Leonard 00
Latitudinal Variation in Species Interactions: A Test in the New England Rocky Intertidal Zone
Author(s): George H. Leonard
Source: Ecology, Vol. 81, No. 4 (Apr., 2000), pp. 1015-1030
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/177175
Accessed: 25/01/2010 14:13
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=esa.
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.
Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.
http://www.jstor.org
Ecology, 81(4), 2000, pp. 1015-1030
? 2000 by the Ecological Society of America
LATITUDINAL VARIATION IN SPECIES INTERACTIONS:
A TEST IN THE NEW ENGLAND ROCKY INTERTIDAL ZONE
GEORGE H. LEONARD'
Brown University, Department of Ecology and Evolutionary Biology, Providence, Rhode Island 02912 USA
Abstract. How species interactions vary as a function of habitat characteristics con-
tinues to be an important debate in ecology. Using the barnacle-seaweed assemblage com-
mon in New England rocky intertidal habitats, I tested the hypothesis that species inter-
actions switch from negative to positive at sites across the Cape Cod faunal barrier because
of latitudinal variation in thermal stress and predation intensity between these regions. I
manipulated Ascophyllum nodosum canopies in the high zone of two sites from each region
and then determined the consequences for underlying Semibalanus balanoides recruits and
adults for two years (January 1995 through January 1997). In both years, algal canopies
reduced barnacle recruitment and growth rates at all sites but greatly increased survival
rates only at the southern sites. When integrated in a demographic framework, these data
showed that the reproductive fitness potential of individuals was facilitated by the algal
canopy at southern sites but was reduced under similar conditions at northern sites. At
southern sites, this was likely the result of buffering from physical stresses in the absence
of predators. At northern sites, any buffering from physical stress was likely offset by an
increase in mortality in the presence of predators. Interestingly, this variability in species
interactions appeared to be associated with subtle variation in climate. Facilitation was
evident only during 1995, the warmest year on record since 1900. In contrast, interactions
were entirely negative during 1996, a slightly cooler year. These results suggest that species
interactions in the intertidal zone may be sensitive to even subtle changes in climate.
Interspecific buffering of neighbors from thermal stress is likely to be common in other
systems and suggests that both aquatic and terrestrial vegetation may become increasingly
important to local species persistence as climates change during the next century.
Key words: Ascophyllum nodosum; direct vs. indirect effects; fitness consequences; habitat
amelioration; interactions, positive and negative; latitude effects on species interactions; New England
(USA) rocky intertidal zone; Nucella lapillus; predation; Semibalanus balanoides; thermal stress.
INTRODUCTION ness et al. [1981], Menge and Lubchenco [1981] for
Pattern and process across biogeographic spatial exceptions). This scarcity of experimental data is at
scales have been of interest to ecologists since the time least partly due to the logistical difficulties of con-
of Darwin. Increases in diversity across latitudinal gra- ducting manipulative experiments at large spatial
dients and between ocean basins are well known (Fi- scales.
scher 1960, MacArthur 1965, 1972, Spight 1976, Wal- Over the last twenty years, work within single bio-
lace 1878), although the ultimate mechanism for their geographic regions has highlighted how abiotic con-
ditions can alter the relative importance of biological
origin and maintenance remains unclear (Rhode 1992,
Rosenzweig 1995). Biological processes have long processes in governing community structure (Connell
been argued to vary across large spatial scales (Mac- 1961, Menge 1976, Menge and Sutherland 1987).
Arthur 1972, Vermeij 1978), and latitudinal variation These results, largely from marine habitats, predict that
in predation, in particular, has been associated with as physical stress increases, predation pressure decreas-
differences in morphology in many organisms (Mayr es and intraspecific competition increases (i.e., the con-
sumer stress models of Menge and Olson [1990]). Work
1963, Bakus 1969, Vermeij and Veil 1978). Although
on vascular plants in terrestrial habitats has similarly
species interactions have been hypothesized to vary
biogeographically (e.g., Vermeij 1978, Dethier and suggested that the role of competition varies with stress
Duggins 1988), experimental tests of these hypotheses (Grime 1973, 1977, Tilman 1988) although there is
have been relatively rare (but see Jeanne [1979], Bert- much controversy surrounding this assertion (Grime
1979, Weldon and Saulson 1986, Wilson and Keddy
Manuscriptreceived 25 January1999 (originally submitted30 1986, Moloney 1990, Grace 1991, Goldberg and Barton
April 1998); revised 24 February 1999; accepted 25 February 1992). When physical stress becomes extreme, how-
1999. ever, competitive interactions may be replaced by fa-
a Present address: Stanford University, Department of Bi-
cilitative interactions (Bertness and Callaway 1994) if
ological Sciences, Hopkins Marine Station, Oceanview Bou-
levard, Pacific Grove, California 93950-3094 USA. the "competitor" modifies the physical habitat and al-
E-mail: gleonard@leland.stanford.edu leviates the stressful conditions (e. g., Bertness and
1015
1016 GEORGEH. LEONARD Ecology,Vol. 81, No. 4
Shumway 1993, Bertness and Hacker 1994). In these which to investigate spatial and temporal variation in
cases, a species may have higher fitness when in as- species interactions. Most intertidal organisms are
sociation with a "competitor" than when it is living known to be sensitive to thermal and desiccation stress-
alone. es, which can vary at both small and large spatial scales
While productive, this debate on the dichotomy be- (Lewis 1964, Wethey 1983, 1984). Most importantly,
tween competition and facilitation ignores the fact that these physical stresses can often be alleviated by the
species can influence each other in ways that do not presence of other species. For example, intertidal algal
strictly involve limiting resources. One organism can canopies often keep the substrate moist at low tide and
influence the recruitment, survival, growth, or repro- can thus reduce the physiological stress of the organ-
duction of another and these effects can vary in both isms underneath (Dayton 1971, Menge 1978, Under-
space and time. For example, subtidal algae can alter wood and Denley 1984 and references therein). Be-
the recruitment and growth of benthic invertebrates by cause of the large amount of work done in these com-
altering propagule delivery and food acquisition with- munities (see Lewis 1964, Paine 1994, Little and Kitch-
out actually competing for resources (Duggins et al. ing 1996), there is also an ecological context in which
1990, Eckman and Duggins 1991). to place experimental work done at broader scales. Ex-
The debate between competition and facilitation perimental manipulations to understand how species
should be broadened further to address the many pos- interactions vary at larger scales may also be a pow-
itive and negative ways in which organisms influence erful way to predict how species and communities will
each other's fitness. This is critical because the overall respond to global climate change in the future.
effect of one species on another may be positive, neg- In this study, I hypothesized that interactions be-
ative, or neutral depending on the magnitude and di- tween a large, canopy-forming intertidal alga (Asco-
rection of the individual effects (e. g., Underwood phyllum nodosum) and a sessile, filter-feeding inver-
1986). Demography offers an excellent framework in tebrate (Semibalanus balanoides) should vary at lati-
which to address these types of multiple effects tudinal spatial scales in New England, USA, because
(McPeek and Peckarsky 1998). Simultaneous variation of predictable differences in environmental stress (tem-
in recruitment, survival, and fecundity as a function of perature) and predation by the carnivorous whelk, Nu-
the presence or absence of a species can be entered cella lapillus. Seasonal temperature fluctuations in New
into a standard life table and the net effect evaluated England are some of the largest in the world (Menge
as the product of these individual effects. Many of the 1976) with summer air temperatures greater than in
previous studies of facilitation, however, have focused either Great Britain or the west coast of North America
only on survival (e. g., see review by Callaway [1995]) (Barnes 1958a). In addition, subtle but important dif-
and few have considered these multiple effects. This ferences in summer air temperature have been hypoth-
is especially true of interactions that vary as a function esized to influence the distribution of, and interactions
of life-history stage. To more fully understand the role among, intertidal organisms between sites to the north
of positive and negative interactions in natural com- and south of the Cape Cod peninsula in Massachusetts
munities requires a focus on these types of multiple, (Barnes and Barnes 1959, Wethey 1983). Observations
interactive effects (Greenlee and Callaway 1996, Cal- I made in 1994 at exposed intertidal habitats in Rhode
laway and Walker 1997). Island (south of Cape Cod) indicated that the upper
The debate must also move beyond documenting the distribution of the Semibalanus zone was generally co-
direct effect of one species on another and begin to incident with the top of the Ascophyllum zone but that
incorporate indirect effects. Although there are excep- at similar sites in Maine (north of Cape Cod) it ex-
tions, many of the previous studies have focused on tended beyond it (see Methods: Study sites and zona-
pairwise interactions (e. g., Bertness and Shumway tion patterns, below). The southern pattern is atypical
1993). This has unintentionally disregarded the large of most intertidal habitats where barnacles generally
number of species that comprise most biological com- persist above the algal zone in all but the most protected
munities and the numerous indirect interactions (both habitats (Stephenson and Stephenson 1948, Carefoot
positive and negative) that occur among them. Inter- 1977, Menge 1978, Ricketts et al. 1985; personal ob-
action modifications, where the non-trophic effects of servations). This observation is consistent with a re-
one species alters the interaction between two other duction in mortality from thermal stress north but not
species, may be common and important indirect inter- south of Cape Cod.
actions in many communities (Wootton 1993). It is Cape Cod is also a well-known faunal break (Gould
clear that a synthetic approach, which focuses on how 1840, cited in Allee 1923) that separates the Atlantic
spatial and temporal variability in the environment in- Boreal fauna to the north from the Atlantic Temperate
fluences both direct and indirect interactions, will help fauna to the south (Gosner 1978). Nucella lapillus is
ecologists better understand how species interactions the primary predator of barnacles in New England and
vary in natural habitats (Callaway and Walker 1997, also has a largely boreal distribution (Gosner 1978).
McPeek and Peckarsky 1998). These biogeographic patterns and the experimental
The marine intertidal may be an ideal system in findings of Menge (1976) in northern New England
April 2000 VARIATIONIN SPECIESINTERACTIONS 1017
suggest that predation pressure may be reduced south
of Cape Cod. In addition, by harboring whelks and
increasing their foraging efficiency (Menge 1978), the
algal canopy indirectly increases barnacle mortality at
those sites where whelks are present. This suggests that
the direct positive effects of the algal canopy on bar-
nacle survival (by alleviating temperature stress) may
be critical to the maintenance of barnacle populations
south of the Cape but that the negative, indirect effects
of the canopy on predators (an interaction modification
sensu Wootton 1993) may overwhelm any direct pos-
itive effects at northern sites.
To test these ideas, I experimentally manipulated al-
gal canopies at northern and southern sites and eval-
uated the consequences for both barnacles and their
predators. I quantified canopy effects at different bar-
nacle life-history stages and then integrated these com-
ponent effects in a demographic framework. I hypoth-
esized that the algal canopy would decrease fecundity
at all sites but that it would increase survival only at
the southern sites. If the positive effects of the canopy
were stronger than its negative effects, the association
would result in higher fitness (defined as the product FIG. 1. Map of New England(USA) showing the open-
of survival and fecundity) at the southern sites. I also coast study sites in Rhode Island (RI) and Maine. Cape Cod
took advantage of the differences in climatic conditions is a well-recognizedfaunalbarrier betweenthesetwo regions.
in 1995 and 1996 to determine whether this interaction
was associated with year-to-year variation in climate.
mean lower-low water (MLLW). Because the tidal
METHODS range differs between Rhode Island and Maine (1.4 m
vs. 3.5 m above MLLW, respectively), elevations were
Study sites and zonation patterns
expressed as the percentage of time that zones were
Two sites in Rhode Island and two in Maine (New exposed to aerial conditions within each region. Tidal
England, USA) were chosen to test this hypothesis (Fig. height data were obtained from TideGuide version 1.30
1). All sites were semi-exposed, intertidal habitats con- (Zihua Software, Pacific Grove, California, USA). The
sisting of gently sloping granite benches interspersed tops of the Ascophyllum nodosum and Semibalanus bal-
with large granitic boulders. Each site was oriented anoides zones were quantified by sampling 10 eleva-
approximately south-south east. Sites were protected tions per zone across approximately 20-40 m of each
from the largest ocean swells by either small offshore study site. Differences in exposure time of each zone
islands or large seaward rock benches. The two south- between northern and southern sites were analyzed us-
ern sites, Sakonnet Point (41?27'14" N, 71?11'35" W) ing nested ANOVA with site nested within region.
and Middletown (41?28'31" N, 71014'30" W), were lo-
cated on the eastern and western sides, respectively, of Barnacle demographics and variable species
interactions
Narragansett Bay, Rhode Island. The two northern
sites, Chamberlain (43?53'7" N, 69?28'29" W) and Pe- Positive and negative interactions between the algal
maquid (43?50'8" N, 69?30'29" W), were located along canopy and understory barnacles were examined at
the eastern shore of Pemaquid Neck. Although exper- these four sites for two years, from January 1995
iments in Maine were not conducted exactly where oth- through January 1997. In November 1994 I created
ers have worked (e. g., Menge 1976, 1978), these sites circular clearings in the Ascophyllum canopy (radius
have been the subject of considerable past research. -1.0 m) at its upper border in the high zone of all sites
Observations made during May-September 1994 (n = 8 clearing/site). Canopy plots (n = 8 plots/site)
suggested that the acorn barnacle, Semibalanus bal- were unmanipulated areas that, at low tide, had a 100%
anoides, extended above the Ascophyllum nodosum cover of A. nodosum. The absolute tidal height of all
zone at northern sites, but at southern sites it was pres- plots was set to keep the percentage of time exposed
ent only under the algal canopy. To determine if this to aerial conditions (-60%) constant between regions.
was due to a vertical extension of the barnacle zone Within each canopy and cleared plot, I set up two per-
rather than a contraction of the algal zone, distribution manent quadrats (25 X 25 cm) on the rock substrate
patterns were quantified at all sites using standard sur- marked at their corners with galvanized bolts. Quadrats
veying equipment and then standardizing elevations to were nestled between the A. nodosum holdfasts (in can-
1018 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
opy plots) and initially had an intermediate cover kept on ice in the field, and then frozen at -10?C in
(-50%) of barnacles. the laboratory. Within 4 mo barnacles were thawed in
Each spring one of the two quadrats in each plot was seawater and then each individual was dissected into
scraped to bare rock with a putty knife (without dis- shell, somatic, and reproductive (i.e., larvae) tissue.
turbing A. nodosum holdfasts or canopy cover) and Tissue components were dried at 35?C for 24 h before
used to evaluate the effect of the canopy on barnacle weighing on a microbalance (precision = +0.01 mg).
recruitment, survival, growth, and fecundity. Because
adult barnacles may generally be less susceptible to Statistical analysis
physical stress than new recruits (Foster 1969, 1971), Differences in recruit and adult barnacle survival as
I used the unmanipulated quadrat to evaluate the in- a function of the canopy, region, and site were tested
fluence of the canopy on the "adult" population (de- using nested analysis of covariance (ANCOVA) on sur-
fined as all individuals >1 yr old, sensu Wethey 1984). viving barnacle density in October. Site was considered
In addition, scraped quadrats (n = 8 quadrats/site) were a random factor (nested within region) and canopy and
established in the zone directly above the Ascophyllum region were considered fixed factors. Density the pre-
canopy at all sites to evaluate regional differences in vious spring was the covariate. Because space is often
barnacle recruitment, survival, growth, and reproduc- the limiting resource for barnacles, mortality is fre-
tion above the zone influenced by the algal canopy. quently density dependent (i.e., greater mortality on
In 1996 I incorporated a predation treatment to at- high densities than low densities of recruits; Connell
tempt to partition the mortality of barnacle recruits into 1985). ANCOVA statistically factors out any density-
the direct effect of the algal canopy on reduced thermal dependent effects on survival by removing the influ-
stress and the indirect effect of the canopy on predation ence of initial density on final density. This approach,
intensity (Menge 1978). At all sites I attached 20 X however, has two assumptions that were verified before
20 X 5 cm galvanized wire mesh cages and cage con- the analysis proceeded (Underwood 1997). First, the
trols to cleared quadrats in all canopy and cleared plots. relationship between the dependent variable and the
Unfortunately, this treatment was unsuccessful at ma- covariate must not vary among treatments (homoge-
nipulating predator abundance (unpublished data) be- neity of slopes). Second, the range of the covariate must
cause cages were frequently damaged by waves and be broadly similar among treatments.
corrosion of the wire mesh often compromised the tight Differences in the demographic parameters of bar-
fit of the cages to the substrate. Accordingly, I analyzed nacle recruits and adults as a function of the algal can-
only the results from the uncaged quadrats. Densities opy, region, and site were analyzed with nested analysis
of whelks were measured three times during each sum- of variance for both 1995 and 1996. As before, site
mer at all sites in these uncaged quadrats and average was considered a random, nested factor. Differences in
predator abundance was analyzed using analysis of var- growth were evaluated using total soft tissue (somatic
iance. Differences in predator abundance were related + reproductive tissue) as the dependent variable. Sim-
qualitatively to patterns of barnacle mortality among ilar analyses were performed on fecundity (total larval
regions and years. mass) and reproductive fitness potential. Reproductive
At the end of the settlement season (Rhode Island fitness potential was defined in the demographic sense
end of March, Maine = beginning of May), barnacle as "l,mx" (an individual's probability of survival mul-
recruitment was measured in the field in the cleared tiplied by its reproductive output). Because of the broad
quadrats using sampling grids. Barnacles that survived dispersal capability and open population structure of
through the summer were sampled using similar tech- Semibalanus, I could evaluate neither the complete life
niques in late October in both 1995 and 1996. Change table for this species (but see Eckman [1996] for a
in density of the "adults" was determined from pho- conceptual and empirical approach) nor the contribu-
tographs of the unmanipulated quadrats at the begin- tion of individual effects to population growth rate (i.e.,
ning (March 1995), middle (August 1995), and end a sensitivity analysis of X; McPeek and Peckarsky
(October 1996) of the experiment. 1998).
In both years, growth and reproductive output of In addition, there were often large differences in
recruits and adults were quantified from three individ- these parameters among sites (see Table 1). These dif-
uals haphazardly selected from each of the cleared and ferences could have been due to a number of uncon-
unmanipulated quadrats (total sample size: recruits = trolled factors including (1) differences in larval supply
72 individuals per site per year, adults = 48 individuals (Raimondi 1990), (2) phytoplankton composition
per site per year). In New England, barnacles reproduce (Barnes and Barnes 1959), (3) increased egg size at
in early fall, fertilized eggs mature over the winter, and northern sites (Crisp 1959), (4) temperature-dependent
larvae are released only once a year in early spring or flow-mediated plasticity in growth rates (Southward
(Barnes 1958b). Individuals were sampled in mid-win- 1955, Sanford et al. 1994), or (5) genetic differentiation
ter (December-January) after larvae had fully matured among sites or regions (Crisp 1964). My primary in-
but before they had been released. Barnacles were care- terest was not in the absolute magnitude of these pa-
fully removed from the rock using surgical scalpels, rameters but rather in their variation with the algal
April 2000 VARIATIONIN SPECIESINTERACTIONS 1019
canopy between regions. I therefore standardized the canopy treatment per site). In 1996, sample size was
data for differences among sites before statistical anal- increased to 8 thermometers/site but temperatures were
ysis by converting each datum to reflect its deviation collected only in cleared plots. This design increased
(either larger or smaller) from its site mean. Analyses the power of detecting differences in thermal charac-
were then performed on these deviations (In (x + 1) teristics among sites. Both wave exposure and thermal
transformed to meet the assumption of normality and data were analyzed using repeated-measures, nested
variance heteroscedascity). Positive or negative species analysis of variance with region, site and time as fac-
interactions were defined from the direction of a sta- tors. As before, site was considered a random factor
tistically significant "canopy" effect in the analysis of and was nested within region.
variance. In addition, regional differences in the mag- I used weather-station data to supplement these di-
nitude and direction of these species interactions were rect measures of temperature. Overall differences in
identified by a significant "canopy X region" effect in climatic conditions between Maine and Rhode Island
the same analyses. Because this was a nested design, were quantified using data acquired from the Northeast
the error term for the F ratio for both canopy and can- Regional Climate Center. Maximum daily air temper-
opy X region effects was the canopy X site (region) ature had been recorded for 1995 and 1996 at land-
term. based sites in Newport, Rhode Island (41?30' N, 71?21'
W), and Boothbay, Maine (43?52' N, 69?35' W), both
Abiotic conditions in close proximity to the respective study sites. Daily
Physical factors were recorded during the two years differences in temperature from April to September be-
of the study to relate the biotic results to spatial and tween regions and years were used to quantify spatial
temporal variation in abiotic conditions. I tested for and temporal differences in climatic conditions. Pat-
variation in wave exposure among sites because bar- terns in thermal regime, wave exposure, and predator
nacle mortality is known to be reduced at sites of high abundance were compared qualitatively to the exper-
wave splash (Lewis 1964). This was done by quanti- imental results on variation in species interactions.
fying wave exposure at all sites approximately monthly
from December 1995 through August 1996 using RESULTS
spring-loaded dynamometers (Denny 1983, Bell and Zonation patterns
Denny 1994). These instruments measure the maximum
force imposed by breaking waves over the sampling The upper limit of the Ascophyllum canopy on the
interval. Unfortunately, these instruments do not mea- shore was similar between study sites in Maine and
sure the average conditions that exist at a site and it is Rhode Island (Fig. 2; F1,2 = 0.90, P = 0.442) but the
these average conditions that may be most important upper limit of the barnacle zone differed between re-
to organisms that suffer daily emersion. Estimating gions (F, 2 = 19.85, P = 0.047). At southern sites, the
wave exposure by eye, however, can be misleading (see upper limit of barnacle zone was coincident with the
Bell and Denny 1994) and dynamometers, although not upper limit of the algal canopy, but at northern sites it
flawless, are probably the best technique currently extended beyond the algal canopy (Fig. 2, daily emer-
available to estimate wave splash. sion = 62% vs. 75%, respectively).
Dynamometers (n = 5 instruments/site) were bolted Variable species interactions: barnacle recruits
to the rock in the center of circular plots (1-m radius)
cleared of macroalgae in the same area as the canopy The algal canopy strongly influenced barnacle re-
manipulations. Measurements of spring extension (in cruitment, survival, growth, and fecundity during the
millimeters) were converted to maximum force (in two years of this study and this had large effects on
newtons) using the equations in Bell and Denny (1994). their reproductive fitness potential. Recruitment at the
By late May 1996, dynamometers at the Pemaquid site four study sites ranged from 4.06-24.41 individuals/
were being continuously vandalized because of heavy cm2 in 1995 and 1996 (Table 1). The canopy decreased
foot traffic. I therefore removed the wave meters from barnacle recruitment at all sites largely because of me-
this site and no data were collected there throughout chanical abrasion of the substrate (i.e., "algal whip-
the rest of the summer. lash"; Dayton 1971, Menge 1976, Leonard 1999a) and
To test the hypothesis that regional variation in spe- this effect did not vary between regions (Leonard
cies interactions was associated with differences in 1999b).
thermal stress, I quantified the overall thermal regime In contrast, in both 1995 and 1996 the effect of the
at all sites during the summers of 1995 and 1996. In canopy on survival of barnacle recruits varied signif-
1995, maximum rock-surface temperatures were col- icantly between regions (Table 2). The canopy in-
lected approximately every 2 wk from June through creased survival at the southern sites but not at the
October using min/max thermometers (Taylor Scien- northern sites (Fig. 3). At the southern sites, survival
tific, model number 5458). Thermometers were placed was always highest under the canopy, intermediate
under the Ascophyllum canopy and in cleared plots in where the canopy had been removed, and lowest above
the high zone of all sites (n = 2 thermometers per the algal zone (Fig. 3). At northern sites, survival was
1020 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
TABLE 1. Recruitment, survival, growth, and fecundity of Semibalanus balanoides recruits and adults during 1995 and 1996
at four rocky intertidal sites in New England.
Recruitment: Survival?
Stage Year Sitet (no./cm2) n (%) n
Recruits 1995 Sakonnet, RI 14.38 + 2.01 22 6.3 + 2.3 22
Middleton, RI 14.77 + 2.60 24 9.0 + 3.3 22
Chamberlain, ME 9.36 + 1.03 23 30.6 + 3.0 23
Pemaquid, ME 6.22 + 0.96 24 14.1 ? 3.5 22
1996 Sakonnet, RI 24.41 + 2.12 24 10.3 + 2.4 24
Middleton, RI 17.36 + 1.95 24 12.6 + 6.2 24
Chamberlain, ME 7.65 + 1.05 24 46.1 + 6.0 24
Pemaquid, ME 4.06 + 0.77 24 38.0 + 6.4 21
Adults 1995 Sakonnet, RI NA NA -60.01 + 28.05 14
Middleton, RI NA NA -57.00 + 26.52 16
Chamberlain, ME NA NA -45.91 + 20.72 15
Pemaquid, ME NA NA 12.15 + 6.27 16
1996 Sakonnet, RI NA NA 66.90 + 15.56 16
Middleton, RI NA NA 18.50 ? 21.05 16
Chamberlain, ME NA NA 3.90 + 17.41 15
Pemaquid, ME NA NA 33.92 + 18.36 15
Notes: For each parameter, both the site means + 1 SE and the sample size (n) are given. These site means were used to
generate the proportional deviations used in the ANOVAs to test for variable species interactions between regions (see
Methods: Statistical analysis for further clarification). Reproductive fitness potential (survival X fecundity) was also calculated
from these data.
t RI = Rhode Island, USA; ME = Maine, USA.
:. Recruitment was not applicable (NA) for adult barnacles because recruitment refers only to input from the planktonic
larval pool.
? For adult barnacles, "survival" refers to the net change in adults (measured as no./100 cm2) between time periods. This
overall measure is the sum of an increase due to recruits that survive beyond the first year (and hence, become reclassified
as adults) and a decrease due to mortality of established adults.
generally similar under the canopy and in cleared plots cleared plots at southern sites (Fig. 3) could not there-
but was always highest above the algal zone (Fig. 3). fore be attributed to predation in these treatments.
Most importantly, the strength of the positive effects While the canopy had regionally variable effects on
of the canopy at southern sites varied among years (Fig. recruit survival, it had universally negative effects on
3: compare 1995 to 1996 for southern sites). In 1995 growth and reproduction. The canopy inhibited recruit
at southern sites there was very low survival in cleared growth in both 1995 and 1996 (Fig. 5; 1995, F,2
plots (in fact, 1 SE of the mean overlaps 0 individuals/ 20.23, P = 0.046; 1996, F,2 = 249.20, P = 0.004),
cm2) and 100% mortality above the algal zone. How- but this effect did not vary between regions in either
ever, recruit mortality at the southern sites in 1996 was year (1995, Fl,2 = 2.06, P = 0.287; 1996, F1,2 = 0.04,
less severe, as evidenced by the considerable survival P = 0.855). Similarly, the canopy decreased recruit
in the cleared plots and especially that above the algal fecundity in both years (Fig. 5; 1995, F ,2 = 31.42, P
zone. = 0.030; 1996, Fl,2 = 30.22, P = 0.032) and this did
Although the exclusion cages did not effectively con- not vary between regions (1995, F,2 = 2.36, P = 0.264;
trol predator densities (see Methods: Barnacle demo- 1996, F1,2 = 0.03, P = 0.873).
graphics ..., above), barnacle mortality patterns in When integrated, these data indicate that the net ef-
uncaged quadrats at northern sites were related to dif- fect of the algal canopy on recruit reproductive fitness
ferences in predator densities in cleared, canopy, and potential varied between northern and southern sites
above-canopy plots. During both years, Nucella lapil- and between years. In 1995 the canopy increased re-
lus at northern sites were in greatest abundance under productive fitness potential at southern sites but de-
the algal canopy, intermediate in cleared plots and ab- creased it at northern sites (Fig. 5, F,]2 = 4.17, P
sent above the algal canopy (Fig. 4; 1995, F,27 = 17.88, 0.046). In 1996 the canopy decreased reproductive fit-
P < 0.001; 1996, F,28 = 4.551, P = 0.038). The low ness potential at all sites (Fig. 5, F,2 = 7.69, P = 0.008)
recruit survival in canopy and cleared plots was as- and this effect did not vary between regions (F1,2
sociated with the presence of predators while the high 0.04, P 0.868).
survival above the algal canopy was associated with Above the algal canopy, recruit growth, fecundity,
the absence of predators (compare Figs. 3 and 4). In and reproductive fitness potential were generally great-
contrast, although N. lapillus were present at the south- er at northern sites than at southern sites (Fig. 5) al-
ern sites (unpublished data) they were never observed though these relationships were clouded by high var-
in the high zone during the two years when this ex- iability at southern sites. In 1995 no recruits survived
periment was done (Fig. 4). Lower survival in the in this zone at southern sites and fitness was therefore
April 2000 VARIATION IN SPECIES INTERACTIONS 1021
TABLE 1. Extended.
H A. nodosum
Growth Fecundity
L]S. balanoides
80
(mg/ind.) n (mg/ind.) n
0.73 + 0.12 15 0.16 + 0.09 15 60
6
0.60 + 0.07 14 0.09 + 0.04 14 o
3.13 + 0.49 24 1.54 + 0.33 24
2.54 + 0.45 22 1.21 + 0.23 22 |S 40
1.22 + 0.13 23 0.10 + 0.04 23
0.99 + 0.14 20 0.06 + 0.05 20 20
2.23 + 0.37 24 1.17 + 0.22 24
1.70 + 0.24 22 0.74 + 0.13 22
1.58 + 0.16 16 0.75 + 0.10 16 0
O
1.62 + 0.16 15 0.94 + 0.13 15 Sak. Midd. Cham. Pem.
6.69 + 1.05 15 3.86 ? 0.80 15
2.86 ? 0.62 13 1.59 + 0.41 13
2.57 + 0.37 16 1.20 + 0.26 16 South North
2.28 + 0.40 14 1.19 + 0.31 14 FIG. 2. Zonation patterns at replicate study sites north
1.05 + 0.31 16 0.54 + 0.20 16 and south of Cape Cod. Data are the percentages of days
2.86 + 0.62 13 1.59 + 0.41 13
(mean ? 1 SE) that the top of the Ascophyllum nodosum and
Semibalanus balanoides zones are exposed to aerial condi-
tions at each site. Sak. = Sakonnet Point, Rhode Island; Midd.
= Middleton, Rhode Island; Cham. = Chamberlain, Maine;
zero. In 1996 neither recruit growth, fecundity, nor and Pem. = Pemaquid Lighthouse, Maine.
fitness were significantly different between northern
and southern sites above the algal canopy (all F, 2 <
7.39, P > 0.113). This was largely due to the high was statistically insignificant (Table 3). This was due
variability in fecundity and reproductive fitness poten- to the high variation in canopy effects among sites
tial at southern sites but not at northern sites (Fig. 5). within regions (Table 3). This year-to-year variation in
canopy effects on adult survival at southern sites was
Variable species interactions: adult barnacles
analogous to that seen for barnacle recruits.
The strength of negative and positive interactions As with the recruits, adult growth and fecundity were
between the canopy and the underlying adult barnacles reduced in the presence of the canopy during both years
also differed between regions and years. During 1995, (Fig. 7; Growth: 1995, F1,2 = 36.69, P = 0.024; 1996,
=
survival was elevated under the canopy at southern Fl,2 = 102.59, P = 0.010; Fecundity: 1995, F,2
sites but reduced by the canopy at northern sites (Fig. 35.08, P = 0.027; 1996, = 177.68, P = 0.006).
F,2
6, Table 3). The overall pattern of survival in 1996 was Similarly, these effects did not vary between region
similar to that in 1995 (Fig. 6) but its magnitude was (Growth: 1995, F,2 = 4.69, P = 0.156; 1996, F,2 =
lower and the resulting canopy X region interaction 6.77, P = 0.122; Fecundity: 1995, F,2 = 8.02, P =
TABLE2. Survival of barnacle recruits as a function of region and site during 1995 and 1996, analyzed using nested analysis
of covariance.
Source of variation df MS F Denominator MSt P
1995
Region 1 3.51 0.42 a 0.583
Site(Region) 2 8.32 14.02 b <0.001
Canopy 1 14.22 38.61 c 0.025
Canopy X Region 1 18.47 50.15 c 0.019
Canopy X Site(Region) 2 0.37 0.62 b 0.542
Recruit density 1 3.49 5.88 b 0.019
Residual 52 0.59
1996
Region 1 0.22 1.20 a 0.338
Site(Region) 2 0.18 1.12 b 0.336
Canopy 1 1.27 102.25 c 0.010
Canopy X Region 1 4.02 324.16 c 0.003
Canopy X Site(Region) 2 0.01 0.08 b 0.928
Recruit density 1 2.02 12.18 b 0.001
Residual 55 0.17
Notes: Recruit density [ln(x + 1)-transformed] was the covariate, and final density [ln(n + 1)-transformed] was the dependent
variable. The statistical assumption of homogeneity of slopes was satisfied in both years: 1995, F,45 = 1.81, P = 0.109;
1996, F748 = 1.06, P = 0.406. In addition, the range of the covariate was broadly similar among the study sites.
t Denominator MS for the F ratios are: a = Site(Region), b = Residual, c = Canopy X Site(Region).
1022 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
* * Canopy
- Canopy
Cleared I_ Cleared
mi Above Canopy Ei Above
Canopy
3
1995 199
2.0-
2 1.5-
4-
I
1
E 0.5-
NA .
0
o O 0.0-
1996 2.5-
A
4
2.0-
I
3- is-
C) 1.0-
2- a' 0.5-
1- 0.0- NA
~ 1
0- q-
North South
QB 2-
2-
FIG. 3. Survival of barnacles that recruited during the io
spring of each study year at the end of October in 1995 and 0 "
1996 in 25 X 25 cm2 quadrats located under the canopy, in
areas cleared of the canopy or above the zone influenced by
1-
'J l I - rll F
f
the canopy. Data are adjusted densities (mean ? 1 SE) from 0-
the analysis of covariance using recruitment density as the North South North South
covariate and final density as the dependent variable. See
Table 2 for statistical analyses. FIG. 5. Demographic parameters of barnacle recruits dur-
ing 1995 and 1996. Parameters measured were growth, fe-
cundity, and reproductive fitness potential (survival X fe-
cundity). Values are the deviation (mean + 1 SE) of each pa-
rameter from its site mean. NA indicates that no individuals
3.6- 1995 survived and that growth and fecundity could therefore not
be measured.
2.4-
0.105) except in 1996 when fecundity was reduced
more under the canopy in northern sites than in south-
1.2- ern sites (Fl,2 = 31.05, P = 0.031).
trn Like those for recruits, these data for adults show
0 0 0
O
S o.o that canopy effects on adult reproductive fitness varied
o0.0
6 North South between regions and between years. In 1995 adult re-
0
1996 productive fitness potential was facilitated by the can-
. 3.6 opy at southern sites but was reduced by the canopy
* Canopy at northern sites (Fig. 7, F ,2 = 23.15, P = 0.041). In
a 2.4 Cleared 1996 the canopy had a consistently negative effect on
[-
Above Canopy reproductive fitness potential at both northern and
; southern sites (Fig. 7, F, 2 = 24.87, P = 0.038). These
1.2 results for adults were strikingly similar to those ob-
tained for barnacle recruits.
0 0 O0
0.0
North South Abiotic conditions
FIG.4. Predator densities as a function of region and treat- Measurements of physical factors suggested that
ment during 1995 and 1996. Predators (Nucella lapillus) were thermal regime but not wave exposure differed between
sampled in the permanent quadrats three times during each Maine and Rhode Island. There was no evidence that
summer. Densities were averaged over the summer. Values northern and southern sites were of different wave ex-
in the figures are means ? 1 SE. Although present in the low
zone (data not shown), N. lapillus were never found in the posure (Fig. 8) as maximum wave force did not vary
high zone at southern sites or above the canopy at northern between regions from December 1995 through May
and southernsites. 1996 (Table 4). From May through September there
April 2000 VARIATIONIN SPECIESINTERACTIONS 1023
* Canopy
D- Cleared
FIG.6. Survivalof adultbarnacles(non-re- 1995
E 80-
cruits)during1995 and 1996. Valuesarethe net
change (mean + 1 SE) in the numberof adults/
100 cm2 from November 1994 to November
1995 (1995) and from November 1995 to No- 0 0-
vember 1996 (1996). See Table 3 for statistical
analyses.
V -160
Z North South North South
was also no overall difference among regions in max- Like 1995, temperatures also varied significantly be-
imum wave force, but exposure did vary among regions tween regions over time (Table 5). In early summer
over time during this latter half of the summer (Fig. 8, there was little difference in rock temperature between
Table 4). When tropical storm Daniel impacted New regions, but by the end of July temperatures at southern
England in July 1996, there were large waves at Cham- sites were consistently several degrees higher than
berlain (and likely Pemaquid, although not measured those at sites in Maine (Fig. 9).
there) but not at Middletown or Sakonnet (personal Land-based weather-station data corroborated these
observation). Other than this single time period, wave direct measures of temperature and indicated that
exposure differed very little between any of the study Rhode Island was, on average, several degrees warmer
sites. than Maine (Fig. 10). This was true in 1995 and 1996
In contrast, there were subtle but potentially biolog- although the magnitude of this difference was smaller
ically important differences in thermal regime between in 1996 (i.e., 1.69? vs. 2.40?C, Table 6). These data
northern and southern sites (Fig. 9). This was evi- also revealed that 1995 was a warmer summer overall,
denced by both rock surface temperatures (Fig. 9) and especially in Rhode Island (Fig. 10). This finding is in
weather station data (Fig. 10). In summer 1995, overall agreement with other meteorological records that show
rock-surface temperatures did not differ between re- 1995 was the hottest summer on record since 1900
gions but did vary between regions over time (Table (Easterling et al. 1997).
5). The algal canopy reduced rock surface temperatures
DISCUSSION
by -6.6?C at both northern and southern sites (Fig. 9,
Table 5). Contrary to expectations, however, temper- My results contribute to the continuing debate on the
atures in the open plots of the northern sites in early influence of site "quality" on species interactions and
summer were higher than those of the southern sites suggest that the intensity and direction of interactions
(Fig. 9). By the beginning of August this pattern had can change with physical stress and predation intensity
reversed and southern sites were slightly but consis- at large spatial scales. In this study, subtle differences
tently warmer than northern sites until October. in temperature north and south of Cape Cod and re-
In summer 1996, rock temperatures were nearly sig- gional differences in predator abundance were asso-
nificantly different between regions (i. e., P = 0.094; ciated with differences in negative vs. positive inter-
Table 4). Overall, rock temperatures at southern sites actions between the algal canopy and the underlying
were 1.5?C warmer than at the northern sites (Fig. 9). barnacles. In addition, year-to-year variation in the net
TABLE 3. Results of nested ANOVAs on adult barnacle survival (measured as yearly changes in adult barnacle density (no./
100 cm2) during 1995 and 1996.
1995 1996
Denom- Denom-
Source of variation df MS F inator MS P MS F inator MS P
Region 1 24 644.8 1.73 a 0.320 7517.3 0.57 a 0.528
Site(Region) 2 14 284.7 3.96 b 0.025 13 110.3 2.99 b 0.059
Canopy 1 49 104.5 11.69 c 0.076 4290.1 0.28 c 0.650
Canopy x Region 1 165 730.5 39.45 c 0.024 26 164.3 1.71 c 0.322
Canopy x Site(Region) 2 4200.7 1.16 b 0.320 15 340.4 3.50 b 0.037
Residual 54 3611.9 4386.9
Notes: Analysis of covariancewas not used (as in Table 2) because the covariate(initial adult density) was not broadly
similar among the study sites. The denominator for the F ratios were: a = Site(Region), b = Residual, c = Canopy X
MS
Site(Region).
1024 GEORGEH. LEONARD Ecology, Vol. 81, No. 4
* Canopy 120-
-0- South
[7 Cleared
Z -- North
2.0- 1995 1996
80-
0
- 1.5-
+ -t- -- ?- B
o 1.0-
40-
0.5- ed
cX
0-
i I a
I I I I I I I I I
2.0- - 1I Dec Jan Feb Mar Apr May Jun Jul Aug Sep
4 1.5- 1996
= 1.0-
-+
- FIG. 8. Wave exposure at northern and southern sites col-
0 lected from December 1995 through October 1996 using
L; 0.5- spring-loaded dynamometers. Data are maximum force
I (means ? 1 SE) estimated from spring displacement measured
0- at approximately monthly intervals.
- 2.0- - I
; 1.5- cundity) were largely consistent with work done in in-
tertidal systems in the past. The canopy inhibited bar-
1.0-
nacle recruitment at all study sites in both years via
'
0.5- mechanical "whiplashing" of the substrate (Leonard
-j ----r_B-_ 1999a). This effect has often been found on wave-
North South beaten shores (Dayton 1971, Menge 1976, Grant 1977,
FIG. 7. Demographic parameters fo r adultbarnaclesdur- Hawkins 1983) and appears to be a common feature of
ing 1995 and 1996. Data and presenta tion are analogousto large, canopy-forming algal assemblages.
those in Fig. 5. Similarly, barnacle growth and fecundity were sig-
nificantly lower under algal canopies than in cleared
plots. This common observation (Lewis 1964, Wethey
effect of the algal canopy on reprc )ductive fitness po- 1985) was similar at both northern and southern sites
tential occurred because of variati{ in barnacle sur- and did not vary between years. Interestingly, growth
on
vival rather than reproductive oultput. These results rates (and fecundity) were not enhanced under algal
suggest that the local persistence { species in some canopies compared to open plots at southern sites,
of
intertidal habitats may be dependemnt the presence where higher temperatures were hypothesized to im-
on
of other species that can modify the increasingly severe pose severe physiological stress on metabolic and re-
environment predicted under globzal warming scenar- productive processes (cf. Barnes and Barnes 1959, Cos-
ios. sins and Bowler 1987). Artificially shading barnacles
has been shown to alleviate these stresses and result
Constituent effects of the algal calnopy on barnacles in higher growth rates (Bertness 1989, Bertness and
The individual effects of the Asc:ophyllum nodosum Gaines 1993). However, the physical structure provided
canopy on barnacles (i.e., recruitmc growth, and fe- by algal canopies also alters patterns of water flow
ent,
TABLE4. Results of repeated-measures nested ANOVAs on wave-exposure data.
December 1995-May 1996 June-September 1996
Denom- Denom-
inator inator
Source of variation df MS F MS P df MS F MS P
Region 1 58.83 1.40 a 0.359 1 204.80 0.66 a 0.566
Site(Region) 2 42.14 0.15 b 0.861 1 310.08 1.81 b 0.203
Instrument(Site) 16 278.67 12 171.36
Time 3 6.66
121.13 c 0.001 5 2267.87
38.24 c <0.001
Time X Region 3 1.23
224.98 d 0.379 5 984.25
8.57 d 0.017
Time X Site(Region) 6 10.10
183.57 c <0.001 5 1.65
114.90c 0.218
Time x Instrument(Site) 48 18.18 60 69.77
Notes: Analyses were done separately for the period December 1995-May 1996 and June 1996-September 1996 because
instruments were vandalized at Pemaquid (Maine, USA) during the summer of 1996, and data could not be collected there.
The dependent variable was maximum force (in newtons) calculated from the spring-loaded dynamometers. The denominator
MS for the F ratios were: a = Site(Region), b = Instrument(Site), c = Time x Instrument(Site), d = Time X Site(Region).
April 2000 VARIATION IN SPECIES INTERACTIONS 1025
--- South:Cleared to feed on Semibalanus balanoides (Menge 1983). Any
- South: Canopy direct reduction in mortality under the canopy because
-D- North:Cleared of lower thermal stress at northern sites was apparently
--* North:Canopy offset by an indirect increase in mortality from pred-
ators mediated by the canopy (Menge 1978, Minchin-
40-1995
ton and Scheibling 1993).
Counter to my original hypothesis, results for the
35 -
adults were similar to those of the recruits for both
35-c,,
30-v years. Adults, like recruits, were sensitive to the ex-
perimental removal of the canopy and to the ensuing
25 - changes in environmental conditions and predator
-L
abundance. The similarity in response of recruits and
o
20-
adults to canopy manipulation suggests recruits in these
a
habitats are unlikely to escape thermal conditions over
15 time by growing in size (Foster 1971).
E Abiotic conditions associated with variation in
H 1996 barnacle survival
40-
Variation in survival rates was not associated with
E differences in wave exposure (Fig. 8) but may have
35-
ct
30- 8
1995: N vs. S
25-
20- ,n iaA
INA
Jun Jul Aug Sep Oct
V' ri
I I I I I
FIG. 9. Thermal characteristics at northern and southern
sites collected during the summers of 1995 and 1996 with
min/max thermometers. Values are the maximum temperature 1996: N vs. S
(mean + 1 SE) recorded during each preceding interval. In
1995, data were collected in cleared plots and under the can- a)
opy (n = 2 plots per treatment per site). In 1996, data were
collected less frequently and only in cleared plots but with 0
... tA
higher replication within sites (n = 8 plots per site). H
~
-8 ,*~?wrN vv IVV
E
(Eckman et al. 1989) and can modify feeding behavior -8
I I I I I
by interfering with deployment of the feeding structure
(Palmer et al. 1982). In my study, positive effects of CI- 8
the algal canopy on organism growth (through allevi- S: 1995 vs. 1996
ation of physiological stress) were apparently out-
weighed by negative effects on food acquisition.
In contrast to these negative effects, the algal canopy 0
had a large positive impact on organism survival, but
this was evident only at the southern sites. In both
years, barnacle survival in Rhode Island was highest
-8
,A^AI
under the canopy, intermediate in cleared plots, and Apr May Jun
Apr May Jun Jul Aug Sep
Jul Aug Sep
lowest in the zone above the algal canopy. Gastropod
predators were rare at these sites and mortality rates FIG. 10. Temperature anomalies (deviations in ?C) be-
were consistent with those `expected from variation in tween weather stations located near the open coast in the north
thermal stress alone. This contrasts sharply with north- (Maine) and south (Rhode Island). Data are 7-d running av-
erages of the differences in daily maximum air temperature
ern sites, where survival was generally similar under collected from 1 April through 1 October. Top panel: Dif-
the canopy and in cleared plots but was always highest ference between northern and southern sites during 1995.
above the algal zone. The abundance of the carnivorous Middle panel: Similar comparison for 1996. Bottom panel:
Difference between 1995 and 1996 at the southern sites. Neg-
gastropod Nucella lapillus at these northern sites was ative differences (in the middle and top panels) indicate that
high under the canopy, intermediate in cleared plots, the north was cooler than the south. Positive differences (in
and low at higher tidal heights. These predators often the bottom panel) indicate that 1995 was warmer than 1996.
seek refuge under the algal canopy and are well known Summary statistics are shown in Table 6.
1026 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
TABLE 5. Results of repeated-measures nested ANOVAs on substrate temperature (?C) as a function of region, site, canopy,
and time.
Denominator
Source of variation df MS F MS P
Summer 1995
Region 1 0.48 0.01 a 0.935
Site(Region) 2 56.50 2.76 b 0.123
Canopy 1 1935.55 324.30 c 0.003
Canopy x Region 1 0.54 0.09 c 0.793
Canopy X Site(Region) 2 5.97 0.29 b 0.755
Thermometer(Canopy x Site) 8 20.47
Time 10 92.64 20.47 d <0.001
Time X Region 10 60.50 11.05 e <0.001
Time x Site(Region) 20 5.47 1.21 d 0.326
Time x Canopy 10 6.27 1.34 f 0.277
Time X Canopy x Region 10 3.28 0.70 g 0.715
Time x Canopy X Site(Region) 20 4.69 1.04 d 0.430
Time X Thermometer(Canopy X Site) 80 4.53
Summer 1996
Region 1 119.31 9.12 a 0.094
Site(Region) 2 13.09 0.58 h 0.566
Thermometer(Site) 28 22.52 <0.001
Time 6 69.36 23.22 i 0.019
Time X Region 6 43.21 4.02 e 0.001
Time x Site(Region) 12 10.76 3.60 i
Time X Thermometer(Site) 168 2.99
Notes: The dependent variable was maximum temperature of the rock substrate over 2-wk periods during the summers of
1995 and 1996. In 1996 the canopy treatment was eliminated, and within-site replication was increased from 2 to 8 ther-
mometers/site. Denominator MS for the F ratio were: a = Site(Region), b = Thermometer(Canopy x Site), c = Canopy X
Site(Region), d = Time X Thermometer(Canopy X Site), e = Time x Site(Region), f = Canopy x Site(Region), g = Time
x Canopy X Site(Region), h = Thermometer(Site), and i = Time x Thermometer(Site).
been related to subtle differences in climatic conditions air and rock temperatures at southern sites commonly
in 1995 and 1996 in the absence of predators. Weather- approached the thermal limits of Semibalanus balan-
station data from Rhode Island and Maine indicated oides (i.e., 40?C; Southward 1958, Foster 1969, 1971),
that 1995 was a warmer summer than 1996 (Fig. 10), further suggesting a causal link between yearly vari-
in agreement with global climate data showing 1995 ation in thermal stress and variation in species inter-
was the warmest year since 1900 (Easterling et al. actions.
1997). The correspondence between my experimental These conclusions, however, should be taken cau-
results and these yearly climatic conditions suggests tiously because the direct measures of rock temperature
that these species interactions may be sensitive to sub- did not closely match the indirect measures of local
tle overall differences in temperature in regions where conditions made using weather-station data. For ex-
the influence of predators is minimal. Differences in ample, weather-station data indicated that Maine was
air temperature of only a few degrees have also been cooler than Rhode Island during June-July in 1996 but
hypothesized by Wethey (1983) to affect the distribu- direct measurements of substrate temperature showed
tion and abundance of intertidal organisms. Although no difference among the regions at this time. This may
tissue temperatures were not measured, both maximum partly be because overall weather patterns may not ac-
curately predict physical conditions at specific study
TABLE6. Summary statistics of daily temperature anomalies sites (see Helmuth 1998, 1999) as well as due to dif-
(difference between Maine and Rhode Island) in maximum ferences in sampling frequency between thermometers
temperature collected at weather stations near the study and weather-station data.
sites duringthe summersof 1995 and 1996. Differences in instrument replication also make com-
Statistic 1995 1996 parisons among years difficult. For example, during
1995 when species interactions varied between regions,
Negative anomalies (%) 70 65
Mean daily temperature diff. (?C) -2.40 -1.69 low thermometer replication and low statistical power
1 SD of mean daily temp. diff. (?C) 4.29 4.15 made it difficult to detect differences in substrate tem-
1 SE of mean daily temp. diff. (?C) 0.32 0.31
-10.30
perature between northern and southern sites. Average
Minimum daily difference (?C) -15.11
Maximum daily difference (?C) 7.83 8.54 differences in temperature of only 1-3?C, however, may
n 183 183 be biologically important but difficult to detect using
Note: Fig. 10 (top and middle panels) graphically presents min/max thermometers. Instead of thermometers,
the 7-d running average of the original 183 daily anomalies. Leonard et al. (1999) used computer-controlled therm-
April 2000 VARIATIONIN SPECIESINTERACTIONS 1027
istors to readily detect differences in substrate tem- predators in a region of reduced temperatures). My
perature <2.5?C among New England intertidal habi- results support the hypothesis that small differences in
tats that had significant effects on the mortality of Sem- physical factors can be important in governing species
ibalanus balanoides. Although expensive, these therm- interactions over 100s of kilometers and that the com-
istors may be the best means to detect subtle differences bination of direct and indirect effects that vary across
in physical conditions among sites. In lieu of these the life history determine the magnitude and direction
instruments, weather-station data may be useful in re- of overall species interactions. Long-term experiments
lating experimental results to broad-scale climatic con- at a series of sites along the New England coastline
ditions (Hargrove and Pickering 1992). (which would increase statistical power) will be nec-
Finally, it should be noted that my study was done essary to evaluate the validity and generality of these
at only two sites in each region over two years. This conclusions.
experimental design resulted in low statistical power
to detect differences among regions and cautions that The changing nature of species interactions during
my findings may not apply generally to all New Eng- global climate change
land intertidal habitats. Despite these limitations, my Few scientists question that humans have signifi-
data are the first to document the previously inferred cantly altered global carbon and nitrogen cycles and
difference in physical conditions across the Cape Cod have had profound effects on climate (Vitousek et al.
peninsula that has been argued to be important in the 1997). Greenhouse gases are expected to increase at
distribution and ecology of rocky-shore organisms least into the 21st Century, resulting in a 1.5-4.5?C
(Wethey 1983). warming trend and an increase in the temporal and
Direct and indirect interactions and the net effect on spatial variability of many aspects of the global climate
(Gates 1993). Temperature-sensitive species and those
organism fitness with whom they interact are likely to be influenced by
Although examples of positive interactions are be- these changing conditions.
coming increasingly common in the literature (e. g., As climate continues to warm, algal canopies in the
Carlsson and Callaghan 1991, Bertness and Shumway high intertidal zone of sites in southern New England
1993, Agular and Sala 1994, Berkowitz et al. 1995, may buffer species from local changes in distribution.
Callaway 1995), few studies have taken a demographic The upper limit of barnacles may largely be set by the
approach to evaluate the nature of species interactions. presence of algal canopies in the future, while in areas
For example, although Carlsson and Callaghan (1991) without canopies the vertical limit may be much lower.
showed that a sedge growing with a shrub in the Arctic Species persistence at sites to the north is harder to
tundra had a higher growth rate than when growing predict because of the indirect effect of the canopy on
alone, there was no attempt to determine effects of the barnacle predators. As in the south, barnacles may
shrub on seed supply, germination success, or mortal- come to rely on the buffering capacity of these common
ity, all of which directly contribute to fitness. Inte- species of algae, especially if the predators shift their
grating these effects as "lxmx" (survival probability X distribution more than that of the barnacles. Alterna-
reproductive output) is critical in determining if a net tively, as climate warms, the refuge from predators
interaction is positive or negative. above the algal canopy at northern sites (Menge 1976)
In my study, integrating component parts revealed may vanish and predator-prey interactions under the
that the overall interaction between the algal canopy canopy may intensify. This could result in local ex-
and barnacles switched from negative to positive be- tinction of either predator, prey, or both. The outcome
tween regions to the north and south of Cape Cod. will depend on the relative susceptibility of predator
Moreover, this interaction differed between the two and prey to regional changes in climate (Menge and
years of the study. Because of an overwhelmingly large Olson 1990) and their interactions with organisms that
direct effect on survival in 1995 at the southern sites, can modify the local conditions. Variable species in-
the algal canopy facilitated the reproductive fitness po- teractions, such as those documented here, are likely
tential of barnacle recruits there. On the other hand, to be evident in other habitats as well, where a single
the net interaction with the canopy was consistently species or suite of species strongly moderates physical
negative at northern sites, evidently due to canopy- conditions. Further work in these and other habitats
mediated indirect effects on predation pressure com- would contribute substantially to our understanding of
bined with lower reproductive output under the canopy. how ecological and evolutionary dynamics in natural
In 1996, when the canopy-mediated effect on survival communities might change over the next century.
was weaker at southern sites, fitness did not vary be-
ACKNOWLEDGMENTS
tween regions.
From these data, I conclude that net species inter- I am indebted to M. Baker, M. Bertness, S. Brewer, J. Eddy,
A. Ingraham, T. Leonard, J. Levine, and P. Schmidt, all of
actions switched from positive at southern sites (be- whom helped in the field during less-than-ideal conditions.
cause of few predators and elevated temperatures) to T. Miller and K. Eckelbarger of the University of Maine fa-
negative at northern sites (because of a suite of boreal cilitated my stays at the Darling Marine Center and the U.
1028 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
S. Fish and Wildlife Service kindly provided access to the Connell, J. H. 1961. The influence of interspecific compe-
Middletown site at the Sachuest Point National Wildlife Ref- tition and other factors on the distribution of the barnacle
uge. M. Bertness, S. Gaines, and D. Rand contributed sig- Chthamalus stellatus. Ecology 42:710-723.
nificantly to the development of the ideas in this paper. T. Connell, J. H. 1985. The consequences of variation in initial
Leonard steadfastly supported me during this work. Earlier settlement vs. post-settlement mortality in rocky intertidal
versions of the manuscript were greatly improved by com- communities. Journal of Experimental Marine Biology and
ments from K. Earls, T. Rand, P. Ewanchuk, J. Witman, J. Ecology 93:11-45.
Bruno, T. Minchinton, B. Menge, and two anonymous re- Cossins, A. R., and K. Bowler. 1987. Temperature biology
viewers. This work was funded, in part, by the Andrew Mel- of animals. Chapman & Hall, London, UK.
lon Foundation and grants from Sigma Xi and The Sounds Crisp, D. J. 1959. Factors influencing the time of breeding
Conservancy to G. H. Leonard and an NSF grant to M. Bert- of Balanus balanoides. Oikos 10:275-289.
ness. Crisp, D. J. 1964. Racial differences between North Amer-
ican and European forms of Balanus balanoides. Journal
LITERATURE CITED
of the Marine Biological Association of the United King-
Agular, M. R., and 0. E. Sala. 1994. Competition, facilitation, dom 44:33-45.
seed distribution and the origin of patches in a Patagonian
Dayton, P. K. 1971. Competition, disturbance, and commu-
steppe. Oikos 70:26-34. nity organization: the provision and subsequent utilization
Allee, W. C. 1923. Studies in marine ecology. IV. The effect of space in a rocky intertidal community. Ecological Mono-
of temperature in limiting the geographical range of in-
graphs 41:351-389.
vertebrates of the Woods Hole littoral. Biological Bulletin
Denny, M. W. 1983. A simple device for recording the max-
IV:341-354. imum force exerted on intertidal organisms. Limnology and
Bakus, G. J. 1969. Energetics and feeding in shallow marine Oceanography 28:1269-1274.
waters. International Review of General and Experimental Dethier M. N., and D. O. Duggins. 1988. Variation in strong
Zoology 4:275-369. interactions in the intertidal zone along a geographical gra-
Barnes, H . 1958a. Regarding the southern limits of Balanus dient: a Washington-Alaska comparison. Marine Ecology
balanoides (L.). Oikos 9:139-157.
Progress Series 50:97-105.
Barnes, H . 1958b. Temperature and the life-cycle of Balanus
balanoides (L.). Friday Harbour Symposium on Limnoria Duggins, D. O., J. E. Eckman, and A. T. Sewell. 1990. Ecol-
and Barnacles. University of Washington, Seattle, Wash- ogy of understory kelp environments. II. Effects of kelps
on recruitment of benthic invertebrates. Journal of Exper-
ington, USA. imental Marine Biology and Ecology 143:27-45.
Barnes, H., and M. Barnes. 1959. A comparison of the annual
Easterling, D. R., B. Horton, P. D. Jones, T. C. Peterson, T.
growth patterns of Balanus balanoides (L.) with particular R. Karl, D. E. Parker, M. J. Salinger, V. Razuvayev, N.
reference to the effect of food and temperature. Oikos 10:
1-18. Plummer, P. Jamason, and C. K. Folland. 1997. Maximum
and minimum temperature trends for the globe. Science
Bell, E. C., and M. W. Denny. 1994. Quantifying "wave 277:364-367.
exposure": a simple device for recording maximum veloc- Eckman, J. E. 1996. Closing the larval loop: linking larval
ity and results of its use at several field sites. Journal of
ecology to the population dynamics of marine benthic in-
Experimental Marine Biology and Ecology 181:9-29. vertebrates. Journal of Experimental Marine Biology and
Berkowitz, A. R., C. D. Canham, and V. R. Kelly. 1995.
Ecology 200:207-237.
Competition vs. facilitation of tree seedling growth and
survival in early successional communities. Ecology 76: Eckman, J. E., and D. O. Duggins. 1991. Life and death
1156-1168. beneath macrophyte canopies: effects of understory kelps
on growth rates and survival of marine, benthic suspension
Bertness, M. D. 1989. Intraspecific competition and facili-
tation in a northern acorn barnacle population. Ecology 70: feeders. Oecologia 87:473-487.
257-268. Eckman, J. E., D. O. Duggins, and A. T. Sewell. 1989. Ecol-
Bertness, M. D., and R. Callaway. 1994. Positive interactions ogy of understory kelp environments. I. Effects of kelps
in communities. Trends in Ecology and Evolution 9:191- on flow and particle transport near the bottom. Journal of
193. Experimental Marine Biology and Ecology 129:173-187.
Bertness, M. D., and S. D. Gaines. 1993. Larval dispersal Fischer, A. G. 1960. Latitudinal variations in organic diver-
and local adaptation in acorn barnacles. Evolution 47:316- sity. Evolution 14:64-81.
320. Foster, B. A. 1969. Tolerance of high temperatures by some
Bertness, M. D., S. D. Garrity, and S. C. Levings. 1981. intertidal barnacles. Marine Biology 4:326-332.
Predation pressure and gastropod foraging: a tropical-tem- Foster, B. A. 1971. Dessication as a factor in the intertidal
zonation of barnacles. Marine Biology 8:12-29.
perate comparison. Evolution 35:995-1007.
Bertness, M. D., and S. D. Hacker. 1994. Physical stress and Gates, D. M. 1993. Climate change and its global conse-
positive associations among marsh plants. American Nat- quences. Sinauer Associates, Sunderland, Massachusetts,
uralist 144:363-372. USA.
Bertness, M. D., and S. W. Shumway. 1993. Competition and Goldberg, D. E., and A. M. Barton. 1992. Patterns and con-
facilitation in marsh plants. American Naturalist 142:286- sequences of interspecific competition in natural commu-
292. nities: a review of field experiments with plants. American
Callaway, R. M. 1995. Positive interactions among plants. Naturalist 139:771-801.
Botanical Review 61:306-349. Gosner, K. L. 1978. A field guide to the Atlantic seashore.
Callaway, R. M., and L. R. Walker. 1997. Competition and Houghton Mifflin, Boston, Massachusetts, USA.
facilitation: a synthetic approach to interactions in plant Grace, J. B. 1991. A clarification of the debate between
communities. Ecology 78:1958-1965. Grime and Tilman. Functional Ecology 5:583-587.
Carefoot, T 1977. Pacific seashores. A guide to intertidal Grant, W. S. 1977. High intertidal community structure on a
ecology. University of Washington Press, Seattle, Wash- rocky headland in Maine, USA. Marine Biology 44:15-25.
ington, USA. Greenlee, J. T., and R. M. Callaway. 1996. Abiotic stress and
Carlsson, B. A., and T. V. Callaghan. 1991. Positive plant the relative importance of interference and facilitation in
interactions in tundra vegetation and the importance of montane bunchgrass communities in western Montana.
shelter. Journal of Ecology 79:973-983. American Naturalist 148:386-396.
April 2000 VARIATION IN SPECIES INTERACTIONS 1029
Grime, J. P. 1973. Competitive exclusion in herbaceous veg- tertidal community. Marine Ecology Progress Series 95:
etation. Nature 242:344-347. 233-244.
Grime, J. P. 1977. Evidence for the existence of three primary Moloney, K. A. 1990. Shifting demographic control of a
strategies in plants and its relevance to ecological and evo- perennial bunchgrass along a natural habitat gradient. Ecol-
lutionary theory. American Naturalist 111:1169-1194. ogy 71:1133-1143.
Grime, J. P. 1979. Plant strategies and vegetation processes. Paine, R. T. 1994. Marine rocky shores and community ecol-
John Wiley & Sons, New York. ogy: an experimentalist's perspective. Ecology Institute,
Hargrove, W. W., and J. Pickering. 1992. Pseudoreplication: Oldendorf/Luhe, Germany.
a sine qua non for regional ecology. Landscape Ecology Palmer, A. R., J. Szymanska, and L. Thomas. 1982. Pro-
6:251-258. longed withdrawal: a possible predator evasion behavior
Hawkins, S. J. 1983. Interactions of Patella and macroalgae in Balanus glandula (Crustacea: Cirripedia). Marine Bi-
with settling Semibalanus balanoides (L.). Journal of Ex- ology 67:51-55.
perimental Marine Biology and Ecology 71:55-72. Raimondi, P. T. 1990. Patterns, mechanisms, consequences
Helmuth, B. S. 1998. Intertidal mussel microclimates: pre- of variability in settlement and recruitment of an intertidal
dicting the body temperature of a sessile invertebrate. Ecol- barnacle. Ecological Monographs 60:283-309.
ogy 68:51-74. Rhode, K. 1992. Latitudinal gradients in species diversity:
Helmuth, B. S. 1999. Thermal biology of rocky intertidal the search for the primary cause. Oikos 65:514-527.
mussels: quantifying body temperatures using climatolog- Ricketts, E. E, J. Calvin, and J. W. Hedgpeth. 1985. Between
ical data. Ecology 80:15-34. Pacific tides. Stanford University Press, Stanford, Califor-
Jeanne, R. L. 1979. A latitudinal gradient in rates of ant nia, USA.
predation. Ecology 60:1211-1224. Rosenzweig, M. L. 1995. Species diversity in space and time.
Leonard, G. H. 1999a. Positve and negative effects of in- Cambridge University Press, Cambridge, UK.
tertidal algal canopies on recruitment and survival of bar- Sanford, E., D. Bermudez, M. D. Bertness, and S. D. Gaines.
nacles. Marine Ecology Progress Series. 178:241-249. 1994. Flow, food supply and acorn barnacle population
Leonard, G. H. 1999b. Population and community conse- dynamics. Marine Ecology Progress Series 104:49-62.
quences of physical forcing mechanisms in New England Southward, A. J. 1955. On the behavior of barnacles. I. The
intertidal habitats. Dissertation. Brown University, Provi- relation of cirral beat and other activities to temperature.
dence, Rhode Island, USA. Journal of the Marine Biological Association of the United
Leonard, G. H., P. J. Ewanchuk, and M. D. Bertness. 1999. Kingdom 34:413-422.
How recruitment, intraspecific interactions, and predation Southward, A. J. 1958. Note on the temperature tolerances
control species borders in a tidal estuary. Oecologia 118: of some intertidal animals in relation to environmental tem-
492-502.
peratures and geographical distribution. Journal of the Ma-
Lewis, J. R. 1964. The ecology of rocky shores. English rine Biological Association of the United Kingdom 37:49-
University Press, London, UK. 66.
Little, C., and J. A. Kitching. 1996. The biology of rocky Spight, T. M. 1976. Censuses of rocky shore prosobranchs
shores. Oxford University Press, Oxford, UK. from Washington and Costa Rica. Veliger 18:309-317.
MacArthur, R. H. 1965. Patterns of species diversity. Bio-
Stephenson, T. A., and A. Stephenson. 1948. The universal
logical Reviews 40:510-533. features of zonation between tide-marks on rocky coasts.
MacArthur, R. H. 1972. Geographical ecology: patterns in Journal of Ecology 37:289-305.
the distribution of species. Harper & Row, New York, New
Tilman, D. 1988. Plant strategies and the dynamics and struc-
York, USA. ture of plant communities. Princeton University Press,
Mayr, E. 1963. Animal species and evolution. Harvard Uni- Princeton, New Jersey, USA.
versity Press, Cambridge, Massachusetts, USA.
Underwood, A. J. 1986. Physical factors and biological in-
McPeek, M. A., and B. L. Peckarsky. 1998. Life histories
and the strengths of species interactions: combining mor- teractions: the necessity and nature of ecological experi-
ments. Pages 372-390 in P. G. Moore and R. Seed, editors.
tality, growth, and fecundity effects. Ecology 79:867-879. The ecology of rocky coasts. Columbia University Press,
Menge, B. A. 1976. Organization of the New England rocky New York, New York, USA.
intertidal community: role of predation, competition, and
environmental heterogeneity. Ecological Monographs 46: Underwood, A. J. 1997. Experiments in ecology. Their log-
355-393. ical design and interpretation using analysis of variance.
Menge, B. A. 1978. Predation intensity in a rocky intertidal Cambridge University Press, Cambridge, U.K.
community: effect of an algal canopy, wave action and Underwood, A. J., and E. J. Denley. 1984. Paradigms, ex-
dessication on predator feeding rates. Oecologia 34:17-35. planations, and generalizations in models for the structure
Menge, B. A. 1983. Components of predation intensity in of intertidal communities on rocky shores. Pages 151-180
the low zone of the New England rocky intertidal region. in D. R. Strong, Jr., D. Simberloff, L. G. Abele, and A. B.
Oecologia 58:141-155. Thistle, editors. Ecological communities: conceptual issues
Menge, B. A., and J. Lubchenco. 1981. Community orga- and the evidence. Princeton University Press, Princeton,
nization in temperate and tropical rocky intertidal habitats: New Jersey, USA.
prey refuges in relation to consumer pressure gradients. Vermeij, G. J. 1978. Biogeography and adaptation: patterns
Ecological Monographs 51:429-450. of marine life. Harvard University Press, Cambridge, Mas-
Menge, B. A., and A. M. Olson. 1990. Role of scale and sachusetts, USA.
environmental factors in regulation of community struc- Vermeij, G. J., and J. A. Veil. 1978. A latitudinal pattern in
ture. Trends in Ecology and Evolution 5:52-57. bivalve shell gaping. Malacologia 17:57-61.
Menge, B. A., and J. P. Sutherland. 1987. Community reg- Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M.
ulation: variation in disturbance, competition, and preda- Melillo. 1997. Human domination of Earth's ecosystems.
tion in relation to environmental stress and recruitment. Science 277:494-499.
American Naturalist 130:730-757. Wallace, A. R. 1878. Tropical nature and other essays. Mac-
Minchinton, T. E., and R. E. Scheibling. 1993. Free space Millan, London, UK.
availability and larval substratum selection as determinants Weldon, C. W., and W. L. Slauson. 1986. The intensity of
of barnacle population structure in a developing rocky in- competition versus its importance: an overlooked distinc-
1030 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
tion and some implications. Quarterly Review of Biology Wethey, D. S. 1985. Catastrophe, extinction, and species di-
61:23-44. versity: a rocky intertidal example. Ecology 66:445-456.
Wethey, D. S. 1983. Geographic limits and local zonation: Wilson, S. D., and P. A. Keddy. 1986. Measuring diffuse com-
the barnacles Semibalanus (Balanus) and Chthamalus in petition along an environmental gradient:results from a shore-
New England. Biological Bulletin 165:330-341. line plant community. American Naturalist 127:862-869.
Wethey, D. S. 1984. Sun and shade mediate competition in Wootton, J. T. 1993. Indirect effects and habitat use in an
the barnacles Chthamalus and Semibalanus: a field exper- intertidal community: interaction chains and interaction
iment. Biological Bulletin 167:176-185. modifications. American Naturalist 141:71-89.
Author(s): George H. Leonard
Source: Ecology, Vol. 81, No. 4 (Apr., 2000), pp. 1015-1030
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/177175
Accessed: 25/01/2010 14:13
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=esa.
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.
Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.
http://www.jstor.org
Ecology, 81(4), 2000, pp. 1015-1030
? 2000 by the Ecological Society of America
LATITUDINAL VARIATION IN SPECIES INTERACTIONS:
A TEST IN THE NEW ENGLAND ROCKY INTERTIDAL ZONE
GEORGE H. LEONARD'
Brown University, Department of Ecology and Evolutionary Biology, Providence, Rhode Island 02912 USA
Abstract. How species interactions vary as a function of habitat characteristics con-
tinues to be an important debate in ecology. Using the barnacle-seaweed assemblage com-
mon in New England rocky intertidal habitats, I tested the hypothesis that species inter-
actions switch from negative to positive at sites across the Cape Cod faunal barrier because
of latitudinal variation in thermal stress and predation intensity between these regions. I
manipulated Ascophyllum nodosum canopies in the high zone of two sites from each region
and then determined the consequences for underlying Semibalanus balanoides recruits and
adults for two years (January 1995 through January 1997). In both years, algal canopies
reduced barnacle recruitment and growth rates at all sites but greatly increased survival
rates only at the southern sites. When integrated in a demographic framework, these data
showed that the reproductive fitness potential of individuals was facilitated by the algal
canopy at southern sites but was reduced under similar conditions at northern sites. At
southern sites, this was likely the result of buffering from physical stresses in the absence
of predators. At northern sites, any buffering from physical stress was likely offset by an
increase in mortality in the presence of predators. Interestingly, this variability in species
interactions appeared to be associated with subtle variation in climate. Facilitation was
evident only during 1995, the warmest year on record since 1900. In contrast, interactions
were entirely negative during 1996, a slightly cooler year. These results suggest that species
interactions in the intertidal zone may be sensitive to even subtle changes in climate.
Interspecific buffering of neighbors from thermal stress is likely to be common in other
systems and suggests that both aquatic and terrestrial vegetation may become increasingly
important to local species persistence as climates change during the next century.
Key words: Ascophyllum nodosum; direct vs. indirect effects; fitness consequences; habitat
amelioration; interactions, positive and negative; latitude effects on species interactions; New England
(USA) rocky intertidal zone; Nucella lapillus; predation; Semibalanus balanoides; thermal stress.
INTRODUCTION ness et al. [1981], Menge and Lubchenco [1981] for
Pattern and process across biogeographic spatial exceptions). This scarcity of experimental data is at
scales have been of interest to ecologists since the time least partly due to the logistical difficulties of con-
of Darwin. Increases in diversity across latitudinal gra- ducting manipulative experiments at large spatial
dients and between ocean basins are well known (Fi- scales.
scher 1960, MacArthur 1965, 1972, Spight 1976, Wal- Over the last twenty years, work within single bio-
lace 1878), although the ultimate mechanism for their geographic regions has highlighted how abiotic con-
ditions can alter the relative importance of biological
origin and maintenance remains unclear (Rhode 1992,
Rosenzweig 1995). Biological processes have long processes in governing community structure (Connell
been argued to vary across large spatial scales (Mac- 1961, Menge 1976, Menge and Sutherland 1987).
Arthur 1972, Vermeij 1978), and latitudinal variation These results, largely from marine habitats, predict that
in predation, in particular, has been associated with as physical stress increases, predation pressure decreas-
differences in morphology in many organisms (Mayr es and intraspecific competition increases (i.e., the con-
sumer stress models of Menge and Olson [1990]). Work
1963, Bakus 1969, Vermeij and Veil 1978). Although
on vascular plants in terrestrial habitats has similarly
species interactions have been hypothesized to vary
biogeographically (e.g., Vermeij 1978, Dethier and suggested that the role of competition varies with stress
Duggins 1988), experimental tests of these hypotheses (Grime 1973, 1977, Tilman 1988) although there is
have been relatively rare (but see Jeanne [1979], Bert- much controversy surrounding this assertion (Grime
1979, Weldon and Saulson 1986, Wilson and Keddy
Manuscriptreceived 25 January1999 (originally submitted30 1986, Moloney 1990, Grace 1991, Goldberg and Barton
April 1998); revised 24 February 1999; accepted 25 February 1992). When physical stress becomes extreme, how-
1999. ever, competitive interactions may be replaced by fa-
a Present address: Stanford University, Department of Bi-
cilitative interactions (Bertness and Callaway 1994) if
ological Sciences, Hopkins Marine Station, Oceanview Bou-
levard, Pacific Grove, California 93950-3094 USA. the "competitor" modifies the physical habitat and al-
E-mail: gleonard@leland.stanford.edu leviates the stressful conditions (e. g., Bertness and
1015
1016 GEORGEH. LEONARD Ecology,Vol. 81, No. 4
Shumway 1993, Bertness and Hacker 1994). In these which to investigate spatial and temporal variation in
cases, a species may have higher fitness when in as- species interactions. Most intertidal organisms are
sociation with a "competitor" than when it is living known to be sensitive to thermal and desiccation stress-
alone. es, which can vary at both small and large spatial scales
While productive, this debate on the dichotomy be- (Lewis 1964, Wethey 1983, 1984). Most importantly,
tween competition and facilitation ignores the fact that these physical stresses can often be alleviated by the
species can influence each other in ways that do not presence of other species. For example, intertidal algal
strictly involve limiting resources. One organism can canopies often keep the substrate moist at low tide and
influence the recruitment, survival, growth, or repro- can thus reduce the physiological stress of the organ-
duction of another and these effects can vary in both isms underneath (Dayton 1971, Menge 1978, Under-
space and time. For example, subtidal algae can alter wood and Denley 1984 and references therein). Be-
the recruitment and growth of benthic invertebrates by cause of the large amount of work done in these com-
altering propagule delivery and food acquisition with- munities (see Lewis 1964, Paine 1994, Little and Kitch-
out actually competing for resources (Duggins et al. ing 1996), there is also an ecological context in which
1990, Eckman and Duggins 1991). to place experimental work done at broader scales. Ex-
The debate between competition and facilitation perimental manipulations to understand how species
should be broadened further to address the many pos- interactions vary at larger scales may also be a pow-
itive and negative ways in which organisms influence erful way to predict how species and communities will
each other's fitness. This is critical because the overall respond to global climate change in the future.
effect of one species on another may be positive, neg- In this study, I hypothesized that interactions be-
ative, or neutral depending on the magnitude and di- tween a large, canopy-forming intertidal alga (Asco-
rection of the individual effects (e. g., Underwood phyllum nodosum) and a sessile, filter-feeding inver-
1986). Demography offers an excellent framework in tebrate (Semibalanus balanoides) should vary at lati-
which to address these types of multiple effects tudinal spatial scales in New England, USA, because
(McPeek and Peckarsky 1998). Simultaneous variation of predictable differences in environmental stress (tem-
in recruitment, survival, and fecundity as a function of perature) and predation by the carnivorous whelk, Nu-
the presence or absence of a species can be entered cella lapillus. Seasonal temperature fluctuations in New
into a standard life table and the net effect evaluated England are some of the largest in the world (Menge
as the product of these individual effects. Many of the 1976) with summer air temperatures greater than in
previous studies of facilitation, however, have focused either Great Britain or the west coast of North America
only on survival (e. g., see review by Callaway [1995]) (Barnes 1958a). In addition, subtle but important dif-
and few have considered these multiple effects. This ferences in summer air temperature have been hypoth-
is especially true of interactions that vary as a function esized to influence the distribution of, and interactions
of life-history stage. To more fully understand the role among, intertidal organisms between sites to the north
of positive and negative interactions in natural com- and south of the Cape Cod peninsula in Massachusetts
munities requires a focus on these types of multiple, (Barnes and Barnes 1959, Wethey 1983). Observations
interactive effects (Greenlee and Callaway 1996, Cal- I made in 1994 at exposed intertidal habitats in Rhode
laway and Walker 1997). Island (south of Cape Cod) indicated that the upper
The debate must also move beyond documenting the distribution of the Semibalanus zone was generally co-
direct effect of one species on another and begin to incident with the top of the Ascophyllum zone but that
incorporate indirect effects. Although there are excep- at similar sites in Maine (north of Cape Cod) it ex-
tions, many of the previous studies have focused on tended beyond it (see Methods: Study sites and zona-
pairwise interactions (e. g., Bertness and Shumway tion patterns, below). The southern pattern is atypical
1993). This has unintentionally disregarded the large of most intertidal habitats where barnacles generally
number of species that comprise most biological com- persist above the algal zone in all but the most protected
munities and the numerous indirect interactions (both habitats (Stephenson and Stephenson 1948, Carefoot
positive and negative) that occur among them. Inter- 1977, Menge 1978, Ricketts et al. 1985; personal ob-
action modifications, where the non-trophic effects of servations). This observation is consistent with a re-
one species alters the interaction between two other duction in mortality from thermal stress north but not
species, may be common and important indirect inter- south of Cape Cod.
actions in many communities (Wootton 1993). It is Cape Cod is also a well-known faunal break (Gould
clear that a synthetic approach, which focuses on how 1840, cited in Allee 1923) that separates the Atlantic
spatial and temporal variability in the environment in- Boreal fauna to the north from the Atlantic Temperate
fluences both direct and indirect interactions, will help fauna to the south (Gosner 1978). Nucella lapillus is
ecologists better understand how species interactions the primary predator of barnacles in New England and
vary in natural habitats (Callaway and Walker 1997, also has a largely boreal distribution (Gosner 1978).
McPeek and Peckarsky 1998). These biogeographic patterns and the experimental
The marine intertidal may be an ideal system in findings of Menge (1976) in northern New England
April 2000 VARIATIONIN SPECIESINTERACTIONS 1017
suggest that predation pressure may be reduced south
of Cape Cod. In addition, by harboring whelks and
increasing their foraging efficiency (Menge 1978), the
algal canopy indirectly increases barnacle mortality at
those sites where whelks are present. This suggests that
the direct positive effects of the algal canopy on bar-
nacle survival (by alleviating temperature stress) may
be critical to the maintenance of barnacle populations
south of the Cape but that the negative, indirect effects
of the canopy on predators (an interaction modification
sensu Wootton 1993) may overwhelm any direct pos-
itive effects at northern sites.
To test these ideas, I experimentally manipulated al-
gal canopies at northern and southern sites and eval-
uated the consequences for both barnacles and their
predators. I quantified canopy effects at different bar-
nacle life-history stages and then integrated these com-
ponent effects in a demographic framework. I hypoth-
esized that the algal canopy would decrease fecundity
at all sites but that it would increase survival only at
the southern sites. If the positive effects of the canopy
were stronger than its negative effects, the association
would result in higher fitness (defined as the product FIG. 1. Map of New England(USA) showing the open-
of survival and fecundity) at the southern sites. I also coast study sites in Rhode Island (RI) and Maine. Cape Cod
took advantage of the differences in climatic conditions is a well-recognizedfaunalbarrier betweenthesetwo regions.
in 1995 and 1996 to determine whether this interaction
was associated with year-to-year variation in climate.
mean lower-low water (MLLW). Because the tidal
METHODS range differs between Rhode Island and Maine (1.4 m
vs. 3.5 m above MLLW, respectively), elevations were
Study sites and zonation patterns
expressed as the percentage of time that zones were
Two sites in Rhode Island and two in Maine (New exposed to aerial conditions within each region. Tidal
England, USA) were chosen to test this hypothesis (Fig. height data were obtained from TideGuide version 1.30
1). All sites were semi-exposed, intertidal habitats con- (Zihua Software, Pacific Grove, California, USA). The
sisting of gently sloping granite benches interspersed tops of the Ascophyllum nodosum and Semibalanus bal-
with large granitic boulders. Each site was oriented anoides zones were quantified by sampling 10 eleva-
approximately south-south east. Sites were protected tions per zone across approximately 20-40 m of each
from the largest ocean swells by either small offshore study site. Differences in exposure time of each zone
islands or large seaward rock benches. The two south- between northern and southern sites were analyzed us-
ern sites, Sakonnet Point (41?27'14" N, 71?11'35" W) ing nested ANOVA with site nested within region.
and Middletown (41?28'31" N, 71014'30" W), were lo-
cated on the eastern and western sides, respectively, of Barnacle demographics and variable species
interactions
Narragansett Bay, Rhode Island. The two northern
sites, Chamberlain (43?53'7" N, 69?28'29" W) and Pe- Positive and negative interactions between the algal
maquid (43?50'8" N, 69?30'29" W), were located along canopy and understory barnacles were examined at
the eastern shore of Pemaquid Neck. Although exper- these four sites for two years, from January 1995
iments in Maine were not conducted exactly where oth- through January 1997. In November 1994 I created
ers have worked (e. g., Menge 1976, 1978), these sites circular clearings in the Ascophyllum canopy (radius
have been the subject of considerable past research. -1.0 m) at its upper border in the high zone of all sites
Observations made during May-September 1994 (n = 8 clearing/site). Canopy plots (n = 8 plots/site)
suggested that the acorn barnacle, Semibalanus bal- were unmanipulated areas that, at low tide, had a 100%
anoides, extended above the Ascophyllum nodosum cover of A. nodosum. The absolute tidal height of all
zone at northern sites, but at southern sites it was pres- plots was set to keep the percentage of time exposed
ent only under the algal canopy. To determine if this to aerial conditions (-60%) constant between regions.
was due to a vertical extension of the barnacle zone Within each canopy and cleared plot, I set up two per-
rather than a contraction of the algal zone, distribution manent quadrats (25 X 25 cm) on the rock substrate
patterns were quantified at all sites using standard sur- marked at their corners with galvanized bolts. Quadrats
veying equipment and then standardizing elevations to were nestled between the A. nodosum holdfasts (in can-
1018 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
opy plots) and initially had an intermediate cover kept on ice in the field, and then frozen at -10?C in
(-50%) of barnacles. the laboratory. Within 4 mo barnacles were thawed in
Each spring one of the two quadrats in each plot was seawater and then each individual was dissected into
scraped to bare rock with a putty knife (without dis- shell, somatic, and reproductive (i.e., larvae) tissue.
turbing A. nodosum holdfasts or canopy cover) and Tissue components were dried at 35?C for 24 h before
used to evaluate the effect of the canopy on barnacle weighing on a microbalance (precision = +0.01 mg).
recruitment, survival, growth, and fecundity. Because
adult barnacles may generally be less susceptible to Statistical analysis
physical stress than new recruits (Foster 1969, 1971), Differences in recruit and adult barnacle survival as
I used the unmanipulated quadrat to evaluate the in- a function of the canopy, region, and site were tested
fluence of the canopy on the "adult" population (de- using nested analysis of covariance (ANCOVA) on sur-
fined as all individuals >1 yr old, sensu Wethey 1984). viving barnacle density in October. Site was considered
In addition, scraped quadrats (n = 8 quadrats/site) were a random factor (nested within region) and canopy and
established in the zone directly above the Ascophyllum region were considered fixed factors. Density the pre-
canopy at all sites to evaluate regional differences in vious spring was the covariate. Because space is often
barnacle recruitment, survival, growth, and reproduc- the limiting resource for barnacles, mortality is fre-
tion above the zone influenced by the algal canopy. quently density dependent (i.e., greater mortality on
In 1996 I incorporated a predation treatment to at- high densities than low densities of recruits; Connell
tempt to partition the mortality of barnacle recruits into 1985). ANCOVA statistically factors out any density-
the direct effect of the algal canopy on reduced thermal dependent effects on survival by removing the influ-
stress and the indirect effect of the canopy on predation ence of initial density on final density. This approach,
intensity (Menge 1978). At all sites I attached 20 X however, has two assumptions that were verified before
20 X 5 cm galvanized wire mesh cages and cage con- the analysis proceeded (Underwood 1997). First, the
trols to cleared quadrats in all canopy and cleared plots. relationship between the dependent variable and the
Unfortunately, this treatment was unsuccessful at ma- covariate must not vary among treatments (homoge-
nipulating predator abundance (unpublished data) be- neity of slopes). Second, the range of the covariate must
cause cages were frequently damaged by waves and be broadly similar among treatments.
corrosion of the wire mesh often compromised the tight Differences in the demographic parameters of bar-
fit of the cages to the substrate. Accordingly, I analyzed nacle recruits and adults as a function of the algal can-
only the results from the uncaged quadrats. Densities opy, region, and site were analyzed with nested analysis
of whelks were measured three times during each sum- of variance for both 1995 and 1996. As before, site
mer at all sites in these uncaged quadrats and average was considered a random, nested factor. Differences in
predator abundance was analyzed using analysis of var- growth were evaluated using total soft tissue (somatic
iance. Differences in predator abundance were related + reproductive tissue) as the dependent variable. Sim-
qualitatively to patterns of barnacle mortality among ilar analyses were performed on fecundity (total larval
regions and years. mass) and reproductive fitness potential. Reproductive
At the end of the settlement season (Rhode Island fitness potential was defined in the demographic sense
end of March, Maine = beginning of May), barnacle as "l,mx" (an individual's probability of survival mul-
recruitment was measured in the field in the cleared tiplied by its reproductive output). Because of the broad
quadrats using sampling grids. Barnacles that survived dispersal capability and open population structure of
through the summer were sampled using similar tech- Semibalanus, I could evaluate neither the complete life
niques in late October in both 1995 and 1996. Change table for this species (but see Eckman [1996] for a
in density of the "adults" was determined from pho- conceptual and empirical approach) nor the contribu-
tographs of the unmanipulated quadrats at the begin- tion of individual effects to population growth rate (i.e.,
ning (March 1995), middle (August 1995), and end a sensitivity analysis of X; McPeek and Peckarsky
(October 1996) of the experiment. 1998).
In both years, growth and reproductive output of In addition, there were often large differences in
recruits and adults were quantified from three individ- these parameters among sites (see Table 1). These dif-
uals haphazardly selected from each of the cleared and ferences could have been due to a number of uncon-
unmanipulated quadrats (total sample size: recruits = trolled factors including (1) differences in larval supply
72 individuals per site per year, adults = 48 individuals (Raimondi 1990), (2) phytoplankton composition
per site per year). In New England, barnacles reproduce (Barnes and Barnes 1959), (3) increased egg size at
in early fall, fertilized eggs mature over the winter, and northern sites (Crisp 1959), (4) temperature-dependent
larvae are released only once a year in early spring or flow-mediated plasticity in growth rates (Southward
(Barnes 1958b). Individuals were sampled in mid-win- 1955, Sanford et al. 1994), or (5) genetic differentiation
ter (December-January) after larvae had fully matured among sites or regions (Crisp 1964). My primary in-
but before they had been released. Barnacles were care- terest was not in the absolute magnitude of these pa-
fully removed from the rock using surgical scalpels, rameters but rather in their variation with the algal
April 2000 VARIATIONIN SPECIESINTERACTIONS 1019
canopy between regions. I therefore standardized the canopy treatment per site). In 1996, sample size was
data for differences among sites before statistical anal- increased to 8 thermometers/site but temperatures were
ysis by converting each datum to reflect its deviation collected only in cleared plots. This design increased
(either larger or smaller) from its site mean. Analyses the power of detecting differences in thermal charac-
were then performed on these deviations (In (x + 1) teristics among sites. Both wave exposure and thermal
transformed to meet the assumption of normality and data were analyzed using repeated-measures, nested
variance heteroscedascity). Positive or negative species analysis of variance with region, site and time as fac-
interactions were defined from the direction of a sta- tors. As before, site was considered a random factor
tistically significant "canopy" effect in the analysis of and was nested within region.
variance. In addition, regional differences in the mag- I used weather-station data to supplement these di-
nitude and direction of these species interactions were rect measures of temperature. Overall differences in
identified by a significant "canopy X region" effect in climatic conditions between Maine and Rhode Island
the same analyses. Because this was a nested design, were quantified using data acquired from the Northeast
the error term for the F ratio for both canopy and can- Regional Climate Center. Maximum daily air temper-
opy X region effects was the canopy X site (region) ature had been recorded for 1995 and 1996 at land-
term. based sites in Newport, Rhode Island (41?30' N, 71?21'
W), and Boothbay, Maine (43?52' N, 69?35' W), both
Abiotic conditions in close proximity to the respective study sites. Daily
Physical factors were recorded during the two years differences in temperature from April to September be-
of the study to relate the biotic results to spatial and tween regions and years were used to quantify spatial
temporal variation in abiotic conditions. I tested for and temporal differences in climatic conditions. Pat-
variation in wave exposure among sites because bar- terns in thermal regime, wave exposure, and predator
nacle mortality is known to be reduced at sites of high abundance were compared qualitatively to the exper-
wave splash (Lewis 1964). This was done by quanti- imental results on variation in species interactions.
fying wave exposure at all sites approximately monthly
from December 1995 through August 1996 using RESULTS
spring-loaded dynamometers (Denny 1983, Bell and Zonation patterns
Denny 1994). These instruments measure the maximum
force imposed by breaking waves over the sampling The upper limit of the Ascophyllum canopy on the
interval. Unfortunately, these instruments do not mea- shore was similar between study sites in Maine and
sure the average conditions that exist at a site and it is Rhode Island (Fig. 2; F1,2 = 0.90, P = 0.442) but the
these average conditions that may be most important upper limit of the barnacle zone differed between re-
to organisms that suffer daily emersion. Estimating gions (F, 2 = 19.85, P = 0.047). At southern sites, the
wave exposure by eye, however, can be misleading (see upper limit of barnacle zone was coincident with the
Bell and Denny 1994) and dynamometers, although not upper limit of the algal canopy, but at northern sites it
flawless, are probably the best technique currently extended beyond the algal canopy (Fig. 2, daily emer-
available to estimate wave splash. sion = 62% vs. 75%, respectively).
Dynamometers (n = 5 instruments/site) were bolted Variable species interactions: barnacle recruits
to the rock in the center of circular plots (1-m radius)
cleared of macroalgae in the same area as the canopy The algal canopy strongly influenced barnacle re-
manipulations. Measurements of spring extension (in cruitment, survival, growth, and fecundity during the
millimeters) were converted to maximum force (in two years of this study and this had large effects on
newtons) using the equations in Bell and Denny (1994). their reproductive fitness potential. Recruitment at the
By late May 1996, dynamometers at the Pemaquid site four study sites ranged from 4.06-24.41 individuals/
were being continuously vandalized because of heavy cm2 in 1995 and 1996 (Table 1). The canopy decreased
foot traffic. I therefore removed the wave meters from barnacle recruitment at all sites largely because of me-
this site and no data were collected there throughout chanical abrasion of the substrate (i.e., "algal whip-
the rest of the summer. lash"; Dayton 1971, Menge 1976, Leonard 1999a) and
To test the hypothesis that regional variation in spe- this effect did not vary between regions (Leonard
cies interactions was associated with differences in 1999b).
thermal stress, I quantified the overall thermal regime In contrast, in both 1995 and 1996 the effect of the
at all sites during the summers of 1995 and 1996. In canopy on survival of barnacle recruits varied signif-
1995, maximum rock-surface temperatures were col- icantly between regions (Table 2). The canopy in-
lected approximately every 2 wk from June through creased survival at the southern sites but not at the
October using min/max thermometers (Taylor Scien- northern sites (Fig. 3). At the southern sites, survival
tific, model number 5458). Thermometers were placed was always highest under the canopy, intermediate
under the Ascophyllum canopy and in cleared plots in where the canopy had been removed, and lowest above
the high zone of all sites (n = 2 thermometers per the algal zone (Fig. 3). At northern sites, survival was
1020 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
TABLE 1. Recruitment, survival, growth, and fecundity of Semibalanus balanoides recruits and adults during 1995 and 1996
at four rocky intertidal sites in New England.
Recruitment: Survival?
Stage Year Sitet (no./cm2) n (%) n
Recruits 1995 Sakonnet, RI 14.38 + 2.01 22 6.3 + 2.3 22
Middleton, RI 14.77 + 2.60 24 9.0 + 3.3 22
Chamberlain, ME 9.36 + 1.03 23 30.6 + 3.0 23
Pemaquid, ME 6.22 + 0.96 24 14.1 ? 3.5 22
1996 Sakonnet, RI 24.41 + 2.12 24 10.3 + 2.4 24
Middleton, RI 17.36 + 1.95 24 12.6 + 6.2 24
Chamberlain, ME 7.65 + 1.05 24 46.1 + 6.0 24
Pemaquid, ME 4.06 + 0.77 24 38.0 + 6.4 21
Adults 1995 Sakonnet, RI NA NA -60.01 + 28.05 14
Middleton, RI NA NA -57.00 + 26.52 16
Chamberlain, ME NA NA -45.91 + 20.72 15
Pemaquid, ME NA NA 12.15 + 6.27 16
1996 Sakonnet, RI NA NA 66.90 + 15.56 16
Middleton, RI NA NA 18.50 ? 21.05 16
Chamberlain, ME NA NA 3.90 + 17.41 15
Pemaquid, ME NA NA 33.92 + 18.36 15
Notes: For each parameter, both the site means + 1 SE and the sample size (n) are given. These site means were used to
generate the proportional deviations used in the ANOVAs to test for variable species interactions between regions (see
Methods: Statistical analysis for further clarification). Reproductive fitness potential (survival X fecundity) was also calculated
from these data.
t RI = Rhode Island, USA; ME = Maine, USA.
:. Recruitment was not applicable (NA) for adult barnacles because recruitment refers only to input from the planktonic
larval pool.
? For adult barnacles, "survival" refers to the net change in adults (measured as no./100 cm2) between time periods. This
overall measure is the sum of an increase due to recruits that survive beyond the first year (and hence, become reclassified
as adults) and a decrease due to mortality of established adults.
generally similar under the canopy and in cleared plots cleared plots at southern sites (Fig. 3) could not there-
but was always highest above the algal zone (Fig. 3). fore be attributed to predation in these treatments.
Most importantly, the strength of the positive effects While the canopy had regionally variable effects on
of the canopy at southern sites varied among years (Fig. recruit survival, it had universally negative effects on
3: compare 1995 to 1996 for southern sites). In 1995 growth and reproduction. The canopy inhibited recruit
at southern sites there was very low survival in cleared growth in both 1995 and 1996 (Fig. 5; 1995, F,2
plots (in fact, 1 SE of the mean overlaps 0 individuals/ 20.23, P = 0.046; 1996, F,2 = 249.20, P = 0.004),
cm2) and 100% mortality above the algal zone. How- but this effect did not vary between regions in either
ever, recruit mortality at the southern sites in 1996 was year (1995, Fl,2 = 2.06, P = 0.287; 1996, F1,2 = 0.04,
less severe, as evidenced by the considerable survival P = 0.855). Similarly, the canopy decreased recruit
in the cleared plots and especially that above the algal fecundity in both years (Fig. 5; 1995, F ,2 = 31.42, P
zone. = 0.030; 1996, Fl,2 = 30.22, P = 0.032) and this did
Although the exclusion cages did not effectively con- not vary between regions (1995, F,2 = 2.36, P = 0.264;
trol predator densities (see Methods: Barnacle demo- 1996, F1,2 = 0.03, P = 0.873).
graphics ..., above), barnacle mortality patterns in When integrated, these data indicate that the net ef-
uncaged quadrats at northern sites were related to dif- fect of the algal canopy on recruit reproductive fitness
ferences in predator densities in cleared, canopy, and potential varied between northern and southern sites
above-canopy plots. During both years, Nucella lapil- and between years. In 1995 the canopy increased re-
lus at northern sites were in greatest abundance under productive fitness potential at southern sites but de-
the algal canopy, intermediate in cleared plots and ab- creased it at northern sites (Fig. 5, F,]2 = 4.17, P
sent above the algal canopy (Fig. 4; 1995, F,27 = 17.88, 0.046). In 1996 the canopy decreased reproductive fit-
P < 0.001; 1996, F,28 = 4.551, P = 0.038). The low ness potential at all sites (Fig. 5, F,2 = 7.69, P = 0.008)
recruit survival in canopy and cleared plots was as- and this effect did not vary between regions (F1,2
sociated with the presence of predators while the high 0.04, P 0.868).
survival above the algal canopy was associated with Above the algal canopy, recruit growth, fecundity,
the absence of predators (compare Figs. 3 and 4). In and reproductive fitness potential were generally great-
contrast, although N. lapillus were present at the south- er at northern sites than at southern sites (Fig. 5) al-
ern sites (unpublished data) they were never observed though these relationships were clouded by high var-
in the high zone during the two years when this ex- iability at southern sites. In 1995 no recruits survived
periment was done (Fig. 4). Lower survival in the in this zone at southern sites and fitness was therefore
April 2000 VARIATION IN SPECIES INTERACTIONS 1021
TABLE 1. Extended.
H A. nodosum
Growth Fecundity
L]S. balanoides
80
(mg/ind.) n (mg/ind.) n
0.73 + 0.12 15 0.16 + 0.09 15 60
6
0.60 + 0.07 14 0.09 + 0.04 14 o
3.13 + 0.49 24 1.54 + 0.33 24
2.54 + 0.45 22 1.21 + 0.23 22 |S 40
1.22 + 0.13 23 0.10 + 0.04 23
0.99 + 0.14 20 0.06 + 0.05 20 20
2.23 + 0.37 24 1.17 + 0.22 24
1.70 + 0.24 22 0.74 + 0.13 22
1.58 + 0.16 16 0.75 + 0.10 16 0
O
1.62 + 0.16 15 0.94 + 0.13 15 Sak. Midd. Cham. Pem.
6.69 + 1.05 15 3.86 ? 0.80 15
2.86 ? 0.62 13 1.59 + 0.41 13
2.57 + 0.37 16 1.20 + 0.26 16 South North
2.28 + 0.40 14 1.19 + 0.31 14 FIG. 2. Zonation patterns at replicate study sites north
1.05 + 0.31 16 0.54 + 0.20 16 and south of Cape Cod. Data are the percentages of days
2.86 + 0.62 13 1.59 + 0.41 13
(mean ? 1 SE) that the top of the Ascophyllum nodosum and
Semibalanus balanoides zones are exposed to aerial condi-
tions at each site. Sak. = Sakonnet Point, Rhode Island; Midd.
= Middleton, Rhode Island; Cham. = Chamberlain, Maine;
zero. In 1996 neither recruit growth, fecundity, nor and Pem. = Pemaquid Lighthouse, Maine.
fitness were significantly different between northern
and southern sites above the algal canopy (all F, 2 <
7.39, P > 0.113). This was largely due to the high was statistically insignificant (Table 3). This was due
variability in fecundity and reproductive fitness poten- to the high variation in canopy effects among sites
tial at southern sites but not at northern sites (Fig. 5). within regions (Table 3). This year-to-year variation in
canopy effects on adult survival at southern sites was
Variable species interactions: adult barnacles
analogous to that seen for barnacle recruits.
The strength of negative and positive interactions As with the recruits, adult growth and fecundity were
between the canopy and the underlying adult barnacles reduced in the presence of the canopy during both years
also differed between regions and years. During 1995, (Fig. 7; Growth: 1995, F1,2 = 36.69, P = 0.024; 1996,
=
survival was elevated under the canopy at southern Fl,2 = 102.59, P = 0.010; Fecundity: 1995, F,2
sites but reduced by the canopy at northern sites (Fig. 35.08, P = 0.027; 1996, = 177.68, P = 0.006).
F,2
6, Table 3). The overall pattern of survival in 1996 was Similarly, these effects did not vary between region
similar to that in 1995 (Fig. 6) but its magnitude was (Growth: 1995, F,2 = 4.69, P = 0.156; 1996, F,2 =
lower and the resulting canopy X region interaction 6.77, P = 0.122; Fecundity: 1995, F,2 = 8.02, P =
TABLE2. Survival of barnacle recruits as a function of region and site during 1995 and 1996, analyzed using nested analysis
of covariance.
Source of variation df MS F Denominator MSt P
1995
Region 1 3.51 0.42 a 0.583
Site(Region) 2 8.32 14.02 b <0.001
Canopy 1 14.22 38.61 c 0.025
Canopy X Region 1 18.47 50.15 c 0.019
Canopy X Site(Region) 2 0.37 0.62 b 0.542
Recruit density 1 3.49 5.88 b 0.019
Residual 52 0.59
1996
Region 1 0.22 1.20 a 0.338
Site(Region) 2 0.18 1.12 b 0.336
Canopy 1 1.27 102.25 c 0.010
Canopy X Region 1 4.02 324.16 c 0.003
Canopy X Site(Region) 2 0.01 0.08 b 0.928
Recruit density 1 2.02 12.18 b 0.001
Residual 55 0.17
Notes: Recruit density [ln(x + 1)-transformed] was the covariate, and final density [ln(n + 1)-transformed] was the dependent
variable. The statistical assumption of homogeneity of slopes was satisfied in both years: 1995, F,45 = 1.81, P = 0.109;
1996, F748 = 1.06, P = 0.406. In addition, the range of the covariate was broadly similar among the study sites.
t Denominator MS for the F ratios are: a = Site(Region), b = Residual, c = Canopy X Site(Region).
1022 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
* * Canopy
- Canopy
Cleared I_ Cleared
mi Above Canopy Ei Above
Canopy
3
1995 199
2.0-
2 1.5-
4-
I
1
E 0.5-
NA .
0
o O 0.0-
1996 2.5-
A
4
2.0-
I
3- is-
C) 1.0-
2- a' 0.5-
1- 0.0- NA
~ 1
0- q-
North South
QB 2-
2-
FIG. 3. Survival of barnacles that recruited during the io
spring of each study year at the end of October in 1995 and 0 "
1996 in 25 X 25 cm2 quadrats located under the canopy, in
areas cleared of the canopy or above the zone influenced by
1-
'J l I - rll F
f
the canopy. Data are adjusted densities (mean ? 1 SE) from 0-
the analysis of covariance using recruitment density as the North South North South
covariate and final density as the dependent variable. See
Table 2 for statistical analyses. FIG. 5. Demographic parameters of barnacle recruits dur-
ing 1995 and 1996. Parameters measured were growth, fe-
cundity, and reproductive fitness potential (survival X fe-
cundity). Values are the deviation (mean + 1 SE) of each pa-
rameter from its site mean. NA indicates that no individuals
3.6- 1995 survived and that growth and fecundity could therefore not
be measured.
2.4-
0.105) except in 1996 when fecundity was reduced
more under the canopy in northern sites than in south-
1.2- ern sites (Fl,2 = 31.05, P = 0.031).
trn Like those for recruits, these data for adults show
0 0 0
O
S o.o that canopy effects on adult reproductive fitness varied
o0.0
6 North South between regions and between years. In 1995 adult re-
0
1996 productive fitness potential was facilitated by the can-
. 3.6 opy at southern sites but was reduced by the canopy
* Canopy at northern sites (Fig. 7, F ,2 = 23.15, P = 0.041). In
a 2.4 Cleared 1996 the canopy had a consistently negative effect on
[-
Above Canopy reproductive fitness potential at both northern and
; southern sites (Fig. 7, F, 2 = 24.87, P = 0.038). These
1.2 results for adults were strikingly similar to those ob-
tained for barnacle recruits.
0 0 O0
0.0
North South Abiotic conditions
FIG.4. Predator densities as a function of region and treat- Measurements of physical factors suggested that
ment during 1995 and 1996. Predators (Nucella lapillus) were thermal regime but not wave exposure differed between
sampled in the permanent quadrats three times during each Maine and Rhode Island. There was no evidence that
summer. Densities were averaged over the summer. Values northern and southern sites were of different wave ex-
in the figures are means ? 1 SE. Although present in the low
zone (data not shown), N. lapillus were never found in the posure (Fig. 8) as maximum wave force did not vary
high zone at southern sites or above the canopy at northern between regions from December 1995 through May
and southernsites. 1996 (Table 4). From May through September there
April 2000 VARIATIONIN SPECIESINTERACTIONS 1023
* Canopy
D- Cleared
FIG.6. Survivalof adultbarnacles(non-re- 1995
E 80-
cruits)during1995 and 1996. Valuesarethe net
change (mean + 1 SE) in the numberof adults/
100 cm2 from November 1994 to November
1995 (1995) and from November 1995 to No- 0 0-
vember 1996 (1996). See Table 3 for statistical
analyses.
V -160
Z North South North South
was also no overall difference among regions in max- Like 1995, temperatures also varied significantly be-
imum wave force, but exposure did vary among regions tween regions over time (Table 5). In early summer
over time during this latter half of the summer (Fig. 8, there was little difference in rock temperature between
Table 4). When tropical storm Daniel impacted New regions, but by the end of July temperatures at southern
England in July 1996, there were large waves at Cham- sites were consistently several degrees higher than
berlain (and likely Pemaquid, although not measured those at sites in Maine (Fig. 9).
there) but not at Middletown or Sakonnet (personal Land-based weather-station data corroborated these
observation). Other than this single time period, wave direct measures of temperature and indicated that
exposure differed very little between any of the study Rhode Island was, on average, several degrees warmer
sites. than Maine (Fig. 10). This was true in 1995 and 1996
In contrast, there were subtle but potentially biolog- although the magnitude of this difference was smaller
ically important differences in thermal regime between in 1996 (i.e., 1.69? vs. 2.40?C, Table 6). These data
northern and southern sites (Fig. 9). This was evi- also revealed that 1995 was a warmer summer overall,
denced by both rock surface temperatures (Fig. 9) and especially in Rhode Island (Fig. 10). This finding is in
weather station data (Fig. 10). In summer 1995, overall agreement with other meteorological records that show
rock-surface temperatures did not differ between re- 1995 was the hottest summer on record since 1900
gions but did vary between regions over time (Table (Easterling et al. 1997).
5). The algal canopy reduced rock surface temperatures
DISCUSSION
by -6.6?C at both northern and southern sites (Fig. 9,
Table 5). Contrary to expectations, however, temper- My results contribute to the continuing debate on the
atures in the open plots of the northern sites in early influence of site "quality" on species interactions and
summer were higher than those of the southern sites suggest that the intensity and direction of interactions
(Fig. 9). By the beginning of August this pattern had can change with physical stress and predation intensity
reversed and southern sites were slightly but consis- at large spatial scales. In this study, subtle differences
tently warmer than northern sites until October. in temperature north and south of Cape Cod and re-
In summer 1996, rock temperatures were nearly sig- gional differences in predator abundance were asso-
nificantly different between regions (i. e., P = 0.094; ciated with differences in negative vs. positive inter-
Table 4). Overall, rock temperatures at southern sites actions between the algal canopy and the underlying
were 1.5?C warmer than at the northern sites (Fig. 9). barnacles. In addition, year-to-year variation in the net
TABLE 3. Results of nested ANOVAs on adult barnacle survival (measured as yearly changes in adult barnacle density (no./
100 cm2) during 1995 and 1996.
1995 1996
Denom- Denom-
Source of variation df MS F inator MS P MS F inator MS P
Region 1 24 644.8 1.73 a 0.320 7517.3 0.57 a 0.528
Site(Region) 2 14 284.7 3.96 b 0.025 13 110.3 2.99 b 0.059
Canopy 1 49 104.5 11.69 c 0.076 4290.1 0.28 c 0.650
Canopy x Region 1 165 730.5 39.45 c 0.024 26 164.3 1.71 c 0.322
Canopy x Site(Region) 2 4200.7 1.16 b 0.320 15 340.4 3.50 b 0.037
Residual 54 3611.9 4386.9
Notes: Analysis of covariancewas not used (as in Table 2) because the covariate(initial adult density) was not broadly
similar among the study sites. The denominator for the F ratios were: a = Site(Region), b = Residual, c = Canopy X
MS
Site(Region).
1024 GEORGEH. LEONARD Ecology, Vol. 81, No. 4
* Canopy 120-
-0- South
[7 Cleared
Z -- North
2.0- 1995 1996
80-
0
- 1.5-
+ -t- -- ?- B
o 1.0-
40-
0.5- ed
cX
0-
i I a
I I I I I I I I I
2.0- - 1I Dec Jan Feb Mar Apr May Jun Jul Aug Sep
4 1.5- 1996
= 1.0-
-+
- FIG. 8. Wave exposure at northern and southern sites col-
0 lected from December 1995 through October 1996 using
L; 0.5- spring-loaded dynamometers. Data are maximum force
I (means ? 1 SE) estimated from spring displacement measured
0- at approximately monthly intervals.
- 2.0- - I
; 1.5- cundity) were largely consistent with work done in in-
tertidal systems in the past. The canopy inhibited bar-
1.0-
nacle recruitment at all study sites in both years via
'
0.5- mechanical "whiplashing" of the substrate (Leonard
-j ----r_B-_ 1999a). This effect has often been found on wave-
North South beaten shores (Dayton 1971, Menge 1976, Grant 1977,
FIG. 7. Demographic parameters fo r adultbarnaclesdur- Hawkins 1983) and appears to be a common feature of
ing 1995 and 1996. Data and presenta tion are analogousto large, canopy-forming algal assemblages.
those in Fig. 5. Similarly, barnacle growth and fecundity were sig-
nificantly lower under algal canopies than in cleared
plots. This common observation (Lewis 1964, Wethey
effect of the algal canopy on reprc )ductive fitness po- 1985) was similar at both northern and southern sites
tential occurred because of variati{ in barnacle sur- and did not vary between years. Interestingly, growth
on
vival rather than reproductive oultput. These results rates (and fecundity) were not enhanced under algal
suggest that the local persistence { species in some canopies compared to open plots at southern sites,
of
intertidal habitats may be dependemnt the presence where higher temperatures were hypothesized to im-
on
of other species that can modify the increasingly severe pose severe physiological stress on metabolic and re-
environment predicted under globzal warming scenar- productive processes (cf. Barnes and Barnes 1959, Cos-
ios. sins and Bowler 1987). Artificially shading barnacles
has been shown to alleviate these stresses and result
Constituent effects of the algal calnopy on barnacles in higher growth rates (Bertness 1989, Bertness and
The individual effects of the Asc:ophyllum nodosum Gaines 1993). However, the physical structure provided
canopy on barnacles (i.e., recruitmc growth, and fe- by algal canopies also alters patterns of water flow
ent,
TABLE4. Results of repeated-measures nested ANOVAs on wave-exposure data.
December 1995-May 1996 June-September 1996
Denom- Denom-
inator inator
Source of variation df MS F MS P df MS F MS P
Region 1 58.83 1.40 a 0.359 1 204.80 0.66 a 0.566
Site(Region) 2 42.14 0.15 b 0.861 1 310.08 1.81 b 0.203
Instrument(Site) 16 278.67 12 171.36
Time 3 6.66
121.13 c 0.001 5 2267.87
38.24 c <0.001
Time X Region 3 1.23
224.98 d 0.379 5 984.25
8.57 d 0.017
Time X Site(Region) 6 10.10
183.57 c <0.001 5 1.65
114.90c 0.218
Time x Instrument(Site) 48 18.18 60 69.77
Notes: Analyses were done separately for the period December 1995-May 1996 and June 1996-September 1996 because
instruments were vandalized at Pemaquid (Maine, USA) during the summer of 1996, and data could not be collected there.
The dependent variable was maximum force (in newtons) calculated from the spring-loaded dynamometers. The denominator
MS for the F ratios were: a = Site(Region), b = Instrument(Site), c = Time x Instrument(Site), d = Time X Site(Region).
April 2000 VARIATION IN SPECIES INTERACTIONS 1025
--- South:Cleared to feed on Semibalanus balanoides (Menge 1983). Any
- South: Canopy direct reduction in mortality under the canopy because
-D- North:Cleared of lower thermal stress at northern sites was apparently
--* North:Canopy offset by an indirect increase in mortality from pred-
ators mediated by the canopy (Menge 1978, Minchin-
40-1995
ton and Scheibling 1993).
Counter to my original hypothesis, results for the
35 -
adults were similar to those of the recruits for both
35-c,,
30-v years. Adults, like recruits, were sensitive to the ex-
perimental removal of the canopy and to the ensuing
25 - changes in environmental conditions and predator
-L
abundance. The similarity in response of recruits and
o
20-
adults to canopy manipulation suggests recruits in these
a
habitats are unlikely to escape thermal conditions over
15 time by growing in size (Foster 1971).
E Abiotic conditions associated with variation in
H 1996 barnacle survival
40-
Variation in survival rates was not associated with
E differences in wave exposure (Fig. 8) but may have
35-
ct
30- 8
1995: N vs. S
25-
20- ,n iaA
INA
Jun Jul Aug Sep Oct
V' ri
I I I I I
FIG. 9. Thermal characteristics at northern and southern
sites collected during the summers of 1995 and 1996 with
min/max thermometers. Values are the maximum temperature 1996: N vs. S
(mean + 1 SE) recorded during each preceding interval. In
1995, data were collected in cleared plots and under the can- a)
opy (n = 2 plots per treatment per site). In 1996, data were
collected less frequently and only in cleared plots but with 0
... tA
higher replication within sites (n = 8 plots per site). H
~
-8 ,*~?wrN vv IVV
E
(Eckman et al. 1989) and can modify feeding behavior -8
I I I I I
by interfering with deployment of the feeding structure
(Palmer et al. 1982). In my study, positive effects of CI- 8
the algal canopy on organism growth (through allevi- S: 1995 vs. 1996
ation of physiological stress) were apparently out-
weighed by negative effects on food acquisition.
In contrast to these negative effects, the algal canopy 0
had a large positive impact on organism survival, but
this was evident only at the southern sites. In both
years, barnacle survival in Rhode Island was highest
-8
,A^AI
under the canopy, intermediate in cleared plots, and Apr May Jun
Apr May Jun Jul Aug Sep
Jul Aug Sep
lowest in the zone above the algal canopy. Gastropod
predators were rare at these sites and mortality rates FIG. 10. Temperature anomalies (deviations in ?C) be-
were consistent with those `expected from variation in tween weather stations located near the open coast in the north
thermal stress alone. This contrasts sharply with north- (Maine) and south (Rhode Island). Data are 7-d running av-
erages of the differences in daily maximum air temperature
ern sites, where survival was generally similar under collected from 1 April through 1 October. Top panel: Dif-
the canopy and in cleared plots but was always highest ference between northern and southern sites during 1995.
above the algal zone. The abundance of the carnivorous Middle panel: Similar comparison for 1996. Bottom panel:
Difference between 1995 and 1996 at the southern sites. Neg-
gastropod Nucella lapillus at these northern sites was ative differences (in the middle and top panels) indicate that
high under the canopy, intermediate in cleared plots, the north was cooler than the south. Positive differences (in
and low at higher tidal heights. These predators often the bottom panel) indicate that 1995 was warmer than 1996.
seek refuge under the algal canopy and are well known Summary statistics are shown in Table 6.
1026 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
TABLE 5. Results of repeated-measures nested ANOVAs on substrate temperature (?C) as a function of region, site, canopy,
and time.
Denominator
Source of variation df MS F MS P
Summer 1995
Region 1 0.48 0.01 a 0.935
Site(Region) 2 56.50 2.76 b 0.123
Canopy 1 1935.55 324.30 c 0.003
Canopy x Region 1 0.54 0.09 c 0.793
Canopy X Site(Region) 2 5.97 0.29 b 0.755
Thermometer(Canopy x Site) 8 20.47
Time 10 92.64 20.47 d <0.001
Time X Region 10 60.50 11.05 e <0.001
Time x Site(Region) 20 5.47 1.21 d 0.326
Time x Canopy 10 6.27 1.34 f 0.277
Time X Canopy x Region 10 3.28 0.70 g 0.715
Time x Canopy X Site(Region) 20 4.69 1.04 d 0.430
Time X Thermometer(Canopy X Site) 80 4.53
Summer 1996
Region 1 119.31 9.12 a 0.094
Site(Region) 2 13.09 0.58 h 0.566
Thermometer(Site) 28 22.52 <0.001
Time 6 69.36 23.22 i 0.019
Time X Region 6 43.21 4.02 e 0.001
Time x Site(Region) 12 10.76 3.60 i
Time X Thermometer(Site) 168 2.99
Notes: The dependent variable was maximum temperature of the rock substrate over 2-wk periods during the summers of
1995 and 1996. In 1996 the canopy treatment was eliminated, and within-site replication was increased from 2 to 8 ther-
mometers/site. Denominator MS for the F ratio were: a = Site(Region), b = Thermometer(Canopy x Site), c = Canopy X
Site(Region), d = Time X Thermometer(Canopy X Site), e = Time x Site(Region), f = Canopy x Site(Region), g = Time
x Canopy X Site(Region), h = Thermometer(Site), and i = Time x Thermometer(Site).
been related to subtle differences in climatic conditions air and rock temperatures at southern sites commonly
in 1995 and 1996 in the absence of predators. Weather- approached the thermal limits of Semibalanus balan-
station data from Rhode Island and Maine indicated oides (i.e., 40?C; Southward 1958, Foster 1969, 1971),
that 1995 was a warmer summer than 1996 (Fig. 10), further suggesting a causal link between yearly vari-
in agreement with global climate data showing 1995 ation in thermal stress and variation in species inter-
was the warmest year since 1900 (Easterling et al. actions.
1997). The correspondence between my experimental These conclusions, however, should be taken cau-
results and these yearly climatic conditions suggests tiously because the direct measures of rock temperature
that these species interactions may be sensitive to sub- did not closely match the indirect measures of local
tle overall differences in temperature in regions where conditions made using weather-station data. For ex-
the influence of predators is minimal. Differences in ample, weather-station data indicated that Maine was
air temperature of only a few degrees have also been cooler than Rhode Island during June-July in 1996 but
hypothesized by Wethey (1983) to affect the distribu- direct measurements of substrate temperature showed
tion and abundance of intertidal organisms. Although no difference among the regions at this time. This may
tissue temperatures were not measured, both maximum partly be because overall weather patterns may not ac-
curately predict physical conditions at specific study
TABLE6. Summary statistics of daily temperature anomalies sites (see Helmuth 1998, 1999) as well as due to dif-
(difference between Maine and Rhode Island) in maximum ferences in sampling frequency between thermometers
temperature collected at weather stations near the study and weather-station data.
sites duringthe summersof 1995 and 1996. Differences in instrument replication also make com-
Statistic 1995 1996 parisons among years difficult. For example, during
1995 when species interactions varied between regions,
Negative anomalies (%) 70 65
Mean daily temperature diff. (?C) -2.40 -1.69 low thermometer replication and low statistical power
1 SD of mean daily temp. diff. (?C) 4.29 4.15 made it difficult to detect differences in substrate tem-
1 SE of mean daily temp. diff. (?C) 0.32 0.31
-10.30
perature between northern and southern sites. Average
Minimum daily difference (?C) -15.11
Maximum daily difference (?C) 7.83 8.54 differences in temperature of only 1-3?C, however, may
n 183 183 be biologically important but difficult to detect using
Note: Fig. 10 (top and middle panels) graphically presents min/max thermometers. Instead of thermometers,
the 7-d running average of the original 183 daily anomalies. Leonard et al. (1999) used computer-controlled therm-
April 2000 VARIATIONIN SPECIESINTERACTIONS 1027
istors to readily detect differences in substrate tem- predators in a region of reduced temperatures). My
perature <2.5?C among New England intertidal habi- results support the hypothesis that small differences in
tats that had significant effects on the mortality of Sem- physical factors can be important in governing species
ibalanus balanoides. Although expensive, these therm- interactions over 100s of kilometers and that the com-
istors may be the best means to detect subtle differences bination of direct and indirect effects that vary across
in physical conditions among sites. In lieu of these the life history determine the magnitude and direction
instruments, weather-station data may be useful in re- of overall species interactions. Long-term experiments
lating experimental results to broad-scale climatic con- at a series of sites along the New England coastline
ditions (Hargrove and Pickering 1992). (which would increase statistical power) will be nec-
Finally, it should be noted that my study was done essary to evaluate the validity and generality of these
at only two sites in each region over two years. This conclusions.
experimental design resulted in low statistical power
to detect differences among regions and cautions that The changing nature of species interactions during
my findings may not apply generally to all New Eng- global climate change
land intertidal habitats. Despite these limitations, my Few scientists question that humans have signifi-
data are the first to document the previously inferred cantly altered global carbon and nitrogen cycles and
difference in physical conditions across the Cape Cod have had profound effects on climate (Vitousek et al.
peninsula that has been argued to be important in the 1997). Greenhouse gases are expected to increase at
distribution and ecology of rocky-shore organisms least into the 21st Century, resulting in a 1.5-4.5?C
(Wethey 1983). warming trend and an increase in the temporal and
Direct and indirect interactions and the net effect on spatial variability of many aspects of the global climate
(Gates 1993). Temperature-sensitive species and those
organism fitness with whom they interact are likely to be influenced by
Although examples of positive interactions are be- these changing conditions.
coming increasingly common in the literature (e. g., As climate continues to warm, algal canopies in the
Carlsson and Callaghan 1991, Bertness and Shumway high intertidal zone of sites in southern New England
1993, Agular and Sala 1994, Berkowitz et al. 1995, may buffer species from local changes in distribution.
Callaway 1995), few studies have taken a demographic The upper limit of barnacles may largely be set by the
approach to evaluate the nature of species interactions. presence of algal canopies in the future, while in areas
For example, although Carlsson and Callaghan (1991) without canopies the vertical limit may be much lower.
showed that a sedge growing with a shrub in the Arctic Species persistence at sites to the north is harder to
tundra had a higher growth rate than when growing predict because of the indirect effect of the canopy on
alone, there was no attempt to determine effects of the barnacle predators. As in the south, barnacles may
shrub on seed supply, germination success, or mortal- come to rely on the buffering capacity of these common
ity, all of which directly contribute to fitness. Inte- species of algae, especially if the predators shift their
grating these effects as "lxmx" (survival probability X distribution more than that of the barnacles. Alterna-
reproductive output) is critical in determining if a net tively, as climate warms, the refuge from predators
interaction is positive or negative. above the algal canopy at northern sites (Menge 1976)
In my study, integrating component parts revealed may vanish and predator-prey interactions under the
that the overall interaction between the algal canopy canopy may intensify. This could result in local ex-
and barnacles switched from negative to positive be- tinction of either predator, prey, or both. The outcome
tween regions to the north and south of Cape Cod. will depend on the relative susceptibility of predator
Moreover, this interaction differed between the two and prey to regional changes in climate (Menge and
years of the study. Because of an overwhelmingly large Olson 1990) and their interactions with organisms that
direct effect on survival in 1995 at the southern sites, can modify the local conditions. Variable species in-
the algal canopy facilitated the reproductive fitness po- teractions, such as those documented here, are likely
tential of barnacle recruits there. On the other hand, to be evident in other habitats as well, where a single
the net interaction with the canopy was consistently species or suite of species strongly moderates physical
negative at northern sites, evidently due to canopy- conditions. Further work in these and other habitats
mediated indirect effects on predation pressure com- would contribute substantially to our understanding of
bined with lower reproductive output under the canopy. how ecological and evolutionary dynamics in natural
In 1996, when the canopy-mediated effect on survival communities might change over the next century.
was weaker at southern sites, fitness did not vary be-
ACKNOWLEDGMENTS
tween regions.
From these data, I conclude that net species inter- I am indebted to M. Baker, M. Bertness, S. Brewer, J. Eddy,
A. Ingraham, T. Leonard, J. Levine, and P. Schmidt, all of
actions switched from positive at southern sites (be- whom helped in the field during less-than-ideal conditions.
cause of few predators and elevated temperatures) to T. Miller and K. Eckelbarger of the University of Maine fa-
negative at northern sites (because of a suite of boreal cilitated my stays at the Darling Marine Center and the U.
1028 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
S. Fish and Wildlife Service kindly provided access to the Connell, J. H. 1961. The influence of interspecific compe-
Middletown site at the Sachuest Point National Wildlife Ref- tition and other factors on the distribution of the barnacle
uge. M. Bertness, S. Gaines, and D. Rand contributed sig- Chthamalus stellatus. Ecology 42:710-723.
nificantly to the development of the ideas in this paper. T. Connell, J. H. 1985. The consequences of variation in initial
Leonard steadfastly supported me during this work. Earlier settlement vs. post-settlement mortality in rocky intertidal
versions of the manuscript were greatly improved by com- communities. Journal of Experimental Marine Biology and
ments from K. Earls, T. Rand, P. Ewanchuk, J. Witman, J. Ecology 93:11-45.
Bruno, T. Minchinton, B. Menge, and two anonymous re- Cossins, A. R., and K. Bowler. 1987. Temperature biology
viewers. This work was funded, in part, by the Andrew Mel- of animals. Chapman & Hall, London, UK.
lon Foundation and grants from Sigma Xi and The Sounds Crisp, D. J. 1959. Factors influencing the time of breeding
Conservancy to G. H. Leonard and an NSF grant to M. Bert- of Balanus balanoides. Oikos 10:275-289.
ness. Crisp, D. J. 1964. Racial differences between North Amer-
ican and European forms of Balanus balanoides. Journal
LITERATURE CITED
of the Marine Biological Association of the United King-
Agular, M. R., and 0. E. Sala. 1994. Competition, facilitation, dom 44:33-45.
seed distribution and the origin of patches in a Patagonian
Dayton, P. K. 1971. Competition, disturbance, and commu-
steppe. Oikos 70:26-34. nity organization: the provision and subsequent utilization
Allee, W. C. 1923. Studies in marine ecology. IV. The effect of space in a rocky intertidal community. Ecological Mono-
of temperature in limiting the geographical range of in-
graphs 41:351-389.
vertebrates of the Woods Hole littoral. Biological Bulletin
Denny, M. W. 1983. A simple device for recording the max-
IV:341-354. imum force exerted on intertidal organisms. Limnology and
Bakus, G. J. 1969. Energetics and feeding in shallow marine Oceanography 28:1269-1274.
waters. International Review of General and Experimental Dethier M. N., and D. O. Duggins. 1988. Variation in strong
Zoology 4:275-369. interactions in the intertidal zone along a geographical gra-
Barnes, H . 1958a. Regarding the southern limits of Balanus dient: a Washington-Alaska comparison. Marine Ecology
balanoides (L.). Oikos 9:139-157.
Progress Series 50:97-105.
Barnes, H . 1958b. Temperature and the life-cycle of Balanus
balanoides (L.). Friday Harbour Symposium on Limnoria Duggins, D. O., J. E. Eckman, and A. T. Sewell. 1990. Ecol-
and Barnacles. University of Washington, Seattle, Wash- ogy of understory kelp environments. II. Effects of kelps
on recruitment of benthic invertebrates. Journal of Exper-
ington, USA. imental Marine Biology and Ecology 143:27-45.
Barnes, H., and M. Barnes. 1959. A comparison of the annual
Easterling, D. R., B. Horton, P. D. Jones, T. C. Peterson, T.
growth patterns of Balanus balanoides (L.) with particular R. Karl, D. E. Parker, M. J. Salinger, V. Razuvayev, N.
reference to the effect of food and temperature. Oikos 10:
1-18. Plummer, P. Jamason, and C. K. Folland. 1997. Maximum
and minimum temperature trends for the globe. Science
Bell, E. C., and M. W. Denny. 1994. Quantifying "wave 277:364-367.
exposure": a simple device for recording maximum veloc- Eckman, J. E. 1996. Closing the larval loop: linking larval
ity and results of its use at several field sites. Journal of
ecology to the population dynamics of marine benthic in-
Experimental Marine Biology and Ecology 181:9-29. vertebrates. Journal of Experimental Marine Biology and
Berkowitz, A. R., C. D. Canham, and V. R. Kelly. 1995.
Ecology 200:207-237.
Competition vs. facilitation of tree seedling growth and
survival in early successional communities. Ecology 76: Eckman, J. E., and D. O. Duggins. 1991. Life and death
1156-1168. beneath macrophyte canopies: effects of understory kelps
on growth rates and survival of marine, benthic suspension
Bertness, M. D. 1989. Intraspecific competition and facili-
tation in a northern acorn barnacle population. Ecology 70: feeders. Oecologia 87:473-487.
257-268. Eckman, J. E., D. O. Duggins, and A. T. Sewell. 1989. Ecol-
Bertness, M. D., and R. Callaway. 1994. Positive interactions ogy of understory kelp environments. I. Effects of kelps
in communities. Trends in Ecology and Evolution 9:191- on flow and particle transport near the bottom. Journal of
193. Experimental Marine Biology and Ecology 129:173-187.
Bertness, M. D., and S. D. Gaines. 1993. Larval dispersal Fischer, A. G. 1960. Latitudinal variations in organic diver-
and local adaptation in acorn barnacles. Evolution 47:316- sity. Evolution 14:64-81.
320. Foster, B. A. 1969. Tolerance of high temperatures by some
Bertness, M. D., S. D. Garrity, and S. C. Levings. 1981. intertidal barnacles. Marine Biology 4:326-332.
Predation pressure and gastropod foraging: a tropical-tem- Foster, B. A. 1971. Dessication as a factor in the intertidal
zonation of barnacles. Marine Biology 8:12-29.
perate comparison. Evolution 35:995-1007.
Bertness, M. D., and S. D. Hacker. 1994. Physical stress and Gates, D. M. 1993. Climate change and its global conse-
positive associations among marsh plants. American Nat- quences. Sinauer Associates, Sunderland, Massachusetts,
uralist 144:363-372. USA.
Bertness, M. D., and S. W. Shumway. 1993. Competition and Goldberg, D. E., and A. M. Barton. 1992. Patterns and con-
facilitation in marsh plants. American Naturalist 142:286- sequences of interspecific competition in natural commu-
292. nities: a review of field experiments with plants. American
Callaway, R. M. 1995. Positive interactions among plants. Naturalist 139:771-801.
Botanical Review 61:306-349. Gosner, K. L. 1978. A field guide to the Atlantic seashore.
Callaway, R. M., and L. R. Walker. 1997. Competition and Houghton Mifflin, Boston, Massachusetts, USA.
facilitation: a synthetic approach to interactions in plant Grace, J. B. 1991. A clarification of the debate between
communities. Ecology 78:1958-1965. Grime and Tilman. Functional Ecology 5:583-587.
Carefoot, T 1977. Pacific seashores. A guide to intertidal Grant, W. S. 1977. High intertidal community structure on a
ecology. University of Washington Press, Seattle, Wash- rocky headland in Maine, USA. Marine Biology 44:15-25.
ington, USA. Greenlee, J. T., and R. M. Callaway. 1996. Abiotic stress and
Carlsson, B. A., and T. V. Callaghan. 1991. Positive plant the relative importance of interference and facilitation in
interactions in tundra vegetation and the importance of montane bunchgrass communities in western Montana.
shelter. Journal of Ecology 79:973-983. American Naturalist 148:386-396.
April 2000 VARIATION IN SPECIES INTERACTIONS 1029
Grime, J. P. 1973. Competitive exclusion in herbaceous veg- tertidal community. Marine Ecology Progress Series 95:
etation. Nature 242:344-347. 233-244.
Grime, J. P. 1977. Evidence for the existence of three primary Moloney, K. A. 1990. Shifting demographic control of a
strategies in plants and its relevance to ecological and evo- perennial bunchgrass along a natural habitat gradient. Ecol-
lutionary theory. American Naturalist 111:1169-1194. ogy 71:1133-1143.
Grime, J. P. 1979. Plant strategies and vegetation processes. Paine, R. T. 1994. Marine rocky shores and community ecol-
John Wiley & Sons, New York. ogy: an experimentalist's perspective. Ecology Institute,
Hargrove, W. W., and J. Pickering. 1992. Pseudoreplication: Oldendorf/Luhe, Germany.
a sine qua non for regional ecology. Landscape Ecology Palmer, A. R., J. Szymanska, and L. Thomas. 1982. Pro-
6:251-258. longed withdrawal: a possible predator evasion behavior
Hawkins, S. J. 1983. Interactions of Patella and macroalgae in Balanus glandula (Crustacea: Cirripedia). Marine Bi-
with settling Semibalanus balanoides (L.). Journal of Ex- ology 67:51-55.
perimental Marine Biology and Ecology 71:55-72. Raimondi, P. T. 1990. Patterns, mechanisms, consequences
Helmuth, B. S. 1998. Intertidal mussel microclimates: pre- of variability in settlement and recruitment of an intertidal
dicting the body temperature of a sessile invertebrate. Ecol- barnacle. Ecological Monographs 60:283-309.
ogy 68:51-74. Rhode, K. 1992. Latitudinal gradients in species diversity:
Helmuth, B. S. 1999. Thermal biology of rocky intertidal the search for the primary cause. Oikos 65:514-527.
mussels: quantifying body temperatures using climatolog- Ricketts, E. E, J. Calvin, and J. W. Hedgpeth. 1985. Between
ical data. Ecology 80:15-34. Pacific tides. Stanford University Press, Stanford, Califor-
Jeanne, R. L. 1979. A latitudinal gradient in rates of ant nia, USA.
predation. Ecology 60:1211-1224. Rosenzweig, M. L. 1995. Species diversity in space and time.
Leonard, G. H. 1999a. Positve and negative effects of in- Cambridge University Press, Cambridge, UK.
tertidal algal canopies on recruitment and survival of bar- Sanford, E., D. Bermudez, M. D. Bertness, and S. D. Gaines.
nacles. Marine Ecology Progress Series. 178:241-249. 1994. Flow, food supply and acorn barnacle population
Leonard, G. H. 1999b. Population and community conse- dynamics. Marine Ecology Progress Series 104:49-62.
quences of physical forcing mechanisms in New England Southward, A. J. 1955. On the behavior of barnacles. I. The
intertidal habitats. Dissertation. Brown University, Provi- relation of cirral beat and other activities to temperature.
dence, Rhode Island, USA. Journal of the Marine Biological Association of the United
Leonard, G. H., P. J. Ewanchuk, and M. D. Bertness. 1999. Kingdom 34:413-422.
How recruitment, intraspecific interactions, and predation Southward, A. J. 1958. Note on the temperature tolerances
control species borders in a tidal estuary. Oecologia 118: of some intertidal animals in relation to environmental tem-
492-502.
peratures and geographical distribution. Journal of the Ma-
Lewis, J. R. 1964. The ecology of rocky shores. English rine Biological Association of the United Kingdom 37:49-
University Press, London, UK. 66.
Little, C., and J. A. Kitching. 1996. The biology of rocky Spight, T. M. 1976. Censuses of rocky shore prosobranchs
shores. Oxford University Press, Oxford, UK. from Washington and Costa Rica. Veliger 18:309-317.
MacArthur, R. H. 1965. Patterns of species diversity. Bio-
Stephenson, T. A., and A. Stephenson. 1948. The universal
logical Reviews 40:510-533. features of zonation between tide-marks on rocky coasts.
MacArthur, R. H. 1972. Geographical ecology: patterns in Journal of Ecology 37:289-305.
the distribution of species. Harper & Row, New York, New
Tilman, D. 1988. Plant strategies and the dynamics and struc-
York, USA. ture of plant communities. Princeton University Press,
Mayr, E. 1963. Animal species and evolution. Harvard Uni- Princeton, New Jersey, USA.
versity Press, Cambridge, Massachusetts, USA.
Underwood, A. J. 1986. Physical factors and biological in-
McPeek, M. A., and B. L. Peckarsky. 1998. Life histories
and the strengths of species interactions: combining mor- teractions: the necessity and nature of ecological experi-
ments. Pages 372-390 in P. G. Moore and R. Seed, editors.
tality, growth, and fecundity effects. Ecology 79:867-879. The ecology of rocky coasts. Columbia University Press,
Menge, B. A. 1976. Organization of the New England rocky New York, New York, USA.
intertidal community: role of predation, competition, and
environmental heterogeneity. Ecological Monographs 46: Underwood, A. J. 1997. Experiments in ecology. Their log-
355-393. ical design and interpretation using analysis of variance.
Menge, B. A. 1978. Predation intensity in a rocky intertidal Cambridge University Press, Cambridge, U.K.
community: effect of an algal canopy, wave action and Underwood, A. J., and E. J. Denley. 1984. Paradigms, ex-
dessication on predator feeding rates. Oecologia 34:17-35. planations, and generalizations in models for the structure
Menge, B. A. 1983. Components of predation intensity in of intertidal communities on rocky shores. Pages 151-180
the low zone of the New England rocky intertidal region. in D. R. Strong, Jr., D. Simberloff, L. G. Abele, and A. B.
Oecologia 58:141-155. Thistle, editors. Ecological communities: conceptual issues
Menge, B. A., and J. Lubchenco. 1981. Community orga- and the evidence. Princeton University Press, Princeton,
nization in temperate and tropical rocky intertidal habitats: New Jersey, USA.
prey refuges in relation to consumer pressure gradients. Vermeij, G. J. 1978. Biogeography and adaptation: patterns
Ecological Monographs 51:429-450. of marine life. Harvard University Press, Cambridge, Mas-
Menge, B. A., and A. M. Olson. 1990. Role of scale and sachusetts, USA.
environmental factors in regulation of community struc- Vermeij, G. J., and J. A. Veil. 1978. A latitudinal pattern in
ture. Trends in Ecology and Evolution 5:52-57. bivalve shell gaping. Malacologia 17:57-61.
Menge, B. A., and J. P. Sutherland. 1987. Community reg- Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M.
ulation: variation in disturbance, competition, and preda- Melillo. 1997. Human domination of Earth's ecosystems.
tion in relation to environmental stress and recruitment. Science 277:494-499.
American Naturalist 130:730-757. Wallace, A. R. 1878. Tropical nature and other essays. Mac-
Minchinton, T. E., and R. E. Scheibling. 1993. Free space Millan, London, UK.
availability and larval substratum selection as determinants Weldon, C. W., and W. L. Slauson. 1986. The intensity of
of barnacle population structure in a developing rocky in- competition versus its importance: an overlooked distinc-
1030 GEORGE H. LEONARD Ecology, Vol. 81, No. 4
tion and some implications. Quarterly Review of Biology Wethey, D. S. 1985. Catastrophe, extinction, and species di-
61:23-44. versity: a rocky intertidal example. Ecology 66:445-456.
Wethey, D. S. 1983. Geographic limits and local zonation: Wilson, S. D., and P. A. Keddy. 1986. Measuring diffuse com-
the barnacles Semibalanus (Balanus) and Chthamalus in petition along an environmental gradient:results from a shore-
New England. Biological Bulletin 165:330-341. line plant community. American Naturalist 127:862-869.
Wethey, D. S. 1984. Sun and shade mediate competition in Wootton, J. T. 1993. Indirect effects and habitat use in an
the barnacles Chthamalus and Semibalanus: a field exper- intertidal community: interaction chains and interaction
iment. Biological Bulletin 167:176-185. modifications. American Naturalist 141:71-89.