species, data, and conservation planning

Species, Data, and Conservation PlanningTHOMAS BROOKS,∗ GUSTAVO A. B. DA FONSECA∗†‡, AND ANA S. L. RODRIGUES∗∗Center for Applied Biodiversity Science, Conservation International, Washington, D.C. 20036, U.S.A.†Departmento de Zoologia, Universidade Federal de Minas Gerais, Belo Horizonte 31270, Brazil

Introduction

Extinction is irreversible. The restoration of habitats, eco-logical processes, and ecosystem services is complex butpossible (Dobson et al. 1997a). We cannot, however, re-store the building blocks of those higher levels of biolog-ical organization once species go extinct. For this reason,above all, we must consider minimization of the speciesextinction rate our most urgent target and the measure ofconservation success. Molnar et al., Higgins et al., Cowl-ing et al., and Pressey (2004, this issue) contribute a vari-ety of insights on this matter, and for clarity of response,we synthesize their insights into five topics. We start byaddressing the proposals of these authors for consideringdata and targets for three other scales of ecological organi-zation: habitats or other spatial classifications (hereafterland types, following Pressey), ecological processes, andecosystem services. We then tackle the issue of surrogacybetween data types. Finally, we address some of the prob-lems with species-based approaches.

We were invited to write our initial essay (Brooks et al.2004a) to build on discussions of setting targets for biodi-versity conservation planning and “argue that, despite arange of shortcomings, species data provide the best basisfor setting these targets.” We were not charged to reporton the World Parks Congress in general (i.e., the roleof the Durban Accord, [http://www.iucn.org/themes/wcpa/wpc2003/pdfs/outputs/wpc/durbanaccord.pdf])or to report on Workshop Stream 7 (Building Compre-hensive Protected Area Systems), which we do elsewhere(Brooks et al. 2004b) in collaboration with a numberof the authors who respond here to our ConservationBiology article. We emphasize that our use of the word“armchair” was not intended as derogatory. Indeed, therespondents to our essay include some of the world’sleading natural historians.

‡Address correspondence to: G. A. B. da Fonseca, email [email protected] submitted August 30, 2004; revised manuscript accepted Au-gust 31, 2004.

Species Data, Environmental Data, or Both

There is considerable agreement between our views andthose of Higgins et al., Cowling et al., and Pressey thatconservation planning should make the best use of bothspecies and environmental data. What are the disagree-ments? There is a difference in our respective concerns.We called for species data to remain central to conserva-tion planning. In contrast, the respondents fear the ex-clusion of the use of environmental data or of land typesdetermined from such data. In our view, there is no dan-ger of the latter, but there is a risk that in this techno-logical age conservation planners may be tempted to relyon remote-sensing data and computer models without in-corporation of the available species data (e.g., Faith et al.2001).

The main disagreement, however, concerns the rea-sons why environmental data are important. From ourperspective, this information is essential because it canhelp us correct biases and fill gaps in species knowledge.A different view, explicit in Molnar et al., at least, is thatland types are biodiversity entities, and therefore conser-vation targets, in their own right.

We presented some arguments as to why we disagreewith this perspective, but we expand on these here. Weview land types as components of classification systems,somewhat arbitrary mind constructs that help us organizeideas in a complex and biodiverse world. We agree withPressey that ideally, if we had perfect species data (onthe distributions and on the ecological interactions), wewould no longer need environmental data for conserva-tion planning. Although we currently need such data, wemust keep in mind their subjective nature and the needto interpret the results of conservation planning basedon environmental data in the context of the particularclassification system chosen to create them.

Take, for example, the broadest land-type system,biomes. A major announcement at the Fifth World ParksCongress was that the 10% target has been surpassed for9 out of 14 major terrestrial biomes (Fig. 1), based on theUdvardy (1975) biome classification system (Chape et al.2003). If the more recent biome classification of Olson et

1682

Conservation Biology, Pages 1682–1688Volume 18, No. 6, December 2004

Brooks et al. Species, Data, and Conservation Planning 1683

Figure 1. Percentage of protected area in each biome according to the 2003 World Database on Protected Areas(WDPA; World Database on Protected Areas Consortium 2003). Results according to the Olson et al. (2001) andUdvardy (1975) classifications (Chape et al. 2003). The results according to the Olson et al. (2001) classificationwere obtained through an overlap of their corresponding biome layer with that of the WDPA. Arrows indicatebroad correspondence between biomes in the different classification systems.

al. (2001) had been used instead, it would have revealedthat the 10% target is surpassed for only 8 of 16 biomes.Additionally, different conclusions would have been ex-tracted about the coverage of particular biomes: for exam-ple, 12.8% protection of Tropical Dry Forests/Woodlandswould have been achieved under Udvardy’s classification(1975), but Olson et al.’s classification (2001) would haveyielded only 7.9% protection of Tropical and SubtropicalDry Broadleaf Forests. It would therefore be unwise to gettoo attached to the components of a particular classifica-tion system and to take these representation figures tooliterally in assessing progress toward global conservationefforts, as Molnar et al. propose. These systems will be re-fined as better species data become available, in the sameway the system proposed by Olson et al. (2001) refinedUdvardy (1975).

Another problem with taking land types too literally asconservation targets is the assumption that they representnatural spatial subdivisions of biological organization. Wemay forget that discontinuities that are very clear to thehuman eye are not necessarily important for species, andwe may fail to detect what species perceive as majorhabitat changes. Consider, for example, the North Cen-tral Rockies Forest and Northern Short Grassland ecore-gions in North America (Fig. 2). To our eyes, they are verydistinct, being dominated by quite different biomes: Tem-perate Coniferous Forests and Temperate Grasslands, Sa-

vannas, and Shrublands, respectively (Olson et al. 2001).In contrast, the Northwestern Andean Montane Forestsecoregion is represented as a single ecological unit. Wefind higher levels of faunal similarity, however, in com-paring the Rockies with the Temperate Grasslands thanin comparing the northern and southern portions of theAndean Montane Forests ecoregion. In other words, fromthe perspectives of amphibian and bird (although notmammal) species, the northern and southern parts of theNorthwestern Andean Montane Forests are more distinctthan are two different ecoregions within two differentbiomes in North America. Pressey argues that planners“use land types as generalizations about environmentalvariation assumed to be relevant for at least a propor-tion of the species lacking descriptions, adequate pointrecords, or reliable models,” but it is difficult to trust inthe relevance of these classes for unknown species whenthey are not obviously adjusted to the distribution pat-terns of the species we do know.

When land types are used as biodiversity units for con-servation planning, targets are typically set such that agiven percentage of each should be protected (e.g., Strit-tholt & Boerner 1995; Powell et al. 2000; Pressey et al.2000). This approach would ensure that we at least rep-resent the Rockies forests and the Northern Short Grass-lands, but nothing would guarantee that both the north-ern and southern part of the Andes Montane Forests and

Conservation BiologyVolume 18, No. 6, December 2004

1684 Species, Data, and Conservation Planning Brooks et al.

Figure 2. Faunal similarity of land types (Olson et al.2001). Tables (on the maps) present informationcomparing the similarity between the faunacomposition of (a) the North Central Rockies Forestsin relation to the Northern Short Grasslandsecoregions and (b) the northern section of theNorthwestern Andean Montane Forests in relation tothe southern section. We assessed similarity with theJaccard coefficient ( J). For two regions A and B, J =a/(a + b + c), where a is the number of speciescommon to both regions, b is the number of speciespresent in (a) but not (b), and c is the number ofspecies present in (b) but not (a). The J varies between0 (minimum) and 1 (maximum similarity). SeeBrooks et al. (2004b) for data sources.

their respective and very different species assemblageswould be represented. This is particularly problematicwhen these well-meant conservation targets become ceil-ings (Soule & Sanjayan 1998).

The example of the Northwestern Andean MontaneForests ecoregion illustrates another shortcoming of us-ing land types as conservation targets—the assumptionthat they are uniform. It makes a difference whether weconcentrate our conservation efforts in one of this ecore-gion’s sections or whether we locate them strategicallyacross the entire ecoregion. We have learned from sys-tematic conservation planning that it matters where pro-tected areas are located, not just how much area they oc-cupy (e.g., Pressey 1994), a lesson that is as valid acrossan entire region as within each of the land types that com-pose such a region. Even sophisticated techniques for set-ting variable percentage targets across land types (Desmet& Cowling 2004) do not address the former question.

Another point of debate concerns the ordering of in-formation and analysis: which should take precedence,species data or environmental data? Higgins et al. andCowling et al. enthuse about the coarse- and fine-filterapproach of Noss (1987), which would incorporate en-vironmental data first. In contrast, we argue that speciesdata should take precedence, with environmental infor-mation being used to expand the value of the species data.Put another way, conservation planning should have solidroots in what we already know about species and buildfrom that information, rather than starting with untestedassumptions, interpretations, or classifications about theway species perceive environmental variation. For exam-ple, the methods mentioned by Cowling et al. of “combin-ing environmental and available species data to produceland classes that reflect biological heterogeneity associ-ated with compositional turnover” include two quite dis-tinct approaches. One is the environmental diversity (ED)method (Faith & Walker 1996; Faith 2003) applied to se-lect areas that are complementary in a multidimensionalenvironmental space. The other is the approach in whichthe observed spatial patterns of species turnover are mod-eled using environmental data (Ferrier 2002; Ferrier et al.2002). The former is based on a set of assumptions onhow species are distributed in the environmental space,and the latter is based on the observed patterns of speciesdistribution in such space. We would avoid the former,but see the latter as a promising approach (as we notedin our essay).

Ecological Process

We fully agree with Pressey that conservation planning re-quires not only representation of ecological pattern butalso of process, on which biodiversity depends for long-term persistence. Our statement (Brooks et al. 2004a)that techniques for mapping and measuring ecologicalprocesses are still in their infancy was not intended tosuggest that the aim of representing process should beabandoned because it is too difficult. The extraordinary,pioneering work in this field by Pressey, Cowling, andothers (e.g., Cowling et al. 2003) has contributed muchto tackling this problem. Rather, we intended simply tohighlight the huge task still ahead. The first challengeis to explicitly define the full set of processes we wantor need to conserve. At any given site the list of candi-date processes could be endless; for example, the inter-actions between every pair of species (e.g., predation),biogeochemical processes (e.g., nitrogen cycle), popula-tion dynamics (e.g., migration), and disturbance dynam-ics (e.g., fire). Although in some cases some processesare clearly vital, such as fire dynamics in Mediterraneansystems (Keeley & Fotheringham 2001), we cannot listall the critical processes in a particular region, let alone



understand how to best address them through conserva-tion planning.

With this said, we argue that one of the main advan-tages of species data is precisely to inform the spatialrequirements for biodiversity persistence. Setting targetsfor the processes mentioned in Pressey’s Table 1 can onlybe based on information on species’ ecological require-ments. Hence, the importance of soil interfaces (e.g.,Rouget et al. 2003) and of areas of climatic stability (e.g.,Fjeldsa et al. 1999) as speciation centers is obvious onlybecause of observed patterns in species endemism. Mi-gratory areas and dispersal barriers are revealed by dataon species distribution and movements (e.g., Powell &Bjork 1995); connectivity is dependent on the dispersalability of species (e.g., D’Eon et al. 2002); and habitat suit-ability and effective population sizes are clearly relativeto particular species (e.g., Allen et al. 2001). This is notto say that ensuring the persistence of the species thatwe know would be a sufficient umbrella or adequate sur-rogate for the persistence of all biodiversity. But it is notclear how the persistence of a land type could be assessedat all without looking at the persistence of its componentspecies.

Ecosystem Services

We consider species an important biodiversity conserva-tion target in their own right, but we fully agree withMolnar et al. on the need to address the maintenance ofthe ecosystem services that biodiversity provides to hu-mankind. What we fail to understand are the claims thata species-based approach is intrinsically contrary to anapproach that values ecosystem services. It is even lessclear to us why an approach based on land-type repre-sentation is better in addressing ecosystem services thanan approach based on species, as Molnar et al. claim. Inthe same way that not all land types are equally valuablefor species representation, presumably not all land typesare equally important for providing ecosystem services.Their argument that “a major advantage of conservationgoals that are based on habitats is that every place on theplanet remains a candidate for conservation protection”is bizarre—every place on the planet has species as wellas habitats.

We were unconvinced by the examples given by Mol-nar et al. as proof that there is an incongruity in an ap-proach focused on retaining global species diversity andone based on retaining ecosystem services. Salt marshesand other highly productive wetlands are not neglectedby a species-based conservation approach because thesame productivity that makes them valuable to us makesthem valuable to a set of other species. For example, 69%of the important bird areas (a species-based approach)identified in Europe are wetlands because wetlands are

crucial breeding, nonbreeding, and stopover habitat formany bird species (Heath & Evans 2000). In contrast toMolnar et al., our proposal does not suggest that conser-vation priorities should be defined based on the lengthof species lists, but rather that species information, es-pecially information about irreplaceability and threat,should underlie decisions on conservation priorities.

We were unable to confirm the origin of Molnar et al.’sstatement that Canada “is in the top 10% in the value ofits ecosystem services per unit area.” The table publishedby Sutton and Costanza (2002; http://www.du.edu/∼psutton/esiindexisee/Table2complete.xls) ranks Canadafiftieth in a list of countries sorted by decreasing valueof land ecosystem services product (ESP) per square kilo-meter (the first country is the Maldives, and the second isMonaco; it is not clear that ESP/km2 is a particularly use-ful measure). Interestingly, in terms of absolute land ESP,12 of the top 20 nations are megadiversity countries (Mit-termeier et al. 1997). Although both species endemismand absolute ESP are related to land area (explaining whycountries such as Brazil and Australia score high on bothlists), land area alone does not explain why a countrysuch as Papua New Guinea ranks twentieth in absoluteESP, whereas it ranks fifty-fifth in total area. A reasonablehypothesis might be that this wealth of ecosystem ser-vices is at least in part driven by its megadiversity.

Surrogacy

A key issue mentioned in all four responses is surrogacy.The surrogacy value of land types or ED in representingspecies diversity is frequently assumed (e.g., Powell etal. 2000; Pressey et al. 2000; Faith et al. 2001) but rarelytested. An adequate test of surrogacy of land types orED in representing species diversity must address the fol-lowing question: If one selects a set of protected areasbased on land types or ED, how well does such a networkperform in representing species diversity? Measuring per-formance requires comparing the levels of species repre-sentation obtained using the surrogate with what wouldbe obtained if the same area had been selected at randomand if the same area had been selected using the speciesinformation directly (Ferrier 2002). Without this informa-tion, the effectiveness of the surrogate being tested can-not be assessed. For example, Araujo et al. (2001) foundthat the selection of the set of 52 cells that maximizedED represented 83% of all bird species. This appears tobe evidence of high surrogacy, but they reported thatrandomly selected sets of 52 cells would likely performeven better (5% upper tail equals 86%) and that it wouldhave been possible to represent 100% of all bird speciestwice in the same area. These results underpin Araujo etal.’s (2001) conclusion that maximizing ED is not an ade-quate surrogacy strategy for maximizing species diversity.



Naturally, as Pressey points out, their test is valid only fora given region (Europe); a given set of species (plants,birds, mammals, and amphibians); a given scale (50 × 50km grid cells); and a given method, ED (Faith & Walker1996). Nevertheless, this remains the only study of whichwe are aware that meets the criteria for an adequate testof environmental surrogacy.

The analyses of Kiester et al. (1996), Ward et al. (1999),and Lombard et al. (2003), among others, present incon-clusive results on the issue. The first study, for example,found that five cells maximizing diversity of vegetationclasses (defined by dominant cover species) represented76% of 357 vertebrate species analyzed. Although Kiesteret al. indicate that a maximum of 93% of the species couldhave been represented in the same number of cells, theyprovide no information on how many species would havebeen represented by an average, randomly selected, setof five cells.

Pressey and Higgins et al. refer to several studies as pro-viding “more encouraging” evidence for the surrogacy ofland types (Wessels et al. 1999; Pharo & Beattie 2001;Cushman & McGarigal 2002; Mac Nally et al. 2002; Oliveret al. 2004; Su et al. 2004). In fact, all these are analysesof the association between land type and species assem-blages, not surrogacy tests. For example, Mac Nally et al.(2002) found positive associations between “biodiversitymanagement units” (based on biotic, climatic, and physi-cal characteristics) and assemblages of tree species, birds,and mammals. This analysis indicates that species distri-butions and abundances respond to the same variablesused to define the land types, which is valuable ecologi-cal information but does not prove or disprove that theseclasses are adequate surrogates for species in reserve plan-ning. Indeed, there are a number of ways in which landtypes may be distinct, with quite different meanings forconservation planning. For example, a land type with animpoverished or degraded species composition would beconsidered quite distinct (Table 1), but does not automat-ically become a biodiversity element that ought to be rep-resented in its own right, and it adds nothing in terms ofspecies representation to a reserve network that alreadyrepresents the more intact species assemblage. A landtype with a different species composition, on the otherhand, adds new species (maybe even endemic species)and deserves a different level of attention altogether fromthe perspective based on complementarity (Faith et al.2001).

We therefore retain our statement that there is currentlyno evidence supporting the use of land types as surro-gates for species representation, and recommend this asa priority line of research. Equally important is to furtherinvestigate the extent to which subsets of species are, orare not, good surrogates for the representation of otherspecies. Previous studies have yielded mixed results inthis regard: Dobson et al. (1997b) present evidence of sur-rogacy, whereas Virolainen et al. (2000) found surrogacy

Table 1. Hypothetical values of abundance for nine species (S1 to S9)across five land types (L1 to L5).∗

S1 S2 S3 S4 S5 S6 S7 S8 S9

L1 10 10 10 10 10 10 0 0 0L2 4 4 4 4 4 4 0 0 0L3 10 10 0 0 0 0 0 0 0L4 4 4 0 0 0 0 0 0 0L5 0 0 10 10 10 10 10 10 10

∗Each of the classes L2 to L5 was compared with L1 with theBray-Curtis dissimilarity index (also used in Wessels et al. [1999],Mac Nally et al. [2002], and Su et al. [2004] to distinguish betweenland types; this index varies between 0 [minimum dissimilarity]and 1 [maximum dissimilarity]). The L2 is a degraded version of L1

(same species but in lower abundances, Bray-Curtis = 0.43); L3 is animpoverished version of L1 ( fewer species but ones remaining occurin the same abundances, Bray-Curtis index = 0.50); L4 is a degradedand impoverished version of L1 (Bray-Curtis index = 0.76); and L5

includes four of the species in L1 (in the same abundances) but addsthree more (Bray-Curtis index = 0.38). Despite the fact that L5 hasspecies that do not occur in L1, it is considered by the Bray-Curtisindex as less dissimilar from L1 than the degraded or impoverishedland types.

value for some taxa but not for others. We are in full agree-ment with the cautionary comment by Higgins et al. that“including all known taxa [in defining conservation pri-orities] will not guarantee coverage of all biodiversity.” Atleast, however, it would guarantee coverage of all knowntaxa as conservation targets in their own right. It remainsto be seen whether using land types as surrogates wouldensure even that.

Difficulties Facing Species-Based Approaches

The responses to our essay highlight two serious diffi-culties facing species-based approaches: the limitationsof existing sampling and the cost of extending sampling.All the respondents concentrate on the former. We agreethat these are limitations but see them as incentives forfurther sampling efforts rather than reasons for neglect-ing to use the existing species data. The variability ofabundance across species ranges might mean that evenlow sampling (if well distributed spatially) can provideuseful information in identifying those areas most impor-tant for species persistence (Gaston & Rodrigues 2003).Pressey is right that false negatives plague species distri-bution data sets, but we contend that these are far lessserious than the false positives introduced by environ-mental data would be (Loiselle et al. 2003). Acceptingfalse negatives is precautionary in assuming that conser-vation efforts should be aimed at places where we knowthat species are present (even if more appropriate placesare subsequently found), whereas accepting false posi-tives could allow a species’ extinction because we as-sume that we are conserving it where it does not actually



occur. Biased taxonomic coverage is a bigger concern, butexciting current initiatives have revealed developments inthis area that would have seemed unattainable a few yearsago. Examples include the emergence of comprehensiveand conservation-relevant species data for plant groupssuch as conifers (Farjon & Page 1999) and cycads (Don-aldson 2003), terrestrial invertebrates as diverse as tigerbeetles (Cassola & Pearson 2000) and ants (Wilson 2003),and a range of coral reef taxa (Roberts et al. 2002). In ouropinion, the gap most urgently requiring attention is forfreshwater: global assessments of taxa such as fish anddragonflies are an extremely high priority.

Molnar et al. and Cowling et al. are also concernedabout the cost of increasing sampling coverage. Thesecosts are certainly high, but it is essential to see them interms of both their immediate value and of current expen-diture in other sectors. Investment in collection and com-pilation of species data yields great value for the moneyin conservation planning (e.g., Balmford & Gaston 1999).More generally, although we suspect that the estimate of$5 billion for describing all species is an underestimate,it is still cheap at the price. The human genome project,a task of comparable biological discovery, cost approxi-mately $5 billion (Wilson 2000) and has generated invalu-able information across topics as diverse as human evolu-tion, physiology, and health. Similarly, we see improveddata for conservation planning as just one of the manybenefits to humanity that would be delivered by the mas-sive increase in the sampling of biodiverse regions andtaxa. For example, such a project would achieve tremen-dous human welfare benefits in its delivery of education,training, and job opportunities for developing countries,where the majority of the species to be described andmapped are found. Finally, it has been pointed out thatfield research is often an effective conservation tool initself (e.g., Oates 1999), so in many cases the direct con-sequence of producing data on species becomes savingthem.

Several of the respondents believe that we claimed thatspecies data on their own are equivalent to conservationresults. We do not claim this: they are not. But these dataare a precondition of conservation because we cannotunderstand the relationships between the componentsof biodiversity without knowing what those componentsare. Collar (1997) summed up the issue succinctly: “Tax-onomy precedes conservation. This is as basic as to saythat language precedes education. Without the formalstructure of names and an agreed system of usage, therecan be no understanding of what exists to be conserved.”

Conclusion

We wholeheartedly agree with Cowling et al. that themain challenges to biodiversity conservation are in the im-

plementation of conservation plans (e.g., Pressey 1998).But even though we recognize the need to act as oppor-tunities arise, we are also wary of the options lost whenconservation planning is driven by opportunity (Pressey1994). Before systematic conservation planning, decisionmakers were most likely acting with the best possibleintentions when they created the current protected-areanetwork that is so biased toward the rocky and icy parts ofthe planet (Scott et al. 2001). We worry that there may besimilar risks in creating networks that are representativeof land types that may prove not to be good biodiversitysurrogates. Efforts to compile more and better speciesdata are indeed needed, not to replace the implementa-tion process but to support it. Above all, we cannot affordto ignore the species data that already exist.

Literature Cited

Allen, C. R., L. G. Pearlstine, and W. M. Kitchens. 2001. Modeling vi-able mammal populations in gap analyses. Biological Conservation99:135–144.

Araujo, M. B., C. J. Humphries, P. J. Densham, R. Lampinen, W. J. M.Hagemeijer, A. J. Mitchell-Jones, and J. P. Gasc. 2001. Would environ-mental diversity be a good surrogate for species diversity? Ecography24:103–110.

Balmford, A., and K. J. Gaston. 1999. Why biodiversity surveys are goodvalue. Nature 398:204–205.

Brooks, T. M., G. A. B. da Fonseca, and A. S. L. Rodrigues. 2004a. Pro-tected areas and species. Conservation Biology 18:616–618.

Brooks, T., et al. 2004b. Coverage of the existing global protected areasystem. BioScience:in press.

Cassola, F., and D. L. Pearson. 2000. Global patterns of tiger beetlespecies richness (Coleoptera: Cicindelidae): their use in conserva-tion planning. Biological Conservation 95:197–208.

Chape, S., S. Blyth, L. Fish, P. Fox, and M. Spalding. 2003. 2003United Nations list of protected areas. World Conservation Union,Gland, Switzerland, and World Conservation Monitoring Centre,Cambridge, United Kingdom.

Collar, N. J. 1997. Taxonomy and conservation: chicken and egg. Bulletinof the British Ornithologists’ Club 117:122–136.

Cowling, R. M., R. L. Pressey, M. Rouget, and A. T. Lombard. 2003. A con-servation plan for a global biodiversity hotspot—the Cape FloristicRegion, South Africa. Biological Conservation 112:191–216.

Cushman, S. A., and K. McGarigal. 2002. Hierarchical, multi-scale de-composition of species-environment relationships. Landscape Ecol-ogy 17:637–646.

D’Eon, R. G., S. M. Glenn, I. Parfitt, and M.-J. Fortin. 2002. Land-scape connectivity as a function of scale and organism vagility in areal forested landscape. Conservation Ecology 6:10. Available fromhttp://www.ecologyandsociety.org/vol6/iss2/art10/index.html (ac-cessed 28 August 2004).

Desmet, P. G., and R. M. Cowling. 2004. Using the species-area relation-ship to set baseline targets for conservation. Conservation Ecology:inpress.

Dobson, A. P., A. D. Bradshaw, and A. J. M. Baker. 1997a. Hopes forthe future: restoration ecology and conservation biology. Science277:515–522.

Dobson, A. P., J. P. Rodriguez, W. M. Roberts, and D. S. Wilcove. 1997b.Geographic distribution of endangered species in the United States.Science 275:550–553.

Donaldson, J. 2003. Cycads: status survey and conservation action plan.World Conservation Union, Gland, Switzerland.



Faith, D. P. 2003. Environmental diversity (ED) as surrogate informationfor species-level biodiversity. Ecography 26:374–379.

Faith, D. P., and P. A. Walker. 1996. Environmental diversity: on the best-possible use of surrogate data for assessing the relative biodiversityof sets of areas. Biodiversity and Conservation 5:399–415.

Faith, D. P., C. R. Margules, P. A. Walker, J. Stein, and G. Natera. 2001.Practical application of biodiversity surrogates and percentage tar-gets for conservation in Papua New Guinea. Pacific ConservationBiology 6:289–303.

Farjon, A., and C. N. Page. 1999. Conifers: status survey and conservationaction plan. World Conservation Union, Gland, Switzerland.

Ferrier, S. 2002. Mapping spatial pattern in biodiversity for regional con-servation planning: where to from here? Systematic Biology 51:331–363.

Ferrier, S., M. Drielsma, G. Manion, and G. Watson. 2002. Extendedstatistical approaches to modelling spatial pattern in biodiversity innortheast New South Wales. II. Community-level modelling. Biodi-versity and Conservation 11:2309–2338.

Fjeldsa, J., E. Lambin, and B. Mertens. 1999. Correlation between en-demism and local ecoclimatic stability documented by comparingAndean bird distributions and remotely sensed land surface data.Ecography 22:63–78.

Gaston, K. J., and A. S. L. Rodrigues. 2003. Reserve selection in regionswith poor biological data. Conservation Biology 17:188–195.

Heath, M. F., and M. I. Evans. 2000. Important bird areas in Europe:priority sites for conservation. BirdLife International, Cambridge,United Kingdom.

Keeley, J. E., and C. J. Fotheringham. 2001. Historic fire regime inSouthern California shrublands. Conservation Biology 15:1536–1548.

Kiester, A. R., J. M. Scott, B. Csuti, R. F. Noss, B. Butterfield, K. Sahr,and D. White. 1996. Conservation prioritization using GAP data.Conservation Biology 10:1332–1342.

Loiselle, B. A., C. A. Howell, C. H. Graham, J. M. Goerck, T. Brooks, K.G. Smith, and P. H. Williams. 2003. Avoiding pitfalls of using speciesdistribution models in conservation planning. Conservation Biology17:1591–1600.

Lombard, A. T., R. M. Cowling, R. L. Pressey, and A. G. Rebelo. 2003.Effectiveness of land classes as surrogates for species in conserva-tion planning for the Cape Floristic Region. Biological Conservation112:45–62.

Mac Nally, R., A. F. Bennett, G. W. Brown, L. F. Lumsden, A. Yen, S. Hink-ley, P. Lillywhite, and D. A. Ward. 2002. How well do ecosystem-basedplanning units represent different components of biodiversity? Eco-logical Applications 12:900–912.

Mittermeier, R. A., P. Robles Gil, and C. G. Mittermeier, editors. 1997.Megadiversity: Earth’s biologically wealthiest nations. CEMEX, Mex-ico City.

Noss, R. F. 1987. From plant communities to landscapes in conservationinventories: a look at The Nature Conservancy (USA). BiologicalConservation 41:11–37.

Oates, J. F. 1999. Myth and reality—how conservation strategies arefailing in West Africa. University of California Press, Berkeley.

Oliver, I., et al. 2004. Land systems as surrogates for biodiversity inconservation planning. Ecological Applications 14:485–503.

Olson, D. M. 2001. Terrestrial ecoregions of the world: a new map oflife on Earth. BioScience 51:933–938.

Pharo, E. J., and A. J. Beattie. 2001. Management forest types as a sur-rogate for vascular plant, bryophyte and lichen diversity. AustralianJournal of Botany 49:23–30.

Powell, G. V. N., J. Barborak, and M. Rodriguez S. 2000. Assessing rep-resentativeness of protected natural areas in Costa Rica for conserv-

ing biodiversity; a preliminary gap analysis. Biological Conservation93:35–41.

Powell, G. V. N., and R. Bjork. 1995. Implications of intratropicalmigration on reserve design—a case-study using Pharomachrusmocinno. Conservation Biology 9:354–362.

Pressey, R. L. 1994. Ad hoc reservations—forward or backward stepsin developing representative reserve systems. Conservation Biology8:662–668.

Pressey, R. L. 1998. Algorithms, politics and timber: an example of therole of science in a public, political negotiation process over newconservation areas in production forests. Pages 73–87 in R. T. Willsand R. J. Hobbs, editors. Ecology for everyone: communicating ecol-ogy to scientists, the public and politicians. Surrey Beatty & Sons,Sydney.

Pressey, R. L., T. C. Hager, K. M. Ryan, J. Schwarz, S. Wall, S. Ferrier, andP. M. Creaser. 2000. Using abiotic data for conservation assessmentsover extensive regions: quantitative methods applied across NewSouth Wales, Australia. Biological Conservation 96:55–82.

Roberts, C. M., et al. 2002. Marine biodiversity hotspots and conserva-tion priorities for tropical reefs. Science 295:1280–1284.

Rouget, M., R. M. Cowling, R. L. Pressey, and D. M. Richardson. 2003.Identifying spatial components of ecological and evolutionary pro-cesses for regional conservation planning in the Cape Floristic Re-gion, South Africa. Diversity and Distributions 9:191–210.

Scott, J. M., F. W. Davis, R. G. McGhie, R. G. Wright, C. Groves, and J.Estes. 2001. Nature reserves: do they capture the full range of Amer-ica’s biological diversity? Ecological Applications 11:999–1007.

Soule, M. E., and M. A. Sanjayan. 1998. Conservation targets: do theyhelp? Science 279:2060–2061.

Strittholt, J. R., and R. E. J. Boerner. 1995. Applying biodiversity gap anal-ysis in a regional nature reserve design for the edge of Appalachia,Ohio (USA). Conservation Biology 9:1492–1505.

Su, J. C., D. M. Debinski, M. E. Jakubauskas, and K. Kindscher. 2004. Be-yond species richness: community similarity as a measure of cross-taxon congruence for coarse-filter conservation. Conservation Biol-ogy 18:167–173.

Sutton, P. C., and R. Costanza. 2002. Global estimates of market and non-market values derived from nighttime satellite imagery, land cover,and ecosystem service valuation. Ecological Economics 41:509–527.

Udvardy, M. D. F. 1975. A classification of the biogeographical provincesof the world. Occasional paper 18. International Union of Conser-vation of Nature and Natural Resources, Morges, Switzerland.

Virolainen, K. M., P. Ahlroth, E. Hyvarinen, E. Korkeamaki, J. Mattila,J. Paivinen, T. Rintala, T. Suomi, and J. Suhonen. 2000. Hot spots,indicator taxa, complementarity and optimal networks of taiga. Pro-ceedings of the Royal Society of London. Series B, Biological Sciences267:1143–1147.

Ward, T. J., M. A. Vanderklift, A. O. Nicholls, and R. A. Kenchington. 1999.Selecting marine reserves using habitats and species assemblages assurrogates for biological diversity. Ecological Applications 9:691–698.

Wessels, K. J., S. Freitag, and A. S. van Jaarsveld. 1999. The use of landfacets as biodiversity surrogates during reserve selection at a localscale. Biological Conservation 89:21–38.

Wilson, E. O. 2000. A global biodiversity map. Science 289:2279.Wilson, E. O. 2003. Pheidole in the New World: a dominant, hyperdi-

verse ant genus. Harvard University Press, Boston.World Database on Protected Areas (WDPA) Consortium. 2003. 2003

World Database on protected areas. The World Conservation Union–World Commission on Protected Areas (IUCN–WCPA) and UnitedNations Environment Program–World Conservation Monitoring Cen-ter (UNEP–WCMC), Washington, D.C.


species, data, and conservation planning

Documents