MARINE ECOLOGY PROGRESS SERIES
  Vol. 298: 143–156, 2005                                           Published August 15
                            Mar Ecol Prog Ser




  Density-dependent sea urchin grazing: differential
    removal of species, changes in community
   composition and alternative community states
  Jeffrey T. Wright1, 3,*, Symon A. Dworjanyn1, 4, Cary N. Rogers1, Peter D. Steinberg 1, 2,
              Jane E. Williamson1, 5, Alistair G. B. Poore1
    1
    School of Biological, Earth and Environmental Sciences, and 2Centre for Marine Biofouling and Bio-Innovation,
               The University of New South Wales, New South Wales 2052, Australia
  3
   Present address: Institute of Conservation Biology and Department of Biological Sciences, University of Wollongong,
                        New South Wales 2522, Australia
4
Present address: NSW Department of Primary Industries, Port Stephens Fisheries Centre, New South Wales 2315, Australia
      5
      Present address: Department of Biological Sciences, Macquarie University, New South Wales 2109, Australia




      ABSTRACT: The grazing activity of consumers causes shifts between alternative states in a variety of
      terrestrial and marine ecosystems. One of the best examples of a consumer-driven shift occurs on
      temperate marine reefs, where grazing by high densities of sea urchins results in a shift from a foliose
      algal- to a crustose algal-dominated state. In this study, we focussed on 2 largely untested but impor-
      tant issues during the transition from a foliose algal- to a crustose algal-dominated state: (1) whether
      sea urchins impact foliose algal community composition by differentially removing species and (2)
      whether any impacts of grazing vary with 2 different densities of aggregating sea urchins. We
      tracked the movement of a high-density front of the sea urchin Heliocidaris erythrogramma and then
      performed experimental manipulations of H. erythrogramma at 2 unusually high but naturally occur-
      ring densities. Non-metric multidimensional scaling (nMDS) followed by analysis of similarities
      (ANOSIM) showed differences in the foliose algal community composition, and therefore differential
      removal of species, between permanent plots before and during grazing (surveys), and between
      grazed and ungrazed plots (experiment). Of the 6 abundant foliose algae, 2 had relatively low sur-
      vivorship (Amphiroa anceps and Zonaria diesingiana), while 2 had relatively high survivorship
      (Delisea pulchra and Corallina officinalis) when grazed by high densities of sea urchins. Grazing by
      different densities of H. erythrogramma resulted in differences in the foliose algal community
      composition and for the chemically-defended D. pulchra there appeared to be a threshold sea urchin
      density required before its removal. Our results show that an intermediate community state
      composed of grazer-resistant foliose algae and crustose algae can occur, which may have important
      consequences for community composition.

      KEY WORDS: Sea urchins · Differential grazing · Community composition · Alternative community
      states · Density-dependent effects · Heliocidaris erythrogramma · Herbivory · Australia
                Resale or republication not permitted without written consent of the publisher



            INTRODUCTION                    variation in consumer activity causes switches
                                     between states in both terrestrial, aquatic and marine
 Many ecosystems can exist in alternative community           systems (Dublin et al. 1990, Van de Koppel et al. 1997,
states and a variety of factors trigger switches between         Duffy & Hay 2001, Bertness et al. 2004, Schrage &
states in different systems (May 1977, Scheffer et al.          Downing 2004). On tropical marine reefs, a reduction
2001). For example, in combination with other factors,          in the abundance, or the total removal of herbivorous


*Email: jeffw@uow.edu.au                         © Inter-Research 2005 · www.int-res.com
144                    Mar Ecol Prog Ser 298: 143–156, 2005




fishes and sea urchins, results in a switch from a coral-   algal-dominated community (although see Andrew
dominated to a foliose algal-dominated state (Hughes     1993, Konar & Estes 2003, Hill et al. 2003). Second,
1994). Similarly, in temperate marine systems, the      studies examining the effects of sea urchin grazing
occurrence of high-density aggregations of grazing      often pool algal species into functional groups (e.g.
sea urchins results in a switch from a foliose algal-     turfing, filamentous, foliose, articulated corallines;
dominated to a crustose algal-dominated state (Dean      Ebeling et al. 1985, Benedetti-Cecchi et al. 1998, Konar
et al. 1984, Ebeling et al. 1985, Fletcher 1987, Andrew    & Estes 2003) rather than examining the response of
1993, Leinaas & Christie 1996, Benedetti-Cecchi et al.    individual species. In addition, to our knowledge, only
1998, Shears & Babcock 2002, Konar & Estes 2003,       1 study has examined how variation in the density of
Gagnon et al. 2004). Thus, the intense level of her-     sea urchins (i.e. variable ‘high’ densities) affects the
bivory that occurs in many marine systems (Hay &       transition to a crustose algal-dominated state (Hill et
Steinberg 1992, Duffy & Hay 2001) has direct implica-     al. 2003). For these reasons, it is unclear whether
tions for the community state, and marine communities     aggregations of temperate sea urchins remove algae
often exist as a mosaic of alternative community states    selectively or whether different sea urchin densities
that are a direct result of spatial and/or temporal vari-   matter during the transition to the crustose algal-
ation in grazing intensity (e.g. Ebeling et al. 1985,     dominated state.
Konar & Estes 2003).                       Heliocidaris erythrogramma (Valenciennes, 1846) is
  The creation of these alternative community states     a common subtidal sea urchin in temperate southeast-
by marine herbivores also directly influences algal      ern Australia (Keesing 2001). It is usually found at den-
community composition. Crustose coralline algae        sities <10 m–2 (Wright et al. 2000) but has been
(hereafter crustose algae) are more abundant where      described in aggregations exceeding 100 urchins m–2
the intensity of grazing is high (Duffy & Hay 2001).     (Keesing 2001, Wright & Steinberg 2001). In 1995, a
Additionally, some tropical species of foliose, fleshy    high-density aggregation of H. erythrogramma was
and turfing algae (hereafter foliose algae) suffer rela-   observed moving in a front through an algal commu-
tively low loss to grazing fishes and are more abundant    nity at a site near Sydney, Australia (Wright et al.
in habitats with high grazing intensity (Hay et al. 1983,   2000). This aggregation persisted at this site for > 3 yr
Hay 1984, Lewis 1986, Thacker et al. 2001). Generally,    (J. T. Wright pers. obs.) although the front only took 3
species of foliose algae that survive in habitats with    to 4 mo to move through fixed quadrats (Wright &
intense grazing pressure in the tropics have spatial or    Steinberg 2001).
temporal refuges, or have chemical and/or structural       In this study, we examined the impacts of these high-
defences that deter grazing (see reviews by Duffy &      densities of H. erythrogramma on the algal community
Hay 2001, Paul et al. 2001). Unpalatable and/or chem-     and asked 2 main questions: (1) Is there differential
ically defended tropical cyanobacteria (Thacker et al.    removal of foliose algal species during grazing? (2) Do
2001) and sponges (Pawlik et al. 1995) are also more     the impacts of grazing (i.e. differential removal of
abundant in habitats with high grazing intensity.       foliose algal species and changes in community
  In contrast to evidence from the tropics, the extent to  composition) vary between 2 different densities of
which different species of temperate foliose algae per-    aggregating sea urchins? If the assumption of non-
sist under intense sea urchin grazing pressure is not     differential removal holds, all species of foliose algae
well understood. Although many temperate foliose       will be removed at the same rate during grazing and
algae possess chemical defences and sea urchins often     no discernable difference in foliose algal composition
have feeding preferences that can be explained by       will occur during the transition from a foliose algal- to
those compounds (Hay et al. 1987, Wright et al. 2004),    a crustose algal-dominated state. Similarly, if the im-
it is generally assumed that because grazing by aggre-    pacts of H. erythrogramma are density-independent,
gations of sea urchins is so intense, they remove foliose   then the hierarchy of species removal and foliose algal
algae non-selectively during the switch from a foliose    community composition should not differ with urchin
algal- to crustose algal-dominated state. There are      density. To examine these questions, we first estab-
probably 2 reasons for this assumption. First, the focus   lished permanent transects perpendicular to the front
of most studies of urchin-dominated communities has      of H. erythrogramma, followed their movement, and
been on following the recovery of the foliose algal      measured changes in algal community composition
community after removal of sea urchins from barrens      during the switch from a foliose algal- to a crustose
habitats (e.g. Fletcher 1987, Andrew & Underwood       algal-dominated state. These permanent transects
1993, Leinaas & Christie 1996, Benedetti-Cecchi et al.    allowed us to determine whether there were differ-
1998, Shears & Babcock 2002); relatively few studies     ences in the foliose algal community composition
have added sea urchins to foliose algal beds and fol-     before, during, and after grazing by high sea urchin
lowed the transition from a foliose algal- to a crustose   densities. Following these surveys, we transplanted
                 Wright et al.: Community effects of sea urchin grazing               145




H. erythrogramma at 2 ‘high’ densities into a foliose     pare changes in algal composition as a function of sea
algal-dominated community to determine their effect      urchin grazing, we defined all quadrats at each sam-
on the rate and hierarchy of foliose algal removal, and    pling time into 1 of 3 categories: ‘before’, ‘during’ or
foliose algal community composition during the transi-     ‘after’ the sea urchin front had passed using a quanti-
tion to a crustose algal-dominated state.           tative definition based on Heliocidaris erythrogramma
                                density. At Bare Island, densities of H. erythrogramma
                                ranged from 10 to 20 m–2 before the front to >100 m–2 at
      MATERIALS AND METHODS               the front before declining to < 40 m–2 after the front had
                                passed (Wright & Steinberg 2001). We defined the
  Study site. This study was undertaken on a sub-       ‘before’ category as < 40 urchins m–2; the ‘during’ cate-
tidal reef approximately 3 to 4 m below MLLW at        gory as > 40 urchins m–2; and the ‘after’ category as < 40
Bare Island, near Sydney, Australia (33° 59’ 32’’ S,      urchins m–2. Because we were following urchin move-
151° 13’ 50’ E). The macroalgal community at Bare       ment, with the exception of the initial sampling time
Island is diverse with the most abundant species being     (December 19, 1996), we could easily separate qua-
the foliose red alga Delisea pulchra, various brown      drats with < 40 urchins m–2 into ‘before’ and ‘after’ cat-
algae (Ecklonia radiata, Sargassum vestitum, Sargas-      egories. For December 19, 1996, we designated
sum linearifolium, Zonaria diesingiana and Dilophus      quadrats with < 40 urchins m–2 as ‘before’ and ‘after’
marginatus), coralline turfing red algae (Corallina      based on their position in the transect relative to the
officinalis and Amphiroa anceps) and crustose red       direction of urchin movement.
algae (Sporolithon durum, Lithamnion spp.). Several        We examined how the algal community composi-
sea urchins and large gastropods occur at Bare Island     tion changed as the front of Heliocidaris erythro-
(Andrew 1991, Wright et al. 2000, Williamson et al.      gramma moved through it using non-metric multidi-
2004) but Heliocidaris erythrogramma is the most        mensional scaling (nMDS). Similarity matrices were
abundant and grazing by high-density aggregations of      constructed using the Bray-Curtis coefficient and
this urchin (> 80 urchins m–2) resulted in lower recruit-   square root-transformed data (Clarke & Warwick
ment and higher mortality of D. pulchra compared to      1994). Algal composition was contrasted between the
sites without H. erythrogramma (Wright & Steinberg       ‘before’, ‘during’ and ‘after’ categories by analysis of
2001).                             similarities (ANOSIM; Clarke & Warwick 1994).
  Sea urchin movement and changes in the algal com-      Analyses were run twice: first, including all foliose
munity. In December 1996, 5 permanent transects each      algal species and crustose algae and second, with
15 m in length were established perpendicular to the      crustose algae excluded. The latter contrast allowed
high-density front of Heliocidaris erythrogramma, and     an examination of changes in foliose algal species
marked at each end with eyebolts. Ten 0.5 m × 0.5 m      composition only during the formation of a crustose
quadrats were placed at 1.5 m intervals along each       algal-dominated barren instead of simply document-
transect. Each transect started with 3 to 4 quadrats      ing that a barren had been formed. In both analyses,
within the foliose algal habitat (where high densities of   samples in which there were no algae present were
sea urchins had yet to invade) and passed through the     excluded. Samples from the ‘after’ category were
sea urchin front into the crustose algal-dominated       excluded from the second analysis because only 4
habitat. These permanent quadrats were sampled on       quadrats in this category contained algae other than
December 19, 1996, February 21, 1997, April 24, 1997,     the crustose corallines.
June 24, 1997 and September 2, 1997. In each quadrat,       Experimental manipulation of sea urchin densities.
the number of H. erythrogramma was counted and the       We manipulated sea urchin density in cages at Bare
cover of each foliose algal species, crustose algae (all    Island on an area of reef approximately 400 m2 in size.
species pooled) and bare substratum was measured        This reef had not been invaded by high densities of
using a 100-point grid where each point of the grid that    Heliocidaris erythrogramma, and contained the typical
fell over algae or bare rock was recorded.           foliose algal community for this site. The experiment
  Although initially all transects were established so    had 5 treatments: (1) high-urchin density (100 m–2), (2)
that they contained a similar number of quadrats posi-     low urchin density (40 m–2), (3) zero urchin treatment,
tioned before, during and after the sea urchin front had    (4) cage control and (5) open plot. Cages were plastic-
passed, the front did not move perpendicular to the      coated steel mesh baskets 52 cm long by 42 cm wide by
transects, but instead crossed at an angle (~45°) to      18 cm high (mesh size = 3 cm) bolted to the substratum
them. By the end of the surveys, the sea urchin front     in an inverted position. Cages for the urchin density
was diffuse and scattered. Consequently, quadrats in      treatments and the zero urchin treatment had mesh
the same spatial position on different transects did not    tops; the cage controls had the mesh tops removed,
always have the same grazing history. In order to com-     while the open plots were areas the same size as the
146                   Mar Ecol Prog Ser 298: 143–156, 2005




cage marked with eyebolts at each corner. Cages were     abundant foliose algae in both urchin density treat-
attached to the substratum using strips of rubber      ments. To make these comparisons, we first calculated
looped around the corners of each cage and bolted to     the mean percentage cover of those 6 species in the
the substratum with dynabolts. Attaching the cages to    high urchin density, low urchin density and zero urchin
the substratum with rubber strips allowed a small      treatments for each time as a function of the mean per-
amount of movement and reduced the effects of drag      centage cover of that species at time zero. Then, using
by waves, but did not allow enough movement for       these standardised (for initial cover) values, we calcu-
urchins to escape. There were 8 replicates per treat-    lated the mean percentage cover of the 6 species in the
ment.                            high and low urchin density treatments, each time as a
 The 2 experimental densities of Heliocidaris erythro-   function of the mean percentage cover of those same
gramma were higher than the densities of adjacent      species in the zero urchin treatment. We then calcu-
uninvaded areas at Bare Island and both can persist for   lated the slope of the relationship between time and
long periods of time (Wright & Steinberg 2001, also see   percentage cover (relative to zero urchin treatment) for
results). Hence, our experimental densities reflect     each species in the 2 sea urchin treatments. Differences
invasion into the foliose algal community by 2 un-      among slopes of the 6 species within each sea urchin
usually high, albeit naturally-occurring, densities of    density category and, between the 2 sea urchin density
H. erythrogramma (Wright et al. 2000, Wright & Stein-    treatments within species, were examined separately
berg 2001). To stock the 2 sea urchin density treat-     using analysis of covariance (ANCOVA; Quinn &
ments, H. erythrogramma (test diameter 50 to 70 mm)     Keough 2002). The significance of these ANCOVAs
were collected by carefully removing them from the      was tested after correcting for multiple comparisons.
substratum to avoid damage, and immediately placed       Temporal change in the composition of the algal
into cages at appropriate densities.             community within each treatment was examined using
 The experiment began on February 27, 1998 and       nMDS. For each sampling time, the algal composition
lasted for 208 d. At time zero, we counted the cover of   was contrasted across treatments using ANOSIM in 2
each foliose algal species, crustose algae and bare sub-   different ways. First, using all foliose algal species and
stratum in each replicate using a 120-point grid on the   crustose algae and then, with crustose algae excluded.
mesh top of each cage. The 120-point grid excluded      The significance level of the pairwise comparisons
the points around the cage perimeter. For the 2 treat-    within ANOSIM was corrected for multiple compar-
ments without mesh tops (cage controls and open       isons.
plots), identical cages were temporarily placed over      Statistical analyses. Univariate analyses were car-
plots so cover could be measured. We determined the     ried out using SYSTAT (Version 10, SPSS). The
total cover of algae and bare substratum in each cage    assumptions of normality and heterogeneity of vari-
approximately every 2 wk for 2 mo, and then approxi-     ance were checked using frequency histograms of
mately every 4 to 6 wk after that. During sampling, we    residuals and plots of residuals versus means, respec-
also counted the number of sea urchins in each cage     tively (Quinn & Keough 2002). Logarithmic transfor-
and added more when required. Points that fell over     mations were made where appropriate. Multivariate
sea urchins were not counted but this was uncommon      scaling and ANOSIM were conducted using Primer
as most sea urchins were around the perimeter of the     (Version 5.2.2, Primer-E). The significance level was
cages during the day.                    taken as p < 0.05 except where stated.
 We used both univariate and multivariate analyses to
examine the effects of the 2 high densities of Helioci-
daris erythrogramma on algal community structure.                  RESULTS
Analysis of variance (ANOVA) was used to determine
whether the cover of the 6 most abundant foliose algal      Sea urchin movement and changes in the algal
species (which comprised approximately 95% of total               community
foliose algal cover), crustose algae and bare substratum
differed among treatments at 4 sampling times: 0, 33,     The front of Heliocidaris erythrogramma moved
92 and 159 d. Because we sampled the same quadrats      across the substratum in densities greater than 100 m–2
at 4 different times, we corrected for multiple compar-   (maximum density: 192 m–2). Initially (December
isons and tested at a significance level of 0.0125 in    1996), there was a transition along transects from low
these analyses. Following ANOVA, we made compar-       (<10 m–2) to high (> 80 m–2) average densities. How-
isons between treatments using Tukey’s tests. To gain    ever, this transition changed as the urchin front
further insight into the survival of foliose algae under   became diffuse. This scattering of urchins was empha-
intense sea urchin grazing, we determined whether      sised by similar average densities of sea urchins in
there was a difference in the decline of the 6 most     quadrats across transects in September 1997 (Fig. 1).
                           Wright et al.: Community effects of sea urchin grazing                 147




               Heliocidaris           Delisea        Corallina         Amphiroa     Sargassum
              erythrogramma            pulchra        officinalis        anceps      vestitum
          200                                       50          30
                           80           60
          160                                       40
December




                           60                                 20
          120                           40           30
 1996




           80                40                       20
                                       20                      10
           40                20                       10
           0                0           0            0          0
          200                                       50          30
                           80           60
          160                                       40
February




          120                60                       30          20
                                       40
 1997




           80                40                       20
                                       20                      10
           40                20                       10
           0                0           0            0          0
          200                                       50          30
                           80           60
          160                                       40
      –2




                           60                                 20
                      % cover
1997
April




          120                                       30
      no. m




                                       40
           80                40                       20
                                       20                      10
           40                20                       10
           0                0           0            0          0
          200                                       50          30
                           80           60
          160                                       40
                           60                       30          20
1997
June




          120                           40
           80                40                       20
                                       20                      10
           40                20                       10
           0                0           0            0          0
          200                                       50          30
                           80           60
          160                                       40
September




                           60                                 20
 1997




          120                           40           30
           80                40                       20
                                       20                      10
           40                20                       10
           0                0           0            0          0
             0 2 4 6 8 10           0 2 4 6 8 10      0 2 4 6 8 10      0 2 4 6 8 10     0 2 4 6 8 10
                                           Quadrat
Fig. 1. Mean (±1 SE) densities of Heliocidaris erythrogramma and the percentage cover of foliose algal species, total crustose
algae and bare substratum in the permanent quadrats at 5 times. The number of quadrats in each category based on grazing
history (‘before’, ‘during’ and ‘after’ grazing) each time was: December 1996 (21, 28, 2); February 1997 (15, 37, 1); April 1997 (14,
          35, 3); June 1997 (13, 32, 7); September 1997 (6, 35, 10). Note the variable scale on the y-axes


Additionally, in quadrats 1 to 4, average H. erythro-                SE H. erythrogramma densities between 60.0 ± 7.0 and
gramma densities generally remained between 20 and                 87.2 ± 3.9 m–2 throughout our sampling (Fig. 1). All
40 m–2 from February to September 1997, considerably                other foliose algae were in very low abundance, or
lower than the very high densities (> 80 m–2) that occur              were completely absent at these sea urchin densities.
at the urchin front.                                The cover of crustose algae and bare substratum
  Each sampling time, quadrats with high densities of               increased with sea urchin density and both covered
Heliocidaris erythrogramma had low cover of all algae                approximately 50% of the substratum at high densities
except crustose algae (Fig. 1). Generally, with low den-              of sea urchins.
sities of urchins, there was a relatively high cover of                When we separated the abundance of each algal
Delisea pulchra, Corallina officinalis and Amphiroa                 species into the 3 sea urchin grazing categories, the
anceps. Other foliose algal species including Sargas-                role of high densities of sea urchins in removing all
sum vestitum, Sargassum linearifolium, Ecklonia radi-                foliose algae was emphasised. With the exception of
ata and Zonaria diesingiana were also present with                 the rare Sargassum vestitum, all non-crustose species
low densities of urchins but in lower abundance.                  had a much lower cover in the ‘during’ and ‘after’ cat-
D. pulchra, and to a lesser extent C. officinalis, main-              egories compared to the ‘before’ category (Fig. 2). In
tained low cover in quadrats 6 to 8, which had mean ±                contrast, crustose algae and bare substratum had high
148                        Mar Ecol Prog Ser 298: 143–156, 2005




           Sargassum       Zonaria       Ecklonia        crustose
           linearifolium     diesingiana       radiata         algae         bare
        8          8           5          80
        7          7                                80
        6          6           4          60
        5          5           3                     60
        4          4                     40
        3          3           2                     40
        2          2           1          20
        1          1                                20
        0          0           0           0           0
        8          8           5          80
        7          7                                80
        6          6           4          60
        5          5           3                     60
        4          4                     40
        3          3           2                     40
        2          2           1          20
        1          1                                20
        0          0           0           0           0
        8          8           5          80
        7          7                                80
        6          6           4          60
   % cover




        5          5           3                     60
        4          4                     40
        3          3           2                     40
        2          2           1          20
        1          1                                20
        0          0           0           0           0
        8          8           5          80
        7          7                                80
        6          6           4          60
        5          5           3                     60
        4          4                     40
        3          3           2                     40
        2          2           1          20
        1          1                                20
        0          0           0           0
                                                    0
        8          8           5          80
        7          7                                80
        6          6           4          60
        5          5           3                     60
        4          4                     40
        3          3           2                     40
        2          2           1          20
        1          1                                20
        0          0           0           0           0
          0 2 4 6 8 10     0 2 4 6 8 10     0 2 4 6 8 10     0 2 4 6 8 10     0 2 4 6 8 10
                                  Quadrat
                              Fig. 1 (continued)

average cover in the ‘during’ and ‘after’ categories.          Experimental manipulation of sea urchin densities
The composition of the algal community varied
strongly among samples in the ‘before’, ‘during’ and           With the exception of an initial period (up to Day 14),
‘after’ categories (Fig. 3A; ANOSIM contrasting            when some Heliocidaris erythrogramma escaped from
‘before’, ‘during’ and ‘after’: R = 0.514, p < 0.001; all       cages, and August when there was a mass mortality of
pairwise comparisons between ‘before’, ‘during’ and          sea urchins after an influx of fresh water at our site fol-
‘after’ were significant: p < 0.001). The changes in the        lowing heavy rain, the cages maintained the 2 densi-
composition of the algal community during the forma-          ties of urchins in the treatments (Fig. 4). In addition,
tion of the urchin barrens were not, however, simply          only 3 urchins in total were ever observed in control
due to the declining cover of foliose algae and increas-        quadrats during the experiment. Following the mass
ing cover of crustose algae. The composition of foliose        mortality (> 88%) of urchins at the end of August, there
algae when considered on their own contrasted             was a period of 3 wk when we were unable to gain
strongly between samples ‘before’ and ‘during’ graz-          access to our site and consequently, cages were not re-
ing by high urchin densities (Fig. 3B, ANOSIM, R =           stocked until September 18 (Day 204). However, these
0.22, p < 0.001). Over the 5 sampling times, quadrats         urchins used to re-stock cages appeared unhealthy
were in the ‘before’ category for 1.36 ± 0.52 of the time,       and had poor survivorship, hence the experiment was
the ‘during’ category for 3.2 ± 0.24 of the time and the        stopped on September 22 (Day 208). Because of the
‘after’ category for 0.44 ± 0.12 of the time (n = 50          potential artefacts after the mass mortality, we have
quadrats; mean ± SE).                         only analysed data up to Day 159, the sampling time
                         Wright et al.: Community effects of sea urchin grazing                  149




     100  Heliocidaris erythrogramma                          cover of Amphiroa anceps and Zonaria
     80                                         diesingiana in the high-density treat-
no. m–2




     60                                         ment (Fig. 5). By Day 92, there was also a
     40                                         significantly lower cover of Corallina
     20                                         officinalis in the high-density treatment.
     0                                          Moreover, the mean ± SE total cover of
                                               foliose algae in the high-density treat-
     30    Delisea pulchra   Corallina officinalis   30 Amphiroa anceps
     25
                                               ment declined from a cover of 64.59 ±
                                 25
% cover




     20                                         3.71 (initially) to 19.13 ± 2.53% (Day 92)
                                 20
     15                           15             and, with the exception of Delisea pul-
     10                           10             chra, the cover of individual species was
      5                            5             < 3.5%. By Day 130, the total cover of
      0                            0             foliose algae in the high-density treat-
      5 Sargassum vestitum     Sargassum linearifolium 40 crustose algae
                                               ment was 5.88 ± 1.41% and all foliose
                                               species remained in very low cover in
% cover




      4                           30
      3
                                               that treatment until the end of the exper-
                                 20             iment. Despite such low cover, there
      2
                                 10             were significant differences between the
      1
                                               high-density and zero urchin treatment
      0                            0
                                               at Day 159 for A. anceps and Sargassum
      5 Ecklonia radiata      Zonaria diesingiana   70  bare          vestitum only. In the low-density treat-
                                 60              ment, there was a significantly lower
      4
% cover




                                 50
      3                           40              cover of A. anceps (Days 33, 92 and 159)
      2                           30              and Z. diesingiana (Day 33, although for
                                 20              Day 92 p = 0.06) compared to the zero
      1                           10
      0                            0             urchin treatment (Fig. 5). There were no
         B    D    A    B     D    A     B   D   A    differences in the cover of crustose
      Fig. 2. Mean (±1 SE) densities of Heliocidaris erythrogramma and the per-
                                               coralline algae between the zero urchin
      centage cover of foliose algal species, total crustose algae and bare substra-   treatment and both urchin density treat-
      tum in quadrats for the 3 sea urchin grazing categories ‘before’ (B; n = 69),   ments throughout the experiment but
             ‘during’ (D; n = 167) and ‘after’ (A; n = 23) grazing        there was a significantly higher cover of
                                               bare substratum in the high-density
      immediately prior to that event, although we have          treatment versus the zero urchin treatment at Days 33
      presented plots of percentage cover and MDS ordina-         and 159, and in the low-density treatment versus the
      tions up to Day 208.                         zero urchin treatment at Day 159. The cover of D. pul-
       Grazing by high densities of Heliocidaris erythro-        chra and Sargassum linearifolium did not differ
      gramma caused a decline in the abundance of all foli-        between the zero urchin treatment and either urchin
      ose algae. By Day 33, there was a significantly lower        density treatment for any dates despite S. linearifolium


        A                      Stress: 0.1    B                       Stress: 0.14




                                           Before
                                           During
                                           After


      Fig. 3. MDS plots of algal community composition in quadrats for the 3 sea urchin grazing categories ‘before’, ‘during’ and ‘after’
           grazing. (A) All algae. (B) Foliose algae only. No ‘after’ data was included in the analysis of foliose algae only
150                                     Mar Ecol Prog Ser 298: 143–156, 2005




      120              high                                being at < 0.13% cover from Day 130. Three other
                      low                                 foliose algae occurred in plots during the experiment;
      100                                                Ecklonia radiata, Dilophus marginatus and Wrangelia
                                                        plumosa. However, both E. radiata and D. marginatus
      80                                                 were always in very low cover (0 to 5%), while W. plu-
No. m–2




                                                        mosa only occurred briefly from Days 0 to 49 as an epi-
      60
                                                        phyte. The cover of these 3 species did not differ
                                                        among treatments at any time. There were 2 occasions
      40
                                                        when caging artefacts were apparent for individual
                                                        species; for A. anceps (Day 92) and S. linearifolium
      20
                                                        (Day 159), there were differences in the cover in the
       0                                                 open plots versus the zero urchin treatment.
         0       30     60     90     120  150   180     210        ANCOVA revealed no difference in regression
                              Day                        slopes among species for either urchin density treat-
                                                        ment (F5,36 = 1.448, p = 0.231 for high; F5,36 = 0.953,
Fig. 4. Heliocidaris erythrogramma. Mean (±1 SE) abundance
in the experimental high and low sea urchin density
                                                        p = 0.459 for low). The relative cover of foliose algal
treatments. The cover of algae was determined on the                              species at Day 159 was greater in the low-density
         same days except for Day 191                                 treatment compared to the high-density treatment


                      Delisea pulchra                      Corallina officinalis
      40    ns ns         ns        ns        20    ns ns   p<0.003         ns
                                                                         ZU
                                                     **
      30                                 15                                OP
 % Cover




                                                                         HU
      20                                 10
                                                                         LU
      10                                  5                                CC

       0                                  0
                     Amphiroa anceps                      Sargassum vestitum
      40    ns p<0.001      p<0.001     p<0.001       40    ns ns         ns     p<0.001
              **,*       **,*       **,*                              **
      30                                 30
 % Cover




      20                                 20

      10                                 10

       0                                  0
                   Sargassum linearifolium                      Zonaria diesingiana
                                           p<0.009
      40    ns  ns       p<0.001     p<0.001       25       p<0.001     p<0.001   p<0.001

                                        20     **,*        **
      30
 % Cover




                                        15                           Fig. 5. Mean (±1 SE) percentage
      20                                                             cover of all foliose algal species, total
                                        10                           crustose algae and bare substratum
      10                                                             in the 5 treatments from the sea
                                         5
                                                                    urchin density experiment. P values
       0                                  0                           above Days 0, 33, 92 and 159 are
                                                                    from ANOVAs done on those times
                     Crustose algae                             Bare
      35    ns p = 0.004      ns        ns        100   p = 0.017        ns     p<0.001   with F 4, 35 for all analyses. ns: no sig-
      30                                 80    * p<0.001             **,*    nificant difference among treat-
                                                **                   ments; ##: significant differences be-
      25
 % Cover




                                        60
                                                                    tween zero and high sea urchin
      20                                                             density treatments; #: significant dif-
      15                                 40                           ferences between zero and low sea
      10                                                             urchin density treatments (p < 0.05;
                                        20                           Tukey’s tests). HU: high sea urchin
       5
       0                                  0                           density; LU: low sea urchin density;
         0    30    60   90   120   150  180 210      0   30    60    90  120 150 180 210  ZU: zero urchin treatment; CC: cage
                         Day                                Day             control; OP: open plot
                        Wright et al.: Community effects of sea urchin grazing                 151




          3                                     0.656, p = 0.157; high-density treatment: r =
            Delisea pulchra        Corallina officinalis         0.701, p = 0.121).
                                                Multidimensional scaling done with all
Relative % cover




       2                                       species demonstrated a rapid change in the
                                               algal community composition with urchin
                                               grazing (Fig. 7A). There were no differences
                                               in community composition among high-
       1                                       density, low-density and zero urchin treat-
                                               ments at 0 and 14 d, but by Day 33, both sea
                                               urchin density treatments were significantly
       0                                       different to the zero urchin treatment
       3                                       (Table 1). Moreover, with the exception of
           Amphiroa anceps           Sargassum vestitum         Days 92 and 159 (for the low density treat-
                                               ment), sea urchin density treatments dif-
Relative % cover




                                               fered to zero urchin treatments for all other
       2
                                               times. There was only 1 occasion (Day 33)
                                      low
                                               when both sea urchin density treatments
                                      high       did not differ to cage controls and open plots
       1                                       as well. There were also differences
                                               between high and low density treatments
                                               for Days 75, 130 and 159 (Table 1), indicat-
       0                                       ing density-dependent effects of grazing on
                                               the algal community composition. These
       3                                       effects were not only due to the decline in
          Sargassum linearifolium       Zonaria diesingiana
                                               foliose algae and an increase in the relative
Relative % cover




                                               cover of crustose algae as the area pro-
       2                                       gressed toward a crustose-dominated bar-
                                               ren. Similar changes in composition were
                                               evident when only foliose algae were
                                               included in the analyses (Fig. 7B, Table 2).
       1
                                               Differences in foliose algal community com-
                                               position between the high-density and the
                                               zero urchin treatment occurred by Day 33
       0
                                               and between the low-density and zero
        0    100   200   300  400 0    100   200   300  400
                                               urchin treatment occurred by Day 49. These
                 Day                 Day          differences remained throughout the exper-
      Fig. 6. Decline in the mean percentage cover of the 6 most abundant        iment with the exception of Day 159 when
      foliose algae under low (– – –) and high (       ) sea urchin densities   low-density and zero urchin treatments did
         relative to their percentage cover in the zero urchin treatments
                                               not differ. Differences in the foliose algal
                                               community composition between high and
      (t-test comparing high versus low density treatments: t        low density treatments occurred for Days 75, 92 and
      = 3.273, df = 10, p = 0.008). The hierarchy of species         159 (Table 2). The only differences in foliose algal com-
      decline was similar for both densities with Delisea          munity composition among the 3 control treatments for
      pulchra predicted to reach 0% cover (the x-intercept of        both analyses, indicating caging artefacts, occurred
      the lines of best fit) last at both densities (Fig. 6).        between the zero urchin treatment and the open plots
      D. pulchra was the only species for which there was          for Days 14 and 159 (Table 2).
      a difference in regression slopes between the high and
      low urchin density treatments (F1,12 = 13.833, p =
      0.003). To examine whether initial cover of species was                   DISCUSSION
      important in its relative decline, correlations were per-
      formed at both sea urchin densities between the aver-          Many sea urchin species trigger a switch in temper-
      age initial cover and the estimated time to removal (the        ate subtidal communities from a foliose algal- to a crus-
      x-intercept of the lines of best fit, Fig. 6) of each spe-       tose algal-dominated state (Dean et al. 1984, Ebeling et
      cies. There were positive but non-significant correla-         al. 1985, Fletcher 1987, Andrew 1993, Leinaas &
      tions at both densities (low-density treatment: r =          Christie 1996, Benedetti-Cecchi et al. 1998, Shears &
152                      Mar Ecol Prog Ser 298: 143–156, 2005




                                  Estes 2003). We have also demonstrated that the
  A                                foliose algal community composition is different in
                                  areas grazed by sea urchins compared to ungrazed
                                  areas, indicating the differential removal of foliose
                                  algae by high densities of sea urchins during grazing.
                                  Different densities of sea urchins also resulted in dif-
                                  ferent foliose algal community composition, indicating
                                  that if low sea urchin densities persist, then an inter-
                                  mediate state composed of less palatable foliose algae
                                  (e.g. D. pulchra) and crustose algae can occur.


                                      Differential removal of foliose algae
                         high urchin
                         low urchin     Under intense sea urchin grazing, some foliose algae
                         open plots    were removed before others. Amphiroa anceps and
                         zero urchin   Zonaria diesingiana, in particular, were quickly re-
                         cage control   moved and within 33 d, both occurred in lower abun-
                                  dance with high and low urchin densities compared to
                                  ungrazed areas. This contrasted to Delisea pulchra and
  B
                                  Corallina officinalis, and to a lesser extent both Sargas-
                                  sum species, particularly in the low urchin density
                                  treatment. Laboratory consumption rates on these
                                  foliose algae by Heliocidaris erythrogramma do not
                                  fully explain the differences in persistence observed in
                                  the field. In no-choice experiments, D. pulchra was
                                  consumed at a low rate compared to C. officinalis and
                                  at a similar rate compared to Z. diesingiana and Sar-
                                  gassum vestitum (Wright et al. 2004). These inconsis-
                                  tencies suggest that for a variety of reasons, results
                                  from feeding experiments with sea urchins may not
                                  always reflect patterns of consumption in the field (also
                                  see Hill et al. 2003). For example, C. officinalis may be
                                  more vulnerable to grazing in the laboratory versus the
                                  field if small unattached pieces used in laboratory
                                  feeding experiments are more susceptible to grazing
Fig. 7. MDS ordination of each treatment in the sea urchin
density experiment (stress = 0.18) for (A) all algae and (B)
                                  than attached thalli.
foliose algae only. Lines connect the centroids of all replicate   Differential consumption of algal species by sea
quadrats within each treatment at each sampling time,       urchins is usually linked to differences in defensive
starting at the beginning of the experiment (circles) and     traits among species, particularly differences in chemi-
         ending at Day 208 (arrows)
                                  cal defences (Steinberg 1992, Paul et al. 2001). Delisea
                                  pulchra is defended against feeding by Heliocidaris
Babcock 2002, Konar & Estes 2003). As expected, graz-       erythrogramma and other macrograzers by non-polar
ing by high densities of Heliocidaris erythrogramma        secondary metabolites (halogenated furanones; Wright
caused such a switch in this community. More specifi-       et al. 2004), and this may be an important reason for
cally, this community went from being dominated by a        the resistance of D. pulchra to grazing. We have not ex-
mix of foliose algae before grazing, through an inter-       amined the effects of the chemical defences of other fo-
mediate state incorporating some of these foliose algae      liose algae from this community against H. erythro-
(particularly Delisea pulchra) and crustose algae, and       gramma, but other sea urchins in this region also
finally to a state containing almost entirely crustose       consume D. pulchra at low levels relative to brown al-
algae. The time for the transition from a foliose algal-      gae, and are strongly deterred by halogenated fura-
to a crustose algal-dominated community was rela-         nones, but weakly deterred by brown algal phlorotan-
tively quick; within 130 d, grazing by high densities of      nins (Steinberg & van Altena 1992, Williamson et al.
H. erythrogramma caused a decline in the total cover        2004, Wright et al. 2004, P. D. Steinberg & R. de Nys
of foliose algae from 64.58 to 5.88% (also see Konar &       unpubl. data). The removal of Zonaria diesingiana,
                    Wright et al.: Community effects of sea urchin grazing           153




Table 1. Results of ANOSIM comparing algal community composition (all    temperate regions against sea urchins
species) during the sea urchin density experiment. Treatments: HU: high sea (Wright et al. 1997). Although temper-
urchin density; LU: low sea urchin density; ZU: zero urchin treatment; CC: cage
                                      ate brown algae off North Carolina that
control; OP: open plot. The level of significance was set at p = 0.005. *p < 0.005;
                ns: non-significant             are unpalatable to grazing fish are dom-
                                      inant where these fish are common
                     Day                (Miller & Hay 1996, Duffy & Hay 2000),
         0   14   33    49   75  92  130   159   this is the first description of a commu-
                                      nity-structuring role for differential
 Global R  0.099 0.173 0.263 0.384 0.382 0.303 0.364 0.446        grazing by temperate sea urchins
 Comparison                                linked to algal chemical defences.
  HU vs. LU  ns   ns   ns    ns    *  ns   *     *
                *    *    *  *   *     *
                                        The persistence of Corallina offici-
  HU vs. ZU  ns   ns
  HU vs. CC   *   ns   *    *    *  *   *     *   nalis during urchin grazing raises the
  HU vs. OP  ns   *    *    *    *  *   *     *   possibility that structural defences (cal-
  LU vs. ZU  ns   ns   *    *    *  ns   *    ns   cium carbonate) of foliose algae may
  LU vs. CC  ns   ns   *    *    *  *   *     *
                                      also deter grazing by Heliocidaris ery-
  LU vs. OP  ns   *   ns    *    *  *   *     *
  ZU vs. CC  ns   ns   ns    ns   ns  ns   ns    ns
                                      throgramma. However, the early re-
  ZU vs. OP  ns   ns   ns    ns   ns  ns   ns    ns   moval of Amphiroa anceps by H. ery-
  CC vs. OP  ns   ns   ns    ns   ns  ns   ns    ns   throgramma indicates that any effects
                                      of calcium carbonate are not consistent
Table 2. Results of ANOSIM comparing algal community composition (foliose  across articulated coralline species.
algae only) during the sea urchin density experiment. Abbreviations and   Calcium carbonate deters feeding by
             symbols as for Table 1              some marine herbivores (Pennings &
                                      Paul 1992, Schupp & Paul 1994, Hay et
                     Day                al. 1994), although the sea urchin
         0   14   33    49   75  92  130   159
                                      Diadema antillarum was only partially
 Global R  0.058 0.169 0.230 0.372 0.465 0.342 0.401 0.511        deterred by calcium carbonate, and
 Comparison                                deterrence increased when secondary
  HU vs. LU  ns   ns   ns    ns    *  *   ns    *   metabolites were also present (Hay et
  HU vs. ZU  ns   ns   *    *    *  *   *     *   al. 1994). Ultimately in our system, it
  HU vs. CC  ns   ns   *    *    *  *   *     *   appears that when urchin densities are
  HU vs. OP  ns   *    *    *    *  *   *     *
                    *    *  *   *        very high and persist for long enough,
  LU vs. ZU  ns   ns   ns                  ns
  LU vs. CC  ns   ns   ns    *    *  *   *     *   even chemically- and/or physically-
  LU vs. OP  ns   ns   ns    *    *  *   *     *   defended foliose algae are removed.
  ZU vs. CC  ns   ns   ns    ns   ns  ns   ns    ns     Several other factors unrelated to
  ZU vs. OP  ns   *   ns    ns   ns  ns   ns    *
                                      palatability may have influenced the
  CC vs. OP  ns   ns   ns    ns   ns  ns   ns    ns
                                      different rates of removal of foliose
                                      algae by Heliocidaris erythrogramma.
Sargassum vestitum and Sargassum linearifolium,      First, rate of removal may simply have been a function
which all contain phlorotannins, by H. erythrogramma    of initial abundance. The non-significant, but positive
suggests phlorotannins may be a weak deterrent to     correlations between initial cover and the estimated
H. erythrogramma. In general, non-polar compounds     time to removal provide some support for that idea.
such as furanones or terpenes appear to be more deter-   However, the finding that Zonaria diesingiana and
rent than phlorotannins to sea urchins in Australia    Corallina officinalis declined at different rates, particu-
(Steinberg & van Altena 1992) even though phlorotan-    larly at ‘low’ urchin densities, even though both had
nins are deterrent to urchins in other temperate      low initial cover indicates that other factors are im-
regions (Steinberg 1992).                 portant too. Second, some species may be removed
 The persistence of Delisea pulchra in the low urchin   more quickly if they are more easily located than other
density treatment indicates that non-polar secondary    species, or if they are more vulnerable to a strategi-
metabolites may play an important community role and    cally-placed bite (e.g. species with a small holdfast
allow unpalatable species to persist under sea urchin   area:thallus size ratio may be more susceptible than a
grazing. Important roles have been described for chem-   species with a relatively large holdfast area:thallus size
ical defences in the persistence of foliose algae,     ratio). Third, differences among species in the ability to
cyanobacteria and sponges in tropical regions against   tolerate or recover from grazing because of different
grazing fishes (Pawlik et al. 1995, Duffy & Hay 2001,   life histories may also be important. There appeared to
Paul et al. 2001) and in the persistence of sponges in   be different growth rates among foliose algae during
154                    Mar Ecol Prog Ser 298: 143–156, 2005




our experiment. For example, in the control treat-        Transitions between alternative community states
ments, both Sargassum species increased cover by up
to 2 to 3 times compared to their initial cover. Similarly,    Many temperate subtidal algal communities exist as
the apparent increase in C. officinalis cover between     a mosaic of alternative community states that fluctuate
days 159 and 208 in both urchin density treatments fol-    in space and time (Ebeling et al. 1985, Konar & Estes
lowing the mass mortality of H. erythrogramma, sug-      2003). Both the survey and experimental components
gests that some small thalli of C. officinalis remained    of this study indicated that an intermediate community
after grazing and grew rapidly once grazing pressure      state occurs in addition to the foliose algal- and crus-
was removed. Tropical foliose algae can tolerate        tose algal-dominated states. This intermediate com-
intense grazing by fishes and regrow rapidly when       munity state consisted of a mix of foliose algae, crus-
grazers are removed (Carpenter 1986, Lewis 1986,        tose algae and bare rock, and occurred as a function of
Lewis et al. 1987).                      both time grazed by sea urchins (the ‘during’ category)
                                and sea urchin density (high versus low urchin density
                                treatments). Although, for the most part, in experimen-
        Density-dependent effects            tal cages, this intermediate state persisted from Day 75
                                to Day 159, even foliose algae more resistant to grazing
  There were important density-dependent effects of      are predicted to eventually decline to zero cover (i.e.
grazing by Heliocidaris erythrogramma on this algal      where the line of best fit intercepts the x-axis, Fig. 6),
community. By the end of our experiment, the areas       suggesting that this intermediate state is transitory.
grazed by 100 urchins m–2 consisted almost entirely of     The complete removal of foliose algae observed in
crustose algae and bare substratum whereas the areas      most quadrats in the ‘after’ category supports this
grazed by 40 urchins m–2 contained a mix of foliose      notion of a transitory intermediate state.
algae, crustose algae and bare substratum. These         An intermediate community state may have impor-
effects were reflected in the differences in the commu-    tant consequences following a mass mortality of sea
nity structure between these 2 urchin densities con-      urchins. Trajectories of community succession after
sidering all algae, and considering just foliose algae.    disturbance are influenced by a variety of stochastic
The latter indicates these effects were not simply due     and deterministic factors including the species compo-
to a decline of foliose algae and an increase in crustose   sition of a site or adjoining sites, a range of site-specific
algae. Grazing by different densities of Centro-        factors, and recruitment (Dudgeon & Petraitis 2001,
stephanus rodgersii also resulted in different algal      Sousa 2001). Here we did not follow community devel-
composition (Hill et al. 2003).                opment after the mass mortality of Heliocidaris erythro-
  Because the hierarchy of species removal was simi-     gramma. However, because there were differences in
lar for the 2 experimental sea urchin densities, differ-    the foliose algal composition between grazed and
ences in community composition between densities        ungrazed areas, and between areas grazed for different
on the same dates may be due to similar processes       times and by different densities of sea urchins, mass
occurring at different rates. However, the difference     mortality of H. erythrogramma may result in different
in the relative decline of Delisea pulchra between       foliose algal community composition because of differ-
high and low urchin densities suggests that a lower      ent grazing histories.
density of Heliocidaris erythrogramma does not just        Clearly, whether subtidal temperate communities
mean a slower rate of removal for this species. In       end-up at the crustose- or foliose algal-dominated state
fact, a threshold density of H. erythrogramma (~80       depends on both the urchin density and its persistence
urchins m–2; see Wright & Steinberg 2001) may be        in time (Ebeling et al. 1985, Fletcher 1987, Andrew
required before Delisea pulchra is consumed. Similar      1993, Shears & Babcock 2002, Konar & Estes 2003). Sea
threshold densities of H. erythrogramma were not        urchin predators may influence the switching between
apparent in the decline of other foliose algae,        these alternative community states (e.g. Estes &
although note the variable slopes between relative       Pamisano 1974, Estes & Duggins 1995). In southeastern
cover and time for Corallina officinalis (Fig. 6). The     Australia, large crustose algal-dominated areas occur
mechanisms responsible for this threshold effect for      on coastal reefs (Andrew & O’Neill 2000), possibly due
D. pulchra are unclear, but may be related to the       to a reduction in predation on herbivorous sea urchins
strong feeding deterrent effects of furanones (Wright     because of a reduction in abundance, or lack, of sea
et al. 2004). Similarly, once sea urchins have estab-     urchin predators (Estes & Steinberg 1988, Steinberg et
lished a crustose-dominated state, total sea urchin      al. 1995). Theory predicts that the switch between al-
removal may be required for the transition to a        ternative community states requires not just a restora-
foliose algal- dominated state (Benedetti-Cecchi et al.    tion of the original conditions, but a catastrophic shift
1998, Hill et al. 2003).                    (Scheffer et al. 2001). Mass mortality of sea urchins due
                    Wright et al.: Community effects of sea urchin grazing                  155




to factors such as storms, reduced salinity or pathogens      Estes JA, Palmisano JF (1974) Sea otters: their role in structur-
(Ebeling et al. 1985, Scheibling 1986, Andrew 1991,           ing nearshore communities. Science 185:1058–1060
                                  Estes JA, Steinberg PD (1988) Predation, herbivory, and kelp
Scheibling & Hennigar 1997) may enable such a cata-
                                    evolution. Paleobiology 14:19–36
strophic shift to occur and allow southeast Australian       Fletcher WJ (1987) Interactions among subtidal Australian
reef communities to switch from a crustose-dominated          sea urchins, gastropods, and algae: effects of experimental
state back to a foliose algal-dominated state.             removals. Ecol Monogr 57:89–109
                                  Gagnon P, Himmelman JH, Johnson LE (2004) Temporal vari-
                                    ation in community interfaces: kelp-bed boundary dynam-
Acknowledgements. J.T.W, S.A.D, C.N.R, J.E.W. and A.G.B.P.       ics adjacent to persistent urchin barrens. Mar Biol 144:
were supported by A. P. A. post-graduate research scholar-       1191–1203
ships. The Australian Research Council and the Centre for      Hay ME (1984) Predictable spatial escapes from herbivory:
Marine Biofouling and Bio-Innovation provided further finan-      How do these affect the evolution of herbivore resistance
cial support.                              in tropical marine communities? Oecologia 64:396–407
                                  Hay ME, Steinberg PD (1992) The chemical ecology of plant-
                                    herbivore interactions in marine versus terrestrial commu-
          LITERATURE CITED                  nities. In: Rosenthal GA, Berenbaum M (eds) Herbivores:
                                    their interaction with plant secondary metabolites. Vol 2:
Andrew NL (1991) Changes in subtidal habitat following         Ecological and evolutionary processes. Academic Press,
  mass mortality of sea urchins in Botany Bay, New South        San Diego, CA, p 371–413
  Wales. Aust J Ecol 16:353–362                  Hay ME, Colburn T, Downing D (1983) Spatial and temporal
Andrew NL (1993) Spatial heterogeneity, sea urchin grazing,       patterns in herbivory on a Caribbean fringing reef: the
  and habitat structure on reefs in temperate Australia.        effects on plant distribution. Oecologia 58:299–308
  Ecology 74:292–302                        Hay ME, Duffy JE, Pfister CA, Fenical W (1987) Chemical
Andrew NL, O’Neill AL (2000) Large-scale patterns in habitat      defense against different marine herbivores: Are
  structure on subtidal rocky reefs in New South Wales. Mar      amphipods insect equivalents? Ecology 68:1567–1580
  Freshw Res 51:255–263                      Hay ME, Kappel QE, Fenical W (1994) Synergisms in plant
Andrew NL, Underwood AJ (1993) Density-dependent forag-         defences against herbivores: interactions of chemistry,
  ing in the sea urchin Centrostephanus rodgersii on shal-       calcification, and defence quality. Ecology 75:1714–1726
  low subtidal reefs in New South Wales, Australia. Mar      Hill NA, Blount C, Poore AGB, Worthington D, Steinberg PD
  Ecol Prog Ser 99:89–98                        (2003) Grazing effects of the sea urchin Centrostephanus
Benedetti-Cecchi L, Bulleri F, Cinelli F (1998) Density depen-     rodgersii in two contrasting rocky reef habitats: effects of
  dent foraging of sea urchins in shallow subtidal reefs on      urchin density and its implications for the fishery. Mar
  the west coast of Italy (western Mediterranean). Mar Ecol      Freshw Res 54:691–700
  Prog Ser 163:203–211                       Hughes TP (1994) Catastrophes, phase shifts, and large-scale
Bertness MD, Trussell GC, Ewanchuk PJ, Silliman BR, Crain        degradation of a Caribbean coral reef. Science 256:
  CM (2004) Consumer-controlled community states on Gulf        1547–1551
  of Maine rocky shores. Ecology 85:1321–1331           Keesing JK (2001) The ecology of Heliocidaris erythro-
Carpenter RC (1986) Partitioning herbivory and its effects on      gramma. In: Lawrence JM (ed), Edible sea urchins: biol-
  coral reef algal communities. Ecol Monogr 56:345–36         ogy and ecology. Elsevier Science, Berlin, p 261–270
Clarke KR, Warwick RM (1994) Change in marine communi-       Konar B, Estes JA (2003) The stability of boundary regions
  ties: an approach to statistical analysis and interpretation,    between kelp beds and deforested areas. Ecology 84:
  1st edn. Plymouth Marine Laboratory, Plymouth            174–185
Dean TA, Schroeter SC, Dixon JD (1984) Effects of grazing by    Leinaas HP, Christie H (1996) Effects of removing sea urchins
  two species of sea urchins (Stronglyocentrotus francis-       (Stronglyocentrotus droebachiensis): stability of the bar-
  canus and Lytechinus anamesus) on recruitment and sur-        ren state and succession of kelp forest recovery in the east
  vival of two species of kelp (Macrocystis pyrifera and        Atlantic. Oecologia 105:524–536
  Pterygophora californica). Mar Biol 78:301–313          Lewis SM (1986) The role of herbivorous fishes in the organi-
Dublin HT, Sinclair AR, McGlade J (1990) Elephants and fire       zation of a Caribbean reef community. Ecol Monogr 56:
  as causes of multiple stable states in the Serengeti-Mara      183–200
  woodlands. J Anim Ecol 59:1147–1164               Lewis SM, Norris JN, Searles RB (1987) The regulation
Dudgeon S, Petraitis PS (2001) Scale-dependent recruitment       of morphological plasticity in tropical reef algae by
  and divergence of intertidal communities. Ecology 82:        herbivory. Ecology 68:636–641
  991–1006                             May RM (1977) Thresholds and breakpoints in ecosystems
Duffy JE, Hay ME (2000) Strong impacts of grazing            with a multiplicity of stable states. Nature 269:471–477
  amphipods on the organization of a benthic community.      Miller MW, Hay ME (1996) Coral/seaweed/grazer/nutrient
  Ecol Monogr 70:237–263                        interactions on temperate reefs. Ecol Monogr 66:
Duffy JE, Hay ME (2001) The ecology and evolution of marine       323–344
  consumer-prey interactions. In: Bertness MD, Gaines SD,     Paul VJ, Cruz-Rivera E, Thacker RW (2001) Chemical media-
  Hay ME (eds) Marine community ecology. Sinauer Associ-        tion of macroalgal-herbivore interactions: ecological and
  ates, Sunderland, MA, p 131–157                   evolutionary perspectives. In: McClintock JB, Baker BJ
Ebeling AW, Laur DR, Rowley RJ (1985) Severe storm distur-       (eds) Marine chemical ecology. CRC Press, Boca Raton,
  bances and reversal of community structure in a southern       FL, p 227–265
  California kelp forest. Mar Biol 84:287–294           Pawlik JR, Chanas B, Toonen RJ, Fenical W (1995) Defenses
Estes JA, Duggins DO (1995) Sea otters and kelp forests in       of Caribbean sponges against predatory reef fish: I.
  Alaska: generality and variation in a community ecologi-       Chemical deterrency. Mar Ecol Prog Ser 127:183–194
  cal paradigm. Ecol Monogr 65:75–100               Pennings SC, Paul VJ (1992) Effect of plant toughness, calci-
156                      Mar Ecol Prog Ser 298: 143–156, 2005




  fication and chemistry on herbivory by Dolabella auricu-      secondary metabolites. Cornell University Press, Ithaca,
  laria. Ecology 73:1606–1619                    NY, p 51–92
Quinn GP, Keough MJ (2002) Experimental design and data      Steinberg PD, Van Altena I (1992) Tolerance of marine in-
  analysis for biologists. Cambridge University Press, Cam-     vertebrate herbivores to brown algal phlorotannins in
  bridge                               temperate Australasia. Ecol Monogr 62:189–222
Scheffer M, Carpenter S, Foley, JA, Folke C, Walker B (2001)    Steinberg PD, Estes JA, Winter FC (1995) Evolutionary conse-
  Catastrophic shifts in ecosystems. Nature 413:591–596       quences of food chain length in kelp forest communities.
Scheibling RE (1986) Increased macroalgal abundance fol-       Proc Natl Acad Sci USA 92:8145–8148
  lowing mass mortality of sea urchins (Stronglyocentrotus    Thacker RW, Ginsburg DW, Paul VJ (2001) Effects of herbi-
  droebachiensis) along the Atlantic coast of Nova Scotia.      vore exclusion and nutrient enrichment on coral reef
  Oecologia 68:186–198                        macroalgae and cyanobacteria. Coral Reefs 19:318–329
Scheibling RE, Hennigar AW (1997) Recurrent outbreaks of      Van de Koppel J, Rietkerk M, Weissing FJ (1997) Cata-
  disease in sea urchins Stronglyocentrotus droebachiensis      strophic vegetation shifts and soil degradation in terres-
  along the Atlantic coast of Nova Scotia: evidence for a link    trial grazing systems. Trends Ecol Evol 12:352–356
  with large-scale meteorologic and oceanographic events.     Williamson JE, Carson DG, de Nys R, Steinberg PD (2004)
  Mar Ecol Prog Ser 152:155–165                   Demographic consequences of an ontogenetic shift by a
Schrage LJ, Downing JA (2004) Pathways of increased water       sea urchin in response to host plant chemistry. Ecology 85:
  clarity after fish removal from Ventura Marsh; a shallow,     1355–1371
  eutrophic wetland. Hydrobiologia 511:215–231          Wright JT, Steinberg PD (2001) Effects of variable recruitment
Schupp PJ, Paul VJ (1994) Calcium carbonate and secondary       and post-recruitment herbivory on local population size of
  metabolites in tropical seaweeds: variable effects on her-     a marine alga. Ecology 82:2200–2215
  bivorous fishes. Ecology 75:1172–1185              Wright JT, Benkendorff K, Davis AR (1997) Habitat associated
Shears NT, Babcock RC (2002) Marine reserves demonstrate       differences in temperate sponge assemblages: the impor-
  top-down control of community structure on temperate        tance of chemical defence. J Exp Mar Biol Ecol 213:
  reefs. Oecologia 132:131–142                    199–213
Sousa WP (2001) Natural disturbance and the dynamics of      Wright JT, de Nys R, Steinberg PD (2000) Geographic varia-
  marine benthic communities. In: Bertness MD, Gaines SD,      tion in halogenated furanones from the red alga Delisea
  Hay ME (eds) Marine community ecology. Sinauer Associ-       pulchra and associated herbivores and epiphytes. Mar
  ates, Sunderland, MA, p 85–130                   Ecol Prog Ser 207:227–241
Steinberg PD (1992) Geographical variation in the interaction   Wright JT, de Nys R, Poore AGB, Steinberg PD (2004) Chem-
  between marine herbivores and brown algal secondary        ical defence in a marine alga: heritability and the potential
  metabolites. In: Paul VJ (ed) Ecological roles for marine     for selection by herbivores. Ecology 85:2946–2959

Editorial responsibility: Otto Kinne (Editor-in-Chief),      Submitted: September 14, 2004; Accepted: March 12, 2005
Oldendorf/Luhe, Germany                      Proofs received from author(s): July 26, 2005