Journal of Experimental Marine Biology and Ecology
                     295 (2003) 63 – 90
                                       www.elsevier.com/locate/jembe




  Establishment of the introduced kelp Undaria
pinnatifida in Tasmania depends on disturbance to
        native algal assemblages
            Joseph P. Valentine, Craig R. Johnson *
    School of Zoology and Tasmanian Aquaculture and Fisheries Institute, University of Tasmania,
              GPO Box 252-05, Hobart, Tasmania 7001, Australia
     Received 21 February 2003; received in revised form 23 May 2003; accepted 30 May 2003



Abstract

  Despite recent rapid increases in the occurrence of nonindigenous marine organisms in the marine
environment, few studies have critically examined the invasion process for a marine species. Here
we use manipulative experiments to examine processes of invasion for the Asian kelp Undaria
pinnatifida (Harvey) Suringar at two sites on the east coast of Tasmania. Disturbance to reduce cover
of the native algal canopy was found to be critical in the establishment of U. pinnatifida, while the
presence of a stable native algal canopy inhibited invasion. In the first sporophyte growth season
following disturbance of the canopy, U. pinnatifida recruited in high densities (up to 19 plants mÀ 2)
while remaining rare or absent in un-manipulated plots. The timing of disturbance was also
important. U. pinnatifida recruited in higher densities in plots where the native canopy was removed
immediately prior to the sporophyte growth season (winter 2000), compared with plots where the
canopy was removed 6 months earlier during the period of spore release (spring 1999). Removal of
the native canopy also resulted in a significant increase in cover of sediment on the substratum. In the
second year following canopy removal, U. pinnatifida abundance declined significantly, associated
with a substantial recovery of native canopy-forming species. A feature of the recovery of the native
algal canopy was a significant shift in species composition. Species dominant prior to canopy
removal showed little if any signs of recovery. The recovery was instead dominated by canopy-
forming species that were either rare or absent in the study areas prior to manipulation of the canopy.
D 2003 Elsevier B.V. All rights reserved.

Keywords: Invasion process; Introduced macroalgae; Establishment; Disturbance; Canopy removal; Undaria
pinnatifida




  * Corresponding author. Tel.: +61-3-6226-2582; fax: +61-3-6226-2745.
  E-mail address: Craig.Johnson@utas.edu.au (C.R. Johnson).


0022-0981/03/$ - see front matter D 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0022-0981(03)00272-7
64       J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

1. Introduction

  The introduction of nonindigenous species into the marine environment is recognised
as a major threat to marine ecosystems, with potentially dramatic effects on biological
diversity, productivity, habitat structure and fisheries (Carlton, 1999; Bax et al., 2001).
Over the past two decades there has been a vast increase in the worldwide spread of
nonindigenous organisms, due mainly to dispersal via human-mediated transport (Bax et
al., 2001). It is estimated that more than 15,000 species of marine organisms may be
transported around the world in ships’ ballast water each week (Carlton, 1999). This
rapid acceleration in spread of nonindigenous marine organisms now poses a major
challenge for management of marine ecosystems. When presented with a large number
of introduced species, managers must decide which species have immediate priority for
control, which to control if time and finances are available, and which to leave alone
(Hiebert, 1997).
  Knowledge of the threat posed by an introduced species is essential to effectively
prioritise species for management purposes (Byers et al., 2002). One important aspect of
threat is associated with the invasion process itself, particularly the role of disturbance in
the establishment of an introduced species. While there is substantial evidence showing
that disturbance can be a key mechanism in the invasion of both terrestrial and freshwater
organisms (e.g. Hobbs and Adkins, 1988; Hobbs and Huenneke, 1992; Lodge, 1993;
Moyle and Light, 1996; D’Antonio et al., 1999), relatively few examples exist for marine
communities (but see Nichols et al., 1990; Reusch and Williams, 1999).
  In recent years the kelp Undaria pinnatifida (Harvey) Suringar has experienced a
global range expansion in temperate waters. Native to Japanese, Korean and Chinese
coasts, U. pinnatifida has spread to the Atlantic and Mediterranean coasts of Europe
(Curiel et al., 1998; Castric-Fey et al., 1993; Fletcher and Manfredi, 1995) and to shores of
New Zealand (Hay and Luckens, 1987), Argentina (Casa and Piriz, 1996) and Australia
(Campbell and Burridge, 1998; Sanderson, 1990). While the plant was intentionally
introduced to the Atlantic Coast of Europe in 1983 (Floc’h et al., 1991), introductions to
other areas are all thought to have occurred accidentally via international shipping activity,
mediated either through hull fouling or discharge of ballast water, or associated with
translocation of aquaculture organisms (Perez et al., 1981).
  While the occurrence and spread of U. pinnatifida has been well documented, the
mechanism of its invasion and impact on native communities has received little attention.
In one of the few experimental studies to date, local kelp species were shown to be
resistant to invasion by U. pinnatifida on the Atlantic coast of France (Floc’h et al., 1996).
Despite this result, it is speculated widely that U. pinnatifida is a highly invasive species,
able to competitively displace native species in sheltered to moderately exposed waters
(Rueness, 1989; Fletcher and Manfredi, 1995).
  In the Mercury Passage, where the plant was first recorded in Tasmania, U. pinnatifida
exhibits an annual growth pattern. Macroscopic sporophytes typically recruit in winter
growing through spring to a length of up to 2 m. Reproduction is thought to occur during
late spring – early summer, after which the plant degenerates. Sporophytes are generally
absent from reefs by the end of summer (Sanderson and Barrett, 1989). U. pinnatifida
occurs most abundantly on urchin ‘barrens’ characterized by high densities of the sea urchin
        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90  65

Heliocidaris erythrogramma and low cover of native algae. In these habitats U. pinnatifida
forms monospecific stands during the sporophyte growth season (Sanderson, 1990). Recent
work has demonstrated a significant negative correlation between sea urchin densities and
native algae, but a significant positive correlation between sea urchins and U. pinnatifida
(Johnson, unpublished). U. pinnatifida also occurs abundantly in other disturbed habitats
such as areas of sandscour at the base of reefs and on unstable substrata, while it occurs
rarely in established macroalgal stands (Sanderson, 1997; C. Johnson, pers. comm.).
  Observations of U. pinnatifida occurring abundantly in disturbed habitats suggest
disturbance is potentially playing a significant role in its establishment. U. pinnatifida also
manifests many characteristics of an opportunistic species, such as short lifespan, high
growth rate, a high biomass invested in reproduction, small propagule size and high
number of propagules released, and a single reproductive episode (Grime, 1977; Clayton,
1990). Species with these features are commonly associated with disturbance (Clayton,
1990). If U. pinnatifida requires disturbance to establish, then there exists a range of
management options, which include targeting the cause of the disturbance rather than the
plant itself. If disturbance is linked to anthropogenic activity, then managing disturbance
may prove a cost-effective option. Alternatively, if U. pinnatifida is capable of displacing
native algae in the absence of any primary mechanism of facilitation such as disturbance,
then it represents a major threat to the integrity of native algal communities. Under this
scenario, management may need to target the plant directly.
  In this study we investigate the role of disturbance as a process facilitating invasion of
dense stands of U. pinnatifida. Manipulative experiments were used to examine the
relationship between disturbance, establishment of U. pinnatifida and subsequent recovery
of native species.


2. Materials and methods

2.1. Study sites

  The experiment was conducted at 7– 12 m depth on rocky reef in the Mercury Passage, on
the east coast of Tasmania (Fig. 1). Reefs in this area support a variety of algal communities,
ranging from sea urchin ‘barrens’ (dominated by H. erythrogramma) seasonally dominated
by U. pinnatifida, to areas dominated by diverse stands of native canopy-forming algae. Our
experiments were conducted at two sites (Flensers Point and Lords Bluff), dominated by
native algal species and as far as practically possible from the nearest dense stands of U.
pinnatifida (ca. 0.2 km at Lords Bluff and 1.0 km at Flensers Point).
  Both sites are characterized by gently sloping rocky substratum to a depth of 12– 14 m,
with moderate topographic relief. Although there is slight variation in aspect between the
two sites, they are similarly exposed to easterly swells, which although infrequent, can be
large. Using the classification scheme proposed for Tasmanian subtidal communities by
Edgar (1984), the sites are described as moderately exposed and support a mixed algal
assemblage.
  Flensers Point was dominated by the fucoid Seirococcus axillaris, however, the
common kelp Ecklonia radiata and the fucoids Carpoglossum confluens, Cystophora
66         J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90




   Fig. 1. Map of Mercury Passage showing the location of study sites at Flensers Point and Lords Bluff.


retroflexa and Sargassum fallax were also distributed patchily throughout the study area.
At Lords Bluff, a range of canopy-forming species were found including E. radiata,
Phyllospora comosa, C. confluens and S. axillaris. The understorey at both sites consisted
of a diverse assemblage of turfing algal species, encrusting algae and invertebrates.

2.2. Experimental manipulations

  Experimental manipulations were applied to fixed 16-m2 quadrats, while response
variables were monitored only in the inner 4-m2 of each quadrat to minimize edge effects.
The experiment followed a three-way factorial design representing all possible combina-
tions of two levels of each of three factors, viz.:

1. Disturbance (two levels; 100% removal of native algal canopy, no removal)
2. U. pinnatifida spore enhancement (two levels; background, enhanced)
3. Site (two sites).

  Treatments requiring manipulation were assigned at random at each site, and there were
three replicates of each treatment. The disturbance treatment, involving physical removal
        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90  67

of the macroalgal canopy, mimics natural disturbance caused by urchin grazing and
storms. Canopy removal was conducted initially in spring (November 1999), during the
period of spore release by U. pinnatifida (Sanderson, 1997). Plants were removed by
carefully cutting stipes immediately above the holdfast, while understorey species were
left intact.
  In treatments involving enhancement of U. pinnatifida spores, mesh bags were filled
with fertile sporophylls and hung over the plots. Fresh material was added every 4 –6
weeks for as long as fresh sporophyll material was available in sufficient quantities.
Containing the sporophylls in a coarse (20 mm) mesh bag prevented their grazing by
toothbrush leatherjackets (Acanthaluteres vittiger) which caused significant damage to
unprotected sporophylls. Spore enhancements were undertaken from Nov. 1999 to Jan.
2000 and from Sep. 2000 to Jan. 2001.
  To minimise confounding of treatments involving U. pinnatifida spores, experimental
plots were separated by a minimum distance of 10 m. We assumed that the effective spore
shadow of U. pinnatifida is limited and predominantly within a few metres of the parent
plant, as has been demonstrated for other large brown algae (Ambrose and Nelson, 1982;
Dayton, 1985; Andrew and Viejo, 1998).
  To examine the effect of timing of disturbance on invasion by U. pinnatifida, an
additional canopy removal was employed in winter (June 2000). In contrast to the initial
canopy removal in spring, this removal of native algae was immediately prior to the
appearance of macroscopic U. pinnatifida sporophytes. At each site there were three
replicate plots of this treatment.

2.3. Assessment of algal abundance

  The algal community was assessed immediately prior to manipulation and at three
monthly intervals thereafter for 24 months. Abundance of canopy-forming species was
measured in terms of stipe counts (i.e. density) and percentage cover. Stipe counts
involved recording all adult plants >30 cm in length in each 4-m2 plot. Abundance of
understorey algae, sessile invertebrates and sediment was assessed in terms of
percentage cover. Percentage cover was estimated with a 0.25-m2 quadrat using a
point intercept method. The quadrat was divided with a grid of 49 evenly spaced
intersections and was laid flat on the reef during algal assessment. Algae occurring
under each intercept and one corner of the quadrat were recorded, to give a total of
50 intersections per quadrat. Four randomly positioned quadrats were assessed in this
way for each plot on every sampling occasion. Where a dense cover of canopy algae
was present in a quadrat, cover was assessed in a two-stage process. First, cover of
canopy algae was estimated. Secondly, the fronds of the canopy species were moved
aside to allow assessment of the cover of understorey algae, sessile invertebrates and
sediment. Accordingly, the total percentage cover for individual quadrats can exceed
100%.
  Organisms were identified in situ to the highest taxonomic resolution possible. For
canopy algae, identification to species level was possible, however, it was necessary to
allocate other species to species complexes or guilds (e.g. foliose red algae, brown turf
algae).
68        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

2.4. Analysis

2.4.1. Univariate analyses
  Densities (i.e. stipe counts) were analysed using a three-way Model I analysis of
variance (ANOVA) with the main factors of canopy removal (two levels), U. pinnatifida
spore enhancement (two levels) and site (two levels) all treated as fixed factors. Site was
considered a fixed factor because possible sites available for the experiment (i.e. of similar
depth, exposure, topography, extent of reef and proximity to nearest dense U. pinnatifida
stand) was essentially limited to the two sites chosen.
  Analysis of responses to treatments assessed in November 2000 (1 year after the initial
canopy removal) revealed no effect of U. pinnatifida spore enhancement on subsequent U.
pinnatifida density (Table 1). In tests conducted on cover of native algae, the effect of U.
pinnatifida spore enhancement was similarly highly nonsignificant. Consequently, treat-
ments of F U. pinnatifida spores were excluded from further analysis, enabling pooling of
treatments and greater power to examine the effect of canopy removal.
  In subsequent analyses in which treatments of F spore enhancement were pooled, data
on stipe counts were analysed by a two-way Model I ANOVA, while a three-factor Model
III nested ANOVA was used for cover data. Both analyses included canopy removal (three
levels) and site (2 levels). There were three levels of canopy removal because these
analyses included the treatment of winter canopy removal. The nested ANOVA included
the effect of plot nested within all combinations of canopy removal*site as a random
factor. The design was unbalanced since there were three replicates of each treatment for
the winter canopy removal treatment, but six replicates of the remaining treatments (after
pooling across treatments with F U. pinnatifida spore enhancement). This analysis was
conducted on data collected during assessment of algal community composition in
November 2000 and November 2001. This allowed examination of the algal response
to canopy removal during the peak period of U. pinnatifida sporophyte development, 1
and 2 years after the initial canopy removals. For both density and cover data, three
planned comparisons were conducted for each site, viz. (i) control vs. spring 1999 canopy
removal, (ii) control vs. winter 2000 canopy removal, and (iii) spring 1999 vs. winter


Table 1
Three-factor Model I ANOVA examining the response of Undaria pinnatifida in November 2000 to experimental
treatments initiated during November 1999. The analysis was conducted on square root transformed stipe counts
of all U. pinnatifida plants >30 cm in length in each experimental plot (n = 3). Note that the effect of the U.
pinnatifida spore enhancement treatment was highly nonsignificant
Source of variation         df         MS           F           P
Canopy removed (C)          1         38.4816         64.27         0.0001
Enhanced spores (E)          1          0.0487         0.08         0.7790
Site (S)               1          6.6502         11.11         0.0042
C*E                  1          1.4553         2.43         0.1385
C*S                  1          5.8066         9.70         0.0067
S*E                  1          0.0846         0.14         0.7120
C*S*E                 1          0.2115         0.35         0.5606
Error                16          0.5987
        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90   69

2000 canopy removal. The Dunn – Sidak adjustment (aadjusted = 1 À (1 À a)p, where
p = number of tests) was used to adjust the significance level associated with planned
comparisons.
  Prior to all univariate tests, transformations to stabilize variances were determined from
the relationship between group standard deviations and means (Draper and Smith, 1981).
Transformations are expressed in terms of the untransformed variate, Y. All univariate tests
were undertaken using the SASR statistical package.

2.4.2. Multivariate analyses
  To describe community responses to treatments and assess the significance of differ-
ences between treatments, nonmetric multi-dimensional scaling (MDS) and nonparametric
MANOVA (np-MANOVA) were used, respectively. The relationship between controls and
canopy removal plots was compared before manipulation and 2 years after manipulation at
each site. To identify species most responsible for any observed differences in community
structure, SIMPER analysis was conducted. These analyses were based on Bray– Curtis
similarity matrices derived from percentage cover data after a fourth root transformation to
reduce the influence of dominant species. MDS and SIMPER analyses were undertaken
using the PRIMER 4.0 software (Carr and Clarke, 1994), while np-MANOVAs were
undertaken as outlined in Anderson (2001). For np-MANOVA, the winter canopy
removals were excluded from the analysis because of the inherent problems of low power
as a result of low replication (n = 3) and, therefore, the small number of permutations
available to determine the distribution of the test statistic.


3. Results

3.1. The effect of canopy removal on the density of U. pinnatifida and native canopy-
forming algae

  The effect of canopy removal had a dramatic effect on U. pinnatifida density in the
spring growth period of the following year (Fig. 2a). While U. pinnatifida remained rare or
absent in controls, plots from which the canopy was removed were characterized by the
appearance of U. pinnatifida sporophytes, to a maximum density of 19 plants mÀ 2 in
some plots. The trend was qualitatively consistent among sites, however, there were
significantly more U. pinnatifida plants associated with the Lords Bluff site, evidenced by
a highly significant ‘‘canopy removal*site’’ interaction ( F = 14.71, df2,24, P < 0.0001). The
timing of disturbance events also influenced U. pinnatifida abundance. Canopy removals
conducted in winter 2000, at the onset of the period of sporophyte growth and
development, exhibited higher numbers of U. pinnatifida plants compared to plots where
the canopy was removed the previous spring. This trend was evident at both sites, although
a statistically significant result was observed at Lords Bluff ( F = 44.41, df1,24, P < 0.0001),
but not at Flensers Point at the adjusted a level ( F = 7.31, df1,24, P < 0.0124; aadjusted =
0.0085).
  Algal assessments conducted in November 2001 (during the second season of U.
pinnatifida sporophyte growth following disturbance) revealed a significant effect of
70         J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90




Fig. 2. Effect of canopy removals on abundance of Undaria pinnatifida and total canopy-forming native algae
assessed in (a) November 2000 and (b) November 2001. Data are means (F SE) of stipe counts (n = 6 plots per
treatment for spring canopy removal and controls; n = 3 plots per treatment for winter canopy removal). Note that
stipe counts represent plants >30 cm total length. Canopy-forming native species include Ecklonia radiata,
Phyllospora comosa, Seirococcus axillaris, Carpoglossum confluens, Cystophora monoliformis, C. retroflexa,
Sargassum fallax and S. vestitum.
        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90  71

‘‘site’’ ( F = 38.31, df1,24, P < 0.0001) but no significant response of U. pinnatifida to
the canopy manipulations conducted 18 and 24 months previously (Fig. 2b). At
Flensers Point, very low levels of U. pinnatifida were observed in ‘canopy removal’
plots in November 2001 while the density of native species increased markedly (Fig.
2b). In contrast, U. pinnatifida plants were observed in moderate levels (mean 5 plants
mÀ 2) in all treatments at Lords Bluff (including controls). The number of U.
pinnatifida plants in plots from which the canopy was removed at Lords Bluff
decreased significantly from November 2000 to the November 2001 assessment, while
density of native species increased to levels comparable with controls (Fig. 2b). It
should also be noted that the density of native canopy-forming algae declined in
control plots at Lords Bluff between November 2000 (mean 7.8 plants mÀ 2) and
November 2001 (mean 5.1 plants mÀ 2). This was due to a decline in P. comosa and
E. radiata associated with above average water temperatures during the 2000/2001
summer.

3.2. Native canopy-forming algae: species composition

  Although densities of native canopy-forming algae had recovered in plots from which
the canopy was removed by November 2001 (24 months after the initial canopy removal),
the species composition in control plots and recovered ‘canopy removal’ plots was
distinctly different. While Seiroccoccus axillaris continued to dominate control areas
throughout the experiment at Flensers Point, the assemblages that developed in areas
where the canopy was removed consisted mainly of S. fallax, C. retroflexa, Sargassum
vestitum and, to a lesser extent, Cystophora monoliformis (Fig. 3). Similarly, at Lords
Bluff, the assemblage in un-manipulated control plots dominated by E. radiata, P. comosa,
S. axillaris and C. confluens was replaced by C. retroflexa and C. monoliformis in the
‘canopy removal’ treatments (Fig. 3). At both sites, species abundant in control areas were
rare or absent in plots from which the canopy was removed, so differences between
treatments could not be tested statistically.

3.3. Recovery of native canopy algae: percentage cover

  While stipe density was appropriate to examine some aspects of the response of U.
pinnatifida and native canopy algae, a more detailed examination of recovery patterns of
the entire community was based on plant cover. Cover can provide greater sensitivity than
data describing density, largely reflecting the different growth forms and densities among
algal species (Johnson and Mann, 1993).
  There were substantial differences among sites in the response of native canopy-
forming algae to canopy removal. During the first year following canopy removal there
was a gradual increase in cover at Flensers Point, although by November 2000 cover in
control plots (73% F 6.5 SE) was still considerably greater than that in plots where canopy
removals had been conducted in spring 1999 (28% F 5.2 SE) and winter 2000 (11% F 1.3
SE) (Fig. 4; Table 2). However, during 2001 the cover of native canopy-forming algae
increased dramatically in plots from which the canopy had been removed in both spring
1999 and winter 2000, reflecting the trend shown for stipe counts. By November 2001,
72         J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90




Fig. 3. Abundance of dominant canopy-forming native algae in relation to canopy removal at two sites in Mercury
Passage, November 2001. Data represent mean stipe densities (+ SE) (n = 6 replicate plots per treatment for spring
canopy removals and controls; n = 3 replicates plots per treatment for winter canopy removals).
          J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90          73




Fig. 4. Effect of removal of native canopy-forming algae on the cover of various algal guilds, invertebrates and
the sediment matrix at Flensers Point. Data are mean percentage cover (F SE) (n = 6 plots per treatment for spring
canopy removals and controls; n = 3 plots per treatment for winter canopy removals). Circles = canopy removed
spring 1999; triangles = canopy removed winter 2000; crosses = control.
                                                                                      74
Table 2
Analysis of the effect of removing native canopy-forming algae on the cover of various algal guilds, invertebrates and the sediment matrix, assessed in November 2000.
Results are of the overall ANOVA examining the effect of canopy removal and site, and of the three planned comparisons for each site. For planned comparisons,
‘‘co’’= control, ‘‘sp’’= spring canopy removal, while ‘‘wi’’= winter canopy removal. Significant P-values are shown in bold face: P-values < 0.05 are significant for the
main analysis; P-values < 0.0085 are significant for the planned comparisons (a adjusted using Dunn – Sidak method). All of the tests presented use the MS Plot (C*S) as




                                                                                      J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90
the error term
Guild (transformation)     Source of variation                         Planned comparisons
                                                  Flensers point           Lords Bluff
                Canopy removal (C) Site (S)      C*S      Plot (C*S)   co vs. sp  co vs. wi sp vs. wi co vs. sp    co vs. wi sp vs. wi
                F (df = 2,24)     F (df = 1,24) F (df = 2,24) F (df = 24,90) F        F      F     F       F      F
                P           P       P       P        P      P      P     P       P      P
Sediment cover       18.50            1.58      1.79     5.05       17.46  19.20  0.94       73.71  42.20  0.25
 [log (Y + 1)]       0.0001           0.2205     0.1879    0.0001      0.0001  0.0001 0.3337       0.0001  0.0001 0.6189
Encrusting algae     117.32            8.28      4.84     2.11      138.70  33.24  14.83       252.78  123.58  3.48
 [log (Y + 1)]       0.0001           0.0083     0.0171    0.0061      0.0001  0.0001 0.0002       0.0001  0.0001 0.0648
Total foliose algae    22.14           12.93      2.38     2.95       16.44  13.57  0.14       52.93  61.18  3.54
 (no transformation)    0.0001           0.0015     0.136     0.0001      0.0001  0.0004 0.7090       0.0001  0.0001 0.0624
Large brown algae     103.46            4.27      3.44     2.50       81.34  102.90  7.73       174.80  210.57  13.81
 (no transformation)    0.0001           0.0498     0.0487    0.0010      0.0001  0.0001 0.0064       0.0001  0.0001 0.0003
Foliose red algae (sqrt)  15.63            6.84      1.53     6.05       68.03  31.19  1.32       16.65  20.81  1.51
              0.0001           0.0152     0.2373    0.0001      0.0001  0.0001 0.2527       0.0001  0.0001 0.2212
Brown turf [log (Y + 1)]  14.67            4.04      4.09     1.43       0.45  18.94  14.45       23.21  17.17  0.04
              0.0001           0.0559     0.0296    0.1143      0.5018  0.0001 0.0002       0.0001  0.0001 0.8335
Undaria pinnatifida    40.66           54.86     17.01     1.60       5.96   5.62  0.14       59.60  146.43  33.61
 [log (Y + 1)]       0.0001           0.0001     0.0001    0.0586      0.0162  0.0194 0.7065       0.0001  0.0001 0.0001
Invertebrates (sqrt)    39.99           10.19      4.76     1.27       20.30   5.94  1.54       83.79  29.06  4.34
              0.0001           0.0039     0.0181    0.2060      0.0001  0.0164 0.2171       0.0001  0.0001 0.0394
Green algae         1.94            3.86      2.45     1.40
 (no transformation)    0.1654           0.0612     0.1074    0.1310
Zonaria/Lobophora complex  0.06            9.46      2.54     1.65        4.41     0.09   1.99     2.72     0.21   0.79
 (no transformation)    0.9435           0.0052     0.0997    0.0471       0.0379    0.7599  0.1615    0.1016    0.6468  0.3762
        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90  75

there was no significant difference in the cover of native canopy species in control plots
(86% F 5.6 SE) and plots from which the canopy was removed in spring 1999 (71% F 4.0
SE). Cover in plots from which the canopy was removed in winter 2000 had increased
markedly (49% F 4.1 SE) but still remained significantly lower than that in controls (Fig.
4; Table 3).
  At Lords Bluff there was also a gradual increase in cover of native canopy-forming
species in the year following canopy removals in spring 1999 (31% F 6.3 SE) and winter
2000 (9% F 1 SE) (Fig. 5). The trend of recovery stalled somewhat in 2001, with spring
1999 (34% F 5.8 SE) and winter 2000 ‘canopy removal’ plots (15% F 4.0 SE) showing
only slight increases in cover. Unlike Flensers Point, where cover in controls remained
consistently high (mean 69-86%) over the entire 24 months of the study, the cover in
control areas at Lords Bluff declined significantly during the study period, averaging 98%
in November 2000 but declining to 54% in February 2001. This was mainly associated
with the declines in P. comosa and E. radiata. Despite this decline in cover in control
plots, cover in ‘canopy removal’ plots was still significantly lower than in controls by
November 2001 (Fig. 5; Table 3).

3.4. Response of understorey algae to canopy disturbance

  In interpreting the response of U. pinnatifida and native canopy-forming algae to
disturbance, it is also important to consider understorey algal species given that occupation
of space by turfing algal species can inhibit recruitment of canopy-forming species
(Dayton et al., 1984; Kennelly, 1987a; Airoldi, 1998). Thus, the response of turfing
species to canopy removal may have significant implications for both invasion of U.
pinnatifida as well as the recovery of native canopy-forming species.

3.4.1. Foliose red algae
  There was a significant response of foliose red understorey algae to canopy removal,
although the response varied significantly among sites and with the time since canopy
removal. At Flensers Point, foliose red algal cover remained at uniformly low levels
( < 5%) in control plots for the duration of the experiment while fluctuating significantly in
treatments in which the canopy was removed (Fig. 4). Cover increased to a peak in
November 2000 for canopy removals conducted in both spring 1999 (38% F 9.8 SE) and
winter 2000 (26% F 4.1 SE), after which a gradual decrease was recorded. No significant
effect on foliose red algae of disturbance to the canopy was detected on completion of the
final assessment in November 2001, 18 and 24 months after implementation of canopy
removals (Table 3).
  At Lords Bluff, cover of foliose red algae remained at low levels in all treatments prior
to November 2000, when cover increased in plots from which the canopy was removed in
spring 1999 (11% F 3.7 SE) and winter 2000 (18% F 8.6 SE) relative to controls
(1% F 1.0 SE) (Fig. 5). Cover in canopy removal treatments remained significantly higher
than in controls for the remainder of 2001, despite a slight increase in cover in the control
areas (Fig. 5). The significant ‘‘site’’ effect evident in the November 2001 assessment
reflected the higher cover of foliose red algae observed in all treatments at Lords Bluff
compared with Flensers Point.
                                                                                       76
Table 3
Analysis of the effect of removing native canopy-forming algae on the cover of various algal guilds, invertebrates and the sediment matrix, assessed in November 2001
Guild (transformation)   Source of variation                           Planned comparisons
                                                  Flensers point            Lords Bluff




                                                                                       J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90
              Canopy removal (C)   Site (S)    C*S       Plot (C*S)    co vs. sp  co vs. wi  sp vs. wi  co vs. sp   co vs. wi  sp vs. wi
              F (df = 2,24)     F (df = 1,24)  F (df = 2,24)  F (df = 24,90)  F      F      F      F       F      F
              P           P        P        P        P      P      P      P       P      P
Sediment cover (sqrt)    8.09          3.03      0.21      7.20       24.32    14.83    0.03    13.85     18.99    1.74
              0.0021         0.0948     0.8093     0.0001       0.0001   0.0002   0.8607    0.0003    0.0001   0.1898
Encrusting algae (sqrt)  85.18         14.28      1.60      1.62       125.82    86.21    0.02    89.02     39.98    1.90
              0.0001         0.0009     0.2230     0.0541       0.0001   0.0001   0.8997    0.0001    0.0001   0.1702
Total foliose algae     5.96          9.30      1.90      2.14        1.05    3.10    0.86     0.03     19.29    20.55
 (no transformation)    0.0079         0.0055     0.1720     0.0053       0.3087   0.0808   0.3560    0.8623    0.0001   0.0001
Large brown algae     13.47         25.19      0.47      2.66        4.74    19.29    6.84    16.08     31.92    5.64
 (no transformation)    0.0001         0.0001     0.6328     0.0005       0.0316   0.0001   0.0101    0.0001    0.0001   0.0192
Foliose red algae (sqrt)  1.94         11.04      1.02      4.17        0.05    2.46    3.10     8.71     7.30    0.09
              0.1658         0.0028     0.3765     0.0001       0.8158   0.1193   0.0810    0.0038    0.0079   0.7704
Brown turf (sqrt)      4.00          5.72      9.09      1.87        1.15    11.72    6.49    15.47     3.29    25.25
              0.0316         0.0249     0.0012     0.0186       0.2854   0.0009   0.0122    0.0001    0.0723   0.0001
Undaria pinnatifida     1.23         18.58      1.23      1.03        0.00    0.00    0.00     3.52     0.10    3.41
 (no transformation)    0.2960         0.0001     0.2960     0.4400       1.0000   1.0000   1.0000    0.0630    0.7550   0.0675
Invertebrates (sqrt)    4.30          1.49      2.21      1.80        9.59    16.93    2.52     0.64     0.29    0.01
              0.0253         0.2346     0.1311     0.0255       0.0025   0.0001   0.1155    0.4261    0.5896   0.9116
Green algae         0.24          0.12      0.50      1.32
 (no transformation)    0.7893         0.7290     0.6100     0.1731
Zonaria/Lobophora      5.81         10.68      5.98      1.36         4.35   4.96    0.27     7.59     6.18    22.41
 complex (sqrt)      0.0088         0.0033     0.0078     0.1498        0.0393  0.0279   0.6013    0.0069    0.0144   0.0001
Results are of the overall ANOVA examining the effect of canopy removal and site, and of the three planned comparisons for each site. For planned comparisons,
‘‘co’’= control, ‘‘sp’’= spring canopy removal, while ‘‘wi’’= winter canopy removal. Significant P-values are shown in bold face: P-values < 0.05 are significant for the
main analysis; P-values < 0.0085 are significant for the planned comparisons (a adjusted using Dunn – Sidak method). All of the tests presented use the MS Plot (C*S) as
the error term.
          J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90         77




Fig. 5. Effect of removal of native canopy-forming algae on the cover of various algal guilds, invertebrates and
the sediment matrix at Lords Bluff. Data are mean percentage cover (F SE) (n = 6 plots per treatment for spring
canopy removals and controls; n = 3 plots per treatment for winter canopy removals). Circles = canopy removed
spring 1999; triangles = canopy removed winter 2000; crosses = control.
78         J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

3.4.2. Brown turf algae
  The guild of ‘brown turf algae’ represented less than 10% cover in control plots at both
sites (Figs. 4 and 5). Cover in plots at Flensers Point subject to canopy removal in winter
2000 displayed consistently higher cover of brown turf than in control plots and in plots
where the canopy was removed in spring 1999 (Fig. 4). In contrast, at Lords Bluff cover of
brown turf in plots from which the canopy was removed in spring 1999 was higher than in
control plots and in plots where canopy removals occurred in winter 2000 (Fig. 5). These
differences are reflected in a significant ‘‘canopy removal*site’’ interaction evident for
assessments in November 2000 and 2001. A notable feature at Lords Bluff was the major
peak in brown turf cover observed in the first assessment following the spring 1999
canopy removal, which was associated with recruitment of Colpomenia spp. (Fig. 5). This
ephemeral species subsequently degenerated and comprised a minor component of algal
cover in all further assessments.

3.4.3. Green algae
  The green algal guild, comprising mainly species of Caulerpa, was a minor
component of the Lords Bluff flora. While they contributed up to 20% cover at Flensers
Point, no significant treatment effects were detected, indicating that abundance of
Caulerpa fluctuated patchily in time and space independent of our experimental treat-
ments (Fig. 4).

3.4.4. Zonaria/Lobophora complex
  In general, responses of algae in the Zonaria/Lobophora complex to experimental
treatments were relatively small. A significant effect of canopy removal was detected
during the November 2001 assessment at Lords Bluff, with cover in plots cleared of
canopy species eventually developing approximately twice the cover of that in control
plots (Fig. 5; Table 3). Cover of this guild at Flensers Point was consistently higher than
at Lords Bluff, however, differences between treatments at Flensers Point were not
significant.

3.4.5. Encrusting algae
  The encrusting alga guild, including nongeniculate coralline algae and Peyssionnella
spp., showed clear responses to experimental manipulations. Removal of the algal canopy
resulted in bleaching of the vast majority of encrusting algae present in experimental plots,
with no subsequent recovery observed over the 24-month study period (Figs. 4 and 5;
Table 3). A ‘‘canopy removal*site’’ interaction was evident at the November 2000
assessment, reflecting that the reduction in cover of encrusting algae at Lords Bluff was
more dramatic than at Flensers Point (Table 2).



Fig. 6. Ordination (MDS) showing relationship between experimental plots from which the algal canopy was
removed (in spring 1999 and winter 2000) and un-manipulated plots over the duration of the study (November
1999 – November 2001) at Flensers Point and Lords Bluff. The analysis is based on a Bray – Curtis matrix of
fourth root transformed percentage cover data. The plots associated with canopy removals and controls have been
outlined for clarity.
J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90  79
80         J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

3.5. Effect of canopy removal on sediment cover

  Cover of sediment, forming a loose matrix on the substratum of variable depth ca. 1– 10
mm, increased significantly immediately after canopy removal at both sites (Figs. 4 and 5).
Sediment cover remained significantly higher in canopy removal plots than in controls
throughout the study period (Table 3). Sediment cover was low in control plots, averaging
less than 4% in control areas at Flensers Point for the duration of the study, while at Lords
Bluff cover was < 2% during 2000, after which there was a slight increase to an average of
7% by November 2001.

3.6. Community level effects

  By November 2001, the total cover of foliose algae in plots from which the canopy
was removed initially (i.e. in spring 1999) had recovered to levels comparable with
controls at both sites (see Table 3). However, despite the recovery of cover, there were
significant differences between treatments in algal community structure. At Flensers
Point in November 2001, algal community structure in control plots and in plots from
which the canopy was removed were clearly separated in MDS space (Fig. 6a) despite
supporting similar cover. Although not as clear as the patterns observed at Flensers
Point, significant patterns in community structure were also apparent at Lords Bluff,
with np-MANOVA indicating differences among treatments in algal community
composition 24 months after the initial canopy removal (Table 4). An interesting
anomaly in algal composition at Lords Bluff was the increased variation in control
treatments in November 2001 relative to the two previous years (Fig. 6b). This reflects
dieback and therefore decreased abundance of P. comosa and E. radiata, which
occurred in the control plots after November 2000. Those control plots subject to
dieback, which initially supported a dense canopy of P. comosa and E. radiata, were
more similar to canopy removal treatments after the dieback, indicating that the changes
associated with the natural decline of these algae were similar to those observed in
artificial disturbances.


Table 4
Comparison of community structure in relation to canopy removal before (November 1999) and 24 months after
(November 2001) experimental manipulations
Site       Source of variation   Time
                     November 1999              November 2001
                     df    MS     F    P     MS      F     P
Flensers point  Canopy removal      (1,10)  2185.94  1.4524  0.2338  29,204.38  21.4214  0.0026
         Plot (canopy removal)  (10,36)  1505.06  1.8880  0.0058   1363.33   1.9413  0.0008
Lords Bluff   Canopy removal      (1,10)   655.91  0.3177  0.8552  22,443.06   9.482   0.0026
         Plot (canopy removal)  (10,36)  2064.67  3.0763  0.0004   2366.92   1.9121  0.0038
Results are two-factor nested np-MANOVAs based on a Bray – Curtis matrix of fourth root transformed data
(4999 permutations used for tests of significance). The level of significance was altered according to the Dunn –
Sidak adjustment, aadjusted = 0.013. Significant tests are shown in bold face. (Note that winter canopy removals
were not included in the analysis due to low replication.)
          J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90         81

Table 5
SIMPER analysis identifying individual species or guilds responsible for the differences in community structure
between treatments assessed in November 2001 at Flensers Point and Lords Bluff
Species               Average abundance (% cover)      % Contribution    Cumulative %
                  Canopy removal     Control
Flensers Point
 Seirococcus axillaris       1.16          66.34     13.19         13.19
 Sargassum fallax         35.92          2.50     10.55         23.73
 Encrusting algae          7.50          54.34      7.87         31.60
 Sargassum vestitum         6.92          1.16      7.76         39.37
 Caulocystis cephalornithos     3.84          0.00      6.35         45.72

Lords Bluff
 Seirococcus axillaris       0.66          29.66     10.02         10.02
 Cystophora monoliformis      12.26          2.00      8.11         18.13
 Ecklonia radiata          0.00          12.84      8.03         26.16
 Cystophora retroflexa       10.66          1.42      8.02         34.18
 Phyllospora comosa         0.00          4.00      7.34         41.52
 Caulocystis cephalornithos     2.76          0.00      6.53         48.06
 Encrusting algae          3.00          30.08      6.14         54.20
The column ‘% Contribution’ quantifies the breakdown of the contributions from each species to the difference in
community structure between canopy removals and controls. Species were included in the table if they
contributed to >5% of the difference in community structure. The analysis does not include plots where the
canopy was removed in winter because the total cover of foliose algae in this treatment was still significantly
lower than in controls by November 2001.

  We used the SIMPER routine (Carr and Clarke, 1994) to identify the species
contributing to these differences in community structure (note that the analysis did not
include treatments in which the canopy was removed in winter 2000, since total foliose
algal cover had not recovered to that in the control plots by November 2001 at either site;
see Table 3). The species contributing to the observed differences (Table 5) were found to
strongly reflect treatment effects described earlier for canopy-forming algae (see results in
Section 4.2). At Flensers Point, of the five macroalgal groups observed to contribute >5%
to the difference between treatments, four were the canopy-forming algae that proliferated
in response to the initial canopy removal. The remaining group, encrusting algae,
contributed 7.60% to the difference between treatments due to the high percentage cover
in control relative to canopy removal plots. At Lords Bluff, lack of recovery of species
dominating control areas (i.e. S. axillaris, E. radiata, P. comosa) and an increase in cover
of C. retroflexa and C. monoliformis in canopy removal plots were the main contributors
to the differences observed between treatments (Table 5).


4. Discussion

4.1. U. pinnatifida: opportunist or super competitor?

  Patterns of abundance of U. pinnatifida observed in this study demonstrate clearly that
disturbance resulting in removal of the native algal canopy is a critical step in the process
82       J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

leading to establishment. The results indicate that microscopic U. pinnatifida gameto-
phytes and/or sporophytes were dispersed throughout the native algal assemblages at both
sites during the study period. These microscopic phases responded opportunistically to
the artificial disturbance of canopy removal at both sites, and to the natural decline of the
E. radiata and P. comosa canopy at Lords Bluff in 2001 (Valentine and Johnson,
unpublished data).
  Given that high densities of U. pinnatifida sporophytes recruited soon after disturbance
to the canopy, the density of microscopic gametophytes present on the reef must have been
sufficiently high to enable fertilisation. Clearly, there is no evidence to suggest that U.
pinnatifida is capable of displacing native algal species through direct competition. A
similar response to canopy removal has been observed for the introduced seaweed
Sargassum muticum in northern Spain (Andrew and Viejo, 1998). In the present study,
two lines of evidence suggest that it is competition for light, rather than for space, that is
the major barrier to invasion. Firstly, U. pinnatifida recruited most strongly to plots where
canopy removals were conducted 4 months after the period of spore release, just prior to
the period of development of the macroscopic sporophyte (i.e. winter 2000). This
demonstrates that the native canopy does not represent a physical barrier preventing
spores from reaching the reef. Secondly, under the native algal canopy there was ample
availability of hard substratum suitable for attachment of U. pinnatifida propagules and
development of sporophytes given that cover of understorey species was generally less
than 20%.
  In relation to the supply of U. pinnatifida propagules, it is also important to consider the
lack of any effect associated with the ‘‘spore enhancement’’ treatment. The most likely
explanation for this result is that high densities of U. pinnatifida propagules had reached
the reef via natural dispersal, so that the additional spores associated with the treatment
had negligible effects on subsequent sporophyte density. An alternative explanation is that
the treatment was unsuccessful in delivering high numbers of viable propagules to the reef.
A problem of this nature might arise if the handling process had a detrimental impact on
source plants, or if spores were released but were carried away from experimental plots by
currents or surge. We consider this unlikely, however, given that a similar technique has
been used previously to successfully seed U. pinnatifida (Saito, 1975).
  The higher levels of U. pinnatifida recruitment observed in November 2000 in plots
where the canopy was removed immediately prior to the sporophyte growth period (winter
2000), compared to canopy removals 6 months earlier during the period of spore release
(spring 1999), raise two possibilities. There may have been higher survivorship of U.
pinnatifida gametophytes and/or microscopic sporophytes beneath the algal canopy than in
the cleared areas, or increased competition of developing U. pinnatifida sporophytes with
native algae that also responded to the spring 1999 canopy removal. In plots where canopy
removals were conducted in spring 1999, native algae had a 6-month window of
development before commencement of the growth phase of the annual U. pinnatifida
sporophyte generation. Proliferation of native species inhibiting the establishment of an
introduced species has been demonstrated previously in experimental manipulations
involving S. muticum (Deysher and Norton, 1982).
  These observations raise key questions relating to dispersal of spores and longevity of
the gametophyte stage in U. pinnatifida. Since there were no macroscopic U. pinnatifida
        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90  83

plants within the study areas at the beginning of the study, dispersal of spores from nearby
plants over distances of at least several hundreds of metres must have occurred (the site at
Flensers Point was f 1 km, and Lords Bluff was f 0.2 km, from the nearest stand of U.
pinnatifida). Recent work conducted in New Zealand has suggested that U. pinnatifida
possesses multiple strategies for natural dispersal. Laboratory experiments and field
observations of spore dispersal confirmed that while spore dispersal is likely to be
important for short-range dispersal (tens of metres), drifting sporophylls or fragments
enable dispersal in the scale of hundreds of metres to kilometres (Forrest et al., 2000). Drift
plants with intact sporophylls are commonly observed throughout the Mercury Passage.
Similar multiple dispersal strategies have been described for S. muticum and it is thought
that they may provide a mechanism to utilize the advantages of both long- and short-
distance dispersal (Andrew and Viejo, 1998; Deysher and Norton, 1982; Kendrick and
Walker, 1991).
  The longevity of the U. pinnatifida gametophyte generation is also a critical question
for managers. While analogies between gametophytes and seed banks in terrestrial plants
have been proposed (Hoffman and Santelics, 1991), there is no experimental evidence of
the phenomenon. Gametophytes of the perennial kelps Macrocystis pyrifera and Pter-
ygophora californica in Southern California appear to live for < 4 weeks, while for the
annual kelp Desmarestia ligulata dormancy of up to 3 –4 months has been observed (Reed
et al., 1997). If U. pinnatifida gametophytes have similar properties to D. ligulata,
disturbance would need to occur during this short period of gametophyte viability for U.
pinnatifida sporophytes to establish. Alternatively, if gametophytes are capable of
surviving for more than 1 year then it is possible that there could be an accumulation
of these stages over successive years. In this scenario, the timing of disturbance would be
less important since there would be a high likelihood that viable gametophytes would be
present in any particular year. In the Mercury Passage, our experiments indicate that the
longevity of gametophytes and/or microscopic sporophytes is at least 4 –5 months.
  The opportunistic nature of U. pinnatifida observed in this study is also characteristic of
other annual canopy-forming algae from the North American coast. These include the
annual laminarian kelps Alaria fistulosa and Nereocystis luetkeana and the annual brown
alga D. ligulata. These species appear unable to invade established kelp beds, but colonize
rapidly when kelp canopies are removed (Vadas, 1972; Duggins, 1980; Reed and Foster,
1984; Edwards, 1998). The establishment of D. ligulata following severe storms can
inhibit recruitment of other kelps, often causing local or patchy delays in kelp recovery
(Dayton et al., 1992). It could be expected that U. pinnatifida establishment may cause
similar delays in the establishment of native canopy-forming species. It should be noted
that there is no native annual canopy-forming algal species in temperate waters in
Australia.

4.2. Maintenance of U. pinnatifida stands post-establishment

  Critical to understanding its invasion dynamics and defining the threat it poses is
whether continued disturbance is required for U. pinnatifida to maintain persistent
populations. While disturbance may be a requirement for its establishment, it does not
necessarily follow that continued disturbance is required for U. pinnatifida populations to
84       J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

persist. For example, on the Atlantic Coast of North America, disturbance to native kelps
either due to destructive urchin grazing or infestation by an epiphyte (Membranipora
membranacea) facilitates establishment of the introduced alga Codium fragile subsp.
tomentosoides. Once established, dense stands of C. fragile subsp. tomentosoides appear
to inhibit kelp recruitment in the absence of continued disturbance, eventually displacing it
(Chapman et al., 2002). Research associated with terrestrial plant invasions also indicates
that persistence may occur in the absence of continued disturbance if an introduced species
changes the disturbance regime to favour its own reproduction, or if there are no species-
specific herbivores or pathogens (Luken, 1997).
  In the present study, U. pinnatifida declined in the second season following canopy
removal, corresponding with the recovery of native canopy-forming species. These results
suggest that, on the east coast of Tasmania, continued disturbance is required to maintain
dense stands of U. pinnatifida, although this conclusion should be viewed with caution
given that only two seasons of U. pinnatifida growth were observed. Further research
should specifically address the ongoing maintenance of dense U. pinnatifida stands after
they establish.

4.3. Recovery of native canopy-forming species following disturbance

  The decline in the abundance of U. pinnatifida after its initial establishment is most
likely explained by recovery of native species, in particular canopy-forming brown algae.
However, while the native species that recruited to cleared areas (predominately Cys-
tophora and Sargassum species) are ostensibly competitors of U. pinnatifida, they were
markedly different to the canopy species dominating control plots. A possible explanation
for differences in the long-established and newly developed canopies of native algae is the
timing of disturbance. The availability of propagules is known to determine early
succession in other algal assemblages (Foster, 1975; Emerson and Zedler, 1978; Dayton
et al., 1984; Kim and DeWreede, 1996), but unfortunately the phenology of the majority of
the native canopy-forming species observed in this study remains poorly understood. We
note, however, that while canopy manipulations were 6 months apart, the species
composition of the resultant canopy was similar for both seasons of canopy removal, at
both sites. Therefore it appears likely that the timing of canopy removal had only a minor
influence on native algal succession and that species which successfully colonized cleared
patches were opportunistic and may represent the initial stages of algal succession.
Interestingly, spatial patchiness in algal community composition at scales of 102 m is a
feature of Mercury Passage, possibly reflecting patches at varying stages of algal
succession.
  Comparison of similar experiments conducted elsewhere reveals that patterns of
recovery of canopy-forming species vary substantially. Similar to our results, removal of
a canopy of E. radiata in Western Australia realised a shift in dominance from E.
radiata to Sargassum spp. (Kirkman, 1981). In contrast, canopy removal in E. radiata
forests on the New South Wales coast facilitated the establishment of dense mats of turf
algae from the Zonaria/Lobophora complex, which persisted for up to 2 years for
canopy removals conducted in all seasons except winter (Kennelly, 1987a). Canopy
removals conducted in winter were colonized by both turf and E. radiata, with the kelp
        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90  85

rapidly developing a closed canopy, eventually resulting in the decline of turf (Kennelly,
1987a).
  Examples from the Northern Hemisphere also reveal a wide variation in response to
canopy disturbances. On the Atlantic Coast of North America, the canopy of Laminaria
longicuris can redevelop rapidly after disturbance, irrespective of timing, dominating both
early and late stages of community development (Johnson and Mann, 1988). In contrast,
on the Pacific coast of North America where a high diversity of canopy-forming species
are present, the canopy is often a mosaic of species depending on the frequency and
intensity of disturbance and proximity to reproductive plants (Dayton et al., 1984, 1992,
1999; Edwards, 1998). Given the patterns observed in response to our manipulations of the
canopy, we speculate that mechanisms similar to those maintaining patch dynamics on the
Pacific coast of North America forests also act on the east coast of Tasmania.

4.4. Canopy removal and the sediment matrix

  There are several mechanisms that may increase sediment deposition on the substratum
following canopy removal. First, the algal canopy represents a large surface area and
removing it allows sediment that would otherwise be trapped in the canopy to be deposited
on the substratum. Additionally, the sweeping motion of canopy algae on the substratum
caused by surge prevents sediment from accumulating on exposed surfaces of the reef
(Kennelly, 1989). This is consistent with observations of higher levels of sediment in the
centre of clearings compared with the edges (Kennelly and Underwood, 1993). It has also
been suggested that the presence of the kelp canopy prevents colonization by small
filamentous algae that facilitate accretion and consolidation of sediment (Melville and
Connell, 2001).
  Previous work has also observed an increase in sediment cover after canopy removal
(Kennelly, 1987a,b; Kennelly and Underwood, 1993; Melville and Connell, 2001). The
increased sediment levels observed in this study persisted throughout the study period in
plots from which the canopy was removed. This is in contrast to previous research where
persistence of the sediment layer after clearing was short-lived, decreasing to similar levels
as that in control areas within a few months (Kennelly, 1987a; Kennelly and Underwood,
1993).
  Sediment accumulation is a potentially important process in the ecology of rocky reefs
for a number of reasons. Sediment burial and scour may affect algal communities by
removing whole organisms, by physically preventing settlement of propagules on stable
substrata, or by limiting newly settled propagules by reducing inputs of light and oxygen
(Airoldi et al., 1995). Experiments have shown recruitment of some algal species to be
negatively affected by sediment deposition (Devinny and Volse, 1978; Kendrick, 1991;
Umar et al., 1998). It is possible that the significant increase in sediment levels observed in
canopy removal plots might influence the response of the algal community. However,
despite the increase and persistence of sediment following canopy removal, both U.
pinnatifida and some native species were able to recruit to these patches. This suggests
that these particular species can tolerate a degree of sediment stress. Increased sediment
may explain the lack of recovery of several of the native canopy-forming species, which
may be more sensitive to sediment stress. Notably, at other sites at Lords Bluff where
86       J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

sediment accumulation occurs on a large spatial scale associated with sea urchin ‘barrens’,
native algae did not recover over a 2-year period in areas where both urchins and U.
pinnatifida were removed (Valentine and Johnson, unpublished data).
  A feature of canopy removal areas at both sites was the increased abundance of C.
monoliformis relative to controls. C. monoliformis is known to grow in a variety of
stressed habitats, including areas subject to sediment stress, while apparently being
outcompeted in more favourable habitats (Edgar, 1984). In South Australia, C. mono-
liformis is abundant on sand scoured reefs including those covered by several centimetres
of sediment (Shepherd and Wommersley, 1981).

4.5. Destructive sea urchin grazing: an important source of disturbance?

  While we have shown that disturbance is necessary for successful establishment of U.
pinnatifida at high densities, an important question is to identify the natural disturbance(s)
facilitating U. pinnatifida establishment. Within our study area, destructive grazing by the
sea urchin H. erythrogramma is the most widespread form of disturbance to native algae,
and in Mercury Passage the only large monospecific stands of U. pinnatifida are
associated with urchin barrens (Johnson, unpublished data). While H. erythrogramma
can feed on U. pinnatifida, the recruitment and growth rates of the kelp clearly exceed the
urchins’ capacity to graze the plant at mean urchin densities of 6– 7 mÀ 2.
  Understanding the mechanisms of urchin barren formation by H. erythrogramma is
therefore an important step in understanding the process of U. pinnatifida invasion. In
temperate seas elsewhere in the world there is evidence supporting the link between
overfishing of sea urchin predators and barren formation (Estes and Palmisano, 1974;
Harrold and Reed, 1985; Watanabe and Harrold, 1991; Estes and Duggins, 1995; Vadas
and Steneck, 1995; Sala et al., 1998; Steneck, 1998; Shears and Babcock, 2002). Recent
work in Tasmania has indicated that the spiny lobster Jasus edwardsii is more important
than reef fishes as a predator of H. erythrogramma and, moreover, that reduced
abundances of lobsters as a result of fishing activity is sufficient to account for barren
formation (Pederson and Johnson, unpublished). It is therefore possible that overfishing of
sea urchin predators is the ultimate cause of reduced native algal cover in the Mercury
Passage that has facilitated the establishment of dense U. pinnatifida stands.

4.6. Conclusions

  This study demonstrates that disturbance to the native algal canopy facilitates the
establishment of Undaria pinnatifida sporophytes, while in the absence of disturbance
native algal communities resist invasion by this introduced kelp. The results suggest that
management of U. pinnatifida populations may be most effective by targeting the cause of
canopy disturbance, rather than the plant itself. Whilst it is not practical to manage natural
disturbances in subtidal habitats such as storm damage, if disturbance is linked to human
activity then options for control may exist. In our study area, the demonstrated links
between fishing of sea urchin predators, urchin barren formation and subsequent
establishment of U. pinnatifida provide a potential management opportunity to control
abundances of this introduced alga.
          J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90           87

Acknowledgements

  We thank the many dive volunteers who assisted with fieldwork operations, particularly
Edward Forbes, Miles Lawler and Liz Pietrzykowski . We are particularly grateful to Hugh
Pederson for valuable assistance in the field. This study was part of JPV’s PhD project
supported by an Australian Postgraduate Award. Research funds were provided by an
ARC grant awarded to CRJ. [RW]


References

Airoldi, L., 1998. Roles of disturbance, sediment stress, and substratum retention on spatial dominance in algal
  turf. Ecology 79, 2759 – 2770.
Airoldi, L., Rindi, F., Cinelli, F., 1995. Structure, seasonal dynamics and reproductive phenology of a filamentous
  turf assemblage on a sediment influenced, rocky subtidal shore. Bot. Mar. 38, 227 – 237.
Ambrose, R.F., Nelson, B.V., 1982. Inhibition of Giant Kelp recruitment by an introduced brown alga. Bot. Mar.
  15, 265 – 267.
Anderson, M.J., 2001. A new method for non-parametric multivariate analysis of variance. Aust. Ecol. 26,
  32 – 46.
Andrew, N.L., Viejo, R.M., 1998. Ecological limits to the invasion of Sargassum muticum in northern Spain.
  Aquat. Bot. 60, 251 – 263.
Bax, N., Carlton, J.T., Mathews-Amos, A., Haedrich, R.L., Howarth, F.G., Purcell, J.E., Rieser, A., Gray, A.,
  2001. The control of biological invasions in the world’s oceans. Conserv. Biol. 15, 1234 – 1246.
Byers, J.E., Reichard, S., Randall, J.M., Parker, I.M., Smith, C.S., Lonsdale, W.M., Atkinson, I.A.E., Seastedt,
  T.R., Williamson, M., Chornesky, E., Hayes, D., 2002. Directing research to reduce the impacts of non-
  indigenous species. Conserv. Biol. 16, 630 – 640.
Campbell, S.J., Burridge, T.R., 1998. Occurrence of Undaria pinnatifida Phaeophyta: Laminariales in Port
  Phillip Bay, Victoria, Australia. Mar. Freshw. Res. 49, 379 – 381.
Carlton, J.T., 1999. Scale and ecological consequences of biological invasions in the world’s oceans. In: Sand-
  lund, O.T., Schei, P.J., Viken, A. (Eds.), Invasive Species and Biodiversity Management. Kluwer Academic
  Publishers, Dordecht, pp. 195 – 212.
Carr, M.R., Clarke, K.R., 1994. Primer, vol. 4.0. Plymouth Marine Laboratories, National Environmental Re-
  search Council, Plymouth, UK.
Casa, G.N., Piriz, M.L., 1996. Surveys of Undaria pinnatifida Laminariales, Phaeophyta in Golfo Nuevo,
  Argentina. Hydrobiologia 326 – 327, 213 – 215.
Castric-Fey, A., Girard, A., L’Hardy-Halos, M.T., 1993. The distribution of Undaria pinnatifida Phaeophyceae,
  Laminariales on the coast of St Malo Brittany, France. Bot. Mar. 36, 351 – 358.
Chapman, A.S., Scheibling, R.E., Chapman, A.R.O., 2002. Species introductions and changes in the marine
  vegetation of Atlantic Canada. In: Claudi, R., Nantel, P., Muckle-Jeffs, E. (Eds.), Alien Invaders in Canada’s
  Waters, Wetlands And Forests. Canadian Forest Service, Ottawa, pp. 133 – 148.
Clayton, M.N., 1990. The adaptive significance of life history characters in selected orders of marine brown
  macroalgae. Aust. J. Ecol. 15, 439 – 452.
Curiel, D., Bellemo, G., Marzocchi, M., Scattolin, M., Parisi, G., 1998. Distribution of introduced Japanese
  macroalgae Undaria pinnatifida, Sargassum muticum Phaeophyta and Antithamnion pectinatum Rhodophyta
  in the Lagoon of Venice. Hydrobiologia 385, 17 – 22.
D’Antonio, C.M., Dudley, T.L., Mack, M., 1999. Disturbances and biological invasions: direct effects and
  feedbacks. In: Walker, L.R. (Ed.), Ecosystems of Disturbed Ground. Elsevier, Amsterdam, pp. 429 – 468.
Dayton, P.K., 1985. Ecology of kelp communities. Annu. Rev. Ecol. Syst. 16, 215 – 245.
Dayton, P.K., Currie, V., Gerrodette, T., Keller, B.D., Rosenthal, R., Ven Tresca, D., 1984. Patch dynamics and
  stability of some Californian kelp communities. Ecol. Monogr. 54, 253 – 289.
Dayton, P.K., Tegner, M.J., Parnell, P.E., Edwards, P.B., 1992. Temporal and spatial patterns of disturbance and
  recovery in a kelp forest community. Ecol. Monogr. 62, 421 – 445.
88         J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

Dayton, P.K., Tegner, M.J., Edwards, P.B., Riser, K.L., 1999. Temporal and spatial scales of kelp demography:
  the role of oceanographic climate. Ecol. Monogr. 69, 219 – 250.
Devinny, J.S., Volse, I.A., 1978. The effects of sediments on the development of Macrocystis pyrifera game-
  tophytes. Mar. Biol. 48, 343 – 348.
Deysher, L., Norton, T.A., 1982. Dispersal and colonisation in Sargassum muticum Yendo Fensholt. J. Exp. Mar.
  Biol. Ecol. 56, 179 – 196.
Draper, N., Smith, H., 1981. Applied Regression Analysis. Wiley, New York.
Duggins, D.O., 1980. Kelp beds and sea otters: an experimental approach. Ecology 61, 447 – 453.
Edgar, G.J., 1984. General features of the ecology and biogeography of Tasmanian subtidal rocky shore com-
  munities. Pap. Proc. R. Soc. Tasm. 118, 173 – 186.
Edwards, M.S., 1998. Effects of long-term kelp canopy exclusion on the abundance of the annual alga Desmar-
  estia ligulata Light F. J. Exp. Mar. Biol. Ecol. 228, 309 – 326.
Emerson, S.E., Zedler, J.B., 1978. Recolonisation of intertidal algae: an experimental study. Mar. Biol. 44,
  315 – 324.
Estes, J.A., Duggins, D.O., 1995. Sea otters and kelp forests in Alaska: generality and variation in a community
  ecological paradigm. Ecol. Monogr. 65, 75 – 100.
Estes, J.A., Palmisano, J.F., 1974. Sea otters: their role in structuring nearshore communities. Science 185,
  1058 – 1060.
Fletcher, R.L., Manfredi, R.L., 1995. The occurrence of Undaria pinnatifida Phaeophyceae, Laminariales on the
  South Coast of England. Bot. Mar. 38, 355 – 358.
Floc’h, J.Y., Pajot, R., Wallentinus, I., 1991. The Japanese brown alga Undaria pinnatifida on the
  coast of France and its possible establishment in European waters. J. Cons. Int. Explor. Mer. 47,
  379 – 390.
Floc’h, J., Pajot, R., Mouret, V., 1996. Undaria pinnatifida Laminariales, Phaeophyta 12 years after its intro-
  duction into the Atlantic Ocean. Hydrobiologia 326/327, 217 – 222.
Forrest, B.M., Brown, S.N., Taylor, M.D., Hurd, C.L., Hay, C.H., 2000. The role of natural dispersal mechanisms
  in the spread of Undaria pinnatifida Laminariales, Phaeophyceae. Phycologia 39, 547 – 553.
Foster, M.S., 1975. Algal succession in a Macrocystis pyrifera forest. Mar. Biol. 32, 313 – 329.
Grime, J.P., 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological
  and evolutionary theory. Am. Nat. 111, 1169 – 1194.
Harrold, C., Reed, D.C., 1985. Food availability, sea urchin grazing, and kelp forest community structure.
  Ecology 66, 1160 – 1169.
Hay, C.H., Luckens, P.A., 1987. The Asian kelp Undaria pinnatifida Phaeophyta: Laminariales found in a New
  Zealand harbour. N. Z. J. Bot. 25, 329 – 332.
Hiebert, R.D., 1997. Prioritizing invasive plants and planning for management. In: Luken, J.O., Thieret, J.W.
  (Eds.), Assessment and Management of Plant Invasions. Springer, New York, pp. 195 – 212.
Hobbs, R.J., Adkins, L., 1988. Effect of disturbance and nutrient addition on native and introduced annuals in
  plant communities in the Western Australian wheatbelt. Aust. J. Ecol. 13, 171 – 179.
Hobbs, R.J., Huenneke, L.F., 1992. Disturbance, diversity, and invasion: implications for conservation. Conserv.
  Biol. 6, 324 – 337.
Hoffman, A.J., Santelics, B., 1991. Banks of algal microscopic forms: hypotheses on their functioning and
  comparisons with seed banks. Mar. Ecol. Prog. Ser. 79, 185 – 194.
Johnson, C.R., Mann, K.H., 1988. Diversity, patterns of adaptation and stability of Novia Scotion kelp beds.
  Ecol. Monogr. 58, 129 – 154.
Johnson, C.R., Mann, K.H., 1993. Rapid succession in subtidal understorey seaweeds during recovery from
  overgrazing by sea urchins in eastern Canada. Bot. Mar. 36, 63 – 77.
Kendrick, G.A., 1991. Recruitment of coralline crusts and filamentous turf algae in the Galapagos archipelago:
  effect of simulated scour, erosion and accretion. J. Exp. Mar. Biol. Ecol. 147, 47 – 63.
Kendrick, G.A., Walker, D.I., 1991. Dispersal distances for propagules of Sargassum spinuligerum Sargassa-
  ceae, Phaeophyta measured directly by vital staining and venturi suction sampling. Mar. Ecol. Prog. Ser. 79,
  133 – 138.
Kennelly, S.J., 1987a. Physical disturbances in an Australian kelp community: I. Temporal effects. Mar. Ecol.
  Prog. Ser. 40, 145 – 153.
          J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90           89

Kennelly, S.J., 1987b. Physical disturbances in an Australian kelp community: II. Effects on understorey due to
  differences in kelp cover. Mar. Ecol. Prog. Ser. 40, 155 – 165.
Kennelly, S.J., 1989. Effects of kelp canopies on understorey species due to shade and scour. Mar. Ecol. Prog.
  Ser. 50, 215 – 224.
Kennelly, S.J., Underwood, A.J., 1993. Geographic consistencies of effects of experimental physical disturbance
  on understorey species in sublittoral kelp forests in central New South Wales. J. Exp. Mar. Biol. Ecol. 168,
  35 – 58.
Kim, J.H., DeWreede, R.E., 1996. Effects of patch size and season of disturbance on algal patch recovery in a
  rocky intertidal community. Mar. Ecol. Prog. Ser. 133, 217 – 228.
Kirkman, H., 1981. The first year in the life history and the survival of the juvenile marine macrophyte, Ecklonia
  radiata (Turn.) J. Agardh. J. Exp. Mar. Biol. Ecol. 55, 243 – 254.
Lodge, D.M., 1993. Biological invasions: lessons for ecology. Trends Ecol. Evol. 8, 133 – 137.
Luken, J.O., 1997. Management of plant invasions: implicating ecological succession. In: Luken, J.O., Thieret,
  J.W. (Eds.), Assessment and Management of Plant Invasions. Springer, New York, pp. 133 – 144.
Melville, A.J., Connell, S.D., 2001. Experimental effects of kelp canopies on subtidal coralline algae. Aust. Ecol.
  26, 102 – 108.
Moyle, P.B., Light, T., 1996. Fish invasions in California: do abiotic factors determine success? Ecology 77,
  1666 – 1670.
Nichols, F.H., Thompson, J.K., Schemel, L., 1990. Remarkable invasion of San Francisco Bay California, USA
  by the Asian clam Potamocorbula amurensis: II. Displacement of a former community. Mar. Ecol. Prog. Ser.
  66, 95 – 101.
Perez, R., Lee, J.Y., Juge, C., 1981. Undaria pinnatifida on the French coasts. Cultivation method. Biochem-
  ical composition of the sporophyte and gametophyte. In: Stadler, T. (Ed.), Algal Biotechnology, France, pp.
  315 – 327.
Reed, D.C., Foster, M.S., 1984. The effects of canopy shading on algal recruitment and growth in a Giant Kelp
  forest. Ecology 65, 937 – 948.
Reed, D.C., Anderson, T.W., Ebeling, A.W., Anghera, M., 1997. The role of reproductive synchrony in the
  colonization potential of kelp. Ecology 78, 2443 – 2457.
Reusch, T.B.H., Williams, S.L., 1999. Macrophyte canopy structure and the success of an invasive bivalve. Oikos
  84, 398 – 416.
Rueness, J., 1989. Sargassum muticum and other introduced Japanese macroalgae: biological pollution of Euro-
  pean coasts. Mar. Pollut. Bull. 20, 173 – 176.
Saito, Y., 1975. Undaria. In: Tokida, J., Hirose, H. (Eds.), Advance of Phycology in Japan. Junk, The Hague,
  pp. 304 – 320.
Sala, E., Boudouresque, C.F., Harmelin-Vivien, M., 1998. Fishing, trophic cascades, and the structure of algal
  assemblages: evaluation of an old but untested paradigm. Oikos 82, 425 – 439.
Sanderson, J.C., 1990. A preliminary survey of the distribution of the introduced macroalga, Undaria pinnatifida
  Harvey Suringer on the East Coast of Tasmania, Australia. Bot. Mar. 33, 153 – 157.
Sanderson, J.C., 1997. Survey of Undaria pinnatifida in Tasmanian coastal waters, January-February 1997.
  Report to Tasmanian Department of Marine Resources.
Sanderson, J.C., Barrett, N., 1989. A Survey of the Distribution of the Introduced Japanese Macroalga Undaria
  Pinnatifida Harvey Suringer in Tasmania, December 1988. Department of Sea Fisheries, Tasmania.
Shears, N.T., Babcock, R.C., 2002. Marine reserves demonstrate top-down control of community structure on
  temperate reefs. Oecologia 132, 131 – 142.
Shepherd, S.A., Wommersley, H.B.S., 1981. The algal and seagrass ecology of Waterloo Bay, South Australia.
  Aquat. Bot. 11, 305 – 371.
Steneck, R.S., 1998. Human influences on coastal ecosystems: does overfishing create trophic cascades? Trends
  Ecol. Evol. 13, 429 – 430.
Umar, M.J., McCook, L.J., Price, I.R., 1998. Effects of sediment deposition on the seaweed Sargassum on a
  fringing coral reef. Coral Reefs 17, 169 – 177.
Vadas, R.L., 1972. Ecological implications of culture studies on Nereocystis leutkeana. J. Phycol. 8,
  196 – 203.
Vadas, R.L., Steneck, R.S., 1995. Overfishing and inferences in kelp-sea urchin interactions. In: Skjoldal, H.R.,
90        J.P. Valentine, C.R. Johnson / J. Exp. Mar. Biol. Ecol. 295 (2003) 63–90

 Hopkins, C., Erickstad, K.E., Leinnas, H.P. (Eds.), Ecology of Fjords and Coastal Waters. Elsevier, Amster-
 dam, pp. 509 – 524.
Watanabe, J.M., Harrold, C., 1991. Destructive grazing by sea urchins Stronglyocentrotus spp. in a central
 Californian kelp forest: potential roles of recruitment, depth, and predation. Mar. Ecol. Prog. Ser. 71,
 125 – 141.