sign in sign up
<<
Home , Overfishing

Historical Overfishing and the Recent Collapse of Coastal Author(s): Jeremy B. C. Jackson, Michael X. Kirby, Wolfgang H. Berger, Karen A. Bjorndal, Louis W. Botsford, Bruce J. Bourque, Roger H. Bradbury, Richard Cooke, Jon Erlandson, James A. Estes, Terence P. Hughes, Susan Kidwell, Carina B. Lange, Hunter S. Lenihan, John M. Pandolfi, Charles H. Peterson, Robert S. Steneck, Mia J. Tegner and Robert R. Warner Reviewed work(s): Source: Science, New Series, Vol. 293, No. 5530 (Jul. 27, 2001), pp. 629-638 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/3084305 . Accessed: 04/10/2012 16:53

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science.

http://www.jstor.org ECOLOGY THROUGH TIME - w Historical Overfishing and the Recent Collapse of Coastal Ecosystems Jeremy B. C. Jackson,'-2* Michael X. Kirby,3 Wolfgang H. Berger,1 Karen A. Bjorndal,4 Louis W. Botsford,5 Bruce J. Bourque,6 Roger H. Bradbury,7 Richard Cooke,2 Jon Erlandson,8 James A. Estes,9 Terence P. Hughes,10 Susan Kidwell,"1 Carina B. Lange,' Hunter S. Lenihan,12 John M. PandoLfi,3 Charles H. Peterson,12 Robert S. Steneck,14 Mia J. Tegner,lt Robert R. Warner'5

Ecologicalextinction caused by overfishing precedes all other pervasive longer term cycles or shifts in oceanographic human disturbanceto coastal ecosystems, includingpollution, degrada- regimes and productivity(15-17). To help ad- tion of water quality, and anthropogenicclimate change. Historicalabun- dress this problem, we describe dances of large consumer species were fantastically large in comparison structurepredating modem ecological studies with recent observations. Paleoecological,archaeological, and historical using well-datedtime seriesbased on biological data show that time lags of decades to centuries occurredbetween the (18, 19), biogeochemical (20, 21), physical onset of overfishingand consequent changes in ecological communities, (22), and historical(23) proxies that are infor- because unfished species of similartrophic level assumed the ecological mative over a varietyof spatialscales and bio- roles of overfished species until they too were overfished or died of geographicrealms (24). Althoughproxies vary epidemic diseases related to overcrowding.Retrospective data not only in precisionand clarityof the signalsthey mea- help to clarify underlyingcauses and rates of ecological change, but they sure, the use of multipleproxies that give the also demonstrate achievable goals for restoration and management of same ecological signal greatlyincreases confi- coastal ecosystems that could not even be contemplated based on the dence in results.Precision in age datingvaries limited perspective of recent observationsalone. fromcenturies to a single year, season,or event in the exceptionalcase of varvedsediments, ice Few modem ecological studies take into ac- manatees,dugongs, cows, monkseals, croc- cores, and writtenhistorical records (25). Pre- count the former naturalabundances of large odiles, codfish,jewfish, ,, and cision decreaseswith the amountof biological marinevertebrates. There are dozens of places rays are other large marinevertebrates that are or physical disturbanceto the sediment ana- in the Caribbeannamed after large sea turtles now functionallyor entirely extinct in most lyzed (26). whose adult populationsnow number in the coastal ecosystems (3-10). Place names for We exploited data from many disciplines tens of thousandsrather than the tens of mil- oysters, pearls, and conches conjureup other that span the period over which anthropogen- lions of a few centuries ago (1, 2). Whales, ecological ghosts of marine invertebratesthat ic changes may have occurred. Because our were once so abundantas to pose hazardsto hypothesis is that humans have been disturb- 'Scripps Institution of Oceanography, University of navigation(11), but are witnessednow only by ing marine ecosystems since they first California, San Diego, La Jolla, CA 92093-0244, USA. massive garbageheaps of empty shells. learnedhow to , our time periods need to 2Center for Tropical Paleoecology and Archeology, Such ghosts representa far more profound begin well before the human occupation or Smithsonian Tropical Research Institute, Box 2072, problem for ecological understanding and European colonization of a coastal region. Balboa, Republic of Panama. 3National Center for Ecological Analysis and Synthesis, 735 State Street, management than currently realized. Evi- Broadly, our data fall into four categories and Suite 300, Santa Barbara, CA 93101, USA. 4Archie dence from retrospective records strongly time periods: Carr Center for Sea Turtle Research and Department suggests that major structuraland functional 1) Paleoecological records from marine of Zoology, University of , Gainesville, FL changes due to overfishing (12) occurred sediments from about 125,000 years ago to 32611, USA. 5Department of Wildlife, Fish, and Con- worldwide in coastal marineecosystems over the present, coinciding with the rise of mod- servation , University of California, Davis, CA 95616, USA. 6Department of Anthropology, 155 Pet- many centuries. Severe overfishing drives ern Homo sapiens. tengill Hall, Bates College, Lewiston, ME 04240, USA. species to ecological extinctionbecause over- 2) Archaeological records from human 7Centre for Resource and Environmental Studies, Aus- fished populationsno longer interact signifi- coastal seftlements occupied after about tralian National University, Canberra, ACT0200, Aus- cantly with other species in the community 10,000 years before the present (yr B.P.) tralia. 8Department of Anthropology, University of Oregon, Eugene, OR 97403, USA. 9U.S. Geological (5). Overfishing and ecological when worldwide sea level approached Survey, A-316 Earth and Marine Sciences Building, predate and precondition modem ecological present levels. These document human ex- University of California, Santa Cruz, CA 95064, USA. investigationsand the collapse of marineeco- ploitation of coastal resources for food and 1'Center for Reef , Department of systems in recent times, raising the possibil- materialsby past populationsthat range from , James Cook University, Townsville, QLD 4811, . "Department of Geophysical ity that many more marine ecosystems may small-scale aboriginalsocieties to towns, cit- Sciences, University of Chicago, 5734 South Ellis Av- be vulnerableto collapse in the near future. ies, and empires. enue, Chicago, IL 60637, USA. "2Institute of Marine 3) Historical records from documents, Sciences, University of North Carolina at Chapel Hill, Importance of Historical Data journals, and charts from the 15th century to 3431 Arendell Street, Morehead City, NC 28557, USA. 13Department of Paleobiology, National Museum of Most ecologicalresearch is based on local field the presentthat documentthe period from the Natural History, Smithsonian Institution, Washington, studies lastingonly a few years and conducted first Europeantrade-based colonial expansion DC 20560-0121, USA. 14School of Marine Sciences, sometime after the 1950s without longer term and exploitation in the Americas and the University of Maine, Darling Marine Center, Orono, historicalperspective (1, 8, 13). Such observa- South Pacific (23). ME 04573, USA. "5Department of Ecology, Evolution, tions fail to encompassthe life-spansof many and Marine Biology, University of California, Santa 4) Ecological records from the scientific Barbara, CA 93106, USA. ecologicallyimportant species (13, 14) andcrit- literatureover the past centuryto the present *To whom correspondence should be addressed. E- ically important environmental disturbances covering the period of globalized exploitation mail: [email protected] such as extremecyclones or ENSO (El Nifo- of marine resources. These also help to cali- tDeceased. Southern Oscillation) events (8), as well as brate the older records.

www.sciencemag.org SCIENCE VOL 293 27 JULY 2001 629 ECOLOGY THROUGH TIME -

Time Periods, Geography, and Analysis mercantile powers incorporatingdistant re- the Americas, , and Australia We recognize three differentbut overlapping sources into a developing market economy. the differentstages are well separatedin time, periods of human impact on marine ecosys- Global use involves more intense and geo- and the aboriginaland colonial periods began tems: aboriginal, colonial, and global. Ab- graphicallypervasive exploitation of coastal, at different times in the different regions. original use refers to subsistence exploitation shelf, and oceanic integrated into Thus, we can distinguish between cultural of near-shore, coastal ecosystems by human global patternsof resourceconsumption, with stages, as well as between humanimpacts and cultureswith relatively simple watercraftand more frequentexhaustion and substitutionof naturalchanges due to changing climate. extractive technologies that varied widely in fisheries. In Africa, Europe, and Asia, these The addition of a deep historical dimen- magnitude and geographic extent. Colonial cultural stages are strongly confounded in sion to analyze and interpretecological prob- use comprises systematicexploitation and de- time and space, so that their differentialsig- lems requires that we sacrifice some of the pletion of coastal and shelf by foreign nificance is difficult to establish. However, in apparent precision and analytical elegance prized by ecologists (1, 13, 14). Paleoeco- BEFORE AFTER FISHING logical, archaeological, and historical data were collected -formany purposes,vary wide- ly in methods of collection and quality, and A Forests B Alaska/California Gulfof Maine Alaska/California Gulfof Maine are less amenableto many types of statistical analysis than well-controlled experiments. Killerwhales Killerwhales People But none of these problems outweighs the 4~~~~~~~~~~~~~~~~~~~~4 benefits of a historical approach.Clearly, we a Sheep- Sea Sea Shee N cannot generate realistic null hypotheses (ottersl oters h head mink about the composition and dynamics of eco- ? ~~~Lobster Lobe systems from our understanding of the present alone, since all ecosystems have al- | Abalones_A SeaA most certainly changed due to both human cows urchinSnwsAaoe urchin and naturalenvironmental factors (8, 16, 27, 28). Here, we briefly review long-term hu- . Kelp Kelp man impacts in several key marine ecosys- tems. These reconstructionsprovide insight Coral Reefs D into the and extent of degraded eco- C Peo le systems that point to new strategies for mit- Birds i-Sha -11 MonkCrocodiles Birds j4 S rocodiles igation and restoration that are unlikely to emerge from modem monitoringprograms.

r red Pred P Kelp Forests I vei rvrts Kelp forests characterizeshallow, rocky hab- itats from warm temperate to subarctic re- Grazing Grazing Sea Sea Grazing Gr ing Sea Sea gions worldwide and provide complex envi- inverts fish cows turtles inverts \fih/ cows rties ronments for many commercially important Macroalgae gaSs ponges Mac Corals eages and invertebrates(29). NorthernHemi- sphere kelp forests have experienced wide- spread reductions in the number of trophic F levels and due to population explosions of herbivores following the re- Whales-. Sharks-. Seals Crocodiles Whales SCroodiles moval of apex predatorsby fishing (Fig. 1, A and B). Phase shifts between forested and Bi les Bi refh e les deforested states (the latter known as "sea urchin barrens")result from intense grazing due to increased abundanceand altered for- ivert hiivert aging patternsof sea urchinsmade possible in turnby humanremoval of their predatorsand Zo-GrazingZoo-\ Graing SeaSe Mrms/oma O t Zoo-\ GGra zing Sea tormsls Oysters fish cows Oystersplanktonfish coWS A phipods plankton A phipods competitors(7, 8, 30-32). The ecosystem of the Northem I \.1 \I 1. ->v Pacific arose duringthe last 20 million years 4 Phytoplankton Benthic S rass <1hytoplankton al agrass aig egas lgaeC with the evolution of , strongylocent- icro ial Detritus Microbial- Detritus rotid sea urchins, sea otters, and the extinct loop ~~~~~~~~~~~loop Steller's sea cow (6). Sea cows were widely distributedacross the northern Pacific Rim Fig. 1. Simplifiedcoastal food webs showing changes in some of the important top-down throughthe Late Pleistocene. They may have interactionsdue to overfishing;before (left side) and after (rightside) fishing.(A and B) Kelpforests been eliminated from most of their range by for Alaskaand southernCalifornia (left box),and Gulfof Maine(right box). (C and D) Tropicalcoral aboriginalhunting at the end of the Pleisto- reefs and seagrassmeadows. (E and F) Temperateestuaries. The representationof food webs after cene and in the early Holocene, because they fishingis necessarilymore arbitrarythan those before fishingbecause of rapidlychanging recent survived events. Forexample, sea urchinsare once again rarein the Gulfof Maine,as they were beforethe thousands.of years longer in the overfishingof cod, due to the recent fishingof sea urchinsthat has also permittedthe recoveryof western Aleutian Islands that were not peo- kelp.Bold font representsabundant; normal font representsrare; "crossed-out" represents extinct. pled until about4000 yr B.P. (6). By the time Thickarrows represent strong interactions;thin arrowsrepresent weak interactions. of Europeancontact in 1741, sea cows per-

630 27 JULY 2001 VOL 293 SCIENCE www.sciencemag.org ECOLOGY THROUGH TIME sisted only in the Commander Islands, the extinction of sheephead, spiny lobsters, and Table 1). Patternsof communitymembership only islands of the Aleutians unoccupied by abalone startingin the 1950s (8, 36) (Table 1 and dominance of coral species were also aboriginalpeople. Europeanfur traderskilled and Fig. 1, A and B). Subsequentfishing of highly predictable(44), so thatthere is a clear the last sea cow 27 years later in 1768. We the largest sea urchinspecies in the 1970s and baseline of pristine coral community compo- have no idea to what extent abundant sea 1980s resulted in the return of well-devel- sition before human impact. cows grazed kelp forests, although their ap- oped kelp forests in many areas that, as in the WesternAtlantic reef corals suffered sud- parent inability to dive deeply probably lim- Gulf of Maine, effectively lack trophic levels den, catastrophicmortality in the 1980s due ited their grazing to the surface canopy of higher than that of primary producers (36, to overgrowth by macroalgae that exploded kelps and to seaweeds lining the shore (6). 39). in abundanceafter mass mortality of the su- Northern Pacific kelp forests presum- perabundantsea urchin Diadema antillarum ably flourished before human settlement Coral Reefs that was the last remaininggrazer of macroal- because predation by sea otters on sea ur- Coral reefs are the most structurallycomplex gae (42, 47). Early fisheries reports suggest chins prevented the urchins from overgraz- and taxonomically diverse marine ecosys- that large herbivorous fishes were already ing kelp (30). Aboriginal Aleuts greatly tems, providing habitatfor tens of thousands rare before the 20th century (48). However, diminished sea otters beginning around of associated fishes and invertebrates(40). macroalgaewere held in check until the last 2500 yr B.P., with a concomitant increase Aboriginalfishing in environments major herbivore,Diadema, was lost from the in the size of sea urchins (31). Fur traders began at least 35,000 to 40,000 years ago in system through disease (42, 47). subsequently hunted otters to the brink of the western Pacific (41) but appearsto have Corals on the have extinction in the 1800s with the attendant had limited ecological impact. Recently, cor- experienced recurrentmass mortality since collapse of kelp forests grazed away by sea al reefs have experienced dramatic phase 1960 due to spectacular outbreaks of the urchins released from predation. shifts in dominant species due to intensified crown-of-thorns starfish Acanthaster planci Legal protection of sea otters in the 20th human disturbancebeginning centuries ago that feeds on coral (49). The causes of out- century partially reversed this scenario. (1) (Fig. 1, C and D). The effects are most breaks are controversial,but they are almost However, kelp forests are again being de- pronouncedin the Caribbean(42) but are also certainlynew phenomena.There are no early pleted in areas of Alaska because of in- apparenton the GreatBarrier Reef in Austra- records of Acanthaster in undisturbedfossil creased predation on sea otters by killer lia despite extensive protection over the past deposits, in aboriginalfolklore, or in accounts whales (33). The whales shifted their diet to three decades (43). of European explorers and fishers. Now, in sea otters from seals and sea lions, which Large species of branchingAcropora cor- recent decades, the frequencyand intensityof are in drastic decline. als dominated shallow reefs in the tropical outbreaks have exceeded the capability of A similar sequence of events occurred in western Atlantic for at least half a million longer lived species to recover as outbreaks kelp forests of the Gulf of Maine (7, 34). Sea years (44-46) until the 1980s when they have become more chronic than episodic otters were never present, but declined dramatically(42, 47) (Fig. 2B and (50). and other large ground fish are voracious predators of sea urchins. These fishes kept Fig. 2. Retrospectivedata showing baselines 120- A sea urchin populationssmall enough to allow before ecosystem collapse. (A) Time series of E 100- persistence of kelp forests despite intensive mean body length of Atlantic cod from kelp aboriginaland early Europeanhook-and-line forests in the coastal Gulfof Maine.The earlier >, 80- five data pointsare derivedfrom 60- fishing for at least 5000 years. New mecha- archaeological records,whereas the last three points are from E nized fishing technology in the 1920s set off fisheries C: data (113). Verticalbars represent the CZ a rapid decline in numbers and body size of standarderror. Horizontalbars representthe 20- coastal cod in the Gulf of Maine (7) (Fig. 2A time range of data for a single interval of and Table 1) that has extended offshore to observations.(B) Paleoecologicaland ecological . C B data showing the percentageof Caribbeanlo- . 80- (35). Formerlydominant pred- E i calities with Acropora palmata (A) or A. cervi- C o ? atory fish are now ecologically extinct and 0 comis (U) as the dominant shallow-water coral 60- have been partially replaced by smaller and in the Late Pleistocene, Holocene, before 1983, 2 40 commercially less important species. Lob- and after 1983 (114). Percentagesof localities O Q sters, crabs, and sea urchins rose in abun- are significantlydifferent over the four time 20- dance accordingly (7). Kelp forests disap- periodsfor A. palmata (X2 = 34.0, P <0.0001, o peared with the rise in sea urchins due to df = 3) and A. cervicornis (X2 = 22.4, P 0 10 df = Verticaland horizontalbars 6- C removal of , and then reap- <0.0001, 3). E are as in (A). (C) Paleoecological and fisheries 0, r 8 pearedwhen sea urchinswere in turnreduced data from ChesapeakeBay showing the ratio 'a . v C 4- -6 0 to low abundanceby fishing. in abundance of planktonic to benthic dia- The more diverse food web of southern toms (dotted line) (77) and landings of the a E 3 4 4( 4 0 o Crassostrea . California kelp forests historically included oyster virginica (solid line) (80). L. 2- spiny lobsters and large sheepheadlabrid fish The planktonicto benthic diatom ratio is a '-2 C in addition to sea otters as predatorsof sea proxy for eutrophicationthat shows the rel- ,. < ative amount of planktonicto benthic prima- 0 0 Cu urchins, as well as numerous species of aba- 106 105 1000 100 ry production (77). For over 1200 years this Yr 10p lone that compete with sea urchins for kelps ratio remained fairly constant at about 1:1, Yearsbeforepresent (Fig. 1, A and B) (36). Aboriginal exploita- but then increased threefold coincidentally tion began about 10,000 yr B.P. and may with increased runoff of sediments and nutrients due to Europeanagriculture after 1750. The have had local effects on kelp communities ratio remainedat about 3:1 between 1830 and 1930, after which it increaseddramatically to (37). The fur trade effectively eliminated sea about 8:1. Oyster landingsshow an initial increase in the early 19th century, peak in 1884, and subsequent collapse as deep channel reefs were destroyed by mechanicaldredging (80). These ottersby the early 1800s (38), but kelp forests data strongly imply that oysters were able to limit the potential for eutrophicationinduced by did not begin to disappear on a large scale increased inputs of nutrients between 1750 and 1930 until oyster populationscollapsed as a until the intense exploitation and ecological result of overfishing.

www.sciencemag.org SCIENCE VOL 293 27 JULY 2001 631 ECOLOGY THROUGH TIME

LU ~j N - T W m m 00 It "It 0~ Ln 0 N ~m N m- "r "0

o-v --vUcl) - -vC ~~~~~0 Cl) Cl) -v~~~~~~~~C\U U c ~ ~ ~ ~ ~ l v~C) C)U U

V) m ~ ~ ~ ~ ~ ~ ~ C ) -- N ---

Cl0 CV 0

m ~ ~ ~ U 0 -v

M 4 ) C. Cr Lf; Lf c- -vN ~~~C > -. Cl)~ ~ ~ 0 I C L

Cl)4- 4l) 0 m+10 t t ' 0 0V0 0 w-

Mi ccE0 l U,- - l -c E n~~~~U U, 2 -

> 4-' 0 Cl) Cl) > 0Lrmcl ~ ~ - C E D t E 4( Cl) m (V QL V~~C)l) VU, 0 , E .0> ) o 0 U UL vC)l Cl LI -V) EJ C3 0& 0 Q~xN -6 .- 0 0 ) QL ~~ *-~~~ .~~~~ 0U , -v w t 0VOt Ln EU,Z ' c~~ 0Cl) -o o ~~~~~~~ oc~~c~~-~~0) ~ ~ ~ >~~U ' V) 0 m U0-C~X C C C E 0~ 0 0 c 0 .0 0 4- 4 E00- !C 4X 4- Xo0o w 0 Ln0 0 ~ C> .00mc1 N .im.0 A0 A 0 N700c Cl)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~V 0E 0 0 0 0 0 0 o 4-Jf 0 00 o V00 0 0 0L) 0 *~~u~~ m NV N ri A A A U, -ct E ~~~~~~~~~V t!) tV

E 4-j 0~~~~~~~~~~~~~~rU C 0' Eo CiN G)) u,

a- 0- r-- m >10u C ; 4- - U,~~~~~~~~~~~~&mx .0 e, ?, = , E~~ ~ E -0 CC~~ 0)U,~..a).0~~L -v0> NuvU-U,~~ .--0o C -vU,\ a C UUC~ w co 'ILU LU LU LU LU -I N < LU 6 r ClULIL I I~~~~ ~CC 00>E I &I &I &I m Ao .0 uj Z;in V)M ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ )U C Ca kD ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 .- UC) C a) a) >~~~~~~).u~~~ a)

U,L.n .- U UC C0 -mu Ew E au' 0CCC r~~~0 a)4) V)4- a) V) V) V)~~~~~~~~~~a4i ~~~~~~~~c~~~~~~~~~~~~~~~~~~~~C 0 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ V)-J0 -c c4- w 0 ; . 0 0 C~~-~ 0C 00 0 I--'-- 0~~~~~~tcU 632 VOL~~~~~~-27 JULY2001 29 SCEC ww.cinemgorg ECOLOGY THROUGH TIME One possible explanationfor Acanthaster ' COO ~~~~~~~~~~~~~~ r~~~o"trn~ outbreaks is that overfishing of species that prevy upon larval or juvenile stages of crown-of-thorns starfish is responsible for m m massive recruitmentof the starfish (51). The 4) m ~~~~ N~~r, NJ m m NN4(1 m m m m m highly cryptic, predator-avoidingbehavior of juvenile starfish, their formidable antipreda- tor defenses as subadultsand adults, and the ~~~ro~ reductionof some generalizedpredatory fish- es on the GreatBarrier Reef all point to such 0 0 -0 0 0 0 0~~~~ 1:9 a "top-down"explanation. Commercial and recreationalfishing, as well as indirecteffects of intensive for prawns, are likely E 0 explanations for decreased abundance of 0 0 0 predators of crown-of-thorns starfish (52). 0 VI ON W >1 VI Massive recruitmentof starfish may also be E E due to "bottom-up"increases in productivity oL 0 0 - i due to increasedrunoff of nutrientsfrom the

0 E u .AI land (53). In either case, the explanation is ~~~j I~~~~~~E I D 4 C- 0. tC. .x almost certainlyhistorical and anthropogenic, and cannot be resolved by recent observa- tions alone. Expeditionsoccurred annually to northern 0 0 0 0.0 Australia from the Malay Archipelago -0~ Co0 throughout the 18th and 19th centuries to uo E ~c ~~~~L)Nco co 0 harvestan estimated6 million sea cucumbers each season (54). After European coloniza- 0~~~~~~~ tion, industrial-scalefishing developed along the Great Barrier Reef and subtropical east Australian in the early to mid-19th 00 0n DNO -C0~ century (55). Whales, dugongs, turtles, pearl oysters, and Trochusshell were each heavily ~ ~- N~~Lf~ m 0 0 O VN exploited only to rapidly collapse, and all have failed to regain more than a small frac- tion of their formerabundance (55-57). Fish- ing of pelagic and reef fishes, sharks, and cC CGJC = 't ,u prawnshas continuedto the present,although .r0- i 5Cj oLI has declined greatly (58).

Tropical and Subtropical Seagrass Beds

www.sciencemag.org SCIENCE VOL 293 27 JULY 2001 633 - ECOLOGY THROUGH TIME - seagrass beds in ways that increased their ongs remove up to 96% of above-ground ecological response was a gradualshift in the vulnerability to recent events. and 71% of below-ground biomass taxa responsible for primaryproduction that Vast populationsof very large green turtles of seagrasses(73). Theirgrazing rips up large began in the late 18thcentury. Seagrasses and were eliminatedfrom the Americasbefore the areas of seagrass beds, providing space for benthic diatoms on the bay floor declined, 19th century(1, 2) (Table 1). Formerlygreat colonizationby competitivelyinferior species while planktonic diatoms and other phyto- populationsof green turtles in Moreton Bay, of seagrasses. Dugong grazing also produces planktonin the corresponding- Australia, also were greatly reduced by the massive amountsof floating debris and dung ly increased. However, anoxia and hypoxia early20th century(66). Moreover,there are no that are exportedto adjacentecosystems. The were not widespread until the 1930s when estimatesof abundancesof turtlesin Australia decline in seagrasses in Moreton Bay is cer- phytoplankton populations and the flux of at the dawn of Europeanexploitation, so that tainly due in large partto the dramaticdecline organic matter to the bay floor increased reportedreductions must be only a small frac- in water quality due to and dramaticallywith concomitantloss of benthic tion of the total numberslost. All turtlespecies runoff of sediment (63, 64). Nevertheless, as fauna (75, 77) (Fig. 2C and Table 1). Similar continueto decline at unsustainablerates along noted for green turtlesand turtlegrassin Flor- changes began in the 1950s in the Baltic Sea, the GreatBarrier Reef today (67). ida Bay, the cessation of systematic plowing with widespread expansion of the extent of Abundant green turtles closely crop of the bay floor by once abundantdugongs anoxic laminated sediments (74, 85), and in turtlegrass and greatly reduce the flux of must also have been a major factor. the 1950s to 1970s in Pamlico Sound (84). organic matter and nutrients to sediments Vast oyster reefs were once prominent (59-62, 68). In the near absence of green Oysters and Eutrophication in structures in Chesapeake Bay (11), where turtles today, turtlegrass beds grow longer Estuaries they may have filtered the equivalent of the blades that baffle currents, shade the bot- Temperateestuaries worldwide are undergo- entire water column every 3 days (79). De- tom, start to decompose in situ, and provide ing profound changes in oceanography and spite intensive harvesting by aboriginal and suitable substrate for colonization by the ecology due to human exploitation and pol- early colonial populations spanning several slime molds that cause.turtlegrass wasting lution, renderingthem the most degraded of millennia, it was not until the introductionof disease (65). Deposition within the beds of marineecosystems (74-76) (Fig. 1, E and F). mechanical harvesting with dredges in the vastly more plant detritus also fuels micro- The litany of changes includes increasedsed- 1870s that deep channel reefs were seriously bial populations, increases the oxygen de- imentationand turbidity(77); enhanced epi- affected (79, 80). Oyster catch was rapidly mand of sediments, and promotes hypoxia sodes of hypoxia or anoxia (74, 75, 77); loss reduced to a few percent of peak values by (65). Thus, all the factors that have been of seagrasses (78) and dominant suspension the early 20th century (79, 80) (Fig. 2C and linked with recent die-off of turtlegrass feeders (79), with a general loss of oyster reef Table 1). Only then, after the oyster beds in Florida Bay (65), except for chang- habitat (80); shifts from ecosystems once had collapsed, did hypoxia, anoxia, and other es in temperature and salinity, can be at- dominated by benthic primaryproduction to symptoms of eutrophicationbegin to occur in tributed to the of those dominatedby planktonic primarypro- the 1930s (75, 77), and outbreaksof oyster green turtles (27). duction (77); eutrophication(74-76) and en- parasites became prevalent only in the Europeancolonists did not exploit tropical hanced microbialproduction (81); and higher 1950s (80). Thus, fishing explains the bulk Americanmanatees as systematicallyas they frequencyand durationof nuisance algal and of the decline, whereas decline in water exploited green turtles, so the data related to toxic dinoflagellate blooms (82, 83), out- quality and disease were secondary factors fisheries are poor. We know, however, that breaks of (79), and fish kills (83). (80). However, now that oyster reefs are manateeswere extensivelyfished by aboriginal Most explanationsfor these phenomena em- destroyed, the effects of eutrophication, people andby earlycolonists (68). In Australia, phasize "bottom-up"increases in nutrients disease, hypoxia, and continued dredging aboriginalpeople also harvesteddugongs ex- like nitrogen and phosphorus as causes of interact to prevent the recovery of oysters tensively long before European colonization phytoplankton blooms and eutrophication and associated communities (86). Field ex- (3), yet the numbersreported by earlycolonists (74-76), an interpretationconsistent with the periments in Pamlico Sound demonstrate were vast. Three-or four-mile-longherds com- role of estuaries as the focal point and sewer that oysters grow well, survive to maturity, prising tens of thousandsof large individuals for many land-based,human activities. Nev- and resist oyster disease when elevated were observedin Wide Bay in about 1870 (69) ertheless, long-termrecords demonstratethat above the zone of summer hypoxia-even and in MoretonBay as recentlyas 1893 (70). reduced "top-down" control resulting from in the presence of modern levels of eu- Widespreadcolonial exploitationof dugongs losses in benthic suspension feeders predated trophication and pollution (87). for their flesh and oil along the southern eutrophication. Overfishingof oysters to the point of eco- Queenslandcoast resultedin the crash of the The oldest and longest recordscome from logical extinction is just one example in a dugong fishery by the beginning of the 20th cores in sediments from Chesapeake Bay general patternof removal of species capable century(3) (Table 1). Ironically,scientists re- (77) and Pamlico Sound (84) in the eastern of top-down control of community structure cently reportedthe "discoveryof a largepopu- and from the Baltic Sea (85) in estuaries.Dense populationsof oysters and lation" of dugongs in Moreton Bay-a mere that extend back as far as 2500 yr B.P. (Fig. other suspension-feeding bivalves graze 300 individuals(71). Furthernorth, numbers of 2C and Table 1). A general sequence of planktonso efficiently that they limit blooms dugongs in the vast southernhalf of the Great ecological change is apparent in all three of phytoplanktonand prevent symptoms of BarrierReef had dwindledto fewer than 4000 cases, but the timing of specific ecological eutrophication(88, 89), just as occurs with when they were first accurately counted in transitionsdiffers among estuariesin keeping grazing by zooplanktonin freshwaterecosys- 1986-87, with a further50 to 80% decline in with their unique histories of , ex- tems (90). The ecological consequences of recentyears (72). These increasinglyfragment- ploitation, and human -a uncounted other losses are unknown. Gray ed populationsrepresent the last remnantsof difference that rules out a simple climatic whales (now extinct in the Atlantic), dol- the vast herds of the early 19th century and explanation. Increased sedimentation and phins, manatees, river otters, sea turtles, alli- before. burial of organic carbon began in the mid- gators, giant , sheepshead, sharks, The ecological implications of these re- 18th century in Chesapeake Bay, coincident and rays were all once abundantinhabitants ductions are at least as impressive as those for with widespread land clearance for agricul- of Chesapeake Bay but are now virtually green turtles. Moderate sized herds of dug- ture by European colonists (77). The main eliminated.

634 27 JULY 2001 VOL 293 SCIENCE www.sciencemag.org - ECOLOGY THROUGH TIME

Offshore Benthic Communities century. The full sequence of events is most outbreaksof microbialpopulations (88). Continentalshelves cover more of the characteristic of temperate estuaries like The third important corollary is that floor than all previously discussed environ- Chesapeake Bay. Not all the human distur- changes in climate are unlikely to be the ments combined. Commercially important bances illustrated in Fig. 3 have affected all primary reason for microbial outbreaks and cod, , haddock, turbot, , ecosystems yet. But wherever these events disease. The rise of microbes has occurredat , rays, and a host of other groundfishes, have occurred, the standard chronological different times and under different climatic scallops, cockles, and oysters have been sequence of human disturbance and modi- conditions in differentplaces, as exemplified fished intensively for centuries from conti- fication of ecosystems is recognizable. by the time lag between events in Chesapeake nental shelves of Europe and North America, The second important corollary is that Bay and Pamlico Sound (77, 79, 80, 84). and more recently throughoutthe world (5, 7, overfishing may often be a necessary precon- Anthropogenic may now be 10, 91). Hook-and-line fishing was replaced dition for eutrophication,outbreaks of dis- an importantconfounding factor, but it was by intensive use of the beam trawl duringthe ease, or species introductionsto occur (27). not the original cause. Rapid expansion of 18th century, and industrializedfishing was For example, eutrophicationand hypoxia did introduced species in recent decades (95) further intensified with the advent of large not occur in ChesapeakeBay until the 1930s, may have a similarexplanation, in additionto steam- and diesel-powered vessels and the nearly two centuriesafter clearing of land for increase in frequencyand modes of transport. otter trawl at the end of the 19th century. agriculturegreatly increased runoff of sedi- Massive removal of suspension feeders, graz- Reports of severely depleted and ments and nutrients into the (77). ers, and predatorsmust inevitably leave ma- shifting of fishing groundsfarther and farther Suspension feeding by still enormous popu- rine ecosystems more vulnerableto invasion from home ports into the and the lations of oysters was sufficient to remove (96, 97). outer GrandBanks were commonplaceby the most of the increased production of phyto- beginning of the 19th century. Scientific in- plankton and enhanced turbidity until me- Synergistic Effects of Human vestigation consistently lagged behind eco- chanical harvesting progressively decimated Disturbance nomic realities of depleted stocks and inexo- oyster beds from the 1870s to the 1920s (77, Ecological extinction of entire trophic levels rable exploitation of more-distant fishing 80) (Fig. 2C). makes ecosystems more vulnerable to other grounds. As late as 1883, Thomas Huxley The consequencesof overfishing for out- naturaland human disturbancessuch as nu- claimed that fish stocks were inexhaustible breaksof disease in the next lowertrophic level trient loading and eutrophication, hypoxia, (92), a view discredited by the beginning of fall into two categories.The most straightfor- disease, storms, and climate change. Expan- the 20th century (5). Today, several formerly ward is that populations in the lower level sion and intensificationof different forms of abundant,large fish as well as formerlydense become so dense that they are much more human disturbance and their ecological ef- assemblages of suspension feeders are eco- susceptible to disease as a result of greatly fects on coastal ecosystems have increased logically extinct over vast areas (7-10, 93). increasedrates of transmission(94). This was and accelerated with human population presumablythe case forthe sea urchinDiadema growth, unchecked exploitation of biological The Primacy of Overfishing in Human on Caribbeanreefs and the seagrassThalassia resources, technological advance, and the in- Disturbance to Marine Ecosystems in FloridaBay. In contrast,among oysters dis- creased geographic scale of exploitation Overfishing of large vertebratesand shellfish ease did not become importantin Chesapeake through globalization of markets. Moreover, was the first major human disturbanceto all Bay until oysters had been reduced to a few the effects are synergistic, so that the whole coastal ecosystems examined (Table 1). Eco- percentof theiroriginal abundance (80), a pat- response is much greater than the sum of logical changes due to overfishing are strik- repeatedin Pamlico Sound (86, 87) and individual disturbances(98). This is perhaps ingly similar across ecosystems despite the FoveauxStrait, New Zealand(93). Two factors most apparentin the rise of eutrophication, obvious differences in detail (Fig. 1, A to F). may be responsible.First, oysters may have hypoxia, and the outbreak of toxic blooms Everywhere, the magnitude of losses was become less fit owing to stresseslike hypoxia and disease following the destructionof oys- enormousin terms of biomass and abundance or sedimentation,making them less resistantto ter reefs by mechanical harvestingof oysters of large animals that are now effectively ab- disease (87). Alternatively,suspension feeding (79, 80, 86). Other possible examples are sent from most coastal ecosystems world- by dense populationsof oystersand associated outbreaksof seagrass wasting disease due to wide. These changes predatedecological in- species on oyster reefs may have indirectly the removal of grazers of seagrasses like the vestigations and cannot be understoodexcept limited populationsof pathogensby favoring green turtle (27). by historical analysis. Their timing in the other plankton-an explanationthat may ex- A strikingfeature of such synergisticeffects Americas and Pacific closely tracksEuropean tendto blooms of toxic planktonand most other is the suddennessof the transitionin abundance colonization and exploitation in most cases. However, aboriginaloverfishing also had ef- Fig. 3. Historical se- 5 fects, as exemplified by the decline of sea quence of human distur- Climate change otters (and possibly sea cows) in the northeast bances affecting coastal Pacific thousands of years ago. ecosystems. Fishing(step 4 There are three important corollaries to 1) always precededother Introductions the primacy of overfishing. The first is that human disturbancein all cases examined. This is 3Mcail pollution, eutrophication, physical destruc- the basis for our hypoth- Human Mechanical tion of habitats, outbreaks of disease, inva- Alted esis of the primacy of expansion > habitat e Altered sions of introduced species, and human- overfishingin the deterio- destruction eoytm induced climate change all come much later ration of coastal ecosys- 2 than overfishing in the standard sequence tems worldwide. Subse- Pollution of historical events (Fig. 3). The pattern quent steps 2 through 5 1 holds regardless of the initial of have not been observedin 1 timing every example and may Fishing colonial overfishing that began in the vary in order. Americas in the 16th and 17th centuries and in Australia and New Zealand in the 19th "Then" .)J> "Now"

www.sciencemag.org SCIENCE VOL 293 27 JULY 2001 635 - ECOLOGY THROUGH TIME

of different kinds of organisms and com- pervasive removal of higher trophic levels ChesapeakeBay by massive restorationof oys- munity composition due to threshold ef- and the asynchronousoutbreaks of disease in ter reefs (79). Experimentsin Pamlico Sound fects (99). Ecological diversity and redun- different ecosystems that belie a simple cli- show that this is possible (86, 87, 96), and dancy within trophic levels is probably the matic explanation. modelingof food webs suggeststhat even par- most importantreason for the delay or time tial restorationof oysters would reduce eu- lag between the onset of fishing and the Historical Perspectives for Ecosystem trophicationsubstantially (110). Aquacultureof subsequent threshold response (42, 100). Restoration suspension-feedingbivalves like oystersmight The importance of biodiversity in the form The characteristicsequence of human distur- be promotedto reversethe effects of eutrophi- of ecological redundancy is clearly appar- bance to marineecosystems (Fig. 3) provides cation and to restorewater quality in degraded ent for the delay in the collapse of kelp a frameworkfor remediationand restoration estuaries.Other important examples include the forests in southern California compared that is invisible without a historical perspec- restorationof coral reefs and seagrassbeds by with Alaska after the extirpation of sea tive. More specific paleoecological, archaeo- protectionof fishes, sharks, turtles, and sire- otters. Sheephead fish, spiny lobsters, and logical, and historicaldata should be obtained niansin very largereserves on the scale of all of abalone in the more diverse Californian to refine the histories of specific ecosystems Florida Bay and the Florida Keys-an ap- kelp forests kept sea urchin populations in and as a tool for management,but the overall proachrecently advocated for terrestrialecosys- check until these predatorsand competitors patternsare clear. The historical magnitudes tems (111). Once again, small-scale grazing of sea urchins had also been effectively of losses of large animals and oysters were so experimentswith reef fishes (112) show that eliminated (8, 36). Similarly, the sea urchin great as to seem unbelievablebased on mod- fishes couldreverse the overgrowthof coralsby Diadema kept macroalgae in check long em observationsalone (Table 1). Even seem- macroalgaeon a massive scale. The potential after the extreme overfishing of herbivo- ingly gloomy estimates of the global percent- for reducingdiseases of corals and turtlegrass rous fishes on Caribbean coral reefs (42). age of fish stocks that are overfished (108) by restoringnatural levels of grazingis unprov- A second potentially important mecha- are almost certainly far too low. The shifting en but consistentwith historicalevidence (27). nism for the suddenness of ecosystem col- baseline syndrome is thus even more insidi- In summary,historical documentation of the lapse is the elimination of previously un- ous and ecologically widespreadthan is com- long-termeffects of fishing providesa hereto- fished refuges that were protectedhistorically monly realized. fore-missing perspective for successful man- because of distance or expense of access. For On the other hand, recognition of these agementand restorationof coastalmarine eco- example, reef fishes all aroundJamaica in the losses shows what coastal ecosystems could systems.Previous attempts have failedbecause 1960s rarely reached reproductive maturity be like, and the extraordinarymagnitude of they have focused only on the most recent so that the abundant recruits of fishes on economic resourcesthat are retrievableif we symptomsof the problemrather than on their Jamaicanreefs at that time must have come are willing to act on the basis of historical deep historical causes. Contraryto romantic from undiscoveredpopulations in Jamaicaor knowledge. The central point for successful notions of the as the "last frontier" elsewhere (101). But as more and more reefs restoration is that loss of economically im- and of the supposedly superior ecological have been overfished,the potentialsources of portant fisheries, degradation of habitat at- wisdom of non-Western and precolonial such recruits must have effectively disap- tractiveto landownersand tourists,and emer- societies, our analysis demonstrates that pearedover wider areas (102). A similar sce- gence of noxious, toxic, and life-threatening overfishing fundamentally altered coastal nario has been proposed for the American microbial diseases are all part of the same marine ecosystems during each of the cul- lobster with regard to loss of larvae from standardsequence of ecosystem deterioration tural periods we examined. Changes in eco- deep-wateroffshore stocks (103). that has deep historical roots (27). Respond- system structure and function occurred as ing only to currentevents on a case-by-case early as the late aboriginal and early colo- Microbialization of the Global Coastal basis cannot solve these problems. Instead, nial stages, although these pale in compar- Ocean they need to be addressedby a series of bold ison with subsequent events. Human im- Most recent changes to coastal marine eco- experimentsto test the success of integrated pacts are also accelerating in their magni- systems subsequent to overfishing involve management for multiple goals on the scale tude, rates of change, and in the diversity of population explosions of microbes responsi- of entire ecosystems. With few exceptions, processes responsible for changes over ble for increasingeutrophication (74-76, 81), such as the Caribbeanmonk seal and Steller's time. Early changes increased the sensitiv- diseases of marine species (104), toxic sea cow, most species that are ecologically ity of coastal marine ecosystems to subse- blooms (82, 83), and even diseases such as extinct probably survive in sufficient num- quent disturbance and thus preconditioned cholera that affect human health (104, 105). bers for successful restoration.This optimism the collapse we are witnessing. ChesapeakeBay (81) and the Baltic Sea (74) is in stark contrast with the state of many are now bacterially dominated ecosystems terrestrialecosystems where many or most References and Notes with a trophic structuretotally different from large animals are already extinct (28). More- 1. J. B. C. Jackson, Coral Reefs 16, S23 (1997). 2. K. A. Bjorndal, A. B. Bolten, M. Y. Chaloupka, Ecol. that of a century ago. Microbial domination over, we now have the theoreticaltools (109) AppI. 10, 269 (2000). also has expanded to the open ocean off the to roughly estimate per capita interaction 3. G. C. L. Bertram, C. K. Ricardo Bertram, Biol. J. Linn. mouth of the Mississippi River (106) and to strengthsof survivingindividuals of now rare Soc. 5, 297 (1973). the (107). animals like sea turtles,sirenians, sharks, and 4. K. W. Kenyon,J. Mammal. 58, 97 (1977). 5. D. H. Cushing, The Provident Sea (Cambridge Univ. Nowhere is the lack of historicalperspec- large groupers.We can then use these data to Press, Cambridge, UK, 1988). tive more damaging to scientific understand- build tentativemodels of the consequencesof 6. J. A. Estes, D. 0. Duggins, G. B. Rathbun, Conserv. ing than for microbial outbreaks. Plans for the renewed abundance of these species in Biol. 3, 252 (1989). remediationof eutrophicationof estuariesare their native environmentsthat can in turn be 7. R. S. Steneck, in Proceedings of the Gulf of Maine to Ecosystem Dynamics Scientific Symposium and still based on the belief that eutrophicationis used design large-scale, adaptive experi- Workshop (RARGOM Report 91-1, Regional Associ- caused only by increased nutrients without ments for ecosystem restoration,exploitation, ation for Research in the Gulf of Maine, Hanover, regard to overfishing of suspension feeders. and management(96, 108, 110). NH, 1997), pp. 151-165. Even more remarkableis the attributionof One obviously timely and overdue exper- 8. P. K. Dayton, M. J. Tegner, P. B. Edwards, K. L. Riser, Ecol. AppI. 8, 309 (1998). the rise in marine diseases to climate change iment is to attempt the amelioration of eu- 9. J. M. Casey, R. A. Myers, Science 228, 690 (1998). and pollution (104) without regard to the trophication, hypoxia, and toxic blooms in 10. J. A. Hutchings, Nature 406, 882 (2000).

636 27 JULY 2001 VOL 293 SCIENCE www.sciencemag.org - ECOLOGY THROUGH TIME

11. E. Ingersoll,in The Historyand Present Conditionof 43. R. A. Kenchington,Managing Marine Environments 84. S. R. Cooper, The Historyof Water Quality in North the FisheryIndustries, G. B. Goode, Ed. (U.S. Depart- (Taylor and Francis,New York, 1990). Carolina Estuarine Waters as Documented in the ment of the Interior,Tenth Census of the United 44. J. M. Pandolfi,J. B. C. Jackson,Ecol. Monogr. 71, 49 StratigraphicRecord (Univ. of North CarolinaWater States, Washington, DC, 1881), pp. 1-252. (2001). Resources ResearchInstitute, Raleigh,NC, 2000). 12. "Fishing"is the original and most general term in 45. R. B. Aronson, W. F. Precht, I. G. Macintyre, Coral 85. J. Persson, P. Jonsson, Mar. Pollut. Bull. 40, 122 the English language for hunting and gathering all Reefs 17, 223 (1998). (2000). kinds of organisms in the oceans, including plants, 46. J. Geister, Proc. 3rd Int. Coral Reef Symp. 1, 23 86. H. S. Lenihan,C. H. Peterson, Ecol. Appl. 8, 128 invertebrates, and vertebrates ranging from sharks (1977). (1998). to true fishes to whales. "Overfishing"is the reduc- 47. H. A. Lessios,Annu. Rev. Ecol. Syst. 19, 371 (1988). 87. H. S. Lenihan,F. Micheli,S. W. Shelton, C. H. Peter- tion of a species well below the or 48. J. E. Duerden, West IndianBull. 1901, 121 (1901). son, Limnol.Oceanogr. 44, 910 (1999). to such low abundancethat it may not recover even 49. J. Sapp, What Is Natural? Coral Reef Crisis(Oxford 88. C. B. Officer,T. J. Smayda, R. Mann,Mar. Ecol. Prog. if fishing ceases. Univ. Press, New York, 1999). Ser. 9, 203 (1982). 13. J. H. Brown,Macroecology (Univ. of Chicago Press, 50. R. M. Seymour, R. H. Bradbury,Mar. Ecol. Prog. Ser. 89. J. E. Cloern,Mar. Ecol. Prog. Ser. 9, 191 (1982). Chicago, IL,1995). 176, 1 (1999). 90. S. R.Carpenter et at., Ecol. Monogr. 71, 163 (2001). 14. J. B. C. Jackson,BioScience 41, 475 (1991). 51. R.Ormond et al., in Acanthasterand the CoralReef: 91. D. Pauly, V. Christensen,J. Dalsgaard,R. Froese, F. 15. R. C. Francis, S. R. Hare, Fish. Oceanogr. 3, 279 A Theoretical Perspective, R. H. Bradbury, Ed. TorresJr., Science 279, 860 (1998). (1994). (Springer-Verlag,Berlin, 1990), pp. 189-207. 92. T. H. Huxley, Fish. Exhib.Lit. 4, 1 (1884). 16. A. D. MacCall, Calif. Coop. Fish. Invest. 37, 100 52. A. Steven, An Analysisof FishingActivities on Possi- 93. H. J. Cranfield,K. P. Michael, I. J. Doonan, Aquatic (1996). ble Predatorsof the Crownof ThomsStarfish (Acan- Conserv.Mar. Freshw.Ecosyst. 9, 461 (1999). 17. C. B. Lange,S. K. Burke,W. H. Berger Clim. Change thaster planci) of the Great BarrierReef (Prepared 94. W. M. Hochachka,A. A. Dhondt, Proc. Natl. Acad. 16, 319 (1990). for the Great BarrierReef Marine Park Authority, Sci. U. S. A. 97, 5303 (2000). 18. J. B. C. Jackson,in Biotic Interactionsin Recent and Townsville,Australia, 1988). 95. J. T. Carlton,J. B. Geller, Science 261, 78 (1993). Fossil Benthic Communities, M. J. S. Tevesz, P. L. 53. C. Birkeland,Mar. Biol. 69, 175 (1982). 96. H. S. Lenihan,Ecol. Monogr. 69, 251 (1999). McCall,Eds. (Plenum, New York,1983), pp. 39-120. 54. C. C. MacKnight,The Voyage to Marege: Macassan 97. J. J. Stachowicz, R. B. Whitlatch, R. W. Osman, 19. S. J. M. Davis, The Archaeology of Animals (Yale Trepangersin Northem Australia (Melbourne Univ. Science 286, 1577 (1999). Univ. Press, New Haven, CT, 1987). Press, Melbourne,Australia, 1976). 98. N. Myers,Science 269, 358 (1995). 20. G. Wefer, W. H. Berger,J. Bijima,G. Fischer,in Use 55. D. T. Neil, in Moreton Bay and Catchment, I. R. 99. N. Knowlton,Am. Zool. 32, 674 (1992). of Proxies in Paleoceanography:Examples from the Tibbetts, N. J. Hall, W. C. Dennison, Eds. (Univ. of 100. S. Naeem, Conserv.Biol. 12, 39 (1998). South Atlantic, G. Fischer,G. Wefer, Eds. (Springer- Queensland, Brisbane,Australia, 1998), pp. 3-54. 101. J. Munro,ICLARM (Int. Cent. LivingAquat. Resour. Verlag, Berlin, 1999), pp. 1-68. 56. D. Jones, The Whalersof Tangalooma (Nautical As- Manag.) Stud. Rev. 7, 1 (1983). 21. D. S. Jones, Pa(eontol. Soc. Pap. 4, 37 (1998). sociation of Australia,Melbourne, 1980). 102. J. L. Munro,in Fish and FisheriesSeries; Reef Fisher- 22. C. W. Finkt Jr., J. Coast. Res. Spec. Iss. 17, 402 57. R. Ganter, The Pearl-Shellersof Torres Strait: Re- ies, N. V. C. Polunin,C. M. Roberts,Eds. (Chapman & (1995). source Use, Development and Decline, 1860s-1960s Hall, New York,1996), pp. 1-14. 23. A. W. Crosby,Ecologicat Imperialism: The Biological (MelbourneUniv. Press, Melbourne, Australia, 1994). 103. M. J. Fogarty,Can. Spec. Pub(. Fish.Aquat. Sci. 125, Expansion of Europe 900-1900 (Cambridge Univ. 58. R. H. Quinn, FisheriesResources of the Moreton Bay 273 (1998). Press, Cambridge,UK, 1986). Region (Queensland Fish Management Authority, 104. C. D. Harvellet al., Science 285, 1505 (1999). 24. Biological proxies include fossils or archaeological Brisbane,Australia, 1993). 105. R. R. Colwell, Science 274, 2025 (1996). remains that may indicate presence, abundance, 59. J. C. Ogden, in Handbook of Seagrass Biology: An 106. R. E.Turner, N. N. Rabalais,Nature 368, 619 (1994). age, growth rate, health, or body size, among other Ecosystem Perspective, R. C. Phillips, C. P. McRoy, 107. D. Justic, in Modem and Ancient Continental Shelf factors, of selected taxa. Biogeochemical proxies Eds.(Garland STPM, New York,1980), pp. 173-198. Anoxia, R. V. Tyson, T. H. Pearson, Eds. (Geological include concentrations of inorganic or organic con- 60. G. W. Thayer,D. W. Engel,K. A. Bjorndal,J.Exp. Mar. Society, London,1991), pp. 95-105. stituents and isotopic ratios in sediments or skele- Biol. Ecol. 62, 173 (1982). 108. L.W. Botsford,J. C. Castilla,C. H. Peterson, Science tons that track former abundances of unpreserved 61. J. C. Ogden, L.Robinson, H. Whittock,H. Daganhart, 277, 509 (1997). organisms and variations in parameters such as R. Cebula,J. Exp. Mar. Biol. Ecol. 66, 199 (1983). 109. R. T. Paine, Nature 355, 73 (1992). temperature, salinity, oxygen, nutrient chemistry, 62. G. W. Thayer, K. A. Bjorndahl,J. C. Ogden, S. L. 110. R. E. Ulanowicz, J. H. Tuttle, Estuaries 15, 298 and productivity.Physical proxies include sedimen- Williams,J. C. Zieman, Estuaries7, 351 (1984). (1992). tary structures and composition that reflect water 63. H. Kirkman,Aquat. Bot. 5, 63 (1978). 111. M. E. Soule, J. Terborgh,Eds., Continental Conserva- movements, runoff from the land, and rate of sed- 64. E.G. Abal,W. C. Dennison,Mar. Freshw. Res. 47, 763 tion (Island,Washington, DC, 1999). imentation. Historicalproxies include demographic, (1996). 112. S. M. Lewis,Ecol. Monogr. 56, 183 (1986). customs, and commercial records,as well as explor- 65. J. C. Zieman, J. W. Fourqurean,T. A. Frankovich, 113. Cod body length for samples older than 50 years was ers' and naturalists' descriptions of sights and Estuaries22, 460 (1999). derived from cod vertebrae collected at the Turner events and nautical charts marking reefs, coastal 66. C. J. Limpus,P. J. Couper,M. A. Read,Mem. Queens- Farmshell midden,North Haven,Maine. Five occupa- wetlands, and other landmarks. land Mus. 35, 139 (1994). tions are recognized:Occupations 1 (5300 to 4900 yr 25. R. S. Bradley,Paleoclimatology: ReconstructingCli- 67. C. J. Limpus,in State of the GreatBarrier Reef World B.P.),2 (4500 to 4200 yr B.P.),3 (3700 to 3500 yr B.P.), mates of the Quatemary (Academic Press, San Di- HeritageArea Workshop,D. Wachenfeld,J. Oliver,K. and 4 (= CeramicPeriod, 3000 to 1000 yr B.P.),and ego, ed. 2, 1999). Davis, Eds. (Great BarrierReef MarinePark Author- PlowZone (mixedfrom earlier occupations but mostly 26. S. M. Kidwell,K. W. Ftessa,Annu. Rev. Ecol. Syst. 26, ity, Townsville,Australia, 1995), pp. 258-265. postdating1000 yr B.P.).Regression of body length on 269 (1995). 68. W. Dampier,A New Voyage around the World(Do- vertebradiameter of livingcod allowed estimation of 27. J. B. C. Jackson, Proc. NatI. Acad. Sci. U.S.A. 98, ver, New York, 1968). body lengths from vertebrae of ancient cod [y = 5411 (2001). 69. E.Thorne, The Queen of the Colonies (SampsonLow, 61.0 + 204.6log(x)].This model assumes that we are 28. D. H. Janzen, P. S. Martin,Science 215, 19 (1982). Marson,Searle, and Rivington,London, 1876). usingthe largestvertebra for each fish, and wilt there- 29. P. K. Dayton, Annu. Rev. Ecol. Syst. 16, 215 (1985). 70. G. M. Allen, Am. Comm. Int. Wildl. Protect. Spec. fore tend to underestimateactual body size in cases 30. J. A. Estes,J. F. Palmisano,Science 185, 1058 (1974). Pub. 11, 1 (1942). where smallervertebrae are used. Data from the past 31. C. A. Simenstad, J. A. Estes, K. W. Kenyon,Science 71. G. E. Heinsohn, R. J. Lear,M. M. Bryden, H. Marsh, 50 years are actualbody lengthsreported from coastal 200, 403 (1978). B. R. Gardner,Environ. Conserv. 5, 91 (1978). zones [H. B. Bigelow,W. C. Schroeder,Fish. Bull. Fish 32. E. Sala, C. F. Boudouresque, M. Harmelin-Vivien, 72. H. Marsh,P. Corkeron,I. Lawler,J. Lanyon,A. Preen, Wildl.Serv. 53, 1 (1953); J. S. Hacunda,Fish. Bull. 79, Oikos 82, 425 (1998). Great BarrierReef MarinePark Authority Rep. 41, 1 775 (1981); F. P. Ojeda,thesis (Univ.of Maine,1987)]. 33. J. A. Estes, M. T. Tinker,T. M. Williams, D. F. Doak, (1996). 114. Percent Caribbeanlocalities with A. palmata or A. Science 282, 473 (1998). 73. A. Preen, Mar. Ecol. Prog. Ser. 124, 201 (1995). cervicornis as the dominant or codominant corals 34. B. J. Bourque,Diversity and Complexityin Prehistoric 74. R. Elmgren,Ambio 18, 326 (1989). were derived from 50 studies from Antigua, Ba- MaritimeSocieties: A Gulfof MainePerspective (Ple- 75. C. B. Officer et al., Science 223, 22 (1984). hamas, Barbados, Belize, Bonaire, Cayman Islands, num, New York, 1995). 76. S. W. Nixon, Ophelia 41, 199 (1995). Colombia, Dominican Republic, Florida, Haiti, Ja- 35. K. Sherman,Ecol. Appl. 1, 349 (1991). 77. S. R. Cooper, G. S. Brush,Estuaries 16, 617 (1993). maica, Mexico, Netherlands Antilles, Panama, 36. M. J. Tegner, P. K. Dayton, ICES(Int. Counc. Expl. 78. R. J. Orth, K. A. Moore, Science 222, 51 (1983). Puerto Rico, and U.S. Virgin Islands. Studies con- Sea) J. Mar. Sci. 57, 579 (2000). 79. R.I. E.Newell, in Understandingthe Estuary:Advanc- tained either paleoecological data from outcrops 37. J. M. Erlandsonet al., Radiocarbon38, 355 (1996). es in Chesapeake Bay Research, M. P. Lynch, E. C. of fossil reefs or from sediment cores, or ecolog- 38. A. Ogden, The Califomia Sea Otter Trade (Univ. of Krome,Eds. (ChesapeakeBay ResearchConsortium, ical data. For A. palmata, only localities described CaliforniaPress, Berkeley, 1941). Baltimore,MD, 1988), pp. 536-546. as reef crest or between 0- and 10-m water depth 39. M. J. Tegner, P. K. Dayton, Mar. Ecol. Prog. Ser. 77, 80. B. J. Rothschild,J. S. Ault, P. Goulletquer, M. Heral, were included (131 localities). For A. cervicornis, 49 (1991). Mar. Ecol. Prog. Ser. 111, 29 (1994). only localities described as forereef, reef slope, or 40. N. Knowlton,Proc. Natl. Acad. Sci. U. S. A. 98, 5419 81. R. B. Jonas, Am. Zool. 37, 612 (1997). between 10- and 20-m water depth were included (2001). 82. T. J. Smayda, in Toxic Marine Phytoplankton, E. (72 localities). Leeward and windward environ- 41. J. Allen, C. Gosden, J. P. White, Antiquity 63, 548 Graneli et al., Eds. (Elsevier Science, New York, ments were not distinguished. The percentage of (1989). 1990), pp. 29-40. localities that contained A. palmata or A. cervi- 42. T. P. Hughes, Science 265, 1547 (1994). 83. J. M. Burkholderet al., Nature 358, 407 (1992). cornis as the most abundant coral was estimated

www.sciencemag.org SCIENCE VOL 293 27 JULY 2001 637 - ECOLOGY THROUGH TIME -

for four time intervals: Late Pleistocene (before Tegner who died while diving after this paper was University of California,Santa Barbara).Additional humans arrived in the Americas), Holocene (when submitted. This work was conducted as part of the support was also provided for the PostdoctoralAs- only aboriginal populations were present), pre- Long-TermEcological Records of Marine Environ- sociate MXKin the Group.L.W.B. was also support- 1983 (before the mass mortality of Diadema ments, Populations and Communities Working ed by NSFgrant OCE-9711448.We thank A. Bolten, antillarum), and post-1983 (after the Diadema Group supported by the National Center for Eco- S. Cooper, N. Knowlton,B. Mitterdorfer,E. Sala, and mortality). logical Analysisand Synthesis (funded by NSFgrant two anonymous reviewers for discussions and com- 115. We dedicate this paper to the memory of Mia DEB-0072909,the University of California,and the ments on the manuscript.

R EV IE W

Noisy Clockwork: Time Series Analysis of Population Fluctuations in Animals Ottar N. Bj0rnstadl* and Bryan T. Grenfell2 Both biotic interactionsand abiotic randomforcing are crucialinfluences The dynamics of marine stocks serve as on population dynamics. This frequently leads to roughly equal impor- an illustrationof the currentparadigm. Most tance of deterministic and stochastic forces. The resulting tension be- commercial fish stocks vary greatly in abun- tween noise and determinism makes ecological dynamics unique, with dance and the associated time series exhibit conceptualand methodologicalchallenges distinctive from those in other complex spectra,with combinationsof high- dynamicalsystems. Thetheory for stochastic, nonlinearecological dynam- frequencyoscillations and longer term trends ics has been developed alongside methods to test models. A range of (4, 5) (Fig. 1). High-frequency oscillations dynamicalcomponents has been considered-density dependence, envi- are thoughtto arise from environmentalvari- ronmentaland demographicstochasticity, and climaticforcing-as well as ability particularly affecting reproduction their often complex interactions.We discuss recent advances in under- [through expatriation of eggs, temperature- standingecological dynamicsand testing theory using long-termdata and induced mortality,etc. (4)], as well as inter- review how dynamicalforces interact to generate some central field and actions between individuals( and laboratorytime series. cannibalism)or between species (fish-fish or plankton-fishinteractions). The low-frequen- The century of studies in populationecology long-term studies (often 10 or more gener- cy oscillations and trends are usually related has been dominatedby a nested set of debates ations) through time series analysis. to extemal forcing such as overfishing, cli- regardingthe importanceof various dynami- There has been much parallel and inter- matic changes, and decadal, supra-,or super- cal forces. The first controversy concerned twined developmentof these three dynamical decadal oscillations in climate. The most re- the relative impact of biotic versus abiotic themes, and history testifies to a succession cent studies that combine theoretical model- control of population fluctuations. The key of popularity of the various positions (1). ing with time series analysis indicate that the question was the relative importance of Crudelysummarized, early focus on extrinsic full variability in marine stocks can only be "noise"(small-scale, high-frequencystochas- influences was replaced by the "density-de- explained by considering the interactionbe- tic influences) versus climatic forcing (larger- pendent paradigm" (2) in the 1950s and tween nonlinear dynamics and stochastic scale, often lower-frequency signals) versus 1960s. This accelerated in the late 1970s, forcing (5, 6), often in the face of strong nonlinear interactions between individuals with May's cri de coeur (3) about the poten- human influences (7, 8) and obscured by of the same or different species. The second tial of dynamical complexity even in simple measurementerror (5, 7). question concerned the impact of intrinsic models, leading to a focus in the 1980s on The relative importanceof differentcom- (i.e., intraspecific) processes, as opposed to nonlinearityand the detection of determinis- ponents of ecological dynamics differs some- extrinsic or community-level interactions, tic chaos (Taken's embedology, Lyapunov what between systems-notably between ter- an argumentthat has been particularlyheat- exponents, etc.). Research has focused on restrial versus marine, vertebrateversus in- ed with reference to population cycles. A two frontsin the past decade:(i) the impactof vertebrate,simple versus complex life-cycle, third debate, nested within the latter, con- large-scale climatic forcing, coinciding with etc. However, evidence is mounting that all cerns the "dimensionality" of population the rise in popularityof climate change stud- components contributeand interactat partic- fluctuations; given that most populations ies throughthe early 1990s, and (ii) stochas- ular spatial and temporal scales in most sys- are embedded in rich communities and af- tic nonlinearmodels that combine the nonlin- tems. Here we review the currentunderstand- fected by numerous interspecific interac- ear deterministicand (largely) linear stochas- ing of the different forces that drive ecolog- tions, can simple (low-dimensional) models tic theories. The goal in synthesizing these ical dynamics. involving one or a few species capture the approaches in recent years is to understand Simple density-dependent interactions. patterns of fluctuations? All these questions how population fluctuations arise from the Nonlinear, density-dependentinteractions can have been studied through a number of interplayof noise, forcing, and nonlineardy- potentially stabilize or promote fluctuations detailed analyses of specific systems in namics. The comparableimportance of deter- in abundancebecause such interactions can which theoretical models are linked with ministic and stochastic forces makes ecolog- either result in stable equilibria(point attrac- ical dynamics unique. In particular,the inter- tors, namely "the carrying capacity") or cy- action between noise and nonlineardetermin- clic or chaotic attractors, associated with 'Departmentof Entomology,501 ASIBuilding, Penn ism in ecological dynamics adds an extra strongly overcompensatory density depen- State University,University Park, PA 16802, USA. 2Departmentof Zoology, Universityof Cambridge, level of complexity comparedwith the large- dence (3). About 25 years ago, Hassell et al. CambridgeCB2 3EJ, UK. ly stochastic dynamics of, say, economic sys- (9) and Gumey et al. (10) took the bold step *To whom correspondenceshould be addressed.E- tems or the largely deterministicdynamics of of insisting that the then-qualitative,strategic mail:[email protected] many physical and chemical processes. theory ought to be testable by analyses of

638 27 JULY2001 VOL 293 SCIENCE www.sciencemag.org