999

b

). These and other data(Munasinghe 1993; Raymond 1995; Rakotoarisoa 1997;Kremen et al. 1998) were extrapolated to estimate the

Table 2. Description of the zones compared within each environmental gradient surveyed for biodiversity, and the restrictions applied to each comparison.

Elevationalgradient Localities

a

Numberof sites

b

Restrictions tothe comparison

ElevationLow,

,

400 m W9L2 10 high rainfallgranitic soilsMedium,

400–700 m W9L1 10High,

.

700 m W9L3 5Rainfall

Low, eastern W2L1, W2L2,W7L1, W7L2

26 low elevationgranitic soils

High, western W9L2 20Peninsular effect

Tip, northern W2L1, W2L2 14 low rainfalllow elevationgranitic soils

Base, southern W7L1, W7L2 12

Parent rock typeGranite W2L1, W2L2,

W7L1, W7L215 low rainfall

,

250 m elevationBasalt W2L3 8Sand W7L3 5

Edge effectInterior W2L2, W7L2 14 low rainfall

low elevationgranitic soils

Periphery W2L1, W7L1 12

a

See Fig. 1 for locations.

b

Not all sites at each locality were included in a zonal comparison;inclusion depended on the restriction applied.



1059

annual per-household wood volume. To calculate annualproduction of useful wood, we determined the percent-age of useful species from forest inventories and as-sumed a conservative growth rate for the entire forest of2 m

3

/ha/year (Silva et al. 1995; Bruenig 1996).

Timber Inventory and Economic Analyses

There are no roads on the peninsula, and selective log-ging of hardwoods occurs by hand-felling. Planks arecarried to rivers and rafted on canoes to seaports. Thisunmechanized, selective logging has a relatively lowimpact on the forest (personal observations) comparedto that of other sites in the tropics, where road build-ing alone caused 60% of total damage ( Johns 1997). Toidentify the region where unmechanized, roadless, se-lective logging can occur, we chose the forested areawithin 5 km of the navigable portions of rivers withports. For erosion control we selected areas withslopes

,

10

8

.Within this zone of 71,400 ha, we conducted a forest

inventory to assess the timber resource. Twenty-six in-ventory sites were established, with three circular plotsof 500 m

2

and subplots of 100 m

2

at each site. Tree iden-tity, wood-use category (Direction des Eaux et Forêtsclassification), height to first branch, and diameter atbreast height (dbh) were measured for all stems withdbh

.

20 cm in the large plot and dbh

.

5 cm in thesmall plot. Commercial volumes per ha were estimatedby calculating the roundwood volume for stems

.

40 cmdbh (the minimum cutting diameter in Madagascar) andthen assuming a 30% conversion ratio to sawn wood(from personal observations of improved pit-sawingtechniques in Madagascar).

We collected economic data to analyze the domestic,export, and certified markets for Malagasy timber; to de-velop a cost-benefit analysis of a community-based, sus-tainable forest management plan; and to compare thenet benefits of this plan with a business-as-usual scenario(illegal timber extraction followed by slash-and-burnfarming). Local wood prices were obtained through mar-ket surveys in nearby towns, and free-on-board (FOB)export prices were obtained from importers or deter-mined (for unknown timber species) from the pricepaid for similar wood. Costs (Tables 5 & 7) and laborproductivity were obtained from local woodcutters, ex-porters, foresters, the Service Provincielle des Eaux etForêts, and Rakotoarisoa (1997).

In each case (sustainable management and business-as-usual), we computed benefits based on a communityof 100 households and 20 woodcutters that could pro-duce a maximum of 130 m

3

of sawn wood per year bypit-sawing (Rakotoarisoa 1997; R. Lemaraina, unpub-lished data). In the community-based scenario (based ona pilot project now occurring on the Masoala Peninsula),the management plan called for selective harvest of 12

species over long rotation periods (60 years). In the ab-sence of site- and species-specific growth rates, we useda conservative growth rate (2 m

3

/ha/year) to estimatesustainable, annual allowable cut. We assumed that onlyhalf of the wood produced would be of sufficient qualityfor export and that the other half would be sold domes-tically. The business-as-usual scenario assumed local saleof wood and that forests would be slashed and burnedfollowing wood collection at a constant annual rate of1.5% (based on current national rates; Green & Sussman1990). Net present values (NPVs) for both scenarioswere calculated at 10, 60, and 120 years using a discountrate of 10% (based on the economic rate of return forMadagascar; World Bank 1995) and are reported on in1996 U.S. dollars.

Results

Biological Diversity

High levels of complementarity, indicating high

b

diver-sity, were found across all environmental gradients, par-ticularly for insects and small mammals (Table 3). Eachzone of the rainfall, peninsular, and elevational gradients(defined in Table 2) had a complement of restricted spe-cies (Table 4). Some of these are not yet known fromany other localities in Madagascar and may represent lo-calized endemics; (for example, four of five new butter-fly species discovered during the Masoala inventorywere found in a single zone (Kremen et al. 1999

a

).In contrast, the high complementarity values for

the soils comparison (Table 3) resulted chiefly fromthe absence of species in sandy and basaltic localitiesrelative to granitic. Also, all but two of the speciesfound only in these localities during the inventory(Table 4) had been observed casually in other habitatson the peninsula. Of the two apparently restricted spe-cies, the bird

Caprimulgus enarratus

, is widely distrib-uted throughout Madagascar (Langrand 1990).

Similarly, few species were observed only in theperipheral zone localities during the inventory. Ofthese, all butterfly and bird species were known fromcasual observations elsewhere on the peninsula andmany favored disturbed conditions (Kremen 1992;Kremen et al., 1999

a

). In contrast, 41% of cicindelidswere inventoried only at the peripheral zone localities;but none of these species were restricted to Masoala(Andriamampianina 1996). Ten percent of scarab specieswere found only in the peripheral zone; their habitat as-sociations and distributions are unknown.

We examined the species distribution data set to de-termine the number and type of localities required toprotect species of known special concern. Four locali-ties harbored all of the rare or forest-restricted butter-flies, and these localities included the extremes of rain-

1060



fall, elevation, and peninsular gradients (Fig. 4, definitionof rarity modified from Rabinowitz et al. [1986] asdescribed in Kremen [1999

a

]). The same set of locali-ties contained all the vulnerable and endangered birdsand primates (Fig. 4). None of the peripheral-zone lo-calities (i.e., the areas most threatened by deforesta-tion) were required to protect these species of specialconcern.

Analysis of Threats to Biodiversity

The Masoala project identified and ranked four principaldirect threats to biodiversity: (1) slash-and-burn agricul-ture, (2) overharvest of ebony and rosewood, (3) over-fishing of lagoons, and (4) overharvest of nontimberforest products (plants and animals). Of these, slash-and-burn farming is the largest in scale and multiplicity of ef-fects, destroying terrestrial habitats through conversion

to non-native grasslands (

Aristida

species) and marinehabitats through sedimentation (Sammarco 1996; Hodg-son 1997).

We investigated the pattern and rate of past deforesta-tion to predict which areas were at greatest risk of defor-estation in the future. Only 5% (22,400 ha) of the penin-sula was deforested by 1957, but an additional 23%(98,800 ha) was cut by 1991, mirroring similar increasesin deforestation during this period elsewhere in Mada-gascar (Green & Sussman 1990; Kramer et al. 1997

b

)and the tropics (Dobson et al. 1997). Masoala’s annualdeforestation rate of 0.72% (2900 ha/year) was just un-der half the national rate (Green & Sussman 1990), re-flecting the isolation of the Masoala Peninsula.

The physiographic inventory showed that deforestedareas were primarily at low slope (

,

5

8

) and elevation(

,

300 m, Fig. 2). We selected these characteristics aspredictors for the area most likely to be deforested in com-

Table 3. Complementarity measures for each taxonomic group and environmental gradient showing the degree of difference in faunal composition across zones (value of 1 indicates faunas are entirely different).

Taxon

Totalspecies

observedEndemism

(%)

Rainfallgradient

Peninsulargradient Elevational gradient

a

Parent rock gradient Edge effect

east

3

westnorth

3

southlow

3

midmid

3

highlow

3

highgranite

3

sandgranite

3

basaltbasalt

3

sandperipheral

3

interior

Cicindelids

b

22 100 0.82 0.53 — — — 0.95 0.80 1.00

c

0.55Scarabs

b

97 100 0.78 0.71 0.65 0.78 0.77 0.94 0.90 0.88 0.62Butterflies

d

99 66.7 0.14 0.29 0.33 0.26 0.30 0.22 0.22 0.05 0.18Small

mammals

e

19 89.5 0.44 0.50 0.60 0.77 0.76 0.86 0.50 0.86 0.40Birds

f

87 90.8

g

0.20 0.22 — — 0.22

h

0.40 0.28 0.41 0.26Primates

i

10 100 0.10 0.20 — — — 0.25 0.00 0.25 0.20

a

Low,

#

400 m; mid,

.

400 m and

,

700 m; high,

.

700 m.

b

Data from Andriamampianina (1995).

c

Score of 1 means no species are found in common to both zones.

d

One hundred thirty-five species known from the Peninsula, data from Kremen et al. (1999

a

).

e

Data from Razafindrakoto (1995).

f

Data from Thorstrom and Watson (1997).

g

Endemic to the Malagasy subregion.

h

For birds the low- and mid-elevation zones are combined.

i

Data from Sterling and Rakotoarison (1998).

Table 4. The number of species restricted to single zones within environmental gradients for each taxonomic group.

a

Taxon

Totalspecies

observed

Rainfall gradient

Peninsular gradient

Elevation gradient

b

Parent rock type Edge effect

Low(east)

High(west)

Base(north)

Tip(south) Low Mid High Granite Basalt Sand Peripheral Interior

Cicindelids 22 15 3 8 2 — — — 14 1 0 9 3Scarabs 97 22 48 19 11 20 10 8 25 2 2 9 61Butterflies 99 10 2 20 4 7 10 5 16 2 1 5 11Small mammals

(nonvolant) 19 1 6 2 3 4 4 0 2 2 1 0 8Birds

c

87 8 6 9 5 10 — 5 6 3 6 2 16Primates 10 1 0 0 2 — — — 0 0 0 0 2

a

Restriction within one gradient does not imply restriction within another.

b

Low,

,

400 m; mid,

.

400 m and

,

700 m; high,

.

700 m.

c

For birds, low- and mid-elevation zones were combined.



1061

ing decades (Fig. 3). This region also contains a high con-centration of villages. With a constant deforestation rate, thisarea of 70,500 ha would be completely deforested within24 years (or within 9 years if deforestation increased ex-ponentially as

D

t

5

4.24 e0.052t, where Dt was the totalarea deforested on the Peninsula at time t, assuming t0 50 in 1794, the reported date of the first settlement).

The northwestern watersheds are steep, and defores-tation has been restricted to riparian areas. Nonetheless,riparian deforestation has fragmented the forest intothree blocks joined by narrow corridors (arrows in Fig.3) that link the peninsular forests to Madagascar’s north-south chain of humid forests (Fig. 1).

Human Population, Land Tenure, and Natural Resource Use

We observed 259 permanent villages (average numberof inhabitants, 208 6 582; largest village, 6500 inhabit-ants) located chiefly near the coast and approximately190 temporary villages chiefly in the forest block (Fig.3). Approximately 20% of the population lived in thearea of forest at greatest risk of deforestation.

Few residents had requested permits to clear new for-ests (,30 per year on average) or title to their lands (n 59). Thus, most of the land holdings and village territorieson the peninsula had no legal basis. In designating theirterritories, villagers included only small forest borders(,0.5 km) around fields for gathering forest products.Patterns of settlement and land use were consistentacross the peninsula (Lance et al. 1995; R. Lemaraina,unpublished data).

The average household consumed 8.1 6 3.4 m3/yearof firewood, 0.42 6 0.001 m3/year for house construc-tion and maintenance, and 0.48 6 0.45 m3/year forbuilding canoes, for a total of 8.9 6 3.9 m3/ year. From

the forest inventory, we determined that 90% of the for-est volume of stems .5 cm dbh consisted of useful spe-cies in one or more of these categories. Assuming a con-servative forest growth rate (2.0 m3/ha/year), the forestwould produce 0.9 3 2.0 m3/ha/year or 1.8 m3/ha/yearof useful wood. For sustainable use (i.e., in which con-sumption does not exceed production), each householdwould therefore need approximately 5 ha for collectionof forest products (8.9 m3/year of wood divided by 1.8m3/ha/year of production). The current population ofabout 8500 households would therefore require approx-imately 42,000 6 16,000 ha (SE based on householdvariation in wood consumption).

This forest was highly heterogeneous in species com-position (G. Rahajasoa, unpublished data), and these cal-culations would have little validity for individual species.Because there are many interchangeable species per usecategory (Rakotoarisoa 1997; Kremen et al. 1998) andbecause the forest contains a high proportion of useful

Figure 2. Slope and elevation characteristics of the zone deforested between 1957 and 1991. The histo-grams show the percent of the total deforested area (96,204 ha) in each slope category and 200-m eleva-tion range.

Figure 3. Risks of deforestation on the Masoala Penin-sula. The areas of forest depicted in dark gray (70,490 ha) were most similar in slope (,58 ) and elevation (,300 m) to the already deforested zone and would likely be deforested in the next 9–24 years (estimates from exponential to linear models). Permanent and temporary villages shown by size class. Deforestation has already resulted in three forest corridors in the northwestern portion of the peninsula (arrows).

1062 Designing a National Park Kremen et al.


wood, however, we can assume that each forest hectarewould be roughly equivalent in wood production forthese uses.

Feasibility of Natural Forest Management

The area selected as feasible for natural forest manage-ment (71,400 ha; Fig. 5) overlaps by 70% with the areapredicted to be at greatest risk of deforestation (Fig. 3).Within the selected area, we found 31 commercial spe-cies with a mean commercial volume of 83.9 6 7.6 m3/ha. Of these species, only several are exported regularly(ebony and rosewood species) at FOB values of $600–

2000/m3; these are in decline in some areas of the penin-sula due to overharvesting. Several others are harvestedfor domestic markets at $75–100/m3. Here the profitmargin is so low that harvest occurs illegally to avoidpayment of taxes and stumpage fees (Table 5). There isno incentive for communities to manage forests for do-mestic timber production alone.

Forest management plans can be certified as sustain-able if they meet a set of social and environmental stan-dards leading to minimal environmental impacts, sus-tainable production of timber and nontimber forestproducts, and a high level of local community participa-tion (Forest Stewardship Council 1994). Forest certifica-tion can provide a mechanism for developing exportmarkets for unknown timber species due to rising de-mand for certified wood products. On Masoala 12 spe-cies (3 exported and 9 that are not) met the wood qual-ity standards for export, and certified timber companies(e.g., Ecotimber) have expressed interest in importingsome of the unknown species. Sale of these species

Figure 4. Cumulative protection of species as inven-tory sites were added. The order of addition of sites was determined by the sites maximizing the addition of rare butterflies (Kremen et al.). Using this site or-der, other forest-dependent butterflies, threatened birds (see Thorstrom & Watson 1997), and endan-gered primates (see Mittermeier et al. 1994) showed a similar trajectory. In contrast, the trajectories for all butterflies, which includes a large number of species tolerating or favoring disturbed conditions, and that of all species (n 5 341) required inclusion of the pe-ripheral and fragmented sites to level off. The overall richness index (Sisk et al. 1994) was calculated. But-terfly data included eight species found outside of the formal inventory sites.

Figure 5. The Masoala National Park (211,230 ha, hatched) overlaid onto the area where forest manage-ment can be practiced without roads (71,400 ha, dark gray). Seventy percent of the potential forestry zone was left outside of the park. Areas of marine reserves and detached littoral forest reserves are also shown. No permanent villages were located inside of the park, although some temporary settlements were relocated.


Kremen et al. Designing a National Park 1063

through the certified market could increase their pricesfrom the domestic value of $70–95/m3 to an exportvalue of $350–600/m3.

These 12 species made up about 15% of the forest byvolume; therefore, estimated sustainable productionduring one harvest cycle would be (60 years) 3 2 m3/ha/year 3 0.15 5 18m3/ha of roundwood, or 6 m3/ha ofsawn wood (about one-third the production rate for sus-tainable forest management assumed by Bowles et al.1998). The community would be able to attain its maxi-mum productivity (130 m3/ha) by harvesting a 22-ha par-cel each year in a forest concession of 1300 ha.

This community forestry scheme presented advan-tages over business-as-usual, despite the payment oftaxes and certification fees (Table 5) and the conserva-tive estimation of sustainable production. It would gainnot only superior profits from the sale of export timberby a village association but also from the employment ofcommunity members (Table 6). Returns to the commu-nity from the sustainable forestry operation would be

$20,800/year, compared to $8,100/year for business asusual. This advantage translates to an annual gain of$130 per household, a significant benefit to householdlivelihood compared to average per capita income($230; Population Reference Bureau 1997).

Designing the Masoala National Park

The biological survey data showed that the full length ofrainfall and peninsular and elevational gradients neededto be incorporated within the park to protect the fullrange of biodiversity, especially localized endemics andspecies known to be rare or threatened elsewhere inMadagascar (design criteria 3 and 5 in Table 1; Fig. 4).To protect a large area of primary forest (criteria 1 and2), we included most of the forest except for the highlythreatened peripheral areas (Fig. 3), which showed lessbiological uniqueness than the other areas (Fig. 4).

Many other design criteria (6, 7, 8, and 9 in Table 1)supported the exclusion of the peripheral, threatenedzone from the park. It has a high edge-to-area ratio (Fig. 3),and deforestation models showed that this area of70,490 ha would not survive the next 9–24 years undercurrent deforestation trends (Fig. 3). There are many set-tlements in this area, so attempts to protect it wouldprobably also fail over the long term.

As an alternative to full protection, leaving this areaoutside of the park would allow it to serve both as anecological buffer to the park and an economic supportzone for local people. This region included 70% of thearea best suited for community forest management. Ouranalyses suggested that sustainable forest managementin this region could provide better economic returns tolocal communities than slash-and-burn agriculture (de-sign criterion 10, Table 6). Using economic incentives toprotect these resources from slash and burn would im-

Table 5. Profits from illegal versus legal forestry for low-value woods on the domestic timber market* in Madagascar.

Costs and benefitsIllegal logging

($/m3 sawn wood)

Legal forestryconcession

($/m3 sawn wood)

Extraction cost 52.00 52.00Transport cost 32.00 32.00Storage cost 3.00 3.00Local tax — 2.80Stumpage fee — 2.20Domestic market

value (e.g., Nanto) 95.00 95.00Net benefit 8.00 3.00

*Figures calculated for the closest market to the peninsula (Anta-laha).

Table 6. Annual flows and net present values (NPV) at 10, 60, and 120 years from sustainable community forestry versus business-as-usuala to 100 households using 1300 ha of forest.

Annual flows NPV 10 years NPV 60 years NPV 120 years

Net benefitsCommunity

forestryb BAU cCommunity

forestryb UsualCommunity

forestryb UsualCommunity

forestrya Usual

Profits: rice production 0 1,400 0 8,296 0 13,457 0 13,477Profits: timber

production 12,700 6,800 46,004 39,242 79,113 63,655 79,399 63,744Employment: timber

harvest 8,100 0 42,538 0 73,597 0 73,863 0Total 20,800 8,200 88,542 47,538 152,710 77,112 153,262 77,220aThe business-as-usual scenario would deforest the entire area in 66 years at the national deforestation rate of 1.5%, whereas revenues fromsustainable forestry would continue indefinitely.bFor calculation of net benefits from sustainable forestry, weighted values for timber production for domestic and certified markets were deter-mined from species commercial volumes and projected certified values. Costs are found in Table 5. Additional costs related to export and certifi-cation are transport $100/m3, storage $30/m3, export tax $20/m3, transaction costs $8/m3, and certification fee $23/m3 after year 5 (assumingthat the integrated conservation and development program pays certification fees until year 5).cNet benefits for the business-as-usual scenario are calculated for rice production at $69/ha (R. Lemaraina, unpublished data) and for timber,assuming sale to middlemen at the cost of extraction ($52/m3, Table 5).



prove the likelihood of encouraging people to respectpark boundaries and would provide habitat for speciestolerant of reduced-impact forest management (e.g.,Ganzhorn 1995; Faith & Walker 1996).

Both biological and socioeconomic data thereforepointed to the same solution: to include the large andenvironmentally heterogeneous core zone to protect theunique biodiversity of the area and to protect the forestsof the peripheral zone through community-based eco-nomic incentives rather than legal mechanisms (Fig. 5).Next, we brought in other information layers and designcriteria to fine-tune the boundaries.

Some of the lowest-elevation forest in the peripheralzone (approximately 20,000 ha in the southeast; Fig. 5)was included within the park as a safeguard, despite itsapparent lack of biological uniqueness. Forest under 200m elevation is poorly represented in Madagascar’s re-serve network (Nicoll & Langrand 1989). Also, invento-ries are never exhaustive, and future surveys may dis-cover that species of concern are limited to the peripheralzone (e.g., the poorly known scarab beetles).

Littoral forest is a highly threatened habitat in Mad-agascar (Du Puy & Moat 1996) and contains uniqueplant species (R. Rabevohitra, unpublished data). Severalfragments were therefore included in the park (designcriteria 3 & 5). Littoral forests differed markedly in plantcommunity composition from east to west on the penin-sula. Eastern littoral forests were dominated by the plantfamily Euphorbiaceae (32 6 4% of stems .10 cm dbh inthe three forest fragments studied), whereas Pandan-aceae (15%) and Lauraceae (13%) dominated the west-ern littoral forest (G. Rahajasoa, unpublished data). Wetherefore included examples of each type of littoral for-est (design criterion 3). Including these littoral forest ar-eas created management problems. On the west coast,many small settlements were dispersed throughout thisforest, requiring negotiations with villagers to consoli-date settlements so as to include an area free of settle-ments in the park (design criterion 8). On the east coast,where littoral forests were disjunct, separate reserveswere required, increasing the patrolling burden for thepark.

We emphasized protecting transitional zones and habi-tat mosaics to provide for species requiring multiple hab-itat types and for important environmental interactionsbetween habitat types (design criteria 1 & 4, Bennet1993; Fenton 1997; Polis et al. 1997; Smith et al. 1997b).For example, we selected littoral forest areas based ontheir association with other habitat types. On the eastcoast, only littoral forest fragments consisting of mosaicsof littoral forest, mangrove, marsh, and permanentlyflooded forest were selected. In one fragment, the terres-trial park could also be linked with a marine reserve in-cluding beaches, estuaries, lagoons, and coral reefs. Onthe west coast, including the littoral forest preserved aunique littoral-to-lowland humid forest transition, now

extremely rare throughout Madagascar, and a littoral-to-beach and coral-head transition (Fig. 5).

Three small forest corridors and the northwestern sec-tion (Fig. 5) were included within the park to guard fu-ture options for linking the Masoala forests with pro-tected areas to the north (design criterion 4), wheremany species known from Masoala also occur (Fisher1998; Lees et al. 1999). Including the corridors compli-cated the park’s configuration and increased the man-agement burden (e.g., we did not address criteria 7 &13): several communities were enclosed, necessitatingspecial negotiations (communities agreed not to expandexisting village territories in exchange for communalland titles legalizing their rights to the land).

In sum, the proposed park (Kremen et al. 1995)was 211,230 ha, including approximately 97,000 ha oflowland forest (defined here as ,400 m). Over half of theavailable lowland habitat on the peninsula would be pro-tected, a larger area than exists in all of Madagascar’s hu-mid forest reserves combined (Direction des Eaux etForêts et al. 1993). The park consisted almost entirely(97%) of primary forests. Approximately 121,840 ha offorest were excluded from the park, of which 50,100 hafit the criteria for natural forest management (Fig. 5).The area left outside of the park would thus be sufficientto meet local needs for collection of subsistence prod-ucts and could also provide revenues through sustain-able forestry.

Discussion

The Masoala National Park was gazetted in October1997, culminating 2 years of work with local, re-gional, and national councils. Only minor changeswere made to the proposed boundaries during theapproval process. Below we comment on the park’sdesign in terms of its likely ecological and socioeco-nomic sustainability.

This park is Madagascar’s largest protected area(Direction des Eaux et Forêts et al. 1993). The large sizeof Masoala will contribute to the park’s viability and itsimportance for conservation of humid evergreen ecosys-tems in Madagascar, especially given the increasing iso-lation and the risk of biotic collapse in Madagascar’sexisting system of small reserves (Green & Sussman1990). It is the only park in Madagascar to protect a sig-nificant piece of lowland humid forests and is thus ofglobal conservation importance.

The park’s size was determined largely by the needto include the environmental gradients and habitat diver-sity that support the diverse species found in this locality.Thus, we turned to area-demanding species (Landres1983) to assess whether the park will be large enough tomaintain viability. The population of the park’s largestlemur, Varecia variegata rubra (a subspecies restricted



to the peninsula), was estimated at 16,200 6 3600 (Mer-enlender et al. 1998), which would appear sufficient forlong-term survival. Two rare forest raptors, however,the Madagascar Serpent Eagle (Eutriorchis astur) andRed Owl (Tyto soumagnei), although unlikely to havebetter opportunities for survival elsewhere in Madagas-car (Thorstrom et al. 1995; O. Langrand & R. Watson,personal communication), appear to have small popula-tions in Masoala Park (,400 and 1600 individuals basedon territory size, respectively; Thorstrom et al. 1995, un-published data). This suggests that Madagascar’s largestprotected area may still be too small to protect the mostarea-demanding species. Increasing the size of MasoalaPark would have provided only marginal increases in thesize of protected populations. Both raptors, however,were recently discovered in the Anjanaharibe reserve(32,000 ha) to the north (D. Hallieux, personal commu-nication). The best option for assuring their survival, andperhaps that of many other species in these protectedareas, would probably be to maintain the wide forestcorridor between Masoala National Park and Anjanaha-ribe (approximately 70 3 32 km).

Protected areas must have the support of local com-munities and the nation if they are to survive (Hough &Sherpa 1989). Accordingly, we took care to develop thepark proposal in consultation with people at local andnational levels. At the local level, the participation of vil-lagers and local field staff in mapping village territoriesensured that traditional views on ownership and landuse were respected. Villagers were never evicted fromlands they had settled, despite their lack of legal tenure,unless social norms permitted eviction. For example, so-cial norms dictated that the communities illegally estab-lished near the corridors should not be evicted becausethese communities had invested heavily in developing ir-rigated rice paddies over a long time period. We know-ingly increased the park’s management burden to re-spect these norms and gain local support. In contrast,other villagers (47 households) were practicing slash-and-burn farming illegally out of temporary settlementsnear the headwaters of several watersheds. Here, socialnorms encouraged the reintegration of these people intopermanent villages because they had made no invest-ment in permanent agriculture and their activities poten-tially threatened the water supply of downstream com-munities.

The integrated conservation and development ap-proach to park design was essential for winning sup-port for the park at both local and national levels.This approach consisted of designing a larger conser-vation landscape that provided economically viable,alternative uses for forests beyond the park. Our feasibil-ity studies suggested that profitable sustainable forestmanagement in this zone could be achieved by develop-ing export markets for unknown timber species throughcertification. They showed that this zone (50,100 ha)

could generate an annual flow of $647,000 to the regionand an additional $413,600 to the nation through taxes.About half of the population of the peninsula, includingthe entire population currently living within the threat-ened peripheral zone, could potentially benefit from thisactivity. Other activities such as ecotourism and im-proved agriculture and fishing would enhance local andnational revenues, as would indirect benefits from wa-tershed protection and the sustained production of for-est products for subsistence (Kremen et al., unpublishedobservation).

Two caveats must be mentioned in assessing socioeco-nomic sustainability. First, if population increases on theMasoala Peninsula at the national rate of 3.2% (Popula-tion Reference Bureau 1997), it is doubtful that the Ma-soala ICDP will be able to protect the park ecosystem ef-fectively or sustain the productivity of managed forests,agricultural lands, and marine resources outside thepark. Until now, members of the Masoala ICDP consor-tium have disagreed on the appropriateness and impor-tance of including a family planning and health compo-nent and have taken no action in this arena. We considerthis missing component to be a major flaw in this ICDP.Second, many of the activities that the ICDP is imple-menting are complex and experimental. The successrate of development projects in Africa is only 50%(World Bank 1995). Project success will require contin-ued assistance to local communities over a long time pe-riod, with continued monitoring and adaptation of man-agement plans (Kremen et al. 1994; Margoluis & Salafsky1998).

Certain information could not be obtained during thetime-span available for reserve design, including the lo-cal rate of human population growth, identities of speci-mens from botanical inventory, fine-scaled mapping ofvegetation by plant associations, and differentially cor-rected GPS points (accuracy of 61 m). Also, we knowlittle about the key regulators of ecosystem processes inMalagasy forests. Similarly, data will be lacking for mostpoorly known ecosystems of the world. The reserve de-sign process may often need to fill in for deficiencies inlocal ecological knowledge with general conservationbiology principles, and it may need to allow for a higherdegree of spatial error (e.g., accuracy of 6100 m) thandesired.

Lessons Learned from Masoala

The following lessons from our project on the MasoalaPeninsula can be applied to other areas:

(1) Biodiversity survey. Birds and primates showedless differentiation than most insects across envi-ronmental zones in Masoala (Tables 3 & 4) and hadlarger range sizes (Lees 1997; Lees et al. 1999) andlower rates of species turnover along elevational



and latitudinal gradients (Fisher 1996) in Madagas-car. If we had restricted our surveys to taxa withlow b diversity, we might have erroneously con-cluded that Masoala’s forests lacked heterogeneityand that there was little need to consider comple-mentarity across environmental zones in develop-ing a conservation strategy. This example empha-sizes the need to consider multiple taxonomicgroups when designing reserves (see also Prender-gast et al. 1993; Willis et al. 1996; Flather et al.1997), especially groups with high b diversity(Kremen et al. 1993; Colwell & Coddington 1994;Kremen 1994).

(2) Threats analysis. The strategy used at Masoala,from park and multiple-use zone design to the se-lection of economic development activities, fo-cused on understanding existing threats and theirspatial distribution and developing project activi-ties to counter each threat (for a clear explanationof this methodology, see Margoluis & Salafsky1998). The relationship between reserve design,project activities, and threats is essential to creat-ing real links between conservation and develop-ment and to avoiding the problems experiencedby an earlier generation of ICDPs (Wells et al.1992; Kramer et al. 1997a) in which conservationand development activities were often carried outin parallel.

(3) Sustainable forestry. Although our analyses showedthat sustainable forestry could be a viable eco-nomic alternative to the business-as-usual scenario,the margin was small. Competition from nonsus-tainable commercial enterprise could underminethese efforts (Howard et al. 1996). The ICDPstherefore need to become more diverse in their ap-proaches to capturing forest values in buffer zones(e.g., ecotourism, nontimber forest products).Also, ICDPs should encourage reforestation withcommercial and subsistence species in degradedareas (Bawa & Reinman 1998) so as to be preparedfor a future with larger human populations.

Acknowledgments

We thank C. Boggs, G. Daily, J. Hellman, J. Niles, M.Powers, J. Sarukhan, T. Ricketts, and two anonymousreviewers for helpful suggestions on the manuscript.The design and establishment of the Masoala NationalPark was funded principally by a U.S. Agency for Inter-national Development SAVEM grant and was carried outby the Masoala Project consortium: CARE InternationalMadagascar, Wildlife Conservation Society, and the Pere-grine Fund under the guidance of the Direction des Eauxet Forêts and the Association National pour la Gestiondes Aires Protégées. The geographic information system

support was donated by Stanford University’s Center forConservation Biology with Arc/Info software donated byEnvironmental Systems Research Institute. The SPOTsatellite images were provided by the Missouri BotanicalGarden, and image processing was completed at the U.S.Geological survey EROS Data Center. We thank themany individuals who helped bring this project to com-pletion by providing data or technical assistance or car-rying out field work. Finally, we thank the former U.S.Ambassador V. Huddlestone for promoting the park pro-posal and President D. Ratsiraka and his Cabinet of Min-isters for final approval.

Literature Cited

Abraham, J., R. Benja, M. Randrianasolo, J. U. Ganzhorn, V. Jeannoda,and E. G. Leigh. 1996. Tree diversity on small plots in Madagascar:a preliminary review. Revue d’Ecologie (Terre Vie) 51:93–116.

Alpert, P. 1996. Integrated conservation and development projects.BioScience 46:845–855.

Andriamampianina, L. M. 1995. Contribution à l’étude de la repartitionet de la structure des peuplements des insectes Scarabeidae (Co-leoptères) de la Presqu’île de Masoala. Faculté des Sciences. Univer-site d’Antananarivo, Antananarivo, Madagascar.

Andriamampianina, L. 1996. Biogeography of enariine (Melolonthidae)and cicindelid beetles in Madagascar. Durrell Institute of Conserva-tion and Ecology. University of Kent, Canterbury, United Kingdom.

Bawa, K. S., and S. Reinman. 1998. Natural forest management andconservation of biodiversity in tropical forests. Conservation Biol-ogy 12:46–55.

Bennet, A. F. 1993. Microhabitat use by the long-nose potoroo, Potor-ous tridactylus, and other small mammals in remnant forest vegeta-tion of south-western Victoria. Wildlife Research 20:267–285.

Bowles, I. A., R. E. Rice, R. A. Mittermeier, and G. A. B. da Fonseca.1998. Logging and tropical forest conservation. Science 280:1890–1899.

Bruenig, E. F. 1996. Conservation and management of tropical rainfor-ests. CAB International, Wallingford, United Kingdom.

Colwell, R. K., and J. A. Coddington. 1994. Estimating terrestrial biodi-versity through extrapolation. Philosophical Transactions of theRoyal Society of London B Biological Sciences 345:101–118.

Direction des Eaux et Forêts, Projet COEFOR, and Conservation Inter-national. 1993. Repertoire et carte de distribution: domaine fores-tier national de Madagascar. Conservation International, Antanan-arivo, Madagascar.

Dobson, A. P., A. D. Bradshaw, and J. M. Baker. 1997. Hopes for the fu-ture: restoration ecology and conservation biology. Science 277:515–522.

Du Puy, D. J., and J. Moat. 1996. A refined classification of the primaryvegetation of Madagascar based on the underlying geology: usingGIS to map its distribution and to assess its conservation status.Pages 205-218 in W. R. Lourenco, editor. Biogeographie de Mada-gascar. ORSTOM, Paris.

Ehrlich, P. R. 1997. A world of wounds: ecologists and the human di-lemma. Ecology Institute, Oldendorf, Germany.

Faith, D. P., and P. A. Walker. 1996. Integrating conservation and de-velopment: incorporating vulnerability into biodiversity-assessmentof areas. Biodiversity and Conservation 5:417–429.

Fenton, M. B. 1997. Science and the conservation of bats. Journal ofMammology 78:1–14.

Fisher, B. L. 1996. Ant diversity patterns and conservation planning inMadagascar. University of California, Davis.

Fisher, B. L. 1998. Ant diversity patterns along an elevational gradient



in the Réserve Spéciale d’Anjanaharibe-Sud and on the western Ma-soala Peninsula, Madagascar. Fieldiana 0:39–67.

Flather, C. H., K. R. Wilson, D. J. Dean, and W. C. McComb. 1997.Identifying gaps in conservation networks: of indicators and uncer-tainty in geographic-based analyses. Ecological Applications 7:531–542.

Forest Stewardship Council. 1994. Principles and criteria. Forest Stew-ardship Council, Richmond, Vermont.

Ganzhorn, J. U. 1995. Low-level forest disturbance effects on primaryproduction, leaf chemistry and lemur populations. Ecology 76:2084–2096.

Gentry, A. H. 1993. Diversity and floristic composition of lowland trop-ical forest in Africa and South America. Pages 500–547 in P. Gold-blatt, editor. Biological relationships between Africa and SouthAmerica. Yale University Press, New Haven, Connecticut.

Goodman, S. M. 1997. Natural change and human impact in Madagas-car. Smithsonian Institution Press, Washington, D.C.

Green, G. M., and R. W. Sussman. 1990. Deforestation history of theeastern rain forests of Madagascar from satellite images. Science248:212–215.

Hannah, L., B. Rakotosamimanana, J. Ganzhorn, R. A. Mittermeier, S.Olivieri, L. Iyer, S. Rajaobelina, J. Hough, F. Andriamialisoa, I.Bowles, and G. Tilkin. 1998. Participatory planning, scientific pri-orities and landscape conservation in Madagascar. EnvironmentalConservation 25:30–36.

Hodgson, G. 1997. Resource use: conflicts and management solutions.Pages 386–410 in C. Birkeland, editor. Life and death of coral reefs.International Thomson Publishing, New York.

Hottin, G., and E. Liandrat. 1964. Géologie de la presqu’île de Masoala.Service Géologique, Antananarivo, Madagascar.

Hough, J. L., and M. N. Sherpa. 1989. Bottom up vs. basic needs: inte-grating conservation and development in the Annapurna andMichiru Mountain conservation areas of Nepal and Malawi. Ambio18:434–441.

Howard, A. F., R. E. Rice, and R. E. Gullison. 1996. Simulated financial re-turns and selected environmental impacts from four alternative silvi-cultural prescriptions applied in the neotropics: a case study of theChimanes Forest, Bolivia. Forest Ecology and Management 89:43–57.

Jenkins, M. D. 1990. Madagascar: profil de l’environnement. WorldConservation Union, Gland, Switzerland.

Johns, A. G. 1997. Timber production and biodiversity conservation intropical rain forests. Cambridge University Press, Cambridge,United Kingdom.

Kramer, R., C. van Schaik, and J. Johnson. 1997a. Last Stand: protectedareas and the defense of tropical biodiversity. Oxford UniversityPress, New York.

Kramer, R. A., D. D. Richter, S. Pattanayak, and N. P. Sharma. 1997b.Ecological and economic analysis of watershed protection in east-ern Madagascar. Journal of Environmental Management 49:277–295.

Kremen, C. 1992. Assessing the indicator properties of species assem-blages for natural areas monitoring. Ecological Applications 2:203–217.

Kremen, C. 1994. Biological inventory using target taxa: a case studyof the butterflies of Madagascar. Ecological Applications 4:407–422.

Kremen, C., R. K. Colwell, T. L. Erwin, D. D. Murphy, R. F. Noss, andM. A. Sanjayan. 1993. Terrestrial arthropod assemblages: their usein conservation planning. Conservation Biology 7:796–808.

Kremen, C., A. M. Merenlender, and D. D. Murphy. 1994. Ecologicalmonitoring: a vital need for integrated conservation and develop-ment programs in the tropics. Conservation Biology 8:388–397.

Kremen, C., J. Rakotomalala, V. Razafimahatratra, O. Rakotobe, P. Guil-lery, A. Zakandraina, and Jaomanana. 1995. Proposition des limitesdu Parc National Masoala. CARE International, Wildlife Conserva-tion Society, and Peregrine Fund, Antananarivo, Madagascar.

Kremen, C., I. Raymond, and K. Lance. 1998. An interdisciplinary tool

for monitoring conservation impacts in Madagascar. ConservationBiology 12:549–563.

Kremen, C., D. Lees, V. Razafimahatratra, and H. Raharitsimba. 1999a.Biodiversity surveys in Madagascar rain forest: using butterfly indi-cators to evaluate the design of a new national park. In W. Weber,A. Vedder, H. S. Morland, L. White, and T. Hart, editors. In Press.African rain forest ecology and conservation. Yale University Press,New Haven, Connecticut.

Kremen, C., I. Raymond, K. Lance, and A. Weiss. 1999b. Monitoringnatural resource use on the Masoala Peninsula, Madagascar: a toolfor managing integrated conservation and development projects.Pages 63–84 in K. Saterson, R. Margoluis, and N. Salafsky, editors.Measuring conservation impacts: an interdisciplinary approach toproject monitoring and evaluation. Biodiversity Support Program,Washington, D.C.

Lance, K., C. Kremen, and I. Raymond. 1995. Annexe 1. Reconnais-sance préliminaire de la peninsule de Masoala: utilisation de la terreet des ressources. Pages 1–17 in C. Kremen, editor. Proposition deslimites du Parc National Masoala. CARE International Madagascar,Wildlife Conservation Society, and The Peregrine Fund, Antanan-arivo, Madagascar.

Landres, P. B. 1983. Use of the guild concept in environmental impactassessment. Environmental Management 7:393–398.

Langrand, O. 1990. Guide to the birds of Madagascar. Yale UniversityPress, New Haven, Connecticut.

Lees, D. C. 1997. Systematics and biogeography of Madagascar my-calesine butterflies and applications to conservation evaluation.University of London, London.

Lees, D. C., C. Kremen, and L. Andriamampianiana. 1999. A null modelfor species richness gradients: bounded range overlap of butterfliesand other rainforest endemics in Madagascar. Biological Journal ofthe Linnean Society. In Press.

Margoluis, R., and N. Salafsky. 1998. Measures of success: designing,managing, and monitoring community-oriented conservation projects.Island Press, Covelo, California.

Margules, C. R., and J. L. Stein. 1989. Patterns in the distributions ofspecies and the selection of nature reserves: an example from Eu-calyptus forests in south-eastern New South Wales. Biological Con-servation 50:219–238.

Margules, C. R., A. O. Nicholls, and R. L. Pressey. 1988. Selecting net-works of reserves to maximize biological diversity. Biological Con-servation 43:63–76.

Merenlender, A., C. Kremen, M. Rakotondratsima, and A. Weiss. 1998.Monitoring impacts of natural resource extraction on lemurs onthe Masoala Peninsula, Madagascar. Conservation Ecology 2:(article5 www.consecol.org/journal).

Mittermeier, R. A., I. Tattersall, W. R. Konstant, D. M. Meyers, and R. B.Mast. 1994. Lemurs of Madagascar. Conservation International,Washington, D.C.

Munasinghe, M. 1993. Environmental economics and biodiversity man-agement in developing countries. Ambio 22:126–135.

Myers, N. 1988. Threatened biotas: “hotspots” in tropical forests. TheEnvironmentalist 8:1–20.

Nelson, R., and N. Horning. 1993. AVHRR-LAC estimates of forest areain Madagascar, 1990. International Journal of Remote Sensing 14:1463–1475.

Newmark, W. D. 1985. Legal and biotic boundaries of western NorthAmerican national parks: a problem of congruence. Biological Con-servation 33:197–208.

Newmark, W. D. 1996. Insularization of Tanzanian parks and the localextinction of large mammals. Conservation Biology 10:1549–1556.

Nicoll, M. E., and O. Langrand. 1989. Madagascar: révue de la conserva-tion et des aires protégées. World Wide Fund for Nature, Gland,Switzerland.

Noss, R. F. 1987. Protecting natural areas in fragmented landscapes.Natural Areas Journal 7:2–13.

Noss, R. F., H. B. Quigley, M. G. Hornocker, T. Merrill, and P. C.



Paquet. 1996. Conservation biology and carnivore conservation inthe Rocky Mountains. Conservation Biology 10:949–963.

Pernetta, J., R. Leemans, D. Elder, and S. Humphrey, editors. 1994. Im-pacts of climate change on ecosystems and species. World Conser-vation Union, Gland, Switzerland.

Pickett, S. T. A., and J. N. Thompson. 1978. Patch dynamics and the de-sign of nature reserves. Biological Conservation 13:27–37.

Pielou, E. 1984. The interpretation of ecological data: a primer on clas-sification and ordination. Wiley, New York.

Polis, G. A., W. B. Anderson, and R. D. Holt. 1997. Toward an integra-tion of landscape and food web ecology: the dynamics of spatiallysubsidized food webs. Annual Review of Ecological Systems 28:289–316.

Population Reference Bureau. 1997. World population data sheet1995. Population Reference Bureau, Washington, D.C.

Prendergast, J. R., R. M. Quinn, J. H. Lawton, and B. C. Eversham.1993. Rare species, the coincidence of diversity hotspots and con-servation strategies. Nature 365:335–337.

Pressey, R. L., S. Ferrier, T. C. Hager, C. A. Woods, S. L. Tully, and K. M.Weinman. 1996. How well protected are the forests of north-east-ern New South Wales? Analyses of forest environments in relationto formal protection measures, land tenure, and vulnerability toclearing. Forest Ecology and Management 85:311–333.

Rabinowitz, D., S. Cairns, and T. Dillon. 1986. Seven forms of rarityand their frequency in the flora of the British Isles. Pages 182–204in M. E. Soulé, editor. Conservation biology: the science of scarcityand diversity. Sinauer Associates, Sunderland, Massachusetts.

Rakotoarisoa, J. N. 1997. Etude de la structure de la filière bois dans laforêt de Masoala: investigation sur le bassin versant Ianobe et laville d’Antalaha, dans une perspective de mise en place d’une ges-tion communautaire. Ecole Superieure des Sciences Agronomiques.Université d’Antananarivo, Antananarivo, Madagascar.

Raymond, I. 1995. Approche phytoecologique sur l’évaluation qualita-tive et quantitative des utilisations villageoises des ressources na-turelles en forêt dense humide sempervirente. Universitéd’Antananarivo, Antananarivo, Madagascar.

Razafimahatratra, V., and L. Andriamampianina. 1995. Annexe V.4. In-ventaire des deux groupes d’insectes Coleoptères: Scarabeidae etCicindelidae de la Presqu’Ile Masoala et comparaisons biogéogra-phiques. Pages 1–64 in C. Kremen, editor. Proposition des limitesdu Parc National Masoala. CARE International Madagascar, WildlifeConservation Society, and The Peregrine Fund, Antananarivo,Madagascar.

Razafindrakoto, Y. 1995. Contribution à l’étude des rongeurs et insecti-vores de la presqu’île de Masoala. Université d’Antananarivo,Antananarivo, Madagascar.

Rich, P. M., S. B. Weiss, D. M. Debinski, and J. F. McLaughlin. 1992.Physiographic inventory of a tropical reserve. Pages 197–208 inProceedings of the 12th ESRI user conference. Environmental Sys-tems Research Institute, Redlands, California.

Ricklefs, R. E. 1973. Ecology. W.H. Freeman, New York.Sammarco, P. W. 1996. Comments on coral reef regeneration, bioero-

sion, biogeography, and chemical ecology: future directions. Jour-nal of Experimental Marine Biology and Ecology 200:135–168.

Scott, J. M., B. Csuti, J. D. Jacobi, and J. E. Estes. 1987. A geographic ap-proach to protecting future biological diversity. BioScience 37:782–788.

Shaffer, M. L. 1981. Minimum population sizes for species conserva-tion. BioScience 31:131–134.

Silva, J. N. M., J. O. P. De Carvalho, J. D. C. A. Lopes, B. F. De Almeida,D. H. M. Costa, L. C. De Oliveira, J. K. Vanclay, and J. P. Skovs-gaard. 1995. Growth and yield of a tropical rain forest in the Brazil-

ian Amazon 13 years after logging. Forest Ecology and Management71:267–274.

Sisk, T., S. E. Launer, K. R. Switky, and P. R. Ehrlich. 1994. Identifying ex-tinction threats: global analyses of the distribution of biodiversity andthe expansion of the human enterprise. BioScience 44:592–604.

Smith, A. P., N. Horning, and D. Moore. 1997a. Regional biodiversityplanning and lemur conservation with GIS in western Madagascar.Conservation Biology 11:498–512.

Smith, T. B., R. K. Wayne, D. J. Girman, and M. W. Bruford. 1997b. Arole for ecotones in generating rainforest biodiversity. Science 276:1855–1857.

Soulé, M. E. 1986. Conservation biology: the science of scarcity and di-versity. Sinauer Associates, Sunderland, Massachusetts.

Soulé, M. E., and D. Simberloff. 1986. What do genetics and ecologytell us about the design of nature reserves? Biological Conservation35:19–40.

Sterling, E. J., and N. Rakotoarison. 1998. Rapid assessment of primatespecies richness and density on the Masoala Peninsula, easternMadagascar. Folia Primatologica 69:109–116.

Terborgh, J., and B. Winter. 1983. A method for siting parks and re-serves with special reference to Colombia and Ecuador. BiologicalConservation 27:45–58.

Thorstrom, R., and R. T. Watson. 1997. Avian inventory and key spe-cies of the Masoala Peninsula, Madagascar. Bird Conservation Inter-national 7:99–115.

Thorstrom, R., R. T. Watson, B. Damary, F. Toto, M. Baba, and V. Baba.1995. Repeated sightings and first capture of a live Madagascar Ser-pent-eagle Eutriorchis astur. Bulletin of the British Ornithologists’Club 115:40–45.

Thorstrom, R., J. Hart, and R. T. Watson. 1997. New record, rangingbehavior, vocalization and food habits of the Madagascar Red Owl.Ibis 139:477–481.

Usher, R. R. 1986. Wildlife conservation evaluation. Chapman andHall, London.

van Schaik, C. P., J. Terborgh, and B. Dugelby. 1997. The silent crisis: thestate of rain forest nature preserves. Pages 64–89 in R. Kramer, C. P.van Schaik, and J. Johnson, editors. Last stand: protected areas and thedefense of tropical diversity. Oxford University Press, New York.

Wallis De Vries, M. F. 1995. Large herbivores and the design of large-scale nature reserves in western Europe. Conservation Biology 9:25–33.

Wells, M., K. Brandon, and L. Hannah. 1992. People and parks: linkingprotected area management with local communities. The WorldBank, Washington, D.C.

Western, D., and H. Gichohi. 1993. Segregation effects and the impov-erishment of savanna parks: the case for ecosystem viability analy-sis. African Journal of Ecology 31:269–281.

Western, D., and M. C. Pearl. 1989. Conservation for the twenty-firstcentury. Oxford University Press, New York.

Wilkie, D. S., and J. T. Finn. 1996. Remote sensing imagery for naturalresources monitoring. Columbia University Press, New York.

Willis, C. K., A. T. Lombard, R. M. Cowling, B. J. Heydenrych, and C. J.Burgers. 1996. Reserve systems for limestone endemic flora of theCape Lowland fynbos: iterative versus linear programming. Biologi-cal Conservation 77:53–62.

World Bank. 1995. 1993 evaluation results. A World Bank OperationsEvaluation Study. World Bank, Washington, D.C.

World Bank, U.S. Agency for International Development, CooperationSuisse, United Nations Educational, Scientific, and Cultural Organi-zation, United Nations Development Program, and Worldwide Fundfor Nature. 1988. Madagascar environmental action plan. World Bank,Washington, D.C.

designing the masoala national park in madagascar based on

Documents