Forgot your password?
Document Actions
Vinueza et al 06
Ecological Monographs, 76(1), 2006, pp. 111–131
2006 by the Ecological Society of America
˜
TOP-DOWN HERBIVORY AND BOTTOM-UP EL NINO EFFECTS ON
´ PAGOS ROCKY-SHORE COMMUNITIES
GALA
L. R. VINUEZA,1,4 G. M. BRANCH,2 M. L. BRANCH,2 AND R. H. BUSTAMANTE3
1Charles Darwin Foundation, The Galapagos Islands, Ecuador
´
2 Zoology Department, University of Cape Town, Rondebosch 7701, South Africa
3CSIRO Marine Research, Cleveland 4163, Australia
Abstract. We evaluated the effects of marine iguanas, sally lightfoot crabs, and fish
on rocky-shore sessile organisms at two sites at Santa Cruz Island, Gala pagos Islands,
´
Ecuador, for 3–5 years during and after the 1997–1998 El Nin o, using exclusion cages to
˜
separate the effects. Plots exposed to natural grazing were dominated either by encrusting
algae or by red algal turf and articulated corallines. Algae fluctuated in response to El Nin o
˜
in the following way. During an early phase, crustose Gymnogongrus and/or red algal turf
were dominant. In the heart of El Nino, grazers had limited effects on algal cover but
˜
influenced algal sizes substantially. Most algae (particularly edible forms) were scarce or
declined, although warm-water ephemeral species (notably Giffordia mitchelliae) flourished,
increasing diversity and overgrowing crusts. Iguana mortalities were high, and crab densities
low. When normal conditions returned, warm-water ephemerals declined, crab densities
rose, and grazers had significant but site-specific effects on algae. At one site, any com-
bination of grazers diminished most erect species, reducing diversity and restoring domi-
nance of competitively inferior grazer-resistant crusts. At a second site, only the combined
effect of all grazers had this effect. Laboratory experiments confirmed that crabs could
control erect algae and promote crustose forms, and crustose Gymnogongrus developed
into an erect form in the absence of crabs. Differences between sites and large-scale temporal
changes associated with El Nino indicate that tropical shores are not all as constant in time
˜
and space as previously suggested. Mobile grazers did affect algal communities, but over
the period of our observations far greater effects were attributable to intersite differences
and temporal shifts in oceanographic conditions. El Nin o events reduce nutrients, intensify
˜
wave action, and raise sea levels, affecting food availability for intertidal herbivores and
their influence on benthic algae. Thus, the dramatic transformations of communities during
El Nino presage the impacts of global climate change.
˜
Key words: algae; bottom-up top-down effects; climate change; ENSO; Galapagos Islands; graz-
´
ing; rocky shores.
INTRODUCTION bivory on community structure relative to (b) changes
Two current focuses of ecology are the effects of in bottom-up effects associated with oceanographic
global climate change and distinguishing the relative shifts.
importance of bottom-up environmental effects versus The effectiveness of grazers can be restricted spa-
the top-down influences of consumers (Menge and tially and temporally by physical factors such as wave
Branch 2001). The Galapagos Archipelago provides
´ action and thermal stress (Dayton 1971, Cubit 1984),
unique circumstances to explore both issues because it habitat complexity or predation (Hockey and Branch
lies at the juncture of three major currents and is par- 1994), or plant defenses (Lubchenco 1980). These pro-
ticularly susceptible to oceanographic shifts associated cesses act locally but can be affected by large-scale
with El Nino. Its intertidal rocky shores are thermally
˜ oceanographic conditions. Thus, changes in tempera-
stressed by the equatorial climate and herbivory ap- ture, nutrient levels or propagule supply associated
pears intense. In this paper, we examine the impact of with patterns of ocean circulation can affect the set-
herbivory over periods of contrasting oceanographic tlement, growth and reproduction of algae and the ac-
conditions during and after the 1997–1998 El Nin o, to
˜ tivities of herbivores and, hence, interactions between
examine (a) the magnitude of top-down effects of her- them (Lubchenco and Gaines 1981). However, if graz-
ing intensifies beyond a threshold, herbivores can de-
Manuscript received 30 December 2004; revised 10 June crease diversity, reducing the landscape to a few grazer-
2005; accepted 15 July 2005; final version received 26 July 2005. resistant species (Paine and Vadas 1969, Vance 1979,
Corresponding Editor: P. T. Raimondi.
4 Present address: Zoology Department, 3029 Cordley Menge et al. 1985, Hixon and Brostoff 1996).
Hall, Oregon State University, Corvallis Oregon 97331 USA. Tropical intertidal systems contain a range of her-
E-mail: vinuezal@science.oregonstate.edu bivore guilds, and there are few predictions as to how
111
112 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
their plant assemblages are organized by the combined between different phases of El Nino, and to test the
˜
effects of numerous herbivores. The barren aspect of generality of previous models accounting for the com-
many tropical latitudes has been attributed to the pres- position of tropical rocky-shore communities. To
ence of diverse consumers (Menge and Lubchenco achieve this, we (1) quantified herbivores, (2) estab-
1981, Menge et al. 1986a, b, Brosnan 1992). In Pan- lished whether differences existed in community com-
ama, intertidal rocky shores are dominated by encrust- position among years, and (3) measured the effects of
ing forms; sessile animals and macroalgae are scarce. grazers on algae by field experiments excluding igua-
Both fast-moving consumers (e.g., fish and crabs) and nas, crabs and fish, and by laboratory experiments with
slow-moving consumers (e.g., limpets) are abundant crabs.
and reduce diversity on open surfaces, maintaining a
consistent community because their effects are impor- MATERIALS AND METHODS
tant year-round (Menge and Lubchenco 1981, Lub-
chenco et al. 1984). In Hong Kong, the effect of con- Study sites and conditions
sumers diminishes during summer because of reduced The field experiments were replicated at two inter-
food availability and thermal stress (Williams 1993, tidal, south-facing, gently sloping, lava reefs in Acad-
1994, Kennish et al. 1996). In Brazil, lush foliose algae emy Bay, Santa Cruz Island, Ecuador. Site 1 lay in
and zoanthids cover the low intertidal, despite physical front of the Charles Darwin Research Station Marine
conditions there being comparable to those in Panama. Laboratory on the outer, wave-exposed face of a lava
The role of consumers is inconsistent and sometimes bench forming an embayment (0 44 37 S, 90 18 18
crabs and fish even enhance the growth of Ulva. Bare W). Site 2 was located at La Ratonera, 200 m to the
space is rare and competition between sessile forms east in front of the meteorological station (0 44 46 S,
often important (Sauer Machado et al. 1992, 1996). 90 18 09 W). Site selection was based on (a) logistic
Thus, there is conflicting evidence of the role of accessibility, as the experiments had to be monitored
herbivory on tropical rocky shores. Generalizations regularly over a prolonged period, and (b) replication
have been based on a limited geographic coverage, and
at sites that were sufficiently close by as to be regarded
seldom integrate stochastic environmental changes and
as physically comparable, but far enough apart to be
their possible modulation of herbivory (Brosnan 1992).
independent. Both sites were subjectively rated as
Rocky shores in Galapagos are characterized by a
´
semi-exposed, but measurements with dynamometers
black, basaltic substratum of volcanic origin that reach-
(Palumbi 1984) revealed that Site 1 was slightly more
es temperatures as high as 60 C. Thermal stress strong-
exposed than Site 2 (experiencing maximum wave forc-
ly affects community structure, but its influence differs
es of 11.45 2.32 N and 9.11 2.84 N, respectively;
at small scales related to intertidal height, wave ex-
ANOVA, F1.42 18.19, P 0.001). Densities of igua-
posure, and relief (Wellington 1984). In addition, three
nas and crabs and standing stocks of algae were mea-
main currents converge on the archipelago, and can
sured at these two sites and at four sites at Punta Nun ez,
˜
shift in position and intensity, resulting in fluctuating
gradients of temperature and productivity. Even larger- the latter being selected to span the spectrum of algal
scale disturbances are associated with El Nin o-South-
˜ availability during the El Nino (see Appendix A for
˜
ern Oscillation (ENSO) events, which occur at intervals map of sites). The tides in Galapagos are semidiurnal,
´
of 3–7 yr and are characterized by unusually warm spanning 1.8–2.5 m (Houvenaghel and Houvenaghel
waters, a rise in sea level, greater wave action, and a 1977). Our studies were confined to the two lower-most
depletion of nutrients that reduces primary productivity intertidal zones on the shore, which Wellington (1975)
and ultimately affects all trophic levels (Cane 1983, termed the midlittoral (0.5–1.5 m) and the infralittoral
Houvenaghel 1984, Glynn 1988, 1990, Laurie 1990, zones (0.25–0.5 m above chart datum).
Bustamante et al. 2002). This dynamic environment The 1997–1998 ENSO began in March 1997, when
creates an ideal scenario for the study of top-down sea surface temperature (SST) rose to 3–5 C warmer
influences of herbivores on plant assemblages against than normal from about May 1997 to May 1998 before
a background of environmental stochasticity that alters declining precipitously, and from September 1998 con-
nutrient inputs and thus bottom-up effects. ditions returned to being close to the long-term norm
Our work reports on a series of observations and (see Appendix B for temperature records). Detailed
experiments conducted on rocky intertidal shores in studies of nutrient concentrations were independently
Galapagos, to examine the influence of fast-moving
´ undertaken at the time of our studies (Chavez et al.
herbivores on algal assemblages and to compare com- 1999), so we did not measure nutrients.
munity structure during and after El Nin o. The exper-
˜ We began observations in June 1997, initiated our
iments ran throughout the 1997–1998 ENSO event and caging experiment in August 1997, and ran the caging
for a year after its passage, and monitoring of control experiment for a year in which sea temperatures
plots continued to February 2003. reached their greatest departures from normal (termed
Our goals were to assess the relative influences of the ‘‘El Nino period’’), and the following year when
˜
herbivory by mobile species and temporal differences temperatures were again normal (‘‘post-El Nin o’’). ˜
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 113
FIG. 1. Design of the experimental treatments to test the effects of grazers. The notation indicates grazers that were
excluded ( ) or could gain access ( ). ‘‘Crabs’’ refers to Grapsus grapsus; ‘‘some’’ indicates that only a portion of that
group of grazers would have gained access.
Monitoring of control sites continued for a further 4.5 centage of composition of all identifiable items and
yr of ‘‘normal’’ conditions. estimating the total volume. Gut fullness was calcu-
lated as ([volume of gut contents/volume of stomach]
Monitoring densities of consumers 100).
Densities of marine iguanas and sally lightfoot crabs Selectivity of G. grapsus for different functional
at Sites 1 and 2 were censused once every three months groups of algae was assessed by comparing their con-
during low spring tides from August 1997 to September tribution to the diet with their contribution to the per-
1999 and again in July 2001, five instantaneous counts centage of cover in control plots during weeks 2, 4,
being made in areas of 40 5 m at 15-min intervals. and 8 of the caging experiment (when diet was as-
Counts were made with binoculars at a distance of 100 sessed). Selectivity was measured by the odds ratio, O
m to avoid disturbing the animals. Fish were censused p1q1/p2q2, where p1 the percentage of diet con-
during high tide by snorkeling a 15 2 m transect tributed by a particular taxon, p2 the percentage of
parallel to the shore, at seven roughly equally spaced that taxon available in the environment, q1 the per-
times between April 1998 and April 1999. Logistic centage of diet composed of all other taxa, and q2
constraints prevented replication within sites. Seden- the percentage of all other taxa available in the envi-
tary grazers (e.g., chitons, winkles, and limpets) were ronment. Expressed as ln(O), this yields positive values
always scarce and were not a central focus of our study. for items that are selectively consumed relative to their
To quantify their abundance, they were surveyed once abundance, and negative values for those selected
in June 2001 in 10 1-m2 quadrats in each of three shore against.
heights at each site (0.5, 1.0, and 2.0 m above chart
datum). The first two of these shore heights overlapped Densities of iguanas and crabs relative
with the zone in which our experiments were done. We to algal standing stock
surveyed the uppermost zone in case there were species Densities of iguanas and the crab G. grapsus were
that could migrate down into the low shore during high assessed in August 1997 at Sites 1 and 2 and at four
tide. In calculating the sedentary herbivore abundance sites 100 m apart at Punta Nunez, from instantaneous
˜
and biomass, we used only the data from the two lower scans of five areas of 10 m2 in each site. Ash-free dry
zones. To approximate biomasses, densities were mul- mass (AFDM) of algae was determined from five 100-
tiplied by the estimated wet masses of each species. cm2 samples per site, scraped to bedrock and ashed at
Estimates for fish were obtained from Fishbase (avail- 360 C for 12 h.
able online),5 and averaged 91.05 g per herbivorous
Caging experiment
fish. For iguanas we used 883 g for small individuals
and 1200 g for large individuals feeding in the intertidal The caging experiment was set up at Sites 1 and 2
zone (Wikelski et al. 1993). Direct measurements yield- in the low intertidal at heights that were equivalent at
ed values for Grapsus grapsus (mean 24 g), Pachy- both sites, spanning 0.3–0.8 m above chart datum, and
grapsus transversus (mean 0.8 g), and sedentary her- the positions of different treatments were randomly as-
bivores (4.8 g). signed within this height range. The cages and fences
were designed to differentially exclude iguanas, G.
Crab diet grapsus, and fish to test their relative effects on algal
The diet of G. grapsus was determined for 30 in- composition and abundance. Each treatment plot was
dividuals at each experimental site. Stomach volumes 30 30 cm and was surrounded by an iron fence,
were measured by displacement and the stomach con- totally or partially covered by stainless steel netting of
tents then extracted, spread on a slide, and compressed mesh size 1.0 1.0 cm (Fig. 1), with five randomly
to a thickness of 0.5 mm before measuring the per- interspersed replicates per site. Pilot studies indicated
that 96% of the diversity of the shore was incorporated
5 www.fishbase.org in five replicate plots of this size. The total exclosure
114 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
treatment (T) excluded all grazers larger than 1 cm. and the mean lengths of the upright branches measured
The half wall roof (H) left a gap too small for iguanas (n 5 independent replicates).
and fish to enter but allowed crabs access; only crabs
were ever seen in these plots. The wall treatment (W) Univariate data analyses
effectively excluded iguanas, but fish and at least some
For univariate analyses, the percentage of cover of
crabs could enter it. The roof treatment (R) constituted
sessile organisms was arcsine transformed for the pur-
a procedural control that theoretically allowed all graz-
poses of ANOVA to meet assumptions of normality
ers access while shading the plot like other roofed ex-
and equality of variance. There were no differences in
clusion plots. All three types of mobile grazers were
the abundance of sessile organisms among treatments
observed in the roof controls, and no significant dif-
at the beginning of the experiment in August 1997
ferences were recorded between the responses of algae
(two-way ANOVA with site as a random factor and
in them and in the controls (C) for four of the seven
functional groups examined. In the remaining three, treatment as a fixed factor; P 0.05). Because the data
differences emerged in only two to four instances (out for successive measures of each plot were not inde-
of 18 sets of observations). Thus, the roof is unlikely pendent, a repeated-measures (RM) ANOVA (using
to have had serious shading effects. None of the treat- SYSTAT 9; Systat Software, Port Richmond, Califor-
ments excluded the crab Pachygrapsus transversus nia, USA) was used to test the effects of site, treatment,
(mean densities 10–20/m2) or small blennies, which and temporal variation on algal abundance, with site
could pass through the mesh. We assumed that their and treatment (between subjects) as random and fixed
effect was similar in all treatments. No sedentary ben- factors, respectively. Time was analyzed as a within-
thic grazers were ever recorded in any of the treatments. subject effect. This model assessed the percentage of
The percentage of cover of the following functional variation ascribable to each of the factors (and their
groups of macroalgae and sessile invertebrates was interactions). Mauchly’s tests were run for all RM AN-
scored every three months in quadrats with 400 inter- OVAs. Adjusted degrees of freedom were applied to
secting grid points: algal crusts (Gymnogongrus and the F tests (using Greenhouse-Geisser and Huynh-Feldt
Lithothamnium), foliose greens (Ulva and Enteromor- corrections) for data sets that did not meet the sphe-
pha), filamentous greens (mostly Chaetomorpha an- ricity assumptions (Kuehl 2000). Post-hoc contrasts,
tennina), red algal turf (intermingled species forming using Bonferroni/Dunn adjusted probabilities (P
a dense turf 10 mm tall), filamentous brown algae 0.005 for significant differences) identified differences
(notably Giffordia mitchelliae), articulated corallines among the main effects of treatments on each sampling
(mainly Jania spp.), the anemone Isoactinia sp., and date. Successive measurements of frond lengths were
the barnacle Tetraclita milliporosa. Frond lengths of independent because the chance of sampling fronds
Ulva, red algal turf, filamentous greens, and articulated coming from the same plants or the same fronds on
corallines (n 5 independent replicates per plot) were successive dates was remote. The data were also normal
measured monthly initially and then once every three and had equal variances; so two-way ANOVA was ap-
months for two years. plied to untransformed data for each site, with time and
treatment as variables.
Aquarium experiments on the effects Diversity (Shannon-Weiner H ), species richness
of crabs on algae (Margalef’s d ), and evenness (Pielou’s J ) were based
The effects of G. grapsus on algal communities were on percentage of cover of functional groups, using
additionally determined in aquarium experiments. PRIMER version 5.2.2 (PRIMER-E, Roborough, UK).
Twelve rocks (upper surface areas 725 cm2) were To test the effect of grazing, site, and time on these
prized from the low shore adjacent to Site 1, installed measures, repeated measures ANOVA was employed,
in separate cages (50 40 25 cm with a 2-mm mesh), using the procedure described for percentage of cover.
and held in outdoor aquaria with a continuous supply Differences between ENSO (from week 16 to week
of ambient seawater. Water depth was adjusted to sim- 40), post-ENSO (from week 52 to week 100), and nor-
ulate twice-a-day tides that left the upper half of each mal (only controls, weeks 195 and 277) conditions were
rock exposed for 2 h coincident with natural low tides. tested using ANOVA and post-hoc Bonferroni/Dunn
One crab was added to each of six cages; the remaining adjusted probabilities contrasts. The results must be
cages lacked crabs. The rock area per crab was equiv- interpreted with caution, given the possibility of a type
alent to 10 crabs/m2, the highest density recorded in I error, as normality and homogeneity of variance were
the field. not always observed (Bartlett’s and Cochran’s C test).
The upper surfaces of the rocks were monitored on However, provided the design is balanced, ANOVA is
days 0, 3, 6, 16, 27, and 33 to compare the changes in not generally affected when these assumptions are not
cover of encrusting versus erect forms. During the ex- met (Day and Quinn 1989), and graphical inspection
periment, the normally crustose Gymnogongrus devel- of the data (Quinn and Keough 2002) indicated that
oped into an erect stage with upright knobbly branches. the heterogeneity of variance was not sufficient to com-
Cover of the crustose and erect phases was estimated, promise the ANOVAs.
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 115
Multivariate analysis
Community structure within each treatment was an-
alyzed by averaging the percentage of cover data for
the five replicates of each treatment for each sampling
date; the data were then double root transformed and
analyzed by Bray Curtis similarity and nonmetric mul-
tidimensional scaling (MDS). Clusters recognized in
the MDS were based on 80% similarity cutoff in the
Bray Curtis similarity. SIMPER (similarity percentage)
was used to calculate the contributions of each func-
tional group to dissimilarities among clusters.
RESULTS
Consumers
On average, 0.10 0.02 marine iguanas/m2 were
recorded at both sites (mean SD). Densities fluctu-
ated, but tended to decline until near the end of the
observations (Fig. 2). Densities of G. grapsus foraging
during low tide averaged 0.44 0.10 individuals/m2
at Site 1, and 0.84 0.21 individuals/m2 at Site 2. Both
sites initially had similar densities of G. grapsus (0.4
individuals/m2), followed by a decline until September
or December 1998. Substantial increases occurred after
the El Nino period, mainly due to the appearance of
˜
juveniles, with numbers at Site 2 rising to double those
at Site 1 (Fig. 2).
G. grapsus consumed mainly red algal turf and Ulva,
but also erect thalli of Gymnogongrus, and small
amounts of filamentous greens and articulated coral-
lines (Fig. 3A). Differences between the sites reflected
relative availability of algae. Filamentous brown algae
FIG. 2. Densities of marine iguanas Amblyrhynchus cris- were virtually absent from both diet and habitat at Site
tatus (upper panel) and sally lightfoot crabs Grapsus grapsus
(lower panel) at Sites 1 and 2. 2, and articulated corallines were near-absent at Site 1
(Fig. 3B). The selectivity indices (Fig. 3C) indicated a
strong preference for red algal turf and Ulva. Erect
thalli of Gymnogongrus were positively selected when
present. The crabs displayed relatively neutral selec-
tivity for articulated corallines and filamentous green
FIG. 3. (A) Percentage composition of gut contents of Grapsus grapsus, (B) availability of algal functional groups, and
(C) selectivity indices indicating preference for particular functional groups (positive values) or avoidance of them (negative
values). In (A) and (B), values are means, and error bars indicate SE.
116 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 4. Densities of herbivorous fish, iguanas, sally lightfoot crabs, and sedentary herbivores compared across times
(mean 95% CI), and approximate conversions to whole wet biomass (g/m2).
algae, and avoided filamentous brown algae, encrusting space appeared intense, unoccupied rock averaging
Gymnogongrus and encrusting corallines. only 0.9%. Site 1 was dominated by crustose Gym-
Nineteen fish species were recorded, including eight nogongrus (40–85% of cover) and red algal turf (ini-
herbivores and three omnivores (see Appendix C for tially 28%). Site 2 was at first dominated by red algal
species list and densities). Numerically, fish were the turf (approximately 75%). Gymnogongrus was scarce
most abundant mobile herbivores (collectively aver- there at the start, but rose to 50–70%.
aging 1.4 individuals/m2), with sally lightfoot crabs In the unmanipulated control plots, three types of
averaging 0.64 individuals/m2 and iguanas 0.10 indi- temporal change occurred (Fig. 5). First, encrusting
viduals/m2 (Fig. 4) Gymnogongrus, Ulva, and Enteromorpha declined or
Twenty-three species of sedentary benthic consum- were scarce during El Nino, and then became more
˜
ers occurred at the two sites, but none was ever abun- abundant. Second, several groups were most abundant
dant (see Appendix D for details of species and den- during the El Nino. Giffordia mitchelliae was absent
˜
sities). Six species of herbivores occurred in the zones initially but increased to 40–60% cover, replacing the
previously dominant crustose Gymnogongrus, and then
where we worked, but were sparse. Chitons and limpets
disappeared. At Site 2, red algal turf, Chaetomorpha
were rare (0.012 individuals/m2). Sea urchins never
antennina, and articulated corallines also peaked dur-
grazed in the intertidal zone. Thus, sedentary herbi-
ing the El Nino. The barnacle Tetraclita milleporosa
˜
vores were scarce (collectively 0.6 individuals/m 2) declined from 14% to 2% in the latter half of the
but mobile consumers were abundant (totaling 2.14 in- study. Encrusting corallines covered 29% early in the
dividuals/m2). Biomass estimates emphasize the scar- El Nino, but declined to 3% and began to recover
˜
city of slow-moving benthic herbivores (Fig. 4). only three years after the El Nino. Third, erect Gym-
˜
The densities of both crabs and iguanas (y, individ- nogongrus and the anemone Isoactinia displayed no
uals/m2) were linearly related to algal standing stocks temporal trends (neither ever exceeding 3% cover).
(x, g AFDM/cm2) when the two sites in Academy Bay The most obvious changes in cover occurred during
and the four sites at Punta Nunez were examined. (For
˜ the El Nino, when G. mitchelliae and C. antennina
˜
iguanas, y 1.636x 0.189; n 6, r2 0.986; for increased and crustose forms diminished. Once tem-
crabs, y 6.848x 0.496; n 6, r2 0.956.) The peratures dropped, Ulva and Enteromorpha replaced
gut fullness index for crabs was 83.3 11.7 (mean G. mitchelliae and C. antennina, rising to cover 48%
SE ) for Site 1, and 50.0 8.7 at Site 2, the differences of primary space; and encrusting Gymnogongrus re-
being significant (t test; t 27.1, P 0.001), sug- covered to become the eventual dominant. Bare space
gesting that food was more difficult to obtain at Site accounted for up to 20% during the El Nino but was
˜
2, despite algal biomass being higher there. almost nonexistent thereafter.
From week 76 onward, almost all of the functional
Natural changes in benthic community structure groups changed scarcely at all in the control plots dur-
Abundances of functional groups were initially com- ing the 4.5 yr of monitoring after El Nino passed.
˜
parable among treatments (ANOVA, df4,40, P 0.05 Changes in community structure after
in all cases). Differences did exist between sites ( P herbivore manipulation
0.001), but there were no significant interactions be- Temporal, site, and grazer effects significantly af-
tween treatment and site (P 0.05). Competition for fected the abundance of all functional groups except
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 117
Lithothamnium (Fig. 5). Temporal differences usually clined over time; but neither the main effects nor their
had the greatest effect, accounting for 21–37% of the interactions had any influence on abundance.
variability. Site effects were also substantial (16–32%), The effects of grazer treatments on all functional
and grazer treatments less important (1–10%), but there groups were clear when the data were contrasted be-
were significant interactions in most cases (see Ap- tween the El Nino and post-El Nino periods (Fig. 6),
˜ ˜
pendix E for details of statistical analyses). and most obvious (and significant) at whichever site
Percentage of cover of encrusting Gymnogongrus supported the highest cover of each group. Four types
differed significantly between El Nin o and post-El Nino
˜ ˜ of responses emerged. (1) Encrusting Gymnogongrus
periods, increasing in the latter (Fig. 5A). At Site 1, was promoted by grazing and suppressed during the El
only total exclusion of all grazers led to a decrease in Nino. (2) Chaetomorpha, Ulva, and probably Entero-
˜
this crust. At Site 2, both partial and total exclusions morpha were inhibited by grazing, with Chaetomorpha
of grazers had this effect. Cover rose to 79% in the peaking during El Nino and the other two after its pas-
˜
controls, but only 6–26% in all other treatments. Over- sage. (3) Red algal turf and articulate corallines yielded
all, encrusting Gymnogongrus declined or was held at ambiguous results, with a suggestion of higher values
low levels during the El Nino period (in all treatments)
˜ at intermediate levels of grazing, at least at Site 2 where
but increased thereafter, most obviously in treatments both were most abundant. El Nino promoted red algal
˜
involving no exclusion or partial exclusion of grazers. turf but had no effect on articulated corallines. (4) En-
Ulva and Enteromorpha responded oppositely to crusting corallines and G. mitchelliae showed no con-
Gymnogongrus, with abundance peaking inside the to- sistent responses to grazing but both peaked during the
tal exclusion treatments (Fig. 5B and C). Site and tem- El Nino.
˜
poral differences were large, and grazing effects sig-
nificant but of lesser magnitude. At both sites, total Effects of grazers on algal size composition
exclusion of grazers increased the cover of Ulva during Differences in frond length were initially nonexistent
the post-El Nino period from week 52 onward (Fig.
˜ among treatment plots (ANOVA, P 0.25 in all cases),
5B), and control plots had low values. Enteromorpha but soon emerged and persisted for 8–52 wk (Fig. 7;
was only abundant at Site 1, where it responded neg- for detailed data and ANOVA, see Appendix F).
atively to grazing (Fig. 5C). In short, Ulva and Enter- Frond lengths of Ulva (Fig. 7A) were significantly
omorpha remained scarce in all treatments during the affected by grazing (P 0.001), being inversely related
El Nino period, but thereafter increased in treatments
˜ to grazing intensity most of the time at both sites. Red
where grazing was diminished or prevented. algal turf (Fig. 7B) showed similar clear-cut responses
No consistent grazing effect was observed for Gif- to grazing for the first 40 wk at Site 1 (P 0.001) and
fordia mitchelliae, with opposite trends at the two sites for 16 wk at Site 2. Chaetomorpha antennina (Fig. 7C)
(Fig. 5D). Most of its variation was explained by time showed striking effects of grazing up to weeks 40–52.
(60%), with treatment, site and interactions between Articulated corallines (Fig. 7D) showed no obvious pat-
factors explaining no more than 7%. Its proliferation terns, although roofed treatments (roof and half wall
was thus clearly linked to El Nino. ˜ roof) had high values in weeks 4–28, similar to the
Red algal turf was initially an important component, pattern for percentage of cover (Fig. 5H).
but progressively declined, and levels were signifi- Time and site always had significant effects, as did
cantly lower post-El Nino than during El Nino (Fig.
˜ ˜ interactions between factors. The percentage of vari-
5E). Grazing had no effect (P 0.05). Thus, cover of ation in the model significantly explained by grazing
red algal turf was more abundant at Site 2 than Site 1, (0.2–9.9%) was always less than that attributable to
and more abundant during the El Nino than afterward,
˜ time (21.0–59.9%) or site (0.3–32.3%). Without ex-
but was virtually unaffected by grazing. ception, the largest responses to grazing were linked
Chaetomorpha antennina (Fig. 5F) did not differ to periods when each taxon was most abundant.
among treatments or sites. Temporal effects dominated,
with Chaetomorpha only being abundant during El Algal diversity
Nino.
˜ There were initially no differences in any of the in-
Articulated coralline algae were virtually absent dices of diversity among treatments. By week 16, how-
from Site 1 (Fig. 5G). At Site 2, grazing had a small ever, diversity, evenness, and richness had increased at
and marginally significant effect. Percentage of cover both sites due to the addition of species such as Gif-
was initially small, but increased as the El Nin o ma-
˜ fordia mitchelliae and Chaetomorpha antennina. After
tured, particularly inside the treatments that provided about weeks 40–52, all three indices progressively de-
partial protection against grazing (roof or half wall clined at both sites and in all treatments, mainly due
roof), and declined post-El Nino. Grazing intensity was
˜ to increased dominance by encrusting Gymnogongrus
not correlated in any simple manner with percentage at Site 1 and by Ulva at Site 2, and because of declines
of cover, with the controls and total exclusion plots in red algal turf, barnacles, and Lithothamnium.
(highest and lowest intensities of grazing) often having Site 1 had less species richness than Site 2 (repeated-
the lowest cover. Encrusting corallines (Fig. 5H) de- measures ANOVA, P 0.001; see Appendix G for
118 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 5. Changes in algal cover (mean SE ) in experimental plots at Sites 1 and 2, August 1997–September 1999, and
in control plots, August 1997–February 2003: (A) encrusting Gymnogongrus sp., (B) Ulva sp., (C) Enteromorpha and
associated diatoms, (D) Giffordia mitchelliae, (E) red algal turf, (F) Chaetomorpha antennina, (G) articulated corallines, and
(H) Lithothamnium sp. The significance of main effects and their contributions to percentage variance pooled across sites
(% var) are shown on the right (see Appendix E for statistical analyses.) Solid horizontal bars indicate times when grazing
effects were significant; dashed bars when they were not (post hoc contrasts; Bonferroni/Dunn adjusted to P 0.005).
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 119
FIG. 5. Continued
120 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 6. Grazing effects (G) on the percent cover of algae comparing ENSO (December 1997–June 1998, solid circles)
vs. post-ENSO (September 1998–September 1999, open circles) conditions (E), and the interaction between grazing and
ENSO effects (GE) (ANOVA, *P 0.05, ** P 0.01, *** P 0.001; NS, P 0.05) at Sites 1 and 2. Letters on the x-
axis represent different treatments (T, total exclosure; H, half wall roof; W, wall; R, roof; and C, control). Within sites,
treatments that share the same lowercase letter (just above the x-axis) were not significantly different (Bonferroni post hoc
tests, P 0.05). Where interactions are reported, arrows indicate the period when differences in grazing effect were found.
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 121
FIG. 7. Differences in the lengths (mean and SE) of algal fronds: (A) Ulva sp., (B) red algal turf, (C) filamentous green
algae (mainly Chaetomorpha antennina), and (D) articulated corallines (which were absent from Site 1). Numbers next to
data points indicate values for truncated portions of the graphs.
statistical analyses and Appendix H for temporal pat- case) were significantly affected by grazing, being in-
terns of diversity between treatments); but diversity versely related to grazing intensity (Fig. 8).
and evenness did not differ between sites (P 0.05).
Grazing affected all three diversity indices (P 0.001). Multivariate analysis of community structure
Total exclusion plots usually had high richness and Treatments were classified into Bray-Curtis hierar-
diversity values compared to other treatments ( P chical clusters and by MDS (see Appendices I and J),
0.005; Fig. 8). The controls (which were most intensely based on the mean percentage of cover of functional
grazed) yielded consistently low values, showed the groups in each treatment on each date. There was a
greatest declines post-El Nino, and had the lowest final
˜ clear pattern dividing the samples into temporal clus-
values. ters, rather than clusters based on treatments. For Site
Overall, contrast analyses showed that all three in- 1, seven clusters emerged, all distinguished by the time
dices were significantly greater during the El Nin o (par-
˜ when samples were taken rather than by treatment. A
ticularly in total exclusion plots), and (in all but one similar predominantly temporal pattern occurred at Site
122 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 8. Grazing effects (G) on species richness (Margalef’s index d ), evenness (Pielou’s evenness index J ) and diversity
(Shannon-Wiener diversity index H ) comparing ENSO (December 1997–June 1998, solid circles) vs. post-ENSO (September
1998–September 1999, open circles) conditions (E) and the interaction between grazing and ENSO effects (GE) (ANOVA,
*P 0.05, ** P 0.01, *** P 0.001; NS, P 0.05). Diversity indexes were derived from percent cover of functional
groups in experimental plots at Sites 1 and 2. Where interactions are reported, arrows indicate the period when differences
in grazing effect were found. Letters on the x-axis represent different treatments (T, total exclosure; H, half wall roof; W,
wall; R, roof; and C, control). Within sites, treatments that share the same lowercase letter (just above the x-axis) were not
significantly different (Bonferroni post hoc tests, P 0.05).
2, where six clusters were identified. Clusters I to V ditions (clusters V–VII). At Site 2, early domination
comprised successive sampling times from weeks 0– by red algal turf in phase 1 (cluster I) gave way to a
100. The only case in which grazing treatments influ- diversification and an expansion of G. mitchelliae in
enced the formation of clusters was the controls. At phase two (clusters II and III), whereas in phase three
weeks 28 and 64, the controls did not cluster with other Ulva/Enteromorpha and encrusting Gymnogongrus
treatments for those dates, and the controls for weeks dominated and diversity declined as ephemeral species
76, 88, and 100 formed a separate group (cluster VI). diminished or disappeared (clusters IV–VI).
Thus, the controls tended to separate out, but apart from
this, grazing had no effect on the clustering of samples. Aquarium experiments with crab grazing
SIMPER identified the taxa characterizing and dis- Rocks held in aquaria developed different floras de-
tinguishing clusters (Fig. 9). At both sites, there were pending on whether they were exposed to crab grazing
three recognizable phases, between which there were or not (Fig. 10). Initially, the rocks were covered by
significant differences in the abundance of almost all roughly equal proportions of encrusting and erect al-
functional groups (post-hoc contrast analyses, P gae, with 5% bare space. Rocks exposed to G. grap-
0.001 for all groups except articulated corallines, for sus ( crab) became progressively dominated by crus-
which differences were not significant, P 0.05). At tose algae (both Gymnogongrus and corallines). Erect
Site 1, Gymnogongrus and red algal turf dominated the algae (notably Ulva, filamentous greens and red algae)
first phase (cluster I), corresponding to the early onset declined proportionally. In treatments lacking crabs
of El Nino, giving way in a second phase to a more
˜ ( crab), the opposite happened: erect algae increased
diverse array and increasing domination by G. mitch- and smothered crusts.
elliae (clusters II–IV) during the mature stage of El Gymnogongrus, the most important space occupier,
Nino. Finally, in phase three, encrusting Gymnogon-
˜ initially almost solely comprised an encrusting phase,
grus (initially with Enteromorpha/Ulva) became prev- which occupied 32% of the rock surface. In the crab
alent and diversity declined during post-El Nin o con-
˜ treatment, this crust increased to 44% (Fig. 11A), but
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 123
FIG. 9. Results of SIMPER (similarity percentage) analyses diagnosing functional groups characterizing clusters identified
by Bray-Curtis similarity and MDS (multidimensional scaling); see Appendices I and J. See Results: Multivariate analysis
of community structure for a definition of phases and clusters. Asterisks identify significant differences in abundance between
successive clusters (* P 0.05; ** P 0.001; *** P 0.001).
FIG. 10. Percentage cover of (A) crustose algae (corallines and Gymnogongrus) and (B) erect algae (filamentous red and
green algae and foliose green algae) in aquaria that either contained or lacked a crab. Values shown are means SE ( n
6 aquaria per treatment). Any differences between the sum of erect and crustose algae and 100% cover were contributed by
bare rock. Asterisks indicate significant differences between treatments with vs. without a crab (* P 0.05; ** P 0.01;
*** P 0.001).
124 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
in the crab treatment, it progressively transformed
into an erect phase (Fig. 11B). Furthermore, thallus
length of the erect phase increased fivefold in the crab
treatment, but remained stunted in the crab treatment
(Fig. 11C).
DISCUSSION
A central message emerging from the results is the
interplay between (1) bottom-up forces, particularly re-
lated to changes in nutrients (Chavez et al. 1999), but
also to other associated factors such as wave action and
sea level during different phases of the El Nin o cycle,
˜
and (2) top-down grazing effects. Moreover, differenc-
es between sites were greater than expected. Tropical
intertidal rocky shores have previously been general-
ized as communities dominated by encrusting forms,
in which temporal and spatial variation in sessile spe-
cies is minimal (Garrity and Levings [1981] and Lub-
chenco et al. [1984] for the coast of Panama; Lub-
chenco and Gaines [1981] and Brosnan [1992] for a
general review; Williams [1993] for the coast of Hong
Kong). This apparent homogeneity has been attributed
to the sustained combined effect of slow-moving ben-
thic grazers coupled with mobile consumers, particu-
larly fish, because they cover substantial ranges
(Gaines and Lubchenco 1982, Lubchenco et al. 1984,
Menge et al. 1985, Brosnan 1992). However, our study
showed substantial temporal and spatial differences in
composition and abundance. On average, 26.1% of the
variance in abundance of functional groups was attrib-
utable to the main effect of time, 13.3% to site but only
4.8% to grazer treatment.
Differences between sites
There were several background differences between
sites, including a high abundance of red algal turf at
Site 2 and its scarcity at Site 1, and the presence of
articulated coralline turf at Site 2 but not Site 1. Re-
sponses to grazers also differed betweens sites. Two
possible causes are differences in wave action and ther-
mal stress. Site 1 was more exposed to waves than Site
2, although the difference was not great. Intersite dif-
ferences in thermal stress could have influenced the
cover of erect algal species. We have no concrete ev-
idence of thermal differences, however, and regard
them as an unlikely explanation of differences in com-
munity structure between sites, as wave action (which
will mitigate thermal stress) was less at Site 2 where
erect algae were most common.
Grazing did seem to be more effective at Site 1,
FIG. 11. Percent cover (mean SE ) for (A) encrusting
Gymnogongrus sp., (B) erect thalli of Gymnogongrus sp., and because each of the individual types of grazers affected
(C) the length of thalli of erect Gymnogongrus sp. in aquaria algal cover there, whereas at Site 2 only the combined
that either held a single Grapsus grapsus or lacked a crab. effects of all grazers influenced algal cover. Almost the
Asterisks indicate significance of the crab effect (* P 0.05; same numbers of marine iguanas occurred at Sites 1
** P 0.01; *** P 0.001). and 2, but smaller individuals visited Site 1 than Site
2. Juveniles are known to be more affected by waves
than larger individuals and forage in more wave-pro-
tected areas or higher on the shore (Trillmich and Trill-
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 125
mich 1986, Wikelski and Trillmich, 1994). G. grapsus Lubchenco et al. 1984, Menge et al. 1985, Brosnan
was more abundant at Site 2 than Site 1 during the 1992). Seasonal shifts of algal composition have been
post-El Nino period, but the gut fullness of crabs was
˜ documented at Hong Kong, where crusts predominate
greater at Site 1 than Site 2. These variations do not in summer whereas filamentous and foliar algae pro-
necessarily reflect differences in wave action, because liferate in winter, profoundly altering food availability
both marine iguanas and G. grapsus feed during low for herbivores (Kennish et al. 1996). Our results con-
tide when wave action has little effect (Kramer 1967, trast with both of these perspectives because they re-
Buttemar and Dawson 1993). veal striking regime shifts between years, fueled by
In short, there were clear differences in community oceanographic anomalies and the bottom-up influences
structure and dynamics between sites, but the causes of El Nino. The magnitude of the interannual changes
˜
remain uncertain although there are plausible expla- was greater than that previously recorded for any other
nations. tropical shores, and is an important reminder of the
extent to which climate change may alter ecosystem
Temporal fluctuations in species composition structure and dynamics.
The clearest pattern was that all species underwent Our observations spanned both a year of El Nin o and
˜
substantial changes in abundance over time. Abun- four years of ‘‘normal’’ conditions. El Ninos recur at
˜
dance of all taxa except Lithothamnium was signifi- irregular intervals of five to 10 years. As a result, our
cantly different during versus after El Nin o, apparently
˜ work covered only a portion of the full cycle. This
related to high temperatures, low nutrients, and strong would probably have had the effect of overestimating
wave action during El Nino (March 1997 to June 1998),
˜ the effect of time relative to herbivory. Temporal
followed by relative stability after the return of ‘‘nor- changes would have been greatest during El Nin o, and
˜
mal’’ conditions (Chavez et al. 1999). These temporal community composition was relatively stable during
differences explained more of the variance in the data the Post-El Nino period. The relative influence of her-
˜
than any other factor, and multivariate analysis divided bivory might have increased had we run our experi-
the community into temporal clusters rather than clus- ments for an entire ENSO cycle. Indeed, the herbivory
ters based on treatment. treatments were continuing to diverge at the end of the
At both sites, three phases existed (Fig. 9). Phase 1 experiment. Nevertheless, our central conclusions re-
was characterized by domination by encrusting Gym- main robust: temporal changes were substantial and, at
nogongrus and/or red algal turf. Phase 2 heralded least over the period in which we operated, overshad-
ephemeral warm-water species, including Giffordia owed the effects of herbivory.
mitchelliae, which proliferated and aggressively over-
grew encrusting algae. Red algal turf, Enteromorpha Consequences of El Nino events
˜
and Ulva (all preferred items in the diet of iguanas and The shores of Galapagos do not have striking zo-
´
sally lightfoot crabs) diminished or disappeared. G. nation patterns (Hedgepeth 1969, Wellington 1984),
mitchelliae seems characteristic of El Nino periods and
˜ mainly due to the absence of obvious belts of algae,
has been linked to previous ENSO events (Laurie mussels or barnacles. In the past, dense mats of Sar-
1990). Phase 3 marked the exit of G. mitchelliae, with gassum and the endemic Bifurcaria ( Blossevillea)
a return to encrusting Gymnogongrus as a clear dom- galapagensis formed a conspicuous band low on the
inant, initially together with Ulva/Enteromorpha. shore (Houvenaghel and Houvenaghel 1977), but, after
Diversity indices were significantly higher during the the 1982–1983 El Nino, these species disappeared and
˜
El Nino because of the addition of warm-water ephem-
˜ have not been observed since (G. Kendrick, personal
erals, but declined thereafter due to the disappearance communication; L. Vinueza, personal observation).
of some taxa and increased domination by one group During the 1982–1983 El Nino, the endemic sea star
˜
(Fig. 8). Both sites were eventually dominated by crusts Heliaster cuminguii declined drastically (Hickman
(79–94% in control plots). No increases in warm-water 1998). Its distribution was closely associated with the
ephemerals occurred at any time in the 4.5 yr during barnacle Tetraclita milleporosa, an important part of
which control plots were monitored after the passage its diet (Wellington 1975, Hickman 1998). T. mille-
of El Nino, so their emergence during 1997–1998 was
˜ porosa declined during the 1982–1983 El Nino and, at
˜
not just seasonal, but rather related to the anomalous our sites, during and after the 1997–1998 El Nin o.
˜
oceanographic conditions (Chavez et al. 1999). The El Nino also seriously affects other marine com-
˜
lack of any significant change in control plots between munities in Galapagos. During the 1982–1983 ENSO,
´
July 1999 (week 100) and February 2003 (week 277) barnacles, gorgonians, giant scallops, ahermatypic cor-
implies stability over this period. Our data do not allow als, and fish declined drastically (Robinson 1985). Sev-
certainty on this point because sampling in the later eral species of birds failed to breed (Valle 1985), and
years was infrequent, but they do show that inter-annual corals were decimated by 97% (Glynn 1988, 1990).
differences were small after the passage of El Nino. ˜ Laurie (1985, 1990) and Laurie and Brown (1990a)
Previous tropical-shore studies have emphasized their recorded 50% mortality of marine iguanas, and drastic
inter-annual stability (Gaines and Lubchenco 1982, reductions of red algal turfs and foliose green algae,
126 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
which are important constituents of the diet of marine The barren aspect of Panamanian shores has been
iguanas and other intertidal grazers. The 1982–1983 El attributed to conspicuous slow-moving consumers (no-
Nino also featured domination by Giffordia mitchel-
˜ tably siphonariid and fissurellid limpets) that graze fo-
liae, never before recorded in the Archipelago (Laurie liose algae down to grazer-resistant forms (Menge et
1985). al. 1986a, b). In Galapagos, and specifically at our ex-
´
Proliferation of G. mitchelliae (and concomitant re- perimental sites, slow-moving grazers were rare and
ductions of edible species) has important implications, their biomass low. In particular, the near absence of
as the digestibility of G. mitchelliae is 21%, compared limpets at all intertidal rocky shores is striking. In short,
with 64% for foliar green algae and 78% for red algal sedentary intertidal grazers that exert an important in-
turf. Its high terpenoid content probably reduces the fluence on algae in other tropical areas are absent or
digestive capabilities of iguanas and may have con- scarce in Galapagos as a whole, and specifically at the
´
tributed to their declining condition and elevated mor- two experimental sites.
tality during El Nino (Laurie 1985, Laurie and Brown
˜
Diets of consumers
1990b). Our data showed that G. mitchelliae was also
avoided by Grapsus grapsus, although it is an impor- Dietary preferences of consumers will also affect
tant constituent of the diet of Grapsus albolineatus in their influence on intertidal communities. With respect
Hong Kong (Kennish 1996). G. mitchelliae was among to iguanas, four generalities emerge from the consid-
several species that increased their abundance or ex- erable literature on their diets (Nagy and Shoemaker
tended their range during the 1997–1998 El Nin o, par-
˜ 1984, Laurie 1985, Wikelski et al. 1993, Wikelski and
ticularly species with affinities to the Indo Pacific and Hau 1995). First, iguanas consume most algae, espe-
Panamic Province, which were probably transported cially foliar green algae. Second, they avoid certain
southward by a warm tongue of water that extended to species. For example, before the 1982–1983 El Nin o, ˜
the whole eastern Pacific (Ruttenberg 2000). Bifurcaria galapagensis was abundant, but apparently
The drastic changes in algal composition and edi- never eaten by iguanas. Third, Giffordia mitchelliae,
bility, accompanied by strong swells and extremely which rises to dominance during El Nin o periods, ap-
˜
high sea levels, reduced food availability and restricted pears to have negative effects on iguanas as described
intertidal feeding opportunities. The correlation be- above. Fourth, low-growing algae are difficult for igua-
tween algal biomass and densities of crabs and iguanas nas (and other herbivores) to graze. Consequently,
during El Nino also suggests they were aggregating
˜ crusts are rarely eaten (Wikelski and Hau 1995), and
where food was most available, implying that food was stand to benefit from the removal of superior compet-
limiting at sites where algal stocks were low. All these itors. Moreover, firmly attached erect algae such as red
factors would have contributed to the heightened mor- algal turf can be grazed down to a thin veneer, but are
tality of marine iguanas ( 50%) observed during difficult to eliminate (Lubchenco et al. 1984, Menge et
1982–1983 (Laurie 1990) and 1997–1998 (Romero and al. 1986b). Marine iguanas can completely remove
Wikelski 2001). After the 1997–1998 El Nino, G. ˜ weakly attached algae such as Ulva and Enteromorpha,
mitchelliae disappeared, being replaced by encrusting but cannot graze on turfs 1 mm in height (Wikelski
Gymnogongrus and edible groups such as Ulva, En- and Hau 1995). Our data showed that grazing reduced
teromorpha, and red algal turf. G. grapsus increased the height of red algal turf, but scarcely influenced its
greatly, and marine iguanas recovered fast, reaching percentage cover.
weights heavier than even those before the El Nin o.˜ Grapsus grapsus showed clear dietary preferences,
with crusts and filamentous browns (notably Giffordia
Types of consumers mitchelliae) being avoided. In contrast, Grapsus al-
bolineatus preferentially feeds on filamentous algae,
Herbivorous fish, sally lightfoot crabs, and iguanas including Giffordia ( Hinksia) mitchelliae in Hong
averaged densities of 1.4, 0.64, and 0.1 individuals/m2, Kong (Kennish 1996, Kennish and Williams 1997).
whereas slow-moving herbivores such as chitons, lim- Overall, grazers displayed four responses to algae.
pets, nerites, and urchins never exceeded 0.03 individ- Encrusting forms were negatively selected because
uals/m2. Their relative impacts on algae will be influ- they are inaccessible; red algal turfs and foliar greens
enced by many factors, including biomass, rate of con- were positively selected; articulated corallines (and
sumption, dietary preferences, duration of tidal for- perhaps filamentous greens) were eaten in proportion
aging, mobility, and the edibility of different algae to availability, and Giffordia mitchelliae was either
(Lubchenco and Gaines 1981). Most of these factors avoided or had adverse effects because of its low di-
lie outside the scope of this investigation, but approx- gestibility.
imate conversions to wet biomass yielded estimations Four corresponding responses to grazing existed
of 178.4, 89.4, 15.4, and 2.9 g/m2 for herbivorous fish, among the algae. (1) Encrusting Gymnogongrus was
iguanas, sally lightfoot crabs, and slow-moving graz- enhanced by grazing, probably as an indirect response
ers, respectively; providing a better perspective of their to grazer-induced reductions of foliar algae that would
potential influence. otherwise have overgrown them. The laboratory ex-
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 127
periments verified that by diminishing foliar and fila- such as Ulva, and indirectly promoted competitively
mentous algae, crabs could enhance encrusting Gym- inferior but grazer-resistant species such as crustose
nogongrus and encrusting corallines. In the field, en- Gymnogongrus. At Site 2, edible foliar algae were more
crusting and erect phases of Gymnogongrus are prob- abundant, and only the combined effects of all grazers
ably adapted to avoiding grazing and the threat of (in the control treatment) significantly reduced ephem-
overgrowth by erect algae respectively (cf. Lubchenco erals and promoted crusts. However, not all crusts re-
and Cubit 1980, Slocum 1980). (2) Some algae such sponded similarly; Gymnogongrus increased substan-
as Ulva, Enteromorpha, and red algal turf were dimin- tially, but Lithothamnium consistently decreased in
ished by grazing. All these algae are preferentially con- abundance and never recovered fully. No grazing effect
sumed by iguanas and crabs. (3) Articulated corallines was detectable for Lithothamnium, although its highest
were scarcely affected by grazing and were eaten in levels of cover were associated with particular times
proportion to availability. (4) G. mitchelliae was un- (December 1997) and treatments (control and roof) in
affected by grazing. which the combined cover of red algal turf, Gymno-
gongrus, and ephemeral algae was lowest. Low com-
Changes in community structure due to petitive ability and slow growth may have retarded the
herbivore manipulation recovery of Lithothamnium. Only three years after El
The effect of grazers fell into three phases. (1) Dur- Nino did its recovery begin.
˜
ing El Nino (March 1997 to June 1998), preferred food
˜ Red algal turf declined from initially moderately
species were scarce, being replaced by G. mitchelliae. high (Site 1) or very high levels of cover (Site 2) down
Heavy swells and high seas would additionally have to very low levels, but the intensity of grazing had
hindered grazing. Densities of G. grapsus were low, little effect on its cover most of the time, in keeping
and mortality of marine iguanas elevated (Romero and with Hay’s (1981) views than the colonial aggregations
Wikelski 2001). (2) In the late stages of El Nin o and
˜ formed by turfs reduce herbivory. There were signifi-
its aftermath (July 1998 to March 1999), ephemeral cant interactions between site and treatment. Grazing
algae (particularly Ulva) grew luxuriantly, reaching effects were undetectable at Site 1 where the cover of
frond lengths of 80 mm and close to 75% coverage red algal turf was low. At Site 2, red algal turf was
even in control plots exposed to all grazers. Chaeto- abundant, and increased in treatments with moderate
morpha antennina achieved frond lengths of 100 mm to high intensities of grazing, perhaps benefiting from
and cover of G. mitchelliae rose substantially. Densities reductions of palatable, competitively superior species.
of mobile grazers such as iguanas and crabs were still Lewis (1986) reported a reduction of turf and crustose
low, and high primary production could have improved species in experimental plots that totally excluded graz-
conditions for consumers and led to the greater body ers, and argued that turfs and crusts are adapted to
mass recorded for iguanas (Wikelski and Trillmich intense grazing as they dominate areas that are heavily
1997). Sessile organisms competed for space, with grazed.
ephemeral algae flourishing, crustose forms initially Articulated corallines were virtually absent from Site
declining and barnacles diminishing to the lowest lev- 1. At Site 2, they were more common, and increased
els recorded. (3) After nutrient and temperature levels within roofed treatments (H, R) at all times. There are
returned to normal (June 1999 onward), grazing effects three possible explanations for this. One is that grapsid
of marine iguanas, crabs, and fish were pronounced and crabs (which would have entered these treatments) se-
provoked changes in community structure, including a lectively removed competitively superior species such
reduction of Ulva and Chaetomorpha antennina, and as Ulva. Second, articulated corallines may have been
an increase in crustose forms. outcompeted in the total exclusion cages. Finally, in-
The magnitude of responses to grazing was greatest tense grazing in the control may have diminished ar-
when the cover of each functional group was high. This ticulated corallines. The net effect was that these algae
implies that during periods of low productivity, algae were most abundant at intermediate levels of grazing.
do not have the capacity to benefit from a reduction or Being heavily calcified, corallines represent a poor
elimination of grazing, and are probably limited by source of energy for grazers (Paine and Vadas 1969,
physicochemical conditions. Conversely, during peri- Duffy and Hay 1990). Relative to their abundance, up-
ods of high productivity dramatically increased growth right coralline algae were not an important part of the
was recorded in grazer-exclusion plots. During the El diet of G. grapsus or marine iguanas (Wikelski et al.
Nino the bottom-up effects of nutrient limitation (Cha-
˜ 1993; L. Vinueza, unpublished data).
vez et al. 1999) were prevalent, and rippled up the food
chain to affect consumers, but when ‘‘normal’’ con- Effect of grazers on algal diversity
ditions returned, the top-down effects of herbivores on Grazers can potentially increase or decrease diver-
algal composition and sizes increased. sity. Some herbivores reduce the landscape to few graz-
The effects of grazing differed between sites, being er-resistant competitively inferior encrusting species.
most obvious at Site 1 where the cumulative or indi- Examples include sea urchins (Paine and Vadas 1969,
vidual effects of grazers reduced ephemeral species Vance 1979), parrotfish (Lewis 1986, Hixon and Bros-
128 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 12. Synthesis of the relative importance of temporal changes, differences between sites, and the effects of grazers
in explaining variance in algal abundance. Changes in Shannon diversity (H ) and the abundance of algae among phases are
averaged across sites and are based on the percent cover reported for control plots. Differences between the sites are averaged
across times and are also based on the percent cover of control plots. Effects of grazers are based on averages across both
times and sites, and they compare control plots with total exclusion plots.
toff 1996), and the diverse guild of consumers present reduction in diversity, particularly post-El Nin o, when
˜
in Panama (Menge and Lubchenco 1981, Lubchenco algae were abundant and growth rates high.
et al. 1984).
We recorded differences in diversity between sites Conclusions and broader implications
over time and among treatments (Fig. 8). First, Site 1 Our study provides new insight into the regulation
had lower species richness than Site 2. Second, sub- of tropical intertidal communities, differing in several
stantial temporal fluctuations in diversity indices oc- respects from previous descriptions (Gaines and Lub-
curred in all treatments. Initially, diversity was inter- chenco 1982, Menge et al. 1985, Brosnan 1992). The
mediate (phase 1, week 0). During phase 2 (weeks 16– near absence of sedentary grazers, significant differ-
20), when temperatures were high and nutrient levels ences between sites, and interannual variation related
low, diversity increased due to the emergence of to the large-scale disturbances of El Nino show that
˜
ephemerals such as Giffordia mitchelliae and Chae- factors additional to grazing have important effects in
tomorpha antennina. In phase 3 (week 52 onward), structuring Galapagos intertidal communities. Tem-
´
decreases of several species and increased dominance poral and intersite differences were substantially great-
by Enteromorpha/Ulva and/or Gymnogongrus were ac- er than grazing effects, at least over the period of our
companied by colder temperatures and high levels of study, reflecting the influences of large-scale oceano-
nutrients (Chavez et al. 1999). Thus, diversity declined graphic events (El Nino) and smaller-scale spatial ef-
˜
when grazing probably became more intense, with crab fects on community structure. This echoes the conclu-
densities rising sharply and the condition of iguanas sions of Sauer Machado et al. (1996) about tropical
improving. Brazilian shores, that although consumers do influence
Regional fluctuations of temperature and nutrients community structure, their effects are limited and var-
levels influenced diversity by addition or loss of iable. Fig. 12 summarizes the relative degrees to which
ephemeral species. Superimposed on this, grazing variances in the data were explained by the main effects
treatment had a significant but relatively small effect, of time (27%), site (13%) and grazing (5%). There
the clearest manifestation of which was to promote were, however, important interactions between time
dominance by crustose Gymnogongrus, with associated and treatment, with herbivores having less effect during
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 129
El Nino than after it. The bottom-up influence of El
˜ Buttemar, W. A., and W. R. Dawson. 1993. Temporal pattern
of foraging and microhabitat use by Galapagos marine igua-
´
Nino and the top-down effects of herbivores thus
˜
nas, Amblyrhynchus cristatus. Oecologia 96:56–64.
changed in relative importance over time. Cane, A. M. 1983. Oceanographic events during El Nino. ˜
In short, the results emphasize the profound trans- Science 222:1189–1195.
formation that global climate change may bring to ma- Chavez, F. P., P. G. Strutton, G. E. Friederich, R. A. Feely,
rine ecosystems, including shifts in composition, in- G. C. Feldman, D. G. Foley, and M. J. McPhaden. 1999.
Biological and chemical response of the equatorial Pacific
teractions, and productivity (Fields et al. 1993). The Ocean to the 1997–98 El Nino. Science 286:2126–2131.
˜
biota of Galapagos, having high endemicity and being
´ Cubit, J. 1984. Herbivory and the seasonal abundance of
isolated and positioned at the juncture of contrasting algae on a high intertidal rocky shore. Ecology 65:1904–
oceanographic currents, is delicately poised and prone 1917.
Day, R. W., and G. P. Quinn. 1989. Comparisons of treatments
to extinction in the face of global changes. If El Nin o
˜ after an analysis of variance in ecology. Ecological Mono-
conditions intensify or become more frequent, we can graphs 59:433–463.
anticipate greater environmental stochasticity, higher Dayton, P. K. 1971. Competition, disturbance, and commu-
temperatures, lower nutrient levels, reduced productiv- nity organization: the provision and subsequent utilization
of spaces in a rocky intertidal community. Ecological
ity and biomass, reductions in the densities of consum-
Monographs 41:351–389.
ers, extinctions of cooler-water species, and probably Duffy, J. E., and M. E. Hay. 1990. Seaweed adaptations to
long-term reductions in diversity. herbivory. BioScience 40:368–376.
These effects will have economic as well as ecolog- Fields, P. A., J. B. Graham, R. H. Rosenblatt, and G. N.
ical consequences, because many of the top consumers Somero. 1993. Effects of expected global climate change
on marine faunas. Trends in Ecology and Evolution. 8:361–
will be affected most, including fish that are harvested, 367.
and birds, iguanas, and sea lions that are charismatic Gaines, S. D., and J. Lubchenco. 1982. A unified approach
attractions for the lucrative ecotourism trade. to marine plant–herbivore interactions. II. Biogeography.
Annual Review Ecology and Systematics 13:111–138.
ACKNOWLEDGMENTS Garrity, S., and S. Levings. 1981. A predatory–prey inter-
actions between two physically and biologically con-
We thank the Galapagos National Park Service and the
´ strained tropical rocky shore gastropods: direct, indirect
Charles Darwin Foundation for the permits and support to and community effects. Ecological Monographs 5:267–
conduct this work. Part of this work comprised a thesis by 286.
L. Vinueza at the University of Wales in Bangor, supported Glynn, P. W. 1988. El Nino–Southern Oscillation 1982–1983:
˜
by a Chevening Scholarship and a student grant from the nearshore population, community and ecosystem respons-
Foundation for Science and Technology of Ecuador to L. es. Annual Review of Ecology and Systematics 19:309–
Vinueza. Ray Seed, Jose Miguel Farina, John Witman, Bruce
˜ 345.
Menge, Paul Dayton, Peter Raimondi, and two anonymous Glynn, P. W. 1990. Coral mortality and disturbances to coral
reviewers constructively improved the manuscript. Paulina reefs in the tropical eastern Pacific. Pages 55–126 in P. W.
Guarderas, Paula Angeloni, Franz Smith, Adelaida Herrera, Glynn, editor. Global ecological consequences of the 1982–
Vanessa Francisco, and Monica Flores provided invaluable 83 El Nino–Southern Oscillation. Elsevier Oceanography
˜
field assistance and continual friendship. Funding for this Series 52. Elsevier, Berlin, Germany.
project was provided by the USAID to the Charles Darwin Hay, M. E. 1981. The functional morphology of turf-forming
Foundation, the Pew Charitable trusts Marine Conservation seaweeds: persistence in stressful marine habitats. Ecology
Fellowship program and the Pew Collaborative Initiative 62:739–750.
Fund Award to R. H. Bustamante. G. M. Branch was funded Hedgepeth, J. 1969. An intertidal reconnaissance of rocky
by the University of Cape Town, the National Research Foun- shores of the Galapagos. Wasmann Journal of Biology
dation, SANCOR, and the Andrew Mellon Foundation. G. 27(1):1–22.
M. Branch and M. L. Branch were kindly accommodated by Hickman, C. P. 1998. Guıa de campo sobre las estrellas de
´
Sol and Rodrigo Bustamante, renewing an old and cherished mar y otros equinodermos de Galapagos. Serie vida marina
´
friendship. This is contribution number 1033 from the Charles de Galapagos. Sugar Spring Press, Lexington, Virginia,
´
Darwin Foundation and contribution number 189 from PIS- USA.
CO, the Partnership for Interdisciplinary Studies of Coastal Hixon, A. M., and W. N. Brostoff. 1996. Succession and
Oceans, funded primarily by the Gordon and Betty Moore herbivory: effects of differential fish grazing on Hawaiian
Foundation and David and Lucile Packard Foundation. coral-reef algae. Ecological Monographs 66:67–90.
Hockey, P. A. R., and G. M. Branch. 1984. Oystercatchers
LITERATURE CITED and limpets: impact and implications. A preliminary as-
sessment. Ardea 72:119–206.
Brosnan, D. M. 1992. Ecology of tropical rocky shores: Houvenaghel, G. 1984. Oceanographic setting of the Gala-
plant–animal interactions in tropical and temperate lati- pagos Islands. Pages 43–54 in R. Perry, editor. Galapagos
tudes. Pages 101–131 in D. M. John, S. J. Hawkins, and key environments series: Galapagos Islands. Pergamon
J. H. Price, editors. Plant animal interactions in the marine Press, Oxford, UK.
benthos. Systematics Association special volume No. 46. Houvenaghel, G. T., and N. Houvenaghel. 1977. Aspectos
Clarendon Press, Oxford, UK. ecologicos de la zonificacion entre marea en las costas ro-
´ ´
Bustamante, R. H., G. M. Wellington, G. M. Branch, G. J. cosas de las islas Galapagos. Pages 79–100 in U. Eberhardt,
´
Edgar, P. Martinez, F. Rivera, and F. Smith. 2002. Out- editor. Compendio de ciencia en Galapagos. Publicacion
´ ´
standing marine features if the Galapagos Archipelago. de la Estacion Cientıfica Charles Darwin Isla Santa Cruz,
´ ´
Pages 61–72 in R. Benson-Smith, editor. A biodiversity Galapagos, Ecuador.
´
vision for the Galapagos Islands. Charles Darwin Foun- Kennish, R. 1996. Diet composition influences the fitness of
dation and World Wide Fund, Puerto Ayora, Galapagos, the herbivorous crab Grapsus albolineatus. Oecologia 105:
Ecuador. 22–29.
130 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
Kennish, R., and G. A. Williams. 1997. Feeding preferences sumers on sessile prey in the low zone of a rocky shore in
of the herbivorous crab Grapsus albolineatus and the dif- the bay of Panama: direct and indirect consequences of
ferential influence of algal nutrient content and morphol- food web complexity. Journal of Experimental Marine Bi-
ogy. Marine Ecology Progress Series 147:87–95. ology and Ecology 100:225–269.
Kennish, R., G. A. Williams, and S. Y. Lee. 1996. Algal Nagy, K. A., and V. H. Shoemaker. 1984. Field measurements
seasonality on an exposed rocky shore in Hong Kong and and food consumption of the Galapagos marine iguana,
the dietary implications for the herbivorous crab Grapsus Amblyrhynchus cristatus. Physiological Zoology 57:281–
albolineatus. Marine Biology 125:55–64. 290.
Kramer, P. 1967. Beobachtungen zur biologie und zum ver- Paine, R. T., and R. Vadas. 1969. The effects of grazing by
halten der klippencrabbe Grapsus grapsus L. (Brachyura sea urchins, Strongylocentrotus spp. on benthic algal pop-
Grapsidae) auf den Galapagos und am Ekuadorianischen ulations. Limnology and Oceanography 14:710–719.
Festland. Zeitschrift fur Tierpsychologie 24:385–402.
¨ Palumbi, S. R. 1984. Measuring intertidal wave forces. Jour-
Kuehl, R. O. 2000. Design of experiments: statistical prin- nal of Experimental Marine Biology and Ecology 81:171–
ciples of research design and analysis. Second edition. Dux- 179.
bury Press, Pacific Grove, California, USA. Quinn, G. P., and M. J. Keough. 2002. Experimental design
Laurie, A. 1985. The effect of the 1982–1983 El Nino on ˜ and data analysis for biologists. Cambridge University
Marine Iguanas. Pages 199–209 in G. Robinson and E. M Press, Cambridge, UK.
del Pino, editors. El Nino en las Islas Galapagos: El Evento
˜ ´ Robinson, G. 1985. Influence of the 1982–1983 El Nino on ˜
de 1982–1983. Publication of the Charles Darwin Foun- Galapagos marine life. Pages 153–190 in G. Robinson and
´
dation for the Galapagos Islands, Quito, Ecuador.
´ E. M. del Pino, editors. El Nino en las Islas Galapagos: El
˜ ´
Laurie, A. 1990. Effects of the 1982–1983 El Nino-Southern
˜ Evento de 1982–1983. Publication of the Charles Darwin
Oscillation event on marine iguanas (Amblyrhynchus cris- Foundation for the Galapagos Islands, Quito, Ecuador.
´
tatus Bell, 1825) populations on Galapagos. Pages 361– Romero, L. M., and M. Wikelski. 2001. Corticosterone levels
380 in P. W. Glynn editor. Global ecological consequences predict survival probabilities of Galapagos marine iguanas
of the 1982–83 El Nino-Southern Oscillation. Elsevier during El Nino events. Proceeding of the National Academy
˜
Oceanography Series 52. Elsevier, Berlin, Germany. of Sciences (USA) 98:7366–7370.
Laurie, W. A., and D. Brown. 1990a. Population biology of Ruttenberg, B. 2000. An unusual pulse of recruitment of two
marine iguanas (Amblyrhynchus cristatus). II. Changes in reef fishes in the Galapagos Islands coincident with the
´
annual survival rates and the effects of size, sex, age and 1997–1998 El Nino. Bulletin of Marine Science 67:869–
˜
fecundity on a population crash. Journal of Animal Ecology 874.
59:529–544. Sauer Machado, K. R. S., A. R. O. Chapman, and R. Coutinho.
Laurie, W. A., and D. Brown. 1990b. Population biology of 1992. Patch structure in a tropical rocky shore community
marine iguanas (Amblyrhynchus cristatus). III. Factors af- in Brazil: a mosaic of successional states? Ophelia 35(3):
fecting survival. Journal of Animal Ecology 59:545–568. 187–195.
Lewis, S. M. 1986. The role of herbivorous fishes in the Sauer Machado, K. R. S., A. R. O. Chapman, and R. Coutinho.
organization of a Caribbean reef community. Ecological 1996. Consumer species have limited and variable roles in
Monographs 56:183–200. community organization on a tropical intertidal shore. Ma-
Lubchenco, J. 1980. Algal zonation in the New England rine Ecology Progress Series 134:78–83.
rocky intertidal community: an experimental analysis. Slocum, C. J. 1980. Differential susceptibility to grazers in
Ecology 61:333–344. two phases of an intertidal alga: advantages of heteromor-
Lubchenco, J., and J. Cubit. 1980. Heteromorphic life his- phic generations. Journal of Experimental Marine Biology
tories of certain marine algae as adaptations to variation in and Ecology 46:99–110.
herbivory. Ecology 6:676–687. Trillmich, K. G. K., and F. Trillmich. 1986. Foraging strat-
Lubchenco, J., and S. D. Gaines. 1981. A unified approach egies of the marine iguana, Amblyrhynchus cristatus. Be-
to marine plant–herbivore interactions. I. Populations and havioral Ecology and Sociobiology 18:259–266.
communities. Annual Review Ecology and Systematics 12: Valle, C. 1985. Alteracion de las poblaciones del cormoran
´ ´
405–437. no volador, el pinguno y otras aves marinas en Galapagos
¨ ´
Lubchenco, J., B. A. Menge, S. D. Garrity, P. J. Lubchenco, por el efecto de El Nino 1982–1983 y su subsecuente re-
˜
L. R. Ashkenas, S. D. Gaines, R. Emlet, J. Lucas, and S. gereracion. Pages 245–258 in G. Robinson and E. M. del
´
Strauss. 1984. Structure, persistence and role of consumers Pino, editors. El Nino en las Islas Galapagos: El Evento
˜ ´
in a tropical rocky intertidal community (Taboguilla Island, de 1982–1983. Publication of the Charles Darwin Foun-
Bay of Panama). Journal of Experimental Marine Biology dation for the Galapagos Islands, Quito, Ecuador.
´
and Ecology 78:23–73. Vance, R. R. 1979. Effects of grazing by the sea urchin Cen-
Menge, B. A., and G. M. Branch. 2001. Rocky intertidal trostephanus coronatus on prey community composition.
communities. Pages 221–251 in M. D. Bertness, S. Gaines, Ecology 60:537–546.
and M. E. Hay, editors. Marine community ecology. Sin- Wellington, G. 1975. Medio ambientes marino costeros de
auer Associates, Sunderland, Massachusetts, USA. Galapagos. Un informe de recursos al departamento de
´
Menge, B. A., and J. Lubchenco. 1981. Community orga- Parque Nacionales y Vida Silvestre, Quito, Ecuador.
nization in temperate and tropical rocky intertidal habitats: Wellington, G. 1984. Marine environment and protection.
prey refuges in relation to consumer pressure gradients. Pages 247–263 in R. Perry, editor. Galapagos key envi-
Ecological Monographs 51:429–450. ronments series: Galapagos Islands. Pergamon Press, Ox-
Menge, B., J. Lubchenco, and L. Ashkenas. 1985. Diversity, ford, UK.
heterogeneity and consumer pressure in a tropical rocky Wikelski, M., B. Gall, and F. Trillmich. 1993. Ontogenetic
intertidal community. Oecologia (Berlin) 63:394–405. changes in food intake and digestion rate of the herbivorous
Menge, B. A., J. Lubchenco, S. D. Gaines, and L. R. Ash- marine iguana (Amblyrhynchus cristatus, Bell). Oecologia
kenas. 1986a. A test of the Menge-Sutherland model of 94:373–379.
community organization in a tropical rocky intertidal food Wikelski, M., and M. Hau. 1995. Is there an endogenous
web. Oecologia (Berlin) 71:75–89. tidal foraging rhythm in marine iguanas? Journal of Bio-
Menge, B. A., J. Lubchenco, S. D. Gaines, and L. R. Ash- logical Rhythms 10(4):335–350.
kenas. 1986b. Experimental separation on effects of con- Wikelski, M., and F. Trillmich. 1994. Foraging strategies of
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 131
the Galapagos marine iguana (Amblyrhynchus cristatus): richness and abundance in the presence of molluscan her-
adapting behavioral rules to ontogenic changes. Behaviour bivores on a tropical rocky shore. Journal of Experimental
128:255–279. Marine Biology and Ecology 167:261–275.
Wikelski, M., and F. Trillmich. 1997. Natural and sexual Williams, G. A. 1994. The relationship between shade and
selection on body size fluctuates between marine iguana molluscan grazing in structuring communities on a mod-
island populations. Evolution 51:922–936. erately-exposed tropical rocky shore. Journal of Experi-
Williams, G. A. 1993. Seasonal variation in algal species mental Marine Biology and Ecology 178:79–95.
APPENDIX A
Map of the study sites (Ecological Archives M076-005-A1).
APPENDIX B
Figure showing the comparison of mean monthly sea surface temperatures in the Galapagos (Ecological Archives M076-005-
´
A2).
APPENDIX C
Table showing mean abundances of fish/m2 at the two study sites (Ecological Archives M076-005-A3).
APPENDIX D
Table showing relative abundance of slow moving invertebrates (no. individuals/m2) at the two study sites (Ecological
Archives M076-005-A4).
APPENDIX E
Table showing the results of a repeated-measures ANOVA on the effect of treatment and site on the abundance of functional
groups of algae (Ecological Archives M076-005-A5).
APPENDIX F
Table showing the results of an ANOVA on the effect of treatment, site, and time on algal frond lengths (mm) (Ecological
Archives M076-005-A6).
APPENDIX G
Table showing the results of a repeated-measures ANOVA to test the effect of treatment and site on species richness,
diversity, and evenness (Ecological Archives M076-005-A7).
APPENDIX H
Figure showing temporal changes in species richness (Margalef’s index d), evenness (Pielou’s evenness index J ), and
diversity (Shannon-Wiener diversity index H ) (Ecological Archives M076-005-A8).
APPENDIX I
Dendrograms of Site 1 and Site 2 (Ecological Archives M076-005-A9).
APPENDIX J
Nonmetrical multidimensional scaling (nMDS) for Site 1 and Site 2 (Ecological Archives M076-005-A10).
2006 by the Ecological Society of America
˜
TOP-DOWN HERBIVORY AND BOTTOM-UP EL NINO EFFECTS ON
´ PAGOS ROCKY-SHORE COMMUNITIES
GALA
L. R. VINUEZA,1,4 G. M. BRANCH,2 M. L. BRANCH,2 AND R. H. BUSTAMANTE3
1Charles Darwin Foundation, The Galapagos Islands, Ecuador
´
2 Zoology Department, University of Cape Town, Rondebosch 7701, South Africa
3CSIRO Marine Research, Cleveland 4163, Australia
Abstract. We evaluated the effects of marine iguanas, sally lightfoot crabs, and fish
on rocky-shore sessile organisms at two sites at Santa Cruz Island, Gala pagos Islands,
´
Ecuador, for 3–5 years during and after the 1997–1998 El Nin o, using exclusion cages to
˜
separate the effects. Plots exposed to natural grazing were dominated either by encrusting
algae or by red algal turf and articulated corallines. Algae fluctuated in response to El Nin o
˜
in the following way. During an early phase, crustose Gymnogongrus and/or red algal turf
were dominant. In the heart of El Nino, grazers had limited effects on algal cover but
˜
influenced algal sizes substantially. Most algae (particularly edible forms) were scarce or
declined, although warm-water ephemeral species (notably Giffordia mitchelliae) flourished,
increasing diversity and overgrowing crusts. Iguana mortalities were high, and crab densities
low. When normal conditions returned, warm-water ephemerals declined, crab densities
rose, and grazers had significant but site-specific effects on algae. At one site, any com-
bination of grazers diminished most erect species, reducing diversity and restoring domi-
nance of competitively inferior grazer-resistant crusts. At a second site, only the combined
effect of all grazers had this effect. Laboratory experiments confirmed that crabs could
control erect algae and promote crustose forms, and crustose Gymnogongrus developed
into an erect form in the absence of crabs. Differences between sites and large-scale temporal
changes associated with El Nino indicate that tropical shores are not all as constant in time
˜
and space as previously suggested. Mobile grazers did affect algal communities, but over
the period of our observations far greater effects were attributable to intersite differences
and temporal shifts in oceanographic conditions. El Nin o events reduce nutrients, intensify
˜
wave action, and raise sea levels, affecting food availability for intertidal herbivores and
their influence on benthic algae. Thus, the dramatic transformations of communities during
El Nino presage the impacts of global climate change.
˜
Key words: algae; bottom-up top-down effects; climate change; ENSO; Galapagos Islands; graz-
´
ing; rocky shores.
INTRODUCTION bivory on community structure relative to (b) changes
Two current focuses of ecology are the effects of in bottom-up effects associated with oceanographic
global climate change and distinguishing the relative shifts.
importance of bottom-up environmental effects versus The effectiveness of grazers can be restricted spa-
the top-down influences of consumers (Menge and tially and temporally by physical factors such as wave
Branch 2001). The Galapagos Archipelago provides
´ action and thermal stress (Dayton 1971, Cubit 1984),
unique circumstances to explore both issues because it habitat complexity or predation (Hockey and Branch
lies at the juncture of three major currents and is par- 1994), or plant defenses (Lubchenco 1980). These pro-
ticularly susceptible to oceanographic shifts associated cesses act locally but can be affected by large-scale
with El Nino. Its intertidal rocky shores are thermally
˜ oceanographic conditions. Thus, changes in tempera-
stressed by the equatorial climate and herbivory ap- ture, nutrient levels or propagule supply associated
pears intense. In this paper, we examine the impact of with patterns of ocean circulation can affect the set-
herbivory over periods of contrasting oceanographic tlement, growth and reproduction of algae and the ac-
conditions during and after the 1997–1998 El Nin o, to
˜ tivities of herbivores and, hence, interactions between
examine (a) the magnitude of top-down effects of her- them (Lubchenco and Gaines 1981). However, if graz-
ing intensifies beyond a threshold, herbivores can de-
Manuscript received 30 December 2004; revised 10 June crease diversity, reducing the landscape to a few grazer-
2005; accepted 15 July 2005; final version received 26 July 2005. resistant species (Paine and Vadas 1969, Vance 1979,
Corresponding Editor: P. T. Raimondi.
4 Present address: Zoology Department, 3029 Cordley Menge et al. 1985, Hixon and Brostoff 1996).
Hall, Oregon State University, Corvallis Oregon 97331 USA. Tropical intertidal systems contain a range of her-
E-mail: vinuezal@science.oregonstate.edu bivore guilds, and there are few predictions as to how
111
112 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
their plant assemblages are organized by the combined between different phases of El Nino, and to test the
˜
effects of numerous herbivores. The barren aspect of generality of previous models accounting for the com-
many tropical latitudes has been attributed to the pres- position of tropical rocky-shore communities. To
ence of diverse consumers (Menge and Lubchenco achieve this, we (1) quantified herbivores, (2) estab-
1981, Menge et al. 1986a, b, Brosnan 1992). In Pan- lished whether differences existed in community com-
ama, intertidal rocky shores are dominated by encrust- position among years, and (3) measured the effects of
ing forms; sessile animals and macroalgae are scarce. grazers on algae by field experiments excluding igua-
Both fast-moving consumers (e.g., fish and crabs) and nas, crabs and fish, and by laboratory experiments with
slow-moving consumers (e.g., limpets) are abundant crabs.
and reduce diversity on open surfaces, maintaining a
consistent community because their effects are impor- MATERIALS AND METHODS
tant year-round (Menge and Lubchenco 1981, Lub-
chenco et al. 1984). In Hong Kong, the effect of con- Study sites and conditions
sumers diminishes during summer because of reduced The field experiments were replicated at two inter-
food availability and thermal stress (Williams 1993, tidal, south-facing, gently sloping, lava reefs in Acad-
1994, Kennish et al. 1996). In Brazil, lush foliose algae emy Bay, Santa Cruz Island, Ecuador. Site 1 lay in
and zoanthids cover the low intertidal, despite physical front of the Charles Darwin Research Station Marine
conditions there being comparable to those in Panama. Laboratory on the outer, wave-exposed face of a lava
The role of consumers is inconsistent and sometimes bench forming an embayment (0 44 37 S, 90 18 18
crabs and fish even enhance the growth of Ulva. Bare W). Site 2 was located at La Ratonera, 200 m to the
space is rare and competition between sessile forms east in front of the meteorological station (0 44 46 S,
often important (Sauer Machado et al. 1992, 1996). 90 18 09 W). Site selection was based on (a) logistic
Thus, there is conflicting evidence of the role of accessibility, as the experiments had to be monitored
herbivory on tropical rocky shores. Generalizations regularly over a prolonged period, and (b) replication
have been based on a limited geographic coverage, and
at sites that were sufficiently close by as to be regarded
seldom integrate stochastic environmental changes and
as physically comparable, but far enough apart to be
their possible modulation of herbivory (Brosnan 1992).
independent. Both sites were subjectively rated as
Rocky shores in Galapagos are characterized by a
´
semi-exposed, but measurements with dynamometers
black, basaltic substratum of volcanic origin that reach-
(Palumbi 1984) revealed that Site 1 was slightly more
es temperatures as high as 60 C. Thermal stress strong-
exposed than Site 2 (experiencing maximum wave forc-
ly affects community structure, but its influence differs
es of 11.45 2.32 N and 9.11 2.84 N, respectively;
at small scales related to intertidal height, wave ex-
ANOVA, F1.42 18.19, P 0.001). Densities of igua-
posure, and relief (Wellington 1984). In addition, three
nas and crabs and standing stocks of algae were mea-
main currents converge on the archipelago, and can
sured at these two sites and at four sites at Punta Nun ez,
˜
shift in position and intensity, resulting in fluctuating
gradients of temperature and productivity. Even larger- the latter being selected to span the spectrum of algal
scale disturbances are associated with El Nin o-South-
˜ availability during the El Nino (see Appendix A for
˜
ern Oscillation (ENSO) events, which occur at intervals map of sites). The tides in Galapagos are semidiurnal,
´
of 3–7 yr and are characterized by unusually warm spanning 1.8–2.5 m (Houvenaghel and Houvenaghel
waters, a rise in sea level, greater wave action, and a 1977). Our studies were confined to the two lower-most
depletion of nutrients that reduces primary productivity intertidal zones on the shore, which Wellington (1975)
and ultimately affects all trophic levels (Cane 1983, termed the midlittoral (0.5–1.5 m) and the infralittoral
Houvenaghel 1984, Glynn 1988, 1990, Laurie 1990, zones (0.25–0.5 m above chart datum).
Bustamante et al. 2002). This dynamic environment The 1997–1998 ENSO began in March 1997, when
creates an ideal scenario for the study of top-down sea surface temperature (SST) rose to 3–5 C warmer
influences of herbivores on plant assemblages against than normal from about May 1997 to May 1998 before
a background of environmental stochasticity that alters declining precipitously, and from September 1998 con-
nutrient inputs and thus bottom-up effects. ditions returned to being close to the long-term norm
Our work reports on a series of observations and (see Appendix B for temperature records). Detailed
experiments conducted on rocky intertidal shores in studies of nutrient concentrations were independently
Galapagos, to examine the influence of fast-moving
´ undertaken at the time of our studies (Chavez et al.
herbivores on algal assemblages and to compare com- 1999), so we did not measure nutrients.
munity structure during and after El Nin o. The exper-
˜ We began observations in June 1997, initiated our
iments ran throughout the 1997–1998 ENSO event and caging experiment in August 1997, and ran the caging
for a year after its passage, and monitoring of control experiment for a year in which sea temperatures
plots continued to February 2003. reached their greatest departures from normal (termed
Our goals were to assess the relative influences of the ‘‘El Nino period’’), and the following year when
˜
herbivory by mobile species and temporal differences temperatures were again normal (‘‘post-El Nin o’’). ˜
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 113
FIG. 1. Design of the experimental treatments to test the effects of grazers. The notation indicates grazers that were
excluded ( ) or could gain access ( ). ‘‘Crabs’’ refers to Grapsus grapsus; ‘‘some’’ indicates that only a portion of that
group of grazers would have gained access.
Monitoring of control sites continued for a further 4.5 centage of composition of all identifiable items and
yr of ‘‘normal’’ conditions. estimating the total volume. Gut fullness was calcu-
lated as ([volume of gut contents/volume of stomach]
Monitoring densities of consumers 100).
Densities of marine iguanas and sally lightfoot crabs Selectivity of G. grapsus for different functional
at Sites 1 and 2 were censused once every three months groups of algae was assessed by comparing their con-
during low spring tides from August 1997 to September tribution to the diet with their contribution to the per-
1999 and again in July 2001, five instantaneous counts centage of cover in control plots during weeks 2, 4,
being made in areas of 40 5 m at 15-min intervals. and 8 of the caging experiment (when diet was as-
Counts were made with binoculars at a distance of 100 sessed). Selectivity was measured by the odds ratio, O
m to avoid disturbing the animals. Fish were censused p1q1/p2q2, where p1 the percentage of diet con-
during high tide by snorkeling a 15 2 m transect tributed by a particular taxon, p2 the percentage of
parallel to the shore, at seven roughly equally spaced that taxon available in the environment, q1 the per-
times between April 1998 and April 1999. Logistic centage of diet composed of all other taxa, and q2
constraints prevented replication within sites. Seden- the percentage of all other taxa available in the envi-
tary grazers (e.g., chitons, winkles, and limpets) were ronment. Expressed as ln(O), this yields positive values
always scarce and were not a central focus of our study. for items that are selectively consumed relative to their
To quantify their abundance, they were surveyed once abundance, and negative values for those selected
in June 2001 in 10 1-m2 quadrats in each of three shore against.
heights at each site (0.5, 1.0, and 2.0 m above chart
datum). The first two of these shore heights overlapped Densities of iguanas and crabs relative
with the zone in which our experiments were done. We to algal standing stock
surveyed the uppermost zone in case there were species Densities of iguanas and the crab G. grapsus were
that could migrate down into the low shore during high assessed in August 1997 at Sites 1 and 2 and at four
tide. In calculating the sedentary herbivore abundance sites 100 m apart at Punta Nunez, from instantaneous
˜
and biomass, we used only the data from the two lower scans of five areas of 10 m2 in each site. Ash-free dry
zones. To approximate biomasses, densities were mul- mass (AFDM) of algae was determined from five 100-
tiplied by the estimated wet masses of each species. cm2 samples per site, scraped to bedrock and ashed at
Estimates for fish were obtained from Fishbase (avail- 360 C for 12 h.
able online),5 and averaged 91.05 g per herbivorous
Caging experiment
fish. For iguanas we used 883 g for small individuals
and 1200 g for large individuals feeding in the intertidal The caging experiment was set up at Sites 1 and 2
zone (Wikelski et al. 1993). Direct measurements yield- in the low intertidal at heights that were equivalent at
ed values for Grapsus grapsus (mean 24 g), Pachy- both sites, spanning 0.3–0.8 m above chart datum, and
grapsus transversus (mean 0.8 g), and sedentary her- the positions of different treatments were randomly as-
bivores (4.8 g). signed within this height range. The cages and fences
were designed to differentially exclude iguanas, G.
Crab diet grapsus, and fish to test their relative effects on algal
The diet of G. grapsus was determined for 30 in- composition and abundance. Each treatment plot was
dividuals at each experimental site. Stomach volumes 30 30 cm and was surrounded by an iron fence,
were measured by displacement and the stomach con- totally or partially covered by stainless steel netting of
tents then extracted, spread on a slide, and compressed mesh size 1.0 1.0 cm (Fig. 1), with five randomly
to a thickness of 0.5 mm before measuring the per- interspersed replicates per site. Pilot studies indicated
that 96% of the diversity of the shore was incorporated
5 www.fishbase.org in five replicate plots of this size. The total exclosure
114 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
treatment (T) excluded all grazers larger than 1 cm. and the mean lengths of the upright branches measured
The half wall roof (H) left a gap too small for iguanas (n 5 independent replicates).
and fish to enter but allowed crabs access; only crabs
were ever seen in these plots. The wall treatment (W) Univariate data analyses
effectively excluded iguanas, but fish and at least some
For univariate analyses, the percentage of cover of
crabs could enter it. The roof treatment (R) constituted
sessile organisms was arcsine transformed for the pur-
a procedural control that theoretically allowed all graz-
poses of ANOVA to meet assumptions of normality
ers access while shading the plot like other roofed ex-
and equality of variance. There were no differences in
clusion plots. All three types of mobile grazers were
the abundance of sessile organisms among treatments
observed in the roof controls, and no significant dif-
at the beginning of the experiment in August 1997
ferences were recorded between the responses of algae
(two-way ANOVA with site as a random factor and
in them and in the controls (C) for four of the seven
functional groups examined. In the remaining three, treatment as a fixed factor; P 0.05). Because the data
differences emerged in only two to four instances (out for successive measures of each plot were not inde-
of 18 sets of observations). Thus, the roof is unlikely pendent, a repeated-measures (RM) ANOVA (using
to have had serious shading effects. None of the treat- SYSTAT 9; Systat Software, Port Richmond, Califor-
ments excluded the crab Pachygrapsus transversus nia, USA) was used to test the effects of site, treatment,
(mean densities 10–20/m2) or small blennies, which and temporal variation on algal abundance, with site
could pass through the mesh. We assumed that their and treatment (between subjects) as random and fixed
effect was similar in all treatments. No sedentary ben- factors, respectively. Time was analyzed as a within-
thic grazers were ever recorded in any of the treatments. subject effect. This model assessed the percentage of
The percentage of cover of the following functional variation ascribable to each of the factors (and their
groups of macroalgae and sessile invertebrates was interactions). Mauchly’s tests were run for all RM AN-
scored every three months in quadrats with 400 inter- OVAs. Adjusted degrees of freedom were applied to
secting grid points: algal crusts (Gymnogongrus and the F tests (using Greenhouse-Geisser and Huynh-Feldt
Lithothamnium), foliose greens (Ulva and Enteromor- corrections) for data sets that did not meet the sphe-
pha), filamentous greens (mostly Chaetomorpha an- ricity assumptions (Kuehl 2000). Post-hoc contrasts,
tennina), red algal turf (intermingled species forming using Bonferroni/Dunn adjusted probabilities (P
a dense turf 10 mm tall), filamentous brown algae 0.005 for significant differences) identified differences
(notably Giffordia mitchelliae), articulated corallines among the main effects of treatments on each sampling
(mainly Jania spp.), the anemone Isoactinia sp., and date. Successive measurements of frond lengths were
the barnacle Tetraclita milliporosa. Frond lengths of independent because the chance of sampling fronds
Ulva, red algal turf, filamentous greens, and articulated coming from the same plants or the same fronds on
corallines (n 5 independent replicates per plot) were successive dates was remote. The data were also normal
measured monthly initially and then once every three and had equal variances; so two-way ANOVA was ap-
months for two years. plied to untransformed data for each site, with time and
treatment as variables.
Aquarium experiments on the effects Diversity (Shannon-Weiner H ), species richness
of crabs on algae (Margalef’s d ), and evenness (Pielou’s J ) were based
The effects of G. grapsus on algal communities were on percentage of cover of functional groups, using
additionally determined in aquarium experiments. PRIMER version 5.2.2 (PRIMER-E, Roborough, UK).
Twelve rocks (upper surface areas 725 cm2) were To test the effect of grazing, site, and time on these
prized from the low shore adjacent to Site 1, installed measures, repeated measures ANOVA was employed,
in separate cages (50 40 25 cm with a 2-mm mesh), using the procedure described for percentage of cover.
and held in outdoor aquaria with a continuous supply Differences between ENSO (from week 16 to week
of ambient seawater. Water depth was adjusted to sim- 40), post-ENSO (from week 52 to week 100), and nor-
ulate twice-a-day tides that left the upper half of each mal (only controls, weeks 195 and 277) conditions were
rock exposed for 2 h coincident with natural low tides. tested using ANOVA and post-hoc Bonferroni/Dunn
One crab was added to each of six cages; the remaining adjusted probabilities contrasts. The results must be
cages lacked crabs. The rock area per crab was equiv- interpreted with caution, given the possibility of a type
alent to 10 crabs/m2, the highest density recorded in I error, as normality and homogeneity of variance were
the field. not always observed (Bartlett’s and Cochran’s C test).
The upper surfaces of the rocks were monitored on However, provided the design is balanced, ANOVA is
days 0, 3, 6, 16, 27, and 33 to compare the changes in not generally affected when these assumptions are not
cover of encrusting versus erect forms. During the ex- met (Day and Quinn 1989), and graphical inspection
periment, the normally crustose Gymnogongrus devel- of the data (Quinn and Keough 2002) indicated that
oped into an erect stage with upright knobbly branches. the heterogeneity of variance was not sufficient to com-
Cover of the crustose and erect phases was estimated, promise the ANOVAs.
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 115
Multivariate analysis
Community structure within each treatment was an-
alyzed by averaging the percentage of cover data for
the five replicates of each treatment for each sampling
date; the data were then double root transformed and
analyzed by Bray Curtis similarity and nonmetric mul-
tidimensional scaling (MDS). Clusters recognized in
the MDS were based on 80% similarity cutoff in the
Bray Curtis similarity. SIMPER (similarity percentage)
was used to calculate the contributions of each func-
tional group to dissimilarities among clusters.
RESULTS
Consumers
On average, 0.10 0.02 marine iguanas/m2 were
recorded at both sites (mean SD). Densities fluctu-
ated, but tended to decline until near the end of the
observations (Fig. 2). Densities of G. grapsus foraging
during low tide averaged 0.44 0.10 individuals/m2
at Site 1, and 0.84 0.21 individuals/m2 at Site 2. Both
sites initially had similar densities of G. grapsus (0.4
individuals/m2), followed by a decline until September
or December 1998. Substantial increases occurred after
the El Nino period, mainly due to the appearance of
˜
juveniles, with numbers at Site 2 rising to double those
at Site 1 (Fig. 2).
G. grapsus consumed mainly red algal turf and Ulva,
but also erect thalli of Gymnogongrus, and small
amounts of filamentous greens and articulated coral-
lines (Fig. 3A). Differences between the sites reflected
relative availability of algae. Filamentous brown algae
FIG. 2. Densities of marine iguanas Amblyrhynchus cris- were virtually absent from both diet and habitat at Site
tatus (upper panel) and sally lightfoot crabs Grapsus grapsus
(lower panel) at Sites 1 and 2. 2, and articulated corallines were near-absent at Site 1
(Fig. 3B). The selectivity indices (Fig. 3C) indicated a
strong preference for red algal turf and Ulva. Erect
thalli of Gymnogongrus were positively selected when
present. The crabs displayed relatively neutral selec-
tivity for articulated corallines and filamentous green
FIG. 3. (A) Percentage composition of gut contents of Grapsus grapsus, (B) availability of algal functional groups, and
(C) selectivity indices indicating preference for particular functional groups (positive values) or avoidance of them (negative
values). In (A) and (B), values are means, and error bars indicate SE.
116 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 4. Densities of herbivorous fish, iguanas, sally lightfoot crabs, and sedentary herbivores compared across times
(mean 95% CI), and approximate conversions to whole wet biomass (g/m2).
algae, and avoided filamentous brown algae, encrusting space appeared intense, unoccupied rock averaging
Gymnogongrus and encrusting corallines. only 0.9%. Site 1 was dominated by crustose Gym-
Nineteen fish species were recorded, including eight nogongrus (40–85% of cover) and red algal turf (ini-
herbivores and three omnivores (see Appendix C for tially 28%). Site 2 was at first dominated by red algal
species list and densities). Numerically, fish were the turf (approximately 75%). Gymnogongrus was scarce
most abundant mobile herbivores (collectively aver- there at the start, but rose to 50–70%.
aging 1.4 individuals/m2), with sally lightfoot crabs In the unmanipulated control plots, three types of
averaging 0.64 individuals/m2 and iguanas 0.10 indi- temporal change occurred (Fig. 5). First, encrusting
viduals/m2 (Fig. 4) Gymnogongrus, Ulva, and Enteromorpha declined or
Twenty-three species of sedentary benthic consum- were scarce during El Nino, and then became more
˜
ers occurred at the two sites, but none was ever abun- abundant. Second, several groups were most abundant
dant (see Appendix D for details of species and den- during the El Nino. Giffordia mitchelliae was absent
˜
sities). Six species of herbivores occurred in the zones initially but increased to 40–60% cover, replacing the
previously dominant crustose Gymnogongrus, and then
where we worked, but were sparse. Chitons and limpets
disappeared. At Site 2, red algal turf, Chaetomorpha
were rare (0.012 individuals/m2). Sea urchins never
antennina, and articulated corallines also peaked dur-
grazed in the intertidal zone. Thus, sedentary herbi-
ing the El Nino. The barnacle Tetraclita milleporosa
˜
vores were scarce (collectively 0.6 individuals/m 2) declined from 14% to 2% in the latter half of the
but mobile consumers were abundant (totaling 2.14 in- study. Encrusting corallines covered 29% early in the
dividuals/m2). Biomass estimates emphasize the scar- El Nino, but declined to 3% and began to recover
˜
city of slow-moving benthic herbivores (Fig. 4). only three years after the El Nino. Third, erect Gym-
˜
The densities of both crabs and iguanas (y, individ- nogongrus and the anemone Isoactinia displayed no
uals/m2) were linearly related to algal standing stocks temporal trends (neither ever exceeding 3% cover).
(x, g AFDM/cm2) when the two sites in Academy Bay The most obvious changes in cover occurred during
and the four sites at Punta Nunez were examined. (For
˜ the El Nino, when G. mitchelliae and C. antennina
˜
iguanas, y 1.636x 0.189; n 6, r2 0.986; for increased and crustose forms diminished. Once tem-
crabs, y 6.848x 0.496; n 6, r2 0.956.) The peratures dropped, Ulva and Enteromorpha replaced
gut fullness index for crabs was 83.3 11.7 (mean G. mitchelliae and C. antennina, rising to cover 48%
SE ) for Site 1, and 50.0 8.7 at Site 2, the differences of primary space; and encrusting Gymnogongrus re-
being significant (t test; t 27.1, P 0.001), sug- covered to become the eventual dominant. Bare space
gesting that food was more difficult to obtain at Site accounted for up to 20% during the El Nino but was
˜
2, despite algal biomass being higher there. almost nonexistent thereafter.
From week 76 onward, almost all of the functional
Natural changes in benthic community structure groups changed scarcely at all in the control plots dur-
Abundances of functional groups were initially com- ing the 4.5 yr of monitoring after El Nino passed.
˜
parable among treatments (ANOVA, df4,40, P 0.05 Changes in community structure after
in all cases). Differences did exist between sites ( P herbivore manipulation
0.001), but there were no significant interactions be- Temporal, site, and grazer effects significantly af-
tween treatment and site (P 0.05). Competition for fected the abundance of all functional groups except
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 117
Lithothamnium (Fig. 5). Temporal differences usually clined over time; but neither the main effects nor their
had the greatest effect, accounting for 21–37% of the interactions had any influence on abundance.
variability. Site effects were also substantial (16–32%), The effects of grazer treatments on all functional
and grazer treatments less important (1–10%), but there groups were clear when the data were contrasted be-
were significant interactions in most cases (see Ap- tween the El Nino and post-El Nino periods (Fig. 6),
˜ ˜
pendix E for details of statistical analyses). and most obvious (and significant) at whichever site
Percentage of cover of encrusting Gymnogongrus supported the highest cover of each group. Four types
differed significantly between El Nin o and post-El Nino
˜ ˜ of responses emerged. (1) Encrusting Gymnogongrus
periods, increasing in the latter (Fig. 5A). At Site 1, was promoted by grazing and suppressed during the El
only total exclusion of all grazers led to a decrease in Nino. (2) Chaetomorpha, Ulva, and probably Entero-
˜
this crust. At Site 2, both partial and total exclusions morpha were inhibited by grazing, with Chaetomorpha
of grazers had this effect. Cover rose to 79% in the peaking during El Nino and the other two after its pas-
˜
controls, but only 6–26% in all other treatments. Over- sage. (3) Red algal turf and articulate corallines yielded
all, encrusting Gymnogongrus declined or was held at ambiguous results, with a suggestion of higher values
low levels during the El Nino period (in all treatments)
˜ at intermediate levels of grazing, at least at Site 2 where
but increased thereafter, most obviously in treatments both were most abundant. El Nino promoted red algal
˜
involving no exclusion or partial exclusion of grazers. turf but had no effect on articulated corallines. (4) En-
Ulva and Enteromorpha responded oppositely to crusting corallines and G. mitchelliae showed no con-
Gymnogongrus, with abundance peaking inside the to- sistent responses to grazing but both peaked during the
tal exclusion treatments (Fig. 5B and C). Site and tem- El Nino.
˜
poral differences were large, and grazing effects sig-
nificant but of lesser magnitude. At both sites, total Effects of grazers on algal size composition
exclusion of grazers increased the cover of Ulva during Differences in frond length were initially nonexistent
the post-El Nino period from week 52 onward (Fig.
˜ among treatment plots (ANOVA, P 0.25 in all cases),
5B), and control plots had low values. Enteromorpha but soon emerged and persisted for 8–52 wk (Fig. 7;
was only abundant at Site 1, where it responded neg- for detailed data and ANOVA, see Appendix F).
atively to grazing (Fig. 5C). In short, Ulva and Enter- Frond lengths of Ulva (Fig. 7A) were significantly
omorpha remained scarce in all treatments during the affected by grazing (P 0.001), being inversely related
El Nino period, but thereafter increased in treatments
˜ to grazing intensity most of the time at both sites. Red
where grazing was diminished or prevented. algal turf (Fig. 7B) showed similar clear-cut responses
No consistent grazing effect was observed for Gif- to grazing for the first 40 wk at Site 1 (P 0.001) and
fordia mitchelliae, with opposite trends at the two sites for 16 wk at Site 2. Chaetomorpha antennina (Fig. 7C)
(Fig. 5D). Most of its variation was explained by time showed striking effects of grazing up to weeks 40–52.
(60%), with treatment, site and interactions between Articulated corallines (Fig. 7D) showed no obvious pat-
factors explaining no more than 7%. Its proliferation terns, although roofed treatments (roof and half wall
was thus clearly linked to El Nino. ˜ roof) had high values in weeks 4–28, similar to the
Red algal turf was initially an important component, pattern for percentage of cover (Fig. 5H).
but progressively declined, and levels were signifi- Time and site always had significant effects, as did
cantly lower post-El Nino than during El Nino (Fig.
˜ ˜ interactions between factors. The percentage of vari-
5E). Grazing had no effect (P 0.05). Thus, cover of ation in the model significantly explained by grazing
red algal turf was more abundant at Site 2 than Site 1, (0.2–9.9%) was always less than that attributable to
and more abundant during the El Nino than afterward,
˜ time (21.0–59.9%) or site (0.3–32.3%). Without ex-
but was virtually unaffected by grazing. ception, the largest responses to grazing were linked
Chaetomorpha antennina (Fig. 5F) did not differ to periods when each taxon was most abundant.
among treatments or sites. Temporal effects dominated,
with Chaetomorpha only being abundant during El Algal diversity
Nino.
˜ There were initially no differences in any of the in-
Articulated coralline algae were virtually absent dices of diversity among treatments. By week 16, how-
from Site 1 (Fig. 5G). At Site 2, grazing had a small ever, diversity, evenness, and richness had increased at
and marginally significant effect. Percentage of cover both sites due to the addition of species such as Gif-
was initially small, but increased as the El Nin o ma-
˜ fordia mitchelliae and Chaetomorpha antennina. After
tured, particularly inside the treatments that provided about weeks 40–52, all three indices progressively de-
partial protection against grazing (roof or half wall clined at both sites and in all treatments, mainly due
roof), and declined post-El Nino. Grazing intensity was
˜ to increased dominance by encrusting Gymnogongrus
not correlated in any simple manner with percentage at Site 1 and by Ulva at Site 2, and because of declines
of cover, with the controls and total exclusion plots in red algal turf, barnacles, and Lithothamnium.
(highest and lowest intensities of grazing) often having Site 1 had less species richness than Site 2 (repeated-
the lowest cover. Encrusting corallines (Fig. 5H) de- measures ANOVA, P 0.001; see Appendix G for
118 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 5. Changes in algal cover (mean SE ) in experimental plots at Sites 1 and 2, August 1997–September 1999, and
in control plots, August 1997–February 2003: (A) encrusting Gymnogongrus sp., (B) Ulva sp., (C) Enteromorpha and
associated diatoms, (D) Giffordia mitchelliae, (E) red algal turf, (F) Chaetomorpha antennina, (G) articulated corallines, and
(H) Lithothamnium sp. The significance of main effects and their contributions to percentage variance pooled across sites
(% var) are shown on the right (see Appendix E for statistical analyses.) Solid horizontal bars indicate times when grazing
effects were significant; dashed bars when they were not (post hoc contrasts; Bonferroni/Dunn adjusted to P 0.005).
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 119
FIG. 5. Continued
120 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 6. Grazing effects (G) on the percent cover of algae comparing ENSO (December 1997–June 1998, solid circles)
vs. post-ENSO (September 1998–September 1999, open circles) conditions (E), and the interaction between grazing and
ENSO effects (GE) (ANOVA, *P 0.05, ** P 0.01, *** P 0.001; NS, P 0.05) at Sites 1 and 2. Letters on the x-
axis represent different treatments (T, total exclosure; H, half wall roof; W, wall; R, roof; and C, control). Within sites,
treatments that share the same lowercase letter (just above the x-axis) were not significantly different (Bonferroni post hoc
tests, P 0.05). Where interactions are reported, arrows indicate the period when differences in grazing effect were found.
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 121
FIG. 7. Differences in the lengths (mean and SE) of algal fronds: (A) Ulva sp., (B) red algal turf, (C) filamentous green
algae (mainly Chaetomorpha antennina), and (D) articulated corallines (which were absent from Site 1). Numbers next to
data points indicate values for truncated portions of the graphs.
statistical analyses and Appendix H for temporal pat- case) were significantly affected by grazing, being in-
terns of diversity between treatments); but diversity versely related to grazing intensity (Fig. 8).
and evenness did not differ between sites (P 0.05).
Grazing affected all three diversity indices (P 0.001). Multivariate analysis of community structure
Total exclusion plots usually had high richness and Treatments were classified into Bray-Curtis hierar-
diversity values compared to other treatments ( P chical clusters and by MDS (see Appendices I and J),
0.005; Fig. 8). The controls (which were most intensely based on the mean percentage of cover of functional
grazed) yielded consistently low values, showed the groups in each treatment on each date. There was a
greatest declines post-El Nino, and had the lowest final
˜ clear pattern dividing the samples into temporal clus-
values. ters, rather than clusters based on treatments. For Site
Overall, contrast analyses showed that all three in- 1, seven clusters emerged, all distinguished by the time
dices were significantly greater during the El Nin o (par-
˜ when samples were taken rather than by treatment. A
ticularly in total exclusion plots), and (in all but one similar predominantly temporal pattern occurred at Site
122 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 8. Grazing effects (G) on species richness (Margalef’s index d ), evenness (Pielou’s evenness index J ) and diversity
(Shannon-Wiener diversity index H ) comparing ENSO (December 1997–June 1998, solid circles) vs. post-ENSO (September
1998–September 1999, open circles) conditions (E) and the interaction between grazing and ENSO effects (GE) (ANOVA,
*P 0.05, ** P 0.01, *** P 0.001; NS, P 0.05). Diversity indexes were derived from percent cover of functional
groups in experimental plots at Sites 1 and 2. Where interactions are reported, arrows indicate the period when differences
in grazing effect were found. Letters on the x-axis represent different treatments (T, total exclosure; H, half wall roof; W,
wall; R, roof; and C, control). Within sites, treatments that share the same lowercase letter (just above the x-axis) were not
significantly different (Bonferroni post hoc tests, P 0.05).
2, where six clusters were identified. Clusters I to V ditions (clusters V–VII). At Site 2, early domination
comprised successive sampling times from weeks 0– by red algal turf in phase 1 (cluster I) gave way to a
100. The only case in which grazing treatments influ- diversification and an expansion of G. mitchelliae in
enced the formation of clusters was the controls. At phase two (clusters II and III), whereas in phase three
weeks 28 and 64, the controls did not cluster with other Ulva/Enteromorpha and encrusting Gymnogongrus
treatments for those dates, and the controls for weeks dominated and diversity declined as ephemeral species
76, 88, and 100 formed a separate group (cluster VI). diminished or disappeared (clusters IV–VI).
Thus, the controls tended to separate out, but apart from
this, grazing had no effect on the clustering of samples. Aquarium experiments with crab grazing
SIMPER identified the taxa characterizing and dis- Rocks held in aquaria developed different floras de-
tinguishing clusters (Fig. 9). At both sites, there were pending on whether they were exposed to crab grazing
three recognizable phases, between which there were or not (Fig. 10). Initially, the rocks were covered by
significant differences in the abundance of almost all roughly equal proportions of encrusting and erect al-
functional groups (post-hoc contrast analyses, P gae, with 5% bare space. Rocks exposed to G. grap-
0.001 for all groups except articulated corallines, for sus ( crab) became progressively dominated by crus-
which differences were not significant, P 0.05). At tose algae (both Gymnogongrus and corallines). Erect
Site 1, Gymnogongrus and red algal turf dominated the algae (notably Ulva, filamentous greens and red algae)
first phase (cluster I), corresponding to the early onset declined proportionally. In treatments lacking crabs
of El Nino, giving way in a second phase to a more
˜ ( crab), the opposite happened: erect algae increased
diverse array and increasing domination by G. mitch- and smothered crusts.
elliae (clusters II–IV) during the mature stage of El Gymnogongrus, the most important space occupier,
Nino. Finally, in phase three, encrusting Gymnogon-
˜ initially almost solely comprised an encrusting phase,
grus (initially with Enteromorpha/Ulva) became prev- which occupied 32% of the rock surface. In the crab
alent and diversity declined during post-El Nin o con-
˜ treatment, this crust increased to 44% (Fig. 11A), but
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 123
FIG. 9. Results of SIMPER (similarity percentage) analyses diagnosing functional groups characterizing clusters identified
by Bray-Curtis similarity and MDS (multidimensional scaling); see Appendices I and J. See Results: Multivariate analysis
of community structure for a definition of phases and clusters. Asterisks identify significant differences in abundance between
successive clusters (* P 0.05; ** P 0.001; *** P 0.001).
FIG. 10. Percentage cover of (A) crustose algae (corallines and Gymnogongrus) and (B) erect algae (filamentous red and
green algae and foliose green algae) in aquaria that either contained or lacked a crab. Values shown are means SE ( n
6 aquaria per treatment). Any differences between the sum of erect and crustose algae and 100% cover were contributed by
bare rock. Asterisks indicate significant differences between treatments with vs. without a crab (* P 0.05; ** P 0.01;
*** P 0.001).
124 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
in the crab treatment, it progressively transformed
into an erect phase (Fig. 11B). Furthermore, thallus
length of the erect phase increased fivefold in the crab
treatment, but remained stunted in the crab treatment
(Fig. 11C).
DISCUSSION
A central message emerging from the results is the
interplay between (1) bottom-up forces, particularly re-
lated to changes in nutrients (Chavez et al. 1999), but
also to other associated factors such as wave action and
sea level during different phases of the El Nin o cycle,
˜
and (2) top-down grazing effects. Moreover, differenc-
es between sites were greater than expected. Tropical
intertidal rocky shores have previously been general-
ized as communities dominated by encrusting forms,
in which temporal and spatial variation in sessile spe-
cies is minimal (Garrity and Levings [1981] and Lub-
chenco et al. [1984] for the coast of Panama; Lub-
chenco and Gaines [1981] and Brosnan [1992] for a
general review; Williams [1993] for the coast of Hong
Kong). This apparent homogeneity has been attributed
to the sustained combined effect of slow-moving ben-
thic grazers coupled with mobile consumers, particu-
larly fish, because they cover substantial ranges
(Gaines and Lubchenco 1982, Lubchenco et al. 1984,
Menge et al. 1985, Brosnan 1992). However, our study
showed substantial temporal and spatial differences in
composition and abundance. On average, 26.1% of the
variance in abundance of functional groups was attrib-
utable to the main effect of time, 13.3% to site but only
4.8% to grazer treatment.
Differences between sites
There were several background differences between
sites, including a high abundance of red algal turf at
Site 2 and its scarcity at Site 1, and the presence of
articulated coralline turf at Site 2 but not Site 1. Re-
sponses to grazers also differed betweens sites. Two
possible causes are differences in wave action and ther-
mal stress. Site 1 was more exposed to waves than Site
2, although the difference was not great. Intersite dif-
ferences in thermal stress could have influenced the
cover of erect algal species. We have no concrete ev-
idence of thermal differences, however, and regard
them as an unlikely explanation of differences in com-
munity structure between sites, as wave action (which
will mitigate thermal stress) was less at Site 2 where
erect algae were most common.
Grazing did seem to be more effective at Site 1,
FIG. 11. Percent cover (mean SE ) for (A) encrusting
Gymnogongrus sp., (B) erect thalli of Gymnogongrus sp., and because each of the individual types of grazers affected
(C) the length of thalli of erect Gymnogongrus sp. in aquaria algal cover there, whereas at Site 2 only the combined
that either held a single Grapsus grapsus or lacked a crab. effects of all grazers influenced algal cover. Almost the
Asterisks indicate significance of the crab effect (* P 0.05; same numbers of marine iguanas occurred at Sites 1
** P 0.01; *** P 0.001). and 2, but smaller individuals visited Site 1 than Site
2. Juveniles are known to be more affected by waves
than larger individuals and forage in more wave-pro-
tected areas or higher on the shore (Trillmich and Trill-
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 125
mich 1986, Wikelski and Trillmich, 1994). G. grapsus Lubchenco et al. 1984, Menge et al. 1985, Brosnan
was more abundant at Site 2 than Site 1 during the 1992). Seasonal shifts of algal composition have been
post-El Nino period, but the gut fullness of crabs was
˜ documented at Hong Kong, where crusts predominate
greater at Site 1 than Site 2. These variations do not in summer whereas filamentous and foliar algae pro-
necessarily reflect differences in wave action, because liferate in winter, profoundly altering food availability
both marine iguanas and G. grapsus feed during low for herbivores (Kennish et al. 1996). Our results con-
tide when wave action has little effect (Kramer 1967, trast with both of these perspectives because they re-
Buttemar and Dawson 1993). veal striking regime shifts between years, fueled by
In short, there were clear differences in community oceanographic anomalies and the bottom-up influences
structure and dynamics between sites, but the causes of El Nino. The magnitude of the interannual changes
˜
remain uncertain although there are plausible expla- was greater than that previously recorded for any other
nations. tropical shores, and is an important reminder of the
extent to which climate change may alter ecosystem
Temporal fluctuations in species composition structure and dynamics.
The clearest pattern was that all species underwent Our observations spanned both a year of El Nin o and
˜
substantial changes in abundance over time. Abun- four years of ‘‘normal’’ conditions. El Ninos recur at
˜
dance of all taxa except Lithothamnium was signifi- irregular intervals of five to 10 years. As a result, our
cantly different during versus after El Nin o, apparently
˜ work covered only a portion of the full cycle. This
related to high temperatures, low nutrients, and strong would probably have had the effect of overestimating
wave action during El Nino (March 1997 to June 1998),
˜ the effect of time relative to herbivory. Temporal
followed by relative stability after the return of ‘‘nor- changes would have been greatest during El Nin o, and
˜
mal’’ conditions (Chavez et al. 1999). These temporal community composition was relatively stable during
differences explained more of the variance in the data the Post-El Nino period. The relative influence of her-
˜
than any other factor, and multivariate analysis divided bivory might have increased had we run our experi-
the community into temporal clusters rather than clus- ments for an entire ENSO cycle. Indeed, the herbivory
ters based on treatment. treatments were continuing to diverge at the end of the
At both sites, three phases existed (Fig. 9). Phase 1 experiment. Nevertheless, our central conclusions re-
was characterized by domination by encrusting Gym- main robust: temporal changes were substantial and, at
nogongrus and/or red algal turf. Phase 2 heralded least over the period in which we operated, overshad-
ephemeral warm-water species, including Giffordia owed the effects of herbivory.
mitchelliae, which proliferated and aggressively over-
grew encrusting algae. Red algal turf, Enteromorpha Consequences of El Nino events
˜
and Ulva (all preferred items in the diet of iguanas and The shores of Galapagos do not have striking zo-
´
sally lightfoot crabs) diminished or disappeared. G. nation patterns (Hedgepeth 1969, Wellington 1984),
mitchelliae seems characteristic of El Nino periods and
˜ mainly due to the absence of obvious belts of algae,
has been linked to previous ENSO events (Laurie mussels or barnacles. In the past, dense mats of Sar-
1990). Phase 3 marked the exit of G. mitchelliae, with gassum and the endemic Bifurcaria ( Blossevillea)
a return to encrusting Gymnogongrus as a clear dom- galapagensis formed a conspicuous band low on the
inant, initially together with Ulva/Enteromorpha. shore (Houvenaghel and Houvenaghel 1977), but, after
Diversity indices were significantly higher during the the 1982–1983 El Nino, these species disappeared and
˜
El Nino because of the addition of warm-water ephem-
˜ have not been observed since (G. Kendrick, personal
erals, but declined thereafter due to the disappearance communication; L. Vinueza, personal observation).
of some taxa and increased domination by one group During the 1982–1983 El Nino, the endemic sea star
˜
(Fig. 8). Both sites were eventually dominated by crusts Heliaster cuminguii declined drastically (Hickman
(79–94% in control plots). No increases in warm-water 1998). Its distribution was closely associated with the
ephemerals occurred at any time in the 4.5 yr during barnacle Tetraclita milleporosa, an important part of
which control plots were monitored after the passage its diet (Wellington 1975, Hickman 1998). T. mille-
of El Nino, so their emergence during 1997–1998 was
˜ porosa declined during the 1982–1983 El Nino and, at
˜
not just seasonal, but rather related to the anomalous our sites, during and after the 1997–1998 El Nin o.
˜
oceanographic conditions (Chavez et al. 1999). The El Nino also seriously affects other marine com-
˜
lack of any significant change in control plots between munities in Galapagos. During the 1982–1983 ENSO,
´
July 1999 (week 100) and February 2003 (week 277) barnacles, gorgonians, giant scallops, ahermatypic cor-
implies stability over this period. Our data do not allow als, and fish declined drastically (Robinson 1985). Sev-
certainty on this point because sampling in the later eral species of birds failed to breed (Valle 1985), and
years was infrequent, but they do show that inter-annual corals were decimated by 97% (Glynn 1988, 1990).
differences were small after the passage of El Nino. ˜ Laurie (1985, 1990) and Laurie and Brown (1990a)
Previous tropical-shore studies have emphasized their recorded 50% mortality of marine iguanas, and drastic
inter-annual stability (Gaines and Lubchenco 1982, reductions of red algal turfs and foliose green algae,
126 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
which are important constituents of the diet of marine The barren aspect of Panamanian shores has been
iguanas and other intertidal grazers. The 1982–1983 El attributed to conspicuous slow-moving consumers (no-
Nino also featured domination by Giffordia mitchel-
˜ tably siphonariid and fissurellid limpets) that graze fo-
liae, never before recorded in the Archipelago (Laurie liose algae down to grazer-resistant forms (Menge et
1985). al. 1986a, b). In Galapagos, and specifically at our ex-
´
Proliferation of G. mitchelliae (and concomitant re- perimental sites, slow-moving grazers were rare and
ductions of edible species) has important implications, their biomass low. In particular, the near absence of
as the digestibility of G. mitchelliae is 21%, compared limpets at all intertidal rocky shores is striking. In short,
with 64% for foliar green algae and 78% for red algal sedentary intertidal grazers that exert an important in-
turf. Its high terpenoid content probably reduces the fluence on algae in other tropical areas are absent or
digestive capabilities of iguanas and may have con- scarce in Galapagos as a whole, and specifically at the
´
tributed to their declining condition and elevated mor- two experimental sites.
tality during El Nino (Laurie 1985, Laurie and Brown
˜
Diets of consumers
1990b). Our data showed that G. mitchelliae was also
avoided by Grapsus grapsus, although it is an impor- Dietary preferences of consumers will also affect
tant constituent of the diet of Grapsus albolineatus in their influence on intertidal communities. With respect
Hong Kong (Kennish 1996). G. mitchelliae was among to iguanas, four generalities emerge from the consid-
several species that increased their abundance or ex- erable literature on their diets (Nagy and Shoemaker
tended their range during the 1997–1998 El Nin o, par-
˜ 1984, Laurie 1985, Wikelski et al. 1993, Wikelski and
ticularly species with affinities to the Indo Pacific and Hau 1995). First, iguanas consume most algae, espe-
Panamic Province, which were probably transported cially foliar green algae. Second, they avoid certain
southward by a warm tongue of water that extended to species. For example, before the 1982–1983 El Nin o, ˜
the whole eastern Pacific (Ruttenberg 2000). Bifurcaria galapagensis was abundant, but apparently
The drastic changes in algal composition and edi- never eaten by iguanas. Third, Giffordia mitchelliae,
bility, accompanied by strong swells and extremely which rises to dominance during El Nin o periods, ap-
˜
high sea levels, reduced food availability and restricted pears to have negative effects on iguanas as described
intertidal feeding opportunities. The correlation be- above. Fourth, low-growing algae are difficult for igua-
tween algal biomass and densities of crabs and iguanas nas (and other herbivores) to graze. Consequently,
during El Nino also suggests they were aggregating
˜ crusts are rarely eaten (Wikelski and Hau 1995), and
where food was most available, implying that food was stand to benefit from the removal of superior compet-
limiting at sites where algal stocks were low. All these itors. Moreover, firmly attached erect algae such as red
factors would have contributed to the heightened mor- algal turf can be grazed down to a thin veneer, but are
tality of marine iguanas ( 50%) observed during difficult to eliminate (Lubchenco et al. 1984, Menge et
1982–1983 (Laurie 1990) and 1997–1998 (Romero and al. 1986b). Marine iguanas can completely remove
Wikelski 2001). After the 1997–1998 El Nino, G. ˜ weakly attached algae such as Ulva and Enteromorpha,
mitchelliae disappeared, being replaced by encrusting but cannot graze on turfs 1 mm in height (Wikelski
Gymnogongrus and edible groups such as Ulva, En- and Hau 1995). Our data showed that grazing reduced
teromorpha, and red algal turf. G. grapsus increased the height of red algal turf, but scarcely influenced its
greatly, and marine iguanas recovered fast, reaching percentage cover.
weights heavier than even those before the El Nin o.˜ Grapsus grapsus showed clear dietary preferences,
with crusts and filamentous browns (notably Giffordia
Types of consumers mitchelliae) being avoided. In contrast, Grapsus al-
bolineatus preferentially feeds on filamentous algae,
Herbivorous fish, sally lightfoot crabs, and iguanas including Giffordia ( Hinksia) mitchelliae in Hong
averaged densities of 1.4, 0.64, and 0.1 individuals/m2, Kong (Kennish 1996, Kennish and Williams 1997).
whereas slow-moving herbivores such as chitons, lim- Overall, grazers displayed four responses to algae.
pets, nerites, and urchins never exceeded 0.03 individ- Encrusting forms were negatively selected because
uals/m2. Their relative impacts on algae will be influ- they are inaccessible; red algal turfs and foliar greens
enced by many factors, including biomass, rate of con- were positively selected; articulated corallines (and
sumption, dietary preferences, duration of tidal for- perhaps filamentous greens) were eaten in proportion
aging, mobility, and the edibility of different algae to availability, and Giffordia mitchelliae was either
(Lubchenco and Gaines 1981). Most of these factors avoided or had adverse effects because of its low di-
lie outside the scope of this investigation, but approx- gestibility.
imate conversions to wet biomass yielded estimations Four corresponding responses to grazing existed
of 178.4, 89.4, 15.4, and 2.9 g/m2 for herbivorous fish, among the algae. (1) Encrusting Gymnogongrus was
iguanas, sally lightfoot crabs, and slow-moving graz- enhanced by grazing, probably as an indirect response
ers, respectively; providing a better perspective of their to grazer-induced reductions of foliar algae that would
potential influence. otherwise have overgrown them. The laboratory ex-
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 127
periments verified that by diminishing foliar and fila- such as Ulva, and indirectly promoted competitively
mentous algae, crabs could enhance encrusting Gym- inferior but grazer-resistant species such as crustose
nogongrus and encrusting corallines. In the field, en- Gymnogongrus. At Site 2, edible foliar algae were more
crusting and erect phases of Gymnogongrus are prob- abundant, and only the combined effects of all grazers
ably adapted to avoiding grazing and the threat of (in the control treatment) significantly reduced ephem-
overgrowth by erect algae respectively (cf. Lubchenco erals and promoted crusts. However, not all crusts re-
and Cubit 1980, Slocum 1980). (2) Some algae such sponded similarly; Gymnogongrus increased substan-
as Ulva, Enteromorpha, and red algal turf were dimin- tially, but Lithothamnium consistently decreased in
ished by grazing. All these algae are preferentially con- abundance and never recovered fully. No grazing effect
sumed by iguanas and crabs. (3) Articulated corallines was detectable for Lithothamnium, although its highest
were scarcely affected by grazing and were eaten in levels of cover were associated with particular times
proportion to availability. (4) G. mitchelliae was un- (December 1997) and treatments (control and roof) in
affected by grazing. which the combined cover of red algal turf, Gymno-
gongrus, and ephemeral algae was lowest. Low com-
Changes in community structure due to petitive ability and slow growth may have retarded the
herbivore manipulation recovery of Lithothamnium. Only three years after El
The effect of grazers fell into three phases. (1) Dur- Nino did its recovery begin.
˜
ing El Nino (March 1997 to June 1998), preferred food
˜ Red algal turf declined from initially moderately
species were scarce, being replaced by G. mitchelliae. high (Site 1) or very high levels of cover (Site 2) down
Heavy swells and high seas would additionally have to very low levels, but the intensity of grazing had
hindered grazing. Densities of G. grapsus were low, little effect on its cover most of the time, in keeping
and mortality of marine iguanas elevated (Romero and with Hay’s (1981) views than the colonial aggregations
Wikelski 2001). (2) In the late stages of El Nin o and
˜ formed by turfs reduce herbivory. There were signifi-
its aftermath (July 1998 to March 1999), ephemeral cant interactions between site and treatment. Grazing
algae (particularly Ulva) grew luxuriantly, reaching effects were undetectable at Site 1 where the cover of
frond lengths of 80 mm and close to 75% coverage red algal turf was low. At Site 2, red algal turf was
even in control plots exposed to all grazers. Chaeto- abundant, and increased in treatments with moderate
morpha antennina achieved frond lengths of 100 mm to high intensities of grazing, perhaps benefiting from
and cover of G. mitchelliae rose substantially. Densities reductions of palatable, competitively superior species.
of mobile grazers such as iguanas and crabs were still Lewis (1986) reported a reduction of turf and crustose
low, and high primary production could have improved species in experimental plots that totally excluded graz-
conditions for consumers and led to the greater body ers, and argued that turfs and crusts are adapted to
mass recorded for iguanas (Wikelski and Trillmich intense grazing as they dominate areas that are heavily
1997). Sessile organisms competed for space, with grazed.
ephemeral algae flourishing, crustose forms initially Articulated corallines were virtually absent from Site
declining and barnacles diminishing to the lowest lev- 1. At Site 2, they were more common, and increased
els recorded. (3) After nutrient and temperature levels within roofed treatments (H, R) at all times. There are
returned to normal (June 1999 onward), grazing effects three possible explanations for this. One is that grapsid
of marine iguanas, crabs, and fish were pronounced and crabs (which would have entered these treatments) se-
provoked changes in community structure, including a lectively removed competitively superior species such
reduction of Ulva and Chaetomorpha antennina, and as Ulva. Second, articulated corallines may have been
an increase in crustose forms. outcompeted in the total exclusion cages. Finally, in-
The magnitude of responses to grazing was greatest tense grazing in the control may have diminished ar-
when the cover of each functional group was high. This ticulated corallines. The net effect was that these algae
implies that during periods of low productivity, algae were most abundant at intermediate levels of grazing.
do not have the capacity to benefit from a reduction or Being heavily calcified, corallines represent a poor
elimination of grazing, and are probably limited by source of energy for grazers (Paine and Vadas 1969,
physicochemical conditions. Conversely, during peri- Duffy and Hay 1990). Relative to their abundance, up-
ods of high productivity dramatically increased growth right coralline algae were not an important part of the
was recorded in grazer-exclusion plots. During the El diet of G. grapsus or marine iguanas (Wikelski et al.
Nino the bottom-up effects of nutrient limitation (Cha-
˜ 1993; L. Vinueza, unpublished data).
vez et al. 1999) were prevalent, and rippled up the food
chain to affect consumers, but when ‘‘normal’’ con- Effect of grazers on algal diversity
ditions returned, the top-down effects of herbivores on Grazers can potentially increase or decrease diver-
algal composition and sizes increased. sity. Some herbivores reduce the landscape to few graz-
The effects of grazing differed between sites, being er-resistant competitively inferior encrusting species.
most obvious at Site 1 where the cumulative or indi- Examples include sea urchins (Paine and Vadas 1969,
vidual effects of grazers reduced ephemeral species Vance 1979), parrotfish (Lewis 1986, Hixon and Bros-
128 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
FIG. 12. Synthesis of the relative importance of temporal changes, differences between sites, and the effects of grazers
in explaining variance in algal abundance. Changes in Shannon diversity (H ) and the abundance of algae among phases are
averaged across sites and are based on the percent cover reported for control plots. Differences between the sites are averaged
across times and are also based on the percent cover of control plots. Effects of grazers are based on averages across both
times and sites, and they compare control plots with total exclusion plots.
toff 1996), and the diverse guild of consumers present reduction in diversity, particularly post-El Nin o, when
˜
in Panama (Menge and Lubchenco 1981, Lubchenco algae were abundant and growth rates high.
et al. 1984).
We recorded differences in diversity between sites Conclusions and broader implications
over time and among treatments (Fig. 8). First, Site 1 Our study provides new insight into the regulation
had lower species richness than Site 2. Second, sub- of tropical intertidal communities, differing in several
stantial temporal fluctuations in diversity indices oc- respects from previous descriptions (Gaines and Lub-
curred in all treatments. Initially, diversity was inter- chenco 1982, Menge et al. 1985, Brosnan 1992). The
mediate (phase 1, week 0). During phase 2 (weeks 16– near absence of sedentary grazers, significant differ-
20), when temperatures were high and nutrient levels ences between sites, and interannual variation related
low, diversity increased due to the emergence of to the large-scale disturbances of El Nino show that
˜
ephemerals such as Giffordia mitchelliae and Chae- factors additional to grazing have important effects in
tomorpha antennina. In phase 3 (week 52 onward), structuring Galapagos intertidal communities. Tem-
´
decreases of several species and increased dominance poral and intersite differences were substantially great-
by Enteromorpha/Ulva and/or Gymnogongrus were ac- er than grazing effects, at least over the period of our
companied by colder temperatures and high levels of study, reflecting the influences of large-scale oceano-
nutrients (Chavez et al. 1999). Thus, diversity declined graphic events (El Nino) and smaller-scale spatial ef-
˜
when grazing probably became more intense, with crab fects on community structure. This echoes the conclu-
densities rising sharply and the condition of iguanas sions of Sauer Machado et al. (1996) about tropical
improving. Brazilian shores, that although consumers do influence
Regional fluctuations of temperature and nutrients community structure, their effects are limited and var-
levels influenced diversity by addition or loss of iable. Fig. 12 summarizes the relative degrees to which
ephemeral species. Superimposed on this, grazing variances in the data were explained by the main effects
treatment had a significant but relatively small effect, of time (27%), site (13%) and grazing (5%). There
the clearest manifestation of which was to promote were, however, important interactions between time
dominance by crustose Gymnogongrus, with associated and treatment, with herbivores having less effect during
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 129
El Nino than after it. The bottom-up influence of El
˜ Buttemar, W. A., and W. R. Dawson. 1993. Temporal pattern
of foraging and microhabitat use by Galapagos marine igua-
´
Nino and the top-down effects of herbivores thus
˜
nas, Amblyrhynchus cristatus. Oecologia 96:56–64.
changed in relative importance over time. Cane, A. M. 1983. Oceanographic events during El Nino. ˜
In short, the results emphasize the profound trans- Science 222:1189–1195.
formation that global climate change may bring to ma- Chavez, F. P., P. G. Strutton, G. E. Friederich, R. A. Feely,
rine ecosystems, including shifts in composition, in- G. C. Feldman, D. G. Foley, and M. J. McPhaden. 1999.
Biological and chemical response of the equatorial Pacific
teractions, and productivity (Fields et al. 1993). The Ocean to the 1997–98 El Nino. Science 286:2126–2131.
˜
biota of Galapagos, having high endemicity and being
´ Cubit, J. 1984. Herbivory and the seasonal abundance of
isolated and positioned at the juncture of contrasting algae on a high intertidal rocky shore. Ecology 65:1904–
oceanographic currents, is delicately poised and prone 1917.
Day, R. W., and G. P. Quinn. 1989. Comparisons of treatments
to extinction in the face of global changes. If El Nin o
˜ after an analysis of variance in ecology. Ecological Mono-
conditions intensify or become more frequent, we can graphs 59:433–463.
anticipate greater environmental stochasticity, higher Dayton, P. K. 1971. Competition, disturbance, and commu-
temperatures, lower nutrient levels, reduced productiv- nity organization: the provision and subsequent utilization
of spaces in a rocky intertidal community. Ecological
ity and biomass, reductions in the densities of consum-
Monographs 41:351–389.
ers, extinctions of cooler-water species, and probably Duffy, J. E., and M. E. Hay. 1990. Seaweed adaptations to
long-term reductions in diversity. herbivory. BioScience 40:368–376.
These effects will have economic as well as ecolog- Fields, P. A., J. B. Graham, R. H. Rosenblatt, and G. N.
ical consequences, because many of the top consumers Somero. 1993. Effects of expected global climate change
on marine faunas. Trends in Ecology and Evolution. 8:361–
will be affected most, including fish that are harvested, 367.
and birds, iguanas, and sea lions that are charismatic Gaines, S. D., and J. Lubchenco. 1982. A unified approach
attractions for the lucrative ecotourism trade. to marine plant–herbivore interactions. II. Biogeography.
Annual Review Ecology and Systematics 13:111–138.
ACKNOWLEDGMENTS Garrity, S., and S. Levings. 1981. A predatory–prey inter-
actions between two physically and biologically con-
We thank the Galapagos National Park Service and the
´ strained tropical rocky shore gastropods: direct, indirect
Charles Darwin Foundation for the permits and support to and community effects. Ecological Monographs 5:267–
conduct this work. Part of this work comprised a thesis by 286.
L. Vinueza at the University of Wales in Bangor, supported Glynn, P. W. 1988. El Nino–Southern Oscillation 1982–1983:
˜
by a Chevening Scholarship and a student grant from the nearshore population, community and ecosystem respons-
Foundation for Science and Technology of Ecuador to L. es. Annual Review of Ecology and Systematics 19:309–
Vinueza. Ray Seed, Jose Miguel Farina, John Witman, Bruce
˜ 345.
Menge, Paul Dayton, Peter Raimondi, and two anonymous Glynn, P. W. 1990. Coral mortality and disturbances to coral
reviewers constructively improved the manuscript. Paulina reefs in the tropical eastern Pacific. Pages 55–126 in P. W.
Guarderas, Paula Angeloni, Franz Smith, Adelaida Herrera, Glynn, editor. Global ecological consequences of the 1982–
Vanessa Francisco, and Monica Flores provided invaluable 83 El Nino–Southern Oscillation. Elsevier Oceanography
˜
field assistance and continual friendship. Funding for this Series 52. Elsevier, Berlin, Germany.
project was provided by the USAID to the Charles Darwin Hay, M. E. 1981. The functional morphology of turf-forming
Foundation, the Pew Charitable trusts Marine Conservation seaweeds: persistence in stressful marine habitats. Ecology
Fellowship program and the Pew Collaborative Initiative 62:739–750.
Fund Award to R. H. Bustamante. G. M. Branch was funded Hedgepeth, J. 1969. An intertidal reconnaissance of rocky
by the University of Cape Town, the National Research Foun- shores of the Galapagos. Wasmann Journal of Biology
dation, SANCOR, and the Andrew Mellon Foundation. G. 27(1):1–22.
M. Branch and M. L. Branch were kindly accommodated by Hickman, C. P. 1998. Guıa de campo sobre las estrellas de
´
Sol and Rodrigo Bustamante, renewing an old and cherished mar y otros equinodermos de Galapagos. Serie vida marina
´
friendship. This is contribution number 1033 from the Charles de Galapagos. Sugar Spring Press, Lexington, Virginia,
´
Darwin Foundation and contribution number 189 from PIS- USA.
CO, the Partnership for Interdisciplinary Studies of Coastal Hixon, A. M., and W. N. Brostoff. 1996. Succession and
Oceans, funded primarily by the Gordon and Betty Moore herbivory: effects of differential fish grazing on Hawaiian
Foundation and David and Lucile Packard Foundation. coral-reef algae. Ecological Monographs 66:67–90.
Hockey, P. A. R., and G. M. Branch. 1984. Oystercatchers
LITERATURE CITED and limpets: impact and implications. A preliminary as-
sessment. Ardea 72:119–206.
Brosnan, D. M. 1992. Ecology of tropical rocky shores: Houvenaghel, G. 1984. Oceanographic setting of the Gala-
plant–animal interactions in tropical and temperate lati- pagos Islands. Pages 43–54 in R. Perry, editor. Galapagos
tudes. Pages 101–131 in D. M. John, S. J. Hawkins, and key environments series: Galapagos Islands. Pergamon
J. H. Price, editors. Plant animal interactions in the marine Press, Oxford, UK.
benthos. Systematics Association special volume No. 46. Houvenaghel, G. T., and N. Houvenaghel. 1977. Aspectos
Clarendon Press, Oxford, UK. ecologicos de la zonificacion entre marea en las costas ro-
´ ´
Bustamante, R. H., G. M. Wellington, G. M. Branch, G. J. cosas de las islas Galapagos. Pages 79–100 in U. Eberhardt,
´
Edgar, P. Martinez, F. Rivera, and F. Smith. 2002. Out- editor. Compendio de ciencia en Galapagos. Publicacion
´ ´
standing marine features if the Galapagos Archipelago. de la Estacion Cientıfica Charles Darwin Isla Santa Cruz,
´ ´
Pages 61–72 in R. Benson-Smith, editor. A biodiversity Galapagos, Ecuador.
´
vision for the Galapagos Islands. Charles Darwin Foun- Kennish, R. 1996. Diet composition influences the fitness of
dation and World Wide Fund, Puerto Ayora, Galapagos, the herbivorous crab Grapsus albolineatus. Oecologia 105:
Ecuador. 22–29.
130 L. R. VINUEZA ET AL. Ecological Monographs
Vol. 76, No. 1
Kennish, R., and G. A. Williams. 1997. Feeding preferences sumers on sessile prey in the low zone of a rocky shore in
of the herbivorous crab Grapsus albolineatus and the dif- the bay of Panama: direct and indirect consequences of
ferential influence of algal nutrient content and morphol- food web complexity. Journal of Experimental Marine Bi-
ogy. Marine Ecology Progress Series 147:87–95. ology and Ecology 100:225–269.
Kennish, R., G. A. Williams, and S. Y. Lee. 1996. Algal Nagy, K. A., and V. H. Shoemaker. 1984. Field measurements
seasonality on an exposed rocky shore in Hong Kong and and food consumption of the Galapagos marine iguana,
the dietary implications for the herbivorous crab Grapsus Amblyrhynchus cristatus. Physiological Zoology 57:281–
albolineatus. Marine Biology 125:55–64. 290.
Kramer, P. 1967. Beobachtungen zur biologie und zum ver- Paine, R. T., and R. Vadas. 1969. The effects of grazing by
halten der klippencrabbe Grapsus grapsus L. (Brachyura sea urchins, Strongylocentrotus spp. on benthic algal pop-
Grapsidae) auf den Galapagos und am Ekuadorianischen ulations. Limnology and Oceanography 14:710–719.
Festland. Zeitschrift fur Tierpsychologie 24:385–402.
¨ Palumbi, S. R. 1984. Measuring intertidal wave forces. Jour-
Kuehl, R. O. 2000. Design of experiments: statistical prin- nal of Experimental Marine Biology and Ecology 81:171–
ciples of research design and analysis. Second edition. Dux- 179.
bury Press, Pacific Grove, California, USA. Quinn, G. P., and M. J. Keough. 2002. Experimental design
Laurie, A. 1985. The effect of the 1982–1983 El Nino on ˜ and data analysis for biologists. Cambridge University
Marine Iguanas. Pages 199–209 in G. Robinson and E. M Press, Cambridge, UK.
del Pino, editors. El Nino en las Islas Galapagos: El Evento
˜ ´ Robinson, G. 1985. Influence of the 1982–1983 El Nino on ˜
de 1982–1983. Publication of the Charles Darwin Foun- Galapagos marine life. Pages 153–190 in G. Robinson and
´
dation for the Galapagos Islands, Quito, Ecuador.
´ E. M. del Pino, editors. El Nino en las Islas Galapagos: El
˜ ´
Laurie, A. 1990. Effects of the 1982–1983 El Nino-Southern
˜ Evento de 1982–1983. Publication of the Charles Darwin
Oscillation event on marine iguanas (Amblyrhynchus cris- Foundation for the Galapagos Islands, Quito, Ecuador.
´
tatus Bell, 1825) populations on Galapagos. Pages 361– Romero, L. M., and M. Wikelski. 2001. Corticosterone levels
380 in P. W. Glynn editor. Global ecological consequences predict survival probabilities of Galapagos marine iguanas
of the 1982–83 El Nino-Southern Oscillation. Elsevier during El Nino events. Proceeding of the National Academy
˜
Oceanography Series 52. Elsevier, Berlin, Germany. of Sciences (USA) 98:7366–7370.
Laurie, W. A., and D. Brown. 1990a. Population biology of Ruttenberg, B. 2000. An unusual pulse of recruitment of two
marine iguanas (Amblyrhynchus cristatus). II. Changes in reef fishes in the Galapagos Islands coincident with the
´
annual survival rates and the effects of size, sex, age and 1997–1998 El Nino. Bulletin of Marine Science 67:869–
˜
fecundity on a population crash. Journal of Animal Ecology 874.
59:529–544. Sauer Machado, K. R. S., A. R. O. Chapman, and R. Coutinho.
Laurie, W. A., and D. Brown. 1990b. Population biology of 1992. Patch structure in a tropical rocky shore community
marine iguanas (Amblyrhynchus cristatus). III. Factors af- in Brazil: a mosaic of successional states? Ophelia 35(3):
fecting survival. Journal of Animal Ecology 59:545–568. 187–195.
Lewis, S. M. 1986. The role of herbivorous fishes in the Sauer Machado, K. R. S., A. R. O. Chapman, and R. Coutinho.
organization of a Caribbean reef community. Ecological 1996. Consumer species have limited and variable roles in
Monographs 56:183–200. community organization on a tropical intertidal shore. Ma-
Lubchenco, J. 1980. Algal zonation in the New England rine Ecology Progress Series 134:78–83.
rocky intertidal community: an experimental analysis. Slocum, C. J. 1980. Differential susceptibility to grazers in
Ecology 61:333–344. two phases of an intertidal alga: advantages of heteromor-
Lubchenco, J., and J. Cubit. 1980. Heteromorphic life his- phic generations. Journal of Experimental Marine Biology
tories of certain marine algae as adaptations to variation in and Ecology 46:99–110.
herbivory. Ecology 6:676–687. Trillmich, K. G. K., and F. Trillmich. 1986. Foraging strat-
Lubchenco, J., and S. D. Gaines. 1981. A unified approach egies of the marine iguana, Amblyrhynchus cristatus. Be-
to marine plant–herbivore interactions. I. Populations and havioral Ecology and Sociobiology 18:259–266.
communities. Annual Review Ecology and Systematics 12: Valle, C. 1985. Alteracion de las poblaciones del cormoran
´ ´
405–437. no volador, el pinguno y otras aves marinas en Galapagos
¨ ´
Lubchenco, J., B. A. Menge, S. D. Garrity, P. J. Lubchenco, por el efecto de El Nino 1982–1983 y su subsecuente re-
˜
L. R. Ashkenas, S. D. Gaines, R. Emlet, J. Lucas, and S. gereracion. Pages 245–258 in G. Robinson and E. M. del
´
Strauss. 1984. Structure, persistence and role of consumers Pino, editors. El Nino en las Islas Galapagos: El Evento
˜ ´
in a tropical rocky intertidal community (Taboguilla Island, de 1982–1983. Publication of the Charles Darwin Foun-
Bay of Panama). Journal of Experimental Marine Biology dation for the Galapagos Islands, Quito, Ecuador.
´
and Ecology 78:23–73. Vance, R. R. 1979. Effects of grazing by the sea urchin Cen-
Menge, B. A., and G. M. Branch. 2001. Rocky intertidal trostephanus coronatus on prey community composition.
communities. Pages 221–251 in M. D. Bertness, S. Gaines, Ecology 60:537–546.
and M. E. Hay, editors. Marine community ecology. Sin- Wellington, G. 1975. Medio ambientes marino costeros de
auer Associates, Sunderland, Massachusetts, USA. Galapagos. Un informe de recursos al departamento de
´
Menge, B. A., and J. Lubchenco. 1981. Community orga- Parque Nacionales y Vida Silvestre, Quito, Ecuador.
nization in temperate and tropical rocky intertidal habitats: Wellington, G. 1984. Marine environment and protection.
prey refuges in relation to consumer pressure gradients. Pages 247–263 in R. Perry, editor. Galapagos key envi-
Ecological Monographs 51:429–450. ronments series: Galapagos Islands. Pergamon Press, Ox-
Menge, B., J. Lubchenco, and L. Ashkenas. 1985. Diversity, ford, UK.
heterogeneity and consumer pressure in a tropical rocky Wikelski, M., B. Gall, and F. Trillmich. 1993. Ontogenetic
intertidal community. Oecologia (Berlin) 63:394–405. changes in food intake and digestion rate of the herbivorous
Menge, B. A., J. Lubchenco, S. D. Gaines, and L. R. Ash- marine iguana (Amblyrhynchus cristatus, Bell). Oecologia
kenas. 1986a. A test of the Menge-Sutherland model of 94:373–379.
community organization in a tropical rocky intertidal food Wikelski, M., and M. Hau. 1995. Is there an endogenous
web. Oecologia (Berlin) 71:75–89. tidal foraging rhythm in marine iguanas? Journal of Bio-
Menge, B. A., J. Lubchenco, S. D. Gaines, and L. R. Ash- logical Rhythms 10(4):335–350.
kenas. 1986b. Experimental separation on effects of con- Wikelski, M., and F. Trillmich. 1994. Foraging strategies of
February 2006 ´
HERBIVORY AND ENSO ON GALAPAGOS SHORES 131
the Galapagos marine iguana (Amblyrhynchus cristatus): richness and abundance in the presence of molluscan her-
adapting behavioral rules to ontogenic changes. Behaviour bivores on a tropical rocky shore. Journal of Experimental
128:255–279. Marine Biology and Ecology 167:261–275.
Wikelski, M., and F. Trillmich. 1997. Natural and sexual Williams, G. A. 1994. The relationship between shade and
selection on body size fluctuates between marine iguana molluscan grazing in structuring communities on a mod-
island populations. Evolution 51:922–936. erately-exposed tropical rocky shore. Journal of Experi-
Williams, G. A. 1993. Seasonal variation in algal species mental Marine Biology and Ecology 178:79–95.
APPENDIX A
Map of the study sites (Ecological Archives M076-005-A1).
APPENDIX B
Figure showing the comparison of mean monthly sea surface temperatures in the Galapagos (Ecological Archives M076-005-
´
A2).
APPENDIX C
Table showing mean abundances of fish/m2 at the two study sites (Ecological Archives M076-005-A3).
APPENDIX D
Table showing relative abundance of slow moving invertebrates (no. individuals/m2) at the two study sites (Ecological
Archives M076-005-A4).
APPENDIX E
Table showing the results of a repeated-measures ANOVA on the effect of treatment and site on the abundance of functional
groups of algae (Ecological Archives M076-005-A5).
APPENDIX F
Table showing the results of an ANOVA on the effect of treatment, site, and time on algal frond lengths (mm) (Ecological
Archives M076-005-A6).
APPENDIX G
Table showing the results of a repeated-measures ANOVA to test the effect of treatment and site on species richness,
diversity, and evenness (Ecological Archives M076-005-A7).
APPENDIX H
Figure showing temporal changes in species richness (Margalef’s index d), evenness (Pielou’s evenness index J ), and
diversity (Shannon-Wiener diversity index H ) (Ecological Archives M076-005-A8).
APPENDIX I
Dendrograms of Site 1 and Site 2 (Ecological Archives M076-005-A9).
APPENDIX J
Nonmetrical multidimensional scaling (nMDS) for Site 1 and Site 2 (Ecological Archives M076-005-A10).