DOI: 10.1126/science.1122667, 1261 (2006);314 Science

K. J. Willis and H. J. B. BirksBiodiversity ConservationWhat Is Natural? The Need for a Long-Term Perspective in

This copy is for your personal, non-commercial use only.

clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others

  here.following the guidelines

can be obtained byPermission to republish or repurpose articles or portions of articles

  ): December 9, 2013 www.sciencemag.org (this information is current as of

The following resources related to this article are available online at

http://www.sciencemag.org/content/314/5803/1261.full.htmlversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org/content/314/5803/1261.full.html#ref-list-1, 4 of which can be accessed free:cites 46 articlesThis article

63 article(s) on the ISI Web of Sciencecited by This article has been

http://www.sciencemag.org/content/314/5803/1261.full.html#related-urls23 articles hosted by HighWire Press; see:cited by This article has been

http://www.sciencemag.org/cgi/collection/paleoPaleontology

subject collections:This article appears in the following

registered trademark of AAAS. is aScience2006 by the American Association for the Advancement of Science; all rights reserved. The title

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience

on

Dec

embe

r 9,

201

3w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

o

n D

ecem

ber

9, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Dec

embe

r 9,

201

3w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

o

n D

ecem

ber

9, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Dec

embe

r 9,

201

3w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

o

n D

ecem

ber

9, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

What Is Natural? The Need fora Long-Term Perspectivein Biodiversity ConservationK. J. Willis1* and H. J. B. Birks2

Ecosystems change in response to factors such as climate variability, invasions, and wildfires. Mostrecords used to assess such change are based on short-term ecological data or satellite imageryspanning only a few decades. In many instances it is impossible to disentangle natural variabilityfrom other, potentially significant trends in these records, partly because of their short time scale.We summarize recent studies that show how paleoecological records can be used to provide alonger temporal perspective to address specific conservation issues relating to biological invasions,wildfires, climate change, and determination of natural variability. The use of such records canreduce much of the uncertainty surrounding the question of what is “natural” and thereby start toprovide important guidance for long-term management and conservation.

Paleoecological records (e.g., fossil pollen,seeds and fruits, animal remains, treerings, charcoal) spanning tens to millions

of years provide a valuable long-term perspec-tive on the dynamics of contemporary ecologi-cal systems (1). Such studies are increasinglybecoming part of community and landscapeecological research (2). In contrast, conservation-related research largely ignores paleoecologicalrecords. For example, there are no temporalrecords spanning more than 50 years included inany of the key biodiversity assessments pub-lished over the past 7 years (3). Paleoecologicalrecords have been considered too descriptiveand imprecise, and therefore of little relevance tothe actual processes of conservation and man-agement. Such criticisms may have been valid30 years ago, but there is now a wealth ofinformation in paleoecological records pro-viding detailed spatial and temporal resolutions(1, 4–7) that match in detail most recordscurrently used in conservation research.

The potential of paleoecological records inconservation biology has been highlightedseveral times, including their application to bio-diversity maintenance, ecosystem naturalness,conservation evaluation, habitat alteration, chang-ing disturbance regimes, and invasions [e.g.,(8–14)]. Conservation of biodiversity in achanging climate (15) and the relevant tempo-ral and spatial scales for ecological restoration(16) have also been considered to warrant alonger-term temporal perspective. Most of

these studies are descriptive and provide littlepractical application. A number of recentapplied paleoecological studies, however, havebegun to provide direct management infor-mation for biodiversity conservation at local,regional, and global scales. These includerecommendations relating to biological inva-sions, wildfires, climate change, and conserva-tion management within thresholds of naturalvariability. The overriding message from thesestudies is that such temporal perspectives areessential for meaningful modeling, prediction,and development of conservation strategies inour rapidly changing Earth.

Biological InvasionsBiological invasions are of critical concern toconservation organizations worldwide, with ageneral perception that many invasives areresponsible for widespread community changeand even extinctions (17). At the Rio Earth Sum-mit Convention on Biological Diversity in 1992,for example, binding signatories were made “toprevent the introduction of, control or eradicatethose alien species which threaten ecosystems,habitats or species” (18). However, biologicalinvasions are complex. Some regions are moreprone to invasion, certain species are moresuccessful invaders than others, and sometimesit is even unclear whether a species is alien ornative. The importance of the historical record inimproving our ability to predict the outcome of

non-native introductions has been acknowledged[e.g., (13, 14)], but several recent paleoecologicalstudies provide direct guidelines for the identifi-cation and management of invasives.

The distinction between what is native andwhat is not is often unclear. A species is usuallyclassified as either native or exotic according towhether it is located in its presumed area ofevolutionary origin and/or whether human agen-cy is responsible for its current distribution. Inthe absence of a temporal record to assess a spe-cies history, the distinction can often becomeblurred (16). For example, in a reexamination ofthe British flora, several discrepancies betweenpublished records were found, with the samespecies being classified as “alien” or “native” de-pending on personal interpretation (19) (Table 1).There is also the question of how far back onetakes “human” activity in determining whether aspecies is native or alien.When using evidence offirst occurrences of species based on paleoeco-logical records to reassess “doubtful natives” inthe British flora, Preston et al. (19) determinedthat at least 157 plant species had been intro-duced to Britain by humans, intentionally orunintentionally, from the start of the Neolithicperiod (about 4000 years ago) to 500 years ago,yet the terminology used for their classificationaccording to different floras is highly variable(Table 1). Preston et al. proposed that suchspecies should be classified separately as“archaeophytes.” They acknowledged, however,that this causes problems with their conservationstatus because this “non-native” label excludesthem from the British Red Data Book ofthreatened or near-extinct species, and auto-matically deems them to be of lower conserva-tion value—even though some are in seriousdecline and have been part of the British flora forat least 500 years.

A similarly conflicting conservationmessagewas reached in an applied paleoecological studyon the origin of an invasive form of the com-mon reed (Phragmites australis) in the marshesof the inland wetlands of Lake Superior, NorthAmerica (20) (Fig. 1). Over recent decades,P. australis populations have expanded rapidlythroughout the coastal wetlands of NorthAmerica, creating substantial changes in com-munity structure and composition. In thisstudy, paleoecological and genetic analyseswere used to determine when the common reedbecame established in this region and whetherthe source was from a native or non-native

REVIEW

1Long-Term Ecology Laboratory, Oxford University Centrefor the Environment, South Parks Road, Oxford OX1 3QY,UK. 2Ecological and Environmental Change ResearchGroup, Department of Biology, and Bjerknes Centre forClimate Research, University of Bergen, N-5007 Bergen,Norway, and Environmental Change Research Centre,University College London, London WC1E 6BT, UK.

*To whom correspondence should be addressed. E-mail:kathy.willis@ouce.ox.ac.uk

Table 1. Classification of 157 species of British plants that were probably introduced more than500 years ago (archaeophytes) according to three published floras (54–56).

Publishedflora Native Doubtful

native Introduced Probablyintroduced

Uncertain oruntreated Total

Dunn, 1905 (54) 31 — 103 — 23 157Clapham et al., 1952 (55) 85 19 30 10 13 157Stace, 1991 (56) 77 27 39 14 0 157

www.sciencemag.org SCIENCE VOL 314 24 NOVEMBER 2006 1261

population. A 4000-year paleoecological recordindicated that reeds were not part of the localflora until very recently (several decades), andthat their recent expansion was probably linkedto changes in water levels in the wetlands andhuman-induced changes to the landscape. Thesimple conservation message from this study istherefore to eradicate or control reed populations,because the expansion was recent and is likely tocause serious changes to the wetlands communi-ty. However, genetic data from these reed pop-ulations add another level of complexity becausethey indicate that the reeds are a native variety,raising the question of whether this is an exotic ornatural invasion.

Oceanic islands are particularly liable to in-vasions, and it is often difficult to assess whetherparticular species are native or introduced. Theinvasive ornamental club moss Selaginellakraussiana, for example, is widely planted in theNeotropics, southern United States, Australasia,and western Europe. It is common on the AzoresIslands in a range of habitats, but is it native there?Paleoecological records (21) (Fig. 2) clearly showthat S. kraussiana had been present on Flores inthe Azores for several thousand years beforePortuguese discovery and Flemish settlement inthe 15th century, thereby establishing beyonddoubt its native status on Flores Island. Paleoecol-ogy again helped here to resolve a question inbiodiversity conservation.

Another key question is whether invasivespecies are the triggering mechanism for eco-system change, or merely opportunists takingadvantage of environmental change caused byother biotic or abiotic factors? Also, are thereparticular factors that make a habitat moresusceptible to invasion? A study of the coloniza-tion and spread of invasive shrubs in nativeshrublands and early successional forests in thenortheastern United States, for example, foundthat prevalence of agricultural fields (historic andpresent-day) was the most influential factoraffecting the colonization and spread of invasiveshrubs (22). These native shrublands and earlysuccessional forests currently have high conser-vation status because of their diversity of ter-restrial vertebrates. By considering the temporaldimension, the authors argue that it should bepossible to identify those early successionalhabitats that may be especially prone to exoticinvasion and ought to be of higher conservation

priority. This study used only 40 years of temporaldata, but studies incorporating longer temporaltime scales have also illustrated persistent legaciesof ancient land use that may influence thevulnerability of a site to invasion (12), includingdifferences in soil pH, C, and N values. Theseimprints can last for decades to centuries. Theidentification of former land use by paleoecolog-ical records can thus be a tool for understand-ing and determining a habitat’s vulnerability toinvasion.

Introductions of non-native species often ap-pear to fail a number of times before they even-tually succeed; therefore, there is a lag betweenfirst colonization and population expansion ofthe invasive species (23). The reasons for re-sistance to invasion are complex and can have asmuch to do with environmental variables andextreme events as with demographic and bioticfactors (6, 7). A study using paleoecologicalrecords has shown that consideration should begiven to biological inertia (24), whereby a nativecommunity occurs where environmental con-ditions are no longer optimal but will remain insitu without any triggering mechanism (e.g.,hurricanes, windthrow, etc.) to “remove” thisresident population. Thus, the life history char-acteristics and biology of the resident species,and not the properties of the invading species,are responsible for invasion lags. This phenom-enon is particularly apparent in forest ecosys-tems. In many current old-growth forests inwestern North America, paleoecological studieshave shown that these stands were establishedduring the cooler and moister climate of theLittle Ice Age (about 650 to 150 years ago) andtherefore reflect recruitment responses to former

climate conditions (25). Suchinformation about ecologicallegacies (1) is directly relevantto conservation because suchforests may be at a criticalthreshold and may be particu-larly vulnerable to invasion aftera disturbance event, eithernatural or human-induced.

WildfiresWildfires have been importantin shaping the structure andfunction of fire-prone com-munities throughout Earth’shistory (26). Of particular con-cern to conservationists, howev-er, are changes in the frequency,severity, and extent of burningfrom those perceived as the“norm” (27). What processesare driving this change (humanor climate)? How will it affectthe composition of plants andanimals in ecosystems, in partic-ular those already identified asvulnerable? And are there par-ticular management techniques

Fig. 1. Native (?) common reed (Phragmitesaustralis) growing in Bark Bay Slough on LakeSuperior, North America [photo: E. A. Lynch];rattlesnake (Crotalus mitchellii stephensi) in thewarm desert of western North America [photo:Blake L. Thompson]; wood grouse or westerncapercaillie (Tetrao urogallus) in the CantabrianMountains, northern Spain [photo: E. Menoni].

20340

320

300

280

260

240

220

200

180100 ± 70 BP-

505 ± 55 BP-

705 ± 35 BP-

895 ± 35 BP-

2435 ± 41 BP-

160

140

120

40 60 80 1 2 3100

cm14C datesModern-

Humanimpact

Pristine

Tree

s

Shrubs

Herbs

Selagin

ella

krau

ssian

a

Fig. 2. Simplified pollen diagram from Lagoa Rasa, Flores Is-land, Azores, for the past 3000 years showing the percentage oftree, shrub, and herb pollen and of Selaginella kraussianaspores before and after human occupation of the island.[Modified from (21)]

24 NOVEMBER 2006 VOL 314 SCIENCE www.sciencemag.org1262

REVIEW

that can be implemented to alter fire regimes?Fundamental to these questions is establishingthe natural variability of wildfires so that thiscan be used as a benchmark against which toevaluate contemporary conditions and futurealternatives (28). Assessments based on short-term records (<50 years) can easily lead tomisguided management plans (29).

Although climate change and human activ-ities have long been acknowledged as driversof wildfires, results from recent paleoecologicalstudies show that these relationships arecomplex. For example, although it is notunreasonable to assume that an increase inaridity would result in more fires, several studiesindicate otherwise. In the Alaskan boreal forest,fires occurred more frequently under wetterclimatic conditions (30). A similar conclusionwas reached in a paleoecological study of firecycles in the Northern Great Plainsgrasslands of North America (31).Here, the highest charcoal fluxoccurred during past moist inter-vals when grass cover was exten-sive and fuel loads were high.Shifts in fuel quantity and qualitycan cause changes in fire regimes.Both studies show that there is acomplex climate-fuel-fire relation-ship determining the variability ofwildfires (32). Such studies (33)should be taken into accountwhen predicting future ecosystemchange within climate changeconservation strategies.

Prehistoric and historic human-induced wildfires are often as-sumed to have caused changes inecosystem structure and degrada-tion, especially in tropical forestswhere natural fires are rare andtend to be limited in extent. Man-agement plans to control such firesare usually implemented, however,without paleoecological evidenceto confirm such an assumption. One suchexample is in the tropical dry forests of thesouthern Ratanakiri Province, northeasternCambodia (34). Here, regional conservationpolicy is based on the premise that burning byhumans has degraded the dense forests andresulted in the present open forest–savannamosaic. However, a paleoecological studyshows that present-day fire activity is nowlower than it has been for the past 9300 years(Fig. 3). Rather, the forest-savanna shift isprobably a consequence of monsoonal activity,and the high-frequency but low-intensity firescaused by humans may, in fact, conserve forestcover. In this case, the current conservation man-agement plan is clearly at odds with evidencefrom the paleoecological record.

Interesting conclusions have also emergedfrom studies examining ecosystem compositionin response to fire regimes. One of the main

findings of the work on the North Americangrasslands described above, for example, is thatfire is not necessarily a universal feature of thisecosystem but oscillates through time with cli-mate (31). The impact of such variability inburning regimes through time on ecosystemcomposition can have conservation implica-tions. This is well illustrated in a study on thelong-term record of fire and open canopy in aforest in southern Sweden that contains an ex-ceptionally large number of endangered speciesof beetle (35). Of the 105 beetle species recordedat this site living on or in rotting wood that are inthe Swedish Red Data Book of threatened ornear-extinct species, many are associated withopen forest, forest fires, or structures createdby fire. Yet a site-scale paleoecological studyindicates that the forest is more closed todaythan at any time in the past 2500 years; although

there had been a significant amount of burning inthe past, there has been a large reduction in firesover the past 200 years. The authors concludedthat openness of the site in the past as aconsequence of burning is an important ex-planation for the high conservation value of thesite today (35). To conserve the diverse beetleassemblage of this site, they suggested that openforest conditions needed to be restored and thatprescribed burns would be the most appropriateway to achieve this.

Climate VariabilityMost conservation organizations have devel-oped climate change conservation strategies [asdescribed in (36)] designed to conserve bio-diversity in a changing climate. Two questionscentral to current conservation strategies arise.Where will biota move to in response to futureclimate change? Which species and regions are

most at risk from future climate change?Underlying these questions are key managementand planning issues—for example, ensuring thatreserve boundaries allow for potential species-range shifts (37) and that the species and regionsmost at risk are identified and protected (38).

In the evaluation of predictive models todetermine the biogeographic effects of climatechange, several studies have used paleoecolog-ical records for backward prediction (hindcast-ing) to assess errors potentially inherent inspecies-envelope bioclimatic modeling (39).This involves running models for past intervalsof time, using present-day species data but mod-eling the species’ response to climate changeagainst paleoclimatic data as opposed to present-day climatic data. The predicted distributions arethen tested against the distribution of the speciesapparent in the fossil record for the time interval

covered by the paleoclimatic datato assess model robustness (40).In a study of 23 extant mammalspecies in the United States (39),for example, an ecological nichemodel was run backward for thetime interval of the Last FullGlacial (14,500 to 20,500 yearsbefore the present) and predicteddistributions were compared toactual distribution records obtainedfrom the FAUNMAP fossil data-base (41). The model was also runin reverse (i.e., using fossil dataand paleoclimatic data to predictpresent distributions) and similarcomparisons were made. Resultsindicated that for nine species themodel was able to predict accu-rately the Pleistocene distributionsfrom the present-day data, and viceversa. Not only did this confirmthat the model was robust for thesespecies, it also provided a test forthe underlying assumption of thesemodels that the species’ ecological

niche characteristics have remained constantthrough time. A similar pattern was recentlyfound for several North American plant species(42). The remaining species, however, either hadsignificant predictions only one way but not theother (nine species) or were not significant ineither direction (five species).

The question of why some species’ distribu-tions cannot be accurately predicted by species-climate modeling can also be answered, at leastfor some species, from paleoecological studies.A study of the spread of Picea abies (spruce)and Fagus sylvatica (beech) over the past 4000years in southern Scandinavia, for example,showed that at the local-stand scale the spreadof Picea closely tracked the changing area ofsuitable regional climate, whereas the spread ofFagus was more directly linked to anthropogen-ic activities and disturbance by fire (43). Thus,caution may be needed in using the results of

Age (years BP)

300

250

200

150

100

50

00 1000 2000 3000 4000 5000 6000 7000 8000 9000 9300

Ch

arco

al c

on

cen

trat

ion

(m

m3

ml-1

)

Fig. 3. Reconstructed fire regimes in northeastern Cambodian monsoonalforests over the past 9300 years, using microfossil charcoal concentrationfrom a dated sedimentary sequence (34). The record indicates that present-day charcoal input is the lowest of the entire period. Conservation policiesthat suggest that human burning has increased and resulted in the openforest–savanna mosaic in this region are clearly misguided, as are man-agement recommendations for fire suppression.

www.sciencemag.org SCIENCE VOL 314 24 NOVEMBER 2006 1263

REVIEW

predictive species-envelope models in conser-vation planning, because the distributions ofsome species today or in the past may be poorlypredicted.

Bioclimatic models are particularly relevantto conservationists in determining and under-standing the dynamics of the leading edge ofspecies-range margins and the potential spacethat will be needed for future reserve boundaries(40). There is also a considerable literature onmodeling to determine which species will goextinct [e.g., (38)]. However, there are fewstudies of the likely fate of rear-edge popula-tions, that is, the source populations from whichthe leading-edge populations migrate (Fig. 4)(44). A key conservation objective should be thepreservation of conditions necessary for specia-tion (45). Evidence from paleoecological andgenetic records indicates that the maintenance ofpopulations in these rear-edge regions could, infact, be critical for conservation of long-termgenetic diversity (44). Evidence also suggeststhat these regions tend to be where plants andanimals were geographically and geneticallyisolated in refugia during the cold stages of thePleistocene. In Europe, for example, refugiallocalities have been recognized in Iberia, theBalkans, and Italy and in mountain ranges suchas the Carpathians (46–48).

With the use of a combination of paleoeco-logical and genetic evidence, other such regionshave been identified, and this information isfeeding into conservation policy. For example,in a study on Eurasian populations of westerncapercaillie (grouse)—a keystone species ofPalearctic boreal and high-altitude coniferousforests (49)—a combined genetic and temporalrecord enabled the identification of two regionsthat should be classified as ecologically signif-icant units (ESUs) because of the geneticdistinctiveness of the populations within themfrom the rest of Europe. The distinctiveness ofthe populations in these ESUs, located in thePyrenees and Cantabrian Mountains (Fig. 1), isalmost certainly related to their Pleistocenerefugial isolation. Similar historically relatedgenetic patterns have been identified in thesetwo regions for a number of plants and animals,and this knowledge is now leading to interna-tional recognition of the conservation impor-tance of these areas (49).

In the United States, a similar approach usinga molecular and deep-time historical perspectiveas a primary mechanism to frame biodiversityreserves (50) has been applied to a number ofgroups of plants and animals. Distinctivepatterns of genetic diversity related to geologicalevents in deep time (Pliocene/Miocene) and toPleistocene refugial isolation have been demon-strated, for example, in four rattlesnake species(Fig. 1; genus Crotalus) in the warm deserts ofwestern North America (50). Here it is arguedthat an approach that seeks to understand thecausation of genetic patterns would be moreeffective in encapsulating biodiversity than

current measures (based on the use of geologicalfeatures as a surrogate for diversity) and thatsuch studies should be routinely used in de-veloping integrated regional conservation poli-cies (50).

Determination of Thresholds WithinNatural VariabilityVariability through time is an inherent part ofecosystem behavior. It is thus essential to in-corporate variability into management policies.To do this reliably in our rapidly changing worldrequires answers to several questions. What arethe baseline or “reference” conditions before re-cent times? What is the range of natural var-iability? Under what conditions do negativeimpacts become apparent? How can thresholdsbe determined beyond which specific manage-ment plans should be implemented?

Gillson and Duffin (51) used paleoecologicalrecords from savannas in Kruger National Park,South Africa, to determine the natural variabilityof woody vegetation cover during the past 5000years. They used this information to addresswhether woody cover has decreased below 80%of its “highest ever value”—a threshold set byecosystem managers to define the upper andlower level of accepted variation in this eco-system. Paleoecological results indicated thatduring the past 5000 years, the estimated woodyvegetation cover had remained at about 20%of its “highest ever value,” and therefore thatmanagement intervention in this part of thepark is unnecessary at present.

Other examples where paleoecologicalrecords have been used to identify where naturalthresholds have recently been exceeded includeriver ecosystems in Australia (52) and Colorado(53). The large deep billabongs in the middlereaches of the Murray River, Australia, for ex-ample, do not currently support submerged mac-

rophyte beds. Yet paleoecological analysesindicate that these were an important part ofthe ecosystem before the arrival of Europeans(52). In the Colorado delta ecosystem (53), pa-leoecological studies suggest that there has beena decline of up to 94% of shelly benthic mac-roinvertebrates over the past 75 years. This de-cline is probably associated with a reduction offresh water and nutrients resulting from thediversion of the Colorado River by dams andirrigation projects. Both studies provide quan-titative assessments of the relative health (4)of these river ecosystems and indicate thresh-olds that have been exceeded—informationthat is critical to their restoration and long-termconservation.

ConclusionsConservation biology and nature managementare primarily concerned with the present and in-creasingly with the future. Paleoecology primar-ily considers the past but can provide a historicalperspective to the present (1). It can also con-tribute to key questions in conservation and man-agement such as habitat naturalness, biologicalinvasions, disturbance regimes, natural variabil-ity, and ecosystem health. With increasingamounts of paleoecological data of a high spatialand/or temporal resolution (4, 5), there is po-tential for synergy between conservation biologyand paleoecology. There are, however, severalresearch needs and challenges that need to bemetbefore an effective synergy can fully develop.These include the following:

1) Paleoecological studies in biodiversityhotspots with a high density of species. At presentthere are few studies from these critical areas.

2) Improved taxonomic resolution of thefossils found, because improved resolution in-variably enhances the biological value of fossilrecords (5, 21).

Continuousrange

Intermixingzone

Leading edge Dominant processes

Rear edge

Refugial isolationPopulation stabilityGenetic driftLocal adaptation

Long-distance dispersalFounder eventsPopulation events

Lineage mixing

Between-populationgenetic diversity

Population age

Within-population genetic diversity

Fig. 4. Schematic representation of the leading and rear-edge populations in response to climatechange (44). Paleoecological and genetic evidence suggests that the rear-edge populations may beextremely important in the conservation of long-term genetic diversity and that more attentionmust be given to modeling the impacts of future climate change on these populations and theirprotection.

24 NOVEMBER 2006 VOL 314 SCIENCE www.sciencemag.org1264

REVIEW

3) Assessing terrestrial paleoecological datain terms of “ecosystem health” to provide anecosystem’s health history (4). Some taxa inpaleoecological records are “indicators” of par-ticular ecological conditions that can provideuseful “symptoms” about the ecosystem’s health.Paleolimnologists (4) have effectively applied theconcept of ecosystem health to lakes in relationto critical loads of pollutants. The same conceptcould be usefully applied to forests, heathlands,grasslands, wetlands, tundra, and savannas.

4) Greater discussion and collaboration be-tween paleoecologists and conservation biolo-gists, so that the most pertinent and urgentresearch questions are addressed together andthe most relevant paleoecological data are col-lected at the spatial and temporal scales of directconcern in conservation.

Paleoecology provides a historical perspec-tive that can help put present and future con-servation and management policies into context.The time is ripe for the two disciplines to workmore closely together and to develop a commonagenda for biodiversity conservation.

References and Notes1. National Research Council, The Geological Record of

Ecological Dynamics: Understanding the Biotic Effects ofFuture Environmental Change (National Academies Press,Washington, DC, 2005).

2. L. Graümlich, S. Sugita, L. B. Brubaker, V. M. Card,Ecology 86, 1667 (2005).

3. K. J. Willis, L. Gillson, T. Brncic, B. Figueroa-Rangel,Trends Ecol. Evol. 20, 107 (2005).

4. J. P. Smol, Pollution of Lakes and Rivers—APaleoenvironmental Approach (Arnold, London, 2002).

5. H. J. B. Birks, Grana 44, 1 (2005).6. M. E. Lyford, S. T. Jackson, J. L. Betancourt, S. T. Gray,

Ecol. Monogr. 73, 567 (2003).

7. S. T. Gray, J. L. Betancourt, S. T. Jackson, R. G. Eddy,Ecology 87, 1124 (2006).

8. H. J. B. Birks, J. Veg. Sci. 7, 89 (1996).9. P. B. Landres, P. Morgan, F. J. Swanson, Ecol. Appl. 9,

1179 (1999).10. T. W. Swetnam, C. D. Allen, J. L. Betancourt, Ecol. Appl.

9, 1189 (1999).11. L. Gillson, K. J. Willis, Ecol. Lett. 7, 990 (2004).12. D. R. Foster et al., Bioscience 53, 77 (2003).13. G. M. MacDonald, Adv. Ecol. Res. 24, 67 (1993).14. S. T. Jackson, in Assessment and Management of Plant

Invasions, J. O. Luken, J. W. Thieret, Eds. (Springer-Verlag, New York, 1997), p. 37.

15. L. Hannah et al., Conserv. Biol. 16, 264 (2002).16. J. B. Calicott, J. Biosci. 27, 409 (2002).17. J. Gurevitch, D. K. Padilla, Trends Ecol. Evol. 19, 470

(2004).18. S. Henderson, T. E. Dawson, R. J. Whittaker, Prog. Phys.

Geogr. 30, 25 (2006).19. C. D. Preston, D. A. Pearman, A. R. Hall, Bot. J. Linn. Soc.

145, 257 (2004).20. E. A. Lynch, K. Saltonsall, Wetlands 22, 637 (2002).21. J. F. N. Van Leeuwen et al., Neobiota 6, 27 (2005).22. V. S. Johnson, J. A. Litvaitis, T. D. Lee, S. D. Frey,

For. Ecol. Manage. 228, 124 (2006).23. D. F. Sax, J. H. Brown, Glob. Ecol. Biogeogr. 9, 363

(2000).24. B. Von Holle, H. R. Delcourt, D. Simberloff, J. Veg. Sci.

14, 425 (2003).25. C. I. Millar, W. B. Woolfendon, Ecol. Appl. 9, 1207

(1999).26. W. J. Bond, J. E. Keeley, Trends Ecol. Evol. 20, 387 (2005).27. D. McKenzie, Z. Gedalof, D. L. Peterson, P. Mote, Conserv.

Biol. 18, 890 (2004).28. P. Z. Fulé, W. W. Covington, M. M. Moore, Ecol. Appl. 7,

895 (1997).29. C. Whitlock, Nature 432, 28 (2004).30. J. A. Lynch, J. L. Hollis, F. Sheng Hu, J. Ecol. 92, 477 (2004).31. K. J. Brown et al., Proc. Natl. Acad. Sci. U.S.A. 102, 8865

(2005).32. J. S. Clark, P. D. Royall, C. Chumbley, Ecology 77, 2148

(1996).33. J. S. Clark, E. C. Grimm, J. Lynch, P. G. Mueller, Ecology

82, 620 (2001).34. A. L. Maxwell, J. Biogeogr. 31, 225 (2004).

35. M. Lindbladh, M. Niklasson, S. G. Nilsson, Biol. Conserv.114, 231 (2003).

36. L. Hannah, G. F. Midgley, D. Millar, Glob. Ecol. Biogeogr.11, 485 (2002).

37. M. B. Araujo, M. Cabeza, W. Thuiller, L. Hannah, P. H.Williams, Glob. Change Biol. 10, 1618 (2004).

38. C. D. Thomas et al., Nature 427, 145 (2004).39. E. Martínez-Meyer, A. Townsend Peterson, W. W. Hargrove,

Glob. Ecol. Biogeogr. 13, 305 (2004).40. M. B. Araújo, C. Rahbek, Science 313, 1396 (2006).41. FAUNMAP Working Group, Illinois State Mus. Sci. Pap. 25

(1994).42. E. Martinez-Meyer, A. Townsend Peterson, J. Biogeogr.

33, 1779 (2006).43. R. H. W. Bradshaw, M. Lindbladh, Ecology 86, 1679

(2005).44. A. Hampe, R. J. Petit, Ecol. Lett. 8, 461 (2005).45. D. Jablonski, Nature 427, 589 (2004).46. K. D. Bennett, P. C. Tzedakis, K. J. Willis, J. Biogeogr. 18,

103 (1991).47. P. C. Tzedakis, I. T. Lawson, M. R. Frogley, G. M. Hewitt,

R. C. Preece, Science 297, 2044 (2002).48. K. J. Willis, J. C. McElwain, The Evolution of Plants (Oxford

Univ. Press, Oxford, 2002).49. O. Duriez et al., Conserv. Genet., in press.50. M. E. Douglas, M. R. Douglas, G. W. Schuett, L. W. Porras,

Mol. Ecol. 15, 3353 (2006).51. L. Gillson, K. I. Duffin, Philos. Trans. R. Soc. London Ser. B,

in press.52. M. A. Reid, R. W. Ogden, River Res. Appl. 22, 167 (2006).53. M. Kowalewski, G. E. A. Serrano, K. Flessa, G. A.

Goodfriend, Geology 28, 1059 (2000).54. S. T. Dunn, Alien Flora of Britain (West Newman, London,

1905).55. A. R. Clapham, T. G. Tutin, E. F. Warburg, Flora of the

British Isles (Cambridge Univ. Press, Cambridge, 1952).56. C. A. Stace, New Flora of the British Isles (Cambridge

Univ. Press, Cambridge, 1991).57. We thank K. D. Bennett, S. Bhagwat, and H. H. Birks for

helpful comment and discussion on this manuscript, andE. A. Lynch, M. E. Douglas, and O. Duriez for help infinding suitable photographic images. This is publicationA147 from the Bjerknes Centre for Climate Research.

10.1126/science.1122667

www.sciencemag.org SCIENCE VOL 314 24 NOVEMBER 2006 1265

REVIEW