19828 - World Bank Documents

19828February 1994

LATEN Dissemination Note # 10

Assessing the Conservation Potential andDegree of Threat Among Ecoregionsof Latin Amenrca and the Caribbean:A Proposed Landscape Ecology Approach

February 1994

~~~~~~~~~~~~~~~-& -

David M. OlsonEric Dinerstein

The World BankLatin America Technical Department

Environment DivisionFIEC P

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

I

I

LATEN Dissemination Note # 10

Assessing the Conservation Potential andDegree of Threat Among Ecoregions

of Latin America and the Caribbean:A Proposed Landscape Ecology Approach

February 1994

David M. OlsonEric Dinerstein

Please address correspondence to either:

Dr. David M. Olson Douglas J. GrahamConservation Science Program Environment DivisionWorld Wildlife Fund Latin America Technical Department1250 24th St., NW The World BankWashington, DC 20037 1818 H Street, NW

Washington, DC 20433

I

Foreword

The 'Dissemination Note" Series of the Latin America & CaribbeanRegion's Environment Division (LATEN) seeks to share the results of ouranalytical and operational work, both completed or in progress. Through thisSeries, we present the preliminary findings of larger studies in an abbreviatedform as well as describe 'best practiceso with regard to major environmentalissues currently confronting LAC Countries.

In most cases, these notes represent "work in progress' and as suchhave not been subject to either substantial intemal review or editing.Therefore the findings, interpretations, and conclusions expressed in thesenotes are entirely those of the authors and should not be attributed to theWorld Bank, members of its Board of Executive Directors, or the countriesthey represent.

This Dissemination note, authored by David Olson and EricDinerstein of the World Wildlife Fund-US, presents the methodologicalframework for an ongoing biodiversity study coordinated by Douglas J.Graham of LATEN. The study is being funded by the Bank's GlobalEnvironment Coordination Division, by LATEN and by WWF-US.

The final products of this study, due in late 1994, will be an originalecoregions map of all of Latin America and the Caribbean and an evaluationof the conservation status of each of those ecoregions. The conservationstatus of an ecoregion is defined as its long-term conservation potential andby a measure of threats undermining that potential. Future steps, possibly incollaboration with other organizations, will involve overlaying biologicalvalue and conservation feasibility information to provide an important inputinto the determination of conservation priorities as carried out by LACcountries and by international funding organizations.

Development of the methodological framework benefited from theinput of many regional and intemational biodiversity specialists. Collectingand compiling the data in order to complete the study will involve an evencloser collaboration. We are well aware that there are as many approachesin this area as there are biodiversity specialists. Although we feel theapproach developed in this study enjoys support from much of theconservation community, many readers will disagree with parts of themethodology. We are disseminating the methodology at this time to facilitatefeedback from all interested parties in the hope that the utility of the endproduct will be maximized.

Dennis J. MaharDivision Chief

Environment DivisionLatin America and the Caribbean Region

The World Bank

Contents

L Introduction 1

IL Objectives 2

IIL Building on Past Efforts 2

IV. Approach 3Project Flow and Structure 4Terminology and Classification Scheme 4Criteria for Calculating Index Values and Ranking Ecosystems 5Graphical Analyses and the Development of Appropriate 5

Conservation StrategiesEcoregion Maps and Text 5

V. Identifying Ecoregions 6Ecoregion Base Maps 6Extant Original Habitat 6Disjunct Ecoregions 6Endemism & "Hotspots" 6The hnportance of Beta Diversity 7Marine and Freshwater Ecosystems 8

VL Auesing Conservation Potentialand Degree of Threat among Ecoregkn 8

Coneraton Potential 9Dogree of Threat 10Alternative Schemes 11Index Range 11Weighting Criteria 12Multi-National Ecoregions 12Criteria for DSE's 12

1. Tropical Broadleaf Forests 122. Conifer and Temperate Broadleaf Foests 243. Grasslands, Savannahs, Wetlands, Shrublands 304. Xeic Formations 365. Mangroves 41

VIL Literature Cited

Appendix L Ecoregion Claifications

I

Assessing Conservation Potential and Degree of Threat Among Ecoregions inLatin America and the Caribbean: A Proposed Landscape Ecology Approach

David Olson and Eric DinersteinConservation Science Program, World Wildlife Fund- US

The World BankLatin America Technical Department, Enviromnental Division

March 1994

I. INTRODUCTION

The scarcity of funds earmarked for biodiversity conservation requires that donor agencies and conservationgroups choose wisely among potential investnent opportunities. A prerequisite for establishing a portfolio ofcost-effective investments is an assessment of conservation potential and threats among biogeographic units withina region. A subsequent requirement is to ensure that the regional investment strategy maintains a representativenetwork of conservation areas of sufficient size, shape, number, and connectedness to conserve ecologicalprocesses, viable populations, species, and communities across diverse landscapes. Without addressing theseconsiderations, investments run the risk of being channeled in an ad hoc manner to areas not under serious threat,of relatively low biological value, or which fail to persist over the long-term.

In order to more effectively carry out its role as an implementing organization of the Global Environmental Facility(GEF), and to incorporate biodiversity considerations into its portfolio of projects, the World Bank hasundertaken this study on the conservation status of terrestrial ecoregions of Latin America and the Caribbean. TheConservation Science Program of WWF-US has been contracted to develop the method for this study and to bethe lead group for its implementation. The study also benefits from substantial co-financing from WWF-US.

TIhe Conservation Science Program of WWF-US, has developed a conservation potential-threat index (hereafter,CMI) for countries and selected biogeographic units in the Indo-Pacific region to increase the objectivity ofconservation planning by integrating biological value, conservation potendal, and threat in the same analysis(Dinerstein & W-kramanayake 1993). A second CPTI analysis assesses the conservation status of tropical moistforests by country and biogeographic unit in the same region (Dinerstein et al. 1993). For the current WorldBank/WWF study, the Conservation Science Program has developed a similar method adapted for use among thediverse habitat rypes found in Latin America and the Caribbean Region (LAC). In a broad sense, this methodexamines the viability of habitat and ecosystem conservation efforts in the LAC region by:

* estimating the potential effecdtveness of conservation effors for each ecoregion;

* assessing the degree of disturbance and altemion of natural habitats within ecoregions; and

* providing information on landscape parameters and biological factors that is critical for selecting whichwildlands would be appropriate sites for conservation units.

Implementation of this method will require the expertise of many biologists, conservationists, and other specialistsbased in the LAC region. This project gives these individuals, research units, and conservation agencies the

* opportunity to influence decision-making among several major funding agencies, including the World Bank. Thepool of data generated by this project will be made widely available by the World Bank.

This project provides a layer of information on landscape-level parameters that, in concert with information ongeographic pattems of biodiversity and social, economic, and political factors, is critical for intelligent decision-making and priority-settng for conservation. The results of this study are not intended to be used by themselvesto set priorities. This layer of information is the first of several important filters that together provide an effectiveanalytical tool for conservation planning to guide investments.

1

II. OBJECTIVES

The overall goal of this project is to inprove the quality and objectivity of conservation planning in the LatinAmerica and Caribbean region.

Specific objectives are to:

* develop an ecologically sound method for assessing the conservation potential and degree of thrat ofecoregions;

* define clearly the conservation status of ecoregions;

* develop appropriate conservation strategies for different landscape scenarios and ecoregion types;

* generate a detailed series of ecoregion maps accompanied by relational databases.

Strategic benefits include:

* helping to ensure that the decision-making processes of major conservation donors, such as the WorldBank, have a sound biological basis by emphasizing principles of landscape ecology, ecosystemdynamics, and representation, critical conservation issues that have not been adequately addressed in thepast;

improving the process by which biologists, conservadonists, and other specialists based in the LAC-region can both inform and influence the decision-making process of some major conservation donors;

* providing landscape-level information that can pemuit governments and conservation organizations tobetter coordinate their activities within the context of whole ecosystem rquirements.

III. BUILDING ON PAST EF1FORTS

This study complements previous attempts at setting regional and national priorties for biodiversity conservationin the Latin American and Caribbean region (LAC). Studies focusing on bird endemism (ICBP 1992) deepen ourunderstanding of the distribution patterns of resticted range species across ecoregions. The Parks in PerilPrograms of The Nature Conservancy help demrmine where protected areas rcquire investment to maintainbiological diversity. The documentation found in national strategies to conserve biodiversity provide usefulguidelines for designing investment portfolios for LAC countries (e.g, Ruben Vila & 'Bertonatti 1993). Otherimportant studies, such as the Paseo Pantera Project of the Caribbean Conservation Corportation, idenifyimportant comdors for linking remaining blocks of original habitat in Central America.

To be effective, regional conservation strategies must be based upon a tiered analysis (Fig.1). The first tier, thefocus of the WWF/World Bank study, identifies areas of original habitat within discrete ecoregions whoselandscape features contribute to the persistence of biodiversity. Consideration of landscape feannes such ashabitat size, configuration, degree of intacmess, and connectedness with other habitat blocks, are essential toaddress the four main goals of biodiversity conservation: maintaining representation of all ecosystem types inconservation areas, viable populations, ecosystem processes, and to retain resiliency to short-term and long-termdisturbance events such as fire, drought, and climate change (see Noss 1983, 1992, Saunders et al. 1991 forfurther discussion of these concepts).

The second-tier of a priority setting exercise should overlay known or estimated patterns of biodiversity (ie.,zones of endemism, richness "hotspots", beta diversity gradients) and critical habitats for important ecologicalphenomena (e.g., migration corridors, disturbance refugia, critical resource or breeding habitats) onto thelandscape-level analysis. This method refines the landscape approach by recognizing unique biological attributeswithin each ecoregion (e.g., Austin & Margules 1986, Margules et a. 1988). This step requires detailed analysesat finer geographical scales. Much of the necessary information for this tier already resides in nationalconservation strategies, regional analyses, and species-based studies (e.g., Campbell & Hammond 1989,

2

Ramamoorthy et al. 1993, Twilley et al. 1993), but for many ecoregions, distribution data for most taxa areuneven or lacking. To gather these data for all of the ecoregions under consideration is beyond the budgetarycontraints of this study. Similarly, to begin by focusing on the third tier of analysis, the protected areas of theregion, is also unsatisfactory. Many ecoregions lack protected areas and some may never achieve formalprotection. Some protected areas were created for reasons other than to conserve biodiversity. And finally,setting priorities among protected areas without considering patterns of fragmentation and connectivity runs therisk of focusing too much on isolated blocks of habitat while ignoring the critical investments needed to link orbuffer some of the most threatened reserves. Thus, the landscape approach employed in this study adds anotheressential layer of analysis to complement national biodiversity conservation plans and regional studies that take amore species oriented approach or set priorities only among existing protected areas. The results of previousstudies will be integrated with the landscape analyses in the text accompanying the assessment of conservatonstaus of ecoregions of LAC.

Macr-economic policies, debt burdens, land tenure, migration patterns, civil unrest, and strength of local NGOand government institutions impinge upon the ability to conserve biodiversity over the long-term (e.g., Aylward &Barbier 1992, Buschbacher 1990, Goodland 1987, McNeely 1988, Myers 1993). Analyses of these factors arecritical for effective biodiversity conservation, but are mostly beyond the scope of this project and should beapplied after landscape-level biological assessments. However, we do consider such factors as population densityand development schemes as modifiers for rates of habitat loss and the long-term effectiveness of protected areas,rspectively.

IV. APPROACH

We.assess conservation potential and degree of threat among ecoregions (see definition below). We defineconservation potential as the degree of intactness of original ecosystems within an ecoregion. We define threats asthe factors that undermine the long-term prospects for biodiversity conservation. Threats can be categorizedfurther into short and long-term factors, but many obvious short-term factors, such as hunting, contribute to thelong-tern alteration of natual communities. Alternatively, many long-term thmats, such as fragmentation, impactecosystems over time scales of days to centuiries. Conservation potential and degree of threat are more fullydefined in section VI.

The CPTI approach is designed to beflexible in order to provide greater analyical rigor. The CP77 outlined inthis study erphasizes strict landscape pararneters. It also incorporates several more subjective critenafrom otheranalydtical iers in order to assess their influence on landscape-level processes. Analyses using differentcombinatons of the CPTI criteria will also be conducted to better understand the influence of differentfactors onconservation. Thus, seeningly nore subjective criteria, such as management of protected areas, can be deleted insome analyses. A synthesis of the resuktsfrom a suite of analyses wUIl provide better insight into general trendsand relationships arnong conservazion pararmeters than will any single index.

Most conservation planning efforts use political rather than biogeographic units largely because conservationprograms and funding are administered by national governments and because essential data are often recorded bycountry. We find both approaches essential to intelligent planning. However, ecorgions often span severalcountries along mountain ranges, coastal plains, or river valleys. GIS-based analysis, such as is proposed here,allows us to illustt both political and biogeogaphic boundaries.

Our approach recognizes that all ecoregions have intrinsic biological value, regardless of their species richness. Informulating our approach, we considered ranking ecoregions by their level of endemism, such as in the ICBPanalysis (1992). However, endemic species distributions are typically nested within the boundaries of ecoregionsdefined by this study. To avoid circularity, we do not rak the biological value of ecoregions by their level ofendemism. Instead, our comparative analyses are based on the degree of threat and conservation potential amongecoregions. Analyses of species richness "hotspots" and endemism patterns are important, but most useful ifevaluated within ecoregions (Fig.1). For a more detailed discussion of the endemism issue see section V.

3

priori habitatblks(bhck) i

g. Combining all of the layers of data (or coverages) in a CIS-analysis helps to7'U' _ i C 3 _ * - /~identify conservation priorities for the ecoregion.

e.g., land tenure f. The Fifth tier considers land-use, land tenure, economic activities, and other

-social, economic, or poiltical variables that potentially influence biodiversity._- consce vation, Unlike landscape features and patterns of biodiversity (layers a and

b), negative trends in this tier are potentially reversible or can be miitigated bychanges in legislation.

e.., zones of low

4kFdt - _<gz the.The fourth tier of the lysis examines the impact of human demography onthe conservation of biological landscapes.

Protected 'd. 'e third tier of the analysis exanines the size numnber, and configuration of------- __ f /protected areas within the ecoregion nd the effectivenes, of management

ar ea _ , _1 -within and aound conservation units. Most studies begin setting priorities at this/ I ?-;;)of;t ;t}6level even though most protectcd uta systems have major gaps and some

countris or ecoregiong lack protectd areas. In some cass, the existing networkmay correspond litdte with conservable eas from a landscape perpective or they

eC, elide_ neglect tareas of hig bological value.

c The onalysis ovedays geographic patterns of blodiversity.;A ~ ~ ~ ~ ~ ~ ~~ such as zones of vasying spaces richmness and ndmism, beta-diversity graients

<:T 62 T C=9 B'-/ (ie.. the change in species composition with distance or elevation), unique bioticassemblages, critical habitats for rnigrtoTy species or refugia from periodic

blocks of otluel- disturbance. Unfortnately, data for this lyer are often incomplete.

b. Importnt landscape eatures ae analyzed for each ecoregion. We record theI( C 3.,- J 9 number and size of lrge remaining blocks of original habitat, their proximity and

connectvity, and the pesence of intact watersheds. These features are thebuilding blocks for conservation potential nd ulimately, the long-termpersistence of biodiversity and ecological processes within an ecoregion. Their

habitat fragments loss is essaetially irreversible. We also examirnc other landscape paramtetes that'. __;< ; 0 iS:: ; ; > 7and potential corrldors undermine conservation of biodivcrsity, such as the level of fragmentation and the

extent of habitat loss. This essential analysis has largely been ignored in previousattwWempts to set prioritiesfor biodiversity conservation.

______________._______C Q____ a. The first step is to classify a biogeographic region into distind ecoreglonsthrugh consultation with experts and a review of the literature. This achieves thegoal of repmresenttion -- tht all ecoregions ar considertd in a regional analysis.

Fig. 1. Analyzing several layers of information is essential to developing effective conservation strategies. integration of the variables in this figure provide the basis forsetting priorities within and among ecoregions. By using a GIS-based approach, the most revealing variables can be isolated and overlaid to yield a better view ofconservation potential and threat.

Project Flow & Structure

The successful completion of this project will require extensive collaboration among a broad coalition ofconservation organizations, govenment agencies, local experts, academic and research institutions, and otheragencies (Fig.2). Many of these parties hold databases, remote sensing imagery, and other information essentialfor this study. Outputs of the project and benefits to participants are also illustrated in Fig.2.

The World Bank is providing the bulk of funding for this study and has responsibility for its overall coordination.More specifically, the Bank will be coordinating contacts with LAC govemments, with the GEF, and with variousother intemational and national organizations. The Bank is also supplying data and expertise critical to thesuccessful completion of the study.

WWF will assume the lead role in:

* developing and refining the method to be used;

* circulating the method to regional experts and collating the responses;

* acquuing approprite satellite imagery, coverages, and databases needed to conduct the analysis;

* conducting analyses and producing graphs to detemin the conservation status of ecoregions, andproviding accompanying text.

Terminology and Classification Scheme

We-define an ecoregion as a geographically distinct assemblage of communities that share a large majority of theirspecies and similar environmental conditions, and whose ecological interactions are critical for their long-termpersistence. On the basis of this definiton, some ecoregions dtat contain a mosaic of disinct habitat types, (e.g.,the gallery forests, woodlands, and savannas of the Brazilian cerrado), will be considered as a single ecologicalunit. At the populaton level, an ecoregion boundary broadly delneates the geographical limits of gene flow andmetapopulation dynamics for the large majority of its species. ITe original extent of an ecoregion's habitat typesis used to delineate its boundaries. Cleariy, thin lines on a map are biologically umealistic as boundaries for manyecoregions, particularly those whem ecosystem distinctions are blurrd by gradual enviamnnial grdients,habitat mosaics, or complex topography. In these cases, ecoregions will be distinguished on the basis of majorbiological and physiognomic features identified by previous regional studies. Wherever possible, we have basedecoregion boundaries on the classification system adopted by regional biologists and conservationists in order tobenefit friom the long tradition of biological investigations throughout the LAC region.

The spectrum of ecoregions contained in LAC spans the world's driest deserts and some of its wettest forests.Assessing conservation potential and theats among these diverse cogwo rquires an analytial fmework thatclassifies ecoregions in a biologically meaningful way. We propose to first categorize ecoregions withinDynamically Similar Ecosystems (DSEs) (Fig.3). A DSE is a set of ecoregions that (1) share comparableecosystem dynamics in terms of both function and scale, (2) display similar ecological responses to habitat loss,fragnentation, and degradation, and (3) all require a similar suite of conservation activities appropriate for thatspecific ecosystem type. DSEs are intended to classify ecoregions on the basis of their ecosystem dynamics andnot on their vegeation structure, species and higher taxonomic composition, or geographic proximity.

DSEs are further subdivided into Structurally Simlar Habitats (SSHs) (Fig.3), here defined as a set of ecoregionsthat experience comparable climatic regimes, have similar physiognomic structure, and whose flora and faunashow similar guild structures and life histories. Use of SSHs avoids mideading ecological comparisons andfacilitates the development of robust ecoregion-specific strategies for conservation.

In some cases, several adjacent ecoregions with strong ecological inteactions, but with different DSE or SSHclassifications (e.g., tropical dry forest, grasslands, and wetlands of the Chaco; lowland and montane forests andpiamo of the Sierra Nevada de Santa Mart, Colombia), are secondarily analyzed as an endre ecosystem unit inorder to consider the seasonal movements or migrations of animals, and to analyze more broad biological regionscommonly recognized by biologists and conservationists.

4

|Pardipation by LRAC Exp : | Development of Methodology

Evalua(~tion of Rcftneview

~~ _ Rcspons~ibilidies

___ _ Refine Methodology & Database

Remote Sensing Souces:_______________ ,NGOs, EPA, NASA,

.b Identify Ecoregions GIS Centers

Develop Regional Contactsfor Expert Evaluadon &

Infonnation L Obtain Coverages &Non-Spatial Information

* l I | Ecoregion Analysis Development of Maps, GIS.~~~~~~~~~~~crgo . ,t -i7 databases, and Relational Databases

-~~~ I Graphica Analysis &TX __ GEF: _ t

P1ning InvestmentsT ~~~~~for Futrc Progams

Worid Bank (LATEN), Multi-latal,& Bi-lateral Organizations:

L-_ _ _ A~Cnservation Planng, Prioritzing,& Siting Projects

Conservation Organizatons,I Government Agencies, & Academic I

Institutions inLatin Amenca &United States: Regional Planning ofAcivities, Investments, & Research

Figure 2. A_nang Conervato Potesa and Degree of Threat among Ecoregons of LAtinAmerica & the Caribbean: Project Structure & Flow.

I ~ ~ ~ ~ h I W ! r~~. .. . .. .. . . ... _ _....... ...

TI opkal Broud_ Foresus )A. Tropical Moist Bradla Forests

B. Tropical Dry Broadea Forsts

U. Conifer/Tempee Broalaf Frests ) A. Tenpeate Fors (Examples)

_ . T; upical & $pbo*iplcal FtausW _ 1. Madean Montane ConfferFores (Mexico)

11 GrasslanhWSavmnahsWetlands/Shrublands A. Grssands, Savannahs, Shrblands 2. Misirto Pine-Savannah_UL B. Flooded Grasslands Forst (Nicapa/Hnduras)

. _ 3. San Lun Evergen_ C. Montas onse _s Woodand (Mexico)E A. Meiraneap _ 4. Guenran Evergreen Poest(Mexico)

B. Desers & Xeric Shrublandsetc...

C. Restingas & Dune Vegetadon

(V. =fr No A. Mangroves

Deflnitio; A SB Is a m of cogi tst (1) dwa ooarabh D- id : An SSH is a st of ecor@gions that expede Defnition: An ecoraion is aecsystem dyamiraI, (2) dislay simir ecological raponse to habita comparable climatic regimes. have sinilar physlognomic geographically distinct assemblage of106l, awd d () r e an : - . : whpmc fl and faun sow iailar pild communities tha share a large majority of

, _r ,pgc. , .''-',,'',:,'- trUcte aso i . . :. -. heir species and sismia envirmnientac--ditions, uid wbse cologicalIn=eraeui sitre cidcal farlongsnum perdsstece.

Figure 3. Our classification schemne identifies three hierarchical levels defined in the boxes above. At the broadest scale, we identify five DynamicallySimilar Ecosystems (DSEs). Then we separate DSEs into 11 Structurally Similar Habitats (SSHs) to provide an interrnediate level of resolution. Finally,we subdivide SSHs into ecoregions. This graphic illustrates how DSE 11 (Conifer/Temperate Broadleaf Porests) can be subdivided into 2 SSHs, andhow SSH "B" (Tropical & Subtropical Forests) splits into several ecoregions. The unit of analysis for this study will be the ecoregion. This methodorganizes ecoregions into a formnat that encourages appropriate comparisons which will lead to ecoregion-specific strategies for conservation.

Criteria for Calculating Index Values and Ranking Ecosystems

Conservation potential and degree of threat indices pemit classification of ecoregions into rad categoies ofecosystem viability. To effectively address the distinct ecological dynarnics of the five DSEs, different sets ofindex criteria, threshold values, and weightings have been developed for each one. All of the ecoregions within agiven DSE will have their conservation potential and degree of threat analyzed using the same set of criteria andweightings.

Comparative ecoregion analyses (primarily descripdive analyses) conducted at the SSH level will yield the mostbiologically meaningful results because of similar ecological characteristics among ecoregions within an SSHDSE-level analyses will also be carnied out in order to provide a broader biogeographic perspective (e.g.,answering questions such as how much more threatened are tropical dry forests than wet forests).

We recogmze that comparisons among disparate ecoregions, essentially between different DSEs, may be viewedas a practicable planning tool by some users. However, it defeats the purpose of The World Bank's originalintention for this study, which was to ensure representaiveness among all ecoregions. Further, substantialdifferences in the dynamics and structure of LAC ecosystems heighten the probability of deceptive andmeaningless conclusions in all-inclusive analyses. The calculation of conservation and threat indices is discussedin greater detail in section VL

The inrdices are employed to help us more clearty understand the ecological status of an ecoregion relative to that ofits original state. Classifying ecoregions on the basis of the degree of intauness can guide conservation strategiesfor each ecoregion (Fig.4). Although this analysis emphasizes the influence of landscape features on the long-term persistence of ecoregion biodiversity and ecosystem dynamics, it is not intended to identify ecoregions thatare too disturbed or too intact to warrant conservaion investment. Each ecoregion is assuned to be equalyimportant in its contribution to biodiversity. The conservation potenta and threat indices are usefiu in identifyingthe paricular set of conservation actions that are most appropriate for different situanons. In largely intactecosystems, a proportionately smaller investment may provide a cost effective way to preserve substantial tracts oforiginal habitat (Fig.4, a&b). Ecoregions that are significantty altered, degraded, andfragmented may requiresignificant conservation investment in order to increase theirprobability ofpersistence and ensure theirrepresention in the future (Fig.4, c&d).

Graphical Analyses and the Development of Appropriate Conservation Strategies

The CPTI graph of related ecoregions (ie., those that fall within the same SSH categories) provides a usefulplaning tool for assessing conservation status within each of the categories (Fig.5). The suite of accompanyingillustrative graphs (i.e., based on hypothetical data) establishes the historical loss of critical habitats (Fig.6); thetrajectories of current trends in habitat loss and efforts at protection of biodiversity in large conservable areas(Fig-7); and threats posed by fragmentation of habitats (Fig.8). The fist graph in this suite (Fig.5) represents anaggregate of multiple influences on biodiversity conservation; the latter graphs analyze individual components.These more specific graphs pinpoint threats and gaps in coverage. A variety of statistical analyses wuill be used totest the relative contribution of fators to vaiation in conservation potential and threat, and the significance ofcorrelation among factrs.

Ecoregion Maps and Text

The map coverages msulting from this analysis, data permittng, will illustrate: ecoregion boundaries; the DSE andSSH classification of each ecoregion through color or patten coding, large conservable blocks of habitat; zonesshowing different degrees of fragmentation or habitat alteration (e.g., urbanization or intensive agriculture);location, boundaries, and classification of protected areas; and a symbolic designation of the threat status andconservaton potential of each ecoregion. They will also highlight unique ecoregions that are difficult to classifyand warrant special consideraion. The accompanying text will discuss in more detail the biological value,distribudon of biodiversity, locaton and general prioritization of conservable areas, conservation potental andthreats for each paricular ecoregion, and ecological linkages arnong ecoregions.

5

A _ Bi hab .itoa

ecoregims beudary

Ecueglo A: large condtgous blocks of original habita Ecoregio B: Cluster of rativdey large ad snail blocks bf orgina habitat with vayingCmrvadon StrateU: Ensunm that te proteted mm systm coatains multde lge leves of connectivity and prbability of penistencewell-connected resrve or coesmation a tat adequaely rprent the eccionas CO_mval Straey: Idetify lae blocks of habitat with highest penistence probabiity tochcterisdc bological feat. For Wffer ara. develop guide for fture lad and arAct core of protectd ar system Incorae other rginal habitat block or fragmts intoresoee extracton cdvities an conp e with the vonueviIon of biodiversity. protected aa system to act as codor betwee eserves. Resoe coridrs between contguousConervatik CbaUeogs: CAmmalkae lo al Nteldd te beane of atinmng hbitut wherc ppropriate. Rep senttin sad buffer zone guidleunesa in A.ecosem services aW conserving auml reoures in Itcd ,_- and effecive buffer Consevakdon Ch:alluaus Simila to those of A. but with gawr inediate attntion tozonm Develop bkbdiversty dabae to idntify gap in praotd are m s nu ratioalizing conservaton with exracton of resourcea Consdeable costs associated withadequate mnagmt of Imp potced pro4teion. nensn t. ed resatio of cridca coridos and fram en.Co Kfetbvmm High cot effeciveum with low to nudaate invent requid Codt ecdlv Higb to moeat cost effetivenss with modezte investaut reqrd for thefor th _se f_r nse f

C A< D 104 D

Fawon C: A few laer blocsb of origin babitat ud several naler, widey scaterd Kereglim D: Oginal habitat remidns only in small, isolated fiagments.fagmnts Co_svatdhn Strategy: Includ remaining fagments of original habitat within protected aCo _nvadom Stteg: Incrpote remaining lage blocks of rginal habitat nd critkl system Improve management within reserves and in surrounding areu to reves ecosystcmconidor fragments in a protcted arca system Repqenution, stomtion, and buffer area degradation Expand sves wher appropriate. Reste secondary habitsu to expand size ofguideline as in 3. funcdonal ecosystemn, increase species populadons, crtate conidors, ad enhance the persistenceConserva Cballeges: Low persistence value in most remaining fragments due to of ecosystem dynamicLisolaim snil size, mad edge effects. Major thrU posed by signifrcant expluitatlon of Ceratidon Cbalniage: Similar to those of C but more intense. Low persismnce probabilitiesnata resources contained in original habitat fragnmnt. Lack of redundancy for unique in all remaining frgments without intensive musagemcnt and restoration.biologkal conuutite with proected Cast Effeelveme: Reldvely low cost effectiveness with high Investments requiredCodt EffhctlveS: Modeate cost cffecdvcnes with high to moderate investmnent equired immediatly.i_nndaey.

Fig. 4. Four hypothetical conservation scenarios that illustrate how landscape analyses can guide priority setting. For each, we outline conservation strategies,chalknges, and levels of cost effecliveness and investment requirements. To aclieve the goal of representation, investnenis in all ecoregions are necessary.

categoryI&6 Eco 1 ' Eco 3

Eco2

_----------------------------------- ------- ----------

0 g Eco 4 '*I* Eco 6

0e 0

3 > m~ Eco7

Conservation Potential g(sum of criteria 16

Figure 5. Graphical analysis of the threat to and conservationpotential of hypothetical tropical moist forest ecoregions.Quadrants (I-I) represent four distinct categories whichwarrant different conservation approaches. For example, tropicalmoist forests fall under the Tropical Broadleaf Forest DSEfor which we have identified six criteria to assess conservationpotential and seven criteria to assess degree of threat. Thenumber of criteria will vary by DSE.

(A) - rema ) forested habitat

[Z origial forested habitat

Ecoregion I

Ecoregion 3

Ecoregion 4S ~~~~~~~~Ecoregion 4 / %

Ecoregion 5

Ecoregion 6

Ecoregion 7

Ecoregion 8

90 80 70 60 50 40 30 20 10 0 0 10 20 30 40 50 60 70 80 90

% of Original Forest Remaining Forest Area (x 1000 km2)

Figure 6. (A) Percentage of original. forest remaining for hypothetical ecoregions compared to (B) theabsolute area of remaining fomested habitat. These graphical analyses will be easier to conduct forthe three forested DSEs and more difficult for the non-forested DSEs.

40

35 -

320 ho°

~~~3O ~~~~~AEco 3 -A

25 iEco4

14 ~~~~~~~~~~Eco 620 EcoS 0

o ~~Eco 7

* ofFrsReannin1yer

o 10 Ecoc 0~~~~~

~~~ 0 ~~~~~~Eco 50 6~~~

001 0 6

0 1 0 20 30 40 50 60 70% of Forest Remaining in 10 years

Figure 7. Percentage of remaining forest in protectedareas (PAs), and the subset located in PAs > 1000 lin2,graphed against the estimated percentage of forestremaining in 10 years of set of hypothetical ecoregions.The relative proximity of the hollow circles to the filledcircles indicates the extent to which remaining forests canbe conserved in large, relatively intact areas that have a highprobability of persistence.

100

bO .Ouam Region

0~~~~~

at I~~~Worst Ca*IcategoryI

CA Dry Fort

00 50 100

Degree of Connectivity

Figure 8. The percentage of original forest remaini'ng comparedtO the degree of connectiity of the patches widhin a set of hypotheticalecoregions. Category I ecoregions represent the best case: forestedareas and large and neighboring patches. Category IV ecoregionsrepresent the worst case: forested areas consist of small, isolated patches.

V. IDENTIFYING ECOREGIONS

Classification of Ecoregions

Many different regional experts, maps, and texts were consulted during the process of delineating the ecoregionsfor this study. These include scientists, resource managers, and conservationists familiar with differentecoregions, vegeation maps, ecological zone maps, and maps showing zones of endemism for different taxa.Wherever possible, we have tried to follow the ecological classifications followed by regional experts. All mapsand texts consulted will be fully referenced in the final document. Some modification of existing ecoregion mapshas taken place to better meet the requirements of this study. Ecoregion boundaries and classifications will bereviewed by regional experts. Ecoregions have been mapped onto the digiml geographic base map, the DigitalChart of the World (DCW).

Extant Original Habitat

A variety of databases are being consulted for identifying larger blocks of original habitat for the analysis. Severalnational atases have maps and satellite images depicting remaining habitaL Some of the satellite imagerydatabases being consulted include the Landsat-based Pathfinder project for the Amazon basin, the AVHR dambasefor South America ceated at Woods Hole Institute and Goddad Space Flight Center, and several TM datasets forCentral America. Additional databases are currently being reviewed. Each data set will be carefully interpreted inlight of its date, resolution, gaps, and estimation and classification forest, vegetation, or land use types. All ofthese crtical attributes will be available in the final document to facilitate interpretation by others if desired.Some DSEs and SSHs present significant difficulties in interpretation from satellite or remote sensing imagery. Inthese cases, we must rely largely upon the khowledge of regional experts to gather information on the extent andintergrity of remaining blocks of original ecosysterns.

Disjunct Ecoregions

Some ecoregions are conposed of several discontinuous blocks of habitat due to bioclimatic factos, large riversor mountain ranges, soils, hydrographic conditions, or historical processes. Such formations will be consideredas a single unit for the purposes of this analysis if they are geographically clustered and are lilely to maintain somelevel of biotic interactions between the habitat blocks. Habitat types that are natually disjct, such asmangroves, varzea forests, gallery forests, Amazonian terra firme savannas, and Amazonian swamp forests widtpalms (Junk 1983, Pires & Prance 1985), or similar, but discrete, island habitats (e.g., Windward Islands tropicalmoist broadleaf forest) will be analyzed as biogeographic complexes.

Endemnism & "Hotspots"

Both species richness "hotspots" and foci of endemism have been used to identify and pnoritize natural areas forconservation action (McNeely et al. 1990). These parameters are most important for identfying priorities withinecoregions. The analysis proposed here does not explicitly employ richness and endemism for discriminatingamong ecoregions for sevemal reasons:

* We do not attempt to assess the biological value of ecoregions other than in tes of conservationpotential and degree of threat because we presuppose that all ecoregions have invinsic biological value,regardless of their species richness and endemism. This approach helps avoid the situation where adistinctive, highly threatened, but "species-poor", ecoregion does not roeive conservation action fastenough to avoid losing a unique ecosystem (see Mares 1992, Redford et al. 1990).

* The majority of "hotspots", foci of endemism, and regions of evolutionary potential (Erwin 1991, Vane-Wright et al. 1991) are likely to be nested within the ecoregions defined here, given the scale of thisanalysis and the habitat specificity typically displayed by most endemic species.

* The spatial corelation between richness and endemism is unclear, signaling a cautous usc of them asconservation parameters. Brown (1989) suggests that community richness and endemism in Amazonianbutterflies are negatively correlated. LalW regions of high species richness may represent areas wherewide-ranging species of different biotic zones overlap. This phenomenon is in contrast to the

6

Pleistocene Refugia Theory which suggests that pockets of endemism and richness in the tropical moistforests of South and Central America represent mesic forest refugia within which there was extensivespeciation during drier epochs (Prance 1982). If Brown is correct, the apparent high levels of endemismassociated with species-rich areas may be a consequence of incomplete regional surveys (Heywood &Stuart 1992). Similarly, areas identified as "hotspots" of diversity or high-priority areas may primailyreflect heterogeneity in sampling effort rather than real paterns of biodiversity (Nelson et al. 1990,Peres & Terborgh 1993).

The Importance of Beta Diversity

Beta diversity refers to the rate at which new species are encountered or disappear over distance or environmentalgradients. Tropical moist broadleaf forests, both lowland and montane, and mediterranean-type communities ofChile and Mexico, typically display high levels of both beta diversity and endemism (Gentry 1986, Mooney1977). High levels of beta diversity or the presence of disdnct zones of high endemism within an ecoregionsuggests that, on an evolutionary and ecological time scale, there has been limited interaction between the biodccommunities of distant areas. This lack of interaction may imply that the large-scale ecosystem dynamics ofdifferent areas are relatively independent, and the conservation potential and degree of thrat for each area shouldbe assessed independently. However, many tropical birds, insects, and mammals display seasonal movements ormigations over broad altitudinal or horizontal distances that are often characterized by high beta diversity for othergroups of plants, invertebrates, and birds (Brown 1989, D'Arcy 1977, Kattan 1992, Powell & 1993, Thomas1991). Thus, linkages among apparently distinct ecoregions may be critical for the long-term persistence of teircommunity structure. Because these linkages and patterns of beta diversity for many regions are poorly Inown,we analyze some tropical moist broadleaf forests and ars with complex topography, both predictors of beadiversity, as a single ecoregion unit. Information from future studies may provide a basis for analyzing theseeco.systems with greater resolution.

Marine and Freshwater Ecosystems

This study does not explicitly evaluate aquatic ecosystems; however, it is likely that the level of teresWtrial habitatalteration is corelated to some degree to the viabilty of la-kes, steams, and rivers found within an ecoregiotnIntensive harvesting of aquatic species (McClanhan 1987), increasing rates of pollution and habitat degradation,and the close linkages that exist with bordering terrestrial systems demand that freshwater and marine ecosystemsbe evaluated using a comprehensive method with a unique set of evaluation critera For example, ecoregionboundaries for aquatic systems would be conceptually different than for terrestrial habitats, excepting some lakes.Unlike terrestrial communities, there is often a low correspondence between patens of endemism and otherbiogeographic factors (Allendorf 1988, Sheldon 1988, Sioli et al. 1969). The nature and scale of ecologicaldynamics, ranging from near linear in rivers to multi-dimensional, global-scale processes in marine systems, arenot adequately addressed using terrestrial criteria (Allan & Flecker 1993, Kaufnan 1992). Fmally, harvestingpressure and pollution are often influenced by remote factors and processes such as multinational fishing fleetsand pesticide use (Salm & Clark 1984). WWF has consulted with the Wetlands for the Amenrcas project to ensurethat analyses of wetland ecosystems are ecologically sound and yield practical information. The World Bank, incollaboration with the IUCN and other groups, is preparing a global marine biodiversity conservation strategy.This study is identifying and defining priority protected areas in 18 biogeographic regions, including all of LAC

VI. ASSESSING CONSERVATION POTENTIAL AND DEGREE OF THREAT AMONGECOREGIONS

Biodiversity is best maintained in regions where populations, large-scale ecosystem dynamics, and bioticinteractions persist within their natural range of variation. Natural population levels, both in terms of absolutenumbers and densities, and large-scale ecosystem dynamics are typically altered after landscapes have sustainedsignificant habitat loss, degradation, or fragmentation. In such areas, the loss of populations and dynamics isassumed to be irreversible unless massive (and unlikely) ecosystem restoration efforts are undertaken. Thus, wedefine conservation potential as the probability of maintaining orginal, large-scale ecosystem dynamics andpopulations over a long period of time given the present stae of the original habitat Conservation potentalrepresents the potential maximum effectiveness of habitat conservation, independent of rates of habitat loss. Forplanning purposes, conservation potential can be assessed by existing "building blocks" that (1) contribute tolong-term biodiversity conservation, and (2) estimate the degree to which extant ecosystems still sustain natual

7

population levels and ecosystem dynamics. Positive factors are emphasized in the conservation potential index.

Threat factors are used to estinate the trajectory of the viability of the ecosystem. This trajectory can be thought ofas a relative assessment of the current stability (in the dynamic sense), rate of change, and resiliency of anecosystem's populations and dynamics. These parameters are estimated by negative factors, particularly thosethat are largely irreversible and affect broad geographic areas.

The conservation potential and degree of threat indices emphasize objective criteria, largely based on satelliteimagery, that detemiine the biological state of ecoregions and allow a high level of confidence in results (Fig.9).Our study will include altemative analyses that will enable us to test the accuracy of results and examinerelationships among variables. For example, we will conduct a strict ecological evaluation of ecoregions basedon objective landscape criteria, followed by a subsequent analysis of more subjective human activities thatinfluence conservation. Recommendations from the primary analysis will be remassessed if there is significantdisparity between results. By selectively removing or lumping criteria we can test for broad correlation amongvariables, identify relationships anong factors that are unique to particular DSEs or SSHs, and determine strongpredictors for conservation potential and degree of threat among DSEs.

Conservation Potential

Large Blockr of Habifat

The critical parameters for assessing conservasion potential arm the numnber and extent of lar blocks ofcontiguous habitat within which populations and ecosystem dynamics function naturally. Larger blocks of habitatsustain larger populations of species, an important deteminant of population viability, and they permit a broaderrange of species and ecosystem dynamics with different minimum area requirements to persist The geographiccoverage of larger blocks also allows for a wider range of habitats, environmental gradients, and species ranges tobe included within intact habita

The number of large blocks of habitat in different size categories is an important component of this criterion.Redundancy theory suggests that the presence of three or more systems, ecosystem types in this case,significantly increases the probability of its long-term persistence globally. Factors such as fre, disease,pollution, deforestation or degradation can eliminate species or natral habitats within blocks. The presence ofseveral reserves with similar communities allows for recolonization and the persistence of particular habitat typesand species over time. Multiple reserves, and the redundancy they confer, are particularly important forconserving species and habitats in ecoregions that are characterized by a high degree of beta diversity (speciesturnover with distance or along environmental grdients).

The threshold size for viable habitats is broadly tailored to the maximum scale of important ecosystem dynamicsfor different DSEs. In order to avoid misleading conclusions by applying continental size thresholds to islandecoregions (or very small continental systems), different sets of size trsholds are employed for each broad sizecategory. Scale problems are similarly addressed in the protected area criterion (see below).

Intact Watersheds

The presence of intact watersheds within original habitat blocks enhances conmervation potential because they (1)preserve the full range of altitude-, edaphic-, or fluvial-specific species and communities, (2) allow altitudinalmigrations and other biotic interactions to continue, and (3) maintain environmental parameters and processesdependent upon landscape condition (e.g., hydrography, soil erosion, rainfall, microclimates) within naturallevels. In terms of conservation management, intact watersheds are easier to defend against human disturbances.Forested ecosystems are likely to benefit most from approaches emphasizing intact watersheds as conservationunits because of the important hydrographic properies conferred by intact forests and the high habitat specificityof forest associated species.

Protected Areas

The remaining criteria under conservation potential attempt to answer the question: How adequat are existingconservation units for preserving large-scale ecosystem dynamics and multiple, large blocks of original habitat?Protected areas, broadly defined to include extractive reserves, indigenous reserves, or other areas managed for

8

High -: Itab t Los (CP)

Fira-menten (1'), ,,~~~~~~ ~~~~~~ . ..... .. ..,-B.' m :':... ...Itact Wa...e.

:~~~ag Block of Haita (C?

Wildif Ep (T

Irreversibility Population Density (F)

Protected Areas (CP)

Development Scbemes (I)

Management of Protected Areas (CP)

Low

Low Confidence Level in Data High

Figure 9. Relative irreversibility and confidence levels in data for criteria. Low confidence levels in data may indicatehigh subjectivity or reflect broad confidence intervals in the original data. The shaded box identifies criteria withgreatest predictive value for conservation potential and degree of threat. (CP) conservation potential, (T) threat.

biodiversity conservation, embody this portion of the analysis. The cnteria attempt to identify how well humanshave incorporated large blocks of oaginal habitat into aras that are fundamentally compatible with conservadionstrategies. Strictly protected areas are prefened as conservation units than other areas only partialy managed forbiodiversity conservation. However, due to the difficulties associated with measuring the effectiveness ofconservation units other than strict reserves, both extractive reserves (non-timber products) and indigenousreserves are classified as protected areas in this analysis. The descriptive analysis of each ecoregion will identimportant gaps in the protected area system, that is, natural areas containing large blocks of habitat, critical habitatfor species and ecosystem dynamics, or unique biological features that are currently outside of a formal protectedarea system.

Protected areas are given less weight than the presence of large blocks of original habitat or intact watershedsbecause (1) their statistics do not necessarily reflect the current extent of original habit, large habitat blocks, orviability-of large-scale ecosystem dynamics or species populations, (2) their legal status,'number, size, andgeographic location can rapidly be changed orreversed unlike the natural state of an ecosystem, and (3) mostprotected areas are currently too few and small to encompass complete ecosystems and will only be effective if thesurrounding landscape is managed intelligently for biodiversity conservation. One could potentially use the lackof formally protected areas as a threat factor for an ecoregion. However, a lack of protected areas may notthreaten some ecoregions, due to the region's remoteness, diseases, or rugged teTain. Assessing threats usingnegative criteria (i.e., absence of protected areas) increases the probability of making a poor decision (akin to aType II error) compared to basing conclusions on existing parameters.

Management of Protected Areas

Designated protected areas may also be ineffective for biodiversity conservation if they are poorly managed orhave insufficient support from local communities or national governments. Therefore, we introduce amanagement criterion which modifies the value of protected areas based upon their current effectiveness. Thiscriterion is weighted to identify those parks whose protection services or level of support arc clearly incompatiblewith a conservation strategy. The parameters for this criterion are necessarily broad because all managementfactors are not effective in all situations, and the management of protected areas within an ecoregion may be quitevariable, particularly if the ecoregion crosses national borders. This critrion has proven to be one of the momcontroversial ones used, but its relatively low weighting in the index scheme and its broad ranking categoriessuggest that it is unlikely tO significantly influence conservation potential assessments. To reiterate, the strucueand protocol of the CMI1 allows more subjective criteria to be selectively removed in particular analyses.

Connectivity (text analysis only)

The connectivity among large habitat blocks is a critical issue in determining conscrvation potential of anecoregion. Several large blocks of habitat with some degree of connectivity increases the probability thatmetapopulation interactions will persist, gene flow between populadons will continue, and that species can findrefugia during extreme disntrbance events. The converse of connectivity, i.e. fragmentation, is examined underthe degree of threat index. However, strictly measuring fragmentation will not necessarily provide an accurateassessment of connectivity because clusters and series of fragments may act as effective dispersal cornidors orstepping stones for a wide variety of taxa. Connectivity is also difficult to quantify. Its benefits to maintainingecosystem dynamics and viable populations varies widely depending upon the configuration, placement, and scaleof corridors, intervening land use or habitats, and the taxa under consideration. Therefore, connectivity is bestassessed on a case by case basis in the text analysis rather than thrugh a single ecoregion index. The prevalenceof seasonal movements or migrations by many species necessitates special consideration be given to themaintenance of linkages among habitats and ecoregions.

Management Practices in Bzqfer and Matrix Areas (text analysis only)

Habitats outside of protected areas are critical for biodiversity conservation because they contribute to the effectivepopulation size of species. They also facilitate connectivity and persistence of ecosystem processes. Someunprotected habitats incorporate distinct biological communities not represented within protected aremas. Finally,buffer zones and the resources they provide to humans can take pressure off protected areas. For these reasons,the management of natural resources within matrix and buffer zones can gratly influence the effectiveness ofprotected areas and other conservation programs. The diversity of different resource use patterns and managementactivities throughout the LAC region necessitates a case by case analysis of this criteria in the text descriptions.

9

Degree of Threat

The degme of threat index for an ecoregion describes the trajectory of the long-term potential for ecosystemviability. It is based on existing levels of habitat loss, fragmentation, biotic degradation, and trends in humandemographics and land and resource use. This index reflects past pattems and current trends in habitat loss andalteradon. It also reflects the probability of future changes in the absence of effective conservation action. As forconservation potential, the original state of the ecosystem, in terms of biodiversity and ecological dynamics, formsthe baseline standard for threat assessments and comparisons.

HabitxLoss

Habitat loss has been widely recognized as one of the primary factors contributing to the reduction and loss ofterrestrial populations, species, and ecosystems. This criterion underscores the rapid loss of species predicted tooccur in ecosystems when the total area of remaining habitat falls below a minimun critical level. Although thereis no consensus on the mechanisms or exact thresholds for species loss in different ecosystems, both theoreticaland empirical studies support this general pattern of habitat loss and species loss. Loss of habitat reducesbiodiversity by eliminating species or communities limited to particular geographic localities, by sending the areaof available original habitat below the minimum size needed to maintain cridcal, large-scale ecosystem dynamics,and though the degradation and fragmentation of remaining habitat such that they become too small or isoLated toindividually or collectively support viable populations or maintain important ecological processes.

Habitat Fragmention

Habitat fragmentation is addressed under degree of threat, rather than conservation potential, because the largeblocks of original habitat in a fragmented landscape contribute most to the conservation of large-scale ecosystemdynamics and viable populations. Smaller habitat patches can be valuable in this regard only if they persist withina matrix of disturbed habitat intensively managed for conservation. The conservation value of large blocks shouldbe recognized regardless of the fragmented state of original surrounding habiat. According to WWF analyses, theavailable suite of habitat fragmentation indices do not sufficiently discriminate among fragmentation scenarios toadequately reflect the conservation value of large blocks. The presence of large blocks of habitat is not necessarilycorrelated with the degree of habitat fragrnentation, particularly at intermediate levels of fragmentation.Furthermor, the absence of fragmentation in itself is not a building block for a conservation strategy.Fragmentation is better suited as a criterion under the degree of threat analyses because it helps estimate thetrajectory of an ecosystem's large-scale dynamics and viability of populations, since the most profound effectsfrom fragmentation take place over time scales from days to centuries.

Lumping of the habitat loss and fragmentation criteria is ecologically untenable due to ther potential lack ofcorrelation. For example, sorne ecoregions may have a mlatively large total percentage of original habitatremaining, but be so highly fragmented that they will have low long-term conservation value. Another ecoregionmay have little total area of original habitat remaining, yet have a high persistence value because it is located in asingle block above the estimated minimum critical size for ecosystem viability. Moreover, lumping of criteriagready reduces the discriminative power of the analysis and the real benefit of analyzing ecoregions on the basis oftheir ecosystem-specific dynamics.

Habitat Conversion

Current levels of habitat loss and fragmentation are assessed with the assumption that associated changes in theecosystem have already occurred or will be reached in the future mgardless of conservation action. The remainingcriteria for the threat index measure the potential for fmrher habitat loss and degradation above and beyond thatwhich is estimated by extant levels of habitat loss and fragmentation. The rate of habitat conversion criterionprovides an approximate estimate of future habitat alteration based on records from the previous decade. Becauseof the inherent uncertainty of these estimates, only broad categories of habitat conversion are used for this criterionin order to avoid misrepresentation of an ecoregion's status.

HabitatDegradation

Habitat degradation resulting from human activities such as selective logging, pesticide exposure, burning, and

10

overgrazing can have profound impacts on the long-term viability of ecosystens. However, the ecologicalintegrity of many namural communities can be difficult to assess because degraded systems can be indistinguishablefrom pristine areas in remote sensing imagery. Moreover, the measurement of habitat degradation is problematicalbecause (1) habitat degradation is often patchy, (2) degraded states form a continuum and ame not easily classified,(3) the ecological effects of different forms of degradation are unclear and may occur on the scale of decades tOcenturies. For these reasons, habitat degradation will be primarily addressed in the descriptive analyses of eachecoregion. The degree of habitat degradation caused by burning and gazing in altred areas is considered underxeric and grassland/savannas/wedand/shrubland DSEs.

Wildlife Exploitation

Curent levels of wildife exploitation may also influence the degree to which ecosystems can rcbound to theiroriginal state. The loss and reduction of top predators, large frugivores, keystone tee species, and other criticalspecies thrugh heavy exploitation can have proound and enduring impacts on ecosystems through secondaryextinctions and shifts in species abundance and biotic intcrations. In most cases, top predators and keystonespecies rarely rebound to original population levels, even after well-supported species conservation programs.Thus, the level of wildlife exploitation is an additional modifier of potential changes in ecosystems over time.

Human Density and Development

Population density and the presec of development projects reflect human demands for an ecosystem's nauralproducts, land, or underlying geologic resources. These pamameters are not given geat weight in the index;effective conservation and development progrms have the potential to alleviate human pressures on naturalhabitats in many situations. The conservation implications of these two factors are discussed more fully undertheir respective criterion descriptions.

Alternative Schemes

The effects of many of the factors contibuting to the altration and conservation of natural ecosystems arc difficultto predict because of the frequency of correlations among them. Several fars could be argued as most relevantto conservation potential or to the degree of bieat depending upOn whetber they are assessed in positive ornegative terms, or upon the experience and taining of their proponents. For these reasons, we acknowledge thata number of altnative analyt schemes ae possible, but we argue that the present method represents a rigorousand scientifically sound approach for detrmining the large-ale ecological status of ecoregions given the availabledata and time. Although we recognize the critical role and influence of factors associated with human activities,they are not employed as primary discrinao because of the project's focus on original ecosystem dynamics,population levels, and biodiversity patterns. In response to concerns that we are proceeding with landscape-levelanalyses prior to the availability of comprehensive data bases, we contend that the prnciples of conservationbiology, although young and under continual rfmnement, ae supported by numerous empirical case studies andmuch theory. To hesitate in implementing these prnciples because comprehensive data are lacking would bedisastrous given the pace of disturbance of natual ecosystems. Over the next few years, decisions will, andmust, be made on the basis of the best available information about what steps are necessary for the conservation ofbiodiversity.

We intend to retainflexibility in our approach and will modify weightings, thresholds, and even criteria, asimportant panerns emerge from the data andfrom dialogue with ewperts. Staisical analyses, such as correlationand analysis of variance, will be applied to the data set to test assuwtions and idendfy relationships amongfactors. Several different analyses using different combinations of criteria will be carried out in order to betterunderstand the importance and intrrelationships of difrentfactors.

Index Range

Each index has a point range from 0 to 60, with higher values denoting a higher level of conservation potential ordegree of threaL The range from 0 to 60 was deemed apopriate because previous experience with rioridzingand ranking large numbers of conservation amas has shown that a namower index scale (e.g., from 0 to 30) yieldslittle discrimination among units (WWF Russia Biodiversity Project). Broader index scales (e.g., 0 to 100) creatca false precision given the accuracy and level of resolution of the available data and the limited nunber of criteriaused.

11

Weighting Criteria

Within each criterion, the distribution of points awarded is designed to reflect real biological processes or therelative contribution of a particular situation to long-term biodiversity conservation. For exarnple, both empiricalevidence and theoretical ecology suggest that species loss and secondary extinctions increase dramatically withextensive habitat loss and fragmentation (Simberloff 1992, Terborgh 1992). In some cases, the assigned pointvalues closely reflect the relanonship between the causal fator and the ecological response (ie., plotting of thepoints will approximate the curve of the general reladonship). More subjective criteria are divided intostraightforward categories that facilitate classification into one of a few broad conditions. If misclassification doesoccur, these criteria are typically assigned few total points and, relatively speaking, have less effect on assessingthe overall status of an ecoregion. Certainly, we expect reviewers of this document to suggest changes in theexact point values and spread for the crition ranings. However, we are confident that the relative randdng andweighting of criteria will provide a sufficient level of discrinination among ecoregions to reach biologically soundassessments of their conservation status. The original databases will be available to all future investigators whomay wish to reanalyze the data using different indices, criteria weightings, or index ranges.

Multi-National Ecoregions

Because some ecoregions span the boundaries of two or more countries, it will be necessary to estimate the overallecoregion stus for criteria that requim country-specific infoimation (e.g, managemnent of protected areas).Estimation techniques are described under criteria descriptions where appropriate. We use broad categories tofacilitate classification.

Criteria for Five Dynanically Similar Ecosystems (DSEs)

As stated previously, these criteria are not intended to be used as a basis for setting prioritiesfor conservation investments by themselves, but only in concert with the other critical layersof information needed to develop conservation strategies (Fig. 1). The criteria can becombined or aggregated to suit the needs of different analyses. The criteria used for analyzingeach DSE and their relative weighting are summarized in Table 1. They are combined here intwo indices, conservation potential and degree of threat; these indices help identify theconserYation status of each ecoregion. Only by examdning the other critical fillers for eachecoregion (Fig. 1) can detailed conservation strategies be defined.

1. TROPICAL BROADLEAF FORESTS

A. Conservation Potential

Conserving biodiversity in tropical broadleaf forests (moist, dry, and montane) requires that viable ecosystemsand populations are preserved over a geographic area sufficient to encornpass the majority of species. Tropicalforests typically display a high degree of beta diversity, the change in species composition occuning with distanceor along environmental gradients (Gentry 1986, Olson 1993, Terborgh & Wimter 1982). Therefore, a protectedarea system must be sufficiently dispesed over the landscape to conserve an adequate rpresentation of theoriginal biodiversity of the region. Because many tropical trees and animals ypically occur at low densities (Aritet al. 1990, Elton 1975, Hubbell & Foster 1986), large tracts of undisturbed habitat are necessary to maintainviable populations, as well as important biological interaCtions.

Criterion 1: Large Blocks of Original Habitat

Rationale, Tropical forest conservation hinges on preserving blocks of habitat of sufficient size, shape, andlandscape orientation to permit ecosystem dynamics and populations to fluctuate naturally. Only large, contiguoustracts of forest provide sufficient area of core habiau for populations to withstand depletion by hunting,harvesting, competition with exotic species, and physical edge effect. For example, Cacids (guans and theirrelatives), monkeys, and large cats may be depleted or extirpated within 10 to 40 Im from the edge of the blocksdue to hunting (Redford 1992, Robinson & Wilcove 1989, Thiollay 1992). For populations of most large forest-dwelling vertebrates, large tracts must be preserved with substantial buffer zones to maintain viable populations.

12

Table 1. Relative weighting of diffaent citesia used for anaysis of each Dynamically Similar Ecosystem (DSE).

L CeMrvba PeeAWiladx

5d& Totdl Poift % Total Poits % Totl Pobts Total PIaits % Total Points 9bLarge bloimk oforIginshabita 36 60% 36 60% 45 75% 45 75% 45 75%Iaact Waenhads 10 17% 10 17% - - -

ProtectedAnn 12 2D9 12 20% 12 20% 12 20% 12 20%Ma"_me PA'a 3 5% 3 5% 3 5% 3 5% 3 5%

U. DqIV. Of T,wsludex

Lossof uKa 20 33% 20 33% 18 30% 18 30% 35 58%Habitat 17_neuL 15 25% 15 25% 11 18% 11 18% -

Habitat Convaion 10 19% 10 17%. 8 13% 8 13% 15 25%Grazg & Buing - 11 18% 11 18% -

Dewvdopm Schemes 6 10% 6 10% 5 8% 5 89% 10 17%WOiO Exploitatio 6 10% 6 10% 4 7% 4 7% -

Poputi Denity 3 5% 3 5% 3 5% 3 5% _

Moreover, populatons of some species may be reduced in habitats adjacent to hunted or deforested areas due tolarge-scale population movements (Ihiollay 1992). Moreover, large verbrate predators whose prodadon patenshelp shape community assemblages and structre often require very large areas of undisturbed forest to maintainviable population sizes. For example, pairs of harpy eagles have been estimated to require 100 km2 home ranges(see Robinson & Wilcove 1989), and much larger areas would be needed to maintain viable populations over thelong-term.

Large unbroken blocks of trpical forest conserve several important components of biodiversity, and because oftheir relative rarity, they should be highly valued for their conservation potential (McCloskey 1993). The buildingof roads into roadless areas generally has resulted in increased hunting and deforestation (Gradwohl & Greenberg1988, Sader & Joyce 1988). The presence of a road in a large block of habita is a cause for alam, and shlould beevaluated for possible increase of fragmentation and human access.

me Remote sensing (e.g., satellite imagery, aerial photographs) and maps will help identify remaining largeblocks of habitat within each ecoregion. Local experts will be consulted to assure accurate interpretation. Forexample, in the Atlantic forest region in Brazil native overstry trees are used to shade cocoa plantations makingthese areas difficult to distinguish from undisturbed forest through satellite imagery.

Ecoregions with original extent of habitat a 3,000 km2

Large blocks of original habitat remain within ecoregion:one or more blocks 2 3,000 km2, or 3 or more blocks 2 1000 km2 36

Two blocks of habitat > 1000 kan2 but < 3,000 km2 28

One habitat block > 1000 km2 but < 3,000 kn2 24

3 or more blocks > 500 hn2 but < 1000 km2 currently intact 20

One or 2 blocks > 500 km2 but < 1000 km2 curently intact 10

3 or more blocks Ž 250 km2 but < 500 km2 currendy intact 5

No blocks 2 250 km2 remain 0

Ecoregions with original extent of habitat < 3,000 kn 2, but 2 1,000 km2

Large blocks of original habitat remain within ecoregion:one or more blocks > 1,000 km2 , or 3 or more blocks > 500 km2 36

Two blocks of habitat 2 500 km2 but < 1,000 hn2 30

One block of habitat 2 500 km2 but < 1,000 km2 20

Three or more blocks of habitat 250 km2 but < 500 km2 10

One or 2 blocks > 250 km2 but < 500 km2 5

No habitat blocks > 250 km2 mnain 0

13

Ecoregions with original extent of habitat < I,000 km2 but > 100 km2

Large blocks of criginal habitat remain within ecoregion:one or more blocks > 500 km2 36

T`1ee or mow large blocks of habitat 250 km2 but < 500 km2 30

One or 2 blocks > 250 km2 but < 500 km2 20

Three or more blocks Ž 100k h 2 but < 250 m2 10

One or 2 blocks Ž 100 km2 but < 250 km2 5

No habitat blocks > 100 km2 0

Ecoregions with original extent of habitat S 100 km2

MOM than 80% of oiginal habitt remain intact 36

Between 61% and 80% of orgna habitat remain intc 30

Between 41% and 60% of original habitat remains intact 20

Beween 10% and 40% of original habitat remains intact 10

Between 1% and 10% of original habitat remains intact 5

Less than 1% of original habitat remains 0

Criterion 2: Presence of Intact Watersheds

Ratdaje: Intact watersheds rweceive promience because theirpy and biological structure promotebiodiversity conservation i a number of ways: intact altitxdinal transects preserve altitudinally restrictedcommunities and allow alftiudinal, horizontal, and fluvial migraions to proceed; upsuteam pollution and siltationare eluninard hydrographic and microclimatic conditions ae maiunain; protected area boundanis are easilyidentified and legally defended (this is more difficult in lowland regions); river access by humans can besomewhat controlled by rsource managers; fluvial-based paterns of genetic divesiy am peserved; and largearas are typically contained within entire watersheds (see Pems & Taborgh 1993).

In many tropical and temperate regions, intact watsheds represent the last refuge for viable populations of manyvertebrat species (Peterson et al. 1993). Iarge aes of intact habitat do not necessarily mean that intactwatersheds are present. This is pardcula true in regions where only montane forsts are left and the blocks ofremaining habitat run along the steep crests of ranges.

Medtod: Coverages of remainig original habitat will be overlayed with watershed coverages within eachecoregion. Different ecoregions will be assessed using appropriate tributary levels. Terry tributy watershedsmay be appropriate for Amazonia while primary river waterheds may be applicable to smallUer islands in theCaribbean.

Three or more intact watersheds 9

One or 2 intact watersheds 6

No intact watersheds 0

14

Criterion 3: Protected Area Network

RadonalL Landscape-scale conservation strategies use protected areas as important core habitat because (1) theyprovide habitat for species unable to coexist with human activities; (2) dispersing individuals and propagules frompopulations in these core habitats may maintain populations in non-protected matrix areas; (3) large tracts of forestin protected areas modulate the rainfall regime, maintain stzam volume, control floods, reduce soil erosion, andmodify local temperatures for both the regional ecosystem and local communities; (4) and they engender acommitment to conservation on the part of the local communities and national governments.

The rankings used for this criterion reflect the need for redundancy, or the presence of multiple examples ofhabitat types within protected areas (Quinn & Hastings 1987, Walker 1992, Quinn et al. 1992). The contributionof a protected area system to conserving a representative armay of species and communities will be enhanced if itencompasses a wide range of altitudinal zones.

A protected area is defined here as a conservation unit managed primnarily for biodiversity conservation and thateffectively maintains species assemblages, ecosystem processes, and population numbers close to their naturalstate. Indigenous and extractive reserves are included within this definition even though there can be considerablevariation in their contribution to biodiversity conservation. National forests or forest reserves whose function isprimarily for the production or storage of timber are not consided here.

Many of the reserves established for indigenous people can act as reserves for regional biodiversity. Indigenousgroups are often assumed to have a thorough knowledge of their local ecosystems and utilize the resources moreefficiently and sustainably than do colonizing peoples (see Denslow & Padoch 1988, Dove 1993, Redford 1991).However, the presence of indigenous groups may create very intense hunting pressure in a region, particularly ifhudting is tied to a market economy and access to the full extent of traditional hunting grounds is reduced(Redford & Stearman 1993). The trade of wildlife (e.g., parrots and tropical hardwoods) may be similarlyeffected.

Extractive reserves often cover significantly larger areas than protected areas in some regions, and may act as theprimary land area for conserving viable populations and biodiversity in some cases (see Browder 1990,Hartshom 1989, Lugo et al. 1981, Salafsky etal. 1993). Many species of plants and animals are able to persistin managed areas (Brown & Lugo 1990). While providing suboptimal habitat for some taxa, managed areas mayact as effective corridors for dispersing or migrtory species, provide additional rsources for populations inprotected areas, and allow species to maintain adequate population size and genetic variability. Extractive reservesmay also contribute to the stability of local economies and, consequently, protected areas (McNeely 1988).

Metho: Protected area maps and daabases will identify reserves. Reserves that have not been fully demarcated,but have a high likelihood of attaining reserve status in the nearffture, will be included.

Raa igs

Ecoregions with original extent of habitat 2. 3,000 km2

Three or more protected areas: 3 or more > 500 km2 12

Three or more protected areas: with 2 a 500 km2 10

Three or more protected areas: no rserves > 500 km2

but has at least 3 reserves > 250 km2 7

Three or more protected areas with at least 2 reserves a 250 km2 4

Protected areas present, but no reserves> 2.50 km2 2

No protected areas exist 0

15

Ecoregions with original extent of habitat 2 1,000 i, but < 3,000 km2

Tbree or more protected areas: at least 2 > 500 km2 12

Three or more protected areas: with 1 2 500 km2 10and 2 or more > 250 km2

Three or more protected areas: no reserves > 500 km2

but has at least 2 reserves Ž 250 kmn2 7

Three or more protected areas: at least 2 rsmeves > 250 km2 4

Protected areas present, but no reserves > 250 km2 2

No protected areas exist o

Ecoregions with original extent of habitat > 250 km2, but < 1,000 k 2

Three or more protected areas: 3 or more > 250 km2 12

Three or more protected areas: with 1 or 2 > 250 kan2 10

Thmree or more protected areas: no reserves Ž250 kn2

buthas at least 3 reserves > 100 bIn2 7

Protected area system has only 1 or 2 reserves > 100man2 4

Protected auta system has no reserves Ž 100 km2 2

No protected areas exist 0

Ecoregions with original extent of habitat < 250 ka,2

Over 80% of original habitat under protected status 12

Between 60% and 80% of original habitat incorporated into a prouxted area system 10

Between 409%o and 59% of oiginal habitat incorporated into a protected area sysem 7

Between 21% and 39% of original habitat incorporated into a protecred area system 4

Less than 20% of original habitat incoporated ino a protected area symem 2

No original habitat incorporated into protcted areas 0

Criterion 4: Management of Conservation Areas

Badnnale: The ability of a protected area to contribute to consrvation depends to some degree on the commitmentcompetence, and reliability of the reserve admninistmtion and the continuing economic and political support of thenational and regional authorities.

We view conservation arta management in a broad context Impcrtant o siratio include support from localauthorities, the potential for cooperation with acive grassroots NGOs and local communities in buffer zones, andthe level of scientific and technical expertise of local resource managen assigned to consevation nits. Ibiscriterion will be less important for very large or rmote serves that have some dme befome they are significandy

16

disturbed. The management of a protected area will directly affect the frequency and intensity of violations in theconservation zone and the potential for degazettng. The effectiveness of management ia important forest tracts thatlack formal protection will likely reflect the quality of management for protected areas.

Because of the diversity of ecosystems and human societies in the region, there is no single set of factors thatpromotes effective management. However, seven factors have been identified that can provide an indication of theadequacy of management (Cifuentes 1993): (1) the protected area is legally esmablished; (2) the borders are clearlydemarcated, (3) there is a permanent physical presence of reserve managers; (4) annual budgets are available; (5)trained technical staff are present; (6) a management plan exists; and (7) annual operational plans are developed. Awell managed protected area should meet the majority of these requirements.

Mieto: Regional experts (e.g., academic and research institutions, local NGOs, govermment agencies) wil beconsulted and appropriate literature will be reviewed. A low management ranking does not necessarily imply alack of commitment on the part of resource managers, but may solely reflect insufficient resources. The rankingscheme of this criterion is potentially biased against those ecoregions where more protected areas exist because, insome cases, a single responsible agency may have to distribute its limited resources to a greater numnber of units.However, we felt that there was no satisfactory way to correct for this in the ranking scheme and that overall, anybias that may occur would be infrequent and unlikely to affect the general accuracy of the index.

Percentage of conservation areas reasonably well managed over 50%(i.e., meets at least 6 of the 7 management criteria oudined above) 3

Percentage of conservation areas reasonably well managed between 25% and 50% 2

Less than 25% reasonably well managed 1

No conservation areas reasonably well managed 0

B. Threats

Threat criteria are assigned points in proportion to their relative influence on the long-term conservation ofbiodiversity. For forested habitats, we distinguish between habitat loss, fragmentation, and rate of deforestation.We do not address forest degradation because of the limited available data and conflicting approaches to measuringdegradation and its ecological consequences. However, there is evidence that degradation processes, such asselective logging, understory fires, and harvesting of wildlife and plants, have the potential to seriously alterecosystems (see Home & Hickey 1991, Johns 1988, Robinson & Redford 1991, Uhl & Kauffman 1990, Uhl &Veira 1989). Habitat loss and fagmentation are particularly strong threats for tropical broadleaf forists relative toother DSEs. Tropical forests typically have high levels of beta diversity and species commonly have restrictedgeographic ranges, (i.e., even a small loss of habitat has the potential to eliminate a species) (Gentry 1986, Kattan1992, Terborgh & Winter 1982, Thomas 1991). The large habitat areas required to maintain viable populations ofspecies and ecosystem dynamics are a consequence of low population densities and low reproductive rates typicalof tropical broadleaf forest species. The point system ranges from 0 to 60, with higher values denoting a higherdegree of threat.

Criterion 1: Loss of Original Habitat

Rationale: Two thirds of all species designated as endangered or threatened have declined in population as a resultof habitat loss. Habitat loss can cause extinction in a variety of ways: through the complete elimination of aspecies' habitat; by reducing the availability of resources below an adequate level to maintain populations; byeliminating important climatic, seasonal, or disturbance refugia; by removing important corridors for migration orbreeding grounds; by allowing exotic species, human activities, anthropogenic fires, and edge effects to reducepopulations below a critical size; by sending the area of available original habitat below the minimum siz needed tomaintain critical ecosystem processes; and through the degradation and fragmentation of remaining habitat (seeJanzen 1986, Saunders et al. 1991, Terborgh 1992). The weighting is intended to reflect the accelerating loss ofspecies with increasing habitat loss as predicted by species-area relationships (see Heywood & Stuart 1992,

17

Humphreys & Kitchner 1982, Simberloff 1992).

Methd Estimates of the remaining habitat area will be compared to estimates of the original habitat cover. Mapsand remote sensing imagery help to estimate current cover, while literature provides original cover estimates.

>90% of original habitat lost 20

50% to 89% lost 16

25% to 49% lost 10

10% to 24% lost 5

0% to 10% lost 0

Criterion 2: Degree of Habitat Fragmentation

Rationale: Persistent small population size is now widely perceived as the major threat to conservation of terrestrialspecies. For threatened species, habitat fragmentation places many low-density species in demographic jeopardy(Berger 1990, Laurance 1991, Newmark 1991, Wilcove et at. 1986). Fragmented ecosystems experience edgeeffects from hunting pressure, fires from surrounding human activity, changes in microclimates, and invasion ofexotic species over a large percentage of their intact habitat area (Lovejoy 1980, Saunders et al. 1991, Skole &Tucker 1993). In other words, as fragmentation increases, the amount of critical core habitat area decreases.Fragments under 100 km2 are inadequate for maintaining viable populations of most large vertebrates. Somespecies of birds, trees, and butterflies in tropical forests that typically occur in very low densities, or haveextremely patchy distributions, may also be lost in fragments of this size (Bierregaard 1986, Lovejoy 1980).Tropical forest ecosystems are most vulnerable to fragmentation because of the high proportion of species that arerare or maintain patchy distibudons, the importance of long-distance pollination and seed dispersal (Janzcn 1983,1987, Kubitzki 1985), the abundance of mutualistic interactions (Forsyth &8 Miyata 1984, Janzen 1983), largearea reuirements for maintamiing viable populations even in small vertebrate species, and the sensitivity of thebiota to edge effects. Certainly, the problems of large-scale ecosystem disruption related to fragmentation are lessadvanced in ecoregions that still maintain large blocks (>1000 km2) of intact original habitat. The ranking systemof this criterion rflects the greater severity of ecosystem threats in landscapes where habitat fragmentation is moreadvanced (Groom & Shumaker 1993).

The degree of connectivity among remaining habitat patches will vary depending upon the type of taxa orecological processes being considered and the type of land cover in areas surrounding blocks of original habitatSome birds, mammals, plants, and invertebrates can easily cross large expanses of non-native habitat or physicalbarriers, while a single road or narrow clearing may represent a significant dispersal barrier for many forestspecies because of physical or behavioral constraints. Many forest understory invertebrates, plants, and birds fallinto this latter category.

Metho: Satellite imagery and maps of extant original habitat cover will be examined to assess the degree of habitatfragmentation in each ecoregion. The accuracy of these data for particular ecoregions will vary depending on thedate of the information. The components of habitat fragmentation that need to be addressed are number offragments, the number of fragments relative to total area, size distribution of fragments, the shape of fragments,time since isolation, edge effect parameters, altitudinal and topographic orientation of fragments, land usecharacteristics of the surrounding matrix, and the isolation and connectivity of fragments. No availablefragmentation index has been able to adequately address even the most. We will examine inter-fragment distancesdirectly, and, indirectly through our model, fragment size distribution, number, and isolation. These reprsentsome of the most critical fragmentation parameters, other important variables such as intrvening habitat types amdifficult to quantify within the time frame and scale of this analysis.

18

Inter-Fragment Distance Analysis

The relative connectivity amnong fragmnents within different ecoregions can be estimated using GIS-based inter-fragment distance analysis. The proximity of neighboring fragments in all directions can be assessed bycalculating inter-fragrnent distance curves (Fig. 10). External buffers over a range of widths are constructedaround each fragment The group ratio (ie., number of new discrete units formed after buffering divided by theoriginal number of units or fragrnents) is calculated for each buffer width. An inter-fragment distance curve isestimated by plotting the group ratio against increasing buffer widths. The buffer widths used are 2, 4, 6, 8, 10,15, 20, 30, and 40 km. These widths largely represent potential dispersal distances of larger vertebrates, strongflyers, and some wind or animal dispersed plants, fungi, and invertebrates. The majority of species in moisttropical forests, primarily plants, invertebrates, and fungi specialized on forest environments, are poor dispersersacross even narrow distances of non-forest habitat Fragments smaller than 30 km2 will be buffered out at 1/2 theprescribed buffer width to reflect their lower value as long-term source pools or transit habitat for dispersing largervertebrates and other species.

The Model

The index will be calculated from the following model. The inter-fragment distance curve can be considered theresult of joining two different linear functions of the form F(x)= ax + b, and the inflection point of the curve willbe the johiing point (Fig. 11). Extrapolating from the inflection point for x=0 and F(x)=0 we obtain a, b, and h.The first linear function represents the joining of local fragments, most important for short-distance dispersal andthe movement of most taxa and associated ecological processes. The second linear function represents the joiningof relatively isolated clusters of fragmented habitat which are important for long-distance dispersal of largervertebrates, strong flyers, and some wind or animal dispersed species. The slope of the first line (ml) is the ratioof h/b and, although g is unknown, it can be calculated from the slope of the second curve (m2), which itself isestimated using two points generated in the previous calculations (m2'). Therefore, the estimated value of g (g')will be calculated from the product of mr' and a, as the following equations show:

hgm 1=-, m2 = *

b ~~a

The model pedicts that better case scenaros would fulfil the following conditons:

a-0; h-0; g0; m1,A,m2 0.

b

The amount of grouping in relation to distance, i.e., the Inter-Fragment Distance Index (IFD), would then becalculated as:

I (100b)(lOh)g

Multiplying b by 100 transforms a ratio to a whole number, while multiplying h by 10 is intended to reflect thepoor dispersal ability of most species and, consequently, the importance of short distances between fragments forthe persistence of dispersal and other spatially-dynanic processes.

19

In t e r - fr a gme n t Di s t a n c e Cu r v e s

*.. f f t r D I staac

Figure 1 0. Examples of inter-fragmnent distance curves for thre scenarios, of fragmented landscapes.

1.0a

a ==tia lie uco

inflection point

Best cae scanio:2 b ml andm2 - 1

ml= hlb and g==n2whweg'is de estimated value of S. andm2' is the esimad value of m2

hi g

Buffer Distance

Figure 11. Graphical interpretation of important parameters in hypotheticalinter-fragment distance curve.

Interpretaton of the Model

Although large blocks of contiguous original habitat represent the best situation to conserve biodiversity, severalfeatures of fragmented landscapes can contribute to the persistence of species dispersal and other large-scaleecological processes:

* landscapes with short distances (e.g., between 0.5 and 5 kin) between nearby fragments;* fragments having multiple, nearby neighbors that are well-distributed around its perineter,* clusters of fragments that are relatively close to other clusters of fragmnents or large blocks of habitat (e.g., 20

km or less); and* intervening habitats closely resembling original habitat

The model described above captures the first three of these feaus to some degree. The slope of the first portionof the curve reflects the joining of nearby fragments. Steep curves signify that average nearest-neighbor distancesare relatively short The height of the inflection point reflects the proporion of the total number of fragments thatare relatively close to another fragment. How quickly the top pordon of the curve reaches unity gives a relativeassessment of how far apart clusters of fragments are from one another, on average. Again, it is the first portionof the curve and the inflection point that is most important to the majority of species because of their limiteddispersal abilities. Inflection points that occur prior to 10 km buffer distances would generally reflect a betterconservation scenario than those found at greater distances. The index value should range from 0 to 1 with valuesclosest to 1 representing a better conservation scenario. Modifications of index thresholds will be made asdifferent conservation scenarios are interpreted. Thus, the index ranges given below are largely illustrative.

Lowv connectivity: Index values between 0 and 0.33 15

Intermediate connectivity: Index values between 0.33 and 0.66 9

Hfigher connectivity: Index values between 0.66 and 0.9 4

Well-connected landscape: Index values between 0.9 and 1 0

Criterion 3: Rate of Deforestation

Rationale: Deforestation rates assess the overall loss and fragrnentation of habitat, but at the ecoregion level, it iscrucial that the rates be looked at in relation to the total amount of remaining habitaL For example, in regions thatstill contain large areas of intact forest, high deforestaion rates do not necessarily portend the immediatedestruction of that ecosystem. Change in land use pattems may reverse the trend before the area of habitatbecomes too small for long-term persistence. Slower deforestation rates may present much more of an immediatethreat in ecoregions where little original habitat remains.

Method We will develop an average deforestation ratte for each ecoregion based on govermment agency studies,FAO documents, articles in the scientific literature (e.g., Sader & Joyce 1988, Sayer & Whitmore 1991, Skole &Tucker 1993, Whitmnore & Sayer 1992), World Bank databases, World Conservation Monitoring Centredatabases (WCMC 1993), EPA studies, NGO literature, and consultation with regional experts. This analysis willonly consider deforestation rates estimated from data obtained within the last decade.

Rankngs:

Ecoregions with less than 3,000 km2 of original habitat remaining

Deforestation > 4% per annum 10

Deforestation from 3.1% to 4% per annum 9

Deforestation from 2.1% to 3% per annum 8

21

Deforestation from 1% to 2% per annum 6

Deforestation < 1% per annum 1

No measurable amount of deforestation in ecoregion 0

Ecoregions with more than 3,000 km2 of original habitat remaining

Deforestation > 4% per annum 8

Deforestation from 3.1% to 4% per annum 7

Deforestation from 2. 1% to 3% per annum 6

Deforestation from 1% to 2% per annum 3

Deforestation <1% per annum 1

No measurable amount of deforestation in ecoregion 0

Criterion 4: Wildlife Exploitation

Rationale: Iigh intensity or uncontrolled hunting and other forms of wildlife exploitation (or harvesting) canquickly extirpate populations of a species or reduce their numbers tu such an extent that the populations will notrecover. Typically, larger vertebrate species are most threatened by hunting because of their naturally lowpopulion densities, low reproductive rates, and attractiveness to hunters (Redford 1992, Robinson & Ramirez1982, Terborgh 1992). Although a forest ecosystem may appear intact from a structural standpoint, hunting canrapidly eliminate the larger species (Redford 1992). The long-temr effects of losing large vertebrate species areunclear, but some evidence shows that it leads to changes in species compositions of forests (both plant andanimals) and impoverishment of communities (Dirzo & Miranda 1990, Terborgh 1992). Many of these largerspecies serve as keystone species in trpical forest ecosystems (Dirzo & Miranda 1990, Terborgh 1992). Huntingless directly affects smaller species of vertebrates (but see Dourojeanni 1985), invertebrates, and plants, with theexception of some birds (Infigo-Elias & Ramos 1991), mammals, reptiles, butterflies (Collins & Morris 1985),and plants with economic value. Although large blocks of intact forest generally have low hunting pressure neartheir interiors, this is not always the case, as rivers and trails provide access to hunters.

Metho: Consultation with regional experts and review of appropriate literatr

Rankings:

High intensity or uncontrolled hunting or wildlife exploitation in region 6(elimination of local populations of most hunted species imminent or complete)

Moderate levels of hunting/exploitation, populations of game/trade species persisting 3(reduced populations of the majority of hunted species)

Very low to no hunting/exploitation of common target species in region 0(Fauna apparently intact with natural population densities)

Criterion 5: Development Schemes

Rationale: Development projects, particularly those that introduce roads into previously remote regions, catalyzeloss, fragmentation, and degradation of intact ecosystems (see Fearnside 1987, Goldsmith & Hilyard 1984,Myers 1993, Oren 1987). Implementation of these projects may constrain establishment of new conservationareas, maintenance of existing units, or development of sustainable resource use in the region The presence of

22

development schemes in a region also indicates that strong economic, political, and/or social forces are at work topress for further exploitation of the area. However, in some cases they can be a catalyst for conserving speciesand original habitaL

Mthod: Govemment records, multilateral/bilateral organization records, NGO records.

Development schemes planned or underway in the region that may 6seriously alter 25 % or more of the remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 3seriously alter between 10% and 24% of the remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 1seriously alter between 5% and 9% of the remaining orginalhabitat within two decades

No development schemes identified for region 0

Criterion 6: Population Density in Region

Rationale: Ecoregions with high human population densities will experience, on average, a greater demand forresources in both protected and non-protected habitats. Rates of habitat alteration can be broadly correlated tohuman densities, but there are many exceptions and spatial and temporal scale considerations greatly influence therelationship (Hecht 1993, Mahar 1989, Myers 1993). Overall, in regions of low population density, effectiveconservation measures may be easier to implement and threats to original habitat should be less intense.

MeFob: Population density estimates will be based upon published records. The average population density forthe entire ecoregion will be calculat& Urban and non-urban areas will be includod since we assume that largeurban areas contribute to an efflux of people into undeveloped regions and reflect demand for some naturalresources taken from original habitaL We will use population density rather than populaton growth rate figureswhich are often based on birth/death rates because the population density may more accurately reflect recentregional or tans-national migrations. A large influx of economically or politically displaced people intoundeveloped areas is a major factor contributing to the degradadon of ecosystems.

Average

Average population density 2100 individuals/ km2 3

Average population density between 26 and 99 individuals/ km2 2

Average population density between 2 and 25 individuals/kma2 1

Average population density <1 individua/ km2 0

23

2. CONIFER & TEMPERATE BROADLEAF FORESTS

A. Conservation Potential

Criterion 1: Large Blocks of Original Habitat

Rationale: Many of the sarne considerations described for the presence of large contiguous blocks of habitat intropical broadleaf forests also apply for conifer forests. Conifer forests typically have lower beta diversity andhigher species densities than do tropical broadleaf forests. Mutualistic intractions for pollination and seeddispersal are also less important in these systems, thereby slightly lowering their sensitivity to fragmentation andhabitat loss compared to tropical forests. Therefore, functioning ecosystems may be maintained in smaller areasthan those required for tropical broadleaf forests, a situation reflected in the ranking categories.

Methd: Satellite imagery and maps will be consulted to identify remaining large blocks of habitat within eachecoregion.

Ecoregions with original extent of habitat 2 3,000 km2

Large blocks of original habitat remain within ecoregion:either one or more blocks > 2,000 km2. or 3 or more blocks > 800 km2 36

Two blocks of habitat > 800 km2 but < 2,000 km2 30

One habitat block > 800 km2 but < 2,000 km2 25

3 or more blocks > 500 km2 but < 800 km2 20

One or 2 blocks > 500 km2 but < 800 krn2 10

3 or more blocks > 250 kmn2 but < 500 km2 5

No contiguous blocks > 250 km2 remain 0

Ecoregions with original extent of habitat < 3,000 km2, but 2 1,000 km2

Large blocks of original habitat remain within ecoregion:either one or more blocks > 800 kin2, or 3 or more blocks > 500 km2 36

Two blocks of habitat > 500 km2 but < 800 km2 30

One block of habitat > 500 km2 but < 800 km2 20

Three or more blocks of habitat 2 250 km2 but < 500 km2 10

One or 2 blocks > 250 km2 but < 500km2 5

No habitat blocks > 250 km2 remain 0

24

Ecoregions with original extent of habitat < 1,000 km2 but > 100 km2

Large blocks of contiguous, unfragrnented, habitat remain within ecoregion:either one or more blocks > 500 km2 36

Three or mome large blocks of habitat > 250 km2 but < 500 km2 30

One or 2 blocks > 250 km2 but < 500 km2 20

Three or more blocks > 100 km2 but < 250 km2 10

One or 2 blocks > 100 km2 but-< 250 km2 5

No habitat blocks > 100 kmn2 remain 0

Ecoregions with original extent of habitat c 100 km2

More than 80% of original habitat remains intact 36

Between 61% and 80% of orgingl habitat remains intact 30

Between 41% and 60% of original habitat remains intact 20

Betuween 10%7 and 40% of original habitat remains intact 10

Less than 10% or original habitat remains intact 5

Less than 1% of original habitat remains 0

Criterion 2: Presence of Intact Watersheds

Rationale: As with tropical broadleaf forests, watersheds are important conservation units for conifer andtemperate broadleaf forests. Because of the reduced area requirements for maintaining viable ecosystems inconiferous forests versus tropical forests (see criterion 1 rationale), a single watershed may contain sufficientforested area to maintain a viable coniferous or temperate forest ecosystem. In addition, access issues may becritical for preservation of these forests since illegal logging and buming are both major threats (Parsons 1976).The physical structure of watersheds can sometimes serve to control access to a protected area.

Metho: Coverages of remaining original habitat will be overlayed with watershed coverages within eachecoregion. Different ecoregions will be assessed using appropriate tributary levels.

Rankings:

Three or more intact watersheds 9

One or 2 intact watersheds 6

No intact watersheds 0

Criterion 3: Protected Area Network

Rationale: Many of the same considerations discussed for tropical bmadleaf forsts are appropriate here.

Meth: Protected area databases and maps from a variety of sources will identify reserves.

25

Rankngs

Ecoregions with original extent of habitat 2 3,000 kM2

ThTee or more protected areas: 3 or more 2 500 km2 12

Three or more protected areas: with 2 > 500 km2 10

Three or more protected areas: no reserves > 500 km2 7but has at least 3 reserves > 250 km2

Three or more protected areas with at least 2 reserves Ž 250 km2 4

Protected area system has no reserves > 250 km2 2

No protected areas exist system 0

Ecoregions with original extent of habitat 2 1,000 kin2 , but < 3,000 km2

Three or more protected uas: at least 2> 500 km2 12

Three or more protected areas: with 1 > 500 km2 10and 2 or more > 250 km2

Three or more protected areas: no. reserves > 500 km2 7but has at least 2 reserves > 250 an2

Three or more protected areas: at least 2 reserves > 250 km2 4

Protected area system has no reserves > 250 km2 2

No protected area system 0

Ecoregions with original extent of habitat > 250 km2, but < 1,000 km2

Three or more protected areas: 3 or more > 250 km2 12

Three or more protected areas: with 1 or 2 Ž 250 km2 10

Three or more protected areas: no reserves Ž 250 km2

but has at least 3 reserves > 100 km2 7

Protected area system has only 1 or 2 reserves Ž 100k mn2 4

Protected area system has no reserves > 100 km2 2

No protected area system 0

Ecoregions with original extent of habitat < 250 km2

Over 80% of original habitat under protected status 12

Between 60% and 80% of original habitat incorporated into a protected area system 10

26

Between 40% and 59% of original habitat incorporated into a protected area system 7

Between 21% and 39% of originaI habitat inorporated into a protected area system 4

Less dtn 20% of onginal habitat incorporated into a protected area system 2

No original habitat incorporated into a protected area system 0

Criterion 4: Management of Conservation Areas

Ratonale: The considons oued for tpical broadleaf forests are generlly applicable here.

Mced: Regional experts will be consulted and appropriate literature reviewed.

At least 50% of conservation areas reasonably well managed 3

Less than 25% reasonably well-managed 2

Virtually none of the existng conservation areas well-managed 0

B. Threats

The major thmats to conifer and temperate broadleaf forests are habitat loss and fragmnentation due to logging,bumning, and conversion to agriculture; and high levels of wildlife exploitation thrughout the region (Faijon et al.1993, Perry 1991). Many forests have already been extensively logged and degraded.

Criterion 1: Loss of Original Habitat

Rationale: The considerations outlined for tropical broadleaf forests are generally applicable here.

M i: Estimates of the remaining habitat area will be comnpared to estimates of the original habitat cover. Mapsand satellite imagery will be used to estimate current cover, while original cover estimates are available in theliterature.

>90% of original habitat lost 20

50% to 89% lost 16

25% to 49% lost 10

5% to 24% lost 5

0% to 4% lost 0

Criterion 2: Degree of Habitat Fragmentation

Rationale: The considerations outlined for tropical broadleaf foests are generally applicable here.

Method: The method outlined for tropical broadleaf forests is also applied here.

27

LanIangi

Low connectivity:.Index values between 0 and 0.33 15

Intermediate connectivity: Index values between 0.33 and 0.66 9

Higher connectivity: Index values between 0.66 and 0.9 4

Well-connected landscape: Index values between 0.9 and 1 0

Criterion 3: Rate of Deforestation

Rationale: The considerations outlined for tropical broadleaf forests are generally applicable here.

Metho: The method outlined for tropical broadleaf forests is also applied here.

Ecoregions with less than 3,000 km2 of original habitat remaining

Deforestation > 4% per annum 10

Deforestation from 3.1% to 4% per annum 9

Deforestation from 2.1% to 3% per annum 8

Deforestation from 1% to 2% per annum 6

Deforestation < 1% per annum 1

No measurable amount of deforestation in ecoregion 0

Ecoregions with more than 3,000 km2 of original habitat remaining

Deforestation > 4% per annum 8

Deforestation from 3.1% to 4% per annum 7

Defoestation from 2.1 % to 3% per annum 6

Deforestation from 1% to 2% per annum 3

Deforestation < 1% per annum 1

No measurable amount of deforestation in ecoregion 0

Criterion 4: Wildlife Exploitation

Rationale: Wildlife exploitation can be intense in conifer and temperate broadleaf forests because of the proximityof human populations and the access provided by roads and trails in or near these ecoregions.

Metho The method outlined for tropical broadleaf forests is also applied here.

28

,Rankings:-

High intensity of wildlife exploitation in region (elimination of local populations 6of most target species imminent or complete)

Moderate levels of wildlife exploitation, populations of game/trade species persisting 3(reduced populations of the majority of hunted species)

Very low to no wildlife exploitation of common target species in region 0(Fauna apparently intact with natural population densities)

Criterion 5: Development Schemes

EadQa&: The considerations outlined for tropical broadleaf forests are generally applicable here. The number ofdevelopment schemes in the first two ranking categories has been reduced because conifer and temperate broadleafforests have a smaller geographic extent than tropical broadleaf forests.

Metho: The method outlined for tropical broadleaf forests is also applied here.

Development schemes planned or underway in the region that may 6seriously alter 25% or more of the remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 3seriously altcr between 10% and 24% of the remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 1seriously alter between 5% and 9% of the remaining originalhabitat within two decades

No development schemes identified for region 0

Criterion 6: Population Density in Region

Rationale: The considerations outlined for tropical broadleaf forests ar generally applicable here.

Method The method outlined for tropical broadleaf forests is also applied here.

Average population density > 100 individuals/ man2 3

Average population density between 26 and 99 individuaLs/ km2 2

Average population density between 2 and 25 individuals/ km2 1

Average population density <1 individual km2 0

29

3. GRASSLANDS, SAVANNAHS, WETLANDS, SHRUBLANDS

A. Conservation Potential

Criterion 1: Large Blocks of Original Habitat

Rainale: Large blocks of grassland, savannah, wedand, and shrubland (GSWS) habitat should have sufficientarea to maintain hydrographic processes, seasonal movements of animals, and be of adequate size to absorbpredictable disturbance events such as fire. Original blocks of habitat in GSWS communities can be more difficultto identify than for forests because ecosystem aleration is often not as visually dramatic as deforestaiion and mayrequire considerable ecological expertise for interpretation. Therefore, we broadly define onginal habitat as thatwhich maintains species compositions, species abundances, and disturbance regimes close to those of nativeecosystems. The allowable level of variation in these parameters will be tailored t the ecological dynamiics andlinkages as assessed by regional experts.

Method: Regional experts, satellite imagery, and maps will be consulted to identify remaining large blocks ofhabitat within each ecoregion.

Ecoregion with original extent of habitat 2 1,000 km2

Large blocks of original habitat remain within ecoregion:either one or more blocks > 900 km2.or at least 2 blocks a 750 km2 or 3 or more blocks > 500 km2 45

At least I block > 750 km2 and 1 > 500 km2 35

One or 2 blocks of habitat > 500 km2 currently intact 25

No blocks of habitat > 500 km2 remain, but at least 3 > 250 km2 20

No blocks of habitat > 500 km2 remain, but 1 to 2 blocks > 250 km2 15

No habitat blocks > 250 km2 remain, but 3 or more > 100 km2 10

No habitat blocks > 100 km2 remain 0

Ecoregion with original extent of habitat < 1,000 km2, but > 100 km2

Large blocks of original habitat remain within ecoregion: at least I block 2Ž750 km2.or at least 1 block 2 500 km2 and lblock Ž 250 km2 45

Three blocks of habitat > 250 km2 but < 500 km2 35

One or 2 blocks > 250 km2 but < 500 km2 25

No habitat blocks > 250km2 remain, but at least 3> 100 km2 15

No habitat blocks > 250 km2 remain, but 1 to 2 blocks > 100 kn2 10

No habitat blocks> 100 km2 rmain 0

30

Ecoregions with original extent of habitat S 100 km2

More than 80% of onginal habitat remains intact 45

Between 61% and 80% of original habitat rmains intact 38

Between 41% and 60% of original habiat remains in 28

Between 20% and 40% of original habitat remains intact 20

Between 10% and 19% of original habitat remains intact 10

Less than 10% or original habitat is curTently intact 5

Less than 1% of original habitat remains 0

Criterion 2: Protected Area Network

Rationale: The considerations for conifer-dominated and temperate forests are generally applicable here.

Method Protected area databases and maps will identify reserves.

£anEnes:

Ecoregions with original extent of habitat 2 3,000 km2

Thue or mom protected areas: 3 or more 2 200 kin2 12

Three or more protected areas: with 2 > 200 km2 10

Three or more protected areas: no reserves > 200 km2 8but has at least 3 reserves > 75 km2

Three or more protected areas: with at least 2 reserves 2 75 km2 6

Protected areas exist, but no reserves > 75 knm2 4

No protected areas exist 0

Ecoregions with oniginal extent of habitat 2. 1,000 In2, but < 3,000 km2

Three or more protected areas: at least 3 > 200 km2 12

Three or more protected areas: with 2 > 200 lan2 10or 1 > 200 kn 2 and 2 or more 2 75 hn2

Three or more protected areas: no reserves > 200 km2 8but has at least 2 reserves 2Ž75 km2

Three or more protected areas: at least 1 reserve > 75 km2 6

Protected area system has fewer than 3 reserves with no reserves > 75 km2 4

No protected areas exist 0

31

Ecoregions with original extent of habitat > 250 km2 , but < 1,000 km2

Three or more protected areas: at least 3 > 150 km2 12

hree or more protected areas: with 2 > 150 km2 10or 1 > 150 km2 and 2 or more > 75 km2

Three or more protected areas: no reserves > 150 km2 8but has at least 2 reserves > 75 mn2

Tlree or more protected areas: no reserves > 150 km2 but at least I reserve > 75 kan2 6

Protected areas exist but no reserves 2Ž75 km2 4

No proected areas exist 0

Ecoregions with original extent of habitat < 250 km2

Over 80% of original habitat under protected statos 12

Between 60% and 80%o of original habitat incorporated into a protected area system 10

Between 40% and 59% of original habitat incorporated into a protected area system 8

Between 21% and 39% of oiiginal habitat incorporated into a protected area system 6

Less than 20% of original habitat incorporated into a protectd area system 4

No original habitat incorporated into a protected area system 0

Criterion 3: Management of Conservation Areas

Ratonale: The considerations oudtjed for conifer-dominated and temperate forests are generally applicable here.

Mflhxl: Regional experts will be consulted and appropriate literature reviewed.

lRan~1L

At least 509o of conservation areas reasonably well managed 3

Less than 25% reasonably well managd 2

Virtually none of the existing conservatdon areas well managed 0or no conservation areas exist

B. Threats

Criterion 1: Loss of Original Habitat

Ratale: In forested habitats, the loss of habitat is largely defined as the removal of me cover. However, inGSWSs, habitat loss is more difficult to define and can be much more subtle. Certainly, plowing, conversion toagriculture, and draining of wetlands constitute habitat loss, but the effects of overgazing and intensive burningcan vary. Overgrazed and buned areas may stll have some native plant species present, but they are oftenseverely reduced in number (or percent cover) and replaced by exotic species that are better conpetitors indisturbed environments (Budowski 1956). Ihus, original habitat is defined here as GSWS communities that ill

32

maintain the vast majority of their original flora in abundance, characteristic of relatively undisturbed conditions,with most of their large vertebrate species.

Estimates of the remaining habitat area will be compared to estimates of the original habitat cover. Mapsand satellite imagery will be used to estimate currcnt cover, while original cover estimates are available in theliterature.

>90% of original habitat lost 18

50% to 89% lost 14

25% to 49% lost 9

5% to 24% lost 5

0% to 4% lost 0

Criterion 2: Degree of Habitat Fragmentation

Ratonale: Many grassland, savannah, wetland, and shrubland habitats are naturally patchy and form a mosaic ofgrassy habitats, shrubby vegetation, open forests, waterways, lakes, and riverine forests (Bourliere 1989,Cabrera 1968, Eiten 1982, Numata 1979). We define fragmentation here as the separation of blocks of the natralhab¶tat mosaic by incompatible land use such as intensive agriculture; fenced pasture wivt high densities oflivestockc; large urban and nrul communities; roads and canals that prohibit the movement of wildlife and waterbetween blocks of orginal habitat; and heavily degraded areas caused by overgazing and frequent burning.

Method: The method outlined for tropical broadleaf forests is also applied here.

Low connectivity: Index values between 0 and 0.33 11

Internediate connectivity: Index values between 0.33 and 0.66 6

Higher connectivity: Index values between 0.66 and 0.9 2

Well-connected landscape: Index values between 0.9 and 1 0

Criterion 3: Grazing and Burning Pressure

Batinale: Grzing and buning (e.g., burning for new forage growth, reduction of undesirable foragc species)pose major dthrats to GSWS habitats (WCMC 1992, Sala 1981, Sims 1988). Native species that are not welladapted to fire and grazing can be eliminated and more resistant species can be displaced by aggressive exoticspecies. The loss of soil and nutrients through erosion and buning can significantly reduce available nutrients andconsequently affect productivity and biological communiis. It is difficult to quantify the effects of grazing andburning because they can be patchy and some native species do persist at most stages.

Mkch: The method outlined for tropical broadleaf forests is also applied here.

High intensity of grazing and buming in areas not condered to be original habitat(i.e., areas containing degrded habitats): few native plant species persist, large nativeherbivores eliminated 11

33

Moderate levels of grazing and burning in areas not considered to be original habitatpopulations of native plant species and large herbivores persist in reduced numbers 6

Very low to no grazing and buming in areas not considered to be original habitat 0

Criterion 4: Rate of Habitat Conversion

Rtijnale: Conversion is defined here as GSWS habitat that has been (1) plowed; (2).used for urban setdements,industrial complexes, agriculture, or replanted for pasture; (3) grazed or burned to such an extent that native plantcommunities are unlikely to ttmrn without restoraton (4) drained in the case of wetands; and (5) cleared ofwoody vegetation in the case of savannahs.

Method: The method outlined for conifer and temperate broadleaf forests is also applied here.

Ecoregions with less than 1000 1an2 of original habitat remaining

Habitat conversion > 4% per annum 8

Habitat conversion from 3.1% to 4% per annum 7

Habitat conversion from 2.1% to 3% per annum 6

Habitat conversion from 1% to 2% per ainum 3

Habitat conversion < 1% per annun 1

No measurable amount of habitat conversion in ecoregion 0

Ecoregions with more than 1000 km2 of original habitat remaining

Habitat conversion > 4% per annum 6

Habitat conversion from 3.1 % to 4% per annum 5

Habitat conversion from 2.1% to 3% per annum 4

Habitat conversion from 1% to 2% per annum 3

Habitat conversion < 1% per annum 1

No measurable amount of habitat conversion in ecoregion 0

Criterion 5: Development Schemes

Rationale: The considerations outlined for conifer and temperate forests are generally applicable here.

Method: The method outlined for tropical broadleaf forests is also applied here.

Ranings:

Development schemes planned or underway in the rgion that may 5seriously alter 25% or more of the remaining originalhabitat within two decades

34

Development schemes planned or underway in the region that may 4seriously alter between 10% and 24% of the remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 2senously alter between 5% and 9% of the remaining originalhabitat within two decades

No development schemes identified for region 0

Criterion 6: Wildlife Exploitation

Rarionai: Wildlife exploitation can qWcidy extirpate or thraten populations of lage herbivores, insectivores, andpredators. When larger herbivors are hunted out they are commonly replaced with domesticated largeherbivores. Domesticated animals often damage the original flora because of high stocking rates.

M:bgd Consultation with regional experts or appropriate literature.

Ran}ia;L

High intensity of wildlife exploitation in ecoregion (eimnation of most local populationsof most hunted species imminent or complete) 4

Mdderatc levels of wildlife exploitation, populations of game/tade species persisting(reduced populations of the majority of hunted species) 2

Very low to no hunting or exploitation of common target species in region(Fauna apparently intact with natuTal population densities) 0

Criterion 7: Population Density in Region

Bationa1e: The consideradons outlined for tropical broadleaf forsts are generally applicable here.

MCIh: The method outlined for tropical brdeaf frsts is also applied here.

Population density 2 100 individuals/km2 3

Population density between 26 and 99 individs/ m2 2

Populaton density between 2 and 25 individuals/ kn 2 1

Population density sl individual/ km2 0

35

4. XERIC ECOSYSTEMS

A. Conservation Potential

Criterion 1: Large Blocks of Original Habitat

Rale:M The same consideratons for GSWS ecosystems apply here except that the geographic dispersion ofhabitat blocks may be more impornt than the size for biodiversity conseaon becaus there can be consideablebeta diversity in xeric regions. Deser dune, and chaparrd often contain species that are either local endemics orhave patchy distributions becas they m ha specialisU (Hanes 1977, Mooney 1977).

Method: Regional expes, satellite imagay, and maps will be consulted to identify taining large blocks ofhabitat within each ecoregion.

Ecoregion with original extent of habitat 2 W1,00 km2

Larg blocks of original haitat remain within ecoreeither one or more blocks a 500 kn2,or at least 2 blocks - 300 km2 or 3 or more blocks 2 200 km2 45

At least I block > 300 km2 and 1 2 200 km2 35

One or 2 blocks of habitat > 200 km2 currently intact 25

No blocks of habitat > 200 km2 rmain, but at least 3 75 kmn2 20

No blocks of habitat > 200 km2 remain, but 1 to 2 blocks Ž 75 km2 15

No habitat bWcks 200 km2 remain, but 3 or more a 75 hm2 10

No habitat blocks > 75 km2 remain 0

Ecoregion with original extent of habitat < 1,000 kIm2, but > 100 km2

Large blocks of original habitat rman vthin ecoregion. at least 1 block 2 500k m2 ,or at least I block _ 300 km2 and lblock 2 200 kn 2 45

Three blocks of habitat > 200 kIm2 but < 300 km2 35

One or 2 blocks a 200km2 but < 300k1m2 25

No habitat blocks > 200 km2 remain, but at least 3 2 75 k 2 15

No habitat blocks > 200 km2 remain, but 1 to 2 blocks > 75 km2 10

No habitat blocks a 75 krn2 remain 0

Ecoregions with original extent of habtat I 100 km2

More dtan 80% of orginal habitat remains intact 45

36

Between 61% and 809o of onginni habitat remains intact - 38

Between 4 1% and 60% of original habitat remains intact 28

Between 20% and 40% of original habitat mnais intact 20

Between 10% and 19% of original habita n amains intact 10

Less than 10% or original habitat is currently intact 5

Less than 1% of original habitat remains 0

Criterion 2: Protected Area Network

Ratnale The considerations for conifer-dominated and temperate forests are applicable her except that desert,dune, and chaparal systems are often not as resilient to human disturbance as are GSWSs. Therefore, humanactviies must be stnctly limited in xenc protected areas.

Mch Protected area datbss and maps will identify rsres.

Ecoregions with original extent of habitat 2 3,000 kn2

Three or mor protected aeas: 3 or more > 200 km2 12

Three or more protected areas: with 2 > 200 kmn2 10

Three or more protected areas: no reserves > 200 man2 8but has at least 3 reserves Ž 75 km2

Three or more protected areas: with at least 2 reserves 2 75 kam2 6

Protected areas exist, but no reserves 2 75 knm2 4

No protected areas exist 0

Ecoregions with origind extent of habita 1,000 kin, but < 3,000 km2

Three or more protected areas: at least 3 2 200 km2 12

Three or mor procted awas: with 2 a 200 kn 2 10or 1> 200 km2 and 2 or moe > 75 km2

Tlree or more protected areas: no reserves > 200 kam2 8but has at least 2 reserves > 75 km2

Tbree or more protected areas: at least 1 reserve > 75 km2 6

Protected area system has fewer than 3 reserves with no reserves Ž75 kcm2 4

No protected areas exist 0

37

Ecoregions with original extent of habitat > 250 kim2 , but < 1,000 km2

Three or more protected areas: at least 3 > 150 km2 12

Three or more protected areas: with 2 > 150 hn2 10or I > l50 kn2 and 2 or more > 75 km2

Three or more protected areas: no reserves > 150 hn2 8but has at least 2 reserves 2 75 km2

Three or more protectd areas: no rserves > 150 km2 but at least 1 reswe > 75 km2 6

Protected areas exist but no reserves > 75 km2 4

No protected areas exist 0

Ecoregions with original extent of habitat < 250 km2

Over 80% of original habitat under protected status 12

Between 60% and 80% of original habitat incorporated into a protected area system 10

Between 40o and 59% of original habitat incorporated into a protected area system 8

Between 21% and 39% of original habitat incorporated into a protected area system 6

Less than 20% of original habitat incorporated into a protected area system 4

No original habitat incorporated into a protcted area system 0

Criterion 3: Management of Conservation Areas

Rational: The considerations outlined for GSWS ecoregions are generally applicable here.

Methoxo: Regional experts will be consulted and appropriate iterwe reviewed.

At least 50% of conservation areas reasonably well managed 3

Less than 25% reasonably well managed 2

Virtually none of the existing conservation areas well managed 0or no conservation areas exist

B. Threats

Criterion 1: Loss of Original Habitat

Rationale: The considerations outlined for GSWS ecoregions ae generally applicable hem

Medb~ Estimates of the remaining habitat area will be compamed to estimates of the original habitat cover.Consultations with regional experts, maps, and satellite inagery will be used to estimate cumnt cover, whileoriginal cover estimates are available in the litetue.

38

>90% of originaI habitat lost 18

50% to 89% lost 14

25% to 49% lost 8

5% to 24% lost 4

0% to 4% lost 0

Criterion 2: Degree of Habitat Fragmentation

BRnaa: The considerations outlined for GSWS ecoregions are also applicable here. Mher is much evidene tathabitat fragmentation in desert, dune, and chaparral communities can cause local extrpauons of a variety ofvertebrate species (Cody 1986, Soule et al. 1988). Many species natve to these ecoregions ar stict habitatspecialists and will be unable or unwilling to cross even narrow bariers (Soul6 et a. 1988).

M :bad The method outlined for tropical bradleaf forests is also applied here.

Low connectvity: Index values between 0 and 0.33 11

Itermcdiate connectivity: Index values between 0.33 and 0.66 6

EigBher connectivity: Index values between 0.66 and 0.9 2

Well-connected landscape: Index values between 0.9 and 1 0

Criterion 3: Grazing and Burning Pressure

R&dainC: The consideraions outlined for GSWS ecoregions are generally applicable here.

MetW The method out1ined for trpical bradleaf forsts is also applied here.

High intensity of grazing and burning in areas not considered to be original habitat(i.e., areas containing degraded habitats): few native plant species persist, large nativeherbivors cminatd 11

Moderate levels of grazing and buming in areas not considered to be original habitacpopulations of native plant species and large herbivores persist in reduced numbers 6

Very low to w grazing and buring in areas not considered to be caiginal habitat 0

Criterion 4: Rate of Habitat Conversion

Badagl: Conversion is defined here as xeric habitat that has been (1) plowed; (2) used for urban setiments,industrial complexes, agricuture, orreplanted for pase; (3) grazed or burned tO such an extent that naivevegeution will not adily retum; and (4) cleared of woody vegetaton, including cactus.

Mczhi: The method outlined for conifer and temperate broadleaf forests is also applied here.

39

Ecoregions with less than 1000 km2 of original habtat remaining

Habitat conversion > 4% per annum 8

Habitat conversion from 3.1% to 4% per annum 7

Habitat conversion from 2.1% to 3% per annum 6

Habitat conversion fron 1% to 2% per annum 3

Habitat conversion < 1% per annum 1

No measurable amount of habitat conversion in ecoregion 0

Ecoregions with more than 1000 km2 of original habitat remaining

Habitat converon > 4% per annum 6

Habitat conversion from 3.1% to 4% per annum 5

Habitat conversion fom 2.1% to 3% per annum 4

Habitat conversion from 1% to 2% per annum 3

Habitat conversion < 1% per annum 1

No measurable amount of habitat conversion in ecoregion 0

Criterion 5: Development Schemes

Ratonale: The considerations outlined for GSWS ecoregions are generally applicable here.

Metho The considerations outlined for GSWS ecoregions are also applicable here.

Development schemes planned or underway in the region that may 5seriously alter 25% or morm of the remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 4seriously alter between 10% and 24% of the remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 2seriously alter between 5% and 9% of the remaining orginalhabitat within two decades

No development schemes planned in region 0

40

Criterion 6: Wildlife Exploitation

Rtna: Wildlife exploitaion can be intense in xeric regons because of the close proximity of hwnanpopulations, easy access, and the lack of cover for hunted species.

g The method oudined for tropcal broadleaf forests is also applied bert.

RAWd=

High intensity of wildlife exploitation in region (elimination of local populadons 4of most target species imminent or complete)

Moderate levels of wildife exploitation, populations of game/tade species persisting 2(reduced populations of the majority of hunted species)

Very low to no wildlife exploitation of common target species in region 0(Fauna apparendy intact with natural population densities)

Criterion 7: Population Density in Region

ldglnnak: The considertions outlined for GSWS ecoregions are generally applicable her.

Mrj:d: The considerations outlined for GSWS ecoregions are also applicable bere.

Population density > 100 individuals kn 2 3

Pnpulation density between 26 and 99 individuals/km2 2

Population density between 2 and 25 individuals/ kan2 1

Population density <1 individual/km2 0

S. MANGROVE FORESTS

Mangrove forests are patchily distributed in a linear fashion along coasdines (Chapman 1975, Hartshorn 1988).Large regional complexes of mangrove fores are used tO delineate mangrove ecoregions. Although themangrove blocks within ecoregion complexes are assumed to have frquent exchanges of propagules, the paucityof data on propagule dispersal means that designations we based more upon geographic proximity and expediencyfor conservation assessments.

A. Conservation Potential

Criterion 1: Large Blocks of Habitat

BRAfnak: Large blocks of mangrove forests present the best situation for maintaining viable populations ofmangrove and mangrove-associated species. In most cases, large blocks are areas where the hydrographiccondidions requied by mangroves will persist for long periods of dme (de la Cruz 1984).

Methd Satellite inagery and maps will be used to identify remaining large blocks of habitt within eachecoregion.

41

Between 80% and 100% of each disdnct mangrove unit within complex 45has > 75% of original mangrove intact

Between 60% and 79% of each distinct mangrove unit within complex 35has > 75% of original mangrove intact

Between 40% and 59% of each distinct mangrove unit within complex 27has > 75% of original mangrove intact

Between 20% and 39% of each distinct mangrve unit within complex. 18has > 75% of original mangrove intact

Between 10% and 19% of each disin mangrove unit wthin complex 9has > 75% of ariginal mangrove intact

10% or less of each distinct mangrove unit within complex 0has > 75% of arignal mangrove intact

Criterion 2: Protected Area Network

RaMnalc The considerations for conifer and temperate forests are generally applicable here. Some humanactivities such as regulated fishing and mollusk coUlecdon may be compatible with conservation in mangroveprotected areas.

Mczta: Protected area databases and maps will identify reserves.

Raninas

Three or more protected areas: 3 or more > 50 im2 12

Three or more protected areas: 2 reserves > 50 Im2 10

Thme or more protected areas: 1 50 km2 9

Three or more protected areas: no reserves 2 50 kn 2 ,but at least 3 reserves > 10 km2 7

Only I to 2 resves > 10 k 2 4

Protected area system but no reseres Ž 10 km2 2

No protected areas exist 0

Criterion 3: Management of Conservation Areas

Rtnale: The considerations outlined for GSWS ecoregions are generally applicable here.

Mthod: Regional experts will be consulted and approprie literature reviewed

R"ns .

At least 50% of conservation areas rteasonably well managed 3

42

Less than 25% reasonably well managed 1

Virtually none of the existng consenruaon aeas wel managed 0or no conservaton aes exist

B. Threats

Criterion 1: Loss of Habitat

RAnf NAl L: Mangrove fsts are dynamic due to the Continiing creation and deswtucion of potential substratby the river and ocean systms (Lugo & Snedaker 1974, Pbol tal. 1977). Habitat loss in mangoves is dcfinedhere as the total removal of mangrove vegetaton or alteration of dte subste such that remaining tees will notsurvive (e.g., changing water flow, poisoning).

Mcdbd: Estimates of the remaning habitat area will be compared to esimates of the orial habitat cover. Mapsand satellite imagery will be used to estimat cumrnnt cover, while oiginal cover estimates are available in theliterature.

>90% of original habitat lost 35

50% to 89% lost 28

25% to49% lost 18

5% to 24% lost 8

0% to 4% lost 0

Criterion 2: Rate of Habitat Conversion

BAtn&: Habitat conversion includes mangrve deforestation due to wood and charcoal prduction, saltproduction, development of shrimp fanns, canaizadon that alters tidal pattems, and killing of tr frompollutants, defoliants, or burning. The development of shrimp fams, salt industies, and harvesting of mangrovetimber repesent the major threats to mangrove forests in the region (Pons & Fiselier 1991, Walsh 1977).

Metho: The method outlined for conifer and tempert broadleaf fes is also applied here.

RBakn=

Habitat conversion > 5% per annum 15

Habitat conversion from 4.1% to 5% per annum 13

Habitat conversion frm 3.1% to 4% per annwn 12

Habitat conversion from 2% to 3% per annum 6

Habitat conversion < 1 % per annum 2

No measuable amount of habitat conversion in ecoregion 0

43

Criterion 3: Development Schemes

Rationale: Many mangrove complexes are curendy being, or are slaed to be, convered into shrimp farms or saltpans.

Medtd The consideratons oudined for GSWS ecorgions are also applicable here.

Development schemes planned or underway in the region that may 10seriously alter 25% or more of th remaining orignhabitat within two decades

Development schemes planned or underway in the region dt may 7seriously alter between 10% and 24% of fte remaining originalhabitat within two decades

Development schemes planned or underway in the region that may 3seriously aler between 5% and 9% of the remaing ordginalhabitat within two decades

No development schemes identified that will convert anginal habitat 0

44

VIIm LITERATURE C1TED

The literature cited here primarily reflects backpround literature on generalecological theory that the first author was most familiar with when initiallydeveloping this proposaL The project executants are actively acquiring andreviewing numerous studies , books, and documents produced by scientists, resourcemanagers, and conservationists from the LAC region. Much of the detailedinformation found in these works will be referred to in the final ecoregion analysis.It is our intention to ensure that the information contained in this study is primarilybaed on the extensive work of experts from the region.

Allan, J.D., & Flecker, A.S. 1993. Biodiversity conservation in running waters.BioScience 43:32-43.

Allendorf, F.W. 1988. Conservation biology of fishes. Conservation Biology 2:145-147.Arita, H.T., Robinson, J.G., & Redford, KM. 1990. Rariry in neotropical forest

mammals and its ecological correlates. Conservation Biology 4:181-192.Austin, M.P. & Margules, C.R. 1986. Assessing representativeness. Pages 45-67 in

Usher, MB. (ed.). Wildlife conservation evaluaton. Chapman & HalL London,England-

Aylward, B. & Barbier, E.B. 1992. Valuing environmental functions in developingcountries. Biology and Conservaton 1:34-50.

Berger, J. 1990. Persistence of different-sized populations: an empirical assessmentof rapid extinctions in Bighorn Sheep. Conservation Biology 4:91-98.

Bierregaard, R.O. 1986. Changes in bird communities in virgin forest and isolatedAmazonian forest fragments. Ibis 128:166-167.

Bourliere, F. 1989. Tropical savannas. In Lieth, H.& Werger, MJ.A. (eds.).Ecosystems of the World, Volume 13. Elsevier Scientific Publishing Co.,Amsterdam.

Browder, J.0. 1990. Exdative reserves will not save tropics. BioScience 40:626.Brown, DE., Reichenbacher, R., & Franson, S. 1993. A classification system and map of

the biotic communities of North America. USEPA, Las Vegas, Nevada.Unpublished.

Brown, KS., Jr. 1989. Conservaton of neotropical environments: insects as indicators.Pages 350-401 in Collins, N.M., & Thomas, J.A. (eds.). The conservation ofinsects and their habitats. 15th Symposium of the Royal Entomological Societyof London, 14-15 September, 1989. Academic Press, London, England.

Brown, S. & Lugo, AX. 1990. Tropical seconday forests. Journal of Tropical Ecology6:1-32.

Budowski, G. 1956. Tropical savannas, a sequence of forest felling and repeatedburmings. Turrialba 6:23-33.

Buschbacher, RJ. 1990. Natural forest management in the humid tropics: ecological,social, and economic considratons. Ambio 19:253-258.

Cabrera, A. 1968. Ecologfa vegetal de la puna. Pages 91-116 in Troll, C. (ed.).Geoecology of the mountainous regions of the tropical Americas.

Campbell, D.G. & Hammond, HD. (eds.). 1989. Floristic inventory of tropical countries.The New York Botanical Garden, New York, New York, U.S.A. 545 pp.

Campbell, J.A., & Lamar, W.W. 1989. The venomous reptiles of Latin America.Comstock Publishing Associates. Cornell University Press, Ithaca, USA.

Chapman, V.J. 1975. Mangrove biogeography. Pages 3-22 in Walsh, G.E., Snedaker,S.C., & Teas, H.J. (eds.). Proceedings of the International synWosium on thebiology and management of mangroves. University of Florida Institute of Foodand Agricultal Sciences, Gainesville, Florida, U.S.A.

Cfuentes, M. 1993. Evaluaci6n de manejo de areas protegidas. WWF, unpublishedmanuscripL

Cody, TJ. 1986. Diversity, rarity, and conservation in mediterranean-climate regions.Pages 123-152 in Sould, M.E. (ed.). Conservation biology: the science of scarcityand diversity. Sinauer Associates, Sunderland, U.S.A.

Collins, N.M. & Moris, M.G. 1985. Threatened Swallowtail Butterflies of the world:IUCN red data book. IUCN, Gland, Switzerland.

de la Cruz. 1984. A realistic approach to the use and management of mangrove areas inSoutheast Asia. In Teas, H.J. (ed.). Physiology and management of mangroves.Dr. W. Junk Publishers, The Netherlands.

Denslow, J.S. & Padoch, C 1988. People of the tropical rain forest. University ofCalifornia Press, Berkeley, USA.

Dinerstein, E. & Wikramanayake, E.D. 1993. Beyond "hotspots": how to pnioritizeinvestments to conserve biodiversity in the Indo-Pacific Region. ConservationBiology 7:53-65.

Dinerstein, E., Wikramanayake, E.D., & Forney, M. 1993. From reservoirs to remnants:conserving the tropical moist forests of the Indo-Pacific region. In Primack, R. &Lovejoy, T. (eds.). The ecology, management, and conservation of SoutheastAsian rainforests. In press.

Dlizo, R., & Miranda, A. 1990. Contemporary neotropical defaunation and foreststructure, function, and diversity - a sequel to John Terborgh. ConservationBiology 4:444-447.

Dourojeanni, KJ. 1985. Over-exploited and under-used animals in the Amazon region.Pages 419-433 in Prance, G.T. & Lovejoy, T.E. (eds.). Amazonia. PergamonPress, Oxford, England. 442 pp.

Dove, M.R. 1993. A revisionist view of tropical deforestation and development.Environmental Conservation 20:17-24.

Eiten, G. 1982. BraziLan 'savannas'. Pages 2547 in Hundey, H.B. & Walker, BX(eds.). Ecology of tropical savannas. Springer-Verlag, New York, USA.

Elton, C.S. 1975. Conservation and the low population density of invertebrates insideneotropical rainforest. Biological Conservation 7:3-15.

Erwin, TL. 1991. An evoluionary basis for conservation strategies. Science 253:750-752

Farjon, A., Page, C.N., & ScheLevis, N. 1993. A preliminary world list of threatenedconifer taxa. Biodiversity and Conservation 2:304-326.

Fearnside, P.M. 1987. Deforestation and international economic development projects inBrazilian Amazonia. Conservation Biology 1:214-221.

Forsyth, A. &8 Miyata, K. 1984. Tropical nature: life and death in the rain forests ofCentral and South America. Charles Scribner's Sons, New York, USA. 248 pp.

Gentry, A.H. 1986. Endemism in tropical versus temperate plant communities. Pages153-181 in Soulk, M.E. (ed.). Conservation biology: the science of scarcity anddiversity. Sinauer, Sunderland, U.S.A.

Goldsmith, E., & Hildyard, N. 1984. The social and environmental effects of largedtams. Sierra Club Books, San Francisco, U.S.A.

Goodland, RJ.A. 1987. The World Bank's wildlands policy: a major new means offinancing conservation. Conservation Biology 1:210-213.

Graham, A. (ed.). 1973. Vegetation and vegetation history of northern Lain America.Gradwohl, J. & Greenberg, R. 1988. Saving the tropicalforests. Island Press,

Washington, D.C., USA. 215 pp.Groom, M.J. & Schumaker, N. 1993. Evaluating landscape change: patterns of

worldwide deforestation and local fragmentation. Pages 24-44 in Kareiva, P.M.,Kingsolver, J.G., & Huey, RB. (eds.). Biotic interactions and global change.Sinauer Associates, Inc., Publishers. Sunderland, U.S.A.

Hanes, T.L. 1977. California chapafral. Pages 232-260 in Barbour, M.G., & Billings,W.D. (eds.). North American terrestrial vegetation. Cambridge University Press,Cambridge, England. 425 pp.

Hartshorn, G.S. 1989. Application of gap theory to tropical forest management: naturalregeneration on strip clear-cuts in the Peruvian Amazon. Ecology 70:567-569.

Hecht, S.B. 1993. The logic of livestock and deforestaion in Amazonia. Bioscience43:687-695.

Heywood, V.H., & Stuart, S.N. 1992. Species extinctions in tropical forests. Pages91-117 in Whitmore, T.C. & Slayer, J.A., (eds.). Tropical deforestation andspecies extinction. Chapman & Hall, London, England.

Horne, R. & Hickey, J. 1991. Ecological sensitivity of Australian rainforests to selectivelogging. Australian Journal of Ecology 16:119-129.

Hubbell, S.P., & Foster, R.B. 1986. Biology, chance, and history and the structure oftropical rain forest communities. Pages 314-329 in Diamond, J., & Case, TJ.(eds.). Community ecology. Harper & Row, New York, U.S.A.

Humphreys, WY. & Kitchener, D.J. 1982. The effect of habitat utilization on species-area curves: implications for optimal reserve area. Journal of Biogeography9:391-396.

ICBP. 1992. Putting biodiversity on the map: priority areasfor global conservation.Cambridge, England.: Interaional Council for Bird Preservation.

Illigo-Elias, E.E. & Ramos, MiA. 1991. The Pssaticine twade in Mexico. Pages 380-392 inRobinson, J.G. & Redford, KH. (eds.). Neotropical wildlife use andconservaton. The University of Chicago Press, Chicago, Illinois, U.S.A.

Janzcn, D.H. 1967. Why mountain passes are higher in the topics. American Naturalist101:233-249.

Janzen, D.H. 1983. Costa Rican natural history. University of Chicago Press, Chicago,U.S.A.

Janzen, D.H 1986. The etemal external theat. Pages 296-303 in Soul6, M.E. (ed.).Conservation biology: the science of scarcity and diversity. Sinauer,Sunderland, USA.

Janzen, D.H. 1987. Insect diversity of a Costa Rican dry forest: why keep it, and how?Biological Journal of the Linnean Society 30:343-356.

Johns, A.D. 1988. Effects of "selective" timber extraction on rain forest structure andcomposition and some consequences for frugivores and folivores. Biotropica20:31-37.

Junk, W.J. 1983. Mires: swamps, bog, fen, and moor ecology of swamps on the middleAmazon. Pages 269-294 in Gore, AJ.P. (ed.). Ecosystens of the world, Volume4B. Elsevier Scientific Publishing Co., Amsterdam.

Kaufman, L. 1992. Catastrophic change in species-rich freshwater ecosystems.BioScience 42:846-858.

Kubitzki, K. 1985. The dispersal of forest plants. Pages 192-206 in Prance, G.T. &Lovejoy, T.E. (eds.). Amazonia. Pergamon Press, Oxford, England.

Laurance, WI. 1991. Ecological correlates of extinction proneness in Auswtian tropicalrain forest mammals. Conservation Biology 5:79-89.

Loveoy, TE. 1980. Discontinuous wilderness: minimum areas for conservation. Parks5:13-15.

Lugo, A.E. & Snedaker, S. 1974. Ecology of mangroves. Annual Review of EcologicalSystens 5:39-64.

Lugo, A.E., Schmidt, R., & Brown, S. 1981. Tropical forests in the Caribbean. Arnbio10:318-324.

Mahar, D. 1989. Government policies and deforestation in Brazil's Amazon region.World Bank, Washington, D.C.

Mares, M.A. 1992. Neotropical mammals and the myth of Amazonian biodiversity.Science 255:976-979.

Margules, C.R., Nicholls, A.O., & Pressey, R.L. 1988. Selecting networks of reserves tomaximize biological diversity. Biological Conservation 43:63-76.

McClenahan, T.R. 1987. Overfishing and coal reef degadation: a preliminary report fronEast Africa. Conservation Biology 1:97-99.

McCloskey, M. 1993. Note on the fragmentation of primary rainforest. Ambio 22:250-251.

McNeely, J.A. 1988. Economics and biological diversity: developing and usingecononic incentives to conserve biological diversity. IUCN, Gland,Switzerland.

McNeely, J.A., Miller, K.R., Reid, W.V., Mitterneier, R.A., & Werner, T.B. 1990.Conserving the world's biological diversity. IUCN, Gland, Switzerland: WRI,CI, WWF-US, & WB, Washingto, D.C 193 pp.

Mooney, H.A. 1977. Convergent evolution of Chile and California - mediterraneanclimate ecosystens. Dowden, Hutchinson, & Ross, Stroudsburg, USA.

Myers, N. 1988. ThIeatened biotas: hotspots in tropical forests. The Enviromnentalist8:187-208.

Myers, N. 1993. Tropical forests: the main deforestaion fronts. EnvironmentalConservation 20:9-16.

Nelson, B.W., Fermeira, C.A., da Silva, MYP. & Kawasaki, M. 1990. Refugia, endemismcenters, and botanical collecting density in Brazilian Amazon. Nature 345:714-716.

Newmark, WD. 1991. Tropical forest entaon and the local extinction ofunderstory birds in the eastern Usambara mountains, Tanzania. ConservationBiology 5:67-78.

Noss,-R.F. 1983. A regional landscape approach to maintain diversity. BioSciece33:700-706.

Noss, R.F. 1992. Application of conservation biology to wilderness recovery. WildEarth, Specieal Issue:11-21.

Numata, M. (ed.). 1979. Ecology of grasslands and bamboolands in the world. Dr. W.Junk Publishers, Boston, USA.

Oren, D.C. 1987. Grande Carajas, inenational financing agencies, and biologicaldiversity in southeastern Braailian Amazonia. Conservation Biology 1:222-227.

Parsons, JJ. 1976. Forest to pasture: development or destuction? Revista de BiologlaTropical [SuppL] 24:121-138.

Peres, CA. & Terborgh, J.W. 1993. Redesigning Amazonian nature reserves: an analysisof the defensibility status of existing conservation units. Submitted toConservation Biology.

Peterson, A.T., Flores-Villela, O.A, Le6n-Paniagua, L.S., Llorente-Bousquets, JE.,Luis-Martinez, M A., Navarro-Sigienza, A.G., Torres-Chvez, M.G., Vargas-Fernandez, I.. 1993. Conservadon priorities in Mexico: moving up in the worldBiodiversity Letters 1:33-38.

Pires, J.M. & Prance, G.T. 1985. The vegetation types of the Brazilian Amazon. Pages109-145 in Prance, G.T. & Lovejoy, TE (eds.). Amawonia. Pergamon Press,Oxford, England.

Pons, LJ. & Fiselier, J1. 1991. Sustinable development of mangroves. LandscapeUrban Plan 20:103-109.

Pool, DJ., Snedaker, S.C. & Lugo, AE. 1977. Structure of mangrove forests in Florida,Puerto Rico, Mexico, and Costa Rica Biotropica 9:195-212.

Prance, G.T. (ed.). 1982. Biological diversificaton in the tropics. Columbia UniversityPress, New York, USA.

Pressey, R.L., Humphries, CJ., Margules, CR., Vane-Wright, R.I. & Williams, P.H.1993. Beyond opporunism: key pninciples for systematic reserve selection.Trends in Ecology and Evolution 8:124-128.

Quinn, J.F., & Hastings, A. 1987. Extinction in subdivided habitats. ConservationBiology 1:198-208.

Quinn, J.F., Wing, S.R., & Botsford, L.W. 1992. Harvest refugia in maine inverebratefisheries: models and applications to the Red Sea Urchin, Strongylocentrotusfranciscanus. Symposium on the Crisis in Invemebrate Conservation, Vancouver,Canada.

Ramamoorthy, T.P., Bye, R., Lot, A., & Fa, A. 1993. Biological diversity of Mexico:origins and distribution. Oxford University Press, New York, New York, U.S.A.777 pp.

Redford, KH. 1991. The ecologically noble savage. Cultural Survival Quarterly 15:46-48.

Redford, K.H. 1992. The empty forest. BioScience 42:412-422.Redford, KMH., Taber, A., & Simonetti, J.A. 1990. There is more to biodiversity than the

tropical rain forests. Conservation Biology 4:328-330.Redford, K.H. & Stearrnan, A.M. 1993. Forest-dwelling native Amazonians and the

conservation of biodiversity: interests in common or in collision. ConservationBiology 7:248-255.

Robinson, J.G. & Ramirez, J. 1982. Conservation biology of neotropical primates. Pages329-344 in Mares, M.A. & Genoways, H.H. (eds.). Mammalian biology in SouthAmerica. Special Publication Series, Pymatuning Laboratory of Ecology,University of Pittsburg, Linersville, U.S.A.

Robinson, J.G. & Redford, K.H (eds.). 1991. Neotropical wildlife use andconservanion. The University of Chicago Press, Chicago, Illinois, U.S.A.

Robinson, S.K., & Wllcove, D.S. 1989. Conserving tropical raptors and game birds.Conservation Biology 3:192-193.

Rub6n Vila, A. & Bertonatti, C. 1993. Situacion ambiental de la Argentina:recomendaciones y prioridades de accion. Fundacion Vida Silvestre Argentina,Boledn Tecnico 14.

Sader, S.A. & Joyce, A.T. 1988. Deforestation rates and trends in Costa Rica, 1940 to1983. Biotropica 20:11-19.

Sala, O.E., Deregibus, V.A., Schlichter, T., & Alippe, H. 1981. Productivity dynamics ofnative temperate grasslands in Argentina Journal of Range Management 34:48-51.

Salafsky, N., Dugelby, BL., & Terborgh, J.W. 1993. Can exactve reserves save therain forest? An ecological and socioeconomic comparison of non-timber forestproduct extraction sysems in Peten, Guatemala, and West Kalimantan, Indonesia.Conservation Biology 7:39-52.

Salm, R.V. & Clark, JR. 1984. Marine and coastal protected areas: a guide forplanners and managers. IJCN, Gland, Switzerland.

Saunders, D.A., Hobbs, R.J., & Margules, C.R. 1991. Biological consequences ofecosystem fagmentation: a review. Conservation Biology 5:18-32.

Sayer, J.A. & Whitnore, T.C. 1991. Tropical moist forests: destruction and speciesextinction. Biological Conservation 55:199-213.

Sheldon, A.I. 1988. Conservation of stream fishes: patterns of diversity, rarity, and risk.Conservation Biology 2:149-156.

Sims, P.L 1988. Grasslands. Pages 232-260 in Barbour, M.G., & Billings, W.D.(ods.). North American terrestrial vegetation. Cambge University Press,Cambridge, England.

Sioli, H, Schwabe, G.H., & Klingee, H. 1969. Limnological outlooks on landscapeecology in Latn Amenca. Tropical Ecology 10:72-82.

Skole, D. & Tucker, C. 1993. Deforestation and habitat fragmentation in the Amazon:satellite data from 1978 to 1988. Science 260:1905-1910.

Simberloff, D. 1992. Do species-area curves predict extinction in fragmented forest?Pages in Whitmome, T.C. & Sayer, J.A. (eds.). Tropical deforestation and speciesextincton, Chapter 4. Chapman & Hall, London, England.

Soule, ME., Bolger, D.T., Alberts, A.C, Wright, J., Sorice, M. & Hill, S. 1988.Reconstructed dynamics of rapid extinctions of chaparral-requiring birds in urbanhabitat islands. Conservation Biology 2:75-92.

Stevens, G.C. 1989. The latitudinal gradient in geographical range: how so many speciescoexist in the tropics. American Naturalist 133:240-256.

Terborgh, J. 1992. The maintenance of diversity in tropical forests. Biotropica 24:283-292.

Terborgh, J. & Wmter, B. 1982. Evolutionary circumstances of species with small ranges.Pages 587-600 in Prance, G.T. (ed.). Biological diversfication in the tropics.Columbia University Press, New York, U.S.A.

Thiollay, J. 1992. Influence of selective logging on bird species diversity in a Guiananrain forest Conservation Biology 6:47-63.

Twilley, RR., Bodero, A., & Robadue, D. 1993. Mangrove ecosystem biodiversity andconservation in Ecuador. Pages 105-127 in Potter, CS., Cohen, J1., &Janczewski, D. (eds.). Perspective on biodiversity: case studies of geneticresource conservation and development. AAAS Prcss, Washington, D.C., U.S.A.

Uhl, C. & Veira, I.CG. 1989. Ecological impacts of selective logging in the BrazilianAmazon: a case study from the Paragominas region of the state of ParaBiotropica 21:98-106.

Uhl, C. & Kauffman, J.B. 1990. Deforestation, fire susceptibility, and the potential treeresponses to fire in the eastern Amazon. Ecology 71:437-449.

UNESCO, 1980. Vegetation map of South America. 1:5,000,000. Institut de la Carieantnationale de Tapis VegetaL Toulouse, France.

Vane-Wright, R.L, Humphries, CJ., & Williams, PR 1991. What to protect?-Systematics and the agony of choice. Biological Conservaton 55:235-254.

Walker, BH. 1992. Biodiversity and ecological redundancy. Conservation Biology6:118-123.

Walsh, GE. 1977. Wet coastal ecosystems: exploitation of mangal. In Chapman, BJ.(ed.). Ecosystens of the world, Volume 1. Elsevier Scienific Publishing Co.,Amsterdam.

WCMC. 1992. Global Biodiversity. Chapman & Hall,. New York, USA. 585 pp.WCMC. 1993. The WCMC bodversity mW librwry: availability and distribution of

GIS datasets. World Conservation Monitoring Center. Cambridge, EnglandVhiatmore, T.C. & Sayer, J.A 1992. Tropical deforestation and species extiction.

Chapman & Hall, Laoion, England. 147 pp.Wilcove, D.S., McLellan, CR, & Dobson, A.P. 1986. Habitat fragmentation in the

temperate zone. Pages 237-256 in Souls, ME. (ed.). Conservation biology: thescience of scarcity and diversity. Sinauer, Sunderland, USA.

APPENDIX I. vList of ecoregions to be analyzed by this study. Ecoregions are classifiedaccording to the DSE and SSH category in which they occur (see Figure 2). Whereverpossible, we have used recognizable geographical names to describe the ecoregions.

Ecoregion List Structure:

Major Biogeographic Region (S,M,C)

1. Dynamically Similar Ecosystems (DSE)A. Structurally Similar Habitat (SSH)

(Secondary classification, not used in analyses)1. Ecoregion Name - Country

South America (S)

1. TROPICAL BROADLEAF FORESTSA. Tropical Moist Forests

(Lowland Forests).S1AO1. Trinidad & Tobago - Trinidad & TobagoS1A02. Choc6 Region - Colombia/ EcuadorS1A03. Western Ecuadorian Moist Forest - EcuadorS1A04. Rio Tocantins/Rio Gurupa Region - BrazilS1A05. Rio Tapajos/Rio Xingu Region - BrazilSlA06. Rio Juni/ Rio Tapajos Region - BrazilS1A07. Rio Itacaidna (west along southern Rond6nia) - BrazilSIA08. Acre Region - Brazil/Peru/BoliviaSlA09. Marafion Region - Brazil/Peru/EcuadorSlAIO. Napo Region - Peru/Ecuador/ColombiaSlAl 1. Orinoco Complex (including areas east of Bogota) -

Colombia/Venezuela/BrazilSlA12. Upper Rio Negro Region - BrazilSIA13. Rio Trombetas/Watamah Region - BrazilSIA14. Guianan Shield Forest (Northern Watershed Basin of the Serra Acarai/

Serra Tumucumaque) - Guyana/Suriname/Brazil/French GuianaSlA15. Magdalena Medio/Urabg - ColombiaSlA16. Maracay Humid Forest - Venezuela/ColombiaSlA17. Brazilian Lowland Atlantic Coastal Forest - BrazilSIAI8. Brazilian Interior Atlantic Coastal Forest - Brazil

(Tropical Swamp Forests)SlA19. Varzea [also include Rfo Amazonas Coastal Swamp Forest] - BrazilSlA20. Delta Orinoco - VenezuelaS1A21. Paramaribo Complex - SurinameS1A22. Choc6 Coastal Swamp Forest - ColombiaS1A23. Maracay Swamp Forest - Venezuela

S1A24. Western Amazon Swamp Forest - Brazil(Submontane Forests)

SlA25. Brazilian Highland Atlantic Coastal Forest - BrazilSIA26. Northem Venezuelan Submontane Forest - VenezuelaSlA27. Western Andes Submontane Forest - Colombia/EcuadorS 1A28. Cauca Valley Submontane Forest - ColombiaSlA29. Magdalena Valley - ColombiaSlA30. Sierra Nevada de Santa Marta - ColombiaS 1A3 1. Cordillera Merida - ColombiaS1A32. Sierra de Macarena - ColombiaS1A33. Eastern Andes of Colombia - ColombiaS1A34. Eastem Andes of Ecuador - EcuadorS1A35. South Eastern Peruvian Andes - PeruS1A36. Northem Yungas - BoliviaS1A37. Orinoco Block - Venezuela/Brazil

(Montane Forests)S1A38. Sierra Nevada de Santa Marta - ColombiaSlA39. Cordillera de M&rida - VenezuelaSIA40. Cordillera Oriental - Colombia/VenezuelaSlA41. Cordillera Central - ColombiaS1A42. Cordillera Occidental - ColombiaS1A43. Cordillera Real - Ecuador/PeruS1A44. Peruvian Cordillera Central - PeruSlA45. Peruvian Cordillera Oriental - Peru/BoliviaS1A46. Tepuis - Venezuelan/Brazil/GuyanaSIA47. S. Andean Montane Forest - Argentina/BoliviaS1A48. Ceja de Monte - Bolivia

B. Tropical Dry Broadleaf ForestsSIBO1. Cerrado - Brazil

including; Mato Grosso do Sol Complex - BrazilAmapa Complex - BrazilSerra do Espinahaqo - Brazil

SlB02. Western Guayaquil Forest - EcuadorS lB03. Golfo de Guayaquil Dry Forestl N. Peruvian Deciduous Thom Forest

Ecuador/PeruSIB04. Upper Cauca Valley Dry Forest - ColombiaSlB05. Sing Valley Dry Forest - ColombiaSlB06. Upper Rfo Magdalena Deciduous Thorn Forest - ColombiaS1B07. Venezuelan Dry Forest Complex [S. of M&rida Mtns.] -

VenezuelaSlB08. Peruvian Dry Montane Forest - PeruSlB09. Chaco Thom Forest - ArgentinaSIB1O. Paraguayan Seasonal Lowland Forest - ParaguaySlB11. Bolivian Seasonal Lowland Forest - BoliviaSlB12. Rio de la Plata Thorn Forest - ArgentinaSIB13. Bolivian Montane Dry Forest - Bolivia

2. CONIFER OR TEMPERATE BROADLEAF FORESTSA. Temperate Forests

S2A01. Islas Malvinas Subpolar Forest - ArgentinaS2A02. Chilean Winter Rain Forest - ChileS2A03. Subpolar Nothofagus Forest - Chile/ArgentinaS2A04. Valdivian Forest - ChileS2A05. Brazilian Aracuaria Forest - Brazil

3. GRASSLANDS/SAVANNAHS/WETLANDS/SHRUBLANDSA. Grasslands/Savannahs/Shrublands

S3AO1. Uruguayan/Brazilian Grass Savannah - Uruguay/BrazilS3A02. Deciduous Woodland and Chaco - ArgentinaS3A03. Fiambala Woodland - ArgeninaS3A04. Pampas - ArgentinaS3A05. Suriname Savannas & Rio Branco - Suriname/Guyana/BrazilS3A06. Llanos - Venezuela/ColombiaS3A07. Argentine grassland with palms - ArgentinaS3A08. Beni Savannahs of Bolivia - BoliviaS3A09. Monte de Argentina - ArgentinaS3A.10. Espinal de Argentina - Argentina

B. Flooded Grasslands & SavannahsS3B01. Wet Chaco - Argentina/Paraguay/Uruguay/BrazilS3B02. Pantanal - Brazil/Bolivia/ParaguayS3B03. Rio Negro - ArgentinaS3B04. Laguna Mar Chiquita (Salina Grandes & Ambargasta] - ArgentinaS3B05. Rfo Deseado - ArgentinaS3B06. Rfo Atuel - ArgentinaS3B07. Beni Swamp Gallery Forest - BoliviaS3B08. Amazonia Campos de Varzea - BrazilS3B09. Peruvian Amazon - PeruS3B10. Sio Luis Flooded Grasslands - BrazilS3B1 1. Upper Rio Xingu/ Araguaia Complex - BrazilS3B12. Rio Atrato - ColombiaS3B13. Guayaquil Flooded Grasslands - EcuadorS3B14. Maracay Flooded Grsslands - VenezuelaS3B15. Cienaga Grande Complex - ColombiaSB316. Orinoco Complex - VenezuelaS3B17. Andean Salt Marsh - Argentina/Bolivia/ChileS3B18. Malargue Flooded Grasslands - ArgentinaS3B19. Rio Deseado - ArgentinaS3B20. Camborombo - Argentina

C. Montane Grasslands(Temperate)

S3CO1. Puna de Argentina - ArgentinaS3C02. Patagonian Steppe - Argentina

S3C03. Southern Andean Steppe - Argentina/Chile(Tropical)

S3C04. Paramo de Sierra Nevada (Santa Martal - ColombiaS3C05. Pnamo de Cordillera Central - ColombiaS3C06. Pramo de Cordillera Oriental - ColombiaS3C07. Cordillera de Merida - VenezuelaS3C08. P&ramo de Cordillera Real - EcuadorS3C09. Paramo de Cordillera Central - PeruS3C10. Central Andean Wet Puna - Peru/BoliviaS3C1 1. Central Andean Puna - Bolivia/Chile/Argentina/PeruS3C12. Central Andean Dry Puna - Argentina/Bolivia/Colombia

4. XERIC FORMATIONSA. Mediterranean Chaparral

S4A01. Chilean Chaparral - ChileB. Deserts & Xeric Shrublands

(Xeric Shrublands)S4B01. Peninsula de Guajira/Barranquilla - ColombiaS4B02. Paraguana - VenezuelaS4B03. Peninsula de Araya y Paria - VenezuelaS4B04. Western Andean Shrublands - Chile/PeruS4B05. Caatinga - BrazilS4B06. Lara-Falc6n Xeric Shrublands - Venezuela

(Deserts)S4B07. Sechura Desert - Ecuador/Peru/ChileS4B08. Atacama Desert/Lomas - Chile

C. Restingasl Dune VegetationS4C01. Rio Grande Complex - BrazilS4C02. Bahfa Blanca Complex - ArgentinaS4C03. Paraguana Complex - BrazilS4C04. Northeastern Brazil Dune Vegetation [near Fortaleza] - BrazilS4C05. Paraguana Complex - Venezuela/Colombia

5. MANGROVESA. Mangroves

S5A01. Amazonian Delta/Tocantins - BrazilS5A02. Sio Luis Delta - BrazilS5A03. Orinoco/Guayanian Coast Complex [from Paramaribo to Caracas] -

Venezuela/Guyana/Suriname/French Guiana/Trinidad & TobagoS5A04. Salvador/Paraguana Complex - BrazilS5A05. Guayaquil Complex - EcuadorS5A06. Choc6/Gulf of Panama/San Miguel Complex - Colombia/PanamaS5A07. Maracay Complex - VenezuelaS5AO8. Colombian Basin Mangrove Complex - Panama/Colombia

- Rfo Atrato Complex - Colombia- Barranquilla Complex - Colombia

Mexico and Central Amerna (M)

1. TROPICAL BROADLEAF FORESTS;A. Tropical Moist Forests

(Lowland Forests)MlAOI. Atlantic Coast Evergreen Rain Forest

- Honduras/Nicaragua/Costa Rica/PanamaMIA02. Pacific Coast Evergreen Forest - Costa Rica/PanamaMlA03. Veracruz Semi-Evergreen Forest - MexicoMlA04. Nayarit Semi-Evergreen Forest - MexicoMlA05. Costa Rican Semi-Evergreen Forest - Costa Rica

(Submontane Forests)MlA06. Veracruz Rain Forest - MexicoMlA07. Mid-elevation Pacific Coast Evergreen Rain Forest

M- exico/Guatemala/EI SalvadorMlA08. Guatemalan Forest - Guatemala/El Salvador/Honduras/BelizeMlA09. Talamancan Forest - Costa Rica/Panama

(Montane Forests)MlA10. Veracruz Cloud Forest - Mexico

B. Tropical Dry Broadleaf ForestsMIBOI. Sinaloan Deciduous Forest - MexicoMlB02. Tamaulipan Semi-Deciduous Forest (Selva baja caducifolio)

MexicoMlB03. San Lucan Deciduous Forest - MexicoM1B04. Cental American Pacific Dry Forest -

El Salvador/Honduras/Nicaragua/Costa RicaMlB05. Yucathn Semi-Deciduous Forest - MexicoMIB06. Yucattn Deciduous Forest - MexicoMlB07. Guerreran Thornscrub-Deciduous Forest - MexicoMlB08. Azuero Dry Forest - Panama

2. CONIFER OR TEMPERATE BROADLEAF FORESTS;A. Temperate ForestsB. Tropical & Subtropical Forests

M2B01. Sierra Madre Oriental Montane Conifer Forest - MexicoM2B02. Sierra Madre Occidental Montane Conifer Forest - MexicoM2B03. Mexican Transvolcanic Montane Conifer Forest - MexicoM2B04. San Lucan Evergreen Woodland - MexicoM2B05. Sierra Madre Oriental Pine/Oak Forest - MexicoM2B06. Sierra Madre Occidental Pine/Oak Forest - MexicoM2B07. Mexican Transvolcanic Pine/Oak Forest - MexicoM2B08. Guatemalan Montane Forest - GuatemalaM2B09. Altiplano Evergreen Woodland

- Guatemala/EI Salvador/Honduras[also includes S. Veracruz Oak Woodland patches]

M2B10. Guerreran Pine/Oak Forest - Mexico- Rlo Coahuayana to Tawantapec

M2B1 1. Honduran Semi-Evergreen Forest - HondurasM2B12. Miskito Pine-Savannah Forest - Nicaragua/HondurasM2B13. Pacific Coast Upland Pine Forest - NicaraguaM2B14. Sierra de Juarez/San Pedro Martir Conifer-Oak Forest - MexicoM2B15. Sierra Madre Occidental Oak Forest - MexicoM2B16. Sierra Madre Oriental Oak Forest - MexicoM2B17. Mexican Transvolcanic Oak Forest - MexicoM2B18. Mexican Transvolcanic Pine Forest - MexicoM2Bl9. Eastern Tehuantepec Isthmus Oak Forest - Mexico

3. GRASSLANDS/SAVANNAHS/WETLANDS/SHRUBLANDSA. Grasslands/Savannahs/Shrublands

M3A0l. Mexican Semidesert Grassland - MexicoM3A02. Cuchamatanis Savannah Grassland - GuatemalaM3A03. Santiago Pinotepa Nacional - MexicoM3A04. Grijalva River - Mexico

M3A05. Campechian - Veracruz Savannah - MexicoB. Flooded Grasslands

M3B01. Cuatro Cienegas (marsh) - MexicoM3B02. Miskito Coast Wetlands/Caratasca Lagunes - Nicaragua/HondurasM3B03. Sian Kahn Flooded Grasslands - Mexico

C. Montane Grasslands(Tropical Montane)

M3C0l. Costa Rican Pmramo - Costa RicaM3C02. Mexican Alpine Tundra/Zacatonal - Mexico

4. XERIC FORMATIONSA. Mediterranean Chaparral

M4A0l. California Chaparral - Mexico/USM4A02. California Coastal Scrub - Mexico/US

B. Desert & Xeric Shrublands(Xeric Shrublands)

M4B01. Sinaloan Thornscrub - MexicoM4B02. Tamaulipian Thornscrub [mezquital w/some pastizal]

- Mexico/USM4B03. San Lucan Thomscrub - MexicoM4B04. Centrl Mexican Semi-Desert Grassland [mezquital/pastizal/matormal

crasicaule/matorral desertico rosetofilo] - Mexico(southern block only)

M4B05. Chihuahuan Interior Chaparral [matorral submontano - Mexico(Desert)

M4B06. Mohave Desertscrub - Mexico/USM4B07. Sonoran Desertscrub (matorral crasicaule/mezquital/matorral desertico

microfilol - Mexico/US

M4B08. Baja Desertscrub [matorral crasicaule] - MexicoM4B09. Chihuahuan Desertscrub [matoral desertico microfilo] - Mexico/US

C. Restingas/ Dune Vegetation

5. MANGROVESA. Mangroves

M5AO1. Mexican Pacific Coast Mangrove Complex- Marismal Nacionales Complex [near Nayarit] - Mexico- Bahfa Magdalena Complex [Baja Mexico] - Mexico- Sinaloan Complex - Mexico

M5A02. Northern Pacific Meso-American Mangrove Complex- Bahia de Jiquilisco - El Salvador- Manch6n Region - Guatemala- Golfo de Fonseca - El Salvador/Honduras/Nicaragua- La Encrucijada Region - Mexico

M5A03. Southern Pacific Meso-American Mangrove Complex- Golfo de Nicoya - Costa Rica- Tierra-Sierpe Delta - Costa Rica- Golfo de San Miguel - Panama- Golfo de Chiriqui - Panama

M5A04. Golfo de Mexico/Atlantic Mangrove Complex- Laguna Madre (northeastern Mexico)- Tabasco Complex - Mexico- Celestun - Mexico- YucatAn/Campeche Complex - Mexico

M5A05. YucatAn Basin Meso-American Complex- Quintan Roo Complex - Mexico- Belizean Mangroves - Belize- Golfo de Honduras - Belize/Guatemala/Honduras

M5A06. Meso-American Caribbean Sea Complex- Cartasca/Miskito Coast - Honduras/Nicaragua- SI-A-PAZ/Tortuguero Area - Nicaragua/Costa Rica- Bocas del Toro - Panama

Caribbean Islands (C)

1. TROPICAL BROADLEAF FORESTSA. Tropical Moist Forests

(Bahamas)CIA01. Bahamian Moist Forest

(Greater Antilles)ClAO2. Cuban Moist ForestClA03. Cuban Cloud ForestClA04. Jamaican Moist ForestClAO5. Puerto Rican Subtropical Moist ForestClA06. Puerto Rican Cloud ForestClA07. Hispaniolan Moist ForestClA08. Virgin Islands Moist Forest

(Lesser Antilles)ClA09. Leeward Islands Moist Forest

- Netherland Antilles Evergreen Forest - SabalSt. Eustatius/St. Martin- St. Kitts-Nevis Evergreen Forest- Montserrat Evergreen Forest- Antiguan Moist Evergreen Forest- Guadeloupe Evergreen Forest- Dominican Evergreen Forest

ClA10. Windward Islands Moist Forest- St. Lucian Evergreen Forest- St. Vincent & The Grenadine Islands Evergreen Forest- Barbados Moist Evergreen Forest

B. Tropical Dry Broadleaf Forests(Bahamas)

CIB01. Bahamian Dry Forest - Southern Bahamas/ Turks & Caicos Islands(Greater Antilles)

CIB02. Cuban Semi-Deciduous Forest (including - Isle of Pines Dry Forest)ClB03. Hispaiiiola Dry Forest - Haiti/Dominican RepublicClB04. Jamaican Dry ForestClB05. Puerto Rican Dry ForestClB06. Virgin Island Dry Forest

C. Montane Forests

2. CONIFER OR TEMPERATE BROADLEAF FORESTS;A. Temperate ForestsB. Tropical & Subtropical Forests

(Bahamas)C2BOL. Bahamian Evergreen Forest

(Greater Antilles)C2B02. Haiti/Dominican Republic Semi-Evergreen Forests

- Massif du Sud Semi-Evergreen Forest - Haiti

- Cordillera Central Semi-Evergreen Forest - Dominican R.- Hispaniola Evergreen Forest - Haiti/Dominican Republic

C2B03. Cuban Evergreen ForestsC2B04. Blue Mountains Semi-Evergreen Forest - JamaicaC2B05. Cayman Islands Dry Forest

3. GRASSLANDS/SAVANNAHS/WETLANDS/SHRUBLANDSA. Grsslands/Savannahs/ShrublandsB. Flooded Grasslands

C3BO0. Lago Enriquillo Wetlands of Hispaiiiola - Haiti/Dominican RepublicC3B02. Wetlands of Cuba (including Peninsula de Zapata)

C. Montane Grasslands

4. XERIC FORMATIONSA. Meditean ChaparralB. Desert & Xeric Shrublands

(Greater Antilles)C4BOL. Cuban Cactus ScrubC4B02. Cuban Serpentine ScrubC4B03. Hispaiiolan Thornscrub - Haiti/Dominican RepublicC4B04. Jamaican Thornscrub - JamaicaC4B05. Cayman Islands ScrubC4B06. Puerto Rican Subtopical Dry ForestC4BM7. Virgin Islands Dry WoodlandC4B08. Virgin Islands Cactus Scrub

(Lesser Antilles)C4B09. Leeward Islands Dry WoodlandC4B10. Leeward Islands Cactus ScrubC4B1 1. Windward Islands Dry WoodlandC4BI2. Windward Islands Cactus Scrub

(Offshore Islands/Southern Lesser Antilles)C4B13. Aruba/Curasao/Bonaire Dry WoodlandC4B14. Aruba/Caraio/Bonaire Cactus Scrub

C. Restingas/ Dune Vegetation

5. MANGROVESA. Mangroves

(Lesser Antilles)CSA01. Leeward Islands

- Anguilla Mangroves- St. Barthelemy Mangroves- St. Kitts-Nevis Mangroves- Montserrat Mangroves- Barbuda Mangroves- Antigua Mangroves- Guadeloupe Mangroves

C5A02. Windward Islands- Martinique Mangroves- St. Lucia Mangroves- St. Vincent & The Grenadine Islands- Barbados Mangroves- Grenadian Mangroves

(Greater Antilles)C5AO5. Cayman IslandsCSAO6. Puerto Rican MangrovesCSAO7. Virgin Island MangrovesCOA08. Dominican MangrovesC5A09. Cuban MangrovesC5A10. Jamaican MangrovesC5AI1. Hispafiiolan Mangroves

(Bahamas)C5A12. Bahamian Islands

(Off Shore Islands/Islas de SotaventoC5A13. Aruba/Curacao/Bonaire Mangroves

LATEN DISSEMINATION NOTES

IDat1 Author

1 Sustainability, Yield Loss and Imediatisimo: April 1993 Robert SchneiderChoice of Technique at the Frontier Gunars Platais

David RosenblattMaryla Webb

2 The Potential for Trade with the Amazon in April 1993 Robert SchneiderGreenhouse Gas Reduction

3 Land Abandonment, Property Rights, and April 1993 Robert SchneiderAgricultural Sustainability in the Amazon

4 The Urban Environmental Challenge in August 1993 John DixonLatin America

5 An Analysis of Flooding in the September Robert J. Anderson, Jr.Parana/Paraguay River Basin 1993 Nelson da Franca Ribeiro

dos SantosHenry F. Diaz

6 Ecology and Microeconomics as 'Joint October 1993 John A. DixonProducts": The Bonaire Marine Park in the Louise Fallon ScuraCaribbean Tom van't Hof

7 Forest Management and Competing Land October 1993 Nalin M. KishorUses: An Economic Analysis for Costa Rica Luis F. Constantino

8 Pueblos Indfgenas y Desarrollo en Am6rica December Jorge E. UquillasLatina 1993 Jean-Carlo Rivera

9 Prospects for Improved Management of December Robert KinnseNatural Forests in Latin America 1993 Luis Constantino

George Guess

10 Assessing the Conservation Potential and February 1994 David M. OlsonDegree of Threat Among Ecoregions of Eric DinersteinLatin America and the Caribbean: A (World Wildlife Fund)Proposed Landscape Ecology Approach

For back issues of the above, please contact:

Environment Division (LATEN)Latin America and the Caribbean Technical Department

The World Bank1818 H Street, N.W.

Washington, DC 20433U.S.A.

Tel. (202) 473-7793Fax: (202) 676-9373