MARINE ECOLOGY PROGRESS SERIES
   Vol. 339: 131–141, 2007                                           Published June 6
                            Mar Ecol Prog Ser




 Limpet grazing and loss of Ascophyllum nodosum
     canopies on decadal time scales
         Andrew J. Davies1, 2,*, Mark P. Johnson1, Christine A. Maggs1
          1
            School of Biological Sciences, Queen’s University, 97 Lisburn Road, Belfast BT9 7BL, UK
2
 Present address: Scottish Association for Marine Science, Dunstaffnage Marine Laboratory, Oban, Argyll PA37 1QA, UK




     ABSTRACT: The role of limpet grazing in preventing the development of algal canopies is a recur-
     rent theme in intertidal ecology. Less is known about interactions of limpets with the long-term
     dynamics of established canopies. Aerial photographs indicate that intertidal canopy cover has
     declined over the past 44 yr in Strangford Lough, Northern Ireland. There has been a loss of the
     previously continuous cover of Ascophyllum nodosum (L.) Le Jolis in the mid-shore. A barnacle-
     dominated assemblage now fills gaps in the A. nodosum canopy. The rates at which barnacle patches
     become established and grow have increased since 1990. Changes in canopy cover have been
     accompanied by increases in limpet densities since the 1980s. Measurements between 2003 and 2004
     showed no increase in length of A. nodosum fronds when limpets Patella vulgata had access to the
     algal holdfasts. In contrast, when limpets were experimentally excluded from the holdfasts, there was
     net frond growth. In the Isle of Man, which is climatically similar to Strangford Lough but has fewer
     limpets, growth occurred regardless of limpet grazing. The breaking force for A. nodosum declined
     with increasing local densities of limpets. A. nodosum is a sheltered shore species, potentially vulner-
     able to changes in wave exposure. There is no evidence, however, that Strangford Lough has become
     windier over the past 3 decades. Variation in wave exposure among locations within the lough was
     not related to rates of barnacle patch creation or expansion. Limpet population density has increased
     following a series of mild winters. Climate change may have a role in causing canopy loss, not by
     direct effects on the growth of fucoids, but by increasing the severity of grazing through changes to
     limpet populations.

     KEY WORDS: Limpets · Ascophyllum nodosum · Patella vulgata · Grazing · Climate · Canopy loss ·
     Fucoid
                 Resale or republication not permitted without written consent of the publisher




            INTRODUCTION                    sum is relatively long-lived, with estimated holdfast
                                      ages exceeding 50 yr (Åberg 1992a,b). Given this
  A canopy of fucoid algae is frequently seen as the           longevity, it is perhaps unsurprising that the temporal
defining characteristic of sheltered rocky shores             dynamics of A. nodosum stands are not well under-
(Lewis 1964). Fucoids often act as foundation species,           stood. The majority of temporal work consists of
creating habitat and modulating the flow of resources           studying the responses to catastrophic disturbances,
to other organisms (Dudgeon & Petraitis 2005). A par-           whereby patches of mature plants are removed com-
ticularly striking example of such a foundation species          pletely from the shore (e.g. by scraping and/or burn-
is Ascophyllum nodosum (L.) Le Jolis, a mid-shore             ing: Keser & Larson 1984, Jenkins et al. 1999, Dudgeon
dominant along sheltered coasts of the North Atlantic.           & Petraitis 2001, Bertness et al. 2004). While these
A. nodosum canopies may have a high biomass                experiments can define the recovery time of canopies,
(Cousens 1984), and can influence biodiversity by             they do not necessarily reflect the dynamics of
facilitating other species (Jenkins et al. 1999). A. nodo-         A. nodosum in areas where large-scale removal is rare


*Email: andrew.davies@sams.ac.uk                      © Inter-Research 2007 · www.int-res.com
132                    Mar Ecol Prog Ser 339: 131–141, 2007




or absent, such as the ice-free shores of the NE        the loss of adult plants; (2) a change in strength of the
Atlantic.                           grazing effect of limpets on adult A. nodosum has
  In general, long-term time series studies of intertidal   increased the loss rate of adult plants. The 2 hypothe-
communities are a rarity, leaving ecologists to infer     ses are not mutually exclusive and A. nodosum loss
community dynamics on the basis of relatively short-      may result from an interaction between environmental
term experiments (Underwood 2000, but see Dye         and trophic processes. To further quantify the potential
1998). Where longer-term observations have been        role of limpets we manipulated grazing in 2 regions
made, it is still difficult to assess drivers of ecosystem   with similar climatic conditions but different densities
change. For example, in a 17 yr study of a shore in the    of limpets. The relationship between limpet density
Severn Estuary, England, Little & Kitching (1996)       and the breaking force for adult fronds was also esti-
identified wave action and limpet grazing as possible     mated.
factors involved in the loss of a fucoid canopy. It has
long been known that experimental removal of limpets
leads to proliferation of fucoids (e.g. Jones 1946, 1948).         MATERIALS AND METHODS
Such experiments have contributed to a consensus
that patellid limpets are the dominant grazers on        Aerial photography. Aerial photographic surveys
NE Atlantic shores (e.g. Hawkins & Hartnoll 1983).       of Strangford Lough (Fig. 1) were made in 1969,
Removing limpets commonly results in a bloom of        1994, 1997, 2001 and 2002 by the Environment and
ephemeral green algae, followed by a dense coverage      Heritage Service, Northern Ireland, and in 1962 and
of fucoid species. Hence much research has empha-       1988 by the Ordnance Survey, Northern Ireland.
sized the role of grazing in preventing establishment     Eleven locations photographed at more than one
of canopy-forming Fucus spp. (e.g. Southward 1964,       time were identified from these surveys (Table 1).
Southward & Southward 1978, Hawkins 1981, Haw-         Analysis was restricted to areas of bedrock as
kins & Hartnoll 1983, Jenkins et al. 1999, Thompson      changes in cover on boulder shores were difficult to
et al. 2004, Jonsson et al. 2006). In contrast to cases
involving the grazing of juvenile algae or recruits, the
loss of established canopies has been less studied, and
it is not clear what role grazing may play. Furthermore,
relatively few studies have recorded patellid grazing                             NI
in Ascophyllum nodosum-dominated areas (but see
Jenkins et al. 1999). Loss of adult plants is usually                               IOM
attributed to changes in physical disturbance (Little
& Kitching 1996) and has been observed as part of
                                               A
the recovery of limpet densities following oil spills
and related impacts (Southward & Southward 1978).                     B        K
Limpet grazing on established A. nodosum canopies
has been observed sporadically on NE Atlantic shores
                                                     J
(Fischer-Piette 1948, Southward 1964). It is therefore
possible that limpets may play a role in the long-term
dynamics of A. nodosum canopies by damaging or
                                             C    I
removing established adult algae.
                                           x
  For shores around the coast of Northern Ireland, pre-                 D      H
liminary observations suggested a trend of decreases
                                                   G
in Ascophyllum nodosum canopy cover over the last
                                               E
decade. We were able to use aerial photographs of the
intertidal zone in Strangford Lough (Northern Ireland)                      F
taken at various times between 1962 and 2002 to
assess this suggested change in canopy cover. The pos-
sibility that grazing is associated with this loss could be
assessed using survey data on limpet Patella vulgata L.
                                     12 km
densities between 1979 and 2004. Given these histori-
cal data, 2 potential hypotheses were tested as expla-
                                Fig. 1. Strangford Lough and photographed locations (X =
nations for the observations of A. nodosum canopy loss     Rathcunningham site; other abbreviations as in Table 1). Inset
(1) a change in environmental conditions (increased      shows locations of Strangford Lough (box), Northern Ireland
wave exposure, Little & Kitching 1996) has resulted in             (NI) and Isle of Man (IOM)
                   Davies et al.: Limpet grazing on Ascophyllum nodosum                133




Table 1. Algal and mean barnacle cover estimates for each     mid-shore therefore facilitates automated identifica-
available shore photograph for Strangford Lough. Estimated     tion of gaps in canopy cover (Ekebom & Erkkila 2003).
area of fucoid cover (Fuc. cover) in first photograph of each
                                  This approach was ground-truthed using GPS to trace
series was standardised to 100% cover to allow comparisons
of sites where fraction of intertidal varied (as a proportion of  the outlines of 7 barnacle patches during 2001. In the
the georeferenced 100 ha areas used to overlay photographs     field an average area of 404.04 m2 (SE = 145.8) was
             in the GIS)               recorded. The same patches were identified in aerial
                                  photographs from 2001. An average patch area of
Site         Year Fuc. cover Barnacle patches      351.06 m2 (SE = 108.7) was recorded (error = 13 %). For
                (%)    n   Size (m2)      each photograph, individual barnacle patches were
                                  extrapolated into polygon shapefiles to measure patch
A: Mahee Island N   1962   100     0    0
            1969   100     0    0      surface area (total area of barnacle patches in each
            1988   98.7    14    73.5     photographed location) and frequency (number of bar-
            1994   97.6     9   211.6     nacle patches at each photographed location). The ini-
            1997   98.3    10   135.5     tial cover of algal canopy was measured for the first
            2001   94.7    14   293.0
            2002   95.3    23   158.7     photograph at each site to provide a baseline to esti-
B: Mahee Island S   1994   100     16   114.8     mate canopy loss. The annual rates of change for bar-
            2001   95.9    24   130.2     nacle and fucoid cover were estimated from successive
            2002   93.7    31   122.6     pairs of photographs at each site, with data plotted at
C: Ringdufferin    1994   100     9    94.3
                                  the mid point of the 2 years used to estimate the rate
            2001   99.0    18   128.2
D: Taggart Island   1969   100     0    0      of change.
            1994   98.9    26    56.4      Limpet abundances. Several different sets of survey
            2001   95.7    37   149.4     data were collated to estimate the extent of change in
E: Chapel Island   1994   100     4    91.9     limpet abundance over time. Information on limpet
            2001   99.9     5   113.9
F: Audley’s Castle  1994   100     44    72.2     density for each aerial photograph location was col-
            2001   97.0    28   148.4     lected from 20 haphazardly thrown quadrats (0.25 m2)
G: Marlfield Bay   1994   100     27    40.2     in the intertidal during summer 2002. At Rathcunning-
            2001   90.2    48    86.0     ham Quay, Strangford Lough (Fig. 1, Site X), the abun-
H: Priest Town    1994   100     40    95.7
                                  dance of limpets (Patella vulgata) boulder–1 (n = 232)
            2001   94.0    56   116.8
I: Lady’s Port    1994   100     13    55.5     had been surveyed in 1979 (Boaden & Dring 1980) and
            2001   99.2    11   172.4     the same methodology (Boaden & Dring 1980) was
J: Black Neb     1994   100     27    81.1     replicated in 2000 (n = 10) and 2004 (n = 58) to remain
            2001   99.5    28    95.0     consistent with earlier counts in 1979. Limpets were
K: Kircubbin     1962   100     6    2.3
            1994   99.3    23    32.7     counted on boulders with a horizontal circumference
            2001   97.4    27    99.9     of approximately 1 m. In each case, all boulders of the
                                  target size were examined, as encountered, along a
                                  transect in the mid shore at the Rathcunningham site.
                                   At Taggart Island, Strangford Lough (Fig. 1: Site D)
quantify in photographs. A fixed 100 ha area was          limpet data were available from both the 1986 North-
marked out in photographs for each location using a        ern Ireland Littoral Survey (Wilkinson et al. 1988) and
geographical information system (GIS, ArcInfo). This        the 2003 Strangford Lough Ecological Change Survey
provided a means for standardising comparison of          (Roberts et al. 2004). Both surveys used the same
photographs among different dates. Between 5 and          methodology. Densities were estimated in 0.25 m2
10 control points per location were identified for geo-      quadrats (1986, n = 16; 2003, n = 35), but converted into
referencing areas in photographs from different          an 8-point categorical scale to describe the mean
dates (Caloz & Collet 1997) and recorded in the field       abundance of species within the vertical height
during 2001 and 2002 using geographical positioning        limits in which they were found. As the raw data
systems (GPS).                           were not available, categorical estimates were back-
 Areas of fucoid canopy in photographs contrast          transformed to estimates of population density by tak-
sharply with the barnacle cover that dominates in the       ing the log mid point of each category (thus, a category
absence of macroalgae. Although it was not possible to       indicating abundances between 10 and 99 limpets has
distinguish the Ascophyllum nodosum canopy from          a mid point of 55 or log mid point of 3.45, see Burrows
other fucoids in photographs, the majority of mid-shore      et al. 2002). The loss of information during this process
areas in Strangford Lough are dominated by A. nodo-        is more likely to have obscured differences between
sum (Brown 1990). The presence of pale white or grey        surveys than to have created artefactual changes in
barnacle patches against the darker algal-dominated        abundance.
134                    Mar Ecol Prog Ser 339: 131–141, 2007




  Physical factors. A cartographic method was used to      Manipulative experiments. To determine the effect
estimate temporal changes in wave exposure con-        of limpets on the frond length of adult Ascophyllum
current with the aerial photograph time series. The      nodosum a hierarchical experimental design was
exposure index was calculated on an annual basis for      employed. Experiments were carried out in 2 regions,
400 locations spaced at 0.5 km intervals along the       Strangford Lough and the Isle of Man. The Isle of
shoreline of Strangford Lough. As the lough has a nar-     Man lies 70 km to the SE of Strangford Lough. It was
row connection to the Irish Sea, waves are determined     selected as a second region because previous studies
by local winds without any influence from open water      have shown it to have a lower density of limpets, yet
swells. The model was based upon 2 factors, fetch dis-     it is both climatically and biologically similar to
tance and consensus wind speed. Using a GIS routine,      Strangford Lough. At the sites used for experiments,
fetch distances for each of the locations were calcu-     the average density of limpets was 29 m–2 at sites in
lated as the distance of open water along 36 compass      the Isle of Man compared to 115 m–2 in Strangford
bearings at 10° intervals.                   Lough (0.25 m2 quadrats, n = 20 site–1). Two sites
  Daily wind records consisting of wind speed and       were randomly chosen from those available in each
wind direction spanning 1972 to 2003 were obtained       region that exhibited > 50% cover of A. nodosum
from 3 local wind stations situated around the lough      interspersed with Patella vulgata. At each site, there
(< 2 km from the shoreline). Data from each wind sta-     were 7 replicates of each of 3 treatments. The 3
tion were transformed to a mean speed of zero with       experimental treatments were (1) square enclosures
unit standard deviation to allow records from sites with    of 30 × 30 cm surrounded by 1 cm mesh rabbit-wire
different mean speeds to be averaged. The consensus      fences to prevent limpet access (exclusions), (2) par-
wind speeds took into account variable wind speed       tial fences with a gap of 5 cm in the middle of each
and direction recorded at different sites as a result of    side as a procedural control and (3) an open treat-
modification by the surrounding topography (Klaic et      ment, marked only with screws (controls). Each treat-
al. 2002). Therefore, they were considered to be a more    ment was centred on an individual adult A. nodosum
reliable basis for extrapolation of wave exposure than     plant, randomly allocated to 1 of the experimental
records from any single site.                 treatments. There was a minimum spacing of 1 m
  Relative wave exposure was estimated by multiply-      between experimental replicates. All enclosures and
ing the square of average consensus wind speed by the     controls were initially cleared of grazers (in May
fetch distance along each 10° bearing (modified after     2002) and frond length of the A. nodosum plant stan-
Thomas 1986). To avoid confounding changes in loca-      dardised to 25 cm. Frond length for each A. nodosum
tion with time, estimates of the mean exposure and       individual was recorded at approximately 2 mo inter-
change in exposure were estimated for 1994 to 2001 for     vals for 14 mo. Prior to analysis using a mixed-model
each location. These dates provided the highest num-      ANOVA, data were examined for heteroscedasticity
ber of paired photographs (n = 11) for comparison over     (‘cage type’ and ‘region’ as fixed factors, ‘site’ as
the same time period. The short-term change in expo-      random factor, nested within ‘region’).
sure at each location was calculated as the rank corre-      Breaking force for Ascophyllum nodosum. The
lation between annual relative exposure and year.       breaking force for A. nodosum fronds (n = 94) was esti-
Hence a positive trend in exposure indicates that wind     mated from haphazardly selected individuals close to
speeds have been increasing between 1994 and 2001       the experimental and survey sites. Around the base of
and/or winds along relatively longer fetches are be-      each individual, local limpet densities were recorded
coming more common for a particular location. In addi-     within an 0.25 m2 quadrat centred on the plant. Break-
tion to mean relative exposure the variance of the time    ing force was estimated for 1 randomly selected frond
series was used as a measure of the potential for rela-    by attaching a grommet below the basal internode
tively extreme years to affect algal cover. The 8 yr sum-   (McEachreon & Thomas 1987). A spring scale (0 to
maries (mean, variance and trend) were used as pre-      2500 g) with a maximum force recorder was hooked to
dictor variables for the change in patch frequency or     the grommet and steadily pulled vertically; if the frond
area between 1994 and 2001 at the 11 photographed       did not break, the procedure was repeated on the same
locations. Longer timescale trends in wind speed were     frond using a 0 to 10000 g spring scale. If the breaking
analysed using regression to determine whether mean      force of the second pull did not exceed that of the first
wind speed had decreased or increased over the 1972      pull, weakening of the frond was assumed and the
to 2003 time period. In case mean wind speed was not      frond omitted from analysis. Broken fronds were
a good indicator of the potential for algal loss in storms,  retained to record the number of limpet grazing marks
the overall trend in records of strong winds to gales     (defined as rasped areas formed by the characteristic
(annual proportion of records >12 m s–1) was also       sweeping movements of limpets per centimetre of
examined between 1972 and 2003.                frond).
                 Davies et al.: Limpet grazing on Ascophyllum nodosum                 135




            RESULTS
                                                      1962
    Temporal change in barnacle patches
       and limpet abundances

  There has been a change in algal cover since 1962,
when the photographed shores were almost totally
covered in algae, to a canopy interspersed with
patches of barnacles (Fig. 2). From the predominantly
mid-shore development of patches and from examina-
tion of locations, it is clear that most of the lost canopy
consisted of Ascophyllum nodosum. The rate of patch
formation was estimated for successive pairs of photo-                           1988
graphs at each location. Since 1962, the average
annual barnacle patch formation rate at each location
has been 1.43 yr–1 (SE 0.595, significantly greater than
zero in a Student’s t-test, p < 0.05). Only 3 changes
in patch frequency were negative (i.e. a decrease in
patches between photos, Fig. 3). For these 3 cases, the
corresponding change in barnacle patch area was pos-
itive, implying coalescence and expansion of existing
patches. The correlation between year and change in
patch frequency was positive, implying that the rate of
patch formation has increased over time (Spearman’s                             1994
rs = 0.714, p < 0.05). This pattern does not seem to have
resulted from changes in the locations used at different
times. Individual locations all showed a net increase
in patch formation over time (Locations A, B, D and K
all increased in rate overall).
  As with the patch formation rate, the estimated rate
of loss of canopy cover appears to be increasing
(Fig. 4). The correlation between rate of change in per-
centage cover and year was negative, indicating that
the loss rate of fucoid cover has increased over time
(rs = 0.479, p < 0.05). There was an average estimated
reduction of 3% of canopy between 1994 and 2001 (the                            2002
years for which most data exist). This loss represents a
1.7 ha decrease in algal cover within the lough.
  The available data on limpet density imply large
increases in limpet density at the surveyed sites.
Limpet numbers were greater in recent surveys at both
Rathcunningham and Taggart Island (1-way ANOVAs,
Rathcunningham: F2,299 = 230.59, p < 0.05; Taggart
Island: F1,50 = 51.27, p < 0.05, Fig. 5).


  Barnacle patch and limpet densities in relation
                               Fig. 2. Development of barnacle patches over time (arrowed) in
         to physical factors             the intertidal zone of Mahee Island, Strangford Lough. Spacing
                                    within pairs of arrows is approximately 50 m
 There were no clear changes in wind speeds in the
consensus data for Strangford Lough over the 1972 to
2003 period. Linear regressions of mean annual wind      strong winds to gales [negative slope]: r2 = 0.086, p =
speed and frequency of strong winds to gales (speeds     0.109).
>12 m s–1) were not significant (mean wind speed        Interactions between wind direction and fetch cre-
[negative slope]: r2 = 0.140, p = 0.551; frequency of     ate variation in wave exposure among locations. This
  136                                    Mar Ecol Prog Ser 339: 131–141, 2007




                 10                                              12                           8
                                          A




                                                 Limpet density per boulder
                 8
                                                               10




                                                                                             Log limpet density m–2
                                          B
                                                                                           6
    No. of patches




                 6
                                                                8
                 4
                                       G                        6                           4
                 2                      H
                                       D
                                       CA J
                            A D          B
                 0     A      K          AK                        4
                                       E
                                       I                                                   2
                                   A
                 –2                                              2
                                       F

                 –4                                              0                         0
                 1960     1970  1980   1990     2000   2010                 1960    1970  1980    1990   2000   2010
                                Year                                            Year
  Fig. 3. Rates of change in barnacle patch numbers at each                        Fig. 5. Patella vulgata. Estimated mean (± SE) limpet density
  photographed location. Changes calculated as annual net                         in different years at Rathcunningham (shaded bars, left ordi-
  change between successive dates with point plotted at the                           nate) and Taggart Island (white bars, right ordinate)
   mid point of the 2 years used to estimate rate of change




                 1.0                                    Table 2. Correlations between potential drivers of changes
Change in algal cover (%) yr–1




                                                      in canopy cover and observed change in canopy cover at
                 0.5                                    each location photographed in both 1994 and 2001 (n = 10).
                                                      Patch frequency: number of patches in each photograph;
                 0.0                                    patch area: estimated area of barnacle patches in each photo-
                                                      graph. No correlations were significant; lowest probability
                 –0.5                                          associated with a coefficient was 0.16

                 –1.0                                             Parameter       Relative exposure    Limpet
                                                                            1994–2001       count
                 –1.5                                                       Mean Variance Trend    (2002)

                 –2.0                                             Change in
                                                                patch frequency  0.237   0.499  –0.482  0.315
                 –2.5                                              patch area    0.110  –0.164  –0.419  0.141
                   1960    1970  1980     1990    2000  2010
                              Year
  Fig. 4. Ascophyllum nodosum. Percentage change in algal                                     Effects of limpets on Ascophyllum nodosum
  cover between pairs of dated photographs, plotted at mid-
       term point between dates of photographs
                                                        The frond lengths of Ascophyllum nodosum individu-
                                                      als after 14 mo were similar in the Isle of Man and
                                                      Strangford when limpets were prevented from grazing
  variation did not, however, influence the creation of                          (Fig. 6). The regions differed when comparing across-
  barnacle patches during the 1994 to 2001 period.                             cage designs that allowed limpets access to A. nodosum
  There was no relationship between the mean relative                           (significant cage type × region interaction: F2,4 = 11.9,
  exposure index, the variance or the temporal trend in                          p < 0.05). Frond lengths increased regardless of grazing
  the exposure index and changes in patch frequency                            in the Isle of Man, but there was no net increase in
  or the total area of barnacles (Table 2). Relative expo-                         frond lengths of grazed A. nodosum after 14 mo in
  sures decreased on average over the photographed                             Strangford. There were no differences between sites
  locations between 1994 and 2001 (average correlation                           within a region (cage type × site (region) interaction:
  between annual exposure and year, rs = –0.62, SE                             F4,59 = 0.39, p > 0.05). In Strangford Lough, but not in
  0.080, significantly different from zero, Student’s t =                         the Isle of Man, limpets were observed to trap A. nodo-
  7.72, p < 0.05). Mean limpet densities at the photo-                           sum fronds under the shell and graze on the trapped
  graphed locations were also unrelated to exposure or                           fronds (Fig. 7a,b) and were also commonly observed
  patch variables.                                             aggregating around A. nodosum holdfasts (Fig. 7c).
                         Davies et al.: Limpet grazing on Ascophyllum nodosum                              137




             14                                           8000




                                       A. nodosum breaking strength (g)
             12
A. nodosum growth (cm)




                                                        6000
             10

             8                                           4000

             6
                                                        2000
             4

             2                                            0
             0

             –2                                              0  10   20   30   40   50    60
               Strangford Lough    Isle of Man                                 Limpet density (m2)

Fig. 6. Ascophyllum nodosum. Mean (+ SE) changes in frond          Fig. 8. Ascophyllum nodosum. Relationship between break-
length in experimental treatments after 14 mo. Treatments          ing strength and density of Patella vulgata adjacent to hold-
shown as: fences to exclude limpets (black bars), partial                fast. Fitted line is a linear regression
fences as a procedural control (grey bars) and unfenced
         treatments (white bars)


                                                               Increased densities of limpets were
                                                              associated with weaker Ascophyllum
                                                              nodosum fronds (Fig. 8, r2 = 0.085, p <
                                                              0.01). A. nodosum with higher densi-
                                                              ties of limpets in the immediate area
                                                              had a greater frequency of grazing
                                                              marks (r = 0.239, p < 0.05), structurally
                                                              weakening the individual. Grazing
                                                              therefore seems to increase the sensi-
                                                              tivity of A. nodosum to frond breakage.


                                                                    DISCUSSION

                                                               Aerial photographs clearly show the
                                                              loss of algal canopy from mid-shore
                                                              hard substrata in Strangford Lough.
                                                              Over the last few decades, once-
                                                              continuous canopies of Ascophyllum
                                                              nodosum have become punctuated by
                                                              barnacle-dominated patches. The rate
                                                              at which these barnacle patches are
                                                              created in the algal canopy appears
                                                              to have increased since 1990 and at
                                                              the same time the total area of these
                                                              patches has also been increasing.
                                                              Such changes in canopy cover, in-
                                                              volving the replacement of primary
                                                              producers with filter-feeders, will
                                                              influence the ecosystem functioning
                                                              of the lough, potentially altering the
Fig. 7. Patella vulgata. Limpet grazing behaviour in Strangford Lough, showing
Ascophyllum nodosum frond trapped under limpet shell at (A) Mahee Island
                                                              flows of carbon and/or nutrients be-
and (B) Marlfield Bay. (C) Limpet clumping behaviour around holdfasts of                          tween the intertidal and other coastal
                solitary A. nodosum                                     habitats.
138                    Mar Ecol Prog Ser 339: 131–141, 2007




  Experimental manipulation of limpet grazing de-                       80




                                Frequency of days subzero (°C)
monstrated that the present densities of limpets in the
                                                70
lough are capable of preventing the growth of estab-
lished Ascophyllum nodosum. In addition, increased                       60
limpet density around holdfasts was associated with
decreases in the breaking force of A. nodosum fronds.                      50
Limpets therefore increase the vulnerability of fronds
to wave-induced breakage. In Strangford Lough, the                       40
observed increases in limpet densities may therefore
                                                30
cause loss of A. nodosum through direct grazing of
established plants. Observations after the Torrey                        20
Canyon oil spill and in Brittany have previously sug-
gested that extreme increases in limpet density are                       10
                                                 1960  1970  1980   1990  2000
sufficient to cause the loss of established algal
canopies (Southward & Southward 1978, Hawkins &                                 Year
Southward 1992, Le Roux 2005).                Fig. 9. Frequency of days with surface air temperatures below
  Demographic and environmental processes may also      0°C (Jones & Lister 2004) at Strangford Lough. Dashed line:
contribute to the loss of canopy, but there is little evi-     mean of time-series; solid line: 5 yr running mean
dence for such factors acting in Strangford Lough. A
lack of algal recruitment could potentially lead to a     the Irish Sea. Records from the Port Erin breakwater
decline in canopy as part of an intrinsic long-term      (Isle of Man) show an increase of approximately 1°C in
cycle, but there is little evidence to support suppres-    average Irish Sea surface temperatures over the last
sion of recruitment by the established canopy. Demo-     100 yr, with a current mean of approximately 11°C
graphic analyses of Ascophyllum nodosum populations      (Evans et al. 2003). Maximum temperatures of 15°C
indicate that population growth rate is more sensitive    were recorded in intertidal areas of Strangford Lough
to changes in the survival of existing plants than to     during short-term (7 d) temperature logging in August
variations in recruitment (Åberg 1992a,b). The esti-     (Strong 2003). All these temperatures seem well within
mated lifespan of A. nodosum holdfasts is 50 to 60 yr in   the tolerance limits of A. nodosum, which has been
areas with sea ice and will exceed this in ice-free areas   shown to grow more rapidly with warming until a
(Åberg 1992b). Given such a long lifespan, the loss of    threshold of between 19 and 25°C is reached (Keser et
canopy during the 1990s must have resulted from an      al. 2005). The experimental manipulation in Strangford
increase in the loss of established A. nodosum. As an     confirms that adult fronds are capable of growing
alternative to a trophic interaction (grazing), A. nodo-   under current environmental conditions, as long as
sum may be responding to other changes in the envi-      limpets are excluded.
ronment of Strangford Lough. Eutrophication has been       Ascophyllum nodosum is a sheltered-shore species
associated with decreases in fucoid cover (Vogt &       and canopy loss may therefore be related to increased
Schramm 1991), but investigations during the 1990s      storm frequency (and therefore waves) as a result of
concluded that Strangford Lough was not eutrophic       changing wind climate (Thompson et al. 2002). There
(Service et al. 1996).                    is no evidence for increases in wind speeds and wave
  Climate change may affect the geographic distribu-     exposure over the last few decades in Strangford. The
tion of Ascophyllum nodosum. In the Atlantic, fucoids     rates of canopy loss showed no association with varia-
are more common on shores at higher latitudes.        tion in mean estimated wave exposure among loca-
Increases in air and sea temperatures are therefore      tions in the lough.
expected to cause the ranges of fucoid algae to move       The grazing experiment indicates that the present
northwards as shores at the southern range limits       densities of limpets in Strangford are preventing Asco-
become too warm (Kendall et al. 2004). Strangford       phyllum nodosum growth by grazing, ultimately caus-
Lough, however, is not at the southern range limit of     ing damage to fronds. A number of factors may have
A. nodosum (Lüning 1990). There is recorded evidence     caused the recent increase in mean limpet densities to
for climate change over the last few decades in North-    the level at which canopy cover becomes affected.
ern Ireland. Air temperatures have increased, leading     Changes in climate, particularly from 1990, have led to
to fewer frosts, particularly since 1990 (Fig. 9). There   less severe winter air (Fig. 9) and sea (Evans et al.
are no continuous records of sea surface temperature     2003) temperatures. Limpet recruitment is lower and
for Strangford Lough. However, given the lough’s esti-    mortality is higher in cold winters (Crisp 1964, Bow-
mated flushing time of 1.6 d (Service et al. 1996), tem-   man & Lewis 1986). The milder winters of the 1990s
perature in the lough is expected to be close to that of   may therefore have reduced density-independent
                 Davies et al.: Limpet grazing on Ascophyllum nodosum               139




restrictions on limpet populations, leading to the ob-    spp. (Southward & Southward 1978). Such patterns
served increases in population size. Such thermal lim-    were observed after the Torrey Canyon oil spill. Fol-
its to populations have been suggested for other mol-    lowing periods of canopy loss resulting in food short-
luscs (Thieltges et al. 2004). Thermal effects on grazer   age, large-scale reductions in limpet density occurred
populations are also apparent from increased popula-     (Southward & Southward 1978, Hawkins & Southward
tion density in response to artificial warming by power   1992). If consumption of A. nodosum is subsidising
station discharges (Schiel et al. 2004). Limpet popula-   high limpet densities that cannot be sustained by other
tions could also have increased due to declines in      food sources (see Bustamante et al. 1995), a food short-
predator densities; however, evidence from bird counts    age and reduction of limpet numbers seems likely if
suggests the opposite. Oystercatchers Haematopus       the A. nodosum canopy is totally lost. However, the
ostralegus L. are considered to be important predators    potential benefits that limpets gain from consuming
of limpets (Coleman et al. 1999). Bird counts indicate    A. nodosum have not yet been assessed.
an increase in oystercatcher population size of 83%       The loss of Ascophyllum nodosum in Strangford has
over the past 25 yr, with the steepest rises occurring in  parallels with declines in fucoid canopies in the Baltic.
the 1990s (Maclean et al. 2005). This implies that the    Along with eutrophication in the Baltic, increases in
predation pressure on limpets may have increased       grazers (mesograzers: the isopod Idotea baltica Pallas)
during recent decades.                    following mild winters are thought to have caused
  The trend of canopy loss seems likely to continue if   reduction in algal belt widths and in percentage cover
limpet populations remain at their current levels      along the coasts of Sweden (Engkvist et al. 2000, Nils-
within the lough. In the short term, canopy loss may     son et al. 2004). The damaging level of canopy grazing
accelerate as observed in the frequency of patch for-    observed in Brittany (Le Roux 2005) and the Baltic
mation and changes in algal canopy since the 1990s.     appears to be a regional phenomenon. It is not always
Ascophyllum nodosum canopy indirectly limits limpet     clear why canopy overgrazing should be limited to
populations by supporting an understorey of red algal    particular locations. Limpets have been observed graz-
turf, which is an unsuitable habitat for limpets (Jenkins  ing on A. nodosum fronds in a wide range of locations
et al. 1999, 2004). When the turf breaks down follow-    (e.g. Brittany, France [Le Roux 2005]; Milford Haven,
ing canopy removal, there is often a large increase in    Wales; all Irish coasts; Plymouth, England and west
limpet density (Jenkins et al. 2004). In Strangford     coast of Scotland [C. A. Maggs, M. T. Burrows and
Lough, these new barnacle- and limpet-dominated       S. J. Hawkins, respectively, pers. comm.]). The extent
areas of the mid-shore may be relatively persistent.     to which this grazing affects canopies may depend on
Other authors have shown that switching between       limpet density. Regional variations in climate are likely
assemblages can be stable and may persist for long      to influence the density of limpets at broad scales.
periods. For example, Petraitis & Dudgeon (1999) have    Local hydrography can further modify recruitment pat-
suggested that large-scale removal of A. nodosum       terns (particularly in restricting grazer populations on
from sheltered shores in New England may lead to a      the Isle of Man, see Norton et al. 1990 for littorinids).
stable alternative assemblage dominated by mussels.     The diversity of observed temporal trends in limpet
Other authors have considered that the factors promot-    populations reported by Burrows et al. (2002) presum-
ing A. nodosum beds are more predictable, such that     ably reflects these local influences on larval supply and
any disturbed canopy will eventually revert to A. nodo-   recruitment.
sum dominance (Bertness et al. 2002, 2004), although      An interaction between limpets and the canopy-
the period of recovery may extend over decades (Jenk-    forming alga Ascophyllum nodosum has been recor-
ins et al. 2004). As each of these studies deals with    ded experimentally herein, for the first time. Limpets
different physical habitats, a generalisation cannot be   are increasing in density, perhaps driven by enhanced
made about how shores dominated by A. nodosum        survivorship through recent favourable winter condi-
canopies will respond to disturbance.            tions. Analysis of canopy changes using aerial photo-
  Ascophyllum nodosum recruits have been observed      graphs has shown an accelerated loss of canopy and
in locations within Strangford from which limpets have    the continuing emergence of barnacle patches over the
been removed, indicating that A. nodosum canopies      last 40 yr. Experimental manipulations have indicated
have the potential to recover (C. A. Maggs unpubl.      that limpets may be responsible for the loss of estab-
data). However, it is likely that any future recovery    lished canopy. The limpet –A. nodosum interaction
would not occur rapidly. Limpet densities would not     demonstrates how separate trophic levels can poten-
decrease immediately with the loss of A. nodosum       tially respond differently to climate change. Assess-
canopy, as the limpets would feed upon the microbial     ments of climate change impacts are often made on a
biofilm (Hill & Hawkins 1991, Thompson et al. 2004)     single-species basis with respect to the assumed cli-
and might seek alternative food supplies such as Fucus    mate envelope required by that species. As limpets
140                     Mar Ecol Prog Ser 339: 131–141, 2007




appear not to respond to the same climatic cues as          changes in rocky littoral fauna from South Africa. Mar
algae, the predicted shifts in range may be influenced        Ecol Prog Ser 164:47–57
                                 Ekebom J, Erkkila A (2003) Using aerial photography for
at small scales by mismatches between the responses
                                   identification of marine and coastal habitats under the
of fucoids and their most important grazer. If milder        EU’s Habitats Directive. Aquat Conserv: Mar Freshw
winters are acting as a trigger for local increases in        Ecosyst 13:287–304
limpet abundance, there are likely to be more cases of      Engkvist R, Malm T, Tobiasson S (2000) Density dependent
A. nodosum canopy loss in the temperate NE Atlantic.         grazing effects of the isopod Idotea baltica Pallas on
                                   Fucus vesiculosus L. in the Baltic Sea. Aquat Ecol 34:
                                   253–260
                                 Evans GL, Hardman-Mountford NJ, Hartnoll RG, Kennington
Acknowledgements. E. Pothanikat, J. Hodgkinson, C. Mc-        K, Mitchelson-Jacob EG, Shammon T, Williams PJLeB
Ilveen and C. Hetherington surveyed Rathcunningham in        (2003) Long-term environmental studies in the Irish Sea:
2000. M. Edwards and A. Portig assisted with fieldwork and      a review. Scientific Rep No. 02 Defra Contract CDEP
GIS and data retrieval, respectively. The authors thank Peter    84/5/311, Marine Biological Association, Plymouth
Petraitis, Michael Burrows, S. J. Hawkins and 2 anonymous    Fischer-Piette E (1948) Sur les éléments de prospérité des
reviewers for valuable comments that improved this manu-       patelles et sur leur spécificité. J Conchol 88:45–96
script. A.J.D. was supported by a DEL scholarship.        Hawkins SJ (1981) The influence of Patella grazing on the
                                   fucoid/barnacle mosaic on moderately exposed rocky
                                   shores. Kiel Meeresforsch 5:537–543
          LITERATURE CITED               Hawkins SJ, Hartnoll RG (1983) Grazing of intertidal algae by
                                   marine invertebrates. Oceanogr Mar Biol Annu Rev 21:
Åberg P (1992a) A demographic study of two populations        195–282
  of the seaweed Ascophyllum nodosum. Ecology 73:        Hawkins SJ, Southward AJ (1992) The Torrey Canyon oil
  1473–1487                             spill: recovery of rocky shore communities. In: Thayer GW
Åberg P (1992b) Size-based demography of the seaweed         (ed) Restoring the nation’s marine environment, Maryland
  Ascophyllum nodosum in stochastic environments. Eco-       Sea Grant College, College Park, MD, p 583–631
  logy 73:1488–1501                       Hill AS, Hawkins SJ (1991) Seasonal and spatial variation of
Bertness MD, Trussell GC, Ewanchuk PJ, Silliman BR (2002)      epilithic microalgal distribution and its ingestion by
  Do alternate stable community states exist in the Gulf of     Patella vulgata on a moderately exposed rocky shore.
  Maine rocky intertidal zone? Ecology 83:3434–3448         J Mar Biol Assoc UK 71:403–423
Bertness MD, Trussell GC, Ewanchuk PJ, Silliman BR, Crain    Jenkins SR, Hawkins SJ, Norton TA (1999) Direct and indirect
  CM (2004) Consumer-controlled community states on Gulf      effects of a macroalgal canopy and limpet grazing in struc-
  of Maine rocky shores. Ecology 85:1321–1331            turing a sheltered inter-tidal community. Mar Ecol Prog
Boaden PJS, Dring MJ (1980) A quantitative evaluation of the     Ser 188:81–92
  effects of Ascophyllum harvesting on the littoral ecosys-   Jenkins SR, Norton TA, Hawkins SJ (2004) Long term effects
  tem. Helgol Meeresunters 33:700–710                of Ascophyllum nodosum canopy removal on mid shore
Bowman RS, Lewis JR (1986) Geographical variation in the       community structure. J Mar Biol Assoc UK 84:327–330
  breeding cycles and recruitment of Patella spp. Hydro-    Jones NS (1946) Browsing of Patella. Nature 158:557–558
  biologia 142:41–56                      Jones NS (1948) Observations and experiments on the bio-
Brown R (1990) Strangford Lough, the wildlife of an Irish Sea    logy of Patella vulgata at Port St. Mary, Isle of Man. Proc
  Lough. Institute of Irish Studies, QUB, Belfast          Trans Liverp Biol Soc 56:60–77
Burrows MT, Moore JJ, James B (2002) Spatial synchrony of    Jones PD, Lister D (2004) The development of monthly tem-
  population changes in rocky shore communities in Shet-      perature series for Scotland and Northern Ireland. Int J
  land. Mar Ecol Prog Ser 240:39–48                 Climatol 24:569–590
Bustamante RH, Branch GM, Eekhout S (1995) Maintenance      Jonsson PR, Granhag L, Moschella PS, Åberg P, Hawkins SJ,
  of an exceptional intertidal grazer biomass in South       Thompson RC (2006) Interactions between wave action
  Africa: subsidy by subtidal kelps. Ecology 76:2314–2329      and grazing control the distribution of intertidal macro-
Caloz R, Collet C (1997) Geographic information systems       algae. Ecology 87:1169–1178
  (GIS) and remote sensing in aquatic botany: methodo-     Kendall MA, Burrows MT, Southward AJ, Hawkins SJ (2004)
  logical aspects. Aquat Bot 58:209–228               Predicting the effects of marine climate change on the
Coleman RA, Goss-Custard JD, Durell S, Hawkins SJ (1999)       invertebrate prey of the birds of rocky shores. Ibis 146:
  Limpet Patella spp. consumption by oystercatchers Hae-      40–47
  matopus ostralegus: a preference for solitary prey items.   Keser M, Larson BR (1984) Colonization and growth of Asco-
  Mar Ecol Prog Ser 183:253–261                   phyllum nodosum (Phaeophyta) in Maine. J Phycol 20:
Cousens R (1984) Estimation of annual production by the       83–87
  intertidal brown algae Ascophyllum nodosum. Bot Mar 27:    Keser M, Swenarton JT, Foertch JF (2005) Effects of thermal
  217–227                              input and climate change on growth of Ascophyllum
Crisp DJ (1964) The effects of the severe winter of 1962–63 on    nodosum (Fucales, Phaeophyceae) in Eastern Long Island
  marine life in Britain. J Anim Ecol 33:165–210          Sound (USA). J Sea Res 54:211–220
Dudgeon S, Petraitis PS (2001) Scale-dependent recruitment    Klaic ZB, Nitis T, Kos I, Moussiopoulos N (2002) Modification
  and divergence of intertidal communities. Ecology 82:       of the local winds due to hypothetical urbanization of the
  991–1006                             Zagreb surroundings. Meteorol Atmos Phys 79:1–12
Dudgeon S, Petraitis PS (2005) First year demography of the   Le Roux A (2005) Les patelles et la régression des algues
  foundation species, Ascophyllum nodosum, and its com-       brunes dans le Morbihan. Penn Bed 192:1–22
  munity implications. Oikos 109:405–415            Lewis JR (1964) The ecology of rocky shores. English Univer-
Dye AH (1998) Community-level analyses of long-term         sities Press, London
                   Davies et al.: Limpet grazing on Ascophyllum nodosum                   141




Little C, Kitching JA (1996) The biology of rocky shores.     Southward AJ (1964) Limpet grazing and the control of
  Oxford University Press, New York                vegetation on rocky shores. In: Crisp DJ (ed) Grazing in
Lüning K (1990) Seaweeds: their environment, biogeography,      terrestrial and marine environments. Blackwell Scientific
  and ecophysiology. Wiley-Interscience, New York         Publications, Oxford, p 165–273
Maclean IMD, Austin GE, Mellan HJ, Girling T (2005) WeBS     Southward AJ, Southward EC (1978) Recolonisation of rocky
  alerts 2003/2004: changes in numbers of wintering        shores in Cornwall after use of toxic dispersants to clear up
  waterbirds in the United Kingdom, its constituent coun-     the Torrey Canyon spill. J Fish Res Board Can 35:682–706
  tries, special protection areas (SPAs) and sites of special  Strong J (2003) The environmental impact of Sargassum
  scientific interest (SSSIs). BTO Res Rep. 416 to the WeBS    muticum on the coastline of Northern Ireland. PhD thesis,
  partnership. British Trust for Ornithology (BTO), Thet-     Queen’s University, Belfast
  ford                             Thieltges DW, Strasser M, van Beusekom JEE, Reise K (2004)
McEachreon JCT, Thomas MLH (1987) Attachment strength        Too cold to prosper — winter mortality prevents popula-
  of Ascophyllum nodosum (L.) Le Jolis and exposure to       tion increase of the introduced American slipper limpet
  wave action. Bot Mar 30:217–222                 Crepidula fornicata in northern Europe. J Exp Mar Biol
Nilsson J, Engkvist R, Persson LE (2004) Long-term decline      Ecol 311:375–391
  and recent recovery of Fucus populations along the      Thomas MLH (1986) A physically derived exposure index for
  rocky shores of southeast Sweden, Baltic Sea. Aquat Ecol     marine shorelines. Ophelia 25:1–13
  38:587–598                          Thompson RC, Crowe TP, Hawkins SJ (2002) Rocky intertidal
Norton TA, Hawkins SJ, Manley NL, Williams GA, Watson        communities: past environmental changes, present status
  DC (1990) Scraping a living — a review of littorinid graz-    and predictions for the next 25 years. Environ Conserv 29:
  ing. Hydrobiologia 193:117–138                  168–191
Petraitis PS, Dudgeon SR (1999) Experimental evidence for     Thompson RC, Norton TA, Hawkins SJ (2004) Physical stress
  the origin of alternative communities on rocky intertidal    and biological control regulate the producer–consumer
  shores. Oikos 84:239–245                     balance in intertidal biofilms. Ecology 85:1372–1382
Roberts D, Davies C, Mitchell A, Moore H and 7 others (2004)   Underwood AJ (2000) Experimental ecology of rocky inter-
  Strangford Lough ecological change investigation (SLECI).    tidal habitats: What are we learning? J Exp Mar Biol Ecol
  Environment and Heritage Service, Belfast            250:51–76
Schiel DR, Steinbeck JR, Foster MS (2004) Ten years of      Vogt H, Schramm W (1991) Conspicuous decline of Fucus
  induced ocean warming causes comprehensive changes        vesiculosus in Kiel Bay (Western Baltic): What are the
  in marine benthic communities. Ecology 85:1833–1839       causes? Mar Ecol Prog Ser 69:189–194
Service M, Durrant AE, Mills JA, Taylor JE, Faughey D (1996)   Wilkinson M, Fuller IA, Telfer TC, Moore CG, Kingston PF
  The trophic status of two Northern Irish Sea Loughs.       (1988) Northern Ireland littoral survey. Environment and
  J Coast Conserv 2:159–168                    Heritage Service, Belfast

Editorial responsibility: Roger Hughes (Contributing Editor),   Submitted: July 13, 2006; Accepted: October 23, 2006
Bangor, UK                            Proofs received from author(s): May 21, 2007