Jenkins et al 05
MARINE ECOLOGY PROGRESS SERIES
Vol. 287: 77–86, 2005 Published February 18
Mar Ecol Prog Ser
Regional scale differences in the determinism of
grazing effects in the rocky intertidal
S. R. Jenkins1,*, R. A. Coleman2, P. Della Santina3, S. J. Hawkins1, 3,
M. T. Burrows4, R. G. Hartnoll5
1
Marine Biological Association, Citadel Hill, Plymouth PL1 2PB, UK
2
Marine Biology and Ecology Research Group, School of Biological Sciences, University of Plymouth, Drake Circus,
Plymouth PL4 8AA, UK
3
Biodiversity and Ecology Division, School of Biological Sciences, Southampton University, Southampton SO16 7PX, UK
4
Scottish Association for Marine Science, Dunstaffnage Marine Laboratory, Oban, Argyll PA37 1QA, UK
5
Port Erin Marine Laboratory, University of Liverpool, Port Erin, Isle of Man IM9 6JA, UK
ABSTRACT: Patellid limpets are dominant grazers on intertidal rocky shores of NW Europe with a
key role in structuring the eulittoral community. Localised loss of limpets and the subsequent reduc-
tion in grazing pressure is known to result in important changes in community structure, through the
development of canopy-forming macroalgae, and an associated increase in species diversity and
community complexity. The level of determinism in the community level response to localised loss of
patellid limpets was assessed at spatial scales from 100s of kilometres to 10s of metres and temporal
scales from weeks to months at mid-tide level of exposed rocky shores. Limpets were removed and
excluded from experimental plots to simulate localised limpet loss and appropriate controls estab-
lished. Experimental plots were established in replicate patches at 2 shores at each of 2 regional
locations, separated by approximately 500 km: the Isle of Man and SW England. Removals were con-
ducted on 2 dates within each of 2 seasons (summer and winter) and the community level response
monitored for a period of 12 mo. There was a clear effect of limpet loss at all spatial and temporal
scales, with rapid development of green ephemeral algae followed by a fucoid canopy. However, the
degree of determinism in the development of canopy-forming algae differed markedly between the
2 locations. At the northerly location, the Isle of Man, fucoid algae developed quickly and dominated
all areas of limpet exclusion; there was little variability between plots. In contrast, in SW England, the
abundance of fucoid algae was significantly lower and much more variable. Such geographic
changes in the development of macroalgae in the absence of the dominant grazer are discussed in
relation to rocky shore community dynamics and the latitudinal change in balance between grazers
and algae over the wave exposure gradient.
KEY WORDS: Herbivory · Large scale · Macroalgae · Rocky shore
Resale or republication not permitted without written consent of the publisher
INTRODUCTION community structure of intertidal rocky shores of
NW Europe (see Southward 1964, Hawkins & Hartnoll
Experimental work on rocky shores has frequently 1983, Hawkins et al. 1992 for reviews). Limpet-
demonstrated that intertidal grazers have the ability to removal experiments on the Isle of Man first demon-
almost completely control the species composition, strated that canopy-forming algae, particularly mid-
distribution and dynamics of algal communities (Lub- shore Fucus vesiculosus, were directly prevented from
chenco & Gaines 1981, Paine 2002). Patellid limpets establishing on wave-exposed shores by limpet graz-
are some of the best-known and most studied of ing (Jones 1946, 1948, Lodge 1948, Burrows & Lodge
marine herbivores, and have a profound effect on the 1950, Southward 1956, 1964). Experimental removal of
*Email: sjen@mba.ac.uk © Inter-Research 2005 · www.int-res.com
78 Mar Ecol Prog Ser 287: 77–86, 2005
limpets resulted in the rapid development of various of these including anemones, dogwhelks, littorinids,
ephemeral or opportunistic species (Enteromorpha isopods and amphipods. In addition, the development
spp., Ulva spp., Blidingia spp.) followed by growth of a of fucoid canopy algae on exposed shores, dominated
fucoid canopy which persisted for up to 5 yr. This suc- by barnacles and mussels, results in a distinct change
cessional sequence was confirmed in the British Isles, from a community dominated by the secondary pro-
following the widespread mortality of limpets on the duction of filter feeders to one dominated by primary
shores of west Cornwall following a massive oil spill production of macroalgae. Hence, the eulittoral zone
(Southward & Southward 1978, Southward 1979), and can change from a net importer of primary produc-
in follow-up experiments on the Isle of Man (Hawkins tion, when barnacle-dominated, to a net exporter,
1981a,b, Hartnoll & Hawkins 1985). when dominated by macroalgae (Hawkins et al.
In the British Isles and northern France, moderately 1992).
exposed shores support a mid-shore community com- We aimed to test the determinism of the assemblage
posed of patches of Fucus spp., barnacles and bare response (and particularly the development of a fucoid
rock, interspersed with grazing patellid limpets (Lewis canopy) to limpet loss at a realistic spatial scale, from
1964). On such shores, localised areas of low grazing small (0.5 × 0.5 m) patches. We tested the general
pressure allow ‘escapes’ of macroalgae to occur (Hart- hypothesis that loss of limpets from small patches at
noll & Hawkins 1985). Once beyond a size of around a mid-tide level of moderately exposed shores results in
few centimetres, such macroalgae are rarely con- macroalgal growth. Many factors may affect the likeli-
sumed by microphagous grazers such as limpets and hood of macroalgal development in the absence of
hence, persist to form clumps of canopy. The dynamics limpet grazing, including the supply of macroalgal
of this patchy mosaic community have been the subject propagules (Arrontes 2002), mortality of germlings in
of a number of empirical (Hawkins 1981a, Hartnoll & unfavourable microclimates (Brawley & Johnson 1991),
Hawkins 1985, Johnson et al. 1997) and modelling dislodgement of propagules by wave action (Vadas et
studies (Burrows & Hawkins 1998, Johnson et al. 1998) al. 1990) and grazing of propagules or germlings by
that have demonstrated the importance of individual other non manipulated grazers such as crustacean
limpet behaviour in maintaining the mosaic com- mesoherbivores (Brawley 1992). Given the variability
munity. of these and other ecological processes on rocky
The role of patellid limpets in structuring the mid- shores, from place to place, and time to time, we used
shore assemblage of the NE Atlantic at exposed sites a complex experimental design to test the general
is undoubted, though they have much less influence hypothesis applied over a number of spatial and tem-
amongst canopies, low on the shore (Jenkins et al. poral scales. In addition, we examined the effect of
1999b) or at sheltered sites (Jenkins et al. 1999a). latitude within the British Isles on the community
Ballantine (1961) observed a latitudinal trend in the response to limpet loss by conducting the experiment
balance between fucoid algae and limpet/barnacle- at 2 locations: the Isle of Man and SW England. In this
dominated areas down the coast of western Europe, way, we tested the hypothesis developed from the
with fucoids being restricted further into shelter in the original observations of Ballantine (1961) that loss of
south. Both he and subsequent authors (Hawkins & limpets would have a larger and more consistent effect
Hartnoll 1983, Hawkins et al. 1992) suggested that the at northern compared to southern latitudes.
balance between the effectiveness of grazers and the
ability of fucoids to grow, changed with latitude,
thereby affecting the probability of successful coloni- MATERIALS AND METHODS
sation by fucoids in areas of reduced grazing pres-
sure. The development of a fucoid canopy on barna- Study sites. Experimental work was undertaken at 2
cle-dominated shores has important implications for locations separated by approximately 450 km, the SW
community dynamics and energy flow. Macroalgal of England near Plymouth (50° 19’ N, 04° 06’ W) and the
canopies regulate community structure in a number of south of the Isle of Man (54° 5’ N, 04° 40’ W) in the Irish
ways: (1) by altering the quality and quantity of light Sea. Both locations were in areas of full salinity with
reaching the substratum (e.g. Reed & Foster 1984), maximum tidal ranges of 5 and 6 m, respectively. Two
(2) by whiplash or sweeping effects of fronds (e.g. Ve- moderately exposed rocky shores on open coastlines,
limirov & Griffiths 1979, Jenkins et al. 1999c) or (3) by separated by a minimum of 2 km, were selected at both
providing shelter from wave action (McCook & Chap- locations: Wembury and Heybrook Bay in SW Eng-
man 1991) and from physical extremes such as high land, Port St. Mary and Derbyhaven in the Isle of Man.
temperatures, desiccation or freezing (Leonard 2000). The main criteria in shore selection was topographical
Fucoid clumps provide a complex habitat and shelter simplicity, a gentle slope (< 30°) and domination at
for a wide range of animal species, the more obvious mid-tide level by an extensive cover of barnacles, with
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 79
abundant patellid limpets. The shores in both areas, monthly until termination of the experiment at 12 mo.
the Isle of Man (Southward 1953, Hartnoll & Hawkins At each sampling date, a 0.5 × 0.5 m quadrat, sub-
1985) and the Plymouth area (Colman 1933, Boalch et divided to give 49 intersection points, was used to
al. 1974) have been well described. At both shores at estimate the percentage cover of all macroalgae using
each location, the mid-shore, though dominated by the point intersect method (e.g. Benedetti-Cecchi et al.
barnacle cover, had patches of fucoid canopy algae. 1996). At selected sampling dates, all experimental
These were less common in the Plymouth area. Exper- quadrats were photographed. At least monthly, exper-
imental plots were located in the mid-shore, well imental plots were checked for damage to fences and
within the barnacle zone and at a mean tidal height repairs made. Any limpets that had invaded exclusion
above Chart Datum of 2.3 m at Wembury (range 1.9 to treatments plus any other ‘macro-grazers’ (littorinids,
2.9 m), 2.1 m at Heybrook Bay (range 1.9 to 2.2 m), topshells) were removed.
3.6 m at Port St. Mary (range 3 to 4.3 m) and 3.5 m at Data analysis. The experiment was designed to
Derbyhaven (range 2.6 to 4.8 m). allow partitioning of sources of variance using a 6-
Experimental design. The experiment consisted of factor mixed-model ANOVA, where the factors Sea-
3 treatments: (1) complete removal of all patellid lim- son, Location and Treatment were fixed, and Date,
pets and exclusion using 3 cm high fences of plastic- Shore and Patch were random. This analysis was used
coated wire mesh with 13 mm square openings, (2) a to determine the spatial and temporal consistency of
half-fenced treatment using the same fence structure treatment effects (i.e. differences between the control,
but only encompassing half the quadrat perimeter, and fenced and exclusion treatments). Significant effects of
allowing free movement of limpets, and (3) a control the treatment give no information on the magnitude of
treatment with only the 4 corners of the quadrat the effect and how this varies over different spatial and
marked by screws. The half-fenced treatment was temporal scales. For this, data from exclusion plots
used as a procedural control to determine whether the alone were used in a 5-factor mixed-model ANOVA.
use of fences in the exclusion treatment had any effect Prior to ANOVA, data were examined for heterogene-
on community succession, other than that caused by ity of variance using Cochran’s test and heterogenous
exclusion of limpets. data transformed appropriately. Significant factors
In order to fully explore the variability in the effect of were analysed further using SNK (Student Newman
limpet grazing on mid-shore community structure, this Keuls) multiple comparisons.
basic design was implemented over a number of In any experiment examining development of biota
spatial and temporal scales. Spatial variability was over a number of sampling dates, choice of the depen-
assessed at 3 scales, between locations (100s of kilo- dent variable is paramount to interpretation of the
metres), between shores within each location (kilo- experiment. In theory, a separate analysis could be
metres) and among patches within each shore (10s to performed for each sampling date but this would only
100s of metres). Temporal variability was assessed at lead to an over complex interpretation. For the key
2 scales, between seasons (summer and winter) and species in our analysis, 2 dependent variables were
between dates within seasons. At each shore, experi- selected, maximal cover during the 12 mo period and
ments were established during 2 different seasons, the area under the curve for each individual plot. Max-
summer 1996 and winter 1996/1997, and in each sea- imal cover indicates the peak response to perturbation,
son, 2 start dates were selected at random from within while the area under the curve integrates the pattern
a 3 mo period, with a minimum separation of 4 wk. of development for any particular species over time
Start dates were independently selected at all shores and, thus, takes into account the rate and temporal
over both locations. At each start date, 2 patches were trajectory of algal colonisation.
selected at each of the 2 shores at both locations.
Within each patch, nine 0.5 × 0.5 m quadrats were cho-
sen and the 3 treatments, replicated 3 times, were RESULTS
applied at random. All 8 patches for each shore were
selected in advance of experimental set-up, over a hor- General patterns of colonisation
izontal distance of between 250 and 400 m, with a min-
imum separation of 30 m between individual patches. Three main algal groups developed in experimental
The choice of the 2 patches, at each start date, from plots as a result of limpet removal: ephemeral green al-
amongst the 8 selected was made at random. gae made up of a mixture of Enteromorpha and Blid-
Maintenance and sampling of the experiment. Fol- ingia spp. with some Ulva spp., and Monostroma spp.,
lowing establishment of the experiment, sampling was soft algal crusts, predominantly Ralfsia spp. and fucoid
undertaken at regular intervals after each individual canopy algae made up almost entirely of Fucus vesicu-
start date, monthly for the first 6 mo and then bi- losus. The patterns of colonisation of these 3 groups
80 Mar Ecol Prog Ser 287: 77–86, 2005
SW ENGLAND ISLE OF MAN
SUMMER WINTER SUMMER WINTER
START START Ephemerals START START
100 100 100 100
a) b) c) d)
% cover
75 75 75 75
50 50 50 50
25 25 25 25
0 0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Soft crusts
100 100 100 100
% cover
e) f) g) h)
75 75 75 75
50 50 50 50
25 25 25 25
0 0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Fucoid canopy
100 100 100 100
i) j) k) l)
% cover
75 75 75 75
50 50 50 50
25 25 25 25
0 0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Months
Shore 1 Limpet exclusion Shore 2 Limpet exclusion
Shore 1 Half fenced Shore 2 Half fenced
Shore 1 Control Shore 2 Control
Fig. 1. Percentage cover of the 3 main algal functional groups in experimental plots over the 12 mo period of observation, for
summer and winter seasonal start dates in SW England and the Isle of Man. Each line represents the mean percentage cover for
a single shore in 1 of 3 treatments, with data pooled over start dates within a season and patches within a shore. Error bars are
omitted for clarity. a–d: ephemeral green algae, predominantly Enteromorpha spp.; e–h: soft crustose algae, predominantly
Ralfsia spp.; i–l: fucoid canopy algae, predominantly Fucus vesiculosus
over the 12 mo period of study are shown in Fig. 1. Soft crustose species only developed in SW England,
Data are grouped into mean levels per shore and sea- being completely absent from experimental plots on
son and, thus, do not display smaller temporal (date) the Isle of Man. There was little difference between
and spatial (patch) scale variability. Of immediate note shores or between seasons in the development of soft
is that for all 3 algal groups, there was a marked effect crusts and on average, they covered 50% of exclusion
of limpet removal; algal cover in exclusion plots plots at the end of the 12 mo period in the south west
differed markedly from control and half-fenced treat- for both summer and winter start dates. Despite high
ments. There was very little difference between values of percentage cover in some plots, the biomass
control and half-fenced treatments indicating few or of soft crusts was consistently low. Soft crusts were
no artefacts caused by the placement of fences. This absent from half-fenced and control plots.
was confirmed in statistical analyses. For example, Development of green ephemeral algae in limpet
SNK multiple comparisons of the significant Treatment exclusion plots differed between the 2 locations in
× Patch interaction in the 6-factor mixed-model timing, abundance and variability. In SW England,
ANOVA, showed that fucoid macroalgae (measured as peak ephemeral algal cover occurred at 8 mo (summer
the area under the fucoid curve over 12 mo) were more start date) and 6 mo (winter start date), and generally
abundant in half-fenced compared to control treat- showed the same pattern of colonisation between sea-
ments in only 2 out of 32 patches at both locations. A sons and shores. On the Isle of Man, green algae
similar result was obtained for ephemeral green algae. developed soon after limpet removal, showing peaks at
As a result of the general lack of a fence artefact, the 2 mo (summer start date) and 4 mo (winter start date)
half-fence treatment was ignored in further analyses. before declining as fucoid canopy algae developed.
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 81
Development of Fucus vesiculosus showed marked sion plots leading to little or no algal growth. For both
differences between the 2 locations. Colonisation oc- maximum cover and area under the curve, there was a
curred more quickly and percentage cover reached a significant interaction between treatment and the
higher level on the Isle of Man compared to SW Eng- smallest spatial scale, that of patch (maximum cover:
land. Maximum cover approached 100% for both start F16,128 = 2.88; p < 0.001; Area: F16,128 = 2.27; p < 0.01).
dates on the Isle of Man, while in SW England it was SNK tests of this interaction show that there was no
little over 50%. There was little difference in the timing effect of the treatment at 3 patches at Port St. Mary on
of fucoid development between seasons in SW Eng- the Isle of Man and 1 patch at Heybrook Bay in SW
land; however, on the Isle of Man, fucoid cover devel- England for the summer start dates. At all other
oped more quickly in the summer experiment, on aver- patches, limpet removal resulted in significantly
age 2 mo in advance of the winter experiment. The higher growth of ephemeral algae than in control plots.
decline in cover of fucoids between 8 and 12 mo For maximum fucoid cover, there was also a signifi-
following the winter but not the summer start date on cant interaction between treatment and patch
the Isle of Man is noteworthy. This occurred as large (Table 1); SNK tests of this interaction showed that at
mature plants were lost in autumn and winter storms. all 16 patches, at each location, the exclusion treat-
For both ephemeral green algae and fucoids, there was ment was significantly greater than the control. Thus,
greater development in control plots on the Isle of Man limpet removal always resulted in significantly greater
than SW England. Fucoids were virtually absent from fucoid cover than when limpets were present, at both
control plots in the southern location, whilst on the Isle locations. For area under the fucoid curve, the highest
of Man they developed high cover in a minority of plots. order significant interaction of the factor treatment was
Treatment × Location; SNK tests showed a significant
difference between the 2 treatments at both locations.
Relationship with physical variables
For each shore at each location, the relationships Variation in exclusion plots
between tidal height, substratum roughness, slope and
aspect for each experimental plot, and the main depen- There were no clear differences between locations in
dent variables (maximal cover and area under the curve) the level of ephemeral green algae in limpet exclusion
for ephemeral algae and fucoids were investigated to plots (Table 2). Both locations showed small scale vari-
determine potential causes of variability. There was no ability at the scale of patch, although such differences
correlation between any of the physical variables between patches only occurred in the summer start
measured and either of the measures of algal cover. dates on the Isle of Man (SNK of significant Patch fac-
tor). Differences between shores only occurred on the
Isle of Man; such differences can be clearly seen in the
Spatial and temporal variability of limpet effects summer experiment where ephemerals did not grow at
all at Port St. Mary, but reached up to 50% cover at
The degree to which limpet removal allowed algal Derbyhaven. Temporal differences were only ob-
growth in the experimental manipulations can be exam- served at the seasonal scale with greater maximum
ined in a number of different ways. Firstly, we used data ephemeral cover in winter on the Isle of Man (SNK of
from control and exclusion plots but ignoring those from Location × Season). For soft algal crusts, differences
the half-fenced treatment to determine the level of between locations were clear; there was no develop-
consistency of the treatment (limpet removal). Secondly, ment of this algal group on the Isle of Man. In SW Eng-
we used data solely from limpet exclusion plots to deter- land, variability only occurred at the smallest spatial
mine how the level of algal growth following localised scale; differences between patches occurred in both
loss of limpets varied at all temporal and spatial scales. the winter and summer experiments.
Removal of limpets led to significantly greater fucoid
development (both maximum cover and area under the
Consistency of treatment effect curve) on the Isle of Man compared to SW England
(Fig. 1, Table 3). These differences between locations
A 6-factor mixed-model ANOVA was applied to the were consistent between experimental start dates and
data and the consistency of the treatment determined seasons. There was significant variability at the spatial
by applying the SNK multiple comparison test to the scales of patches and marginally insignificant variabil-
highest order significant interaction involving the fac- ity (at the 5% level) at the scale of the shore; SNK tests
tor ‘treatment’. For ephemeral algae, the effects of of these factors, for both measures of fucoid cover,
limpet removal were variable, with some limpet exclu- showed that such variability only occurred in SW Eng-
82 Mar Ecol Prog Ser 287: 77–86, 2005
Table 1. Mixed-model ANOVA of fucoid canopy cover in experimental plots of the control and grazer exclusion. Half-fenced
treatment is not included. Tr: treatment; Loc: location; Sh(Loc): shore (location); Se: season; Da: date; Pa: patch
Source Maximum cover Area under curve
Transformation: none Transformation: none
df MS F p MS F p F-ratio versus
Tr 1 193675 507.5 < 0.01 2947110 1656 < 0.001 Tr × Sh(Loc)
Loc 1 47313 132.1 < 0.01 1966680 769 < 0.01 Sh(Loc)
Sh(Loc) 2 358 1.8 > 0.2 2554 0.2 > 0.8 Da[Se × Sh(Loc)]
Se 1 0.02 0.007 > 0.9 186750 12.8 > 0.06 Se × Sh(Loc)
Tr × Loc 1 21042 55.1 < 0.05 1386350 779 < 0.01 Tr × Sh(Loc)
Tr × Sh(Loc) 2 381 2.4 > 0.1 1778 0.7 > 0.5 Tr × Da[Se × Sh(Loc)]
Tr × Se 1 295 0.9 > 0.4 103277 9.3 > 0.09 Tr × Se × Sh(Loc)
Loc × Se 1 1485 5.7 > 0.1 257803 17.7 > 0.05 Se × Sh(Loc)
Tr × Se × Loc 1 462.5 1.4 > 0.3 157580 14.2 > 0.06 Tr × Se × Sh(Loc)
Se × Sh(Loc) 2 261.3 1.3 > 0.3 14530 1.2 > 0.3 Da[Se × Sh(Loc)]
Tr × Se × Sh(Loc) 2 329 2.1 > 0.15 11110 4.2 > 0.05 Tr × Da[Se × Sh(Loc)]
Da[Se × Sh(Loc)] 8 193 0.6 > 0.7 11981 1.3 > 0.3 Pa{Da[Se × Sh(Loc)]}
Tr × Da[Se × Sh(Loc)] 8 156 0.4 > 0.8 2673 0.8 > 0.6 Tr × Pa{Da[Se × Sh(Loc)]}
Pa {Da[Se × Sh(Loc)]} 16 320 2.6 < 0.01 9056 3.9 < 0.001 Residual
Tr × Pa{Da[Se × Sh(Loc)]} 16 325 2.6 < 0.01 3358 1.5 > 0.1 Residual
Residual 128 123 2302
land (Table 3). For example, for SNK comparisons of Small-scale variability in fucoid cover
maximum fucoid cover between patches established
on the same date, 5 out of 8 tests were significant in SW Examination of variability at the smallest spatial
England compared to none on the Isle of Man. At the scale, that between individual experimental plots, was
spatial scale of shores, there were significant differ- made by determining the frequency distribution of
ences in SW England but not the Isle of Man; thus, for maximal percentage cover for exclusion and control
fucoid algae, there was less, but more variable cover in plots (Fig. 2). On the Isle of Man, the frequency distri-
SW England in limpet removal plots than in the more bution for maximum fucoid cover in exclusion plots
northerly locality. was skewed strongly to the right; 43 of the 48 exclusion
Table 2. Mixed-model ANOVA of ephemeral green algal cover in limpet exclusion plots. Loc: location; Sh(Loc): shore (location);
Se: season; Da: date; Pa: patch
Source Maximum cover Area under curve
Transformation: arcsin Transformation: ln(x + 1)
Cochran’s C = 0.1913, p > 0.05 Cochran’s C = 0.1570, p > 0.05
df MS F p MS F p F-ratio versus
Loc 1 163 0.03 > 0.8 16.07 1.05 > 0.4 Sh(Loc)
Sh(Loc) 2 5553 9.15 < 0.01 15.31 6.93 < 0.02 Da [Se × Sh(Loc)]
Se 1 10185 56.73 < 0.02 56.35 14.22 > 0.05 Se × Sh(Loc)
Loc × Se 1 2660 14.82 > 0.05 14.79 3.73 > 0.1 Se × Sh(Loc)
Se × Sh(Loc) 2 179 0.3 > 0.7 3.96 1.79 > 0.2 Da [Se × Sh(Loc)]
Da [Se × Sh(Loc)] 8 606 1.62 > 0.1 2.21 1.55 > 0.2 Pa {Da[Se × Sh(Loc)]}
Pa {Da[Se × Sh(Loc)]} 16 375 4.15 < 0.001 1.43 5.21 < 0.001 Residual
Residual 64 90 0.27
SNK test of Sh(Loc) SE = 5.03 SE = 0.304
Isle of Man: Shore 1 < Shore 2 Isle of Man: Shore 1 < Shore 2
SW England: Shore 1 = Shore 2 SW England: Shore 1 = Shore 2
SNK test of Loc × Se SE = 2.74
Isle of Man: Summer < Winter
SW England Summer = Winter
Summer and winter: Isle of Man = SW England
SNK test of Pa SE = 5.49 SE = 0.303
Isle of Man: 2/8 comparisons significant Isle of Man: 2/8 comparisons significant
SW England: 4/8 comparisons significant SW England: 4/8 comparisons significant
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 83
Table 3. Mixed-model ANOVA of fucoid canopy algal cover in limpet exclusion plots. Loc: location; Sh(Loc): shore (location);
Se: season; Da: date; Pa: patch
Source Maximum cover Area under curve
Transformation: none Transformation: ln(x)
Cochran’s C = 0.4604, p < 0.01 Cochran’s C = 0.1834, p > 0.05
df MS F p MS F p F-ratio versus
Loc 1 65730 90.89 < 0.02 92.88 117.28 < 0.01 Sh(Loc)
Sh(Loc) 2 723 3.68 > 0.05 0.79 4.11 > 0.05 Da [Se × Sh(Loc)]
Se 1 150 0.33 > 0.6 0.19 0.08 > 0.8 Se × Sh(Loc)
Loc × Se 1 1802 4.00 > 0.1 5.55 2.42 > 0.25 Se × Sh(Loc)
Se × Sh(Loc) 2 450 2.29 > 0.1 2.29 11.89 < 0.01 Da [Se × Sh(Loc)]
Da [Se × Sh(Loc)] 8 196 0.54 > 0.8 0.19 0.30 > 0.9 Pa {Da[Se × Sh(Loc)]}
Pa {Da[Se × Sh(Loc)]} 16 365 2.98 < 0.01 0.63 4.57 < 0.001 Residual
Residual 64 122 0.14
SNK test of Sh(Loc) SE = 2.86 SE = 0.09
Isle of Man: Shore 1 = Shore 2 Isle of Man: Shore 1 = Shore 2
SW England: Shore 1 < Shore 2 SW England: Shore 1 < Shore 2
SNK test of Pa SE = 6.39 SE = 0.21
Isle of Man: 0/8 comparisons significant Isle of Man: 0/8 comparisons significant
SW England: 5/8 comparisons significant SW England: 4/8 comparisons significant
a) Isle of Man
50 50
Exclusion treatment Control treatment
40
Frequency
40
30 30
20 20
10 10
0 0
0 25 50 75 100 0 25 50 75 100
b) SW England
50 50
40 40
Frequency
30 30
20 20
Fig. 2. Frequency distri-
bution of maximal fucoid 10 10
canopy cover values for
limpet exclusion and 0 0
control plots on (a) the 0 25 50 75 100 0 25 50 75 100
Isle of Man and in (b) SW
England Maximum percentage cover
plots established throughout the experiment had a cover in control plots showed no variation in SW Eng-
maximal cover of fucoids of over 90%. In contrast, the land, with no fucoid growth at all, while on the Isle of
frequency distribution in SW England was approxi- Man, although over half the experimental plots
mately normal, with the maximum cover of fucoids in showed no fucoid growth, there was extensive growth
exclusion plots showing high variability. Maximum in a minority of plots (Fig. 2)
84 Mar Ecol Prog Ser 287: 77–86, 2005
DISCUSSION the dominant patellid limpet grazers (Patella vulgata
on the Isle of Man and both P. vulgata and P. depressa
Local scale experimental studies have provided con- in SW England), regular maintenance of the experi-
siderable insight into the way shallow subtidal and ment prevented other macrograzers, including Litto-
intertidal communities are structured and organised. rina littorea, L. obtusata, Gibbula umbilicalis and Osil-
The tractable nature of rocky intertidal systems has inus lineata, from exerting a large effect. Other grazers
allowed them not only to provide a means of testing within the eulittoral zone, such as crustacean meso-
ecological theory, but means that rocky shore commu- herbivores, which were not manipulated, could poten-
nities are some of the best understood in the world, in tially contribute to the spatial and temporal differences
either terrestrial or marine environments. Despite this, observed. However, from our observations and the
it is well recognised that the results of many commu- known effects of these grazers (see Brawley 1992 for
nity-based field studies are context-dependent to a review), it is unlikely that they contributed signifi-
large extent (Lawton 1999), making generalisations cantly to the large differences between locations.
difficult. The key role played by grazing patellid The probability of early post-settlement stages of
limpets in controlling macroalgal development on fucoid macroalgae (zygotes, young germlings and
rocky shores of NW Europe has been recognised for germlings of Vadas et al. 1992) escaping grazing by
decades. The huge changes in community structure microphagous molluscs is critical in determining com-
after removal of limpets, especially on a large scale munity structure at the mid-tide level of rocky shores.
(Jones 1948, Southward & Southward 1978), suggest We propose that an increase in the probability of
that these are ‘keystone’ grazers (sensu Paine 1966). escape with increasing latitude is the prime driver for
However, the keystone effects of limpets are not uni- the increasing dominance of macroalgae to the north,
versal (e.g. Jenkins et al. 1999a), just as the effects of with fucoids extending further onto wave-exposed
the original keystone predator Pisaster ochraceus are shores. Such probability will change with the abun-
limited to certain habitats (Menge et al. 1994). In addi- dance, activity and possibly diversity of grazers, and
tion, the geographical generality of limpet-grazing the growth rate of algae. In areas of naturally reduced
effects on exposed shores of Europe are not fully grazing intensity or experimental exclusions, the
known. The majority of effective manipulations of probability of fucoid development lies solely with the
patellid limpets in the mid-shore zone have been con- supply of propagules and/or their ability to develop
centrated on the Isle of Man (see Hawkins et al. 1992 and grow. The dominant fucoid in experimental exclu-
for review). More recently, however, extensive experi- sion plots, Fucus vesiculosus, has a European range
mental manipulations of patellid grazers have been stretching from northern Norway as far south as
made in the Mediterranean (Benedetti-Cecchi et al. Morocco (Luhning 1990). A detailed study into the
2000, 2001). This work has shown extremely inconsis- biology of F. vesiculosus on the Isle of Man and Devon
tent grazing effects at different spatial and temporal (SW England) by Knight & Parke (1950) showed little
scales with only occasional strong effects on macroal- difference in the growth rate, reproductive period or
gal abundance. Low on the shore at exposed sites, ability to repopulate areas cleared within stands of
Boaventura et al. (2002) have convincingly shown that adults. Hence, in SW England, F. vesiculosus is well
limpet grazing can limit the vertical extent of turf form- within its distributional range and so expected to
ing and canopy algae in both England and Portugal. respond well to the release of grazing pressure. It
The results of the present study show, that within the appears unlikely that fucoid zygotes and germlings are
British Isles, the community response to small-scale unable to cope with slightly higher air and sea temper-
loss of limpets is similar; at both locations, macroalgae atures at the southern location, although this remains
developed after limpet removal. However, the level of to be tested experimentally.
determinism in response to simulated localised release Another explanation for the lower effect size in SW
of grazing pressure varies considerably. It is clear that England is a generally lower, less predictable supply of
the more northerly locality, the Isle of Man, experi- macroalgal propagules. Dispersal patterns of algal
ences a strong, deterministic community response to propagules are generally poorly understood, but
localised reductions in grazing pressure. In contrast, assessment of the distance of recruits from adult
further south in the British Isles, the response is weak sources have invariably suggested short dispersal
and more variable. These results support, and provide shadows, in the order of metres to 10s of metres (see
a mechanistic insight into, the observations of Ballan- Santelices 1990 for review). Limited dispersal dis-
tine (1961) of a latitudinal gradient in the balance tances can be increased if the number of source plants
between grazers and macroalgae across the wave- is increased. The effect of large stands of adult plants
exposure gradient. It should be noted that while the on propagule supply was graphically demonstrated by
main focus of our experiment was in manipulation of the early experiments of Burrows & Lodge (1950) when
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 85
extensive fucoid development in a large limpet clear- whole quadrat would be expected to follow a binomial
ance (over 10 × 100 m) resulted in high fucoid recruit- distribution defined by the number of possible patches
ment downstream of the original experimental area. In and the likelihood of occupancy of single patches.
mid-shore experiments in northern Spain, increasing Lower likelihood of patch occupancy as seen in SW
distance from stands of Fucus spp. results in a decline England would give much greater variability in percent
in the colonisation and development of Fucus spp. cover as a result of such a binomial expectation.
canopy in grazer exclusion plots (F. Arenas pers. In summary, we show that the level of determinism
comm.). In the present study, no quantitative measures in the community response to small-scale limpet loss,
were made of the distribution of adult stands of F. in particular the development of a fucoid canopy,
vesiculosus in relation to experimental plots at either varies considerably within the British Isles. The level of
location. However, F. vesiculosus was generally abun- fucoid development was consistently high on the Isle of
dant on the Isle of Man and stands of adults were Man, compared to a low and more variable response
rarely if ever more than 30 m away from experimental further south in SW England. This work supports the
plots. In contrast, on the 2 shores of SW England, observations of Ballantine (1961) of a latitudinal gradi-
stands of F. vesiculosus were rarer and more patchily ent in the balance between grazers and macroalgae.
distributed, supporting the hypothesis that propagule Further experimental work is required to determine
supply was limiting. the causal mechanisms driving this change in balance.
It could be argued that the lower response to grazer
loss in SW England indicates a reduction in the role of Acknowledgements. This study was supported by the Mast III
limpet grazing in structuring mid-shore communities. project EUROROCK MAS3-CT95-0012. Thanks to E.
However, Jenkins et al. (2001) demonstrated an in- LaCroix, M. Roberts, S. Kimmance and D. Boaventura for
crease in abundance and overall grazing pressure of assistance with the experimental set-up and sampling. S.R.J.
and S.J.H. were supported during data analysis and write-up
patellid limpets in SW England compared with the Isle by NERC Grant-In-Aid to the MBA.
of Man, consistent with a general increase in grazing
pressure with declining latitude in Europe. These
LITERATURE CITED
observations would suggest an increase in the impor-
tance of grazers further south. One explanatory model Arrontes J (2002) Mechanisms of range expansion in the
to account for these results is that the activity of intertidal brown alga Fucus serratus in northern Spain.
microphagous grazers, of which patellid limpets are Mar Biol 141:1059–1067
Ballantine WJ (1961) A biologically defined exposure scale for
undoubtedly the most important, limits the develop-
the comparative description of rocky shores. Field Stud
ment of fucoid macroalgae in the south to such an 1:1–19
extent that adult plants are rare; thus, when localised Benedetti-Cecchi L, Airoldi L, Abbiati M, Cinelli F (1996)
areas of reduced grazing pressure arise, through nat- Estimating the abundance of benthic invertebrates: a com-
ural or experimental means, there is insufficient supply parison of procedures and variability between observers.
Mar Ecol Prog Ser 138:93–101
of propagules. Small scale experimentation cannot be Benedetti-Cecchi L, Bulleri F, Cinelli F (2000) The interplay of
used to determine the importance of limpet grazing at physical and biological factors in maintaining mid-shore
a shore-wide scale under this scenario. and low-shore assemblages on rocky coasts in the north-
Variability in propagule supply in SW England may west Mediterranean. Oecologia 123:406–417
Benedetti-Cecchi L, Bulleri F, Acunto S, Cinelli F (2001)
be one cause of the high variability in fucoid develop-
Scales of variation in the effects of limpets on rocky shores
ment at this location. Within the experiment, none of in the northwest Mediterranean. Mar Ecol Prog Ser 209:
the physical variables measured, shore height, aspect, 131–141
slope nor surface roughness could explain variability in Boalch G, Holme N, Jephson N, Sidwell J (1974) A resurvey of
either ephemeral or fucoid development. This is, Colman’s intertidal traverses at Wembury, South Devon.
J Mar Biol Assoc UK 54:551–553
however, not surprising as experimental plots at both Boaventura D, Alexander M, Della Santina P, Smith ND, Re P,
locations were chosen with relatively strict physical Da Fonseca LC, Hawkins SJ (2002) The effects of grazing
criteria, which minimised differences in the physical on the distribution and composition of low-shore algal
environment. Another explanation is based simply on communities on the central coast of Portugal and on the
southern coast of Britain. J Exp Mar Biol Ecol 267:185–206
the inherent probabilities of patch occupancy within
Brawley S (1992) Mesoherbivores. In: John DM, Hawkins SJ,
each exclusion quadrat for high and low mean values of Price JH (eds) Plant-animal interactions in the marine
fucoid cover. Percentage cover can be considered as a benthos. Oxford University Press, Oxford
measure of occupancy of a number of possible sites, or Brawley SH, Johnson LE (1991) Survival of fucoid embryos in
patches (Johnson et al. 1997), for Fucus spp. plants the intertidal zone depends upon developmental stage
and microhabitat. J Phycol 27:179–186
within each quadrat. If the presence of Fucus spp. in Burrows EM, Lodge SM (1950) Note on the inter-relationships
single patches is independent of the occupancy of other of Patella, Balanus and Fucus on a semi-exposed coast.
patches in a quadrat, then percentage cover for the Rep Mar Biol Stn Port Erin 62:30–34
86 Mar Ecol Prog Ser 287: 77–86, 2005
Burrows MT, Hawkins SJ (1998) Modelling patch dynamics Lewis JR (1964) The ecology of rocky shores. English Univer-
on rocky shores using deterministic cellular automata. sities Press, London
Mar Ecol Prog Ser 167:1–13 Lodge SM (1948) Algal growth in the absence of Patella on an
Colman JS (1933) The nature of intertidal zonation of plants experimental strip of foreshore, Port St. Mary, Isle of Man.
and animals. J Mar Biol Assoc UK 18:435–476 Proc Trans Liverpool Biol Soc 56:78–85
Hartnoll RG, Hawkins SJ (1985) Patchiness and fluctuations Lubchenco J, Gaines SD (1981) A unified approach to marine
on moderately exposed rocky shores. Ophelia 24:53–64 plant-herbivore interactions. I. Populations and communi-
Hawkins SJ (1981a) The influence of Patella grazing on the ties. Annu Rev Ecol Syst 12:405–437
fucoid-barnacle mosaic on moderately exposed rocky Luhning K (1990) Seaweeds, their environment, biogeo-
shores. Kiel Meeresforsch 5:537–543 graphy and ecophysiology. John Wiley & Sons, London
Hawkins SJ (1981b) The influence of season and barnacles on McCook, Chapman ARO (1991) Community succession fol-
algal colonization of Patella vulgata (L.) exclusion areas. lowing massive ice scour on an exposed rocky shore:
J Mar Biol Assoc UK 61:1–15 effects of Fucus canopy algae and of mussels during late
Hawkins SJ, Hartnoll RG (1983) Grazing of intertidal algae by succession. J Exp Mar Biol Ecol 154:137–169
marine invertebrates. Oceanogr Mar Biol Annu Rev 21: Menge BA, Berlow EL, Blanchette CA, Navarette SA,
195–282 Yamada SB (1994) The keystone species concept: varia-
Hawkins SJ, Hartnoll RG, Kain JM, Norton TA (1992) Plant tion in interaction strength in a rocky intertidal habitat.
animal interactions on hard substrata in the North-east Ecol Monogr 64:249–286
Atlantic. In: John DM, Hawkins SJ , Price JH (eds) Plant- Paine RT (1966) Food web complexity and species diversity.
animal interactions in the marine benthos. Systematics Am Nat 100:65–75
Association Spec Vol 46. Clarendon Press, Oxford, p 1–32 Paine RT (2002) Trophic control of production in a rocky
Jenkins SR, Hawkins SJ, Norton TA (1999a) Direct and indi- intertidal community. Science 296:736–739
rect effects of a macroalgal canopy and limpet grazing in Reed DC, Foster MS (1984) The effect of canopy shading on
structuring a sheltered inter-tidal community. Mar Ecol algal recruitment and growth in a giant kelp forest.
Prog Ser 188:81–92 Ecology 65:937–948
Jenkins SR, Hawkins SJ, Norton TA (1999b) Interaction Santelices B (1990) Patterns of reproduction, dispersal and
between a fucoid canopy and limpet grazing in structuring recruitment in seaweeds. Oceanogr Mar Biol Annu Rev
a low shore intertidal community. J Exp Mar Biol Ecol 233: 28:177–276
41–63 Southward AJ (1953) The ecology of some rocky shores in the
Jenkins SR, Norton TA, Hawkins SJ (1999c) Settlement and south of the Isle of Man. Proc Trans Liverpool Biol Soc
post-settlement interactions between Semibalanus bal- 59:1–50
anoides (L.) (Crustacea: Cirripedia) and three species of Southward AJ (1956) The population balance between lim-
fucoid canopy algae. J Exp Mar Biol Ecol 236:49–67 pets and seaweeds on wave-beaten rocky shores. Rep Mar
Jenkins SR, Arenas F, Arrontes J, Bussell J and 10 others Biol Stn Port Erin 68:20–29
(2001) A European scale analysis of seasonal variability in Southward AJ (1964) Limpet grazing and the control of
limpet grazing activity and microalgal abundance. Mar vegetation on rocky shores. In: Crisp DJ (ed) Grazing in
Ecol Prog Ser 211:193–203 terrestrial and marine enviroments. Blackwell Scientific
Johnson MP, Burrows MT, Hartnoll RG, Hawkins SJ (1997) Publications, Oxford, p 265–273
Spatial structure on moderately exposed rocky shores: Southward AJ (1979) Cyclic fluctuations in population density
patch scales and the interactions between limpets and during 11 years recolonisation of rocky shores in West
algae. Mar Ecol Prog Ser 160:209–215 Cornwall following the Torrey Canyon oil spill in 1967. In:
Johnson MP, Burrows MT, Hawkins SJ (1998) Individual Naylor E, Hartnoll RG (eds) Cyclic phenomena in marine
based simulations of the direct and indirect effects of plants and animals. Pergamon Press, Oxford, p 85–92
limpets on a rocky shore Fucus mosaic. Mar Ecol Prog Ser Southward AJ, Southward EC (1978) Recolonization of rocky
169:179–188 shores in Cornwall after use of toxic dispersants to clean
Jones NS (1946) Browsing of Patella. Nature 158:557 up the Torrey Canyon spill. J Fish Res Board Can 35:
Jones NS (1948) Observations and experiments on the biol- 682–706
ogy of Patella vulgata at Port St. Mary, Isle of Man. Proc Vadas RL, Wright WA, Miller SL (1990) Recruitment of Asco-
Trans Liverpool Biol Soc 56:60–77 phyllum nodosum: wave action as a source of mortality.
Knight M, Parke MW (1950) A biological study of Fucus vesicu- Mar Ecol Prog Ser 61:263–272
losus and Fucus serratus. J Mar Biol Assoc UK 29:439–514 Vadas RL, Johnson S, Norton TA (1992) Recruitment and
Lawton J (1999) Are there general laws in ecology? Oikos 84: mortality of early post-settlement stages of benthic algae.
177–192 Br Phycol J 27:331–351
Leonard GM (2000) Latitudinal variation in species inter- Velimirov B, Griffiths CL (1979) Wave-induced kelp move-
actions: a test in the New England rocky intertidal zone. ment and its importance for community structure. Bot Mar
Ecology 81:1015–1030 22:169–172
Editorial responsibility: Roger Hughes (Contributing Editor), Submitted: April 6, 2004; Accepted: August 24, 2004
Bangor, UK Proofs received from author(s): January 31, 2005
Vol. 287: 77–86, 2005 Published February 18
Mar Ecol Prog Ser
Regional scale differences in the determinism of
grazing effects in the rocky intertidal
S. R. Jenkins1,*, R. A. Coleman2, P. Della Santina3, S. J. Hawkins1, 3,
M. T. Burrows4, R. G. Hartnoll5
1
Marine Biological Association, Citadel Hill, Plymouth PL1 2PB, UK
2
Marine Biology and Ecology Research Group, School of Biological Sciences, University of Plymouth, Drake Circus,
Plymouth PL4 8AA, UK
3
Biodiversity and Ecology Division, School of Biological Sciences, Southampton University, Southampton SO16 7PX, UK
4
Scottish Association for Marine Science, Dunstaffnage Marine Laboratory, Oban, Argyll PA37 1QA, UK
5
Port Erin Marine Laboratory, University of Liverpool, Port Erin, Isle of Man IM9 6JA, UK
ABSTRACT: Patellid limpets are dominant grazers on intertidal rocky shores of NW Europe with a
key role in structuring the eulittoral community. Localised loss of limpets and the subsequent reduc-
tion in grazing pressure is known to result in important changes in community structure, through the
development of canopy-forming macroalgae, and an associated increase in species diversity and
community complexity. The level of determinism in the community level response to localised loss of
patellid limpets was assessed at spatial scales from 100s of kilometres to 10s of metres and temporal
scales from weeks to months at mid-tide level of exposed rocky shores. Limpets were removed and
excluded from experimental plots to simulate localised limpet loss and appropriate controls estab-
lished. Experimental plots were established in replicate patches at 2 shores at each of 2 regional
locations, separated by approximately 500 km: the Isle of Man and SW England. Removals were con-
ducted on 2 dates within each of 2 seasons (summer and winter) and the community level response
monitored for a period of 12 mo. There was a clear effect of limpet loss at all spatial and temporal
scales, with rapid development of green ephemeral algae followed by a fucoid canopy. However, the
degree of determinism in the development of canopy-forming algae differed markedly between the
2 locations. At the northerly location, the Isle of Man, fucoid algae developed quickly and dominated
all areas of limpet exclusion; there was little variability between plots. In contrast, in SW England, the
abundance of fucoid algae was significantly lower and much more variable. Such geographic
changes in the development of macroalgae in the absence of the dominant grazer are discussed in
relation to rocky shore community dynamics and the latitudinal change in balance between grazers
and algae over the wave exposure gradient.
KEY WORDS: Herbivory · Large scale · Macroalgae · Rocky shore
Resale or republication not permitted without written consent of the publisher
INTRODUCTION community structure of intertidal rocky shores of
NW Europe (see Southward 1964, Hawkins & Hartnoll
Experimental work on rocky shores has frequently 1983, Hawkins et al. 1992 for reviews). Limpet-
demonstrated that intertidal grazers have the ability to removal experiments on the Isle of Man first demon-
almost completely control the species composition, strated that canopy-forming algae, particularly mid-
distribution and dynamics of algal communities (Lub- shore Fucus vesiculosus, were directly prevented from
chenco & Gaines 1981, Paine 2002). Patellid limpets establishing on wave-exposed shores by limpet graz-
are some of the best-known and most studied of ing (Jones 1946, 1948, Lodge 1948, Burrows & Lodge
marine herbivores, and have a profound effect on the 1950, Southward 1956, 1964). Experimental removal of
*Email: sjen@mba.ac.uk © Inter-Research 2005 · www.int-res.com
78 Mar Ecol Prog Ser 287: 77–86, 2005
limpets resulted in the rapid development of various of these including anemones, dogwhelks, littorinids,
ephemeral or opportunistic species (Enteromorpha isopods and amphipods. In addition, the development
spp., Ulva spp., Blidingia spp.) followed by growth of a of fucoid canopy algae on exposed shores, dominated
fucoid canopy which persisted for up to 5 yr. This suc- by barnacles and mussels, results in a distinct change
cessional sequence was confirmed in the British Isles, from a community dominated by the secondary pro-
following the widespread mortality of limpets on the duction of filter feeders to one dominated by primary
shores of west Cornwall following a massive oil spill production of macroalgae. Hence, the eulittoral zone
(Southward & Southward 1978, Southward 1979), and can change from a net importer of primary produc-
in follow-up experiments on the Isle of Man (Hawkins tion, when barnacle-dominated, to a net exporter,
1981a,b, Hartnoll & Hawkins 1985). when dominated by macroalgae (Hawkins et al.
In the British Isles and northern France, moderately 1992).
exposed shores support a mid-shore community com- We aimed to test the determinism of the assemblage
posed of patches of Fucus spp., barnacles and bare response (and particularly the development of a fucoid
rock, interspersed with grazing patellid limpets (Lewis canopy) to limpet loss at a realistic spatial scale, from
1964). On such shores, localised areas of low grazing small (0.5 × 0.5 m) patches. We tested the general
pressure allow ‘escapes’ of macroalgae to occur (Hart- hypothesis that loss of limpets from small patches at
noll & Hawkins 1985). Once beyond a size of around a mid-tide level of moderately exposed shores results in
few centimetres, such macroalgae are rarely con- macroalgal growth. Many factors may affect the likeli-
sumed by microphagous grazers such as limpets and hood of macroalgal development in the absence of
hence, persist to form clumps of canopy. The dynamics limpet grazing, including the supply of macroalgal
of this patchy mosaic community have been the subject propagules (Arrontes 2002), mortality of germlings in
of a number of empirical (Hawkins 1981a, Hartnoll & unfavourable microclimates (Brawley & Johnson 1991),
Hawkins 1985, Johnson et al. 1997) and modelling dislodgement of propagules by wave action (Vadas et
studies (Burrows & Hawkins 1998, Johnson et al. 1998) al. 1990) and grazing of propagules or germlings by
that have demonstrated the importance of individual other non manipulated grazers such as crustacean
limpet behaviour in maintaining the mosaic com- mesoherbivores (Brawley 1992). Given the variability
munity. of these and other ecological processes on rocky
The role of patellid limpets in structuring the mid- shores, from place to place, and time to time, we used
shore assemblage of the NE Atlantic at exposed sites a complex experimental design to test the general
is undoubted, though they have much less influence hypothesis applied over a number of spatial and tem-
amongst canopies, low on the shore (Jenkins et al. poral scales. In addition, we examined the effect of
1999b) or at sheltered sites (Jenkins et al. 1999a). latitude within the British Isles on the community
Ballantine (1961) observed a latitudinal trend in the response to limpet loss by conducting the experiment
balance between fucoid algae and limpet/barnacle- at 2 locations: the Isle of Man and SW England. In this
dominated areas down the coast of western Europe, way, we tested the hypothesis developed from the
with fucoids being restricted further into shelter in the original observations of Ballantine (1961) that loss of
south. Both he and subsequent authors (Hawkins & limpets would have a larger and more consistent effect
Hartnoll 1983, Hawkins et al. 1992) suggested that the at northern compared to southern latitudes.
balance between the effectiveness of grazers and the
ability of fucoids to grow, changed with latitude,
thereby affecting the probability of successful coloni- MATERIALS AND METHODS
sation by fucoids in areas of reduced grazing pres-
sure. The development of a fucoid canopy on barna- Study sites. Experimental work was undertaken at 2
cle-dominated shores has important implications for locations separated by approximately 450 km, the SW
community dynamics and energy flow. Macroalgal of England near Plymouth (50° 19’ N, 04° 06’ W) and the
canopies regulate community structure in a number of south of the Isle of Man (54° 5’ N, 04° 40’ W) in the Irish
ways: (1) by altering the quality and quantity of light Sea. Both locations were in areas of full salinity with
reaching the substratum (e.g. Reed & Foster 1984), maximum tidal ranges of 5 and 6 m, respectively. Two
(2) by whiplash or sweeping effects of fronds (e.g. Ve- moderately exposed rocky shores on open coastlines,
limirov & Griffiths 1979, Jenkins et al. 1999c) or (3) by separated by a minimum of 2 km, were selected at both
providing shelter from wave action (McCook & Chap- locations: Wembury and Heybrook Bay in SW Eng-
man 1991) and from physical extremes such as high land, Port St. Mary and Derbyhaven in the Isle of Man.
temperatures, desiccation or freezing (Leonard 2000). The main criteria in shore selection was topographical
Fucoid clumps provide a complex habitat and shelter simplicity, a gentle slope (< 30°) and domination at
for a wide range of animal species, the more obvious mid-tide level by an extensive cover of barnacles, with
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 79
abundant patellid limpets. The shores in both areas, monthly until termination of the experiment at 12 mo.
the Isle of Man (Southward 1953, Hartnoll & Hawkins At each sampling date, a 0.5 × 0.5 m quadrat, sub-
1985) and the Plymouth area (Colman 1933, Boalch et divided to give 49 intersection points, was used to
al. 1974) have been well described. At both shores at estimate the percentage cover of all macroalgae using
each location, the mid-shore, though dominated by the point intersect method (e.g. Benedetti-Cecchi et al.
barnacle cover, had patches of fucoid canopy algae. 1996). At selected sampling dates, all experimental
These were less common in the Plymouth area. Exper- quadrats were photographed. At least monthly, exper-
imental plots were located in the mid-shore, well imental plots were checked for damage to fences and
within the barnacle zone and at a mean tidal height repairs made. Any limpets that had invaded exclusion
above Chart Datum of 2.3 m at Wembury (range 1.9 to treatments plus any other ‘macro-grazers’ (littorinids,
2.9 m), 2.1 m at Heybrook Bay (range 1.9 to 2.2 m), topshells) were removed.
3.6 m at Port St. Mary (range 3 to 4.3 m) and 3.5 m at Data analysis. The experiment was designed to
Derbyhaven (range 2.6 to 4.8 m). allow partitioning of sources of variance using a 6-
Experimental design. The experiment consisted of factor mixed-model ANOVA, where the factors Sea-
3 treatments: (1) complete removal of all patellid lim- son, Location and Treatment were fixed, and Date,
pets and exclusion using 3 cm high fences of plastic- Shore and Patch were random. This analysis was used
coated wire mesh with 13 mm square openings, (2) a to determine the spatial and temporal consistency of
half-fenced treatment using the same fence structure treatment effects (i.e. differences between the control,
but only encompassing half the quadrat perimeter, and fenced and exclusion treatments). Significant effects of
allowing free movement of limpets, and (3) a control the treatment give no information on the magnitude of
treatment with only the 4 corners of the quadrat the effect and how this varies over different spatial and
marked by screws. The half-fenced treatment was temporal scales. For this, data from exclusion plots
used as a procedural control to determine whether the alone were used in a 5-factor mixed-model ANOVA.
use of fences in the exclusion treatment had any effect Prior to ANOVA, data were examined for heterogene-
on community succession, other than that caused by ity of variance using Cochran’s test and heterogenous
exclusion of limpets. data transformed appropriately. Significant factors
In order to fully explore the variability in the effect of were analysed further using SNK (Student Newman
limpet grazing on mid-shore community structure, this Keuls) multiple comparisons.
basic design was implemented over a number of In any experiment examining development of biota
spatial and temporal scales. Spatial variability was over a number of sampling dates, choice of the depen-
assessed at 3 scales, between locations (100s of kilo- dent variable is paramount to interpretation of the
metres), between shores within each location (kilo- experiment. In theory, a separate analysis could be
metres) and among patches within each shore (10s to performed for each sampling date but this would only
100s of metres). Temporal variability was assessed at lead to an over complex interpretation. For the key
2 scales, between seasons (summer and winter) and species in our analysis, 2 dependent variables were
between dates within seasons. At each shore, experi- selected, maximal cover during the 12 mo period and
ments were established during 2 different seasons, the area under the curve for each individual plot. Max-
summer 1996 and winter 1996/1997, and in each sea- imal cover indicates the peak response to perturbation,
son, 2 start dates were selected at random from within while the area under the curve integrates the pattern
a 3 mo period, with a minimum separation of 4 wk. of development for any particular species over time
Start dates were independently selected at all shores and, thus, takes into account the rate and temporal
over both locations. At each start date, 2 patches were trajectory of algal colonisation.
selected at each of the 2 shores at both locations.
Within each patch, nine 0.5 × 0.5 m quadrats were cho-
sen and the 3 treatments, replicated 3 times, were RESULTS
applied at random. All 8 patches for each shore were
selected in advance of experimental set-up, over a hor- General patterns of colonisation
izontal distance of between 250 and 400 m, with a min-
imum separation of 30 m between individual patches. Three main algal groups developed in experimental
The choice of the 2 patches, at each start date, from plots as a result of limpet removal: ephemeral green al-
amongst the 8 selected was made at random. gae made up of a mixture of Enteromorpha and Blid-
Maintenance and sampling of the experiment. Fol- ingia spp. with some Ulva spp., and Monostroma spp.,
lowing establishment of the experiment, sampling was soft algal crusts, predominantly Ralfsia spp. and fucoid
undertaken at regular intervals after each individual canopy algae made up almost entirely of Fucus vesicu-
start date, monthly for the first 6 mo and then bi- losus. The patterns of colonisation of these 3 groups
80 Mar Ecol Prog Ser 287: 77–86, 2005
SW ENGLAND ISLE OF MAN
SUMMER WINTER SUMMER WINTER
START START Ephemerals START START
100 100 100 100
a) b) c) d)
% cover
75 75 75 75
50 50 50 50
25 25 25 25
0 0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Soft crusts
100 100 100 100
% cover
e) f) g) h)
75 75 75 75
50 50 50 50
25 25 25 25
0 0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Fucoid canopy
100 100 100 100
i) j) k) l)
% cover
75 75 75 75
50 50 50 50
25 25 25 25
0 0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Months
Shore 1 Limpet exclusion Shore 2 Limpet exclusion
Shore 1 Half fenced Shore 2 Half fenced
Shore 1 Control Shore 2 Control
Fig. 1. Percentage cover of the 3 main algal functional groups in experimental plots over the 12 mo period of observation, for
summer and winter seasonal start dates in SW England and the Isle of Man. Each line represents the mean percentage cover for
a single shore in 1 of 3 treatments, with data pooled over start dates within a season and patches within a shore. Error bars are
omitted for clarity. a–d: ephemeral green algae, predominantly Enteromorpha spp.; e–h: soft crustose algae, predominantly
Ralfsia spp.; i–l: fucoid canopy algae, predominantly Fucus vesiculosus
over the 12 mo period of study are shown in Fig. 1. Soft crustose species only developed in SW England,
Data are grouped into mean levels per shore and sea- being completely absent from experimental plots on
son and, thus, do not display smaller temporal (date) the Isle of Man. There was little difference between
and spatial (patch) scale variability. Of immediate note shores or between seasons in the development of soft
is that for all 3 algal groups, there was a marked effect crusts and on average, they covered 50% of exclusion
of limpet removal; algal cover in exclusion plots plots at the end of the 12 mo period in the south west
differed markedly from control and half-fenced treat- for both summer and winter start dates. Despite high
ments. There was very little difference between values of percentage cover in some plots, the biomass
control and half-fenced treatments indicating few or of soft crusts was consistently low. Soft crusts were
no artefacts caused by the placement of fences. This absent from half-fenced and control plots.
was confirmed in statistical analyses. For example, Development of green ephemeral algae in limpet
SNK multiple comparisons of the significant Treatment exclusion plots differed between the 2 locations in
× Patch interaction in the 6-factor mixed-model timing, abundance and variability. In SW England,
ANOVA, showed that fucoid macroalgae (measured as peak ephemeral algal cover occurred at 8 mo (summer
the area under the fucoid curve over 12 mo) were more start date) and 6 mo (winter start date), and generally
abundant in half-fenced compared to control treat- showed the same pattern of colonisation between sea-
ments in only 2 out of 32 patches at both locations. A sons and shores. On the Isle of Man, green algae
similar result was obtained for ephemeral green algae. developed soon after limpet removal, showing peaks at
As a result of the general lack of a fence artefact, the 2 mo (summer start date) and 4 mo (winter start date)
half-fence treatment was ignored in further analyses. before declining as fucoid canopy algae developed.
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 81
Development of Fucus vesiculosus showed marked sion plots leading to little or no algal growth. For both
differences between the 2 locations. Colonisation oc- maximum cover and area under the curve, there was a
curred more quickly and percentage cover reached a significant interaction between treatment and the
higher level on the Isle of Man compared to SW Eng- smallest spatial scale, that of patch (maximum cover:
land. Maximum cover approached 100% for both start F16,128 = 2.88; p < 0.001; Area: F16,128 = 2.27; p < 0.01).
dates on the Isle of Man, while in SW England it was SNK tests of this interaction show that there was no
little over 50%. There was little difference in the timing effect of the treatment at 3 patches at Port St. Mary on
of fucoid development between seasons in SW Eng- the Isle of Man and 1 patch at Heybrook Bay in SW
land; however, on the Isle of Man, fucoid cover devel- England for the summer start dates. At all other
oped more quickly in the summer experiment, on aver- patches, limpet removal resulted in significantly
age 2 mo in advance of the winter experiment. The higher growth of ephemeral algae than in control plots.
decline in cover of fucoids between 8 and 12 mo For maximum fucoid cover, there was also a signifi-
following the winter but not the summer start date on cant interaction between treatment and patch
the Isle of Man is noteworthy. This occurred as large (Table 1); SNK tests of this interaction showed that at
mature plants were lost in autumn and winter storms. all 16 patches, at each location, the exclusion treat-
For both ephemeral green algae and fucoids, there was ment was significantly greater than the control. Thus,
greater development in control plots on the Isle of Man limpet removal always resulted in significantly greater
than SW England. Fucoids were virtually absent from fucoid cover than when limpets were present, at both
control plots in the southern location, whilst on the Isle locations. For area under the fucoid curve, the highest
of Man they developed high cover in a minority of plots. order significant interaction of the factor treatment was
Treatment × Location; SNK tests showed a significant
difference between the 2 treatments at both locations.
Relationship with physical variables
For each shore at each location, the relationships Variation in exclusion plots
between tidal height, substratum roughness, slope and
aspect for each experimental plot, and the main depen- There were no clear differences between locations in
dent variables (maximal cover and area under the curve) the level of ephemeral green algae in limpet exclusion
for ephemeral algae and fucoids were investigated to plots (Table 2). Both locations showed small scale vari-
determine potential causes of variability. There was no ability at the scale of patch, although such differences
correlation between any of the physical variables between patches only occurred in the summer start
measured and either of the measures of algal cover. dates on the Isle of Man (SNK of significant Patch fac-
tor). Differences between shores only occurred on the
Isle of Man; such differences can be clearly seen in the
Spatial and temporal variability of limpet effects summer experiment where ephemerals did not grow at
all at Port St. Mary, but reached up to 50% cover at
The degree to which limpet removal allowed algal Derbyhaven. Temporal differences were only ob-
growth in the experimental manipulations can be exam- served at the seasonal scale with greater maximum
ined in a number of different ways. Firstly, we used data ephemeral cover in winter on the Isle of Man (SNK of
from control and exclusion plots but ignoring those from Location × Season). For soft algal crusts, differences
the half-fenced treatment to determine the level of between locations were clear; there was no develop-
consistency of the treatment (limpet removal). Secondly, ment of this algal group on the Isle of Man. In SW Eng-
we used data solely from limpet exclusion plots to deter- land, variability only occurred at the smallest spatial
mine how the level of algal growth following localised scale; differences between patches occurred in both
loss of limpets varied at all temporal and spatial scales. the winter and summer experiments.
Removal of limpets led to significantly greater fucoid
development (both maximum cover and area under the
Consistency of treatment effect curve) on the Isle of Man compared to SW England
(Fig. 1, Table 3). These differences between locations
A 6-factor mixed-model ANOVA was applied to the were consistent between experimental start dates and
data and the consistency of the treatment determined seasons. There was significant variability at the spatial
by applying the SNK multiple comparison test to the scales of patches and marginally insignificant variabil-
highest order significant interaction involving the fac- ity (at the 5% level) at the scale of the shore; SNK tests
tor ‘treatment’. For ephemeral algae, the effects of of these factors, for both measures of fucoid cover,
limpet removal were variable, with some limpet exclu- showed that such variability only occurred in SW Eng-
82 Mar Ecol Prog Ser 287: 77–86, 2005
Table 1. Mixed-model ANOVA of fucoid canopy cover in experimental plots of the control and grazer exclusion. Half-fenced
treatment is not included. Tr: treatment; Loc: location; Sh(Loc): shore (location); Se: season; Da: date; Pa: patch
Source Maximum cover Area under curve
Transformation: none Transformation: none
df MS F p MS F p F-ratio versus
Tr 1 193675 507.5 < 0.01 2947110 1656 < 0.001 Tr × Sh(Loc)
Loc 1 47313 132.1 < 0.01 1966680 769 < 0.01 Sh(Loc)
Sh(Loc) 2 358 1.8 > 0.2 2554 0.2 > 0.8 Da[Se × Sh(Loc)]
Se 1 0.02 0.007 > 0.9 186750 12.8 > 0.06 Se × Sh(Loc)
Tr × Loc 1 21042 55.1 < 0.05 1386350 779 < 0.01 Tr × Sh(Loc)
Tr × Sh(Loc) 2 381 2.4 > 0.1 1778 0.7 > 0.5 Tr × Da[Se × Sh(Loc)]
Tr × Se 1 295 0.9 > 0.4 103277 9.3 > 0.09 Tr × Se × Sh(Loc)
Loc × Se 1 1485 5.7 > 0.1 257803 17.7 > 0.05 Se × Sh(Loc)
Tr × Se × Loc 1 462.5 1.4 > 0.3 157580 14.2 > 0.06 Tr × Se × Sh(Loc)
Se × Sh(Loc) 2 261.3 1.3 > 0.3 14530 1.2 > 0.3 Da[Se × Sh(Loc)]
Tr × Se × Sh(Loc) 2 329 2.1 > 0.15 11110 4.2 > 0.05 Tr × Da[Se × Sh(Loc)]
Da[Se × Sh(Loc)] 8 193 0.6 > 0.7 11981 1.3 > 0.3 Pa{Da[Se × Sh(Loc)]}
Tr × Da[Se × Sh(Loc)] 8 156 0.4 > 0.8 2673 0.8 > 0.6 Tr × Pa{Da[Se × Sh(Loc)]}
Pa {Da[Se × Sh(Loc)]} 16 320 2.6 < 0.01 9056 3.9 < 0.001 Residual
Tr × Pa{Da[Se × Sh(Loc)]} 16 325 2.6 < 0.01 3358 1.5 > 0.1 Residual
Residual 128 123 2302
land (Table 3). For example, for SNK comparisons of Small-scale variability in fucoid cover
maximum fucoid cover between patches established
on the same date, 5 out of 8 tests were significant in SW Examination of variability at the smallest spatial
England compared to none on the Isle of Man. At the scale, that between individual experimental plots, was
spatial scale of shores, there were significant differ- made by determining the frequency distribution of
ences in SW England but not the Isle of Man; thus, for maximal percentage cover for exclusion and control
fucoid algae, there was less, but more variable cover in plots (Fig. 2). On the Isle of Man, the frequency distri-
SW England in limpet removal plots than in the more bution for maximum fucoid cover in exclusion plots
northerly locality. was skewed strongly to the right; 43 of the 48 exclusion
Table 2. Mixed-model ANOVA of ephemeral green algal cover in limpet exclusion plots. Loc: location; Sh(Loc): shore (location);
Se: season; Da: date; Pa: patch
Source Maximum cover Area under curve
Transformation: arcsin Transformation: ln(x + 1)
Cochran’s C = 0.1913, p > 0.05 Cochran’s C = 0.1570, p > 0.05
df MS F p MS F p F-ratio versus
Loc 1 163 0.03 > 0.8 16.07 1.05 > 0.4 Sh(Loc)
Sh(Loc) 2 5553 9.15 < 0.01 15.31 6.93 < 0.02 Da [Se × Sh(Loc)]
Se 1 10185 56.73 < 0.02 56.35 14.22 > 0.05 Se × Sh(Loc)
Loc × Se 1 2660 14.82 > 0.05 14.79 3.73 > 0.1 Se × Sh(Loc)
Se × Sh(Loc) 2 179 0.3 > 0.7 3.96 1.79 > 0.2 Da [Se × Sh(Loc)]
Da [Se × Sh(Loc)] 8 606 1.62 > 0.1 2.21 1.55 > 0.2 Pa {Da[Se × Sh(Loc)]}
Pa {Da[Se × Sh(Loc)]} 16 375 4.15 < 0.001 1.43 5.21 < 0.001 Residual
Residual 64 90 0.27
SNK test of Sh(Loc) SE = 5.03 SE = 0.304
Isle of Man: Shore 1 < Shore 2 Isle of Man: Shore 1 < Shore 2
SW England: Shore 1 = Shore 2 SW England: Shore 1 = Shore 2
SNK test of Loc × Se SE = 2.74
Isle of Man: Summer < Winter
SW England Summer = Winter
Summer and winter: Isle of Man = SW England
SNK test of Pa SE = 5.49 SE = 0.303
Isle of Man: 2/8 comparisons significant Isle of Man: 2/8 comparisons significant
SW England: 4/8 comparisons significant SW England: 4/8 comparisons significant
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 83
Table 3. Mixed-model ANOVA of fucoid canopy algal cover in limpet exclusion plots. Loc: location; Sh(Loc): shore (location);
Se: season; Da: date; Pa: patch
Source Maximum cover Area under curve
Transformation: none Transformation: ln(x)
Cochran’s C = 0.4604, p < 0.01 Cochran’s C = 0.1834, p > 0.05
df MS F p MS F p F-ratio versus
Loc 1 65730 90.89 < 0.02 92.88 117.28 < 0.01 Sh(Loc)
Sh(Loc) 2 723 3.68 > 0.05 0.79 4.11 > 0.05 Da [Se × Sh(Loc)]
Se 1 150 0.33 > 0.6 0.19 0.08 > 0.8 Se × Sh(Loc)
Loc × Se 1 1802 4.00 > 0.1 5.55 2.42 > 0.25 Se × Sh(Loc)
Se × Sh(Loc) 2 450 2.29 > 0.1 2.29 11.89 < 0.01 Da [Se × Sh(Loc)]
Da [Se × Sh(Loc)] 8 196 0.54 > 0.8 0.19 0.30 > 0.9 Pa {Da[Se × Sh(Loc)]}
Pa {Da[Se × Sh(Loc)]} 16 365 2.98 < 0.01 0.63 4.57 < 0.001 Residual
Residual 64 122 0.14
SNK test of Sh(Loc) SE = 2.86 SE = 0.09
Isle of Man: Shore 1 = Shore 2 Isle of Man: Shore 1 = Shore 2
SW England: Shore 1 < Shore 2 SW England: Shore 1 < Shore 2
SNK test of Pa SE = 6.39 SE = 0.21
Isle of Man: 0/8 comparisons significant Isle of Man: 0/8 comparisons significant
SW England: 5/8 comparisons significant SW England: 4/8 comparisons significant
a) Isle of Man
50 50
Exclusion treatment Control treatment
40
Frequency
40
30 30
20 20
10 10
0 0
0 25 50 75 100 0 25 50 75 100
b) SW England
50 50
40 40
Frequency
30 30
20 20
Fig. 2. Frequency distri-
bution of maximal fucoid 10 10
canopy cover values for
limpet exclusion and 0 0
control plots on (a) the 0 25 50 75 100 0 25 50 75 100
Isle of Man and in (b) SW
England Maximum percentage cover
plots established throughout the experiment had a cover in control plots showed no variation in SW Eng-
maximal cover of fucoids of over 90%. In contrast, the land, with no fucoid growth at all, while on the Isle of
frequency distribution in SW England was approxi- Man, although over half the experimental plots
mately normal, with the maximum cover of fucoids in showed no fucoid growth, there was extensive growth
exclusion plots showing high variability. Maximum in a minority of plots (Fig. 2)
84 Mar Ecol Prog Ser 287: 77–86, 2005
DISCUSSION the dominant patellid limpet grazers (Patella vulgata
on the Isle of Man and both P. vulgata and P. depressa
Local scale experimental studies have provided con- in SW England), regular maintenance of the experi-
siderable insight into the way shallow subtidal and ment prevented other macrograzers, including Litto-
intertidal communities are structured and organised. rina littorea, L. obtusata, Gibbula umbilicalis and Osil-
The tractable nature of rocky intertidal systems has inus lineata, from exerting a large effect. Other grazers
allowed them not only to provide a means of testing within the eulittoral zone, such as crustacean meso-
ecological theory, but means that rocky shore commu- herbivores, which were not manipulated, could poten-
nities are some of the best understood in the world, in tially contribute to the spatial and temporal differences
either terrestrial or marine environments. Despite this, observed. However, from our observations and the
it is well recognised that the results of many commu- known effects of these grazers (see Brawley 1992 for
nity-based field studies are context-dependent to a review), it is unlikely that they contributed signifi-
large extent (Lawton 1999), making generalisations cantly to the large differences between locations.
difficult. The key role played by grazing patellid The probability of early post-settlement stages of
limpets in controlling macroalgal development on fucoid macroalgae (zygotes, young germlings and
rocky shores of NW Europe has been recognised for germlings of Vadas et al. 1992) escaping grazing by
decades. The huge changes in community structure microphagous molluscs is critical in determining com-
after removal of limpets, especially on a large scale munity structure at the mid-tide level of rocky shores.
(Jones 1948, Southward & Southward 1978), suggest We propose that an increase in the probability of
that these are ‘keystone’ grazers (sensu Paine 1966). escape with increasing latitude is the prime driver for
However, the keystone effects of limpets are not uni- the increasing dominance of macroalgae to the north,
versal (e.g. Jenkins et al. 1999a), just as the effects of with fucoids extending further onto wave-exposed
the original keystone predator Pisaster ochraceus are shores. Such probability will change with the abun-
limited to certain habitats (Menge et al. 1994). In addi- dance, activity and possibly diversity of grazers, and
tion, the geographical generality of limpet-grazing the growth rate of algae. In areas of naturally reduced
effects on exposed shores of Europe are not fully grazing intensity or experimental exclusions, the
known. The majority of effective manipulations of probability of fucoid development lies solely with the
patellid limpets in the mid-shore zone have been con- supply of propagules and/or their ability to develop
centrated on the Isle of Man (see Hawkins et al. 1992 and grow. The dominant fucoid in experimental exclu-
for review). More recently, however, extensive experi- sion plots, Fucus vesiculosus, has a European range
mental manipulations of patellid grazers have been stretching from northern Norway as far south as
made in the Mediterranean (Benedetti-Cecchi et al. Morocco (Luhning 1990). A detailed study into the
2000, 2001). This work has shown extremely inconsis- biology of F. vesiculosus on the Isle of Man and Devon
tent grazing effects at different spatial and temporal (SW England) by Knight & Parke (1950) showed little
scales with only occasional strong effects on macroal- difference in the growth rate, reproductive period or
gal abundance. Low on the shore at exposed sites, ability to repopulate areas cleared within stands of
Boaventura et al. (2002) have convincingly shown that adults. Hence, in SW England, F. vesiculosus is well
limpet grazing can limit the vertical extent of turf form- within its distributional range and so expected to
ing and canopy algae in both England and Portugal. respond well to the release of grazing pressure. It
The results of the present study show, that within the appears unlikely that fucoid zygotes and germlings are
British Isles, the community response to small-scale unable to cope with slightly higher air and sea temper-
loss of limpets is similar; at both locations, macroalgae atures at the southern location, although this remains
developed after limpet removal. However, the level of to be tested experimentally.
determinism in response to simulated localised release Another explanation for the lower effect size in SW
of grazing pressure varies considerably. It is clear that England is a generally lower, less predictable supply of
the more northerly locality, the Isle of Man, experi- macroalgal propagules. Dispersal patterns of algal
ences a strong, deterministic community response to propagules are generally poorly understood, but
localised reductions in grazing pressure. In contrast, assessment of the distance of recruits from adult
further south in the British Isles, the response is weak sources have invariably suggested short dispersal
and more variable. These results support, and provide shadows, in the order of metres to 10s of metres (see
a mechanistic insight into, the observations of Ballan- Santelices 1990 for review). Limited dispersal dis-
tine (1961) of a latitudinal gradient in the balance tances can be increased if the number of source plants
between grazers and macroalgae across the wave- is increased. The effect of large stands of adult plants
exposure gradient. It should be noted that while the on propagule supply was graphically demonstrated by
main focus of our experiment was in manipulation of the early experiments of Burrows & Lodge (1950) when
Jenkins et al.: Determinism of grazing effects in the rocky intertidal 85
extensive fucoid development in a large limpet clear- whole quadrat would be expected to follow a binomial
ance (over 10 × 100 m) resulted in high fucoid recruit- distribution defined by the number of possible patches
ment downstream of the original experimental area. In and the likelihood of occupancy of single patches.
mid-shore experiments in northern Spain, increasing Lower likelihood of patch occupancy as seen in SW
distance from stands of Fucus spp. results in a decline England would give much greater variability in percent
in the colonisation and development of Fucus spp. cover as a result of such a binomial expectation.
canopy in grazer exclusion plots (F. Arenas pers. In summary, we show that the level of determinism
comm.). In the present study, no quantitative measures in the community response to small-scale limpet loss,
were made of the distribution of adult stands of F. in particular the development of a fucoid canopy,
vesiculosus in relation to experimental plots at either varies considerably within the British Isles. The level of
location. However, F. vesiculosus was generally abun- fucoid development was consistently high on the Isle of
dant on the Isle of Man and stands of adults were Man, compared to a low and more variable response
rarely if ever more than 30 m away from experimental further south in SW England. This work supports the
plots. In contrast, on the 2 shores of SW England, observations of Ballantine (1961) of a latitudinal gradi-
stands of F. vesiculosus were rarer and more patchily ent in the balance between grazers and macroalgae.
distributed, supporting the hypothesis that propagule Further experimental work is required to determine
supply was limiting. the causal mechanisms driving this change in balance.
It could be argued that the lower response to grazer
loss in SW England indicates a reduction in the role of Acknowledgements. This study was supported by the Mast III
limpet grazing in structuring mid-shore communities. project EUROROCK MAS3-CT95-0012. Thanks to E.
However, Jenkins et al. (2001) demonstrated an in- LaCroix, M. Roberts, S. Kimmance and D. Boaventura for
crease in abundance and overall grazing pressure of assistance with the experimental set-up and sampling. S.R.J.
and S.J.H. were supported during data analysis and write-up
patellid limpets in SW England compared with the Isle by NERC Grant-In-Aid to the MBA.
of Man, consistent with a general increase in grazing
pressure with declining latitude in Europe. These
LITERATURE CITED
observations would suggest an increase in the impor-
tance of grazers further south. One explanatory model Arrontes J (2002) Mechanisms of range expansion in the
to account for these results is that the activity of intertidal brown alga Fucus serratus in northern Spain.
microphagous grazers, of which patellid limpets are Mar Biol 141:1059–1067
Ballantine WJ (1961) A biologically defined exposure scale for
undoubtedly the most important, limits the develop-
the comparative description of rocky shores. Field Stud
ment of fucoid macroalgae in the south to such an 1:1–19
extent that adult plants are rare; thus, when localised Benedetti-Cecchi L, Airoldi L, Abbiati M, Cinelli F (1996)
areas of reduced grazing pressure arise, through nat- Estimating the abundance of benthic invertebrates: a com-
ural or experimental means, there is insufficient supply parison of procedures and variability between observers.
Mar Ecol Prog Ser 138:93–101
of propagules. Small scale experimentation cannot be Benedetti-Cecchi L, Bulleri F, Cinelli F (2000) The interplay of
used to determine the importance of limpet grazing at physical and biological factors in maintaining mid-shore
a shore-wide scale under this scenario. and low-shore assemblages on rocky coasts in the north-
Variability in propagule supply in SW England may west Mediterranean. Oecologia 123:406–417
Benedetti-Cecchi L, Bulleri F, Acunto S, Cinelli F (2001)
be one cause of the high variability in fucoid develop-
Scales of variation in the effects of limpets on rocky shores
ment at this location. Within the experiment, none of in the northwest Mediterranean. Mar Ecol Prog Ser 209:
the physical variables measured, shore height, aspect, 131–141
slope nor surface roughness could explain variability in Boalch G, Holme N, Jephson N, Sidwell J (1974) A resurvey of
either ephemeral or fucoid development. This is, Colman’s intertidal traverses at Wembury, South Devon.
J Mar Biol Assoc UK 54:551–553
however, not surprising as experimental plots at both Boaventura D, Alexander M, Della Santina P, Smith ND, Re P,
locations were chosen with relatively strict physical Da Fonseca LC, Hawkins SJ (2002) The effects of grazing
criteria, which minimised differences in the physical on the distribution and composition of low-shore algal
environment. Another explanation is based simply on communities on the central coast of Portugal and on the
southern coast of Britain. J Exp Mar Biol Ecol 267:185–206
the inherent probabilities of patch occupancy within
Brawley S (1992) Mesoherbivores. In: John DM, Hawkins SJ,
each exclusion quadrat for high and low mean values of Price JH (eds) Plant-animal interactions in the marine
fucoid cover. Percentage cover can be considered as a benthos. Oxford University Press, Oxford
measure of occupancy of a number of possible sites, or Brawley SH, Johnson LE (1991) Survival of fucoid embryos in
patches (Johnson et al. 1997), for Fucus spp. plants the intertidal zone depends upon developmental stage
and microhabitat. J Phycol 27:179–186
within each quadrat. If the presence of Fucus spp. in Burrows EM, Lodge SM (1950) Note on the inter-relationships
single patches is independent of the occupancy of other of Patella, Balanus and Fucus on a semi-exposed coast.
patches in a quadrat, then percentage cover for the Rep Mar Biol Stn Port Erin 62:30–34
86 Mar Ecol Prog Ser 287: 77–86, 2005
Burrows MT, Hawkins SJ (1998) Modelling patch dynamics Lewis JR (1964) The ecology of rocky shores. English Univer-
on rocky shores using deterministic cellular automata. sities Press, London
Mar Ecol Prog Ser 167:1–13 Lodge SM (1948) Algal growth in the absence of Patella on an
Colman JS (1933) The nature of intertidal zonation of plants experimental strip of foreshore, Port St. Mary, Isle of Man.
and animals. J Mar Biol Assoc UK 18:435–476 Proc Trans Liverpool Biol Soc 56:78–85
Hartnoll RG, Hawkins SJ (1985) Patchiness and fluctuations Lubchenco J, Gaines SD (1981) A unified approach to marine
on moderately exposed rocky shores. Ophelia 24:53–64 plant-herbivore interactions. I. Populations and communi-
Hawkins SJ (1981a) The influence of Patella grazing on the ties. Annu Rev Ecol Syst 12:405–437
fucoid-barnacle mosaic on moderately exposed rocky Luhning K (1990) Seaweeds, their environment, biogeo-
shores. Kiel Meeresforsch 5:537–543 graphy and ecophysiology. John Wiley & Sons, London
Hawkins SJ (1981b) The influence of season and barnacles on McCook, Chapman ARO (1991) Community succession fol-
algal colonization of Patella vulgata (L.) exclusion areas. lowing massive ice scour on an exposed rocky shore:
J Mar Biol Assoc UK 61:1–15 effects of Fucus canopy algae and of mussels during late
Hawkins SJ, Hartnoll RG (1983) Grazing of intertidal algae by succession. J Exp Mar Biol Ecol 154:137–169
marine invertebrates. Oceanogr Mar Biol Annu Rev 21: Menge BA, Berlow EL, Blanchette CA, Navarette SA,
195–282 Yamada SB (1994) The keystone species concept: varia-
Hawkins SJ, Hartnoll RG, Kain JM, Norton TA (1992) Plant tion in interaction strength in a rocky intertidal habitat.
animal interactions on hard substrata in the North-east Ecol Monogr 64:249–286
Atlantic. In: John DM, Hawkins SJ , Price JH (eds) Plant- Paine RT (1966) Food web complexity and species diversity.
animal interactions in the marine benthos. Systematics Am Nat 100:65–75
Association Spec Vol 46. Clarendon Press, Oxford, p 1–32 Paine RT (2002) Trophic control of production in a rocky
Jenkins SR, Hawkins SJ, Norton TA (1999a) Direct and indi- intertidal community. Science 296:736–739
rect effects of a macroalgal canopy and limpet grazing in Reed DC, Foster MS (1984) The effect of canopy shading on
structuring a sheltered inter-tidal community. Mar Ecol algal recruitment and growth in a giant kelp forest.
Prog Ser 188:81–92 Ecology 65:937–948
Jenkins SR, Hawkins SJ, Norton TA (1999b) Interaction Santelices B (1990) Patterns of reproduction, dispersal and
between a fucoid canopy and limpet grazing in structuring recruitment in seaweeds. Oceanogr Mar Biol Annu Rev
a low shore intertidal community. J Exp Mar Biol Ecol 233: 28:177–276
41–63 Southward AJ (1953) The ecology of some rocky shores in the
Jenkins SR, Norton TA, Hawkins SJ (1999c) Settlement and south of the Isle of Man. Proc Trans Liverpool Biol Soc
post-settlement interactions between Semibalanus bal- 59:1–50
anoides (L.) (Crustacea: Cirripedia) and three species of Southward AJ (1956) The population balance between lim-
fucoid canopy algae. J Exp Mar Biol Ecol 236:49–67 pets and seaweeds on wave-beaten rocky shores. Rep Mar
Jenkins SR, Arenas F, Arrontes J, Bussell J and 10 others Biol Stn Port Erin 68:20–29
(2001) A European scale analysis of seasonal variability in Southward AJ (1964) Limpet grazing and the control of
limpet grazing activity and microalgal abundance. Mar vegetation on rocky shores. In: Crisp DJ (ed) Grazing in
Ecol Prog Ser 211:193–203 terrestrial and marine enviroments. Blackwell Scientific
Johnson MP, Burrows MT, Hartnoll RG, Hawkins SJ (1997) Publications, Oxford, p 265–273
Spatial structure on moderately exposed rocky shores: Southward AJ (1979) Cyclic fluctuations in population density
patch scales and the interactions between limpets and during 11 years recolonisation of rocky shores in West
algae. Mar Ecol Prog Ser 160:209–215 Cornwall following the Torrey Canyon oil spill in 1967. In:
Johnson MP, Burrows MT, Hawkins SJ (1998) Individual Naylor E, Hartnoll RG (eds) Cyclic phenomena in marine
based simulations of the direct and indirect effects of plants and animals. Pergamon Press, Oxford, p 85–92
limpets on a rocky shore Fucus mosaic. Mar Ecol Prog Ser Southward AJ, Southward EC (1978) Recolonization of rocky
169:179–188 shores in Cornwall after use of toxic dispersants to clean
Jones NS (1946) Browsing of Patella. Nature 158:557 up the Torrey Canyon spill. J Fish Res Board Can 35:
Jones NS (1948) Observations and experiments on the biol- 682–706
ogy of Patella vulgata at Port St. Mary, Isle of Man. Proc Vadas RL, Wright WA, Miller SL (1990) Recruitment of Asco-
Trans Liverpool Biol Soc 56:60–77 phyllum nodosum: wave action as a source of mortality.
Knight M, Parke MW (1950) A biological study of Fucus vesicu- Mar Ecol Prog Ser 61:263–272
losus and Fucus serratus. J Mar Biol Assoc UK 29:439–514 Vadas RL, Johnson S, Norton TA (1992) Recruitment and
Lawton J (1999) Are there general laws in ecology? Oikos 84: mortality of early post-settlement stages of benthic algae.
177–192 Br Phycol J 27:331–351
Leonard GM (2000) Latitudinal variation in species inter- Velimirov B, Griffiths CL (1979) Wave-induced kelp move-
actions: a test in the New England rocky intertidal zone. ment and its importance for community structure. Bot Mar
Ecology 81:1015–1030 22:169–172
Editorial responsibility: Roger Hughes (Contributing Editor), Submitted: April 6, 2004; Accepted: August 24, 2004
Bangor, UK Proofs received from author(s): January 31, 2005