Accelerating Trophic-Level Dysfunction in Kelp Forest Ecosystems of the Western North
Atlantic
Author(s): Robert S. Steneck, John Vavrinec, Amanda V. Leland
Source: Ecosystems, Vol. 7, No. 4 (Jun., 2004), pp. 323-332
Published by: Springer
Stable URL: http://www.jstor.org/stable/3658819
Accessed: 21/12/2009 02:15

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=springer.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

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 support@jstor.org.




        Springer is collaborating with JSTOR to digitize, preserve and extend access to Ecosystems.




http://www.jstor.org
Ecosystems(2004) 7: 323-332
DOI:10.1007/s10021-004-0240-6                                       ECOSYSTEM
                                                        0 2004 Springer-Verlag




     Accelerating   Trophic-level
     Dysfunction Kelpin    Forest
    Ecosystems of the   Western   North
          Atlantic

        Robert S. Steneck,* John Vavrinec, and Amanda V. Leland

        School of Marine Sciences, Darling Marine Center, University of Maine, Walpole, Maine 04573, USA


ABSTRACT
We use archaeological, historical, ecological, and            half-century). Interphase durations have declined
fisheries data to identify three distinctand sequen-           as fishing effects have acceleratedin recent years.
tial phases in the trophic structureof kelp forests in          The naturallylow species diversityof the kelp forest
the western North Atlantic's Gulf of Maine. Phase            ecosystem we studied may facilitaterapid changes
1 is characterized by vertebrate apex predators             because the redundancywithin each trophiclevel is
such as Atlantic cod, haddock, and wolffish and             low. If the biodiversity within controllingtrophic
persisted for more than 4,000 years. Phase 2 is             levels is a buffer against trophic-level dysfunction,
characterized herbivoroussea urchins and lasted
        by                            then our observationsfrom Maine may be predic-
from the 1970s to the 1990s. Phase 3 is dominated            tive of the fate of other, more diverse systems. If
by invertebrate predators such as large crabs and            fishing successively targets most, or all, strong in-
has developed since 1995. Each phase change re-             teractorsat higher trophiclevels, then as those pop-
sulted directly or indirectly from fisheries-induced           ulation densities decline, the potential for trophic-
"trophic-level dysfunction,"in which populations             level dysfunction and associated instabilities will
of functionally important species at higher trophic
                                     increase.
levels fell below the densitiesnecessaryto limit prey
populations at lower trophic levels. By using frac-           Key words: apex predators; ecosystem stability;
tional trophic-level analysis, we found that phase            fisheries effects; food webs; fractionaltrophic-level
changes occurredrapidly(over a few years to a few            analysis; gulf of Maine; phase changes; trophic
decades) as well as relativelyrecently (over the past          cascades.


INTRODUCTION                               spond to both the high per capita interaction
                                     strengthand the population density of the consum-
Food webs define the structure and function of              ers (Sala and Graham2002).
coastal marine ecosystems (Paine 1980). Often,               Trophic cascades result from the functional re-
predators at high trophic levels have a dispropor-            moval of higher trophic levels, shifting the domi-
tionate influence on the distribution,abundance,             nance and effects of consumers to lower levels
dominance, and diversity of organisms at lower              (Paine 1980; Sala and others 1998). Although spe-
trophic levels (Hairston and others 1960; Paine             cies may become rare, at any trophiclevel, they are
1966, Paine 1980). Such trophic cascades corre-             unlikely to go locally or biologicallyextinct. Reduc-
                                     tions in the population density of stronglyinteract-
Received1 October
        2002; accepted9 May2003;publishedonline 27 April     ing species, functional groups, or trophiclevels will
2004.                                  reduce their functional role as consumers in the
         e-mail:
      author;
*Corresponding      Steneck@maine.edu                system (Estes and others 1989). Sufficient reduc-

                                  323
324   AcceleratingTrophic-levelDysfunction

tions of strong interactorswithin a trophic level, or  shores have been frequently glaciated, causing lo-
trophic-level      can
       dysfunction, destabilizethe commu-     calized extinctions at approximately 20,000-year
nity by releasing prey populations that had been     cycles (Adey and Steneck 2001). Western North
suppressedby consumers at higher trophic levels.     Atlantic food webs generally have four trophiclev-
  Fishingis one of the oldest and most conspicuous   els; diversity is low within each level because the
human disturbancesto marine ecosystems (Jackson     species pool is low throughout the region.
and others 2001). Often, highest-order, or "apex,"     Determining the original prehuman composition
predators are targeted because they are large and    of food webs is a challenging task. Marine macro-
conspicuous, or simply because they have the great-   fossils from the Holocene are rare;thus, it is impos-
est food and commercial value. If fishing pressure    sible to estimate food web structure prior to the
reduces the population density of the apex preda-    human colonization of coastal zones in the Gulf of
tors to the point where they become rare or eco-     Maine. The best evidence of what existed in the past
nomically unprofitable, the fishery target will shift  is derived from archaeological sites in the region.
to lower trophic levels. Fishing of lower trophic    Using archaeological evidence and early historical
levels may become profitable if economic markets     accounts, we reconstructedthe "pristine"  food web
develop for the invertebratesat these levels and if   of coastal Maine. Contemporaryecological studies
those species become more abandant due to their     and fisheries landings data provide much better
release from predation by the former apex preda-     resolution of the structure and function of more
tors. This sequential reduction of highest-ordertro-   recent food webs. Below we describe three dis-
phic levels is known as "fishingdown marine food     tinctly different phases in the trophic structure of
webs" (Pauly and others 1998, 2001).           shallow subtidal zones of Maine.
  Most published accounts of fishing down food
                             Phase 1: Predatory Fishes Dominate
webs cover this phenomenon on an oceanic scale
and documentevents occurring over the past sev-     By all accounts, cod and other large predatoryfish
eral decades to a century (see, for example, Pauly    were stable components of coastal zones through-
and others 2001). In this paper, we review and      out the western North Atlantic in phase 1 (Table 1
analyze studies and fishing records to describe the   and Figure 1 These predatoryfishes were abundant
trophodynamics of coastal zones in the western      in the Gulf of Maine, as evident from the bones
North Atlanticover the past severalthousand years.    excavated by archaeologists from Indian middens
We used several categories of information, includ-    dating from 200 to over 4,000 years ago (Carlson
ing long-term archaeologicaland historicalrecords,    1986; Bourque 1995; Steneck 1997). Indigenous
commercial fisheries data, and in situ ecological    hook-and-line fishers subsisted on a varied diet of
studies.                         marine organisms such as cod, other fish, oysters,
  Our results suggest that there have been three    and clams, as well as terrestrialanimalssuch as deer
distinctphases of trophic dominance in this system    and sea mink (now extinct) (Bourque 1995; When
over the past 4,500 years; these three phases were    the first Europeansexplored the Gulfof Maine, the
characterizedby the dominance of predatory fin-     abundance of large fish impressed them (Rosier
fish, herbivorous sea urchins, and predatoryinver-    1605). On his 1526 map of the New World, Ves-
tebrates, respectively. We consider these phases to   pucci labeled the western North Atlantic coast Ba-
be alternate stable states because each persists be-   callaos,which is Portuguese for "land of the cod-
yond the life span of the dominant organisms       fish." In 1602, Bartholomew Gosnoldnamed Cape
(Sousa and Connell 1985), is followed by a rapid    Cod for the myriad fish that "vexed"his ship. Ex-
shift to the next phase once a threshold is reached   tensive fishing grounds for cod and other predatory
(Knowlton 1992), and is maintained by positive     fishes were mapped for coastal zones in Maine first
feedbacks (Petraitisand Latham 1999).          in the 1880s (Collins and Rathbun 1887) and then
                             again in the 1920s (Rich 1929), with remarkably
RECONSTRUCTINGFOOD WEBS OF THE              little change in areal extent or location (Steneck
WESTERN NORTH ATLANTIC                  1997).
                              Predatoryfishes consume and control the distri-
The normallyspecies-diverseshallow rocky subtidal    bution and abundance of large benthic inverte-
zones are speces-depauperate in the western North    brates (Keatsand others 1986; Witmanand Sebens
Atlantic (Steneck and others 2002). This is because    1992; Vadasand Steneck 1995) (Figure 1). Signifi-
the North Atlanticis relativelyyoung, the assembly    cantly, in Maine no large decapod crabsor lobsters
of its biota from the North Pacificis relativelyrecent  have been found in Native Americanmiddens dat-
(that is, 3.5 Mya) (Vermeij 1991), and its rocky     ing between 5,000 and 400 years before present
Table   1.  Functional  Groups Species, and Trophic Levels of Benthic Marine Communities      in Coastal Maine at T
Species                                           TL     Phase 1
                                                     (before 1940)
Crustose coralline algae (e.g., Clathromorphum  spp., Lithothamnion spp., Phymatolithon)  1      Rare?"
Understory algae (e.g., Chondruscrispus,Desmarestiaspp., Ceramiumspp., Corallina       1      Abundant?a
  officinalis,Bonnemaisonia hamifera, Enteromorphasp., Phycodrus rubens, Ptilota serrata)
Kelps (e.g., Laminaria sp., Alaria esculenta,Agarum clathratum, Desmarestiasp.)        1     Abundant"
Gastropods (e.g. Crepidulasp.) Sea cucumber (Cucumaria)                   2.1
Mussels (Mytilus edulis. Modiolus modiolus)                         2.1     Rared
Green sea urchin (Strongylocentrotus droebachiensis)                    2      Rared
Gastropods (e.g., Tecturatestudinalis Tonicellasp., Lacuna vincta)              2.4     Rare?d
Amphipods (e.g., Gammarus sp.)                                2.3     Common?h
Crabs (Cancerspp., Libinia sp., Hyas sp.)                          2.5     Rare to common'di
Gastropods (e.g., Buccinum undatum)                             2.6
Sea stars (Asteriassp.)                                   3.1
American lobster (Homarus americanus)                            3.2     Rare to commondk'·lw
Hake (Urophycissp., Merlucilus sp.)                             3.6-4.2   Abundant?d
Atlantic wolffish (Anarhichas lupus)                             3.2     Abundant?"
Flounder (Pleuronectessp.)                                  3.2     Abundant0
Tomcod (Microgadustomcod)                                  3.3     Commondo°
Sturgeon (Acipenseroxyrchynchus)                               3.4     Common  do
Cunner (Tautogolabrusadspersus)                               3.5     Rared
Rock gunnel (Pholis gunnellus)                                3.5
Haddock (Melanogrammusaeglefinus)                              3.6     Abundanti,q,r,s,t
Sculpin (Myoxocephalus  spp.)                                3.6     Common0
American plaice (Hippoglossoides platessoides)                       3.7
Skates (Raja sp., Torpedo,Dasyatis sp.)                           3.7     Rare d.
Cusk (Brosmebrosme)                                     4      Common'
Seals (Phoca sp., Halichoerusgrypus)                             4      Rare?d
Dogfish (Squalus acanthias)                                 4.3     Commons
Atlantic cod (Gadus morhua)                                 4.4     Very Abundantda'
Radiated shanty (Ulvaria subbifurcata)                            4.4
Pollock (Pallachius virens)                                 4.5     Abundant"'"
Bluefish (Pomatomussaltatrix)                                4.5     Rared ls
Table 1.   Continued
Species                                 TL              Phase 1                 Phase 2
                                                   (before 1940)              1970-19

Atlantic halibut (Hippoglossushippoglossus)               4.6             Common'                 Rares
White shark (Carcharodoncarcharias)                   4.6             Rarev                  Very ra

               and
TL,trophic (fromFroese Pauly2002)
       level
Theearliest     phaseI is based archaeological for 4,300to400yearsago(Carlson
       recordfor       on      data               1986;Spiess Lewis
                                               and    2001),whereaslaterhistorical   are
                                                                  records fromRosier(1605
arereviewed Steneck
        in          and
            1997;Steneck others 2003). Foodwebsduringphases2 and 3 are basedon in situ ecological
                                                      studies otherstudies, follows:
                                                         and       as
aSteneck others
     and    (2002)
bVavrinec (2003)
CSteneck Dethier
    and     (1994)
dSpiess Lewis(2001)
    and
'Witman  (1985)
fSteneck(1982)
ISteneck (1997)
hHacker Steneck
     and     (1990)
'Smith(1879)
'Leland (2002)
kR.S. Steneck  (unpublished)
'Collins Rathbun
    and      (1887)
mDMR  (1971-2000)
"Verrill(1871)
"Carlson (1986)
PLevin(1994)
qMalpass (1992)
     and
rWitman Sebens    (1992)
'Rosier(1605)
'Rich(1929)
    and
"Gilbert Guldager   (1998)
VBigelow Schroeder
     and      (1953)
WNocrabs lobsters
      or    havebeenreported NativeAmerican
                    in       middens Maine,but by the 1600stheywerereported the coast(Rosier
                               in                    on        1605)and by the 1800stheywerecomm
                                                             R. S. Steneck and others      327

              Phase1: Predatory
                      Fishes            Phase2: Herbivorous             Phase3: Predatory
                  Dominate                Sea Urchins
                                          Dominate            Invertebrates
                                                                Dominate
           4.5                                              Pollock
               Pollock                  Pollock                              Cod
                                                                   Atlindc
                    Atlantc Cod                  -Atlntic  Cod
                                                              H<Hwe
                    \ ^\tHake                            Halke



         .  3.5
         >       /lddocyo
              Flounder  s    \
                       ala           nolunodr   /       \
                                                 t        Flounder       Woifiek
                        AmaLosbtt
                          ]                   / 'YAmerican Lobster            /American Lobster

         0
           2.0                        Cbs                    Crabs


           2.0  GelpUnderstor al e C wse l
                    KehlpdnGreen oe     P            sea ur chin           Greensea urchin




           1.5


           1.0Kelp Understoryalgae     CnuitoNconmiUiwli Udentory Crustose coralline algae
                                   K«ip   ,lgws                Kelp Understory algae Cnutoe   algse
                                                                       corlline



Figure 1. Food webs of coastalzones of Maine. All species determinedto have been abundantat one time were plotted
with their assignedtrophiclevel (see Table1). Abundantspecies are shown in bold face;rareor low-abundancespeciesare
shown in smaller regular type. Most trophic linkages (lines connectingspecies)have been demonstratedwith ecological
studies. Apex fish predators(all above TL3.2) feed on invertebrates(TLless than 3). Predatoryinvertebrates(TL2.5-3.0)
feed on the herbivorous sea urchin (TL 2), which feeds on algae (all TL 1). Interaction strengths correspond to the width
of trophiclinkage lines. Note that some species are weak interactorsin this system. Flounderhave no identifiabletrophic
linkage with other species in this system. Lobster'strophic linkages are weak despite their abundancein recent years
because they feed primarilyon lobsterbait (R.S. Steneck unpublished).


(ybp) (Carlson 1986; Bourque 1995; Spiess and                             ported that they often "washed ashore in great
Lewis 2001). It is possible that these decapodssim-                          numbers." Windrows of kelp detritus have been
ply were not preserved in the middens, although                            shown to be a good indicationthat adjacentsubtidal
mussel shells and sea urchin tests were found at                           zones are kelp-forested (Novaczekand McLachlan
these sites. However, early historical records do                           1986). There was no mention of deforestedpatches
document the existence of lobsters (Rosier 1605;                           at that time. The earliest scientific study in the
Smith 1616; Collins and Rathbun 1887) and crabs                            region (Johnson and Skutch 1928) reported that
(Smith 1879); so it is possiblethat even while pred-                         kelps were the "most characteristicplant in the
ators were still dominatingthe system, some inver-                          midlevels of the sublittoralzone." Similarly,Nova
tebrate populations may have begun to expand as                            Scotiawas describedas kelp-dominatedin the early
coastal populations of predatoryfinfish declined.                           1950s (MacFarlane1952).
 Sea urchins, the only functionalherbivoresin the                           Testsof stabilityoften requireproofthat the dom-
system (Steneck and Dethier 1994), were probably                           inant components of the ecosystem can persist for
uncommon (Spiess and Lewis 2001) and cryptic                             more than one generation (Connell and Sousa
(Loweryand Pearse 1973; Harroldand Reed 1985)                             1983). By all accounts, this first phase, which was
(Figure 1). In the few offshorehabitatswhere large                          dominatedby predatoryfinfish, lasted from at least
predatory fishes still persist, urchins and decapods                         4000 ybp to the mid 1960s. Becausethe averagelife
are rareand kelps are abundant(Vadasand Steneck                            span of the residentgroundfishis 1-2 decades (Big-
1988, Vadasand Steneck 1995;Witmanand Sebens                             elow and Schroeder 1953) and Laminaria 2-4 live
1992) and attack rates on tethered sea urchins are                          years (Chapman 1984, Chapman 1986), phase 1
high (Vadasand Steneck 1995). Significantly,very                           can be considered stable.
small urchins (a few millimetersin diameter)were
found at this site, suggestingthat they can settle but                        Phase 2: Herbivorous Sea Urchins Dominate
do not recruit, probably due to high rates of fish                          Stocks of cod and other groundfishpersisted until
predation (Vadasand Steneck 1995).                                  more sophisticated fishing practices were imple-
 Through at least the 1930s, kelp forests domi-                           mented. Mechanized fishing technology and on-
nated the benthos, while predatory finfish were                            boardrefrigerationenabled the targetingof spawn-
abundant in coastal zones (Steneck 1997; Steneck                                   of
                                                   ing aggregations cod in the 1930s (Conklingand
and others 2002). Hervey (1881) describedall three                          Ames 1996). This led to a rapiddeclinein the num-
dominant kelp species (Table 1) as being "very                            bersand body size of coastalcod in the Gulfof Maine
abundant from Greenland to Cape Cod" and re-                             (Steneck 1997;Jacksonand others2001). Dominant
328   AcceleratingTrophic-levelDysfunction

fish predatorsin the coastal zone were replacedby    ing pressure could no longer control macroalgal
small, commerciallyless importantspecies, such as    recruitment. Local reestablishment of macroalgal
sculpins (Malpass1992). By the 1940s, the extirpa-   beds usually occurred in 1-3 years (McNaught
tion of coastal cod and other fishes in the Gulf of   1999), but fish predatorsremained functionallyab-
Maine resultedin the functionalloss of apex preda-   sent (Table1). Most phase shifts to macroalgaldom-
tors,which fundamentallyalteredcoastalfood webs.    inance occurred in the mid- 1990s (McNaught
  As fish populations were exploited, predation    1999).
pressure on lower trophic levels decreased, proba-    Phase 3 also appears to be stable, but it is main-
bly fostering the increase of urchins and other mo-   tained by different predatory mechanisms (Figure
bile benthic invertebrateswithin the kelp beds. As   1). Partsof the coast that are closed to urchin fish-
localized coastal fish predation droppedto low lev-   ing have remained algal beds for at least 6 years
els urchins were able to aggregate into feeding     (Vavrinec2003). Predationby crustaceansliving in
fronts and denude the benthos of fleshy macroalgae   the complex algal habitat probablymaintainsphase
at an estimated rate of 4 m mo-' (for a review, see   3 by preventing repopulationby urchins. Smallcrab
Scheibling and Hatcher 2001). Locally this transi-   (Cancer and Hyas spp.) and gammarid amphipods
tion could happen rapidly, but regionally it proba-   appear to eat newly settling sea urchins, so the
bly took decades. By the mid-1960s, there was a     annual survival rate in these algal beds is reported
mosaic of kelp beds and urchin "barrens"(Lamb      to be less than 1% (McNaught 1999). A study con-
and Zimmerman 1964; W. Adey personal commu-       ducted in 2000-1 showed that large crabs (espe-
nication). Similarpatches were describeda decade    cially Cancer borealis)prey on and eliminate dense
later in Nova Scotia (Breen and Mann 1976). De-     populations  of reintroduced adult sea urchins (Le-
forested areas continued to expand and coalesce,    land 2002). This effectively keeps vast regions free
and by the mid- 1970s to the early 1980s kelp forests  of sea urchins, thus maintaining stable algal com-
reached an all-time low in their distribution and    munities. Functionally, crabs are now the region's
abundance throughout the region (Steneck 1997;     apex predator because there is no higher-order
Steneck and others 2002).                predator to limit their population density (Leland
  Phase 2 was characterizedby expansive sea ur-    2002).
chin barrens, with kelp beds only in urchin-free      Macroalgaemay also prevent the reestablishment
shallow turbulent zones (Steneck and Dethier      of sea urchin populations. Whiplash (sensuDayton
1994) (Table 1). High urchin densities led to high   1975) or the sweeping of the benthos by frondose
grazing pressure, which prevented the establish-    algae may dislodge encroaching populations of sea
ment of young algal sporophytes (Chapman 1981)     urchins (Kennelly 1989; Konar 2000; Konar and
(Figure 1). Grazing-resistant coralline algae domi-  Estes 2003). In many parts of the southern Gulf of
nated the benthos (90%-100% cover in places)      Maine, kelp beds are giving way to diverse native
(Steneck 1982; Steneck and Dethier 1994), but they   understoryassemblages (Vavrinec2003) or to mo-
lacked the spatial complexity of kelp forests and    nocultures of the nonnative bushy green alga Co-
other erect macroalgae that create habitat for ani-   diumfragile (Levin and others 2002). These under-
mals, such as amphipods (Hacker and Steneck       story assemblagesare much denser than kelp beds
1990) and some predatoryfish (Levin 1994). Thus,    (as in phase 1), occupy more of the substrate, and
the coastal benthos had become a good nursery      may further exclude urchins (Levin and others
ground for sea urchins (McNaught1999) but a poor    2002).
nursery habitat for their predators (Levin 1994).
Although this phase did not persistas long as phase  TROPHIC-LEVELDYSFUNCTIONAND FOOD
1, it appearsto have been stable along much of the   WEB INSTABILITY
coast for at least 2 decades, which is more than the
average 15-year life span of the green sea urchin   In all phases, stability was lost when population
(Vadasand Beal 1999).                 densities within a structuring trophic level were
                            reduced below a controlling threshold. Although
Phase 3: Predatory InvertebratesDominate        each speces was extant within each trophic level,
Fishing of the green sea urchin began in 1987      their individual or collective population densities
(Stenek and Carlton 2001) and quickly depleted     could no longer demographicallylimit lower tro-
populations from vast areas along the coast of     phic levels (trophic-level dysfunction had oc-
Maine, thus destabilizingphase 2. When the popu-    curred). Thus, the decline in the abundance of
lation of urchins dropped below a threshold bio-    predatory finfishes in coastal waters destabilized
mass (McNaught 1999; Vavrinec 2003), their graz-    phase 1, releasingsea urchin and other invertebrate
                                     R. S. Steneck and others    329

populations (phase 2). Likewise,declines in sea ur-   consumption of prehistoricpeople. Studies of Na-
chin abundance and the grazing rate led to the     tive Americanmiddens indicatethat the firstcoastal
reestablishmentof algal beds (phase 3).         populations to live on the Maine coast, datingback
  The rapidity of recent phase shifts may have re-   to 4500 ybp, ate primarilymarineorganisms(Spiess
sulted from the inherently low species diversity in   and Lewis 2001; Carlson 1986; Bourque 1995,
this system. Decreased diversity, especially within   Bourque 2001). Fish bones dominate some coastal
functional groups or trophic levels (for example,    middens, with as much as 70%-85% of the bone
Naeem and others 1994; Tillman 1996; McGrady-      mass comprisedof cod bones (Carlson1986; Spiess
Steed and Morin 2000, but see Hairstonand others    and Lewis 2001). In addition, isotope analyses of
1968; Austin and Cook 1974), and strong interac-    nitrogen and carbon from human bones reveal "a
tions of species between trophic levels (Polis and   high intake of marine protein (flesh) in their diets"
Strong 1996; Nuetal and others 2002) can destabi-    (Bourque 2001).
lize ecosystems. Most trophic levels in the Gulf of    Fisheries landings for each species, estimated
Maine consist of only one or a few strongly inter-   from bone and shell fragments and weighted by
acting species (Figure 1). Declines in those species  fractionaltrophic level (see Figure 2 for methods),
can result in trophic-level dysfunction,with cascad-  reveal that predatoryfishes, especiallyAtlanticcod,
ing effects to lower trophic levels. Such naturally   dominated landings for thousands of years (Figure
low-diversity ecosystems may presage the fate of    2A and Table 1). During phase 1, the mean TL
more diverse systems in which anthropogenic re-     ranged between 3.25 and 4.25 (Figure 2A). The
ductions in biodiversity are contributing to the    highest TL was estimated in the oldest portion of
growing frequency of "catastrophic"   phase shifts  the midden, which dates to more than 4000 ybp
(sensu Scheffer and others 2001).            and then declines marginally,indicatingthat even
                            indigenous people were affectingcoastalfood webs.
Temporal Trends in Trophic Levels: The         Other studies in North Pacific coastal ecosystems
Fisheries Signal                    have shown early human-induced phase shifts oc-
Several recent studies offered a global perspective   curring 8,000 to 10,000 years ago (Simenstadand
on the effects of fisheries. Fromthose studies, there  others 1978; Erlandsonand others 2004).
emerged the concept of "fishingdown marine food      In 1927, the mean TL, based on published fish-
webs" (Pauly and others 1998). In the preceding     eries reports, was 3.99 (Rich 1929) (Figure 2A).
sections, we qualitatively reconstructed the food    Unquestionably, large predatory fishes were still
webs of the western North Atlanticusing both fish-   abundant in coastal systems, as evident in mapped
eries-dependent and independent evidence. Below     fishing areasfor each species (quantifiedin Steneck
we revisit the Gulf of Maine ecosystem looking for    1997). In fact, the five species harvestedin greatest
quantitativeevidence of fished-down trophiclevels    number in coastal zones at that time were cod,
and using only fisheries-dependent data. For this    haddock, hake, cusk, and pollock (Rich 1929). This
analysis, we will apply the fractionaltrophic-level   deviation from the previous decline in TLmay have
approachpioneered by Pauly and others (Paulyand     been due to improvements in fishing practices(for
others 1998, Pauly and others 2001, Pauly and      example, trawlers) rather than a change in ecosys-
others 2002).                      tem structure.
 Trophic-level (TL) analysis assigns numbers to      In recent decades, the mean TL,as derivedfrom
species that indicate where on a trophic pyramid    inshore commercialfisheries landings,has declined
the organism feeds, based on studies of its diet.    (Figure 2). During most of phase 2 the mean har-
Thus, autotrophs,such as the algae at the bottom of   vested TL remained relatively high (around 3.2),
the food web, are assigned a TL of 1; herbivorous    but it droppedprecipitouslytoward the end of this
sea urchins feeding on algae are TL 2; fishes or    period (Figure2B). Between phase 2 and phase 3,
invertebratessuch as lobsters and sea stars feeding   landings for 23 of the 28 speces fished in coastal
on sea urchins are around TL 3; and higher-order    zones declined (DMR 1971-2000; NMFS
carnivorousfish whose food is a mixture of herbi-    1971-2000). Most of the harvested stocks (that is,
vores and small carnivoresrange from TL3.5 to TL    22 of the 28 spedes) declined 50% or more. De-
4.6. Fractionaltrophic levels for the dominant spe-   clines were reported for all of the importantpred-
cies of coastal Maine were obtained from published   atory fishes, such as cod, wolffish, hake, and floun-
studies (Froese and Pauly 2002) are are listed in    der, which were already at low levels duringphase
Table 1.                        2. During this phase shift, landings increasedonly
 We appliedfractionaltrophic-levelanalysisto ar-    for lobsters, sea urchins, sea cucumbers,mussels,
chaeological studies to characterizethe patterns of   and skates. Of these, only the lobster had been
330      AcceleratingTrophic-levelDysfunction

A 4.50'              Phase1             targeted as a prime fishery species throughout
                                 phase 2. The emergence of the sea urchin fisheryin
  4.25                             the latter part of phase 2 is reflectedin the decreas-
  4.00                             ing TL (Figure 2B) and ultimately resulted in the
                                 termination of phase 2. In phase 3, the TL has
> 3.75
                                 ranged from 3.0 to 2.7, probablydue to the collapse
  3.50                             of the urchin fishery and a shift in fishing pressure
                                 toward other herbivores (for example, sea cucum-
  3.25                             bers) and primaryproducers.Significantly,in 2001,
  3.00                             marine algae were included by Maine'sDepartment
                                 of Marine Resources in the "fisheries"  landings.
  2.75
                        Phases & 3
                           2       The overall TL decline we report for coastal
                                 Maine over the past 3 decades is similar to that
  205 00 4500 4000 3500 3000 2500 2000 1500 1000 500 0
              YearsBeforePresent          published for Canada (Pauly and others 2001). The
                                 TL decline from 3.4 in 1970 to 2.7 in 2000 (Figure
B          Phase2
                 >    <
                        3
                      Phase
                        >
                                 2B) for Maine is about the same as that for the east
      <
      I
                                 coast of Canada, which decline experienced a TL
                                 from 3.4 to 2.9 over roughly the same time period
                                 (Pauly and others 2001). Such cross-regional
                                 trends, derived from entirely different data sets,
                                 imply that these reductionsin TLare not artifacts  of
                                 the fisheries data.


                                 CONCLUSIONS
                                 Based on data from a number of sources, we con-
                                 dude that fishers in the Gulf of Maine have fished
           1980 1985                 down the food web for millennia. Serial targeting
             Years                 and depletion of abundant top consumers has re-
           trendsin trophic  levels (TL) har-
                           of     peatedly led to trophic-level dysfunction (that is,
Figure2. Temporal                        functional loss of a trophic level), creating trophic
vested speces over the past 4,300 years showingthe        cascadesthat changed the structureand function of
duration the threephases.A Entire
     of               record TLanal-
                         of
             studies the recent
                 to          Area   the ecosystem. Each new phase shift appearsto be
ysisfromarchaeological             period.
                                                          the
withinthe box on the rightincludesphases2 and 3. B        constantuntil fishingpressureagain destabilizes
Expanded  recenttemporal  scaleshowsphases2 and 3,      controlling trophic level. The phase shifts may oc-
withthe rapid  declinein TLsince1970.Thetrophic   levels   cur due to the inherently low diversity in the re-
of fishedspeces frominshorewatersof Maine(0-3 mil)        gion, which renders food webs into food chains. If
weredetermined   frompublished  studies(Pauly oth-
                          and     so, our observationsin Maine may be predictiveof
ers 2001). Theabundances harvested
               of       speces priorto   the fate of other, more diverse systems in which
recordkeepingwere estimatedas the percentof total        fishing successively targetsmost or all of the strong
boneandshellfragments eachsubtidal
              of         marinespedes    interactorswithin upper trophic levels. Therefore,
excavatedfrom Native American     middens(Spiessand     the changes occurringin the Gulf of Maine may be
Lewis2001).In 1927,landings wereintegrated
                 data          with
                  in     Maine(Rich    instructive for managers seeking to understand
detailed mapsof fishinggrounds coastal              what effects reductions in biodiversitymay have in
1929). Landings each speces were calculated the
          of                 as
                                 their own systems.
percentof all coastalfishinggrounds.   Standardized state
of Mainelandingsdata were used from 1971 to 2000
(DMR  1971-2000).Totallandings   caught inshorewa-
                       in         ACKNOWLEDGMENTS
tersforeachspecies estimated
           was      usinga ratioof inshore   We thank Tim McClanahanfor inviting us to write
to totallandings generated fromfederal  landings statistics
                                 this paper and for his helpful critique.The research
(NMFS  1971-2000).Whena fishedspedeswasnot listed
in the federallandingsdata (NMFS    1971-2000),it was    represented here has been funded by several
considered be caughtinshore.
       to                         sources, including the Pew Foundation for Marine
                                 Conservation, NOAA's Sea Grant program to the
                                 University of Maine (to R.S.S.), Maine's Depart-
                                 ment of Marine Resources (to R.S.S.), and the Na-
                                            R. S. Steneck and others      331


tional Undersea Research Program's National Re-           of discovery:examining the past in island environmentsNew
                                   York:Praeger.p 51-83.
search Center at the University of Connecticut at
                                  Estes JA, Duggins DO, Rathbun GB. 1989. The ecology of ex-
Avery Point (grants no. NA46RU0146 and UCAZP             tinctions in kelp forest communities.ConservBiol 3:252-64.
94-121 to R.S.S.). Our colleagues shared insights
                                  Froese R, Pauly D, editors. 2002. FishBase.Availableonline at:
and unpublished data for several kelp forest ecosys-         www.fishbase.org,20 September2002.
tems worldwide and assisted us with analyses. In          GilbertJR, GuldagerN. 1998. Status of harbor and gray seal
particular, we thank Susie Arnold, Chantale Begin,          populations in northern New England. Woods Hole (MA):
Jim Estes, Michael Graham, Jeremy Jackson, Doug           NationalMarineFisheriesService.
McNaught, Daniel Pauly, Bob Scheibling, Bob Va-          HackerSD, SteneckRS. 1990. Habitatarchitecture the abun-
                                                           and
das, our teams of summer interns, and two anony-           dance and body-size-dependenthabitatselection of a phytal
mous reviewers. To all we are grateful.               amphipod.Ecology 71:2269-85.
                                  HairstonNG, Smith FE, SlobodkinLB. 1960. Communitystruc-
                                   ture, populationcontrol, and competition.Am Nat 94:421-5.
REFERENCES
                                  HairstonNG, Allan JD, Colwell RK, Futuyama DJ, Howell J,
Adey WH, Steneck RS. 2001. Thermogeography    over time cre-   Lubin MD, MathiasJ, et al. 1968. The relationshipbetween
 ates biogeographic regions: a temperature/space/time-inte-     species diversityand stability:an experimentalapproachwith
 grated model and an abundance-weightedtest for benthic       protozoaand bacteria.Ecology 49:1091-101.
 marine algae. J Phycol 37:677-98.                HarroldC, Reed DC. 1985. Food availability, urchingrazing,
                                                        sea
Austin MP, CookBG. 1974. Ecosystemstability: resultfrom an
                        a          and kelp forest community structure.Ecology66:1160-9.
 abstractsimulation.J TheoretBiol 45:435-58.           HerveyAB. 1881, Sea mosses: a collector'sguide and an intro-
Bigelow HB, SchroederWC. 1953. Fishes of the Gulfof Maine.      ductionto the study of marinealgae. Salem (MA):SE Cassino.
 Fish Bull 53:1-577.                       JacksonJBC,KirbyMX, BergerWH,Bjordal KA,BotsfordLW,
BourqueBJ. 1995, Diversityand complexityin prehistoricmar-      BourqueBJ, BradburyRH, et al. 2001. Historical  overfishing
 itime societies:a Gulfof Maineperspective. New York:  Plenum   and the recent collapse of coastal ecosystems. Science 293:
 Press.                               629-38.
Bourque BJ. 2001, Twelve thousand years: American Native      Johnson DS, SkutchAF. 1928. Littoral  vegetationon a headland
 Americans in Maine. Lincoln (NE): University of Nebraska      of Mt. Desert Island,Maine. I. Submersible strictlylittoral
                                                         or
 Press,pp 368.                           vegetation.Ecology 9:188-215.
BreenPA, Mann KH. 1976. Changinglobsterabundanceand the      KeatsDW, Steele DH, South GR. 1986. Atlanticwolffish (Anar-
 destructionof kelp beds by sea urchins.Mar Biol 34:137-42.     hichaslupusL.; Pisces:Anarhichidae)predationon green sea
Carlson CC. 1986. Maritime catchment areas: an analysis of      urchins (Strongylocentrotusdroebachiensis Mull.); Echino-
                                                       (O.F.
                                   dermata:Echinoidea)in eastern Newfoundland.Can J Zool
 prehistoricfishing strategiesin the Boothbayregion of Maine
 [thesis]. Orono (ME):Universityof Maine.              64:1920-5.
Chapman,ARO (1981) "Stability sea urchin dominatedbar-
                  of               KennellySJ. 1989. Effectsof kelp canopieson understory species
                                   due to shade and scour. Mar Ecol ProgrSer 50:215-24.
 rengroundsfollowing destructivegrazingof Kelp in St. Mar-
 garet'sBay, EasternCanada."  MarBiol 62: 307-311        KnowltonN. 1992. Thresholdsand multiplestablestatesin coral
     ARO. 1984. Reproduction,  recruitmentand mortality    reef communitydynamics.Am Zool 32:674-82.
Chapman
 in two species of Laminaria southwest Nova Scotia. J Exp
                in                 KonarB. 2000. Seasonal inhibitoryeffects of marine plants on
 Mar Biol Ecol 78:99-109.                      sea urchins:structuring communitiesthe algalway. Oecologia
                                   125:208-17.
ChapmanARO. 1986. Age vs. stage an analysis of age-specific
 and size-specificmortalityand reproduction a populationof
                       in          KonarB, Estes JA. 2003. The stabilityof boundaryregionsbe-
 Laminaria       J
       longicruris. Exp MarBiol Ecol 97:113-22.        tween kelp beds and deforestedareas.Ecology84:174-85.
Collins JW, Rathbun R. 1887. Fishing grounds of the eastern    LambI, Zimmerman 1964. Marinevegetationon CapeAnne,
                                             M.
 coast of North America. In: Goode GB, editor. The fisheries    Essex County,Massachusetts.Rhodora66:217-54.
 and fishingindustriesof the U.S. Sectionm Washington(DC):    LelandAV. 2002. A new apex predatorin the Gulf of Maine:
 US GovernmentPrintingOffice.p 247-313.               crabs (Cancer borealis)control benthic community structure
Conkling PW, Ames T. 1996. Fishing grounds of the eastern      [thesis]. Orono (ME):Universityof Maine.
 coast of North America. In: Platt DD, editor. Penobscot:the   LevinPS. 1994. Small-scalerecruitmentvariationin a temperate
 forest,riverandbay Rockland(ME):IslandInstitute.p 46-65.      fish: the roles of macrophytesand food supply. EnvironBiol
Connell JH, Sousa WP. 1983. On the evidence needed to judge     Fish 40:271-81.
 ecologicalstabilityor persistence.Am Nat 121:789-824.      Levin PS, CoyerIA, PetrikR, Good TP. 2002. Community-wide
Dayton PK. 1975. Experimentalstudiesof algal canopy interac-     effects of non-indigenousspecies on temperaturerockyreefs.
 tions in sea otter-dominated kelp community at Amchitka      Ecology83:3182-93.
 Island,Alaska.Fish Bull 73:230-7.                LoweryLF,PearseJS. 1973. Abalonesand sea urchinsin an area
[DMR] Department of Marine Resources (ME). 1971-2000.        inhabitedby sea otters.Mar Biol 23:213-9.
 Maine fisheries landings. Augusta (ME):Departmentof Ma-     MacFarlane 1952. A surveyof certainseaweedsof commercial
                                        C.
 rine Resources.                          importancein southwest Nova Scotia. Can J Bot 30:78-97.
ErlandsonJM, RickTC,VellanowethRL.2004. Humanimpacts        MalpassWJ. 1992. The ecology and resource use patterns of
 on ancient environments: a case study from California's      small,benthic,predatoryfishes in rockynearshorehabitatsin
 northern ChannelIslands.In: Fitzpatrick editor.Voyages
                      SM,            the Gulfof Maine [thesis]. Orono (ME):Universityof Maine.
332    Accelerating Trophic-level Dysfunction

May RM. 1977. Thresholdsand breakpointsin ecosystemswith      of North Americanorth of CapeCod. TransConn Acad 5:27-
 a multiplicityof stable states. Nature269:471-7.         138.
McGrady-Steed MorinPJ.2000. Biodiversity,
        J,               densitycompen-  Sousa WP,ConnellJH. 1985. Furthercommentson the evidence
 sation, and the dynamics of populations and functional      for multiple stable points in natural communities. Am Nat
 groups. Ecology 81:361-73.                    125:612-5.
McNaught DC. 1999. The indirect effects of macroalgae and     SpiessAE, LewisR. 2001. The TurnerFarmfauna:five thousand
 micropredation the post-settlementsuccessof the green sea
          on                        years of hunting and fishing in Penobscot Bay, Maine. Occ
 urchin in Maine [dissertation]. Orono (ME): University of     Publ Maine Archeol 11:1-120.
 Maine.                             Steneck RS. 1982. A limpet-corallinealga assocation: adapta-
Naeem S, ThompsonLJ, Lawler SP, Lawton JH, Woodfin RM.       tions and defenses between a selectiveherbivoreand its prey.
 1994. Decliningbiodiversitycan alterthe performanceof eco-    Ecology 63:507-22.
 systems. Nature 368:734-7.                   Steneck RS. 1997. Fisheries-inducedbiological changes to the
Neutal A-M, HeesterbeekJAP, de Ruiter PC. 2002. Stabilityin     structure and function of the Gulf of Maine ecosystem In:
 real food webs: weak links in long loops. Science296:1120-3.   Wallace GT, BraaschEF, editors. Proceedingsof the Gulf of
[NMFS] National Marine Fisheries Service (US). 1971-2000.      Maine EcosystemDynamicsScientificSymposiumand Work-
 Fisheriesof the United States. Washington (DC):US Govern-     shop Hanover (NH):RegionalAssociationfor Researchon the
 ment PrintingOffice.                       Gulf of Maine. p 151-65.
Novaczek I, McLachlanJ. 1986. Recolonizationby algae of the    Steneck RS. 2001. FunctionalgroupsIn: Levin S, editor. Ency-
 sublittoralhabitat of Halifax County, Nova Scotia, following   clopedia of biodiversity;vol 1 New York:Academic Press. p
 the demise of sea urchins. Bot Mar 29:69-73.           121-39.
Paine RT. 1966. Food web complexityand species diversity.Am    Steneck, RS, Carlton,JT (2001b) Human alterationsof marine
 Nat 100:65-75.                          communities: Students beware! In: Bertness, M., Gaines, S.,
Paine RT. 1980. Food webs: linkage, interaction strength and    Hay, M. (editors),Marine communityEcology,SinaverPress,
 community infrastructure. Animal Ecol 49:667-85.
               J                   Sunderland,MA
Pauly D, Christensen Dalsgaard FroeseR, TorresF Jr. 1998.
           V,      J,               Steneck RS, Dethier MN. 1994. A functionalgroup approachto
 Fishing down marine food webs. Science 279:860-3.         the structureof algal-dominated  communities.Oikos69:476-
Pauly D, PalomaresR, Froese P, Sa-a M, VakilyD, PreikshotS,     98.
 WallaceD. 2001. Fishing down Canadianaquaticfood webs.     SteneckRS, GrahamMH,BourqueBJ, CorbettD, Erlandson     JM,
 Can J Fish Aquat Sci 58:1-12.                   Estes JA, Tegner MJ. 2003. Kelp forest ecosystems:biodiver-
Pauly D, ChristensenV, Guenette S, Pitcher TJ, Sumaila UR,     sity, stability,resilienceand future. EnvironConserv29:436-
 Walters CJ, Watson R, et al. 2002. Towardssustainabilityin    59.
 world fisheries.Nature 418:689-95.               TillmanD. 1996. Biodiversity:population versus ecosystem di-
Petraitis LathamRE. 1999. The importanceof scale in testing
     PS,                            versity. Ecology 77:350-63.
 the origins of alternativecommunity states. Ecology 80:429-   VadasRL Sr., Beal BF. 1999, Temporaland spatialvariability  in
 42.                                the relationshipsbetween adult size, maturityand fecundityin
Polls GA, StrongDR. 1996. Food web complexity and commu-      green sea urchins:the potential use of a roe-yield standardas
 nity dynamics.Am Nat 147:813-46.                 a conservationtool. Augusta (ME):Maine Departmentof Ma-
RichWH. 1929. Fishinggroundsof the Gulfof Maine. Bureauof      rine Resources.
 FisheriesDocument 1959. p 51-117. US GovernmentPrinting     VadasRLSr., Steneck RS. 1988. Zonationof deep water benthic
 Office,WashingtonD.C.                       algae in the Gulf of Maine. J Phycol 24:338-46.
RosierJ. 1605, The true relations of Capt.GeorgeWeymouth's    VadasRL Sr., Steneck RS. 1995. Overfishingand inferencesin
 voyage. London:GeorgeBishop.                   kelp-sea urchin interactions.In: SkjoldalHR,HopkinsC,Erik-
SalaE, GrahamM. 2002. Community-widedistributionof pred-      stad KE, Leinaas HP, editors. Ecology of fjords and coastal
 ator-prey interactionstrength in kelp forests. Proc Natl Acad   waters Netherlands:Elsevier.p 509-529.
 Sci US 99:3678-83.                       VavrinecJ. 2003. Resilienceof green sea urchin (Strongylocentro-
Sala E, BourdouresqueCF, Harmelin-Vivien 1998. Fishing,
                        M.          tusdroebachiensis) populationsfollowingfishingmortality:ma-
 trophiccascades,and the structureof algal assemblages:eval-    rine protected areas, larval ecology and post-settlementsur-
 uation of an old but untested paradigm.Oikos 82:425-39.      vival [dissertation].Orono (ME):Universityof Maine.
Scheffer M, CarpenterS, Foley JA, Folke C, Walker B. 2001.    Vermeij GJ. 1991. Anatomy of an invasion: the trans-Arctic
 Catastrophic shifts in ecosystems. Nature 413:591-6.       interchange.Paleobiology17:281-307.
ScheiblingRE,HatcherBG. 2001. The ecology of Strongylocentro-   VerrillA. 1871. Marine fauna of Eastport,ME. Bull Essex Inst
 tus droebachiensis LawrenceJM, editor. Ediblesea urchins.
          In:                       8:1-6.
 Netherlands:Elsevier.Sd. p 271-306.               Witman JD. 1985. Refuges, biological disturbance,and rocky
Simenstad CA, Estes JA, Kenyon KW. 1978. Aleut, sea otters,     subtidalcommunity structurein New England.Ecol Monogr
 and alternatestable state communities.Scence 200:403-411.     55:421-45.
Smith J. 1616, The description of New England. London: H.     Witman JD, Sebens KP. 1992. Regionalvariationin fish preda-
 Lownes.                              tion intensity: a historicalperspective in the Gulf of Maine.
Smith SI. 1879. The stalk-eyedcrustaceansof the Atlantic coast    Oecologia90:305-15.