The University of Maine
DigitalCommons@UMaine
Marine Sciences Faculty Scholarship
School of Marine Sciences
7-1-1990
Inference in Ecology: The Sea Urchin
Phenomenon in the Northwestern Atlantic
R. W. Elner
Robert Vadas
University of Maine - Main, [email protected]
Follow this and additional works at: http://digitalcommons.library.umaine.edu/sms_facpub
Repository Citation
Elner, R. W. and Vadas, Robert, "Inference in Ecology: The Sea Urchin Phenomenon in the Northwestern Atlantic" (1990). Marine
Sciences Faculty Scholarship. Paper 60.
http://digitalcommons.library.umaine.edu/sms_facpub/60
This Article is brought to you for free and open access by DigitalCommons@UMaine. It has been accepted for inclusion in Marine Sciences Faculty
Scholarship by an authorized administrator of DigitalCommons@UMaine.
The AmericanNaturalist
Vol. 136, No. 1
INFERENCE
July1990
IN ECOLOGY: THE SEA URCHIN PHENOMENON IN THE
NORTHWESTERN ATLANTIC
R. W.
ELNER AND R.
L. VADAS, SR.
Biological Sciences Branch, Departmentof Fisheries and Oceans, P.O. Box 550, Halifax,
Nova Scotia B3J 2S7, Canada; Departmentof Botany and Plant Pathology,
Universityof Maine, Orono, Maine 04469
SubmittedNovember 21, 1988; Revised July17, 1989; Accepted October 12, 1989
A fundamentalconcern in ecology is the absence of a consensus on how to
"do" this science. Over the past few years, diametricallyopposed treatiseshave
been offeredon establishingground rules both for how to conduct ecological
research and forwhat constitutesevidence to supportthe most basic ecological
premises (see Salt 1983). The appeals of Platt (1964), Strong(1980, 1983), and
Simberloff(1983) forreplicatetesting,the exclusion of null hypotheses,and the
use of "strong inference" to advance ecological understandingare compelling
because they have proved effectivein other sciences. Nonetheless, Quinn and
Dunham (1983) presented convincing argumentsagainst the use of rigid hypothetico-deductivereasoning in ecological research. Similarly,Roughgarden
(1983) attemptedto discreditthe "strong-inference"methodology,arguingfora
"commonsense" approach using simple systemmodels to build understanding.
Salt (1983) seeminglyadded to the dilemma by attemptingto rationalize the
various ecological research procedures. More recently,Slobodkin (1986) advocated working on the simplest manifestationsof problems because of the
extremenatureand complexityof the "big questions" in ecology.
Althoughsuch debate has failed to resolve the fundamentalconcern, it has
served to highlightthe lack of consensus on problemsolvingamong grass-roots
ecologists, who are not philosophers. Althoughsome of the dissentionexists
because even the basic definitionand goals of ecology remainvague and some of
the questions are "too big" to adequately focus upon (Slobodkin 1986), there
ecological research.
appears a need formore rigorin conductingand interpreting
We contendthat,because of the lack of rigorand guidelines,thetendencyto rely
on "weak" inferenceand "common sense" is pervasive,resultingin a propensity
to cling uncriticallyto "pet" concepts ratherthan to test multiplehypotheses
(Chamberlin1897). Such a softapproach has promotedcircularreasoning,aided
in the developmentof paradigms,and retardedecological discovery.
In thispaper, we criticallyanalyze the ecological approach and the interpretation of the resultingdata used in a series of related publicationsto explain the
alternationbetween macroalgal beds and coralline barrensin the sublittoraloff
Am. Nat. 1990. Vol. 136, pp. 108-125.
? 1990 by The Universityof Chicago. 0003-0147/90/3601-0011$02.00.
All rightsreserved.
INFERENCE IN ECOLOGY
109
Nova Scotia, Canada. In the process, we show thatmanyof the studiesreliedon
weak inference,whichcontributedto thedevelopmentof a keystoneparadigmfor
the Americanlobster. We also contendthatthisparadigmhas been retainedby a
series of ad hoc revisions ratherthan experimentation.We express these views
withtrepidation,realizingfullwell thata posterioricriticismis easier thandesigning and carryingout crucial experiments.However, as expressed by Underwood
studies,our criticismis
and Denley (1984) in theirexaminationof rocky-intertidal
intendedto fosterecological progressand not to denigratethose who workedon
this marinesystem.
We do not suggestthatour example is more extremethanotherproblemareas
in ecology; indeed, thisresearchprovoked much discussion and introducedseveral innovativetechniques.However, despiteconsiderableresearchefforts,some
fundamentalanswers to the sea urchinphenomenonremainas elusive today as
they were 15 years ago. Recent commentsby Keats (1986), Breen (1987), and
thequestionableprogressin theresearch.In response
Miller(1986, 1987)highlight
to a perspective by Miller (1985a) that questioned the impact of predatorsin
controllingsea urchinpopulations, Keats argued that we are back to "square
one" concerningthe actual and potentialrole of predation; Breen argued the
experimentalapproach (Connell 1974;
opposite. We believe thata more-rigorous
observations,would have yielded
Paine 1977), based on strongernatural-history
greateradvances in understandingthe ecology of this nearshoresystem.
NATURE AND SOURCE OF THE EVIDENCE
In the early 1970s, explanationswere soughtfora populationexplosion of sea
urchinsand the resultingintensivegrazingthatconvertedmacroalgalbeds into
corallinebarrensalong the Atlanticcoast of Nova Scotia (Mann and Breen 1972).
Concomitantly,commerciallandingsof the Americanlobsterhad declined dramatically,suggesting,withthe hypothesisthatlobsterswere keystonepredators
(in the sense of Paine 1969), a causal relationshipbetween these events. An
analogous explanation, with sea otters acting as the keystone predator,was
advanced to explain differencesin the communitystructureoffAlaska (Estes and
Palmisano 1974; Estes et al. 1978).
Our review examines core papers and one abstract (through1985) directly
relatingto the lobster-sea urchinproblem. To ascertainthe flowof researchspecifically,the questions posed, the ecological approaches used, and the interpretationsmade-we analyzed these publicationschronologically(table 1) accordingto the followingcriteria.What was the purpose or objective(s) of each
study?Did the studyrelyon weak or stronginferences(in thesense ofPlatt1964)?
category,we consideredwhethernullor implicithypothFor the strong-inference
eses were advanced, whetherexperimentswere conducted, and, if so, whether
apcontrols were employed. Classified under weak inferencewere affirmative
proaches (i.e., those designedto prove hypotheses),literaturesurveys,and anecdotal and speculative reasoning.In addition,we surveyedthe sources of data or
informationused in each publication. The informationsources have been
categorized as laboratory experiments,physiologicalindexing(i.e., extending
-w
+
(I
+
v
z
0
Q:V
Cl.
4
+
++
0
+
+~
+
+
+
~~~~~~~
+
w
~~
~~~~~+
w
w
+
+
+
+
++
+l
+
++
+
+
+++
z
z
0
z-
-
t zz+
UQ
F-
0 z
r.
w
cq~
44.
0
>
12.
0~~~~~~
cq re)
o~~o~
10
~~~~~0
~~
~
~
~
0~
0c r
o~(orON
r
U
+~~~~~~
Cl.~~~~~~
+
+
+
b+
+
0CO
CO
+
+
+
+
+
++
+
+
+
+ +
+
++
E
+
cl-
++
0
CO
<
Y
<
.E
+
+
0
C.
++
+
+
+
+
+
Cl. Cv.
++
+
+
E
U
m
<
zE
m
.0
co
s0.>=S
;
Fm
+
U,
+
a,
o
Do
IV0
0
v:;
CZ
o
U!
0>~~~~~~
coc
.0
X
CZ 0
0
0
-o
XZ
co t%.;
0
C
en
t
rm
?
0
cO0
U,
__
V.
X
ON~~~O00
aN
08
>
0
0
-
It
CZ
_
C:
b
0
O
_
~
_
.
_
_4 00
~
OlC
_
~
~N
N
^
-
CZ
06 C~ 6
r-~~
*d
U
co
eO
*_
u
U-
CO0
011
1-1u~~0
t
>
N
CO00
.ON
lCl
00
ON
-
0
~~~~ON co
diN
ON-7.
E
m
+-,
0
0~~
o
00
00
;U
)
ur
m
.D_
0
Cl
U
111
112
THE AMERICAN NATURALIST
physiologicaldata to fieldsituations),fieldsurveys,fieldexperiments,and simulation models (see table 1).
Of the 23 papers, at least 7 addressed the advisoryor managementimplications
of overfishing
lobstersand the resultingchanges in the nearshoreecosystemdue
to the sea urchinpopulationexplosion. Another4 focused on trophicrelations,
includingthose limitinglobster production.Four others discussed the creation
and persistenceof barrengroundscaused by sea urchins.Behavioral aspects of
interactionswere examinedin 5 papers. Sea urchinenergeticsand
predator-prey
or disease factorswere covered in theremaining
systemcontrolby environmental
3 papers.
Eight papers used an experimental(laboratoryor field)approach (7 reported
controls), whereas 11 were based primarilyon the literatureand speculative
comments(table 1). Four otherswere also based on literatureand speculation,as
well as eitherphysiologicalindexingor fieldsurveys.Therefore,almosttwo-thirds
approach. Although
of thepublicationsin thisseries representa nonexperimental
it was difficult
to evaluate the natureof the hypothesespresented,implicittests
seemed apparent in 10 papers, but only 1 paper had a clearly stated null hypothesis. The remainingpublicationsusuallyhad a statedpurposeand reasonably
or generalnature.
clear objectives, but these were oftenof a socio-environmental
conditionof
"The purpose of this paper is to draw attentionto the deteriorating
thelobster-producing
systemof AtlanticNova Scotia and to attemptto changethe
consensus of managementpersonnelfrom'not proven' to somethinglike 'seems
further'" (Whartonand Mann 1981,p. 1340). "This
plausible, worthinvestigating
paper reviews our knowledgeof the stronginteractionsin the kelp ecosystemof
the Atlanticcoast of Nova Scotia" (Mann 1982a, p. 415). "This paper deals with
a problem related to the managementof a valuable resource, the American
lobster.... The questions thatarise ... are at the interfacebetweenpopulation
dynamics and ecosystem theory.
. .
. I hope that this paper will help draw
attentionto the need to build bridgesbetween the two lines of advance" (Mann
and not scientifi1977, p. 455). Many of the stated purposes were confirmatory
cally testable.
In addition,affirmative
or supportinglanguagewas used in 11 papers and may
have been used in 6 others. In most papers (throughthe 1970s and early 1980s),
confirmation
was soughtforthekeystonerole oflobsters."The hypothesisis thus
supported by the evidence: lobsters appear likely as a key predator on sea
urchins" (Breen and Mann 1976b,p. 140). "These . .. factorswere exploredin a
computer model which verifiedthat it was reasonable to suppose that urchin
populations were controlledby lobsters" (Mann 1981, p. 6). The bias of this
approach, and the circularitythat it leads to, is also apparentin otherpapers:
"The intentionwas to lookfor evidence of kelp bed destruction,changes in sea
urchinpopulationsand changes in predatorpopulations,particularlyin stocks of
lobsters" (Mann 1982a, pp. 418-419); "We have thus witnesseda change in the
whole community,at all trophiclevels, triggeredby removalof a key predator"
(p. 421; italics added).
One furtherremarkis needed to show the ambiguoususe of hypothesesand
experimentalresults.In 3 of 10 papers in whichworkinghypotheseswere appar-
INFERENCE IN ECOLOGY
113
ent,resultsthatdid not supportthe hypothesiswere explainedaway. "The finding
in acceptance of the
that lobsters prefercrabs over urchins raises difficulties
theory... thatlobstersare thekeypredatorscontrollingsea urchinsin nature....
The experimentsindicatedthatthe preferenceof lobstersforcrabs is a real one,
even if in naturethe lobsters do not always have the opportunityto indulgeit"
(Evans and Mann 1977,p. 2206). "Sea urchinremainswere foundin only 2% of
crab stomachs, and the crabs showed a clear preferenceof mussels over
urchins.... We should, however, be cautious about jumpingto the conclusion
role in controllingurchinpopulations....
thatrock crabs do notplay a significant
The conclusion, then, is that . . . they [lobsters] could be expected to exact a
significant predatory control over other invertebrate species, including . . . ur-
chins" (Drummond-Daviset al. 1982, pp. 638-639). Althoughthe contradictory
hypothesis(i.e., lobstersdo
resultsdo notabsolutelyrefutethekeystone-predator
not have to preferurchinsto all otherprey in order to act as a keystone,and,
similarly,crabs mightstillcontrolurchinswhere mussel prey are rare), theydo
highlightthe need for careful formulationof hypotheses and appropriateexperimentation.
The 31 proposed system components,definedas the ecological mechanisms
involved in explaininga scenario (in the 23 papers and 1 abstract),are analyzed
chronologicallyin table 2. As system components are introducedwith each
publication,the cumulativenumberof componentsidentifiedgrows. However,
dependingon the frequencywith which retractionsor omissions occurred,the
total numberof systemcomponentsactually invoked in explainingthe scenario
shows an irregularpatternof peaks and declines. Althoughthepatternitselfis not
necessarily unproductive,the changes, as discussed below, are not due to exof
perimentation.Seemingly,the patternreflectsa series of ad hoc reformulations
an early hypothesisthat urchinabundance controlsproductionin the nearshore
ecosystem.
The followingstatementsare the majorcomponentspresentedin these publications,givenchronologicallyand in thephraseologyoftheauthors:(1) lobsteryield
depends on seaweed productionforfood, (2) lobsteryield depends on seaweed
productionforshelter,(3) seaweeds are of trophicvalue to the nearshoreecosystem, (4) man is importantin degradingthe systemby overfishinglobsters, (5)
lobstersare keystonepredators,(6) systemchangesare cyclicphenomena,(7) fish
are importanturchin predators, (8) lobsteringis best in kelp habitats,(9) the
system will recover if lobsteringis stopped, (10) a managementapproach is
advocated for the total nearshore ecosystem, (11) crabs are importanturchin
predators,(12) sea urchinsare the preferredprey of lobsters,(13) sea stars are
importanturchinpredators,(14) a criticalsea urchindensityis requiredforthe
onset of destructivegrazing, (15) sea urchin aggregationleads to destructive
grazing,(16) sea urchin-createdbarrensare irreversiblestates, (17) predationin
generalis importantto the maintenanceof the system,(18) lobstersare important
only as complementarypredators,(19) kelp removalleads to faunalimpoverishcontrol,(21) driftseaweeds
ment,(20) lobsterabundance is underenvironmental
are importantin the system,(22) urchinsaggregatein given conditions,(23) sea
urchins deaggregate in given conditions, (24) a critical sea urchin density is
THE AMERICAN NATURALIST
114
TABLE 2
CHANGES IN THE PROPOSED SYSTEM COMPONENTS OCCURRING IN THE SUBLITTORAL
OFF NOVA SCOTIA, CANADA
Study
and
Yeara
1. 1971
2. 1972
3. 1972
4. 1973b
5. 1973
6. 1976
7. 1976
8. 1976
9. 1977c
10. 1977
11. 1978
12. 1978e
13. 1981
14. 1981
15. 1981
16. 1981
17. 1982
18. 1982
19. 1982f
20. 1982
21. 1983
22. 1984
23. 1985
24. 19859
No. of
New
System
Components
No. of
Previous
System
Components
Maintained
2
5
2
0
1
7
3
1
1
4
2
4
3
5
5
0
1
0
3
3
1
0
0
1
0
1
0
0
0
0
9
6
1
11
15
11
18
10
16
2
14
9
7
11
5
0
3
No. of
System
Components
Omittedor
Retractedfrom
Previous Paper
CumulativeNo.
of ComponentsIdentifiedin
System
No. of
Active
System
Components
1
3
NA
2
4
7
4
NA
3
5
NA
3
7
12
5
9
4
NA
7
9
3
2
7
2
7
9
NA
10
17
20
21
NA
21
22
NA
25
28
29
29
29
30
NA
31
31
31
31
31
2
6
6
NA
5
10
8
6
NA
8
7
NA
14
18
12
18
10
17
NA
15
9
7
11
5
NOTE.-NA, These papers are not applicable to this analysis.
a Published studies as in table 1.
b Energeticstudyof limitedscope and thus not amenable to these analyses.
c Laboratoryfeedingstudyof limitedscope and thus not amenable to these analyses.
d
Increases in the numberof active systemcomponentswithoutthe additionof new components
reflectthe reintroduction
of previouslydismissed components.
e Laboratory studyof limitedscope and thus not amenable to these analyses.
f Laboratorypredationstudyof limitedscope and thus not amenable to these analyses.
g Abstractand thus not fullyamenable to these analyses.
of defensiveaggregations
requiredforthe initiation
againstpredators,
(25) sea
of theecosysteminurchinbehaviorvariesseasonally,(26) spiraldegradation
volves a feedback mechanism,(27) lobsterproductionalong easternNova Scotia
was influenced
createdbytheCanso causeway,(28) ecological
by therestriction
fromothersystems,
inNova Scotiaareuniqueordifferent
conditions
(29) disease
controlssea urchinpopulations,(30) temperature
patternsand changescontrolthe
is required
theoretical
framework
and(31) an improved
abundanceofsea urchins,
to understand
ecosystemfunction.
Essentially,this series of publicationsand the ideas containedtherein,to
explaintheshiftfrommacroalgaeto barrens,can be brokenintofourchronological scenarios(figs.1-4): (1) a 3-4-yrcycle,catalyzedbylobsters;(2) an irrevers-
MACROALGALPHASET'
0
DECRE
PEDA
ONSNAUNICtIIIIS:B. _
INCIREAS
w
OBSERVATION
U IEECE
111OVERFISHING
LOBTE
FEWERLOBSTERBMI
INCREA I LOBSTER
BY DCREASED
\AUN
_
LOOBTEREPOTTO
SEA URCHI
~~OFF
NATURALLY/
\
-t
---\\DIE
SEA URCHIN
GRAZIGCAUSESHOLES
\ NKELP
CANOPY\/
BARRNS PHASETM
in
FiG. 1.-Proposed systemcomponents
duringscenario1 (about1970-1975)occurring
thesublittoral
offNova Scotia,Canada; 3-4-yrcycle(see thetables)betweensea urchindenudedbarrensand macroalgal
phases.
'MACROALGAL PHASE
D
~ ~
I RUN
SE
B
WPEASED
URCHIN
CE(INS
~~AUNAC
FORMTION OF SEA UCH
DEFENUVE AINIIATl"
H
DCESDECRE1ASED
o
rnwrClt
/
LO
N
RECRUTMEN
CW ~
Db
/~~
DECOWASO, RECRUIT"
*
A
IRVRIL
4
VCED
ll
//
~~LOO.tIs
DIE L NE M
LOBSTER AWNDAN
E|
FiG.2.-Proposed sequenceof systemcomponents
duringscenario2 (about1975-1978);
frommacroalgal
phaseto predation-induced,
irreversible-barrens
phase(see thetables).
PHAS1,ETM
ROAFi.LGA2L
' MAC
FROM
PREDATION
ILSIRFISNSRT/
LOUSIERRSHCRA
am ElIDING my
uRCHINDENS1TV
INRAIGSA
^l
nE
~SEAURCHMI
~ DE,ST~
~
E
ET a
A4^~~~~~~~~RN
ma vacuum
OFPREDATR
PRSNMCE
AGGREGATIONS
iF SEAURCHI
FORMATION
LOSTSR PREYONERHNSUNFE.C
I AOGG"TIOII-CATALYST Fru
4
DESTRUCTIVE EAZING RONTS
LOOT
w
^RE4
HASE'
SESTEAYNDNC
PROUCIVIY
DECLINE@
LOBSTER-ABUNDANCE
V/
duringscenario3 (about1978-1983);
components
FIG. 3.-Proposed sequenceof system
phase(see thetables).
induced,irreversible-barrens
frommacroalgalphaseto behaviorally
PHASET
'MACROALGAL
LOW WATER
TEMPEATUIRE
REETASVNENTi
L
ONE
FAVOLE
CONDITIONS
FORINCREASED
SEA URCHNABUNDANCE
RCED
GRAZ
e?f~~~~CYLE
SEA URCHINDISEASE
ST
ENCOURAGED
WATERTEMPERATURE
SNM
BARRN8 PHASE
duringscenario4 (1984-?);temperacomponents
FIG. 4.-Proposed sequenceof system
barrensand macroture-and disease-induced
cyclebetweensea urchin-denuded
15-20-yr
algalphases(see thetables).
INFERENCE IN ECOLOGY
117
ible change in the system,catalyzed by the removalof predators;(3) an irreversible change in the system,catalyzed by complex behavioralinteractions;and (4) a
10-15-yrcycle, catalyzed by extraneousevents. For scenario 1, a simple3-4-yr
cycle was proposed and relatedto theremovalof lobsters,thekeystonepredator,
whichkept sea urchinsin check. In scenario 2, thewhole systemwas statedto be
in a degradative spiral, catalyzed by decreased predationon sea urchinsand
simple behavioral responses (persistence of defensiveaggregations)by sea urchins. The same downward spiral was proposed in scenario 3, but thistimethe
catalystwas determinedto be a seriesofcomplexbehavioralinteractionsbetween
sea urchinsand several predators.Here, the sea urchinsincreaseas a resultof the
decline in lobster predation, but the subsequent formationof aggregationsis
of
reportedlyinducedby the "presence ofpredators."Additionally,theformation
even-largeraggregationsis representedas a result of lobsters' feedingon the
to
smallerdefensiveaggregations.Clearly,the presenceof predatorsis important
the formationof aggregationsin scenarios 2 and 3, but the paradox is thatthese
same predators are supposed to be in lower numbersthan when the sea urchin outbreak began. In scenario 4, a sea urchindie-offbroughtabout yet anothercycle, but thistimeof 10-15 yr. Here, sea urchinsincreaseduringadvantageous (low-temperature?)conditions,formaggregations,and intensivelygraze
kelp beds. The resultingbarrensituationpersistsuntilthe nexthigh-temperature
(>120C) period favorable to a lethal sea urchindisease. The cycle then begins
anew.
We emphasize thatthe abandonmentof scenarios 1 and 3 by the authorswas
not caused by testing the system components throughexperimentationbut,
rather,by the force of extraneousevents. Specifically,the barrensin scenario 1
persistedlongerthanthe postulated3-4-yrcycle, whereas the shiftto scenario4
was due to the sea urchindie-off.The reason forthechangefromscenario2 to 3 is
not readily apparent but may have been due to adverse criticismof the role of
predators(in particular,lobsters)in controllingurchins(Elner 1980; Pringleet al.
1982) and the authors' own experimentalwork(Bernsteinet al. 1983). In general,
we note thatimportantaspects of all fourscenarios (figs.1-4) have continuedto
be based on the "inference" categoryand are not tested. The othercategories
comprise "observation" fromfieldstudies and "mix," subjectivelyassessed as
and/or"obser"inference" coupled withsome degree of eitherexperimentation
vation."
DISCUSSION
Our review of the systematichistoryof researchintothe sea urchinphenomenon offNova Scotia suggeststhatprogressin elucidatingthe ecological mechanismsresponsibleforthese events has been thwartedby thefailureto use a more
scientificapproach. In fact, our review, althoughnot
effective,strong-inference
excluding the possibility that predation is involved in controllingsea urchin
populations, concludes that the evidence to date is not cause to dismiss what
should have been the originalnull hypothesis,"predatorshave no effect."
Initially,a singlehypothesisthatlobstersare a keystonepredatorwas advanced
118
THE AMERICAN NATURALIST
(Mann and Breen 1972; Breen and Mann 1976b) to account for the increased
abundance of sea urchins.The originalnotion,plus various systemcomponents,
was seductiveand remaineda virtuallyunchallengedparadigmuntila workshopin
1980 (Pringleet al. 1982) provided a forumin which studies of the lobster-sea
urchininteractionwere debated. The need forsuch a workshopwas fueledby the
purportedeconomic consequences of the barren grounds on valuable lobsterfisheryresources.
The establishmentof the keystoneexplanationas a paradigm(a pointofview so
dominantthatit makes otherapproaches to a disciplineappear irrelevant;Strong
1980) withoutadequate evidence servedto dividethe scientificcommunity.After
the 1980 workshop, researchersformedtwo camps: proponentsof a keystone
systemversus critics.Proponentscontinuallyurgedfisheriesmanagersto adopt
strategiescommensuratewiththe keystoneparadigm(Mann 1981, 1982a; Wharton and Mann 1981; Bernsteinand Mann 1982). Indeed, at one point,when the
ecosystem was judged to be in irreversibledecline, a proposal was made to
destroy urchins over large areas by using quicklime (Bernsteinand Welsford
1982). In hindsight,fisheriesscientistswere criticalenough not to accept the
natureof the evidence (Pringleet al. 1982)and, by virtueof eithertheirconservativenatureor theirinertia,were stillconsideringthesituation(Pringle1986)by the
timenaturalevents forceda reappraisalof factorscontrollingthe system.Meanwhile, the lobster was identifiedin textbooks(Barnes and Hughes 1982; Mann
1982b; Clark 1983; Andrewarthaand Birch 1984; Valiela 1984; Warner1984; Sze
1986), reviews (Hughes 1980; Paine 1980; but see Harrold and Pearse 1988),
scientificpapers (Duggins 1983a; Briscoe and Sebens 1988), and populararticles
(Duggins 1983b;but see Elner 1983)as a practicalexampleofa keystonepredator.
Unfortunately,popular thinkingis particularlydifficultto change because the
are usuallynotconsidmechanismsof, and evidence for,scientificinterpretations
ered by those outside the debate (Durant et al. 1989).
In review, this literaturestudyhighlightsfourkey areas of concern. First,as
tables 1 and 2 suggest,testingproceduresthatgave resultscontradictory
to theory
were explained away. The resultswere argued to have no directbearingon the
"real-world" situationand thereafter
ignored.Thus, the originalhypothesiswas
retained(see Evans and Mann 1977 and Drummond-Daviset al. 1982 regarding
the impactof lobstersand rock crabs on sea urchins).
Second, systemcomponentsof the scenarios tended to become ingrainedas
"truths" by repetitionover several publications,obscuringweaknesses in the
originalevidence. The role of fishas sea urchinpredatorsis a case in point.
Corollaryevidence between fishabundance, numberof urchinaggregations,broken test material,and exposed urchins,togetherwithlimitedobservationsfrom
fish guts, was argued by Bernstein et al. (1981) as suggestingthat intensive
predationfromfish,particularlywolffish,
mighthave destroyedsea urchinfeeding
fronts.Using the same evidence, Mann (1981) statedthatwolffishare capable of
breakingup aggregationsof urchinsand, further,that aggregationscan persist
numbersof predatoryfish.The importanceof fishas
only ifthereare insufficient
sea urchinpredatorswas repeatedin subsequentpapers. Moreover,thehypothetical role of fish predation as a system component was furtherentrenched;
INFERENCE IN ECOLOGY
119
oflobstersin
to overfishing
overfishing
of wolffishwas invokedas complementary
being instrumentalin the sea urchinpopulationexplosion (Whartonand Mann
1981). Clearly, the predatoryabilitiesof fishon sea urchinswere never tested;
moreover,Miller (1985a), in a review of predator-sea urchininteractions,failed
to demonstratethe importanceof fish,eitheras predatorsper se (but see Keats et
al. 1986, 1987) or as behavioral "catalysts" forthe destructionof aggregations.
Our thirdconcernis thatfidelityto the constituentsystemcomponentsappears
to have been independentof testing(table 2). Certain components, such as
the interactionsamong lobsters, crabs, and sea urchins,have been maintained
throughoutthe unfoldingof the various scenarios. However, othershave been
modified,perhaps as a resultof adverse criticism(Elner 1980; Pringleet al. 1982;
Miller 1985a). The role of lobstershas rangedfromthatof a keystonepredator,to
a complementarypredator,to an aggregationcatalyst. Similarly,the purported
relationshipbetween crabs and sea urchinshas been maintainedthroughseveral
forms,whereas markedlycontrastingexplanationshave been providedfor the
predatoryabilitiesof sea stars. Althoughactual testinghas had variouseffectson
hypotheses,the consequences of extraneousevents have been more conclusive.
phenomIn particular,the hypothesizedtemporalbasis of the macroalgal-barrens
first
grazing
was
after
intensive
cycle,
soon
from
a
3-4-yr
enon has varied
observed (early 1970s), to an irreversible,barren state a few years later, to a
longer, 15-20-yrcycle when macroalgae reestablished(early 1980s, afterwidespread sea urchinmortality;see thefigures).Clearly,thereis a dichotomyin logic;
butextraneous
were arbitrary,
modificationsas a consequence ofexperimentation
eventsforceda posteriorichangesofthescenarios.Althoughsuch nonexperimental evidence can be at least as powerfulas experimentaldata in refutinga
hypothesis,the reliance on naturalevents to compel scenario changes highlights
the comparativeimpotenceof the ecological researchemployedto elucidate the
scenarios.
Our fourthconcern,as illustratedby tables 1 and 2, is thatthe evolutionof the
systemcomponentsintoa burgeoningcomplexcreatesproblemswithplausibility.
Rather than being accepted or replaced, the originalsinglehypothesiswas doggedlyadheredto, manicured,and supplemented(see papers usingscenarios2 and
3). Additionally,problemsof logic and inconsistencyare apparentin some papers.
For example, Breen and Mann (1976b) presentedevidence forsea urchins'being
the major item in lobster diets, whereas Whartonand Mann (1981) argued that
lobsters suffera reduced food supplyon barrensdominatedby urchins.Hence,
we seriouslyquestion the ecological utilityof bothapproaches thatlack appropriate testsand the system"models" theycreate (see also Dayton and Oliver 1980).
Hypothesistestingappears to have been abandoned, and any evidence thatcould
not be ignoredwas incorporatedintothe systemcomponents.The latestscenario
of sea urchin control by disease, for example, was induced by increased sea
temperatureand/orlarge-scalemovementsof warmwatermasses (Scheiblingand
Stephenson 1984; Jonesand Scheibling1985) but has yetto be tested.In a recent
qualitativemodel of Nova Scotian kelp-beddynamics,Johnsonand Mann (1988)
maintainedthe urchin-diseasemechanismforthe succession frombarrenground
to kelp. However, for the switch back to the barrens,they postulatedthat the
120
THE AMERICAN NATURALIST
increase in urchin abundance is due to recruitingsuccess and/ordecreases in
urchinmortalityas a resultof predatorremoval.
On a positive note, it appears worthwhileto highlight
what is knownabout the
Atlantic.The urchinpopulalobster-sea urchin-kelpsystemin the northwestern
tionexplosion and resultingevents promptedmuchresearch,and variousecological techniques were either improved or developed to study the phenomenon
(Bernsteinand Welsford 1982, 1983; Vadas et al. 1986; Margosian et al. 1987).
Documentationof the perturbationsof the systemover time(Arnold 1976; Chapman 1981; Millerand Colody 1983; Miller 1985b;Michaud 1986; Scheibling1986;
Raymond and Scheibling 1987) has resulted in a comprehensivea posteriori
appreciationof the ecological events surroundingintensivegrazingepisodes and
the subsequent reestablishmentof kelp beds, if not the mechanical "trigger(s)"
forthe sequences. Furthermore,the core literaturehas provokeda largebody of
experimentalstudies (Himmelmanet al. 1983; Jonesand Scheibling1985; Miller
1985c; Vadas et al. 1986; Elner and Campbell 1987; Johnsand Mann 1987; Mohn
and Miller 1987; Johnsonand Mann 1988). These workers,whileprobingspecific
pointsfromthe core, have augmentedour knowledgeof the systemand invoked
theirown hypotheses(reviews in Miller 1985a; Pringle1986).
Alternativehypothesesproposed withregardto triggersforchanges fromthe
macroalgalphase to barrenshave advocated factorssuch as fish(Keats 1986)and,
in Pacific waters, sea otters (Estes and Palmisano 1974; Estes et al. 1978) and
1977; Ebert 1983). The
large-scaleepisodic recruitment
of sea urchins(Forerm-an
contributeto thebarrens
possibilitythatmassive pulses in sea urchinrecruitment
phenomenonhas been postulated(Scheibling 1986) but not testedforthe northwesternAtlantic.It is noteworthythatthe latterscenariodoes not involvepredators at all but, rather,draws on the currentecological paradigmof "supply-side
ecology" (Gaines and Roughgarden 1985; Underwood and Fairweather1989).
CONCLUDING
REMARKS
Whileour reviewdoes notresolve thesuperiority
ofone philosophicalapproach
studiesand the creationor
to ecology over another,we warn againstaffirmative
extensionofparadigms,bothofwhichappear to have contributedto theconfusion
in our understandingof the sea urchinphenomenonoffNova Scotia. We strongly
subscribeto theapproaches advocated by Platt(1964) of testingcriticalfulcraand
Strong(1980) of using multifacetedexperimentscoupled withstrictcontrols.We
believe that sound natural-history
observationsand possibly long-termdata sets
(Coull 1985) are prerequisitesto an effectivegenerationof hypothesesin communityecology. So-called "natural experiments"(e.g., Schoener and Spiller 1987)
can, in our opinion, yield only tentativeconclusions and patternsfor further
experimentsto short-term
testing(see McGuinnes 1988). We also preferlong-term
manipulations,the latter of which may provide misleadingconclusions (e.g.,
Jackson and Kaufmann 1987). Our advocacy for a hierarchicalframeworkof
approaches in communityecology parallels the earliercalls forproceduralrules
forstudiesof competitionand coevolution(Connorand Simberloff
1979;Stronget
al. 1979; Connell 1980).
In the finalanalysis, the shrewdnessof the scientificcommunityis essentialin
INFERENCE IN ECOLOGY
121
directingthe progressof science. A highratioof review to researchpapers on a
particulartopic should be regardedas indicativeof data starvationand, perhaps,
an unconscious need to perpetuatea set of paradigmsby repetition(Chamberlin
1897). We thinkthat emphasis on where and how changes are occurringin the
unfoldingof an ecological scenariois important.Specifically,have theconstituent
statementsor proposed systemcomponentsbeen explainedthroughtesting,natural events, or appeals to common sense? From an overall picture(such as presented by tables 1 and 2), it becomes apparent(1) whetherhypotheseshave been
advanced and experimentallyrejected, (2) what systemcomponentshave been
tested, and (3) whetherthe fidelityof componentshas been maintained.
Our review indicatesa need fora morecriticalattitudein ecology and supports
Wiens (1981) in his argumentfora renewed skepticismin science. We should all
learn to be more cautious about acceptingasserted statementsas facts. We also
advocate repeatingkey manipulativeexperimentsperformedby others,as suggested by Connell (1974). Lastly, we advocate thatstudiesbe challengedbothas
entitiesin themselvesand as constituentfragmentsof the biggerpicture.In this
way, ecologists should be able to establishthetrainof truthsnecessaryto unravel
the complexitiesof ecological systems.
SUMMARY
We have reviewed the considerable body of research into the sea urchin
phenomenonresponsibleforthe alternationbetween macroalgalbeds and coralproblems
Atlantic.In doingso, we have identified
line barrensin thenorthwestern
ofresults.Over a periodof
withboththe scientificapproach and theinterpretation
approximately20 years, explanationsforthe phenomenoninvokedfourseparate
scenarios, which changed mainlyas a consequence of extraneousevents rather
than experimentaltesting.Our specificconcerns are thatresultscontraryto the
paradigmforthe Americanlobsterwere circumvented,system
keystone-predator
components of the various scenarios became accepted without testing,and
modificationsof some components appeared arbitrary.Our review illustrates
dilemmas that, we suggest, have hinderedecological progress in general. We
argue fora more rigorousexperimentalapproach,based on sound natural-history
observationsand stronginference.Moreover,we believe thatthe scientificcommunityneeds to be cautious about allowing paradigmsto become established
withoutadequate scrutiny.
ACKNOWLEDGMENTS
We wish to extendour thanksto J. S. Ketch and D. M. Wilburfortheirpatience
for creatingthe
and help duringthe typingof the manuscriptand to G. Jeffery
Vadas,
Jr.,J. D.
R.
L.
W.
D.
Pringle,
Estes,
J.
Dennison,
J.
A.
C.
graphics.
Witman,and threeanonymousreviewersprovidedvaluable criticismofthemanuscript.R.L.V. acknowledges the supportof National Oceanographicand Atmospheric Administration-SeaGrantfunding(NA-81aa-D-00035)to the University
of Maine.
122
THE AMERICAN NATURALIST
LITERATURE CITED
Andrewartha,H. G., and L. C. Birch. 1984. The ecological web. Universityof Chicago Press,
Chicago.
Arnold,D. C. 1976. Local denudationof the sublittoralfringeby the greensea urchin,Strongylocentrotusdro[e]bachiensis(O. F. Muller). Can. Field-Nat. 90:186-187.
Barnes, R. S. K., and R. N. Hughes. 1982. An introductionto marinebiology. Blackwell Scientific,
Oxford.
Bernstein,B. B., and K. H. Mann. 1981. Changes in the near-shoreecosystemof theAtlanticcoast of
Nova Scotia, 1968-1981. Sci. Counc. Res. Doc. 81/ix/134.NorthwestAtlantic Fisheries
Organization,Dartmouth,Nova Scotia.
1982. Changes in the nearshoreecosystemof the Atlanticcoast of Nova Scotia, 1968-1981.
NorthwestAtl. Fish. Organ. Sci. Counc. Stud. 5:101-105.
Bernstein, B. B., and R. W. Welsford. 1982. An assessment of feasibilityof using high-calcium
quicklimeas an experimentaltool forresearchintokelp bed/seaurchinecosystemsin Nova
Scotia. Can. Tech. Rep. Fish. Aquat. Sci. 968:1-51.
1983. Design of a flexiblecage for benthicecological experimentsin high energymarine
environments.Can. Tech. Rep. Fish. Aquat. Sci. 1221:1-26.
Bernstein,B. B., B. E. Williams,and K. H. Mann. 1981.The roleofbehavioralresponsesto predators
in modifyingurchins' (Strongylocentrotus
droebachiensis)destructivegrazingand seasonal
foragingpatterns.Mar. Biol. (Berl.) 63:39-49.
Bernstein, B. B., S. C. Schroeter, and K. H. Mann. 1983. Sea urchin (Strongylocentrotus
droebachiensis) aggregatingbehavior investigatedby a subtidal multifactorial
experiment.
Can. J. Fish. Aquat. Sci. 40:1975-1986.
Breen, P. A. 1987. Commenton "Seaweeds, sea urchins,and lobsters:a reappraisal" by R. J. Miller.
Can. J. Fish. Aquat. Sci. 44:1806-1807.
Breen, P. A., and K. H. Mann. 1976a. Destructivegrazingof kelp by sea urchinsin easternCanada. J.
Fish. Res. Board Can. 33:1278-1283.
1976b.Changinglobsterabundance and thedestructionofkelpbeds by sea urchins.Mar. Biol.
(Berl.) 34:137-142.
Briscoe, C. S., and K. P. Sebens. 1988. Omnivoryin Strongylocentrotus
droebachiensis(Muller)
(Echinodermata: Echinoidea): predationon subtidalmussels. J. Exp. Mar. Biol. Ecol. 115:
1-24.
Chamberlin,T. C. 1897. The methodof multipleworkinghypotheses.J. Geol. 5:837-848.
Chapman, A. R. 0. 1981. Stabilityof sea urchindominatedbarrengroundsfollowingdestructive
grazingof kelp in St. Margaret'sBay, easternCanada. Mar. Biol. (Berl.) 62:307-311.
Clark, J. R. 1983. Coastal ecosystem management.Krieger,Malabar, Fla.
Connell,J. M. 1974. Ecology: fieldexperimentsin marineecology. Pages 21-54 in R. N. Mariscal, ed.
Experimentalmarinebiology. Academic Press, New York.
1980. Diversityand the coevolutionof competitors,or the ghostof competitionpast. Oikos
35: 131-138.
Connor, E., and D. Simberloff.1979. The assemblyof species communities:chance or competition?
Ecology 60:1132-1140.
Coull, B. C. 1985. The use of long-termbiological data to generatetestable hypotheses.Estuaries
8:84-92.
Dayton, P. K., and J. S. Oliver. 1980. An evaluation of experimentalanalyses of populationand
communitypatternsin benthicmarineenvironments.Pages 93-120 in K. Tenore and B. C.
Coull, eds. Marinebenthicdynamics.Belle W. Baruch Libr. Mar. Sci. Publ. 11. Universityof
South Carolina Press, Columbia.
Drummond-Davis,N. C., K. H. Mann, and R. A. Pottle. 1982. Some estimatesof populationdensity
and feedinghabitsof the rock crab, Cancer irroratus,in a kelp bed in Nova Scotia. Can. J.
Fish. Aquat. Sci. 39:636-639.
Duggins, D. 0. 1983a. Starfishpredationand the creationof mosaic patternsin a kelp-dominated
community.Ecology 64:1610-1619.
1983b. Marine dominoes. Equinox 2:42-57.
INFERENCE IN ECOLOGY
123
of science. Nature
Durant, J. R., G. A. Evans, and G. P. Thomas. 1989. The public understanding
(Lond.) 340:11-14.
Ebert, T. A. 1983. Recruitmentin echinoderms.Pages 169-203 in M. Jangouxand J. M. Lawrence,
eds. Echinodermstudies I. Balkema, Rotterdam.
droebachiensis)by the American
Elner, R. W. 1980. Predationon the sea urchin(Strongylocentrotus
lobster (Homarus americanus) and the rock crab (Cancer irroratus).Pages 48-65 in J. D.
Pringle,G. J. Sharp, and J. F. Caddy, eds. Proceedingsof the workshopon the relationship
harvesting.Can. Tech. Rep. Fish.
between sea urchingrazingand commercialplant/animal
Aquat. Sci. 954:1-273.
1983. Keystone controversy.Equinox 2:7.
Elner, R. W., and A. Campbell. 1987. Natural diets of lobsterHomarus americanus frombarren
groundand macroalgalhabitatsoffsouthwesternNova Scotia, Canada. Mar. Ecol. Prog. Ser.
37:131-140.
nearshorecommunities.
Estes, J. A., and J. F. Palmisano. 1974. Sea otters:theirrole in structuring
Science (Washington,D.C.) 185:1058-1060.
organization
Estes, J. A., N. S. Smith,and J. F. Palmisano. 1978. Sea otterpredationand community
in the westernAleutian Islands, Alaska. Ecology 59:822-833.
Evans, P. D., and K. H. Mann. 1977. Selection of preyby Americanlobsters(Homarus americanus)
whenoffereda choice betweensea urchinsand crabs. J. Fish. Res. Board Can. 34:2203-2207.
Foreman, R. E. 1977. Benthic communitymodificationand recoveryfollowingintensivegrazingby
droebachiensis. Helgol. Wiss. Meeresunters.30:468-484.
Strongylocentrotus
of structurein an
Gaines, S., and J. Roughgarden.1985. Larval settlementrate: a leadingdeterminant
ecological communityof the marineintertidalzone. Proc. Natl. Acad. Sci. USA 82:37073711.
Harrold,C., and J. S. Pearse. 1988. The ecological role of echinodermsin kelp forests.Pages 137-233
in M. Jangouxand J. M. Lawrence, eds. Echinodermstudies II. Balkema, Rotterdam.
Himmelman,M. J., A. Cardinal,and E. Bouget. 1983. Communitydevelopmentfollowingremovalof
droebachiensis,fromthe rocky subtidalzone of the St. Lawurchins,Strongylocentrotus
rence Estuary,easternCanada. Oecologia (Berl.) 59:27-39.
Hirtle,R. W. M., and K. H. Mann. 1978. Distance chemoreceptionand visionin the selectionof prey
by Americanlobster(Homarus americanus). J. Fish. Res. Board Can. 35:1006-1008.
Hughes, R. N. 1980. Predationand communitystructure.Pages 699-728 in J. H. Price, D. E. G,
Irvine, and W. F. Farnham,eds. The shore environment.Vol. 2. Syst. Assoc. Spec. Vol.
17(6). Academic Press, New York.
Jackson,J. B. C., and K. W. Kaufmann. 1987. Diadema antillarumwas not a keystonepredatorin
crypticreefenvironments.Science (Washington,D.C.) 235:687-689.
Johns, P. M., and K. H. Mann. 1987. An experimentalinvestigationof juvenile lobster habitat
preferenceand mortalityamonghabitatsof varyingstructuralcomplexity.J. Exp. Mar. Biol.
Ecol. 109:275-285.
droebachiensisforsurvival
Johnson,C. R., and K. H. Mann. 1982. Adaptationof Strongylocentrotus
on barren grounds in Nova Scotia. Pages 277-283 in J. M. Lawrence, ed. International
echinodermsconference,Tampa Bay. Balkema, Rotterdam.
1988. Diversity, patternsof adaptation, and stabilityof Nova Scotian kelp beds. Ecol.
Monogr. 58:129-154.
Jones,G. M., and R. E. Scheibling. 1985. Paramoeba sp. (Amoebida, Paramoebidae) as the possible
causative agent of sea urchinmass mortalityin Nova Scotia. J. Parasitol. 71:559-565.
Keats, D. 1986. Commenton "Seaweeds, sea urchins,and lobsters:a reappraisal" by R. J. Miller.
Can. J. Fish. Aquat. Sci. 43:1675-1676.
Keats, D. W., D. H. Steele, and G. R. South. 1986. Atlanticwolffish(Anarhichas lupus L., Pisces: Anarhichidae) predation on green sea urchins (Strongylocentrotusdroebachiensis
(O. F. Mull.); Echinodermata: Echinoidea) in eastern Newfoundland. Can. J. Zool. 64:
1920-1925.
1987. Ocean pout (Macrozoarces americanus (Block and Schneider) (Pisces: Zoarcidae))
droebachiensis(O. F. Mull.) (Echinoderpredationon greensea urchins(Strongylocentrotus
mata: Echinoidea)) in eastern Newfoundland.Can. J. Zool. 65:1515-1521.
124
THE AMERICAN NATURALIST
Lang, C., and K. H. Mann. 1976. Changes in sea urchinpopulationsafterthedestructionof kelp beds.
Mar. Biol. (Berl.) 36:321-326.
Mann, K. H. 1972. The coastal zone: itsbasic ecological propertiesand subdivisions.Pages 123-131in
Proceedings of a Coastal Zone Seminar, March 1972. BedfordInstituteof Oceanography,
Dartmouth,Nova Scotia.
1973. Seaweeds: their productivityand strategyfor growth.Science (Washington,D.C.)
182:975-981.
1977. Destructionof kelp-bedsby sea-urchins:a cyclicalphenomenonor irreversibledegradation? Helgol. Wiss. Meeresunters.30:455-467.
1981. Managementof resources in the coastal zone: Laminaria and lobstersin Nova Scotia.
LindenberghLect. Ser. Ecol. Ecosystems Center, Marine Biological Laboratory,Woods
Hole, Mass.
1982a. Kelp, sea urchinsand predators:a review of stronginteractionsin rocky subtidal
systemsof eastern Canada, 1970-1980. Neth. J. Sea Res. 16:414-423.
1982b. Ecology of coastal waters. Blackwell Scientific,Oxford.
1985a. Invertebratebehaviourand the structureof marinebenthiccommunities.Pages 227246 in R. M. Sibly and R. H. Smith,eds. Behaviouralecology. Blackwell Scientific,Oxford.
Page 66 in 43d
1985b. Populationoutbreaksof sea urchins:naturalor man-madefluctuations?
Meeting of the American Society of Limnology and Oceanography, June 18-21, 1985.
Universityof Minnesota, Minneapolis.
Mann, K. H., and P. A. Breen. 1972. The relationbetweenlobsterabundance, sea urchins,and kelp
beds. J. Fish. Res. Board Can. 29:603-605.
Mann, K. H., and R. B. Clark. 1978. Long-termeffectsof oil spillson marineintertidalcommunities.
J. Fish. Res. Board Can. 35:791-795.
Mann, K. H., J. L. C. Wright,B. E. Welsford,and E. Hatfield.1984. Responses of the sea urchin
Strongylocentrotusdroebachiensis (O. F. Muller) to water-bornestimulifrompotential
predatorsand potentialfood algae. J. Exp. Mar. Biol. Ecol. 79:223-244.
Margosian,A., F. C. Tan, D. Cai, and K. H. Mann. 1987. Seawater temperaturerecordsfromstable
isotopic profilesin the shell of Modiolus modiolus. EstuarineCoastal Shelf Sci. 25:81-89.
McGuinnes, K. A. 1988. Explainingpatternsin abundances of organismson boulders: the failureof
'naturalexperiments.'Mar. Ecol. Prog. Ser. 48:199-204.
communitiesand food
Michaud, B. J. 1986. Compositionand productionof macrobenthicinvertebrate
resources of the American lobsterHomarus americanus along the Atlanticcoast of Nova
Scotia. Ph.D. diss. Dalhousie University,Halifax.
Miller, R. J. 1985a. Seaweeds, sea urchins,and lobsters: a reappraisal. Can. J. Fish. Aquat. Sci.
42:2061-2072.
. 1985b. Succession in sea urchinand seaweed abundance in Nova Scotia, Canada. Mar. Biol.
(Berl.) 84:275-286.
. 1985c. Sea urchin pathogen: a possible tool for biological control. Mar. Ecol. Prog. Ser.
21:169-174.
1986. Reply to commentby D. Keats. Can. J. Fish. Aquat. Sci. 43:1676.
1987. Reply to commentby P. A. Breen. Can. J. Fish. Aquat. Sci. 44:1807-1809.
Miller,R. J., and A. G. Colodey. 1983. Widespreadmass mortalitiesof the greensea urchinin Nova
Scotia, Canada. Mar. Biol. (Berl.) 73:263-267.
Miller,R. J., and K. H. Mann. 1973. Ecological energeticsof the seaweed zone in a marinebay on the
by sea urchins. Mar. Biol. (Berl.)
Atlantic coast of Canada. III. Energy transformations
18:99-114.
Miller, R. J., K. H. Mann, and D. J. Scarratt. 1971. Productionpotentialof a seaweed-lobster
communityin eastern Canada. J. Fish. Res. Board Can. 28:1733-1738.
Mohn, R. K., and R. J. Miller. 1987. A ration-basedmodelof a seaweed-sea urchincommunity.Ecol.
Model. 37:249-267.
Paine, R. T. 1969. A note on trophiccomplexityand communitystability.Am. Nat. 103:91-93.
1977. Controlled manipulationsin the marine intertidalzone, and their contributionsto
ecological theory.Spec. Publ. Acad. Nat. Sci. Phila. 12:245-270.
INFERENCE
IN ECOLOGY
125
. 1980. Food webs: linkage,interactionstrength,
and communityinfrastructure.
J. Anim. Ecol.
49:667-685.
Platt,J. R. 1964. Stronginference.Science (Washington,D.C.) 146:347-353.
Pringle,J. D. 1986. A review of urchin/macro-algal
associations witha new synthesisfornearshore,
eastern Canadian waters. Monogr. Biol. 4:191-218.
Pringle,J. D., G. J. Sharp, and J. F. Caddy. 1982. Interactionsin kelpbed ecosystemsin thenorthwest
Atlantic:review of a workshop.Can. Spec. Publ. Fish. Aquat. Sci. 59:108-115.
Quinn, J. F., and A. E. Dunham. 1983. On hypothesistestingin ecology and evolution. Am. Nat.
122:602-617.
Raymond,B. G., and R. E. Scheibling.1987. Recruitmentand growthof the sea urchinStrongylocentrotusdroebachiensis (Muller) followingmass mortalitiesoffNova Scotia, Canada. J. Exp.
Mar. Biol. Ecol. 108:31-54.
Roughgarden,J. 1983. Competitionand theoryin communityecology. Am. Nat. 122:583-601.
Salt, G. W. 1983. Roles: theirlimitsand responsibilitiesin ecological and evolutionaryresearch.Am.
Nat. 122:697-705.
Scheibling, R. 1986. Increased macroalgal abundance followingmass mortalitiesof sea urchins
(Strongylocentrotusdroebachiensis) along the Atlantic coast of Nova Scotia. Oecologia
(Berl.) 68:186-198.
droebachiensis
Scheibling,R. E., and R. L. Stephenson. 1984. Mass mortalityof Strongylocentrotus
(Echinodermata: Echinoidea) offNova Scotia, Canada. Mar. Biol. (Berl.) 78:153-164.
Schoener, T. W., and D. A. Spiller. 1987. Effectof lizards on spider populations: manipulative
reconstructionof a naturalexperiment.Science (Washington,D.C.) 236:949-952.
Simberloff,D. 1983. Competitiontheory,hypothesistesting,and othercommunityecological buzzwords. Am. Nat. 122:626-635.
Slobodkin, L. B. 1986. The role of minimalismin art and science. Am. Nat. 127:257-265.
Strong,D. R., Jr. 1980. Null hypothesisin ecology. Synthese43:271-285.
1983. Natural variabilityand the manifoldmechanismsof ecological communities.Am. Nat.
122:636-660.
Strong,D., Jr.,L. Szyska, and D. Simberloff.1979. Tests of community-wide
characterdisplacement
against null hypotheses.Evolution 33:897-913.
Sze, P. 1986. A biologyof the algae. W. C. Brown, Dubuque, Iowa.
Underwood,A. J., and E. J. Denley. 1984. Paradigms,explanationsand generalizationsin modelsfor
the structureof intertidalcommunitieson rockyshores. Pages 151-180 in D. R. Strong,D. S.
Simberloff,L. G. Abele, and A. B. Thistle,eds. Ecological communities:conceptualissues
and evidence. PrincetonUniversityPress, Princeton,N.J.
Underwood,A. J.,and P. G. Fairweather.1989. Supply-sideecologyand benthicmarineassemblages.
Trends Ecol. Evol. 4:16-19.
Vadas, R. L., R. W. Elner, P. E. Garwood, and I. G. Babb. 1986. Experimentalevaluation of
droebachiensis:a reinterpretation.
aggregationbehaviorin the sea urchinStrongylocentrotus
Mar. Biol. (Berl.) 90:443-448.
Valiela, I. 1984. Marine ecological processes. Springer,New York.
Warner,G. F. 1984. Diving and marinebiology. CambridgeUniversityPress, Cambridge.
Wharton,W. G., and K. H. Mann. 1981. Relationshipbetweendestructivegrazingby the sea urchin,
Strongylocentrotus
droebachiensis,and theabundance ofAmericanlobster,Homarus americanus, on the Atlanticcoast of Nova Scotia. Can. J. Fish. Aquat. Sci. 38:1339-1349.
Wiens, J. A. 1981. On skepticismand criticismin ornithology.Auk 98:848-849.