Forgot your password?
Document Actions
BenedettiCecchi et al 06
Temporal Variance Reverses the Impact of High Mean Intensity of Stress in Climate Change
Experiments
Author(s): Lisandro Benedetti-Cecchi, Iacopo Bertocci, Stefano Vaselli, Elena Maggi
Source: Ecology, Vol. 87, No. 10 (Oct., 2006), pp. 2489-2499
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/20069260
Accessed: 29/12/2009 13:52
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, 87(10), 2006, pp. 2489-2499
? 2006 by the Ecological Society of America
TEMPORAL VARIANCE REVERSES THE IMPACT OF HIGH MEAN
INTENSITY OF STRESS IN CLIMATE CHANGE EXPERIMENTS
Lisandro Benedetti-Cecchi,1 Iacopo Bertocci, Stefano Vaselli, and Elena Maggi
Dipartimento di Biolog?a, Via A. Volta, 6, 1-56126 Pisa, Italy
Abstract. Extreme climate events simultaneous
produce changes to the mean and to the
variance of climatic variables overecological time scales. While several studies have
investigated how ecological systems respond to changes in mean values of climate variables,
the combined effects of mean and variance are poorly understood. We examined the response
of low-shore assemblages of algae and invertebrates of rocky seashores in the northwest
Mediterranean to factorial manipulations of mean intensity and temporal variance of aerial
exposure, a type of disturbance whose intensity and of occurrence are
temporal patterning
predicted to change with changing climate conditions. Effects of variance were often in the
opposite direction of those elicited by changes in the mean. Increasing aerial exposure at
regular intervals had negative effects both on diversity of assemblages and on percent cover of
filamentous and coarsely branched algae, but greater temporal variance drastically reduced
these effects. The opposite was observed for the abundance of barnacles and encrusting
coralline algae, where high temporal variance of aerial exposure either reversed a positive
effect of mean intensity (barnacles) or caused a negative effect that did not occur under low
temporal variance (encrusting algae). These results provide the first experimental evidence that
changes in mean intensity and temporal variance of climatic variables affect natural
assemblages of species interactively, suggesting that high temporal variance may mitigate the
ecological impacts of ongoing and predicted climate changes.
Key words: aerial exposure; climate change; disturbance; diversity; mean intensity; rocky shore;
temporal variance.
Introduction (Hughes 2000, Zavaleta et al. 2003, Post and Forch
There is increasing concern
hammer 2004), in the ability of assemblages to resist or
among scientists, policy
to recover from other disturbances and in productivity
makers, and the general public about the societal and
(Grime et al. 2000). Most of these studies have, however,
ecological consequences of climate change. Climate
related average values of ecological variables to trends in
events can affect society directly by causing catastrophes
average values of climate variables, little
and by threatening human health and indirectly by offering
into the relative roles of climate means as
altering the functioning of ecological systems (Easterling insights
compared with climate variances and into their inter
et al. 2000?>). Models ofclimate change generally agree
actions. Although some important studies have inves
that extreme events such as droughts, storms, and floods
are becoming more tigated the effects of changing the spatial or the
frequent (Michener et al. 1997, Allen
et al. 2000, Benestad temporal variability of processes such as competition
2003). Changes in the occurrence of
extreme weather conditions produce concomitant (Hutchings et al. 2003), pr?dation (Butler 1989, Nav
to the mean and to the variance of the
arrete 1996), and disturbance (Collins 2000, McCabe
changes
distributions of climatic vari
and Gotelli 2000), there seems to be little effort to
frequency corresponding
ables et al. understand effects of variance in the context of climate
(Easterling 2000?, b). For example, the
incidence of extremely high temperatures results in long change.
term trends in mean values as well as in more or less Recent attention to these issues has led to the
in temporal formulation of a framework for the
pronounced increases variance, depending investigating
on the time scale considered and and combined effects of in mean
(Hughes 2000, M?ller separate changes
Stone Luterbacher et al. intensity and spatial or temporal variance of
2001, 2004). ecological
Shifts in climate conditions can have processes (Benedetti-Cecchi 2000a, 2003, Bertocci et al.
profound
ecological impacts, including changes in patterns of 2005). The experimental designs proposed for these
analyses can provide a useful to inves
distribution, abundance, and diversity of species starting point
tigate the effects of climate change on natural assem
blages. High temporal variance implies that several
Manuscript received 28 November 2005; revised 14 March
climatic events occur over short intervals of time and
2006; accepted 27 March 2006. Corresponding Editor: P. T.
Raimondi. that these alternate with long periods in which no event
1E-mail: occurs
lbenedetti@biologia.unipi.it (Benedetti-Cecchi 2003). Populations of long
2489
2490
LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
lived species may face more extreme physical conditions We used a factorial experiment to test the hypothesis
when climatic events such as
droughts, floods and that variance and mean of aerial
temporal intensity
storms operate with high compared to low temporal exposure would affect of shores
assemblages rocky
variance over ecological time scales. This may be interactively, with variance more im
temporal being
particularly true for less resilient species that recruit portant under high than low mean intensity of exposure.
episodically and/or that have low dispersal capabilities Intensity of aerial exposure was defined as the time
(Gaines and Roughgarden 1985, Caley et al. 1996). In assemblages remained emersed over the study period.
contrast, fugitive species may benefit from high temporal This was manipulated indirectly by transplanting
variance by taking advantage of newly released resour organisms higher on the shore for defined periods of
ces (Dayton 1971). These effects are, however, not time. Temporal variance was defined with reference to
independent of the average intensity of events. It is the interval of time between successive manipulations.
reasonable to expect that variable events that Organisms were moved up and down the shore either at
temporally
operate with on average, have much regular or at variable intervals of time, while
high intensity larger ensuring
effects on natural populations than events that operate that the overall time spent in the high-shore position did
with the same level of variance, but with lower not differ between levels of temporal variance over the
temporal
mean The idea that interactions between duration of the study. Although this study was
intensity.
intensity and temporal variance of climate events may conducted under peculiar tidal conditions and focused
be for major in populations and on a specific aspect of climate it addressed the
responsible changes change,
has never been more general problem of whether the definition of
assemblages investigated formally.
of algae and invertebrates extreme climate conditions in an ecological context
Assemblages inhabiting
intertidal habitats of shores are to should take into consideration both effects of mean
rocky exposed
variable conditions that can become intensity and temporal variance of events.
physical easily
detrimental to their life (Schonbeck and Norton 1978,
Materials and Methods
Bertness and Leonard 1997, Dethier et al. 2005).
Thermal stress due to aerial and disturbance Study sites and experimental design
exposure
by waves are recurrent events that contribute to This was at
study replicated three sites (stretches of
maintain spatial and temporal variability in the struc coast of 30-40 m to
long and hundreds thousands of
ture of these assemblages (Stephenson and Stephenson meters chosen to represent the exposed
apart) randomly
1949, Dayton 1971, Sousa 1979). These effects can, shores south of Livorno, in the northwest
rocky
however, be magnified by global warming and by the Mediterranean. A first site was established in July 2001
increasing frequency and intensity of storms, leading to and two additional sites were established inMay 2002;
quantifiable changes in patterns of abundance, distribu lasted two at all sites. Assemblages on
experiments years
tion and diversity of species (Davison et al. 1993, these shores are described in Benedetti-Cecchi (20006,
Navarrete et al. 1993, Barry et al. 1995, Bertness et al. low-shore were dominated
2001). Briefly, assemblages
1999, Sanford 1999). by encrusting and filamentous algae (including several
While aerial exposure is a predictable event in systems in the and
species genus Ceramium, Polysiphonia,
characterised by large tides, it is less so where tides have and by branched like
Cladophora) coarsely algae
low amplitude. In the northwest Mediterranean, for Laurencia obtusa Lamouroux and Chondria
(Hudson)
example, where tides rarely exceed 30 cm in amplitude, De Toni. In contrast, the barnacle
boryana (J. Agardh)
weather conditions can variable stellatus
impose temporal Chthamalus (Poli) and cyanobacteria (Rivularia
patterns of aerial exposure and desiccation to sessile dominated on the shore. Although assem
spp.) higher
organisms offsetting any effect of the tide. Prolonged differed among habitats, most of the differences
blages
periods of calm sea and high barometric pressure can were due to changes in relative abundance of taxa rather
push the sea level below the mean low water level than in species A notable was
composition. exception
(MLWL) so that organisms remain exposed to air long the red alga Rissoella verruculosa (Bertoloni) J. Agardh,
enough to dry. Rough conditions, in contrast, maintain which was characteristic of the mid-shore habitat
wet even at low tide.
organisms constantly Therefore, (Benedetti-Cecchi 20006).
changes in the timing of occurrence and duration of At each site, 32 cores 10 cm in diameter were drilled
contrasting climatic conditions can have profound out of the rock from the low-shore habitat (0-5 cm
effects on the biota of these shores. Indeed, some above MLWL) using a diamond-tipped corer mounted
climatic models predict changes in the intensity and on a petrol driller (Tanaka America, Auburn, Wash
temporal variance of storms, with strong events USA). The cores, with intact on
ington, assemblages
concentrating in short periods of time separated by top, were then assigned randomly to the following
longer periods of good weather (M?ller and Stone 2001). treatments (defined according to the terminology
These features make Mediterranean rocky seashores a commonly used in transplant experiments [Chapman
valuable system to explore the ecological consequences 1986]), with eight cores per treatment: (1) disturbed
of changes in mean intensity and temporal variance of cores (i.e., cores that were placed back to their original
climatic events. position immediately after drilling), (2) translocated
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2491
cores cores moved to another within the low artefacts associated with moving the organisms to
(i.e., place
in vertical
shore habitat), (3) cores transplanted to the mid-shore another place, irrespective changes of
habitat (15-20 cm above MLWL), and (4) cores position. This was done by swapping positions between
to the high-shore habitat (25-30 cm above cores in the low-shore habitat.
transplanted
MLWL). Within each treatment, four of the eight cores Cores were attached to the substratum with epoxy
chosen at random were manipulated at almost regular putty (Subcoat S, Veneziani S.p.A, Lodi, Italy) at Site 1.
intervals during the study period, whilst the remaining This procedure ensured a firm attachment, but it was
four cores were manipulated at variable intervals. This logistically demanding because it required new drills
created a factorial experiment with two levels of each time the cores had to be moved up or down the
temporal variance crossed with each treatment. Four shore. A different technique was used at sites 2 and 3.
replicate unmanipulated were also established in Two stainless steel bars were anchored in opposite
plots
the low-shore habitat as controls. Thus, the experiment positions along each core with screws inserted into the
consisted the following experimental
of conditions: rock. Each bar had an outward lip in correspondence to
unmanipulated controls (C), disturbed cores (low the top of the core that was used to screw the whole
temporal variance, DL; high temporal variance, DH), structure to the substratum. Regardless of type of
translocated cores (low temporal variance, TL; high manipulation, all cores were positioned into holes in
temporal variance, TH), cores transplanted to the mid such a way that that their top was approximately at the
shore habitat (low temporal variance, ML; high same level of the natural substratum.
temporal variance, MH) and cores transplanted to the How well realized treatment conditions matched
high-shore habitat (low temporal variance, HL; high intended experimental effects depended on local climate.
temporal variance, HH). For example, rough weather could prevent aerial
Intensity of aerial exposure was manipulated indi exposure of assemblages transplanted both to mid-shore
rectly by transplanting the cores to different heights of and high-shore habitats. Similarly, calm sea and high
the shore with respect to MLWL. Positions were chosen barometric pressure might expose all habitats to aerial
to represent distinct environmental conditions on the conditions. Thus, both good and bad weather could, in
basis of our previous knowledge of the system (Menconi principle, eliminate intended differences among treat
et al. 1999, Benedetti-Cecchi 20006, 2001). With calm ments. We addressed this issue in two ways. First, we
sea, the low-shore habitat was exposed to air only at low collated daily data on the direction and speed of winds
tide concomitant with high barometric pressure, the and on barometric pressure for the entire duration of the
mid-shore habitat was always emersed at low tide and study (courtesy of the Istituto Idrografico e Mareogra
the high-shore habitat was emersed both at high and low fico di Pisa). These data served to determine the
tide. Small waves, like those
produced by boats, washed proportion of days intended experimental conditions
both the low-shore and mid-shore habitats, but not the were offset by adverse weather. Conditions in which
high-shore habitat. All habitats were almost continu rough sea might have prevented aerial exposure of both
ously submersed when the sea was rough. mid-shore and high-shore habitats at low tide were
To obtain the conditiontemporal of low variance of defined as those in which the daily average speed of
aerial exposure, cores were transplanted into new holes winds was equal to or larger than 2.5 kn (1.286 m/s), in
either at mid-shore or high-shore heights for a month directions ranging from southeast to north. In contrast,
and then returned back to their original position in the conditions in which all habitats might have been
low-shore habitat. This manipulation was applied exposed to air due to low sea levels were defined as
approximately every three months, so that the overall those in which barometric pressure was equal to or
length of time low-shore assemblages remained in the larger than 570 and 574 hPa in winter and summer,
high-shore position was about four months in a year. respectively (to correct for seasonal changes in average
Bad weather prevented a perfectly regular distribution aerial temperature) and in which the wind was either
of events in time. The condition of high temporal absent or blew seawards (eastern quadrant) at any
variance of aerial exposure was obtained by trans speed.
planting the cores at irregular intervals, but ensuring Second, we two thermom
deployed replicate digital
that the overall time cores were in the high-shore eters (FT-800 system, Econorma Sas, Treviso, at
Italy)
position was still four months per year. This enabled each tidal height in each site, in the proximity of
the independent manipulation of intensity and temporal transplanted cores. The thermometers were housed in
variance of aerial exposure over the time scale of the PVC that were screwed into the rock and
pipes
study (Fig. 1). measured either water or aerial temperature (depending
Disturbed and translocated treatments were used to on whether or not they were submersed) every hour for
assess several potential artefacts associated with these two months. At the end of the recording period the data
manipulations (Chapman 1986, Kelaher et al. 2003). were downloaded to a computer and the thermometers
Disturbed cores controlled for the effects of drilling, were deployed again in the field. These measurements
manipulating and attaching the cores to the substratum. were used as indicators of the status (emersed or
The translocation treatment controlled for possible submersed) of each habitat through time.
2492
LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
A) Site 1
Low variance
ft t t ttt !
High variance
TT
iJiAiSlOtNlDiJ|FlMlAlMiJlJlAlS|OlNlDiJlFlMlA|Ml
??TF T
2001 2002 2003
A A A A A A
B) Sites 2 and 3
Low variance
4W
f T T Iff
High variance
?
IT m tf
iiJ|J|A|S|0|N|D|J|F|M|A|M|J|J|A|S|0|N|D|J|F|M|A||
2002 2003
A A A A A
Fig. 1. Schematic representation of the experimental design. Black and gray blocks indicate periods in which cores were in
high-shore and low-shore positions, respectively. Down-facing arrows indicate periods in which high barometric pressure and calm
sea conditions exposed the low-shore habitat to air, while up-facing arrows indicate periods in which all habitats were submersed
due to rough sea conditions. The temporal extent of these periods in which intended experimental conditions could not be
maintained is indicated by the thickness of the arrows (with thinnest and thickest arrows corresponding to one and five days,
respectively). Only events that coincided with the periods of transplantation are shown. Time of sampling is indicated by the
arrowheads at the bottom of each panel.
Sampling and statistical analyses achievable in the field. Taxa that could not be identified
at the level of species or genus were lumped into
Cores were sampled six times at site 1 and five times at
sites 2 and 3. At each date of morphological groups (Littler and Littler 1980, Steneck
sampling, transplanted
and Dethier 1994).
cores assigned to the two levels of temporal variance
Data were analysed using population-averaged gen
were in different phases with respect to the actual
eralized estimating equations (PA-GEEs), an extension
position on the shore and in relation to the interval of
of generalized linear models (GLMs) (Liang and Zeger
time since the last manipulation. To avoid the problem
1986, Quinn and Keough 2002). PA-GEEs proved useful
of confounding these contingencies with effects of
in the present context for two reasons. First, because the
variance, dates of sampling were chosen in such a way
hypothesis investigated in this study applied to the entire
that the mean interval of time since the last manipu
duration of the experiment and not to single dates of
lation was maintained as much as possible comparable
sampling, an
analysis that focused on population
across treatments (Fig. 1). averages was desirable. Second, because experimental
The number and percentage cover of dominant taxa
units were repeatedly sampled through time, a technique
on experimental cores and in unmanipulated quadrats was
that could account for temporal autocorrelation
were sampled nondestructively with a 7 X 7 cm plastic a link
necessary. PA-GEEs enable the specification of
frame divided into 25 1.4 X 1.4 cm2 sub-quadrats. Size of function and an error structure for the residuals as in
quadrats was dictated by the area available for sampling and also take into account the correlation
GLMs,
on each core and was appropriate to sample small between observations on the same unit.
experimental
organisms like those targeted in the study. Quantitative This results in robust estimates of variances and
data were obtained by recording the number of sub standard errors that can be used in hypothesis testing.
quadrats that contained a particular taxon and express Several of correlation can be specified,
patterns although
ing final values as percentages. Organisms were identi the procedure is robust to misspecification of the
fied to the most detailed level of taxonomic resolution correlation structure (Hardin and H?be 2003). We used
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2493
Table 1. Summary of results from population-averaged generalized estimating equations (PA-GEEs).
Encrusting Filamentous Chthamalus Coarsely
corallines
Diversity algae stellatus branched algae
Contrasts SI S3 SI S3 SI S3 SI
S3 SI
S3
Control vs. others ns ns NS NS
NS ** NS NS NS
T NS NS
NS **
M NS NS
H NS NS NS
* NS NS NS NS
y
T X V NS NS NS NS NS
M X V ns NS NS NS NS
H X V ns NS NS
** **
M vs. T
H vs. T NS
**
M X V vs. T X V NS NS
H X V vs. T X V NS NS
Notes: If disturbed and translocated cores did not differ significantly (translocated effect), then mid-shore and high-shore effects
were examined by contrasting these treatments with the disturbed cores. If translocated and disturbed cores differed significantly,
then mid-shore and high-shore terms were contrasted with the translocated cores. Similarly, if the effect of variance differed
between translocated and disturbed cores (translocated X variance interaction), then the mid-shore X variance and high-shore X
variance interactions were examined by contrasting these terms with the translocated X variance interaction. Full results of analyses
are reported in Appendix A. Abbreviations: SI, site 1; S3, site 3; T, translocated; M, mid-shore; H, high-shore; V, variance; ns, not
significant. Blank cells indicate that the test was not applicable.
* **
P < 0.05; P < 0.01; *** P < 0.001.
a first order autoregressive model AR(1) in all the analysis to check for strong deviations from the
analyses; this choice was motivated by the absence of distributional assumptions and homogeneity of varian
clear seasonal patterns in the raw data, so that ces.
correlation of residuals was expected to decrease as the
Results
time lag between observations increased.
Two types of response variables were analyzed: Meteorological data indicated that rough sea con
number of taxa (used as a surrogate of
diversity) and ditions might have prevented aerial exposure of cores
percentage covers of species or morphological groups. In transplanted to mid-shore and high-shore habitats in 19
the first case, a log-link function was used to relate the out of 235 days of experimental manipulation (8.1% of
expected value of the response variable to the predictor the time) for the condition of low temporal variance and
variables, assuming a Poisson distribution of the error in 15 out of 240 days (6.2% of the time) for the condition
terms. In the second case, a Gaussian distribution was of high temporal variance (up-facing arrows in Fig. 1) at
assumed for the errors and the identity link was used to site 1. Similar values were obtained at sites 2 and 3, with
relate the linear predictor with expected values of cores transplanted to the mid-shore and low-shore
response variables. Tests of hypotheses were based on habitats been submersed by waves in 17 out of 238 days
"treatment" contrasts and compared the unmanipulated of manipulation for the condition of low temporal
condition vs. all other treatments in first place. Trans variance (7.1% of the time) and in 10 out of 241 days for
located cores were then compared to disturbed cores to the condition high temporal variance (4.1% of the time).
detect possible artefacts associated with moving assem The number of days in which all habitats might have
blages from one place to another within the same been exposed to air during the periods in which the
habitat. If this test was not significant, the effects of transplanted cores were upshore were 0 at Site 1 and 10
transplanting cores to the mid-shore and high-shore out of 482 (2.1% of the time) and 24 out of 479 (5% of
habitats were examined by contrasting these treatments the time) for treatments exposed to low and high levels
with the disturbed cores. If translocated and disturbed of temporal variance, respectively (down-facing arrows
cores differed significantly, the mid-shore and high in Fig. 1) at sites 2 and 3. Collectively, these data
shore treatments were contrasted with the translocated indicated that the match between intended and realized
cores. Similarly, if the effect of variance differed between experimental conditions was always above 90%.
translocated and disturbed cores (translocated X var Similar conclusions can be drawn from the temper
iance interaction), then the mid-shore X variance and ature data, as
presented daily average temperatures
high-shore X variance interactions were examined by recorded between 10:00 and 18:00 (Fig. 2). Unfortu
contrasting these terms with the translocated X variance these data were not available for all habitats over
nately,
interaction (these are the mid-shore X variance vs. the entire study period due to themalfunctioning or loss
translocated X variance and high-shore X variance vs. of some thermometers. Nevertheless, measurements
translocated X variance contrasts in Table 1). Plots of indicated that the high-shore habitat was a distinct
residuals vs. predicted effects were examined after each thermal environment more extreme tern
experiencing
2494 LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
35
30
25
O
ICD20
Q.
E
15
10i
Dec Jan Feb Mar Apr May
2003 2004
=
Fig. 2. Temporal trends in daily average temperatures (between 10:00 and 18:00 hours) in low-shore (dotted line, n 3), mid
=
shore (black line, n=\), and high-shore (gray line, n 3) habitats. Originally there were two replicate thermometers in each habitat
on each shore. Missing data are due to the malfunctioning or loss of thermometers.
values and fluctuations than went lost between time 2 and time 4, probably as a
perature larger temporal
the low-shore habitat. These differences reflected the consequence of strong wave action. At site 3, in contrast,
of aerial exposure of the habitat, there were clear negative effects of transplantation on
patterns high-shore
with emersion indicated by lower temperatures in winter mean number of taxa, but only when cores were moved
(December-February) and by higher temperatures in on a regular basis (Fig. 3). While treatment HL suffered
spring (March-May) (Fig. 2). Analysis of data for the a reduction of 32.4% in mean number of taxa
compared
December-February 2003revealed significantly to controls, this effects was only 16% in treatment HH.
period
= ? X
lower mean temperature values (t 2.5, P < 0.05) and These patterns resulted in significant mid-shore
= X variance
significantly larger mean temporal variances (t 9.8, P variance and high-shore interactions in the
< 0.0004) in the high shore compared to the low-shore analysis (Table 1).
habitat (these are one-tailed tests because of the direc Similar buffering effects of variance were observed for
tional nature of the hypotheses; both tests have 4 df and filamentous and coarsely branched algae (Fig. 3). At
variances are homogenous when checked with Coch sites 1 and 2, effects of variance on filamentous algae
ran's C test). Differences between the low-shore and differed between disturbed and translocated treatments,
mid-shore habitats were less clear. The single reading suggesting the occurrence of artefacts (Fig. 3). An effect
available for the mid-shore habitat in the period between of variance-was, however, still evident at site 1 when
March and May 2004 prevented general conclusions cores transplanted to mid-shore and high-shore habitats
about this environment. were to translocated cores. Percentage cover
compared
to mid-shore and high of filamentous algae was
larger in treatments MH and
Assemblages transplanted
shore habitats changed both in terms of number and HH compared to treatments ML and HL, respectively,
percentage cover of taxa from low-shore assemblages while no such difference occurred between TL and TH.
and in several cases effects differed in relation to the These resulted in significant mid-shore X
patterns
vs. translocated X variance and high-shore X
temporal variance of the manipulation (Fig. 3). Tem variance
variance increased the mean number variance vs. translocated X variance contrasts in the
poral significantly
this effect was not to was
of taxa at site 1, although unique analysis (Table 1). A similar effect of variance
transplanted cores, but was also evident on disturbed observed at site 3 in the high-shore habitat (Fig. 3),
and translocated cores (Fig. 3, Table 1). A similar result although the test was not significant (0.05 < P < 0.06).
was observed at site 2 (Fig. 3), although no statistical Also the percentage cover of coarsely branched algae
was undertaken in this case because many cores was drastically reduced on cores transplanted to mid
analysis
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2495
Site 1 Site 2 Site 3
Diversity
6-U
^ 4
?ilMIlil Filamentous algae
C D T M H CD C D T M H
Fig. 3. Values of response variables (top row, no. taxa; other rows, percent cover by taxa) for the different treatments (mean +
se). Means were calculated by first averaging data from each core over time; n = 4 replicate cores except at site 2, where replication
was sparse due to the loss of a large number of cores between the second sampling date and the fourth sampling date; cores that
were lost contributed data only for the first part of the experiment). Abbreviations: C, controls (gray bars); D, disturbed; T,
translocated; M, transplanted to the mid-shore habitat; H, transplanted to the high-shore habitat. Open bars show low variance;
black bars show high variance.
2496 ANDR?
LIS BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
shore (site 2) and high-shore (site 3) habitats at low, but located treatments usually revealed the existence of
not at high variance. The latter effect was statistically intended effects of treatments over and above any effect
significant (Table 1). of the experimental procedure. Effects of treatments
In contrast to filamentous and coarsely branched could vary in magnitude and/or direction among sites,
algae, the percentage cover of encrusting coralline algae indicating that the sequence with which experimental
and Chthamalus stellatus were adversely affected by manipulations were differed between
applied (sequences
temporal variance (Fig. 3). Although disturbed and site 1 and sites 2 and 3, see Fig. 1), intrinsic differences
translocated treatments responded differently to a among sites or a combination of these factors, con
change in variance, as highlighted by significant trans tributed to some of the variability observed in the data.
located X variance interactions were our current
(Table 1), there Much of understanding of ecological
differences between treatments HL and HH that responses to environmental change comes from studies
emerged over and above any possible effect of the that have contrasted conditions in
experimental differing
experimental procedure. A nonsignificant trend was mean values of predictor variables, e.g., C02 (Zavaleta
evident for encrusting coralline algae at site 1 and a et al. 2003), or in which effects of mean and variance are
significant effect was observed at site 3 (contrast H X V varied simultaneously through changes in the frequency
vs. T X V in Table 1).At this site, the effect of variance of events, e.g., disturbance (Collins 2000). Recent studies
in the high-shore habitat was in opposite direction of have shown that changes in spatial or temporal variance
that observed for the translocated treatment in the low of physical and biological processes can have significant
shore habitat (Fig. 3). C. stellatus was similarly impacted effects on species and assemblages under constant mean
by variance when transplanted to the mid-shore (Site 2) conditions (Cardinale et al. 2002, Day et al. 2003,
and to the high-shore (sites 1 and 3) habitats. The most Hutchings et al. 2003). Our results add the almost
dramatic effect occurred at site 3 in the absence of unique evidence that effects of variance may not be
artefacts, as highlighted by the significant high-shore X independent of the mean.
Therefore, conceptual further
variance interaction in Table 1. At site 1, percentage and methodological steps are required in order to
cover of barnacles was lower on cores translocated at improve our understanding of variable
ecologically
high variance than on cores translocated on a regular phenomena, interactions between the mean
including
basis, but this effect of variance disappeared in the mid and the variance of predictor variables (Benedetti
shore habitat resulting in a significant M X V vs. T X V Cecchi 2003).
interaction (Table 1). Finally, no clear effects of treat We predicted that high temporal variance of aerial
ments were observed for Rivularia sp. and articulated exposure would have produced the largest effects when
coralline algae (Fig. 3, Table 1). assemblages were transplanted to the high-shore habitat,
under the most extreme conditions of aerial exposure
Discussion
(treatment HH). In principle, aerial exposure was
Our results revealed interactive effects of mean expected to produce a variety of ecological responses,
intensity and temporal variance of aerial exposure on including both negative and positive effects on abun
assemblages. Effects of variance were often in opposite dance, depending on the ability of organisms to with
direction of those elicited by changes in the mean. stand desiccation and on their capacity to colonize in
Increasing aerial exposure had negative effects both on emersed compared to submersed conditions. Indeed,
diversity of assemblages and on percentage cover of both types of effects were observed in treatment HH and
filamentous and coarsely branched algae, but these patterns appeared related to the life-history traits of taxa
effects were buffered by high temporal variance. The considered. Encrusting corallines were the most abun
opposite was observed for barnacles (Chthamalus dant taxa in the low-shore habitat, where they resisted
stellatus) and encrusting coralline algae, where high colonization by other organisms. Sloughing of epithal
temporal variance of aerial exposure either reversed a lial cells and grazing (by limpets) probably combined to
positive effect of mean intensity (barnacles) or deter maintain these algae free of epiphytes, as also reported
mined an impact that did not occur under low temporal in other studies (Steneck 1986). Encrusting corallines
variance (encrusting algae). The original hypothesis that are, however, sensitive to desiccation so that extreme
variance of aerial exposure was more impor conditions of aerial exposure might have reduced their
temporal
tant at high than low levels of mean intensity was cover creating favourable conditions for the coloniza
supported for some response variables (e.g., encrusting tion of filamentous and coarsely branched algae. These
coralline and filamentous algae at sites 1 and 3 and C. latter groups included fast-growing species that were
stellatus and coarsely branched algae at site 3; see also capable of recovering quickly from disturbance (Bene
the regression coefficients but not others
in Appendix), detti-Cecchi and Cinelli 1994, Benedetti-Cecchi 20006)
(e.g., barnacles and coarsely branched algae at site 1 and and a turf-like habit probably enabled them to with
number of taxa at sites 1 and 3). In some cases (e.g., stand periodic aerial exposure by maintaining moist
filamentous effects of variance were also ob conditions.
algae),
served in disturbed and translocated treatments. When Barnacles were numerically dominant high on the
this occurred, comparison of transplanted and trans shore at the study sites, as a result of a combination of
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2497
low recruitment and due to cumulative effects of variance on populations and the
greater mortality (possibly
in the low-shore habitat (Benedetti possibility that large temporal fluctuations in environ
algal overgrowth)
Cecchi et al. 2000). Therefore, it was not surprising that mental variables may affect (increase or decrease) the
the abundance of these organisms increased when cores likelihood with which events coincide with important
were transplanted upshore. The interesting result was periods of reproduction and recruitment of organisms
that such an increase occurred under low, but not under (Bertocci et al. 2005).
variance of aerial exposure. This sug A potential caveat must, however, be considered when
high temporal
that the intervals between consecutive interpreting the results of the present study. By trans
gested long
transplantations imposed by treatment HH in which planting the cores upshore, we did not only increase
cores were in the low-shore habitat, were detrimental to exposure of low-shore organisms to aerial conditions,
young barnacles which, in contrast, could withstand the but alsochanged the biological features of the surround
shorter intervals of submersion determined treatment ing habitat. Climate events would obviously affect only
by
HL. aerial exposure and not the surrounding habitat, at least
of encrusting corallines and barnacles on the short term. Thus, for climate conditions to
Replacement by
the more diversified of filamentous and explain our results, it is necessary to rule out possible
groups coarsely
branched algae produced an increase in diversity in effects due to changes in the strength of biological
treatments to high variance of aerial interactions such as grazing and competition across the
exposed temporal
this effect was more at vertical gradient of the shore. We believe our results
exposure. Although pronounced
site 3, it showed were driven because a
that temporal variance alone, i.e., with largely by physical processes
no concomitant changes in mean intensity, could affect companion experiment (data not shown) indicated that
diversity. This result has broad theoretical and empirical most effects of variance associated with treatment HH
because of diversity are disappeared when the cores transplanted to the high
implications patterns commonly
in relation to changes in either mean shore habitat were masked with plastic screens that
explained intensity
of disturbance (Connell 1978,Mackey and Currie 2001) reduced incident radiation and temperature. In addition,
or as a function of of disturbance neither changes in biological interactions nor variation
frequency (Miller
in intensity across
1982, Collins 2000, McCabe and Gotelli 2000), which of recruitment the vertical gradient of
combines both and variance the shore would explain the observed differences
intensity temporal (Bene
detti-Cecchi 2003). The possibility that temporal var between treatments HL and HH, as these treatments
iance alone can affect of diversity new were placed in the same habitat.
patterns opens
avenues of research. In particular, the A large number of ecological effects have been
understanding
time scales over which variance of disturbance be ascribed to recent shifts in climate conditions, including
might
relevant in relation to the life histories, patterns of changes in physiology (Helmuth et al. 2002), distribu
dispersal and colonizing capabilities of species, seems tion (Hughes 2000, Walther et al. 2002), and diversity
pressing in order to enable more accurate predictions of (Kappelle et al. 1999) of native and invading species.
the consequences of climate on These effects have been largely interpreted in terms of
change global diversity
in mean values of climatic as
(Fields et al. 1993, Barry et al. 1995, Hughes 2000). changes variables such
Our data indicated that the temperature, although some studies have explicitly
temperature high-shore
habitat was more variable than the mid-shore and low acknowledged that changes in the incidence of extreme
shore habitats. in addition to the level of events and in their temporal variability can have
Thus, temporal
variance trans profound effects on ecological systems (Gaines and
imposed experimentally, organisms
to the high-shore habitat were also exposed to Denny 1993, Easterling et al. 2000a). These studies also
planted
natural fluctuations of the thermal environment. recognized the difficulties inherent in interpreting and
larger
Future be to quantify the predicting the effects of variable processes due to lack of
experiments might designed
increase in environmental variance that an organism empirical data. Here we have provided the first
would face in the high-shore to the low-shore unequivocal evidence that mean intensity and temporal
compared
the procedures described and variance of climatic events interactively affected natural
habitat, using by Denny
co-workers et al. assemblages of rocky seashores, suggesting that large
(Denny 2004).
Future studies should also the mech temporal variance may mitigate the ecological impacts
clarify specific
anisms variance affects of ongoing and predicted climate changes in these
whereby assemblages. Although
such mechanisms was the of systems. Because shifts in the mean as well as in the
identifying beyond scope
the present a few variance of climatic variables occur at global scales,
study, possibilities deserve particular
attention. First, variance increase the chance that understanding these effects will be key to predict the
may
fluctuations in environmental variables exceed some ecological and societal consequences of climate varia
physiological threshold for any given organism. This is bility.
analogous to the mechanism proposed to relate the risk Acknowledgments
of extinction of natural populations to the variability of
We thank Fabio Bulleri and two anonymous reviewers for
the environment (Inchausti and Halley 2003). Alter on the manuscript
helpful comments and the various graduate
native, but not mutually exclusive explanations, include and undergraduate students that assisted with the field work.
2498
LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
This study was partially supported by a grant from the Day, K. J., M. J. Hutchings, and E. A. John. 2003. The effects
University of Pisa and was carried out in the frame of the of spatial pattern of nutrient supply on yield, structure and
MARBEF Network of Excellence "Marine Biodiversity and mortality in plant populations. Journal of Ecology 91:541
Ecosystem Functioning," which is funded in the Community's 553.
Sixth Framework Programme (contract no. GOCE-CT-2003 Dayton, P. K. 1971. Competition, disturbance, and community
505446). This publication is contribution number MPS-06028 organization: the provision and subsequent utilization of
of MARBEF. space in a rocky intertidal community. Ecological Mono
graphs 41:351-389.
Literature Cited
Denny, M. W., B. Helmuth, G. H. Leonard, C. D. G Harley,
L. J. H. Hunt, and E. K. Nelson. 2004. Quantifying scale in
Allen, M. R., P. A. J. F. B. Mitchell,
Stott, R. Schnur, and T. L.
Delworth. 2000. the uncertainty in forecast of ecology: lessons from a wave-swept shore. Ecological
Quantifying
climate change. Nature 407:617-620. Monographs 74:513-532.
anthropogenic
Dethier, M. N., S. L. Williams, and A. Freeman. 2005.
Barry, J. P., C. H. Baxter, R. D. Sagarin, and S. E. Gilman.
1995. Climate-related, faunal in a Seaweeds under stress: manipulated stress and herbivory
long-term changes
California affect critical functions.
rocky intertidal community. Science 267:672-675. life-history Ecological Monographs
L. 2000a. Variance in ecological consumer 75:403-418.
Benedetti-Cecchi,
resource interactions. Nature 407:370-374. Easterling, D. R., J. L. Evans, P. Ya. Groisman, T. R. Karl, K.
L. 2000&. Predicting direct and indirect E. Kunkel, and P. Ambenje. 2000#. Observed variability and
Benedetti-Cecchi,
effects succession in a midlittoral shore trends in extreme climate events: a brief review. Bulletin of
during rocky
70:45-72. the American Meteorological Society 81:417^125.
assemblage. Ecological Monographs
L. 2001. Variability in abundance of algae Easterling, D. R., G A. Meehl, C. Parmesan, S. A. Changnon,
Benedetti-Cecchi,
and invertebrates at different scales on rocky sea T. R. Karl, and L. O. Mearns. 2000&. Climate extremes:
spatial
shores. Marine Series 215:79-92. observations, modeling, and impacts. Science 289:2068-2074.
Ecology Progress
Benedetti-Cecchi, L. 2003. The of the variance Fields, P. A., J. B. Graham, R. H. Rosenblatt, and G. N.
importance
around the mean effect size of ecological processes. Ecology
Somero. 1993. Effects of expected global climate change on
84:2335-2346. marine faunas. Trends in Ecology and Evolution 8:361-367.
Benedetti-Cecchi, L., S. Acunto, F. Bulleri, and F. Cinelli. 2000. Gaines, S. D., and M. W. Denny. 1993. The largest, smallest,
of the barnacle Chthamalus stellatus in highest, lowest, longest, and shortest: extremes in ecology.
Population ecology
the northwest Mediterranean. Marine Ecology 74:1677-1692.
Ecology Progress
Series 198:157-170. Gaines, S. D., and J. Roughgarden. 1985. Larval settlement
Benedetti-Cecchi, L., and F. Cinelli. 1994. Recovery of patches rate: a leading determinant of structure in an ecological
in an assemblage of geniculate coralline at community of the marine intertidal zone. Proceedings of the
algae: variability
different successional stages. Marine Series National Academy of Sciences (USA) 82:3707-3711.
Ecology Progress
110:9-18. Grime, J. P., V. K. Brown, K. Thompson, G J. Masters, S. H.
Benestad, R. E. 2003. What can present climate models tell us Hillier, I. P. Clarke, A. P. Askew, D. Corker, and J. P. Kielty.
2000. The response of two contrasting limestone
about change? Climatic
climate Change 59:311-331. grasslands
Bertness, M. D.,and G. H. Leonard. 1997. The role of positive to simulated climate change. Science 289:762-765.
interactions in communities: lessons from intertidal habitats. Hardin, J. W., and J. M. H?be. 2003. Generalized estimating
78:1976-1989. equations. Chapman and Hall/CRC, Boca Raton, Florida,
Ecology
Bertness, M. D., G. H. Leonard, J. M. Levine, and J. F. Bruno. USA.
1999. Climate-driven interactions among rocky intertidal Helmuth, B., C. D. G. Harley, P. M. Halpin, M. O'Donnell, G
a rock and a hot place. Oecologia E. Hofmann, and C. A. Blanchette. 2002. Climate
organisms caught between change
120:446-450. and latitudinal patterns of intertidal thermal stress. Science
Bertocci, I., E. Maggi, S. Vaselli, and L. Benedetti-Cecchi. 2005. 298:1015-1017.
Contrasting effects of mean intensity and temporal variation Hughes, L. 2000. Biological consequences of global warming: is
of disturbance on assemblages of rocky shores. Ecology 86: the signal already apparent? Trends in Ecology and
2061-2067. Evolution 15:56-61.
Butler, M. J., IV 1989. Communityresponses to variable Hutchings, M. J., E. A. John, and D. K. Wijesinghe. 2003.
pr?dation: field studies with and freshwater macro
sunfish Toward understanding the consequences of soil heterogeneity
invertebrates. Ecological
Monographs 59:311-328. for plant populations and communities. Ecology 84:2322
Caley, M. J., M. H. Carr, M. A. Hixon, T. P. Hughes, G. P. 2334.
Jones, and B. A. Menge. 1996. Recruitment and the local Inchausti, P., and J. Halley. 2003. On the relation between
dynamics of open marine populations. Annual Review of temporal variability and persistence time in animal popula
Ecology and Systematics 27:477-500. tions. Journal of Animal Ecology 72:899-908.
Cardinale, B. J., M. A. Palmer, C. M. Swan, S. Brooks, and N. Kappelle, M., M. M. I. Van Vuuren, and P. Baas. 1999. Effects
L. Poff. 2002. The influence of substrate heterogeneity on of climate change on biodiversity: a review and identification
biofilm metabolism in a stream ecosystem. Ecology 83:412 of key research issues. Biodiversity and Conservation 8:1383
422. 1397.
Chapman, M. G. 1986. Assessment of some controls in Kelaher, B. P., A. J. Underwood, and M. G Chapman. 2003.
experimental transplants of intertidal gastropods. Journal Experimental transplantations of coralline algal turf to
of Experimental Marine Biology and Ecology 103:181-201. demonstrate causes of differences in macrofauna at different
Collins, S. L. 2000. Disturbance frequency and community tidal heights. Journal of Experimental Marine Biology and
stability in native tallgrass prairie. American Naturalist 155: Ecology 282:23-41.
311-325. Liang, K.-Y., and S. L. Zeger. 1986. Longitudinal data analysis
Connell, J. H. 1978. Diversity in tropical rain forests and coral using generalized linear models. Biometrika 73:13-22.
reefs: high diversity of trees and corals ismaintained only in a Littler, M. M., and D. S. Littler. 1980. The evolution of thallus
nonequilibrium state. Science 199:1302-1310. form and survival strategies in benthic marine macroalgae:
Davison, I. R., L. E. Johnson, and S. H. Brawley. 1993. field and laboratory tests of a functional form model.
Sublethal stress in the intertidal zone: tidal emersion inhibits American Naturalist 116:25-44.
photosynthesis and retards development in embryos of the Luterbacher, J., D. Dietrich, E. Xoplaki, M. Grosjean, and H.
brown alga Pelvetia fastigiata. Oecologia 96:483-492. Wanner. 2004. European seasonal and annual temperature
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2499
variability, trends, and extremes since 1500. Science 303: Post, E., and M. C. Forchhammer. 2004. Spatial synchrony of
1499-1503. local populations has increased in association with the recent
Mackey, R. L.,and D. J. Currie. 2001. The diversity Northern Hemisphere climate trend. Proceedings of the
disturbance relationship: is it generally strong and peaked? National Academy of Sciences (USA) 101:9286-9290.
Ecology 82:3479-3492. Quinn, G P., and M. J. Keough. 2002. Experimental design
McCabe, D. J., and N. J. Gotelli. 2000. Effects of disturbance and data analysis for biologists. Cambridge University Press,
frequency, intensity, and area on assemblages of stream Cambridge, UK.
invertebrates. Oecologia 124:270-279. Sanford, E. 1999. Regulation of keystone pr?dation by small
Menconi, M., L. Benedetti-Cecchi, and F. Cinelli. 1999. Spatial changes in ocean temperature. Science 283:2095-2097.
and temporal variability in the distribution of algae and Schonbeck, M. W., and T. A. Norton. 1978. Factors controlling
invertebrates on rocky shores in the northwest Mediterra the upper limits of fucoid algae on the shore. Journal of
nean. Journal of Experimental Marine Biology and Ecology Experimental Marine Biology and Ecology 31:303-313.
233:1-23. Sousa, W. P. 1979. Experimental investigations of disturbance
Michener, W. K., E. R. Blood, K. L. Bildstein, M. M. Brinson, and ecological succession in a rocky intertidal algal
and L. R. Gardner. 1997. Climate change, hurricanes and community. Ecological Monographs 49:227-254.
tropical storms, and rising sea level in coastal wetlands. Steneck, R. S. 1986. The ecology of
coralline algal crusts:
Ecological Applications 7:770-801. convergent patterns and adaptative strategies. Annual Re
Miller, T. E. 1982. Community diversity and interactions view of Ecology and Systematics 17:273-303.
between the size and frequency of disturbance. American Steneck, R. S., and M. N. Dethier. 1994. A functional group
Naturalist 120:533-536. approach to the structure of algal-dominated communities.
M?ller, R. A., and G W. Stone. 2001. A climatology of tropical Oikos 69:476-498.
storm and hurricane strikes to enhance vulnerability pre Stephenson, T. A., and A. Stephenson. 1949. The universal
diction for the southeast U.S. coast. Journal of Coastal features of zonation between tide marks on rocky coasts.
Research 17:949-956. Journal of Ecology 37:289-305.
Navarrete, S. A. 1996. Variable pr?dation: effects of whelks on Walther, G.-R., E. Post, P. Convey, A. Menzel, C. Parmesan,
a mid-intertidal successional community. Ecological Mono T. J. C. Beebee, J.-M. Fromentin, O. Hoeg-Guldberg, and F.
graphs 66:301-321. Bairlein. 2002. Ecological responses to recent climate change.
Navarrete, S. A., J. Lubchenco, and J. C. Castilla. 1993. Pacific Nature 416:389-395.
Ocean coastal ecosystems and global climate change. Pages Zavaleta, E. S., M. R. Shaw, N. R. Chiariello, H. A. Mooney,
189-193 in H. A. Mooney, E. Fuentes, and B. I. Kronberg, and C. B. Field. 2003. Additive effects of simulated climate
editors. Earth system responses to global change: contrast changes, elevated C02, and nitrogen deposition on grassland
between North and South America. Academic Press, diversity. Proceedings of the National Academy of Sciences
Orlando, Florida, USA. (USA) 100:7650-7654.
APPENDIX
Detailed results of statistical analyses using population-averaged generalized estimating equations (Ecological Archives E087
151-A1).
Experiments
Author(s): Lisandro Benedetti-Cecchi, Iacopo Bertocci, Stefano Vaselli, Elena Maggi
Source: Ecology, Vol. 87, No. 10 (Oct., 2006), pp. 2489-2499
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/20069260
Accessed: 29/12/2009 13:52
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, 87(10), 2006, pp. 2489-2499
? 2006 by the Ecological Society of America
TEMPORAL VARIANCE REVERSES THE IMPACT OF HIGH MEAN
INTENSITY OF STRESS IN CLIMATE CHANGE EXPERIMENTS
Lisandro Benedetti-Cecchi,1 Iacopo Bertocci, Stefano Vaselli, and Elena Maggi
Dipartimento di Biolog?a, Via A. Volta, 6, 1-56126 Pisa, Italy
Abstract. Extreme climate events simultaneous
produce changes to the mean and to the
variance of climatic variables overecological time scales. While several studies have
investigated how ecological systems respond to changes in mean values of climate variables,
the combined effects of mean and variance are poorly understood. We examined the response
of low-shore assemblages of algae and invertebrates of rocky seashores in the northwest
Mediterranean to factorial manipulations of mean intensity and temporal variance of aerial
exposure, a type of disturbance whose intensity and of occurrence are
temporal patterning
predicted to change with changing climate conditions. Effects of variance were often in the
opposite direction of those elicited by changes in the mean. Increasing aerial exposure at
regular intervals had negative effects both on diversity of assemblages and on percent cover of
filamentous and coarsely branched algae, but greater temporal variance drastically reduced
these effects. The opposite was observed for the abundance of barnacles and encrusting
coralline algae, where high temporal variance of aerial exposure either reversed a positive
effect of mean intensity (barnacles) or caused a negative effect that did not occur under low
temporal variance (encrusting algae). These results provide the first experimental evidence that
changes in mean intensity and temporal variance of climatic variables affect natural
assemblages of species interactively, suggesting that high temporal variance may mitigate the
ecological impacts of ongoing and predicted climate changes.
Key words: aerial exposure; climate change; disturbance; diversity; mean intensity; rocky shore;
temporal variance.
Introduction (Hughes 2000, Zavaleta et al. 2003, Post and Forch
There is increasing concern
hammer 2004), in the ability of assemblages to resist or
among scientists, policy
to recover from other disturbances and in productivity
makers, and the general public about the societal and
(Grime et al. 2000). Most of these studies have, however,
ecological consequences of climate change. Climate
related average values of ecological variables to trends in
events can affect society directly by causing catastrophes
average values of climate variables, little
and by threatening human health and indirectly by offering
into the relative roles of climate means as
altering the functioning of ecological systems (Easterling insights
compared with climate variances and into their inter
et al. 2000?>). Models ofclimate change generally agree
actions. Although some important studies have inves
that extreme events such as droughts, storms, and floods
are becoming more tigated the effects of changing the spatial or the
frequent (Michener et al. 1997, Allen
et al. 2000, Benestad temporal variability of processes such as competition
2003). Changes in the occurrence of
extreme weather conditions produce concomitant (Hutchings et al. 2003), pr?dation (Butler 1989, Nav
to the mean and to the variance of the
arrete 1996), and disturbance (Collins 2000, McCabe
changes
distributions of climatic vari
and Gotelli 2000), there seems to be little effort to
frequency corresponding
ables et al. understand effects of variance in the context of climate
(Easterling 2000?, b). For example, the
incidence of extremely high temperatures results in long change.
term trends in mean values as well as in more or less Recent attention to these issues has led to the
in temporal formulation of a framework for the
pronounced increases variance, depending investigating
on the time scale considered and and combined effects of in mean
(Hughes 2000, M?ller separate changes
Stone Luterbacher et al. intensity and spatial or temporal variance of
2001, 2004). ecological
Shifts in climate conditions can have processes (Benedetti-Cecchi 2000a, 2003, Bertocci et al.
profound
ecological impacts, including changes in patterns of 2005). The experimental designs proposed for these
analyses can provide a useful to inves
distribution, abundance, and diversity of species starting point
tigate the effects of climate change on natural assem
blages. High temporal variance implies that several
Manuscript received 28 November 2005; revised 14 March
climatic events occur over short intervals of time and
2006; accepted 27 March 2006. Corresponding Editor: P. T.
Raimondi. that these alternate with long periods in which no event
1E-mail: occurs
lbenedetti@biologia.unipi.it (Benedetti-Cecchi 2003). Populations of long
2489
2490
LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
lived species may face more extreme physical conditions We used a factorial experiment to test the hypothesis
when climatic events such as
droughts, floods and that variance and mean of aerial
temporal intensity
storms operate with high compared to low temporal exposure would affect of shores
assemblages rocky
variance over ecological time scales. This may be interactively, with variance more im
temporal being
particularly true for less resilient species that recruit portant under high than low mean intensity of exposure.
episodically and/or that have low dispersal capabilities Intensity of aerial exposure was defined as the time
(Gaines and Roughgarden 1985, Caley et al. 1996). In assemblages remained emersed over the study period.
contrast, fugitive species may benefit from high temporal This was manipulated indirectly by transplanting
variance by taking advantage of newly released resour organisms higher on the shore for defined periods of
ces (Dayton 1971). These effects are, however, not time. Temporal variance was defined with reference to
independent of the average intensity of events. It is the interval of time between successive manipulations.
reasonable to expect that variable events that Organisms were moved up and down the shore either at
temporally
operate with on average, have much regular or at variable intervals of time, while
high intensity larger ensuring
effects on natural populations than events that operate that the overall time spent in the high-shore position did
with the same level of variance, but with lower not differ between levels of temporal variance over the
temporal
mean The idea that interactions between duration of the study. Although this study was
intensity.
intensity and temporal variance of climate events may conducted under peculiar tidal conditions and focused
be for major in populations and on a specific aspect of climate it addressed the
responsible changes change,
has never been more general problem of whether the definition of
assemblages investigated formally.
of algae and invertebrates extreme climate conditions in an ecological context
Assemblages inhabiting
intertidal habitats of shores are to should take into consideration both effects of mean
rocky exposed
variable conditions that can become intensity and temporal variance of events.
physical easily
detrimental to their life (Schonbeck and Norton 1978,
Materials and Methods
Bertness and Leonard 1997, Dethier et al. 2005).
Thermal stress due to aerial and disturbance Study sites and experimental design
exposure
by waves are recurrent events that contribute to This was at
study replicated three sites (stretches of
maintain spatial and temporal variability in the struc coast of 30-40 m to
long and hundreds thousands of
ture of these assemblages (Stephenson and Stephenson meters chosen to represent the exposed
apart) randomly
1949, Dayton 1971, Sousa 1979). These effects can, shores south of Livorno, in the northwest
rocky
however, be magnified by global warming and by the Mediterranean. A first site was established in July 2001
increasing frequency and intensity of storms, leading to and two additional sites were established inMay 2002;
quantifiable changes in patterns of abundance, distribu lasted two at all sites. Assemblages on
experiments years
tion and diversity of species (Davison et al. 1993, these shores are described in Benedetti-Cecchi (20006,
Navarrete et al. 1993, Barry et al. 1995, Bertness et al. low-shore were dominated
2001). Briefly, assemblages
1999, Sanford 1999). by encrusting and filamentous algae (including several
While aerial exposure is a predictable event in systems in the and
species genus Ceramium, Polysiphonia,
characterised by large tides, it is less so where tides have and by branched like
Cladophora) coarsely algae
low amplitude. In the northwest Mediterranean, for Laurencia obtusa Lamouroux and Chondria
(Hudson)
example, where tides rarely exceed 30 cm in amplitude, De Toni. In contrast, the barnacle
boryana (J. Agardh)
weather conditions can variable stellatus
impose temporal Chthamalus (Poli) and cyanobacteria (Rivularia
patterns of aerial exposure and desiccation to sessile dominated on the shore. Although assem
spp.) higher
organisms offsetting any effect of the tide. Prolonged differed among habitats, most of the differences
blages
periods of calm sea and high barometric pressure can were due to changes in relative abundance of taxa rather
push the sea level below the mean low water level than in species A notable was
composition. exception
(MLWL) so that organisms remain exposed to air long the red alga Rissoella verruculosa (Bertoloni) J. Agardh,
enough to dry. Rough conditions, in contrast, maintain which was characteristic of the mid-shore habitat
wet even at low tide.
organisms constantly Therefore, (Benedetti-Cecchi 20006).
changes in the timing of occurrence and duration of At each site, 32 cores 10 cm in diameter were drilled
contrasting climatic conditions can have profound out of the rock from the low-shore habitat (0-5 cm
effects on the biota of these shores. Indeed, some above MLWL) using a diamond-tipped corer mounted
climatic models predict changes in the intensity and on a petrol driller (Tanaka America, Auburn, Wash
temporal variance of storms, with strong events USA). The cores, with intact on
ington, assemblages
concentrating in short periods of time separated by top, were then assigned randomly to the following
longer periods of good weather (M?ller and Stone 2001). treatments (defined according to the terminology
These features make Mediterranean rocky seashores a commonly used in transplant experiments [Chapman
valuable system to explore the ecological consequences 1986]), with eight cores per treatment: (1) disturbed
of changes in mean intensity and temporal variance of cores (i.e., cores that were placed back to their original
climatic events. position immediately after drilling), (2) translocated
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2491
cores cores moved to another within the low artefacts associated with moving the organisms to
(i.e., place
in vertical
shore habitat), (3) cores transplanted to the mid-shore another place, irrespective changes of
habitat (15-20 cm above MLWL), and (4) cores position. This was done by swapping positions between
to the high-shore habitat (25-30 cm above cores in the low-shore habitat.
transplanted
MLWL). Within each treatment, four of the eight cores Cores were attached to the substratum with epoxy
chosen at random were manipulated at almost regular putty (Subcoat S, Veneziani S.p.A, Lodi, Italy) at Site 1.
intervals during the study period, whilst the remaining This procedure ensured a firm attachment, but it was
four cores were manipulated at variable intervals. This logistically demanding because it required new drills
created a factorial experiment with two levels of each time the cores had to be moved up or down the
temporal variance crossed with each treatment. Four shore. A different technique was used at sites 2 and 3.
replicate unmanipulated were also established in Two stainless steel bars were anchored in opposite
plots
the low-shore habitat as controls. Thus, the experiment positions along each core with screws inserted into the
consisted the following experimental
of conditions: rock. Each bar had an outward lip in correspondence to
unmanipulated controls (C), disturbed cores (low the top of the core that was used to screw the whole
temporal variance, DL; high temporal variance, DH), structure to the substratum. Regardless of type of
translocated cores (low temporal variance, TL; high manipulation, all cores were positioned into holes in
temporal variance, TH), cores transplanted to the mid such a way that that their top was approximately at the
shore habitat (low temporal variance, ML; high same level of the natural substratum.
temporal variance, MH) and cores transplanted to the How well realized treatment conditions matched
high-shore habitat (low temporal variance, HL; high intended experimental effects depended on local climate.
temporal variance, HH). For example, rough weather could prevent aerial
Intensity of aerial exposure was manipulated indi exposure of assemblages transplanted both to mid-shore
rectly by transplanting the cores to different heights of and high-shore habitats. Similarly, calm sea and high
the shore with respect to MLWL. Positions were chosen barometric pressure might expose all habitats to aerial
to represent distinct environmental conditions on the conditions. Thus, both good and bad weather could, in
basis of our previous knowledge of the system (Menconi principle, eliminate intended differences among treat
et al. 1999, Benedetti-Cecchi 20006, 2001). With calm ments. We addressed this issue in two ways. First, we
sea, the low-shore habitat was exposed to air only at low collated daily data on the direction and speed of winds
tide concomitant with high barometric pressure, the and on barometric pressure for the entire duration of the
mid-shore habitat was always emersed at low tide and study (courtesy of the Istituto Idrografico e Mareogra
the high-shore habitat was emersed both at high and low fico di Pisa). These data served to determine the
tide. Small waves, like those
produced by boats, washed proportion of days intended experimental conditions
both the low-shore and mid-shore habitats, but not the were offset by adverse weather. Conditions in which
high-shore habitat. All habitats were almost continu rough sea might have prevented aerial exposure of both
ously submersed when the sea was rough. mid-shore and high-shore habitats at low tide were
To obtain the conditiontemporal of low variance of defined as those in which the daily average speed of
aerial exposure, cores were transplanted into new holes winds was equal to or larger than 2.5 kn (1.286 m/s), in
either at mid-shore or high-shore heights for a month directions ranging from southeast to north. In contrast,
and then returned back to their original position in the conditions in which all habitats might have been
low-shore habitat. This manipulation was applied exposed to air due to low sea levels were defined as
approximately every three months, so that the overall those in which barometric pressure was equal to or
length of time low-shore assemblages remained in the larger than 570 and 574 hPa in winter and summer,
high-shore position was about four months in a year. respectively (to correct for seasonal changes in average
Bad weather prevented a perfectly regular distribution aerial temperature) and in which the wind was either
of events in time. The condition of high temporal absent or blew seawards (eastern quadrant) at any
variance of aerial exposure was obtained by trans speed.
planting the cores at irregular intervals, but ensuring Second, we two thermom
deployed replicate digital
that the overall time cores were in the high-shore eters (FT-800 system, Econorma Sas, Treviso, at
Italy)
position was still four months per year. This enabled each tidal height in each site, in the proximity of
the independent manipulation of intensity and temporal transplanted cores. The thermometers were housed in
variance of aerial exposure over the time scale of the PVC that were screwed into the rock and
pipes
study (Fig. 1). measured either water or aerial temperature (depending
Disturbed and translocated treatments were used to on whether or not they were submersed) every hour for
assess several potential artefacts associated with these two months. At the end of the recording period the data
manipulations (Chapman 1986, Kelaher et al. 2003). were downloaded to a computer and the thermometers
Disturbed cores controlled for the effects of drilling, were deployed again in the field. These measurements
manipulating and attaching the cores to the substratum. were used as indicators of the status (emersed or
The translocation treatment controlled for possible submersed) of each habitat through time.
2492
LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
A) Site 1
Low variance
ft t t ttt !
High variance
TT
iJiAiSlOtNlDiJ|FlMlAlMiJlJlAlS|OlNlDiJlFlMlA|Ml
??TF T
2001 2002 2003
A A A A A A
B) Sites 2 and 3
Low variance
4W
f T T Iff
High variance
?
IT m tf
iiJ|J|A|S|0|N|D|J|F|M|A|M|J|J|A|S|0|N|D|J|F|M|A||
2002 2003
A A A A A
Fig. 1. Schematic representation of the experimental design. Black and gray blocks indicate periods in which cores were in
high-shore and low-shore positions, respectively. Down-facing arrows indicate periods in which high barometric pressure and calm
sea conditions exposed the low-shore habitat to air, while up-facing arrows indicate periods in which all habitats were submersed
due to rough sea conditions. The temporal extent of these periods in which intended experimental conditions could not be
maintained is indicated by the thickness of the arrows (with thinnest and thickest arrows corresponding to one and five days,
respectively). Only events that coincided with the periods of transplantation are shown. Time of sampling is indicated by the
arrowheads at the bottom of each panel.
Sampling and statistical analyses achievable in the field. Taxa that could not be identified
at the level of species or genus were lumped into
Cores were sampled six times at site 1 and five times at
sites 2 and 3. At each date of morphological groups (Littler and Littler 1980, Steneck
sampling, transplanted
and Dethier 1994).
cores assigned to the two levels of temporal variance
Data were analysed using population-averaged gen
were in different phases with respect to the actual
eralized estimating equations (PA-GEEs), an extension
position on the shore and in relation to the interval of
of generalized linear models (GLMs) (Liang and Zeger
time since the last manipulation. To avoid the problem
1986, Quinn and Keough 2002). PA-GEEs proved useful
of confounding these contingencies with effects of
in the present context for two reasons. First, because the
variance, dates of sampling were chosen in such a way
hypothesis investigated in this study applied to the entire
that the mean interval of time since the last manipu
duration of the experiment and not to single dates of
lation was maintained as much as possible comparable
sampling, an
analysis that focused on population
across treatments (Fig. 1). averages was desirable. Second, because experimental
The number and percentage cover of dominant taxa
units were repeatedly sampled through time, a technique
on experimental cores and in unmanipulated quadrats was
that could account for temporal autocorrelation
were sampled nondestructively with a 7 X 7 cm plastic a link
necessary. PA-GEEs enable the specification of
frame divided into 25 1.4 X 1.4 cm2 sub-quadrats. Size of function and an error structure for the residuals as in
quadrats was dictated by the area available for sampling and also take into account the correlation
GLMs,
on each core and was appropriate to sample small between observations on the same unit.
experimental
organisms like those targeted in the study. Quantitative This results in robust estimates of variances and
data were obtained by recording the number of sub standard errors that can be used in hypothesis testing.
quadrats that contained a particular taxon and express Several of correlation can be specified,
patterns although
ing final values as percentages. Organisms were identi the procedure is robust to misspecification of the
fied to the most detailed level of taxonomic resolution correlation structure (Hardin and H?be 2003). We used
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2493
Table 1. Summary of results from population-averaged generalized estimating equations (PA-GEEs).
Encrusting Filamentous Chthamalus Coarsely
corallines
Diversity algae stellatus branched algae
Contrasts SI S3 SI S3 SI S3 SI
S3 SI
S3
Control vs. others ns ns NS NS
NS ** NS NS NS
T NS NS
NS **
M NS NS
H NS NS NS
* NS NS NS NS
y
T X V NS NS NS NS NS
M X V ns NS NS NS NS
H X V ns NS NS
** **
M vs. T
H vs. T NS
**
M X V vs. T X V NS NS
H X V vs. T X V NS NS
Notes: If disturbed and translocated cores did not differ significantly (translocated effect), then mid-shore and high-shore effects
were examined by contrasting these treatments with the disturbed cores. If translocated and disturbed cores differed significantly,
then mid-shore and high-shore terms were contrasted with the translocated cores. Similarly, if the effect of variance differed
between translocated and disturbed cores (translocated X variance interaction), then the mid-shore X variance and high-shore X
variance interactions were examined by contrasting these terms with the translocated X variance interaction. Full results of analyses
are reported in Appendix A. Abbreviations: SI, site 1; S3, site 3; T, translocated; M, mid-shore; H, high-shore; V, variance; ns, not
significant. Blank cells indicate that the test was not applicable.
* **
P < 0.05; P < 0.01; *** P < 0.001.
a first order autoregressive model AR(1) in all the analysis to check for strong deviations from the
analyses; this choice was motivated by the absence of distributional assumptions and homogeneity of varian
clear seasonal patterns in the raw data, so that ces.
correlation of residuals was expected to decrease as the
Results
time lag between observations increased.
Two types of response variables were analyzed: Meteorological data indicated that rough sea con
number of taxa (used as a surrogate of
diversity) and ditions might have prevented aerial exposure of cores
percentage covers of species or morphological groups. In transplanted to mid-shore and high-shore habitats in 19
the first case, a log-link function was used to relate the out of 235 days of experimental manipulation (8.1% of
expected value of the response variable to the predictor the time) for the condition of low temporal variance and
variables, assuming a Poisson distribution of the error in 15 out of 240 days (6.2% of the time) for the condition
terms. In the second case, a Gaussian distribution was of high temporal variance (up-facing arrows in Fig. 1) at
assumed for the errors and the identity link was used to site 1. Similar values were obtained at sites 2 and 3, with
relate the linear predictor with expected values of cores transplanted to the mid-shore and low-shore
response variables. Tests of hypotheses were based on habitats been submersed by waves in 17 out of 238 days
"treatment" contrasts and compared the unmanipulated of manipulation for the condition of low temporal
condition vs. all other treatments in first place. Trans variance (7.1% of the time) and in 10 out of 241 days for
located cores were then compared to disturbed cores to the condition high temporal variance (4.1% of the time).
detect possible artefacts associated with moving assem The number of days in which all habitats might have
blages from one place to another within the same been exposed to air during the periods in which the
habitat. If this test was not significant, the effects of transplanted cores were upshore were 0 at Site 1 and 10
transplanting cores to the mid-shore and high-shore out of 482 (2.1% of the time) and 24 out of 479 (5% of
habitats were examined by contrasting these treatments the time) for treatments exposed to low and high levels
with the disturbed cores. If translocated and disturbed of temporal variance, respectively (down-facing arrows
cores differed significantly, the mid-shore and high in Fig. 1) at sites 2 and 3. Collectively, these data
shore treatments were contrasted with the translocated indicated that the match between intended and realized
cores. Similarly, if the effect of variance differed between experimental conditions was always above 90%.
translocated and disturbed cores (translocated X var Similar conclusions can be drawn from the temper
iance interaction), then the mid-shore X variance and ature data, as
presented daily average temperatures
high-shore X variance interactions were examined by recorded between 10:00 and 18:00 (Fig. 2). Unfortu
contrasting these terms with the translocated X variance these data were not available for all habitats over
nately,
interaction (these are the mid-shore X variance vs. the entire study period due to themalfunctioning or loss
translocated X variance and high-shore X variance vs. of some thermometers. Nevertheless, measurements
translocated X variance contrasts in Table 1). Plots of indicated that the high-shore habitat was a distinct
residuals vs. predicted effects were examined after each thermal environment more extreme tern
experiencing
2494 LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
35
30
25
O
ICD20
Q.
E
15
10i
Dec Jan Feb Mar Apr May
2003 2004
=
Fig. 2. Temporal trends in daily average temperatures (between 10:00 and 18:00 hours) in low-shore (dotted line, n 3), mid
=
shore (black line, n=\), and high-shore (gray line, n 3) habitats. Originally there were two replicate thermometers in each habitat
on each shore. Missing data are due to the malfunctioning or loss of thermometers.
values and fluctuations than went lost between time 2 and time 4, probably as a
perature larger temporal
the low-shore habitat. These differences reflected the consequence of strong wave action. At site 3, in contrast,
of aerial exposure of the habitat, there were clear negative effects of transplantation on
patterns high-shore
with emersion indicated by lower temperatures in winter mean number of taxa, but only when cores were moved
(December-February) and by higher temperatures in on a regular basis (Fig. 3). While treatment HL suffered
spring (March-May) (Fig. 2). Analysis of data for the a reduction of 32.4% in mean number of taxa
compared
December-February 2003revealed significantly to controls, this effects was only 16% in treatment HH.
period
= ? X
lower mean temperature values (t 2.5, P < 0.05) and These patterns resulted in significant mid-shore
= X variance
significantly larger mean temporal variances (t 9.8, P variance and high-shore interactions in the
< 0.0004) in the high shore compared to the low-shore analysis (Table 1).
habitat (these are one-tailed tests because of the direc Similar buffering effects of variance were observed for
tional nature of the hypotheses; both tests have 4 df and filamentous and coarsely branched algae (Fig. 3). At
variances are homogenous when checked with Coch sites 1 and 2, effects of variance on filamentous algae
ran's C test). Differences between the low-shore and differed between disturbed and translocated treatments,
mid-shore habitats were less clear. The single reading suggesting the occurrence of artefacts (Fig. 3). An effect
available for the mid-shore habitat in the period between of variance-was, however, still evident at site 1 when
March and May 2004 prevented general conclusions cores transplanted to mid-shore and high-shore habitats
about this environment. were to translocated cores. Percentage cover
compared
to mid-shore and high of filamentous algae was
larger in treatments MH and
Assemblages transplanted
shore habitats changed both in terms of number and HH compared to treatments ML and HL, respectively,
percentage cover of taxa from low-shore assemblages while no such difference occurred between TL and TH.
and in several cases effects differed in relation to the These resulted in significant mid-shore X
patterns
vs. translocated X variance and high-shore X
temporal variance of the manipulation (Fig. 3). Tem variance
variance increased the mean number variance vs. translocated X variance contrasts in the
poral significantly
this effect was not to was
of taxa at site 1, although unique analysis (Table 1). A similar effect of variance
transplanted cores, but was also evident on disturbed observed at site 3 in the high-shore habitat (Fig. 3),
and translocated cores (Fig. 3, Table 1). A similar result although the test was not significant (0.05 < P < 0.06).
was observed at site 2 (Fig. 3), although no statistical Also the percentage cover of coarsely branched algae
was undertaken in this case because many cores was drastically reduced on cores transplanted to mid
analysis
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2495
Site 1 Site 2 Site 3
Diversity
6-U
^ 4
?ilMIlil Filamentous algae
C D T M H CD C D T M H
Fig. 3. Values of response variables (top row, no. taxa; other rows, percent cover by taxa) for the different treatments (mean +
se). Means were calculated by first averaging data from each core over time; n = 4 replicate cores except at site 2, where replication
was sparse due to the loss of a large number of cores between the second sampling date and the fourth sampling date; cores that
were lost contributed data only for the first part of the experiment). Abbreviations: C, controls (gray bars); D, disturbed; T,
translocated; M, transplanted to the mid-shore habitat; H, transplanted to the high-shore habitat. Open bars show low variance;
black bars show high variance.
2496 ANDR?
LIS BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
shore (site 2) and high-shore (site 3) habitats at low, but located treatments usually revealed the existence of
not at high variance. The latter effect was statistically intended effects of treatments over and above any effect
significant (Table 1). of the experimental procedure. Effects of treatments
In contrast to filamentous and coarsely branched could vary in magnitude and/or direction among sites,
algae, the percentage cover of encrusting coralline algae indicating that the sequence with which experimental
and Chthamalus stellatus were adversely affected by manipulations were differed between
applied (sequences
temporal variance (Fig. 3). Although disturbed and site 1 and sites 2 and 3, see Fig. 1), intrinsic differences
translocated treatments responded differently to a among sites or a combination of these factors, con
change in variance, as highlighted by significant trans tributed to some of the variability observed in the data.
located X variance interactions were our current
(Table 1), there Much of understanding of ecological
differences between treatments HL and HH that responses to environmental change comes from studies
emerged over and above any possible effect of the that have contrasted conditions in
experimental differing
experimental procedure. A nonsignificant trend was mean values of predictor variables, e.g., C02 (Zavaleta
evident for encrusting coralline algae at site 1 and a et al. 2003), or in which effects of mean and variance are
significant effect was observed at site 3 (contrast H X V varied simultaneously through changes in the frequency
vs. T X V in Table 1).At this site, the effect of variance of events, e.g., disturbance (Collins 2000). Recent studies
in the high-shore habitat was in opposite direction of have shown that changes in spatial or temporal variance
that observed for the translocated treatment in the low of physical and biological processes can have significant
shore habitat (Fig. 3). C. stellatus was similarly impacted effects on species and assemblages under constant mean
by variance when transplanted to the mid-shore (Site 2) conditions (Cardinale et al. 2002, Day et al. 2003,
and to the high-shore (sites 1 and 3) habitats. The most Hutchings et al. 2003). Our results add the almost
dramatic effect occurred at site 3 in the absence of unique evidence that effects of variance may not be
artefacts, as highlighted by the significant high-shore X independent of the mean.
Therefore, conceptual further
variance interaction in Table 1. At site 1, percentage and methodological steps are required in order to
cover of barnacles was lower on cores translocated at improve our understanding of variable
ecologically
high variance than on cores translocated on a regular phenomena, interactions between the mean
including
basis, but this effect of variance disappeared in the mid and the variance of predictor variables (Benedetti
shore habitat resulting in a significant M X V vs. T X V Cecchi 2003).
interaction (Table 1). Finally, no clear effects of treat We predicted that high temporal variance of aerial
ments were observed for Rivularia sp. and articulated exposure would have produced the largest effects when
coralline algae (Fig. 3, Table 1). assemblages were transplanted to the high-shore habitat,
under the most extreme conditions of aerial exposure
Discussion
(treatment HH). In principle, aerial exposure was
Our results revealed interactive effects of mean expected to produce a variety of ecological responses,
intensity and temporal variance of aerial exposure on including both negative and positive effects on abun
assemblages. Effects of variance were often in opposite dance, depending on the ability of organisms to with
direction of those elicited by changes in the mean. stand desiccation and on their capacity to colonize in
Increasing aerial exposure had negative effects both on emersed compared to submersed conditions. Indeed,
diversity of assemblages and on percentage cover of both types of effects were observed in treatment HH and
filamentous and coarsely branched algae, but these patterns appeared related to the life-history traits of taxa
effects were buffered by high temporal variance. The considered. Encrusting corallines were the most abun
opposite was observed for barnacles (Chthamalus dant taxa in the low-shore habitat, where they resisted
stellatus) and encrusting coralline algae, where high colonization by other organisms. Sloughing of epithal
temporal variance of aerial exposure either reversed a lial cells and grazing (by limpets) probably combined to
positive effect of mean intensity (barnacles) or deter maintain these algae free of epiphytes, as also reported
mined an impact that did not occur under low temporal in other studies (Steneck 1986). Encrusting corallines
variance (encrusting algae). The original hypothesis that are, however, sensitive to desiccation so that extreme
variance of aerial exposure was more impor conditions of aerial exposure might have reduced their
temporal
tant at high than low levels of mean intensity was cover creating favourable conditions for the coloniza
supported for some response variables (e.g., encrusting tion of filamentous and coarsely branched algae. These
coralline and filamentous algae at sites 1 and 3 and C. latter groups included fast-growing species that were
stellatus and coarsely branched algae at site 3; see also capable of recovering quickly from disturbance (Bene
the regression coefficients but not others
in Appendix), detti-Cecchi and Cinelli 1994, Benedetti-Cecchi 20006)
(e.g., barnacles and coarsely branched algae at site 1 and and a turf-like habit probably enabled them to with
number of taxa at sites 1 and 3). In some cases (e.g., stand periodic aerial exposure by maintaining moist
filamentous effects of variance were also ob conditions.
algae),
served in disturbed and translocated treatments. When Barnacles were numerically dominant high on the
this occurred, comparison of transplanted and trans shore at the study sites, as a result of a combination of
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2497
low recruitment and due to cumulative effects of variance on populations and the
greater mortality (possibly
in the low-shore habitat (Benedetti possibility that large temporal fluctuations in environ
algal overgrowth)
Cecchi et al. 2000). Therefore, it was not surprising that mental variables may affect (increase or decrease) the
the abundance of these organisms increased when cores likelihood with which events coincide with important
were transplanted upshore. The interesting result was periods of reproduction and recruitment of organisms
that such an increase occurred under low, but not under (Bertocci et al. 2005).
variance of aerial exposure. This sug A potential caveat must, however, be considered when
high temporal
that the intervals between consecutive interpreting the results of the present study. By trans
gested long
transplantations imposed by treatment HH in which planting the cores upshore, we did not only increase
cores were in the low-shore habitat, were detrimental to exposure of low-shore organisms to aerial conditions,
young barnacles which, in contrast, could withstand the but alsochanged the biological features of the surround
shorter intervals of submersion determined treatment ing habitat. Climate events would obviously affect only
by
HL. aerial exposure and not the surrounding habitat, at least
of encrusting corallines and barnacles on the short term. Thus, for climate conditions to
Replacement by
the more diversified of filamentous and explain our results, it is necessary to rule out possible
groups coarsely
branched algae produced an increase in diversity in effects due to changes in the strength of biological
treatments to high variance of aerial interactions such as grazing and competition across the
exposed temporal
this effect was more at vertical gradient of the shore. We believe our results
exposure. Although pronounced
site 3, it showed were driven because a
that temporal variance alone, i.e., with largely by physical processes
no concomitant changes in mean intensity, could affect companion experiment (data not shown) indicated that
diversity. This result has broad theoretical and empirical most effects of variance associated with treatment HH
because of diversity are disappeared when the cores transplanted to the high
implications patterns commonly
in relation to changes in either mean shore habitat were masked with plastic screens that
explained intensity
of disturbance (Connell 1978,Mackey and Currie 2001) reduced incident radiation and temperature. In addition,
or as a function of of disturbance neither changes in biological interactions nor variation
frequency (Miller
in intensity across
1982, Collins 2000, McCabe and Gotelli 2000), which of recruitment the vertical gradient of
combines both and variance the shore would explain the observed differences
intensity temporal (Bene
detti-Cecchi 2003). The possibility that temporal var between treatments HL and HH, as these treatments
iance alone can affect of diversity new were placed in the same habitat.
patterns opens
avenues of research. In particular, the A large number of ecological effects have been
understanding
time scales over which variance of disturbance be ascribed to recent shifts in climate conditions, including
might
relevant in relation to the life histories, patterns of changes in physiology (Helmuth et al. 2002), distribu
dispersal and colonizing capabilities of species, seems tion (Hughes 2000, Walther et al. 2002), and diversity
pressing in order to enable more accurate predictions of (Kappelle et al. 1999) of native and invading species.
the consequences of climate on These effects have been largely interpreted in terms of
change global diversity
in mean values of climatic as
(Fields et al. 1993, Barry et al. 1995, Hughes 2000). changes variables such
Our data indicated that the temperature, although some studies have explicitly
temperature high-shore
habitat was more variable than the mid-shore and low acknowledged that changes in the incidence of extreme
shore habitats. in addition to the level of events and in their temporal variability can have
Thus, temporal
variance trans profound effects on ecological systems (Gaines and
imposed experimentally, organisms
to the high-shore habitat were also exposed to Denny 1993, Easterling et al. 2000a). These studies also
planted
natural fluctuations of the thermal environment. recognized the difficulties inherent in interpreting and
larger
Future be to quantify the predicting the effects of variable processes due to lack of
experiments might designed
increase in environmental variance that an organism empirical data. Here we have provided the first
would face in the high-shore to the low-shore unequivocal evidence that mean intensity and temporal
compared
the procedures described and variance of climatic events interactively affected natural
habitat, using by Denny
co-workers et al. assemblages of rocky seashores, suggesting that large
(Denny 2004).
Future studies should also the mech temporal variance may mitigate the ecological impacts
clarify specific
anisms variance affects of ongoing and predicted climate changes in these
whereby assemblages. Although
such mechanisms was the of systems. Because shifts in the mean as well as in the
identifying beyond scope
the present a few variance of climatic variables occur at global scales,
study, possibilities deserve particular
attention. First, variance increase the chance that understanding these effects will be key to predict the
may
fluctuations in environmental variables exceed some ecological and societal consequences of climate varia
physiological threshold for any given organism. This is bility.
analogous to the mechanism proposed to relate the risk Acknowledgments
of extinction of natural populations to the variability of
We thank Fabio Bulleri and two anonymous reviewers for
the environment (Inchausti and Halley 2003). Alter on the manuscript
helpful comments and the various graduate
native, but not mutually exclusive explanations, include and undergraduate students that assisted with the field work.
2498
LISANDRO BENEDETTI-CECCHI ET AL. Ecology, Vol. 87, No. 10
This study was partially supported by a grant from the Day, K. J., M. J. Hutchings, and E. A. John. 2003. The effects
University of Pisa and was carried out in the frame of the of spatial pattern of nutrient supply on yield, structure and
MARBEF Network of Excellence "Marine Biodiversity and mortality in plant populations. Journal of Ecology 91:541
Ecosystem Functioning," which is funded in the Community's 553.
Sixth Framework Programme (contract no. GOCE-CT-2003 Dayton, P. K. 1971. Competition, disturbance, and community
505446). This publication is contribution number MPS-06028 organization: the provision and subsequent utilization of
of MARBEF. space in a rocky intertidal community. Ecological Mono
graphs 41:351-389.
Literature Cited
Denny, M. W., B. Helmuth, G. H. Leonard, C. D. G Harley,
L. J. H. Hunt, and E. K. Nelson. 2004. Quantifying scale in
Allen, M. R., P. A. J. F. B. Mitchell,
Stott, R. Schnur, and T. L.
Delworth. 2000. the uncertainty in forecast of ecology: lessons from a wave-swept shore. Ecological
Quantifying
climate change. Nature 407:617-620. Monographs 74:513-532.
anthropogenic
Dethier, M. N., S. L. Williams, and A. Freeman. 2005.
Barry, J. P., C. H. Baxter, R. D. Sagarin, and S. E. Gilman.
1995. Climate-related, faunal in a Seaweeds under stress: manipulated stress and herbivory
long-term changes
California affect critical functions.
rocky intertidal community. Science 267:672-675. life-history Ecological Monographs
L. 2000a. Variance in ecological consumer 75:403-418.
Benedetti-Cecchi,
resource interactions. Nature 407:370-374. Easterling, D. R., J. L. Evans, P. Ya. Groisman, T. R. Karl, K.
L. 2000&. Predicting direct and indirect E. Kunkel, and P. Ambenje. 2000#. Observed variability and
Benedetti-Cecchi,
effects succession in a midlittoral shore trends in extreme climate events: a brief review. Bulletin of
during rocky
70:45-72. the American Meteorological Society 81:417^125.
assemblage. Ecological Monographs
L. 2001. Variability in abundance of algae Easterling, D. R., G A. Meehl, C. Parmesan, S. A. Changnon,
Benedetti-Cecchi,
and invertebrates at different scales on rocky sea T. R. Karl, and L. O. Mearns. 2000&. Climate extremes:
spatial
shores. Marine Series 215:79-92. observations, modeling, and impacts. Science 289:2068-2074.
Ecology Progress
Benedetti-Cecchi, L. 2003. The of the variance Fields, P. A., J. B. Graham, R. H. Rosenblatt, and G. N.
importance
around the mean effect size of ecological processes. Ecology
Somero. 1993. Effects of expected global climate change on
84:2335-2346. marine faunas. Trends in Ecology and Evolution 8:361-367.
Benedetti-Cecchi, L., S. Acunto, F. Bulleri, and F. Cinelli. 2000. Gaines, S. D., and M. W. Denny. 1993. The largest, smallest,
of the barnacle Chthamalus stellatus in highest, lowest, longest, and shortest: extremes in ecology.
Population ecology
the northwest Mediterranean. Marine Ecology 74:1677-1692.
Ecology Progress
Series 198:157-170. Gaines, S. D., and J. Roughgarden. 1985. Larval settlement
Benedetti-Cecchi, L., and F. Cinelli. 1994. Recovery of patches rate: a leading determinant of structure in an ecological
in an assemblage of geniculate coralline at community of the marine intertidal zone. Proceedings of the
algae: variability
different successional stages. Marine Series National Academy of Sciences (USA) 82:3707-3711.
Ecology Progress
110:9-18. Grime, J. P., V. K. Brown, K. Thompson, G J. Masters, S. H.
Benestad, R. E. 2003. What can present climate models tell us Hillier, I. P. Clarke, A. P. Askew, D. Corker, and J. P. Kielty.
2000. The response of two contrasting limestone
about change? Climatic
climate Change 59:311-331. grasslands
Bertness, M. D.,and G. H. Leonard. 1997. The role of positive to simulated climate change. Science 289:762-765.
interactions in communities: lessons from intertidal habitats. Hardin, J. W., and J. M. H?be. 2003. Generalized estimating
78:1976-1989. equations. Chapman and Hall/CRC, Boca Raton, Florida,
Ecology
Bertness, M. D., G. H. Leonard, J. M. Levine, and J. F. Bruno. USA.
1999. Climate-driven interactions among rocky intertidal Helmuth, B., C. D. G. Harley, P. M. Halpin, M. O'Donnell, G
a rock and a hot place. Oecologia E. Hofmann, and C. A. Blanchette. 2002. Climate
organisms caught between change
120:446-450. and latitudinal patterns of intertidal thermal stress. Science
Bertocci, I., E. Maggi, S. Vaselli, and L. Benedetti-Cecchi. 2005. 298:1015-1017.
Contrasting effects of mean intensity and temporal variation Hughes, L. 2000. Biological consequences of global warming: is
of disturbance on assemblages of rocky shores. Ecology 86: the signal already apparent? Trends in Ecology and
2061-2067. Evolution 15:56-61.
Butler, M. J., IV 1989. Communityresponses to variable Hutchings, M. J., E. A. John, and D. K. Wijesinghe. 2003.
pr?dation: field studies with and freshwater macro
sunfish Toward understanding the consequences of soil heterogeneity
invertebrates. Ecological
Monographs 59:311-328. for plant populations and communities. Ecology 84:2322
Caley, M. J., M. H. Carr, M. A. Hixon, T. P. Hughes, G. P. 2334.
Jones, and B. A. Menge. 1996. Recruitment and the local Inchausti, P., and J. Halley. 2003. On the relation between
dynamics of open marine populations. Annual Review of temporal variability and persistence time in animal popula
Ecology and Systematics 27:477-500. tions. Journal of Animal Ecology 72:899-908.
Cardinale, B. J., M. A. Palmer, C. M. Swan, S. Brooks, and N. Kappelle, M., M. M. I. Van Vuuren, and P. Baas. 1999. Effects
L. Poff. 2002. The influence of substrate heterogeneity on of climate change on biodiversity: a review and identification
biofilm metabolism in a stream ecosystem. Ecology 83:412 of key research issues. Biodiversity and Conservation 8:1383
422. 1397.
Chapman, M. G. 1986. Assessment of some controls in Kelaher, B. P., A. J. Underwood, and M. G Chapman. 2003.
experimental transplants of intertidal gastropods. Journal Experimental transplantations of coralline algal turf to
of Experimental Marine Biology and Ecology 103:181-201. demonstrate causes of differences in macrofauna at different
Collins, S. L. 2000. Disturbance frequency and community tidal heights. Journal of Experimental Marine Biology and
stability in native tallgrass prairie. American Naturalist 155: Ecology 282:23-41.
311-325. Liang, K.-Y., and S. L. Zeger. 1986. Longitudinal data analysis
Connell, J. H. 1978. Diversity in tropical rain forests and coral using generalized linear models. Biometrika 73:13-22.
reefs: high diversity of trees and corals ismaintained only in a Littler, M. M., and D. S. Littler. 1980. The evolution of thallus
nonequilibrium state. Science 199:1302-1310. form and survival strategies in benthic marine macroalgae:
Davison, I. R., L. E. Johnson, and S. H. Brawley. 1993. field and laboratory tests of a functional form model.
Sublethal stress in the intertidal zone: tidal emersion inhibits American Naturalist 116:25-44.
photosynthesis and retards development in embryos of the Luterbacher, J., D. Dietrich, E. Xoplaki, M. Grosjean, and H.
brown alga Pelvetia fastigiata. Oecologia 96:483-492. Wanner. 2004. European seasonal and annual temperature
October 2006 TEMPORAL VARIANCE OF CLIMATE EVENTS 2499
variability, trends, and extremes since 1500. Science 303: Post, E., and M. C. Forchhammer. 2004. Spatial synchrony of
1499-1503. local populations has increased in association with the recent
Mackey, R. L.,and D. J. Currie. 2001. The diversity Northern Hemisphere climate trend. Proceedings of the
disturbance relationship: is it generally strong and peaked? National Academy of Sciences (USA) 101:9286-9290.
Ecology 82:3479-3492. Quinn, G P., and M. J. Keough. 2002. Experimental design
McCabe, D. J., and N. J. Gotelli. 2000. Effects of disturbance and data analysis for biologists. Cambridge University Press,
frequency, intensity, and area on assemblages of stream Cambridge, UK.
invertebrates. Oecologia 124:270-279. Sanford, E. 1999. Regulation of keystone pr?dation by small
Menconi, M., L. Benedetti-Cecchi, and F. Cinelli. 1999. Spatial changes in ocean temperature. Science 283:2095-2097.
and temporal variability in the distribution of algae and Schonbeck, M. W., and T. A. Norton. 1978. Factors controlling
invertebrates on rocky shores in the northwest Mediterra the upper limits of fucoid algae on the shore. Journal of
nean. Journal of Experimental Marine Biology and Ecology Experimental Marine Biology and Ecology 31:303-313.
233:1-23. Sousa, W. P. 1979. Experimental investigations of disturbance
Michener, W. K., E. R. Blood, K. L. Bildstein, M. M. Brinson, and ecological succession in a rocky intertidal algal
and L. R. Gardner. 1997. Climate change, hurricanes and community. Ecological Monographs 49:227-254.
tropical storms, and rising sea level in coastal wetlands. Steneck, R. S. 1986. The ecology of
coralline algal crusts:
Ecological Applications 7:770-801. convergent patterns and adaptative strategies. Annual Re
Miller, T. E. 1982. Community diversity and interactions view of Ecology and Systematics 17:273-303.
between the size and frequency of disturbance. American Steneck, R. S., and M. N. Dethier. 1994. A functional group
Naturalist 120:533-536. approach to the structure of algal-dominated communities.
M?ller, R. A., and G W. Stone. 2001. A climatology of tropical Oikos 69:476-498.
storm and hurricane strikes to enhance vulnerability pre Stephenson, T. A., and A. Stephenson. 1949. The universal
diction for the southeast U.S. coast. Journal of Coastal features of zonation between tide marks on rocky coasts.
Research 17:949-956. Journal of Ecology 37:289-305.
Navarrete, S. A. 1996. Variable pr?dation: effects of whelks on Walther, G.-R., E. Post, P. Convey, A. Menzel, C. Parmesan,
a mid-intertidal successional community. Ecological Mono T. J. C. Beebee, J.-M. Fromentin, O. Hoeg-Guldberg, and F.
graphs 66:301-321. Bairlein. 2002. Ecological responses to recent climate change.
Navarrete, S. A., J. Lubchenco, and J. C. Castilla. 1993. Pacific Nature 416:389-395.
Ocean coastal ecosystems and global climate change. Pages Zavaleta, E. S., M. R. Shaw, N. R. Chiariello, H. A. Mooney,
189-193 in H. A. Mooney, E. Fuentes, and B. I. Kronberg, and C. B. Field. 2003. Additive effects of simulated climate
editors. Earth system responses to global change: contrast changes, elevated C02, and nitrogen deposition on grassland
between North and South America. Academic Press, diversity. Proceedings of the National Academy of Sciences
Orlando, Florida, USA. (USA) 100:7650-7654.
APPENDIX
Detailed results of statistical analyses using population-averaged generalized estimating equations (Ecological Archives E087
151-A1).