Shears and Babcock 02
Oecologia (2002) 132:131–142
DOI 10.1007/s00442-002-0920-x
COMMUNITY ECOLOGY
Nick T. Shears · Russell C. Babcock
Marine reserves demonstrate top-down control
of community structure on temperate reefs
Received: 9 October 2001 / Accepted: 6 March 2002 / Published online: 8 May 2002
© Springer-Verlag 2002
Abstract Replicated ecological studies in marine reserves indirect effects of fishing and re-establish community-
and associated unprotected areas are valuable in examin- level trophic cascades.
ing top-down impacts on communities and the ecosys-
tem-level effects of fishing. We carried out experimental Keywords Kelp communities · Marine protected areas ·
studies in two temperate marine reserves to examine Northeastern New Zealand · Trophic cascades ·
these top-down influences on shallow subtidal reef com- Urchin predation
munities in northeastern New Zealand. Both reserves
examined are known to support high densities of preda-
tors and tethering experiments showed that the chance of Introduction
predation on the dominant sea urchin, Evechinus chlorot-
icus, within both reserves was approximately 7 times Top-down regulation of biological communities has been
higher relative to outside. Predation was most intense on a focal point in ecological theory (Hairston et al. 1960).
the smallest size class (30–40 mm) of tethered urchins, This is ironic, given the efficiency with which humans
the size at which urchins cease to exhibit cryptic behav- have themselves harvested the large-bodied animals
iour. A high proportion of predation on large urchins which may be responsible for the top-down control of
could be attributed to the spiny lobster, Jasus edwardsii. ecosystems, in many cases to extinction (MacPhee
Predation on the smaller classes was probably by both 1999). Examples of top-down regulation or “trophic
lobsters and predatory fish, predominantly the sparid cascades” (see Polis et al. 2000) are increasingly being
Pagrus auratus. The density of adult Evechinus actively identified in a range of terrestrial (Schmitz et al. 2000),
grazing the substratum in the urchin barrens habitat was freshwater (Brett and Goldman 1996) and marine (Sala
found to be significantly lower at marine reserve sites et al. 1998; Pinnegar et al. 2000; Tegner and Dayton
(2.2±0.3 m–2) relative to non-reserve sites (5.5±0.4 m–2). 2000) ecosystems. In the marine environment where
There was no difference in the density of cryptic juve- many fisheries have had to resort to harvesting at lower
niles between reserve and non-reserve sites. Reserve levels of the food chain (Botsford et al. 1997; Pauly et al.
populations were more bimodal, with urchins between 1998), the impacts of fishing on trophic organisation and
40 and 55 mm occurring at very low numbers. Experi- function are substantial [reviewed in Jennings and Kaiser
mental removal of Evechinus from the urchin barrens (1998)]. Removal of top predators has resulted in the
habitat over 12 months lead to a change from a crustose loss of lower-level interactions and consequently many
coralline algal habitat to a macroalgal dominated habitat. trophic cascades have been lost (Pace et al. 1999). Our
Such macroalgal habitats were found to be more exten- ability to understand, manage or restore natural systems
sive in both reserves, where urchin densities were lower, is therefore compromised by our inability to differentiate
relative to the adjacent unprotected areas that were domi- anthropogenic impacts from the “natural” dynamics of
nated by urchin barrens. The patterns observed provide systems (Dayton et al. 1998).
evidence for a top-down role of predators in structuring Trophic cascades are defined as predatory interactions
shallow reef communities in northeastern New Zealand involving three or more trophic levels, whereby primary
and demonstrate how marine reserves can reverse the carnivores indirectly increase plant abundance by sup-
pressing herbivores (Menge 1995). In many subtidal reef
N.T. Shears (✉) · R.C. Babcock systems throughout the world, a reduction in algal forests
Leigh Marine Laboratory, University of Auckland,
P.O. Box 349, Warkworth, New Zealand and an increase in urchin barrens (areas dominated by
e-mail: n.shears@auckland.ac.nz crustose corallines where the grazing activity of sea
Tel.: +64-9-4226111, Fax: +64-9-4226113 urchins has removed all large macroalgae), have been
132
linked to fisheries-related declines in urchin predators For temperate systems there are few examples of the
(Wharton and Mann 1981; Estes and Duggins 1995; use of marine reserves to examine the trophic effects of
Vadas and Steneck 1995; Sala et al. 1998). However, fishing in subtidal kelp communities. In Australia, Edgar
good empirical examples supporting the existence of and Barrett (1999) found an increase in the density of
such trophic effects are generally lacking (Scheibling large fish and lobsters and an increased mean size of aba-
1996). The best known example is that of the role of sea lone in a Tasmanian marine reserve after 7 years of pro-
otters in structuring kelp communities in the northeastern tection, relative to an associated unprotected area. They
Pacific [reviewed by Pinnegar et al. (2000) and Tegner also found some changes in algal assemblages; however,
and Dayton (2000)]. Where sea otters are abundant, her- the cause of these changes was unknown and trophic cas-
bivorous sea urchins are rare and kelp dominates, whereas cade effects were not inferred to be present. The strongest
where otters are absent urchins are abundant and kelp evidence for a key role of predators in controlling subti-
rare (Estes and Duggins 1995). Recent declines in otter dal reef communities in the southern hemisphere is from
numbers in Alaska have been related to an observed two New Zealand marine reserves (Leigh Marine Reserve
increase in killer whale attacks on otters (Estes et al. 1998) and Tawharanui Marine Park) where there has been a
adding another level to this trophic cascade. In some decline in urchin densities and an associated change from
areas where sea otters do not occur, fish and lobsters have urchin barrens to kelp over a 20-year period (Babcock et
been implicated as important predators of urchins [e.g. al. 1999). The density and size of the dominant urchin
southern Califonia (Cowen 1983; Tegner and Levin predators, the snapper Pagrus auratus (Sparidae), blue cod
1983) and the northwestern Atlantic (Bernstein et al. Parapercis colias (Pinguipedidae) and the spiny lobster
1981; Breen and Mann 1976; Wharton and Mann 1981)]. Jasus edwardsii (Palinuridae), are considerably higher in
While the destruction of kelp beds by sea urchins in these these reserves than in adjacent fished areas (Kelly et al.
areas has been linked to overfishing of both lobsters and 2000; Willis et al. 2000; Willis 2001). Both snapper and
fish, the existence of a direct causal linkage has received spiny lobster are heavily targeted by commercial and rec-
much debate (Scheibling 1996). For kelp communities in reational fisherman around New Zealand, and Babcock et
the southern hemisphere it has been widely accepted that al. (1999) suggest that this has ecosystem-level effects,
the absence of a sea otter analogue results in a simpler indirectly resulting in large-scale reduction of macroalgal
two-tiered system with no top-down control of urchins habitats and subsequently benthic primary productivity.
(Estes and Steinberg 1988; Steinberg et al. 1995). While there is strong circumstantial evidence for a
Marine reserves provide a new opportunity for testing topdown effect, experimental evidence supporting a key
the top-down impact of predators and demonstrating predatory role is generally lacking (reviewed in Schiel
indirectly the ecosystem-level effects of fishing. They 1990). Differences in urchin demography, behaviour and
function as an experimental tool where large-scale eco- morphology, and also a higher loss of transplanted
system manipulations are carried out by preventing fish- urchins in the Leigh marine reserve compared to outside
ing and subsequently elevating predator densities. The have been inferred to be due to higher levels of predation
treatments can be viewed as either with or without by Cole and Keuskamp (1998). The subtidal reef commu-
humans as the top predator, or as without or with “natural” nities in northeastern New Zealand are suited to the
predators. This enables comparisons of trophic structure occurrence of community-level cascades (Polis et al.
and further experimental manipulations to be made be- 2000) with a simple trophic structure, discrete habitats
tween reserve and non-protected areas. On coral reefs in and low species diversity. The sea urchin, Evechinus
East Africa, marine reserves have been used in this way. chloroticus is the dominant grazer (Andrew 1988), and
Predatory fish densities have been found to be higher, through its grazing activity can form urchin barrens habitat
urchin densities lower and predation on urchins higher, at depths between approximately 3 and 10 m (Ayling 1981;
in Kenyan marine reserves relative to unprotected areas Choat and Schiel 1982).
(McClanahan and Shafir 1990). Subsequently protected The aim of this study was to demonstrate the indirect
reefs had a higher species diversity and topographic effects of fishing on lower trophic levels by experimen-
complexity, with higher cover of hard coral and calcareous tally examining the top-down role of predators in
algae than unprotected areas. In the Mediterranean Sea explaining the habitat change documented in marine
an expansion of urchin barrens into areas previously reserves in northeastern New Zealand. This was done by:
occupied by erect algae has been linked to overfishing of
urchin predators (Sala and Zabala 1996. Studies utilising 1. An urchin-tethering experiment to test whether rela-
marine reserves in the Mediterranean have shown that tive predation levels on urchins were higher at marine
predatory fish are an important determinant in control- reserve sites and to determine the sources of preda-
ling urchin populations [reviewed by Sala et al. (1998)]. tion.
However, there has not yet been any decline in the extent 2. Comparing the density and demography of urchins in
of urchin barrens in these protected areas. Other factors the urchin barrens habitat at multiple sites in two
such as recruitment, pollution, disease, large-scale reserve and two non-reserve areas.
oceanographic events, urchin harvesting, food subsidies 3. Experimental removal of urchins to test whether the
and availability of shelters may also be important in observed habitat changes in the Leigh reserve were
controlling algal assemblage structure (Sala et al. 1998). consistent with a reduction in urchin populations.
133
that tether-related mortality could be reduced by holding the
urchins in the laboratory for a week prior to experimentation. This
procedure also allowed the urchins to heal, minimising the potential
effects of coelomic fluid leakage on predation (McClanahan and
Muthiga 1989).
Tethering also provided information on the source of predation
through direct observation or from examination of urchin-test
remains. Slow-moving predators such as the starfish, Coscinasterias
muricata, and the gastropod, Charonia lampax were often seen
feeding on the urchin or remained nearby. From trial experiments
we were able to classify the source of predation into the following
categories: (1) unknown (urchin missing with nylon loop still
intact demonstrating urchin had been broken off), (2) lobster (test
had characteristic pattern of lobster predation which involves a
large opening around the Aristotle’s lantern), (3) Coscinasterias
(test intact with patches of freshly stripped spines) and (4)
Charonia (test intact and mucous covered).
The tethering experiments were carried out at three reserve and
three non-reserve sites (Fig. 1); first at Leigh (4 August 1999) and
then repeated at Tawharanui (19 August 1999). Thirty urchins, of
three different size classes (n=10), were tethered at each site and
their survival monitored for 10 days. The three size classes used
for the experiments were: 35–40 mm, representing the size where
urchins move from a sheltered to an exposed habit (Andrew and
Choat 1982; Cole and Keuskamp 1998), 55–60 mm and
75–80 mm, representing the dominant adult size class outside and
Fig. 1 Location of study sites in the Cape Rodney to Okakari Point inside the Leigh reserve, respectively (Cole and Keuskamp 1998).
Marine reserve (CROP) at Leigh and Tawharanui Marine Park Experimental urchins were collected from non-reserve sites at
(TMP). Circles indicate sites where the predation experiment was Leigh where all size classes can be found openly grazing the sub-
carried out. Inset shows general location of study area on New stratum. Urchins were positioned in a 10×10-m2 plot located in the
Zealand’s North Island urchin barrens habitat adjacent to the kelp forest border. Urchins
were tethered on 25-cm monofilament traces, attached to masonry
nails that had previously been embedded in the substratum at
4. Comparing the distribution of macroalgal communities random coordinates. It was important that urchins were attached
among reef habitats between reserve and non-reserve without drawing the attention of diver-positive predatory fish at
marine reserve sites (Cole 1994; Cole and Keuskamp 1998). This
areas. was done by keeping the urchins concealed while one diver created
a disturbance nearby. There were no instances of fish predation on
recently tethered urchins. Daily monitoring enabled detection and
Materials and methods replacement of urchins that appeared to be dying as a result of
tethering. In each experiment only four out of a total of 180 tethered
Study area urchins died as a result of tethering.
Differences in the survival of urchins after 10 days were analysed
This study was carried out at sites located in two marine reserves using a generalised linear mixed model (GLMMIX). The model
and at adjacent unprotected sites in northeastern New Zealand was back-fitted to a binomial distribution using residual (restricted)
(Fig. 1). The two reserves examined are completely no-take and maximum likelihood with the GLMMIX macro in SAS (Littell et
include New Zealand’s oldest marine reserve, the Cape Rodney to al. 1996). This technique was used in preference to ANOVA as
Okakari Point (Leigh) Marine Reserve (549 ha, established in survival data follow a binomial distribution. The factors Area
1976), and Tawharanui Marine Park (350 ha, established in 1982), (Leigh and Tawharanui), Status (reserve and non-reserve) and
8 km to the south. Both marine reserves are subject to similar Size (the three size classes) were treated as fixed effects and
environmental conditions and have extensive subtidal reef commu- Site(Area×Status) as a random effect.
nities typical of moderately exposed coasts in northeastern New
Zealand (Choat and Schiel 1982). Urchin density and size structure
Comparisons of urchin populations were made between reserve
Predation and non-reserve sites at both Leigh and Tawharanui (Fig. 1). Five
sites were sampled within the Leigh reserve and five outside during
Relative predation levels on Evechinus were compared between March and April 1998, while at Tawharanui four sites were sampled
marine reserves and adjacent fished areas using tethering experi- within the reserve and four outside in June 1998. Sites were
ments. Tethering is a simple technique, suited to sedentary benthic selected in areas where urchin barrens habitat was present. At each
organisms (Aronson et al. 2001), that has been used extensively on site ten 1-m2 quadrats were placed haphazardly within the urchin
coral reefs (McClanahan and Muthiga 1989; McClanahan et al. barrens at 4–6 m depth (below MLWS). Within each quadrat we
1999) and in the Mediterranean (Sala and Zabala 1996) to test pre- measured the TD of each urchin using vernier callipers (±1 mm)
dation potential on sea urchins between protected and unprotected and noted whether urchins were located in a crevice (cryptic) or
reefs. were openly grazing the substratum (exposed). In addition, the
The tethering technique involved inserting a hypodermic needle percent cover of dominant encrusting algal forms was visually
(1.2 mm×38 mm) through the dorsal and ventral surface of the estimated to determine if any differences occurred between reserve
urchins test, as far away from the oral-aboral axis as possible and non-reserve sites.
(McClanahan and Muthiga 1989). Nylon monofilament was then Urchin counts were analysed using GLMMIX. A Poisson distri-
threaded through the needle and tied-off. Laboratory trials found bution was used as count data seldom fit the assumptions of normality
100% survival of 80 tethered urchins [ranging in size from 25 to and homogeneity of variance. The factors Area and Status were
75 mm test diameter (TD)] after 10 days. Trials in the field found treated as fixed effects and Site(Area×Status) as a random effect.
134
Fig. 2 Survival of tethered
q
urchins at reserve (q ) and non-
reserve (q) areas. The mean
number of tethered urchins
surviving in each of the three
size classes is given for Leigh
and Tawharanui
Differences in size of exposed urchins between reserve and effect Plot(Treatment) and also for the auto-regressive error struc-
non-reserve areas were tested using mixed-model ANOVA with ture [AR(1)] to account for repeated measures. A binomial distri-
fixed factors Area and Status. Site was treated as a random factor bution with logit-link was used for percent cover data and a Poisson
and nested within Area and Status. Size data were tested for nor- distribution with log-link for count data.
mality using Shapiro-Wilk’s test. Significant interaction terms
were investigated using a multiple comparison (Tukey-Kramer) of
all possible combinations of the main effects. Distribution of urchin barrens habitat
To investigate whether urchin barrens were more abundant in re-
Urchin removal serves, the proportions of habitats were measured at 22 sites located
in and around both reserves using 1-m-wide strip transects (three at
A sea urchin-removal experiment was undertaken to investigate each site). Transects were run perpendicular to the shore from
the role Evechinus plays in maintaining the urchin barrens habitat MLWS to the reef edge or a maximum depth of 12 m. Both depth
and the response of algal communities to a reduction in urchin and habitat type were recorded every metre. Habitat type was
density. The experiment was carried out at Mathesons Bay (Fig. 1) recorded as one of the following categories, based on the density of
near the Leigh marine reserve on an area of reef with extensive plants within each 1-m2 area along the transect: (1) macroalgal hab-
urchin barrens habitat. The reef was dissected by sand-filled itat, >3 adult phaeophytes m–2 e.g. Ecklonia radiata or Carpophyllum
crevices, which form semi-isolated blocks of reef, allowing the flexuosum; (2) urchin barrens, >50% cover of crustose coralline
establishment of discrete experimental plots within a 500-m2 area algae; (3) shallow fucoid zone, >20% cover or 3 adult phaeophytes
of reef in the urchin barrens habitat. Six blocks of reef were selected, m–2 at depths <4 m; (4) turf habitat, >50% cover of turf forming
ranging in size from 10 to 20 m2, at a depth of 4–5 m. All urchins red or green algae with large phaeophytes <3 m–2.
were removed from three randomly selected blocks, the remaining The proportional cover of urchin barrens habitat within three
three were left as controls. The urchins were removed in January depth ranges (0–3, 4–6 and 7–9 m) was examined using GLMMIX
1998, with weekly re-clearances until January 1999. with a binomial distribution. Area, Status and Depth were treated as
The initial density of grazers and macroalgae in the experimental fixed factors, and Site(Area×Status) was treated as a random factor.
areas was estimated in five haphazardly placed 0.25-m2 quadrats.
The percent cover of encrusting algae (crustose coralline algae,
articulated coralline turf, filamentous algae and other encrusting Results
algae) was also measured by estimating the number of 10×10-cm
cells within the 0.25-m2 quadrats each algal type “filled” (Benedetti- Predation
Cecchi et al. 1996). Sampling was repeated monthly to determine
the response of the communities to manipulation. To test for dif- Predation on urchins was significantly higher at reserve
ferences in the dominant species between treatments and between
plots within treatments at the start of the experiment and over time sites than at non-reserve sites (F=9.44, P=0.0133), with
GLMMIX was used. Treatment and Time were set as fixed effects. the relative odds of predation being 6.9 times higher
Covariance parameter estimates were calculated for the random at reserve sites (Fig. 2, Table 1). This was consistent
135
Table 1 Summary statistics for reserve/non-reserve comparisons of urchin barrens and coralline turf (both binomial distribution).
with size of effect expressed as a ratio with 95% confidence limits For count data the ratio indicates the effect size whereas for
(CL). Likelihood ratios calculated by the SAS procedure generalised mortality and percent data the ratio indicates the relative odds
linear mixed model for mortality of tethered Evechinus (binomial [see Willis and Millar (2001) for explanation of interpreting relative
distribution), Evechinus density (Poisson distribution), the cover odds ratio]
Reserve Non-reserve Reserve:non- Upper Lower
mean SE mean SE reserve ratio 95% CL 95% CL
Predation on Evechinus (% mortality) 42.8 (9.0) 12.2 (3.4) 6.88 2.01 23.57
Exposed Evechinus density (m–2) 2.2 (0.3) 5.5 (0.4) 0.60 0.45 0.79
Exposed Evechinus mean size (mm) 69.8 (2.5) 57.3 (1.3) – – –
Cover of coralline turf (%) 29.3 (2.9) 12.6 (1.5) 1.80 0.90 3.60
Extent of urchin barrens (%) 14.8 (4.8) 41.4 (4.2) 0.17 0.07 0.41
Table 2 Source of predation on tethered urchins
Reserve Non-reserve
Size class 35 mm 55 mm 75 mm 35 mm 55 mm 75 mm
Number preyed 40 23 13 11 7 4
Proportion
Unknown 100.0 56.5 46.2 54.5 42.9 0.0
Lobster 0.0 43.3 46.2 0.0 0.0 0.0
Coscinasterias 0.0 0.0 7.7 45.5 42.9 50.0
Charonia 0.0 0.0 0.0 0.0 14.3 50.0
between both areas (F=0.41, P=0.5357). There was a
significant effect of size on predation (F=12.60,
P<0.0001), which was also consistent between areas and
between reserve and non-reserve sites. Predation occurred
on all size classes of tethered urchins, at both reserve and
non-reserve sites, but was highest on the smallest size
class (Fig. 2). The likelihood of predation on the small
and middle size-class urchins was 6.3 [95% confidence
limits (CL) 3.0–13.3] and 2.2 (CL 1.1–4.7) times greater,
respectively, than predation on the largest size class. The
odds of predation did not vary significantly between
reserve and non-reserve sites for each area (Z=1.52,
P=0.0639).
The fate of all small urchins (35 mm) preyed upon at
reserve sites was unknown as the tests were completely
removed from the tethers (Table 2). This could have
been due either to predation by fish, which completely
engulf the urchin, or by lobsters breaking up or removing
small urchins. At reserve sites approximately 45% of
preyed individuals in the larger size classes (55 and
75 mm) showed patterns of damage characteristic of
spiny lobster predation. No urchins showed signs of Fig. 3 Mean density of A exposed and B cryptic urchins, and
spiny lobster predation at non-reserve sites. In most C mean size of exposed urchins from quadrat sampling (n=10) at
cases mortality at non-reserve sites could be attributed q
all reserve (q ) and non-reserve sites (q)
to starfish (Coscinasterias muricata) or the gastropod,
Charonia lampax, both of which are slow-moving
predators. reserve sites for both areas (Fig. 3). The density of
exposed urchins (Fig. 3A) was significantly lower at
marine reserve sites (Tables 1, 3). Exposed urchins were
Urchin density and size structure 1.7 times more abundant overall at non-reserve sites
(Table 1). There was no difference in urchin density
Densities of Evechinus in the urchin-grazed habitat between areas (Leigh and Tawharanui) but there was a
varied widely between sites but were generally lower at significant interaction between Area and Status. This can
136
Table 3 Exposed urchin density statistics. Type 3 tests for counts Table 5 Urchin size statistics. Mixed-model ANOVA results for
of exposed urchins for fixed effects Area (Leigh and Tawharanui) exposed urchin size at Leigh and Tawharanui (Area), reserve and
and Status (reserve and non-reserve). Parameter estimates for the non-reserve site (Status)
random effect Site(Area×Status)
df Mean square F-value Pr>F
Fixed effects df F-value Pr>F
Area 1 3,343.2 5.58 0.0321
Status 1, 14 73.65 <0.0001 Area×Status 1 4,552.8 7.60 0.0147
Area 1, 14 2,21 0.1597 Status 1 19,374.0 32.48 <0.0001
Area×Status 1, 14 13.06 0.0028 Site(Area×Status) 14 726.0 7.33 <0.0001
Covariance parameter Estimate SE Z-value Pr Z
Site(Area×Status) 0.0023 0.0135 0.17 0.4312
The abundance of cryptic urchins (Fig. 3B) did not differ
significantly between reserve and non-reserve locations
Table 4 Cryptic urchin density statistics. Type 3 tests for count of
exposed urchins for fixed effects Area (Leigh and Tawharanui) or between Leigh and Tawharanui (Table 4), although
and Status (reserve and non-reserve). Parameter estimates for the there was significant site-level variability.
random effect (Area×Status) There was significant variation in the size of exposed
urchins between sites (Fig. 3C, Table 5). The mean size
Fixed effects df χ2 P
of exposed urchins was significantly larger at marine re-
Status 1, 14 0.37 0.5529 serve sites (Table 1) although the Area effect was signifi-
Area 1, 14 0.62 0.445 cant and there was a significant Area and Status interac-
Area×Status 1, 14 0.01 0.9274 tion. This can be explained by the larger effect of Status
Covariance parameter Estimate SE Z-value Pr Z at Leigh which results in a significant interaction be-
Site (Area×Status) 0.5194 0.2465 2.11 0.0175 tween Area and Status. There was no difference in size
between the non-reserve sites for each area [Tukey’s
honestly significant difference (HSD) P=0.8112], but for
be explained by examining the size of the effect of Status both areas there were significant differences between
between both areas; for Leigh densities were 3.5 (95% reserve and non-reserve sites (Tukey’s HSD P<0.0001).
CL 2.8–4.4) times higher at non-reserve sites while at There was also significant difference between reserve
Tawharanui densities were 1.7 (1.2–2.4) times higher. populations at Leigh and Tawharanui (Tukey’s HSD
Separate analysis for each area found a significant differ- P<0.0001).
ence in density between reserve and non-reserve sites for The population structure of Evechinus varied between
both Leigh (P<0.0001) and Tawharanui (P=0.0342). reserve and non-reserve sites (Fig. 4). Populations were
Fig. 4 Size frequency distribu-
tions of all Evechinus measured
during quadrat sampling at
each area. Shaded bars indicate
proportion of cryptic urchins
137
Fig. 5 The percent cover of Coralline turf from quadrat sampling
q
(n=10) at all reserve (q) and non-reserve sites (q). Means are given
for each site
more bimodal at reserve sites, with very low numbers of
urchins between 30 and 50 mm, and they generally
remained cryptic to a greater size. This pattern was
stronger in the Leigh marine reserve. Fig. 6A, B Response of encrusting and turfing algae to urchin
Quadrat sampling also revealed that, overall, the removal. The mean proportional cover of A crustose coralline
q
algae and B coralline turf in both control (q) and urchin removal
percent cover of coralline turf (Corallina officinalis) (q) plots following commencement of the experiment in January
was significantly higher at reserve sites (F1,9=14.18, 1998. J January, F February, M March, A April, M May,
P=0.0044) (Table 1, Fig. 5). The relative odds ratio was J June, J July, A August, S September, O October, N November,
1.8 times higher at marine reserve sites (Table 1). This D December
was consistent between areas (F1,9=0.50, P=0.4960) and
while the pattern was clearest at Leigh (Fig. 5) the effect
of reserve status was consistent between areas (F1,9=1.79, crustose coralline algae across all plots was due to a
P=0.2140). large settlement of filamentous algae at the start of the
experiment (Fig. 6). The change from crustose coralline
to articulated coralline algae occurred rapidly for the first
Urchin removal 4 months then remained stable throughout the winter. The
cover of crustose corallines and coralline turf varied
At the commencement of the experiment in January significantly over time (Table 6). While the overall effect
1998 Evechinus densities did not vary between treat- of treatment was not clear, the effect of urchin removal
ments (F1,4=0.06, P=0.8135). Densities of urchins ranged on coralline algae and coralline turf over time was
from 1.2 to 2.4 per 0.25 m2. Crustose coralline algae significant (Table 6).
(Lithothamnion and Lithophyllum spp.) were dominant, A number of brown algal species became established
covering 63–99% of the substratum. Articulated coralline in the urchin-removal plots (Fig. 7). In most cases these
turf was the other dominant encrusting form with cover species remained absent from control plots so differences
ranging between 0 and 35%. There was no significant between treatments could not be statistically tested. Only
difference in either crustose coralline algae (F1,4=0.36, Carpophyllum flexuosum occurred at sufficient densities
P=0.5789) or coralline turf (F1,4=0.63, P=0.4730) in both control and removal plots throughout the experi-
between treatments or between plots within treatments ment for statistical analysis (Fig. 7A). There was no
(Z=1.1, P=0.1367, Z=1.08, P=0.1411). Macroalgae effect of urchin removal on the density of C. flexuosum
were rare at the start of the experiment, with Carpophyllum (Table 6), the numbers remaining stable over time.
flexuosum, which is relatively resilient to urchin grazing Several large Ecklonia radiata sporophytes became
(Cole and Haggitt 2001), being the only conspicuous established (Fig. 7B) within the urchin removal plots
large brown seaweed (<1 per 0.25 m2). There was no while remaining absent in control areas. Survival of
significant difference in the number of C. flexuosum Ecklonia recruits was observed to be low as they appeared
between treatments (F1,4=0.52, P=0.5123) or between to be prime targets for any urchins which did immigrate
plots within treatments (Z=0.53, P=0.2966). Ecklonia into treatment plots. Total exclusion of urchins would
radiata was absent from all plots. probably have resulted in a more rapid response of
After 1 year the control plots remained as urchin barrens Ecklonia. Low numbers of two other species of
dominated by crustose coralline algae, while the urchin- large brown algae, Carpophyllum maschalocarpum and
removal plots had become dominated by coralline turf, Sargassum sinclairii also became established in the
with a mixture of large and small brown algae (a “mixed urchin-removal areas. Small brown algae showed a marked
algal habitat”). A temporary decrease in the cover of response to urchin removal. These included Halopteris
138
Table 6 Response of crustose coralline algae, coralline turf and
Carpophyllum flexuosum following urchin removal. Type 3 analysis
for the percent cover of crustose coralline algae and coralline turf,
and the number of C. flexuosum plants following urchin removal
for fixed effects Treatment and Time. Parameter estimates for the
random effect Plot(Treatment) and the repeated measures effect
[AR(1)]
Crustose coralline
Fixed effects df F-value Pr>F
Treatment 1, 4 7.43 0.0527
Time 9, 276 21.76 <0.0001
Treatment×Time 9, 276 4.57 <0.0001
Covariance parameter Estimate SE Z-value Pr Z
Plot(Treatment) 0.2247 0.1678 1.34 0.0903
AR(1) 0.3961 0.0650 6.10 <0.0001
Coralline turf
Fixed effects df F-value Pr>F
Treatment 1, 4 1.24 0.3278
Time 9, 276 24.74 <0.0001
Treatment×Time 9, 276 9.48 <0.0001
Covariance parameter Estimate SE Z-value Pr Z
Plot(Treatment) 0.8042 0.5754 1.40 0.0811
AR(1) 0.1938 0.0640 3.03 0.0024
Carpophyllum flexuosum
Fixed effects df F-value Pr>F
Treatment 1, 4 1.32 0.3141
Time 9, 276 1.20 0.2947
Treatment×Time 9, 276 1.36 0.2077
Covariance parameter Estimate SE Z-value Pr Z
Plot(Treatment) 1.0384 0.9185 1.13 0.1291
AR(1) –0.0660 0.0630 –1.05 0.2948
Fig. 7A–D Response of macroalgae to urchin removal. The mean
abundance of A Carpophyllum flexuosum, B Ecklonia radiata,
C Halopteris virgata and D small brown seasonal algae in both
q
control (q) and urchin-removal (q) plots following commencement Table 7 Statistics for the proportion of urchin barrens. Type 3
of the experiment in January 1998 analysis for fixed factors Status (reserve/non-reserve), Area (Leigh/
Tawharanui) and Depth (0–3, 4–6 and 7–9 m). Non-significant
interaction terms were removed from the model. Parameter estimates
virgata (Fig. 7C) and a number of short-lived seasonal given for the random effect Site(Area×Status)
species (Fig. 7D), including Dictyota sp. and Colp-
Fixed effects df F-value Pr>F
omenia sinuosa. A few red algal species such as Aspar-
agopsis armata and Champia novaezelandicae also re- Status 1, 19 15.24 0.0010
cruited into urchin-removal plots. Area 1, 19 0.00 0.9740
Within 6 months of completion of the experiment the Depth 2, 174 32.73 <0.001
urchin-removal plots had been heavily grazed and re- Covariance parameter Estimate SE Z-value Pr Z
verted to urchin barrens habitat, dominated by crustose Site(Area×Status) 0.8844 0.3715 2.38 0.0087
coralline algae. The only brown algae present were
stunted Carpophyllum flexuosum plants (personal obser-
vation). (Tables 1, 7). The relative odds ratio for the proportion
of urchin barrens at reserve vs. non-reserve sites was
0.2:1 (Table 1), or inversely, 5.9 times higher at non-
Distribution of urchin barrens habitat reserve sites. This pattern was consistent between both
areas but varied significantly with depth (Table 7).
The general distribution of habitats (Fig. 8) at non- Differences between reserve and non-reserve sites were
reserve sites is typical of northeastern New Zealand greatest in the shallow depth strata (0–3 m and 4–6 m)
with shallow fucoid assemblages, deep kelp forests where urchin barrens covered approximately 54% of
(Ecklonia radiata) and intermediate depths dominated available reef at non-reserve sites and only about 20%
by urchin barrens (Choat and Schiel 1982). However, at of the reef at reserve sites. The deeper strata (7–9 m
marine reserve sites all depth ranges were dominated by and 10–12 m) at both reserve and non-reserve sites
macroalgal habitats. The proportion of urchin barrens were dominated by macroalgal habitat, predominantly
habitat was significantly lower than at non-reserve sites Ecklonia forest.
139
Fig. 8 Mean percent cover of
major habitat types (shallow
fucoids, urchin barrens, turfing
algae and kelp forest), within
each depth range for reserve
and non-reserve sites at Leigh
and Tawharanui
Discussion MacDiarmid 1991). The actual proportion of predation
attributable to lobsters is likely to be higher than estimated,
New Zealand’s two oldest marine reserves at Leigh and as a spiny lobster may completely consume a tethered
Tawharanui support higher predator densities than similar urchin or take it back to shelter, in both cases leaving no
unprotected areas of coastline (Babcock et al. 1999; remains from which the source of predation could be
Kelly et al. 2000; Willis et al. 2000; Willis 2001). Snapper ascertained. In addition, lobster-predated tests of unteth-
are at least 5.8–8.7 times more abundant inside these ered urchins were found regularly at reserve sites during
reserves and spiny lobster 1.6–3.7 times more abundant, the study, demonstrating that spiny lobster were also
than in adjacent unprotected areas (Babcock et al. 1999). feeding on natural urchin populations. The highest levels
Relative rates of predation on sea urchins were found to of predation occurred on the 30- to 40-mm size class,
be higher in the reserves, the densities of actively grazing which are normally sheltered at reserve sites but exposed
urchins lower, and the cover of macroalgal forests higher at non-reserve sites. While the specific predators respon-
relative to adjacent unprotected areas. These patterns sible for attacks on this size class could not be identified
confirmed and extended previous results (Cole and they were likely to have been both lobsters, which pref-
Keuskamp 1998; Babcock et al. 1999) and provide erentially take smaller urchins (Andrew and MacDiarmid
experimental evidence for a top-down impact of predators 1991), and predatory fish. Gut content analysis (Babcock
on subtidal reef communities. The fact that these patterns et al. 1999) and visual observations (personal observa-
were found through comparisons of multiple sites inside tion) indicate that snapper and blue cod only feed on
and outside more than one marine reserve mean that it is small urchins (<50 mm). Cole and Keuskamp (1998)
highly unlikely that the effects were due to spatial varia- suggested that the higher loss of transplanted urchins of
tions in other factors such as nutrients or larval supply. this size in the Leigh reserve sites was due to predation
The spiny lobster, Jasus edwardsii, was found to be by fish.
an important predator of sea urchins at marine reserve Predation of tethered urchins at non-reserve sites
sites. At least 45% of predation on the larger size classes was attributed to a different suite of predators, mainly
of tethered urchins at reserve sites could be attributed to Coscinasterias and Charonia. These species are natural
lobsters. This is noteworthy, considering that spiny lob- predators of urchins (personal observation) but tethering
sters were not previously thought to forage in the urchin reduced the chance of urchins escaping from these slow-
barrens habitat (Andrew and Choat 1982; Andrew and moving predators. At reserve sites both of these predatory
140
species tend to occur at much lower densities than at densities over the 2 years prior to our study, or due to the
non-reserve sites (N. Shears, unpublished data), subse- fact that their sampling was carried out over a larger
quently there was only one instance of predation by depth range (5–10 m) and did not sample sites adjacent
Coscinasterias recorded at reserve sites. Similar patterns to the marine reserve. The smaller effect of reserve status
have been shown in other studies where different preda- seen at Tawharanui compared to Leigh may be due to
tors are important at fished sites where the primary pre- several factors; higher levels of poaching (personal
dators have been removed. In the Gulf of Maine, Vadas observation), younger reserve age and smaller reserve
and Steneck (1995) found high levels of fish predation size. The size structures of Evechinus populations we
on urchins at an offshore reef subject to low fishing pres- found were consistent with those found by Cole and
sure, while at heavily fished coastal sites predation on Keuskamp (1998). At reserve sites the exposed urchins
urchins was attributed to crabs. This was also suggested were larger and populations were more bimodal than at
to be at least in part an artefact of tethering. Likewise, in non-reserve sites. Our tethering experiment provides
the Mediterranean a predatory gastropod was an impor- support for the hypothesis that bimodality in Evechinus
tant predator at fished sites while at protected sites fish populations is related to higher size-specific predation on
were the dominant predators (Sala and Zabala 1996). juveniles moving from a cryptic to exposed lifestyle
Our urchin-tethering experiment provided strong evi- (Andrew and Choat 1982; Cole and Keuskamp 1998).
dence that the lower densities of urchins in both reserves Lower densities of urchins at protected sites com-
are due to relatively higher levels of predation within pared to fished sites has been linked to higher predator
marine reserves. densities in other marine systems (Sala and Zabala 1996;
Previous experimental studies carried out in the Leigh McClanahan and Shafir 1990; McClanahan et al. 1999).
reserve acknowledged that both snapper (Andrew and Reduced densities of urchins in marine reserves have
Choat 1982) and spiny lobsters (Andrew and MacDiarmid implications for the maintenance of the urchin barrens
1991) were important predators of Evechinus in north- habitat and the mechanisms underlying differences
eastern New Zealand, but concluded that predation by between fished and protected areas. While some areas of
these species was not of sufficient magnitude to substan- reef at reserve sites were classified as “urchin barrens”
tially alter urchin populations and cause community-level these areas had a higher cover of articulated coralline
effects. Andrew and Choat (1982) found that the survival turf compared to non-reserve sites. Our urchin-removal
of juvenile urchins was enhanced in caged areas where experiment demonstrated that the initial response to a
predatory fish were excluded. Despite potential caging reduction in grazing pressure was an increase in the cover
artefacts they concluded that sufficient numbers of juve- of coralline turf. With continued removal of urchins from
niles escaped predation by predatory fishes to sustain the the urchin barrens habitat, there was a change from a
adult population and maintain the urchin barrens habitat. crustose coralline-dominated habitat to one dominated
Their study was carried out after only 4 years of marine by macroalgae. These findings are comparable to those
reserve protection. If it were to be repeated now after of previous urchin-removal experiments carried out in
25 years of protection and recovery of predator popula- the Leigh reserve. Ayling (1981) recorded an increase in
tions a larger effect might be expected. Spiny lobsters coralline turf when urchins were removed from small
were also discounted as playing a key regulatory role in caged areas (0.0625 m2) but no response of large brown
controlling urchin populations principally because they algae. Larger scale urchin clearances (1,000 m2) carried
were not thought to occur, or forage, in the urchin barrens out by Andrew and Choat (1982) resulted in a rapid
habitat (Andrew and Choat 1982; Andrew and MacDiarmid increase in large brown algae (Ecklonia radiata and
1991). In addition, Andrew and MacDiarmid (1991) Sargassum sinclairii) as well as coralline turf while the
investigated the relationship between lobsters and urchins control area remained dominated by crustose coralline
in the shallow broken rock habitat and found that, at the algae. The rate of change from urchin barrens to macro-
scale of 9 m2, urchins and lobsters were spatially segre- algal forests therefore depends on the spatial scales at
gated. However, Jasus edwardsii are known to forage which urchins are removed. The decrease in the extent of
over large areas (up to 100 m from their dens) (MacDiarmid urchin barrens habitat in the Leigh reserve (Babcock
et al. 1991). Our study has shown that lobsters do forage et al. 1999) and the greater abundance of macroalgal
in the urchin barrens habitat at marine reserve sites and habitats in both reserves is consistent with a large-scale
predation does occur on adult urchins which are responsible urchin removal.
for maintaining the habitat. This study demonstrates the value of marine reserves
The density of the dominant sea urchin, Evechinus as experimental tools to test ecosystem-level hypotheses
chloroticus, in the urchin barrens habitat was between at ecologically relevant scales, previously unfeasible
1.6 and 3.5 times lower at marine reserve sites compared to using traditional caging and manipulation experiments
equivalent habitats in adjacent areas. Cole and Keuskamp (Andrew and MacDiarmid 1991). Marine reserves have
(1998) carried out sampling in 1996 at Leigh and enabled us to measure the top-down role of predators in
Tawharanui and though they found lower densities in the structuring subtidal reef communities in northeastern
Leigh reserve they reported no difference between New Zealand, as well as the indirect effects of fishing on
Tawharanui Marine Park and Kawau Is, a nearby unpro- the trophic structure of reef communities. It is unclear
tected area. This may be explained by changes in urchin whether, or to what extent, these findings can be extrapo-
141
lated to other regions where urchin barrens are less com- northeastern New Zealand. In: Barker MF (ed) Echinoderms
mon and interactions between trophic levels are weaker 2000. Swets & Zeitlinger, Lisse, pp 425–430
Cole RG, Keuskamp D (1998) Indirect effects of protection from
(unpublished data; Choat and Schiel 1982; Schiel and exploitation: patterns from populations of Evechinus chloroticus
Foster 1986). Manipulations of the scale provided by (Echinoidea) in northeastern New Zealand. Mar Ecol Prog Ser
marine reserves are likely to be of equal if not greater 173:215–226
importance in understanding these systems. Cowen RK (1983) The effect of sheephead (Semicossyphus pulcher)
predation on red sea urchin populations: an experimental analysis.
Dayton et al (1998) stated that current programs Oecologia 58:249–255
aimed at understanding the functioning of kelp commu- Dayton PK, Tegner MJ, Edwards PB, Riser KL (1998) Sliding
nities will fail to distinguish the “ghosts of missing baselines, ghosts, and reduced expectations in kelp forest
animals” resulting in reduced expectations of what is communities. Ecol Appl 8:309–322
“natural”. Our study has demonstrated that the existence Edgar GJ, Barrett NS (1999) Effects of the declaration of marine
reserves on Tasmanian reef fishes, invertebrates and plants.
of reserves increases our expectations of what is natural, J Exp Mar Biol Ecol 242:107–144
and demonstrates that in some systems conservation of Estes JA, Duggins DO (1995) Sea otters and kelp forests in
large predators can lead to the re-establishment of lost Alaska: generality and variation in a community ecological
trophic interactions. paradigm. Ecol Monogr 65:75–100
Estes JA, Steinberg PD (1988) Predation, herbivory and kelp
evolution. Paleobiology 14:19–36
Acknowledgements We are very grateful to Robert Russell and Estes JA, Tinker MT, Williams TM, Doak DF (1998) Killer whale
Jarrod Walker for their invaluable assistance with fieldwork. predation on sea otters linking oceanic and nearshore systems.
Thanks to Drs Nick Tolimieri and Trevor Willis for help with Science 282:473–476
statistical analyses and revision of the manuscript. We would like Hairston NG, Smith FE, Slodbodkin LB (1960) Community struc-
to thank two anonymous reviewers, Dr Richard Taylor and the ture, population control and competition. Am Nat 94:421–425
“Leigh Laboratory discussion group” providing for valuable editorial Jennings S, Kaiser MJ (1998) The effects of fishing on marine
comments on the draft manuscript. Thanks to the staff and students ecosystems. Adv Mar Biol 34:201–352
at the Leigh Marine Laboratory for supporting this study. Kelly S, Scott D, MacDiarmid AB, Babcock RC (2000) Spiny
lobster, Jasus edwardsii, recovery in New Zealand marine
reserves. Biol Conserv 92:359–369
References Littell RC, Milliken GA, Stroup WW, Wolfinger RD (1996) SAS
system for mixed models. SAS, Cary, N.C.
Andrew NL (1988) Ecological aspects of the common sea urchin, MacDiarmid AB, Hickey B, Maller RA (1991) Daily movement
Evechinus chloroticus, in northern New Zealand: a review. NZ patterns of the spiny lobster Jasus edwardsii (Hutton) on a
J Mar Freshwater Res 22:415–426 shallow reef in northern New Zealand. J Exp Mar Biol Ecol
Andrew NL, Choat JH (1982) The influence of predation and 147:185–206
conspecific adults on the abundance of juvenile Evechinus MacPhee RDE (ed) (1999) Extinctions in near time: causes, contexts,
chloroticus (Echinoidea: Echinometridae). Oecologia 54:80–87 and consequences. Kluwer, Plenum, New York
Andrew NL, MacDiarmid AB (1991) Interrelations between sea McClanahan TR, Muthiga NA (1989) Patterns of predation on a
urchins and spiny lobsters in northeastern New Zealand. Mar sea urchin, Echinometra mathaei (de Blainville), on Kenyan
Ecol Prog Ser 70:211–222 coral reefs. J Exp Mar Biol Ecol 126:77–94
Aronson RB, Heck KL, Valentine JF (2001) Measuring predation McClanahan TR, Muthiga NA, Kamukuru AT, Machano H, Kiambo
with tethering experiments. Mar Ecol Prog Ser 214:311–312 RW (1999) The effects of marine parks and fishing on coral
Ayling AM (1981) The role of biological disturbance in temperate reefs of northern Tanzania. Biol Conserv 89:161–182
subtidal encrusting communities. Ecology 62:830–847 McClanahan TR, Shafir SH (1990) Causes and consequences of
Babcock RC, Kelly S, Shears NT, Walker JW, Willis TJ (1999) sea urchin abundance and diversity in Kenyan coral reef
Changes in community structure in temperate marine reserves. lagoons. Oecologia 83:362–370
Mar Ecol Prog Ser 189:125–134 Menge BA (1995) Indirect effects in marine intertidal interaction
Benedetti-Cecchi L, Airoldi L, Abbiati M, Cinelli F (1996) Esti- webs: patterns and importance. Ecol Monogr 65:21–74
mating the abundance of benthic invertebrates: a comparison Pace ML, Cole JE, Carpenter SR, Kitchell JF (1999) Trophic
of procedures and variability between observers. Mar Ecol cascades revealed in diverse ecosystems. Trends Ecol Evol 14:
Prog Ser 138:93–101 483–488
Bernstein BB, Williams BE, Mann KH (1981) The role of Pauly D, Christensen V, Dalsgaard J, Froese R, Torres FJ (1998)
behavioural reponse to predators in modifying urchins’ Fishing down marine food webs. Science 279:860–863
(Stronglyocentrotus droebachiensis) destructive grazing and Pinnegar JK, Polunin NVC, Francour P, Badalamenti F, Chemello
seasonal foraging patterns. Mar Biol 63:39–49 R, Harmelin-Vivien ML, Hereu B, Milazzo M, Zabala M,
Botsford LW, Castilla JC, Peterson CH (1997) The management of D’Anna G, Pipitone C (2000) Trophic cascades in benthic
fisheries and marine ecosystems. Science 277:509–515 marine ecosystems: lessons for fisheries and protected-area
Breen PA, Mann KH (1976) Changing lobster abundance and the management. Environ Conserv 27:179–200
destruction of kelp beds by sea urchins. Mar Biol 34:137–142 Polis GA, Sears ALW, Huxel GR, Strong DR, Maron J (2000)
Brett MT, Goldman CR (1996) A meta-analysis of the freshwater When is a trophic cascade a trophic cascade? Trends Ecol Evol
trophic cascades. Proc Natl Acad Sci USA 93:7723–7726 15:473–475
Choat JH, Schiel DR (1982) Patterns of distribution and abundance Sala E, Zabala M (1996) Fish predation and the structure of
of large brown algae and invertebrate herbivores in subtidal the sea urchin Paracentrotus lividus populations in the NW
regions of northern New Zealand. J Exp Mar Biol Ecol 60: Mediterranean. Mar Ecol Prog Ser 140:71–81
129–162 Sala E, Boudouresque CF, Harmelin-Vivien M (1998) Fishing,
Cole RG (1994) Abundance, size structure, and diver-oriented trophic cascades, and the structure of algal assemblages:
behaviour of three large benthic carnivorous fishes in a marine evaluation of an old but untested paradigm. Oikos 82:425–439
reserve in northeastern New Zealand. Biol Conserv 70:93–99 Scheibling RE (1996) The role of predation in regulating sea
Cole RG, Haggitt TR (2001) Dietary preferences of Evechinus urchin populations in eastern Canada. Oceanol Acta 19:421–430
chloroticus and the persistence of the fucalean macroalga Schiel DR (1990) Macroalgal assemblages in New Zealand: structure,
Carpophyllum flexuosum on coralline-dominated areas in interactions and demography. Hydrobiologia 192:59–76
142
Schiel DR, Foster MS (1986) The structure of subtidal algal stands Wharton WG, Mann KH (1981) Relationship between destructive
in temperate waters. Oceanogr Mar Biol Annu Rev 24:265–307 grazing by the sea urchin, Strongylocentrotus droebachiensis,
Schmitz OJ, Hamback PA, Beckerman AP (2000) Trophic cascades and the abundance of American Lobster, Homarus americanus,
in terrestrial systems: a review of the effects of carnivore on the Atlantic Coast of Nova Scotia. Can J Fish Aquat Sci 38:
removals on plants. Am Nat 155:141–153 1339–1349
Steinberg PD, Estes JA, Winter FC (1995) Evolutionary conse- Willis TJ (2001) Marine reserve effects on snapper (Pagrus auratus:
quences of food chain length in kelp forest communities. Proc Sparidae) in northern New Zealand. PhD thesis. University of
Natl Acad Sci USA 92:8145–8148 Auckland, New Zealand
Tegner MJ, Dayton PK (2000) Ecosystem effects of fishing in kelp Willis TJ, Millar RB (2001) Modified hooks reduce incidental
forest communities. ICES J Mar Sci 57:579–589 mortality of snapper (Pagrus auratus: Sparidae) in the New
Tegner MJ, Levin LA (1983) Spiny lobsters and sea urchins: analysis Zealand commercial longline fishery. ICES J Mar Sci 58:
of a predator-prey interaction. J Exp Mar Biol Ecol 73:125–150 830–841
Vadas RL, Steneck RS (1995) Overfishing and inferences in kelp-sea Willis TJ, Millar RB, Babcock RC (2000) Detection of spatial
urchin interactions. In: Skjoldal HR, Hopkins C, Erickstad variability in relative density of fishes: comparison of visual
KE, Leinaas HP (eds) Ecology of fjords and coastal waters. census, angling, and baited underwater video. Mar Ecol Prog
Elsevier, Amsterdam, pp 509–524 Ser 198:249–260
DOI 10.1007/s00442-002-0920-x
COMMUNITY ECOLOGY
Nick T. Shears · Russell C. Babcock
Marine reserves demonstrate top-down control
of community structure on temperate reefs
Received: 9 October 2001 / Accepted: 6 March 2002 / Published online: 8 May 2002
© Springer-Verlag 2002
Abstract Replicated ecological studies in marine reserves indirect effects of fishing and re-establish community-
and associated unprotected areas are valuable in examin- level trophic cascades.
ing top-down impacts on communities and the ecosys-
tem-level effects of fishing. We carried out experimental Keywords Kelp communities · Marine protected areas ·
studies in two temperate marine reserves to examine Northeastern New Zealand · Trophic cascades ·
these top-down influences on shallow subtidal reef com- Urchin predation
munities in northeastern New Zealand. Both reserves
examined are known to support high densities of preda-
tors and tethering experiments showed that the chance of Introduction
predation on the dominant sea urchin, Evechinus chlorot-
icus, within both reserves was approximately 7 times Top-down regulation of biological communities has been
higher relative to outside. Predation was most intense on a focal point in ecological theory (Hairston et al. 1960).
the smallest size class (30–40 mm) of tethered urchins, This is ironic, given the efficiency with which humans
the size at which urchins cease to exhibit cryptic behav- have themselves harvested the large-bodied animals
iour. A high proportion of predation on large urchins which may be responsible for the top-down control of
could be attributed to the spiny lobster, Jasus edwardsii. ecosystems, in many cases to extinction (MacPhee
Predation on the smaller classes was probably by both 1999). Examples of top-down regulation or “trophic
lobsters and predatory fish, predominantly the sparid cascades” (see Polis et al. 2000) are increasingly being
Pagrus auratus. The density of adult Evechinus actively identified in a range of terrestrial (Schmitz et al. 2000),
grazing the substratum in the urchin barrens habitat was freshwater (Brett and Goldman 1996) and marine (Sala
found to be significantly lower at marine reserve sites et al. 1998; Pinnegar et al. 2000; Tegner and Dayton
(2.2±0.3 m–2) relative to non-reserve sites (5.5±0.4 m–2). 2000) ecosystems. In the marine environment where
There was no difference in the density of cryptic juve- many fisheries have had to resort to harvesting at lower
niles between reserve and non-reserve sites. Reserve levels of the food chain (Botsford et al. 1997; Pauly et al.
populations were more bimodal, with urchins between 1998), the impacts of fishing on trophic organisation and
40 and 55 mm occurring at very low numbers. Experi- function are substantial [reviewed in Jennings and Kaiser
mental removal of Evechinus from the urchin barrens (1998)]. Removal of top predators has resulted in the
habitat over 12 months lead to a change from a crustose loss of lower-level interactions and consequently many
coralline algal habitat to a macroalgal dominated habitat. trophic cascades have been lost (Pace et al. 1999). Our
Such macroalgal habitats were found to be more exten- ability to understand, manage or restore natural systems
sive in both reserves, where urchin densities were lower, is therefore compromised by our inability to differentiate
relative to the adjacent unprotected areas that were domi- anthropogenic impacts from the “natural” dynamics of
nated by urchin barrens. The patterns observed provide systems (Dayton et al. 1998).
evidence for a top-down role of predators in structuring Trophic cascades are defined as predatory interactions
shallow reef communities in northeastern New Zealand involving three or more trophic levels, whereby primary
and demonstrate how marine reserves can reverse the carnivores indirectly increase plant abundance by sup-
pressing herbivores (Menge 1995). In many subtidal reef
N.T. Shears (✉) · R.C. Babcock systems throughout the world, a reduction in algal forests
Leigh Marine Laboratory, University of Auckland,
P.O. Box 349, Warkworth, New Zealand and an increase in urchin barrens (areas dominated by
e-mail: n.shears@auckland.ac.nz crustose corallines where the grazing activity of sea
Tel.: +64-9-4226111, Fax: +64-9-4226113 urchins has removed all large macroalgae), have been
132
linked to fisheries-related declines in urchin predators For temperate systems there are few examples of the
(Wharton and Mann 1981; Estes and Duggins 1995; use of marine reserves to examine the trophic effects of
Vadas and Steneck 1995; Sala et al. 1998). However, fishing in subtidal kelp communities. In Australia, Edgar
good empirical examples supporting the existence of and Barrett (1999) found an increase in the density of
such trophic effects are generally lacking (Scheibling large fish and lobsters and an increased mean size of aba-
1996). The best known example is that of the role of sea lone in a Tasmanian marine reserve after 7 years of pro-
otters in structuring kelp communities in the northeastern tection, relative to an associated unprotected area. They
Pacific [reviewed by Pinnegar et al. (2000) and Tegner also found some changes in algal assemblages; however,
and Dayton (2000)]. Where sea otters are abundant, her- the cause of these changes was unknown and trophic cas-
bivorous sea urchins are rare and kelp dominates, whereas cade effects were not inferred to be present. The strongest
where otters are absent urchins are abundant and kelp evidence for a key role of predators in controlling subti-
rare (Estes and Duggins 1995). Recent declines in otter dal reef communities in the southern hemisphere is from
numbers in Alaska have been related to an observed two New Zealand marine reserves (Leigh Marine Reserve
increase in killer whale attacks on otters (Estes et al. 1998) and Tawharanui Marine Park) where there has been a
adding another level to this trophic cascade. In some decline in urchin densities and an associated change from
areas where sea otters do not occur, fish and lobsters have urchin barrens to kelp over a 20-year period (Babcock et
been implicated as important predators of urchins [e.g. al. 1999). The density and size of the dominant urchin
southern Califonia (Cowen 1983; Tegner and Levin predators, the snapper Pagrus auratus (Sparidae), blue cod
1983) and the northwestern Atlantic (Bernstein et al. Parapercis colias (Pinguipedidae) and the spiny lobster
1981; Breen and Mann 1976; Wharton and Mann 1981)]. Jasus edwardsii (Palinuridae), are considerably higher in
While the destruction of kelp beds by sea urchins in these these reserves than in adjacent fished areas (Kelly et al.
areas has been linked to overfishing of both lobsters and 2000; Willis et al. 2000; Willis 2001). Both snapper and
fish, the existence of a direct causal linkage has received spiny lobster are heavily targeted by commercial and rec-
much debate (Scheibling 1996). For kelp communities in reational fisherman around New Zealand, and Babcock et
the southern hemisphere it has been widely accepted that al. (1999) suggest that this has ecosystem-level effects,
the absence of a sea otter analogue results in a simpler indirectly resulting in large-scale reduction of macroalgal
two-tiered system with no top-down control of urchins habitats and subsequently benthic primary productivity.
(Estes and Steinberg 1988; Steinberg et al. 1995). While there is strong circumstantial evidence for a
Marine reserves provide a new opportunity for testing topdown effect, experimental evidence supporting a key
the top-down impact of predators and demonstrating predatory role is generally lacking (reviewed in Schiel
indirectly the ecosystem-level effects of fishing. They 1990). Differences in urchin demography, behaviour and
function as an experimental tool where large-scale eco- morphology, and also a higher loss of transplanted
system manipulations are carried out by preventing fish- urchins in the Leigh marine reserve compared to outside
ing and subsequently elevating predator densities. The have been inferred to be due to higher levels of predation
treatments can be viewed as either with or without by Cole and Keuskamp (1998). The subtidal reef commu-
humans as the top predator, or as without or with “natural” nities in northeastern New Zealand are suited to the
predators. This enables comparisons of trophic structure occurrence of community-level cascades (Polis et al.
and further experimental manipulations to be made be- 2000) with a simple trophic structure, discrete habitats
tween reserve and non-protected areas. On coral reefs in and low species diversity. The sea urchin, Evechinus
East Africa, marine reserves have been used in this way. chloroticus is the dominant grazer (Andrew 1988), and
Predatory fish densities have been found to be higher, through its grazing activity can form urchin barrens habitat
urchin densities lower and predation on urchins higher, at depths between approximately 3 and 10 m (Ayling 1981;
in Kenyan marine reserves relative to unprotected areas Choat and Schiel 1982).
(McClanahan and Shafir 1990). Subsequently protected The aim of this study was to demonstrate the indirect
reefs had a higher species diversity and topographic effects of fishing on lower trophic levels by experimen-
complexity, with higher cover of hard coral and calcareous tally examining the top-down role of predators in
algae than unprotected areas. In the Mediterranean Sea explaining the habitat change documented in marine
an expansion of urchin barrens into areas previously reserves in northeastern New Zealand. This was done by:
occupied by erect algae has been linked to overfishing of
urchin predators (Sala and Zabala 1996. Studies utilising 1. An urchin-tethering experiment to test whether rela-
marine reserves in the Mediterranean have shown that tive predation levels on urchins were higher at marine
predatory fish are an important determinant in control- reserve sites and to determine the sources of preda-
ling urchin populations [reviewed by Sala et al. (1998)]. tion.
However, there has not yet been any decline in the extent 2. Comparing the density and demography of urchins in
of urchin barrens in these protected areas. Other factors the urchin barrens habitat at multiple sites in two
such as recruitment, pollution, disease, large-scale reserve and two non-reserve areas.
oceanographic events, urchin harvesting, food subsidies 3. Experimental removal of urchins to test whether the
and availability of shelters may also be important in observed habitat changes in the Leigh reserve were
controlling algal assemblage structure (Sala et al. 1998). consistent with a reduction in urchin populations.
133
that tether-related mortality could be reduced by holding the
urchins in the laboratory for a week prior to experimentation. This
procedure also allowed the urchins to heal, minimising the potential
effects of coelomic fluid leakage on predation (McClanahan and
Muthiga 1989).
Tethering also provided information on the source of predation
through direct observation or from examination of urchin-test
remains. Slow-moving predators such as the starfish, Coscinasterias
muricata, and the gastropod, Charonia lampax were often seen
feeding on the urchin or remained nearby. From trial experiments
we were able to classify the source of predation into the following
categories: (1) unknown (urchin missing with nylon loop still
intact demonstrating urchin had been broken off), (2) lobster (test
had characteristic pattern of lobster predation which involves a
large opening around the Aristotle’s lantern), (3) Coscinasterias
(test intact with patches of freshly stripped spines) and (4)
Charonia (test intact and mucous covered).
The tethering experiments were carried out at three reserve and
three non-reserve sites (Fig. 1); first at Leigh (4 August 1999) and
then repeated at Tawharanui (19 August 1999). Thirty urchins, of
three different size classes (n=10), were tethered at each site and
their survival monitored for 10 days. The three size classes used
for the experiments were: 35–40 mm, representing the size where
urchins move from a sheltered to an exposed habit (Andrew and
Choat 1982; Cole and Keuskamp 1998), 55–60 mm and
75–80 mm, representing the dominant adult size class outside and
Fig. 1 Location of study sites in the Cape Rodney to Okakari Point inside the Leigh reserve, respectively (Cole and Keuskamp 1998).
Marine reserve (CROP) at Leigh and Tawharanui Marine Park Experimental urchins were collected from non-reserve sites at
(TMP). Circles indicate sites where the predation experiment was Leigh where all size classes can be found openly grazing the sub-
carried out. Inset shows general location of study area on New stratum. Urchins were positioned in a 10×10-m2 plot located in the
Zealand’s North Island urchin barrens habitat adjacent to the kelp forest border. Urchins
were tethered on 25-cm monofilament traces, attached to masonry
nails that had previously been embedded in the substratum at
4. Comparing the distribution of macroalgal communities random coordinates. It was important that urchins were attached
among reef habitats between reserve and non-reserve without drawing the attention of diver-positive predatory fish at
marine reserve sites (Cole 1994; Cole and Keuskamp 1998). This
areas. was done by keeping the urchins concealed while one diver created
a disturbance nearby. There were no instances of fish predation on
recently tethered urchins. Daily monitoring enabled detection and
Materials and methods replacement of urchins that appeared to be dying as a result of
tethering. In each experiment only four out of a total of 180 tethered
Study area urchins died as a result of tethering.
Differences in the survival of urchins after 10 days were analysed
This study was carried out at sites located in two marine reserves using a generalised linear mixed model (GLMMIX). The model
and at adjacent unprotected sites in northeastern New Zealand was back-fitted to a binomial distribution using residual (restricted)
(Fig. 1). The two reserves examined are completely no-take and maximum likelihood with the GLMMIX macro in SAS (Littell et
include New Zealand’s oldest marine reserve, the Cape Rodney to al. 1996). This technique was used in preference to ANOVA as
Okakari Point (Leigh) Marine Reserve (549 ha, established in survival data follow a binomial distribution. The factors Area
1976), and Tawharanui Marine Park (350 ha, established in 1982), (Leigh and Tawharanui), Status (reserve and non-reserve) and
8 km to the south. Both marine reserves are subject to similar Size (the three size classes) were treated as fixed effects and
environmental conditions and have extensive subtidal reef commu- Site(Area×Status) as a random effect.
nities typical of moderately exposed coasts in northeastern New
Zealand (Choat and Schiel 1982). Urchin density and size structure
Comparisons of urchin populations were made between reserve
Predation and non-reserve sites at both Leigh and Tawharanui (Fig. 1). Five
sites were sampled within the Leigh reserve and five outside during
Relative predation levels on Evechinus were compared between March and April 1998, while at Tawharanui four sites were sampled
marine reserves and adjacent fished areas using tethering experi- within the reserve and four outside in June 1998. Sites were
ments. Tethering is a simple technique, suited to sedentary benthic selected in areas where urchin barrens habitat was present. At each
organisms (Aronson et al. 2001), that has been used extensively on site ten 1-m2 quadrats were placed haphazardly within the urchin
coral reefs (McClanahan and Muthiga 1989; McClanahan et al. barrens at 4–6 m depth (below MLWS). Within each quadrat we
1999) and in the Mediterranean (Sala and Zabala 1996) to test pre- measured the TD of each urchin using vernier callipers (±1 mm)
dation potential on sea urchins between protected and unprotected and noted whether urchins were located in a crevice (cryptic) or
reefs. were openly grazing the substratum (exposed). In addition, the
The tethering technique involved inserting a hypodermic needle percent cover of dominant encrusting algal forms was visually
(1.2 mm×38 mm) through the dorsal and ventral surface of the estimated to determine if any differences occurred between reserve
urchins test, as far away from the oral-aboral axis as possible and non-reserve sites.
(McClanahan and Muthiga 1989). Nylon monofilament was then Urchin counts were analysed using GLMMIX. A Poisson distri-
threaded through the needle and tied-off. Laboratory trials found bution was used as count data seldom fit the assumptions of normality
100% survival of 80 tethered urchins [ranging in size from 25 to and homogeneity of variance. The factors Area and Status were
75 mm test diameter (TD)] after 10 days. Trials in the field found treated as fixed effects and Site(Area×Status) as a random effect.
134
Fig. 2 Survival of tethered
q
urchins at reserve (q ) and non-
reserve (q) areas. The mean
number of tethered urchins
surviving in each of the three
size classes is given for Leigh
and Tawharanui
Differences in size of exposed urchins between reserve and effect Plot(Treatment) and also for the auto-regressive error struc-
non-reserve areas were tested using mixed-model ANOVA with ture [AR(1)] to account for repeated measures. A binomial distri-
fixed factors Area and Status. Site was treated as a random factor bution with logit-link was used for percent cover data and a Poisson
and nested within Area and Status. Size data were tested for nor- distribution with log-link for count data.
mality using Shapiro-Wilk’s test. Significant interaction terms
were investigated using a multiple comparison (Tukey-Kramer) of
all possible combinations of the main effects. Distribution of urchin barrens habitat
To investigate whether urchin barrens were more abundant in re-
Urchin removal serves, the proportions of habitats were measured at 22 sites located
in and around both reserves using 1-m-wide strip transects (three at
A sea urchin-removal experiment was undertaken to investigate each site). Transects were run perpendicular to the shore from
the role Evechinus plays in maintaining the urchin barrens habitat MLWS to the reef edge or a maximum depth of 12 m. Both depth
and the response of algal communities to a reduction in urchin and habitat type were recorded every metre. Habitat type was
density. The experiment was carried out at Mathesons Bay (Fig. 1) recorded as one of the following categories, based on the density of
near the Leigh marine reserve on an area of reef with extensive plants within each 1-m2 area along the transect: (1) macroalgal hab-
urchin barrens habitat. The reef was dissected by sand-filled itat, >3 adult phaeophytes m–2 e.g. Ecklonia radiata or Carpophyllum
crevices, which form semi-isolated blocks of reef, allowing the flexuosum; (2) urchin barrens, >50% cover of crustose coralline
establishment of discrete experimental plots within a 500-m2 area algae; (3) shallow fucoid zone, >20% cover or 3 adult phaeophytes
of reef in the urchin barrens habitat. Six blocks of reef were selected, m–2 at depths <4 m; (4) turf habitat, >50% cover of turf forming
ranging in size from 10 to 20 m2, at a depth of 4–5 m. All urchins red or green algae with large phaeophytes <3 m–2.
were removed from three randomly selected blocks, the remaining The proportional cover of urchin barrens habitat within three
three were left as controls. The urchins were removed in January depth ranges (0–3, 4–6 and 7–9 m) was examined using GLMMIX
1998, with weekly re-clearances until January 1999. with a binomial distribution. Area, Status and Depth were treated as
The initial density of grazers and macroalgae in the experimental fixed factors, and Site(Area×Status) was treated as a random factor.
areas was estimated in five haphazardly placed 0.25-m2 quadrats.
The percent cover of encrusting algae (crustose coralline algae,
articulated coralline turf, filamentous algae and other encrusting Results
algae) was also measured by estimating the number of 10×10-cm
cells within the 0.25-m2 quadrats each algal type “filled” (Benedetti- Predation
Cecchi et al. 1996). Sampling was repeated monthly to determine
the response of the communities to manipulation. To test for dif- Predation on urchins was significantly higher at reserve
ferences in the dominant species between treatments and between
plots within treatments at the start of the experiment and over time sites than at non-reserve sites (F=9.44, P=0.0133), with
GLMMIX was used. Treatment and Time were set as fixed effects. the relative odds of predation being 6.9 times higher
Covariance parameter estimates were calculated for the random at reserve sites (Fig. 2, Table 1). This was consistent
135
Table 1 Summary statistics for reserve/non-reserve comparisons of urchin barrens and coralline turf (both binomial distribution).
with size of effect expressed as a ratio with 95% confidence limits For count data the ratio indicates the effect size whereas for
(CL). Likelihood ratios calculated by the SAS procedure generalised mortality and percent data the ratio indicates the relative odds
linear mixed model for mortality of tethered Evechinus (binomial [see Willis and Millar (2001) for explanation of interpreting relative
distribution), Evechinus density (Poisson distribution), the cover odds ratio]
Reserve Non-reserve Reserve:non- Upper Lower
mean SE mean SE reserve ratio 95% CL 95% CL
Predation on Evechinus (% mortality) 42.8 (9.0) 12.2 (3.4) 6.88 2.01 23.57
Exposed Evechinus density (m–2) 2.2 (0.3) 5.5 (0.4) 0.60 0.45 0.79
Exposed Evechinus mean size (mm) 69.8 (2.5) 57.3 (1.3) – – –
Cover of coralline turf (%) 29.3 (2.9) 12.6 (1.5) 1.80 0.90 3.60
Extent of urchin barrens (%) 14.8 (4.8) 41.4 (4.2) 0.17 0.07 0.41
Table 2 Source of predation on tethered urchins
Reserve Non-reserve
Size class 35 mm 55 mm 75 mm 35 mm 55 mm 75 mm
Number preyed 40 23 13 11 7 4
Proportion
Unknown 100.0 56.5 46.2 54.5 42.9 0.0
Lobster 0.0 43.3 46.2 0.0 0.0 0.0
Coscinasterias 0.0 0.0 7.7 45.5 42.9 50.0
Charonia 0.0 0.0 0.0 0.0 14.3 50.0
between both areas (F=0.41, P=0.5357). There was a
significant effect of size on predation (F=12.60,
P<0.0001), which was also consistent between areas and
between reserve and non-reserve sites. Predation occurred
on all size classes of tethered urchins, at both reserve and
non-reserve sites, but was highest on the smallest size
class (Fig. 2). The likelihood of predation on the small
and middle size-class urchins was 6.3 [95% confidence
limits (CL) 3.0–13.3] and 2.2 (CL 1.1–4.7) times greater,
respectively, than predation on the largest size class. The
odds of predation did not vary significantly between
reserve and non-reserve sites for each area (Z=1.52,
P=0.0639).
The fate of all small urchins (35 mm) preyed upon at
reserve sites was unknown as the tests were completely
removed from the tethers (Table 2). This could have
been due either to predation by fish, which completely
engulf the urchin, or by lobsters breaking up or removing
small urchins. At reserve sites approximately 45% of
preyed individuals in the larger size classes (55 and
75 mm) showed patterns of damage characteristic of
spiny lobster predation. No urchins showed signs of Fig. 3 Mean density of A exposed and B cryptic urchins, and
spiny lobster predation at non-reserve sites. In most C mean size of exposed urchins from quadrat sampling (n=10) at
cases mortality at non-reserve sites could be attributed q
all reserve (q ) and non-reserve sites (q)
to starfish (Coscinasterias muricata) or the gastropod,
Charonia lampax, both of which are slow-moving
predators. reserve sites for both areas (Fig. 3). The density of
exposed urchins (Fig. 3A) was significantly lower at
marine reserve sites (Tables 1, 3). Exposed urchins were
Urchin density and size structure 1.7 times more abundant overall at non-reserve sites
(Table 1). There was no difference in urchin density
Densities of Evechinus in the urchin-grazed habitat between areas (Leigh and Tawharanui) but there was a
varied widely between sites but were generally lower at significant interaction between Area and Status. This can
136
Table 3 Exposed urchin density statistics. Type 3 tests for counts Table 5 Urchin size statistics. Mixed-model ANOVA results for
of exposed urchins for fixed effects Area (Leigh and Tawharanui) exposed urchin size at Leigh and Tawharanui (Area), reserve and
and Status (reserve and non-reserve). Parameter estimates for the non-reserve site (Status)
random effect Site(Area×Status)
df Mean square F-value Pr>F
Fixed effects df F-value Pr>F
Area 1 3,343.2 5.58 0.0321
Status 1, 14 73.65 <0.0001 Area×Status 1 4,552.8 7.60 0.0147
Area 1, 14 2,21 0.1597 Status 1 19,374.0 32.48 <0.0001
Area×Status 1, 14 13.06 0.0028 Site(Area×Status) 14 726.0 7.33 <0.0001
Covariance parameter Estimate SE Z-value Pr Z
Site(Area×Status) 0.0023 0.0135 0.17 0.4312
The abundance of cryptic urchins (Fig. 3B) did not differ
significantly between reserve and non-reserve locations
Table 4 Cryptic urchin density statistics. Type 3 tests for count of
exposed urchins for fixed effects Area (Leigh and Tawharanui) or between Leigh and Tawharanui (Table 4), although
and Status (reserve and non-reserve). Parameter estimates for the there was significant site-level variability.
random effect (Area×Status) There was significant variation in the size of exposed
urchins between sites (Fig. 3C, Table 5). The mean size
Fixed effects df χ2 P
of exposed urchins was significantly larger at marine re-
Status 1, 14 0.37 0.5529 serve sites (Table 1) although the Area effect was signifi-
Area 1, 14 0.62 0.445 cant and there was a significant Area and Status interac-
Area×Status 1, 14 0.01 0.9274 tion. This can be explained by the larger effect of Status
Covariance parameter Estimate SE Z-value Pr Z at Leigh which results in a significant interaction be-
Site (Area×Status) 0.5194 0.2465 2.11 0.0175 tween Area and Status. There was no difference in size
between the non-reserve sites for each area [Tukey’s
honestly significant difference (HSD) P=0.8112], but for
be explained by examining the size of the effect of Status both areas there were significant differences between
between both areas; for Leigh densities were 3.5 (95% reserve and non-reserve sites (Tukey’s HSD P<0.0001).
CL 2.8–4.4) times higher at non-reserve sites while at There was also significant difference between reserve
Tawharanui densities were 1.7 (1.2–2.4) times higher. populations at Leigh and Tawharanui (Tukey’s HSD
Separate analysis for each area found a significant differ- P<0.0001).
ence in density between reserve and non-reserve sites for The population structure of Evechinus varied between
both Leigh (P<0.0001) and Tawharanui (P=0.0342). reserve and non-reserve sites (Fig. 4). Populations were
Fig. 4 Size frequency distribu-
tions of all Evechinus measured
during quadrat sampling at
each area. Shaded bars indicate
proportion of cryptic urchins
137
Fig. 5 The percent cover of Coralline turf from quadrat sampling
q
(n=10) at all reserve (q) and non-reserve sites (q). Means are given
for each site
more bimodal at reserve sites, with very low numbers of
urchins between 30 and 50 mm, and they generally
remained cryptic to a greater size. This pattern was
stronger in the Leigh marine reserve. Fig. 6A, B Response of encrusting and turfing algae to urchin
Quadrat sampling also revealed that, overall, the removal. The mean proportional cover of A crustose coralline
q
algae and B coralline turf in both control (q) and urchin removal
percent cover of coralline turf (Corallina officinalis) (q) plots following commencement of the experiment in January
was significantly higher at reserve sites (F1,9=14.18, 1998. J January, F February, M March, A April, M May,
P=0.0044) (Table 1, Fig. 5). The relative odds ratio was J June, J July, A August, S September, O October, N November,
1.8 times higher at marine reserve sites (Table 1). This D December
was consistent between areas (F1,9=0.50, P=0.4960) and
while the pattern was clearest at Leigh (Fig. 5) the effect
of reserve status was consistent between areas (F1,9=1.79, crustose coralline algae across all plots was due to a
P=0.2140). large settlement of filamentous algae at the start of the
experiment (Fig. 6). The change from crustose coralline
to articulated coralline algae occurred rapidly for the first
Urchin removal 4 months then remained stable throughout the winter. The
cover of crustose corallines and coralline turf varied
At the commencement of the experiment in January significantly over time (Table 6). While the overall effect
1998 Evechinus densities did not vary between treat- of treatment was not clear, the effect of urchin removal
ments (F1,4=0.06, P=0.8135). Densities of urchins ranged on coralline algae and coralline turf over time was
from 1.2 to 2.4 per 0.25 m2. Crustose coralline algae significant (Table 6).
(Lithothamnion and Lithophyllum spp.) were dominant, A number of brown algal species became established
covering 63–99% of the substratum. Articulated coralline in the urchin-removal plots (Fig. 7). In most cases these
turf was the other dominant encrusting form with cover species remained absent from control plots so differences
ranging between 0 and 35%. There was no significant between treatments could not be statistically tested. Only
difference in either crustose coralline algae (F1,4=0.36, Carpophyllum flexuosum occurred at sufficient densities
P=0.5789) or coralline turf (F1,4=0.63, P=0.4730) in both control and removal plots throughout the experi-
between treatments or between plots within treatments ment for statistical analysis (Fig. 7A). There was no
(Z=1.1, P=0.1367, Z=1.08, P=0.1411). Macroalgae effect of urchin removal on the density of C. flexuosum
were rare at the start of the experiment, with Carpophyllum (Table 6), the numbers remaining stable over time.
flexuosum, which is relatively resilient to urchin grazing Several large Ecklonia radiata sporophytes became
(Cole and Haggitt 2001), being the only conspicuous established (Fig. 7B) within the urchin removal plots
large brown seaweed (<1 per 0.25 m2). There was no while remaining absent in control areas. Survival of
significant difference in the number of C. flexuosum Ecklonia recruits was observed to be low as they appeared
between treatments (F1,4=0.52, P=0.5123) or between to be prime targets for any urchins which did immigrate
plots within treatments (Z=0.53, P=0.2966). Ecklonia into treatment plots. Total exclusion of urchins would
radiata was absent from all plots. probably have resulted in a more rapid response of
After 1 year the control plots remained as urchin barrens Ecklonia. Low numbers of two other species of
dominated by crustose coralline algae, while the urchin- large brown algae, Carpophyllum maschalocarpum and
removal plots had become dominated by coralline turf, Sargassum sinclairii also became established in the
with a mixture of large and small brown algae (a “mixed urchin-removal areas. Small brown algae showed a marked
algal habitat”). A temporary decrease in the cover of response to urchin removal. These included Halopteris
138
Table 6 Response of crustose coralline algae, coralline turf and
Carpophyllum flexuosum following urchin removal. Type 3 analysis
for the percent cover of crustose coralline algae and coralline turf,
and the number of C. flexuosum plants following urchin removal
for fixed effects Treatment and Time. Parameter estimates for the
random effect Plot(Treatment) and the repeated measures effect
[AR(1)]
Crustose coralline
Fixed effects df F-value Pr>F
Treatment 1, 4 7.43 0.0527
Time 9, 276 21.76 <0.0001
Treatment×Time 9, 276 4.57 <0.0001
Covariance parameter Estimate SE Z-value Pr Z
Plot(Treatment) 0.2247 0.1678 1.34 0.0903
AR(1) 0.3961 0.0650 6.10 <0.0001
Coralline turf
Fixed effects df F-value Pr>F
Treatment 1, 4 1.24 0.3278
Time 9, 276 24.74 <0.0001
Treatment×Time 9, 276 9.48 <0.0001
Covariance parameter Estimate SE Z-value Pr Z
Plot(Treatment) 0.8042 0.5754 1.40 0.0811
AR(1) 0.1938 0.0640 3.03 0.0024
Carpophyllum flexuosum
Fixed effects df F-value Pr>F
Treatment 1, 4 1.32 0.3141
Time 9, 276 1.20 0.2947
Treatment×Time 9, 276 1.36 0.2077
Covariance parameter Estimate SE Z-value Pr Z
Plot(Treatment) 1.0384 0.9185 1.13 0.1291
AR(1) –0.0660 0.0630 –1.05 0.2948
Fig. 7A–D Response of macroalgae to urchin removal. The mean
abundance of A Carpophyllum flexuosum, B Ecklonia radiata,
C Halopteris virgata and D small brown seasonal algae in both
q
control (q) and urchin-removal (q) plots following commencement Table 7 Statistics for the proportion of urchin barrens. Type 3
of the experiment in January 1998 analysis for fixed factors Status (reserve/non-reserve), Area (Leigh/
Tawharanui) and Depth (0–3, 4–6 and 7–9 m). Non-significant
interaction terms were removed from the model. Parameter estimates
virgata (Fig. 7C) and a number of short-lived seasonal given for the random effect Site(Area×Status)
species (Fig. 7D), including Dictyota sp. and Colp-
Fixed effects df F-value Pr>F
omenia sinuosa. A few red algal species such as Aspar-
agopsis armata and Champia novaezelandicae also re- Status 1, 19 15.24 0.0010
cruited into urchin-removal plots. Area 1, 19 0.00 0.9740
Within 6 months of completion of the experiment the Depth 2, 174 32.73 <0.001
urchin-removal plots had been heavily grazed and re- Covariance parameter Estimate SE Z-value Pr Z
verted to urchin barrens habitat, dominated by crustose Site(Area×Status) 0.8844 0.3715 2.38 0.0087
coralline algae. The only brown algae present were
stunted Carpophyllum flexuosum plants (personal obser-
vation). (Tables 1, 7). The relative odds ratio for the proportion
of urchin barrens at reserve vs. non-reserve sites was
0.2:1 (Table 1), or inversely, 5.9 times higher at non-
Distribution of urchin barrens habitat reserve sites. This pattern was consistent between both
areas but varied significantly with depth (Table 7).
The general distribution of habitats (Fig. 8) at non- Differences between reserve and non-reserve sites were
reserve sites is typical of northeastern New Zealand greatest in the shallow depth strata (0–3 m and 4–6 m)
with shallow fucoid assemblages, deep kelp forests where urchin barrens covered approximately 54% of
(Ecklonia radiata) and intermediate depths dominated available reef at non-reserve sites and only about 20%
by urchin barrens (Choat and Schiel 1982). However, at of the reef at reserve sites. The deeper strata (7–9 m
marine reserve sites all depth ranges were dominated by and 10–12 m) at both reserve and non-reserve sites
macroalgal habitats. The proportion of urchin barrens were dominated by macroalgal habitat, predominantly
habitat was significantly lower than at non-reserve sites Ecklonia forest.
139
Fig. 8 Mean percent cover of
major habitat types (shallow
fucoids, urchin barrens, turfing
algae and kelp forest), within
each depth range for reserve
and non-reserve sites at Leigh
and Tawharanui
Discussion MacDiarmid 1991). The actual proportion of predation
attributable to lobsters is likely to be higher than estimated,
New Zealand’s two oldest marine reserves at Leigh and as a spiny lobster may completely consume a tethered
Tawharanui support higher predator densities than similar urchin or take it back to shelter, in both cases leaving no
unprotected areas of coastline (Babcock et al. 1999; remains from which the source of predation could be
Kelly et al. 2000; Willis et al. 2000; Willis 2001). Snapper ascertained. In addition, lobster-predated tests of unteth-
are at least 5.8–8.7 times more abundant inside these ered urchins were found regularly at reserve sites during
reserves and spiny lobster 1.6–3.7 times more abundant, the study, demonstrating that spiny lobster were also
than in adjacent unprotected areas (Babcock et al. 1999). feeding on natural urchin populations. The highest levels
Relative rates of predation on sea urchins were found to of predation occurred on the 30- to 40-mm size class,
be higher in the reserves, the densities of actively grazing which are normally sheltered at reserve sites but exposed
urchins lower, and the cover of macroalgal forests higher at non-reserve sites. While the specific predators respon-
relative to adjacent unprotected areas. These patterns sible for attacks on this size class could not be identified
confirmed and extended previous results (Cole and they were likely to have been both lobsters, which pref-
Keuskamp 1998; Babcock et al. 1999) and provide erentially take smaller urchins (Andrew and MacDiarmid
experimental evidence for a top-down impact of predators 1991), and predatory fish. Gut content analysis (Babcock
on subtidal reef communities. The fact that these patterns et al. 1999) and visual observations (personal observa-
were found through comparisons of multiple sites inside tion) indicate that snapper and blue cod only feed on
and outside more than one marine reserve mean that it is small urchins (<50 mm). Cole and Keuskamp (1998)
highly unlikely that the effects were due to spatial varia- suggested that the higher loss of transplanted urchins of
tions in other factors such as nutrients or larval supply. this size in the Leigh reserve sites was due to predation
The spiny lobster, Jasus edwardsii, was found to be by fish.
an important predator of sea urchins at marine reserve Predation of tethered urchins at non-reserve sites
sites. At least 45% of predation on the larger size classes was attributed to a different suite of predators, mainly
of tethered urchins at reserve sites could be attributed to Coscinasterias and Charonia. These species are natural
lobsters. This is noteworthy, considering that spiny lob- predators of urchins (personal observation) but tethering
sters were not previously thought to forage in the urchin reduced the chance of urchins escaping from these slow-
barrens habitat (Andrew and Choat 1982; Andrew and moving predators. At reserve sites both of these predatory
140
species tend to occur at much lower densities than at densities over the 2 years prior to our study, or due to the
non-reserve sites (N. Shears, unpublished data), subse- fact that their sampling was carried out over a larger
quently there was only one instance of predation by depth range (5–10 m) and did not sample sites adjacent
Coscinasterias recorded at reserve sites. Similar patterns to the marine reserve. The smaller effect of reserve status
have been shown in other studies where different preda- seen at Tawharanui compared to Leigh may be due to
tors are important at fished sites where the primary pre- several factors; higher levels of poaching (personal
dators have been removed. In the Gulf of Maine, Vadas observation), younger reserve age and smaller reserve
and Steneck (1995) found high levels of fish predation size. The size structures of Evechinus populations we
on urchins at an offshore reef subject to low fishing pres- found were consistent with those found by Cole and
sure, while at heavily fished coastal sites predation on Keuskamp (1998). At reserve sites the exposed urchins
urchins was attributed to crabs. This was also suggested were larger and populations were more bimodal than at
to be at least in part an artefact of tethering. Likewise, in non-reserve sites. Our tethering experiment provides
the Mediterranean a predatory gastropod was an impor- support for the hypothesis that bimodality in Evechinus
tant predator at fished sites while at protected sites fish populations is related to higher size-specific predation on
were the dominant predators (Sala and Zabala 1996). juveniles moving from a cryptic to exposed lifestyle
Our urchin-tethering experiment provided strong evi- (Andrew and Choat 1982; Cole and Keuskamp 1998).
dence that the lower densities of urchins in both reserves Lower densities of urchins at protected sites com-
are due to relatively higher levels of predation within pared to fished sites has been linked to higher predator
marine reserves. densities in other marine systems (Sala and Zabala 1996;
Previous experimental studies carried out in the Leigh McClanahan and Shafir 1990; McClanahan et al. 1999).
reserve acknowledged that both snapper (Andrew and Reduced densities of urchins in marine reserves have
Choat 1982) and spiny lobsters (Andrew and MacDiarmid implications for the maintenance of the urchin barrens
1991) were important predators of Evechinus in north- habitat and the mechanisms underlying differences
eastern New Zealand, but concluded that predation by between fished and protected areas. While some areas of
these species was not of sufficient magnitude to substan- reef at reserve sites were classified as “urchin barrens”
tially alter urchin populations and cause community-level these areas had a higher cover of articulated coralline
effects. Andrew and Choat (1982) found that the survival turf compared to non-reserve sites. Our urchin-removal
of juvenile urchins was enhanced in caged areas where experiment demonstrated that the initial response to a
predatory fish were excluded. Despite potential caging reduction in grazing pressure was an increase in the cover
artefacts they concluded that sufficient numbers of juve- of coralline turf. With continued removal of urchins from
niles escaped predation by predatory fishes to sustain the the urchin barrens habitat, there was a change from a
adult population and maintain the urchin barrens habitat. crustose coralline-dominated habitat to one dominated
Their study was carried out after only 4 years of marine by macroalgae. These findings are comparable to those
reserve protection. If it were to be repeated now after of previous urchin-removal experiments carried out in
25 years of protection and recovery of predator popula- the Leigh reserve. Ayling (1981) recorded an increase in
tions a larger effect might be expected. Spiny lobsters coralline turf when urchins were removed from small
were also discounted as playing a key regulatory role in caged areas (0.0625 m2) but no response of large brown
controlling urchin populations principally because they algae. Larger scale urchin clearances (1,000 m2) carried
were not thought to occur, or forage, in the urchin barrens out by Andrew and Choat (1982) resulted in a rapid
habitat (Andrew and Choat 1982; Andrew and MacDiarmid increase in large brown algae (Ecklonia radiata and
1991). In addition, Andrew and MacDiarmid (1991) Sargassum sinclairii) as well as coralline turf while the
investigated the relationship between lobsters and urchins control area remained dominated by crustose coralline
in the shallow broken rock habitat and found that, at the algae. The rate of change from urchin barrens to macro-
scale of 9 m2, urchins and lobsters were spatially segre- algal forests therefore depends on the spatial scales at
gated. However, Jasus edwardsii are known to forage which urchins are removed. The decrease in the extent of
over large areas (up to 100 m from their dens) (MacDiarmid urchin barrens habitat in the Leigh reserve (Babcock
et al. 1991). Our study has shown that lobsters do forage et al. 1999) and the greater abundance of macroalgal
in the urchin barrens habitat at marine reserve sites and habitats in both reserves is consistent with a large-scale
predation does occur on adult urchins which are responsible urchin removal.
for maintaining the habitat. This study demonstrates the value of marine reserves
The density of the dominant sea urchin, Evechinus as experimental tools to test ecosystem-level hypotheses
chloroticus, in the urchin barrens habitat was between at ecologically relevant scales, previously unfeasible
1.6 and 3.5 times lower at marine reserve sites compared to using traditional caging and manipulation experiments
equivalent habitats in adjacent areas. Cole and Keuskamp (Andrew and MacDiarmid 1991). Marine reserves have
(1998) carried out sampling in 1996 at Leigh and enabled us to measure the top-down role of predators in
Tawharanui and though they found lower densities in the structuring subtidal reef communities in northeastern
Leigh reserve they reported no difference between New Zealand, as well as the indirect effects of fishing on
Tawharanui Marine Park and Kawau Is, a nearby unpro- the trophic structure of reef communities. It is unclear
tected area. This may be explained by changes in urchin whether, or to what extent, these findings can be extrapo-
141
lated to other regions where urchin barrens are less com- northeastern New Zealand. In: Barker MF (ed) Echinoderms
mon and interactions between trophic levels are weaker 2000. Swets & Zeitlinger, Lisse, pp 425–430
Cole RG, Keuskamp D (1998) Indirect effects of protection from
(unpublished data; Choat and Schiel 1982; Schiel and exploitation: patterns from populations of Evechinus chloroticus
Foster 1986). Manipulations of the scale provided by (Echinoidea) in northeastern New Zealand. Mar Ecol Prog Ser
marine reserves are likely to be of equal if not greater 173:215–226
importance in understanding these systems. Cowen RK (1983) The effect of sheephead (Semicossyphus pulcher)
predation on red sea urchin populations: an experimental analysis.
Dayton et al (1998) stated that current programs Oecologia 58:249–255
aimed at understanding the functioning of kelp commu- Dayton PK, Tegner MJ, Edwards PB, Riser KL (1998) Sliding
nities will fail to distinguish the “ghosts of missing baselines, ghosts, and reduced expectations in kelp forest
animals” resulting in reduced expectations of what is communities. Ecol Appl 8:309–322
“natural”. Our study has demonstrated that the existence Edgar GJ, Barrett NS (1999) Effects of the declaration of marine
reserves on Tasmanian reef fishes, invertebrates and plants.
of reserves increases our expectations of what is natural, J Exp Mar Biol Ecol 242:107–144
and demonstrates that in some systems conservation of Estes JA, Duggins DO (1995) Sea otters and kelp forests in
large predators can lead to the re-establishment of lost Alaska: generality and variation in a community ecological
trophic interactions. paradigm. Ecol Monogr 65:75–100
Estes JA, Steinberg PD (1988) Predation, herbivory and kelp
evolution. Paleobiology 14:19–36
Acknowledgements We are very grateful to Robert Russell and Estes JA, Tinker MT, Williams TM, Doak DF (1998) Killer whale
Jarrod Walker for their invaluable assistance with fieldwork. predation on sea otters linking oceanic and nearshore systems.
Thanks to Drs Nick Tolimieri and Trevor Willis for help with Science 282:473–476
statistical analyses and revision of the manuscript. We would like Hairston NG, Smith FE, Slodbodkin LB (1960) Community struc-
to thank two anonymous reviewers, Dr Richard Taylor and the ture, population control and competition. Am Nat 94:421–425
“Leigh Laboratory discussion group” providing for valuable editorial Jennings S, Kaiser MJ (1998) The effects of fishing on marine
comments on the draft manuscript. Thanks to the staff and students ecosystems. Adv Mar Biol 34:201–352
at the Leigh Marine Laboratory for supporting this study. Kelly S, Scott D, MacDiarmid AB, Babcock RC (2000) Spiny
lobster, Jasus edwardsii, recovery in New Zealand marine
reserves. Biol Conserv 92:359–369
References Littell RC, Milliken GA, Stroup WW, Wolfinger RD (1996) SAS
system for mixed models. SAS, Cary, N.C.
Andrew NL (1988) Ecological aspects of the common sea urchin, MacDiarmid AB, Hickey B, Maller RA (1991) Daily movement
Evechinus chloroticus, in northern New Zealand: a review. NZ patterns of the spiny lobster Jasus edwardsii (Hutton) on a
J Mar Freshwater Res 22:415–426 shallow reef in northern New Zealand. J Exp Mar Biol Ecol
Andrew NL, Choat JH (1982) The influence of predation and 147:185–206
conspecific adults on the abundance of juvenile Evechinus MacPhee RDE (ed) (1999) Extinctions in near time: causes, contexts,
chloroticus (Echinoidea: Echinometridae). Oecologia 54:80–87 and consequences. Kluwer, Plenum, New York
Andrew NL, MacDiarmid AB (1991) Interrelations between sea McClanahan TR, Muthiga NA (1989) Patterns of predation on a
urchins and spiny lobsters in northeastern New Zealand. Mar sea urchin, Echinometra mathaei (de Blainville), on Kenyan
Ecol Prog Ser 70:211–222 coral reefs. J Exp Mar Biol Ecol 126:77–94
Aronson RB, Heck KL, Valentine JF (2001) Measuring predation McClanahan TR, Muthiga NA, Kamukuru AT, Machano H, Kiambo
with tethering experiments. Mar Ecol Prog Ser 214:311–312 RW (1999) The effects of marine parks and fishing on coral
Ayling AM (1981) The role of biological disturbance in temperate reefs of northern Tanzania. Biol Conserv 89:161–182
subtidal encrusting communities. Ecology 62:830–847 McClanahan TR, Shafir SH (1990) Causes and consequences of
Babcock RC, Kelly S, Shears NT, Walker JW, Willis TJ (1999) sea urchin abundance and diversity in Kenyan coral reef
Changes in community structure in temperate marine reserves. lagoons. Oecologia 83:362–370
Mar Ecol Prog Ser 189:125–134 Menge BA (1995) Indirect effects in marine intertidal interaction
Benedetti-Cecchi L, Airoldi L, Abbiati M, Cinelli F (1996) Esti- webs: patterns and importance. Ecol Monogr 65:21–74
mating the abundance of benthic invertebrates: a comparison Pace ML, Cole JE, Carpenter SR, Kitchell JF (1999) Trophic
of procedures and variability between observers. Mar Ecol cascades revealed in diverse ecosystems. Trends Ecol Evol 14:
Prog Ser 138:93–101 483–488
Bernstein BB, Williams BE, Mann KH (1981) The role of Pauly D, Christensen V, Dalsgaard J, Froese R, Torres FJ (1998)
behavioural reponse to predators in modifying urchins’ Fishing down marine food webs. Science 279:860–863
(Stronglyocentrotus droebachiensis) destructive grazing and Pinnegar JK, Polunin NVC, Francour P, Badalamenti F, Chemello
seasonal foraging patterns. Mar Biol 63:39–49 R, Harmelin-Vivien ML, Hereu B, Milazzo M, Zabala M,
Botsford LW, Castilla JC, Peterson CH (1997) The management of D’Anna G, Pipitone C (2000) Trophic cascades in benthic
fisheries and marine ecosystems. Science 277:509–515 marine ecosystems: lessons for fisheries and protected-area
Breen PA, Mann KH (1976) Changing lobster abundance and the management. Environ Conserv 27:179–200
destruction of kelp beds by sea urchins. Mar Biol 34:137–142 Polis GA, Sears ALW, Huxel GR, Strong DR, Maron J (2000)
Brett MT, Goldman CR (1996) A meta-analysis of the freshwater When is a trophic cascade a trophic cascade? Trends Ecol Evol
trophic cascades. Proc Natl Acad Sci USA 93:7723–7726 15:473–475
Choat JH, Schiel DR (1982) Patterns of distribution and abundance Sala E, Zabala M (1996) Fish predation and the structure of
of large brown algae and invertebrate herbivores in subtidal the sea urchin Paracentrotus lividus populations in the NW
regions of northern New Zealand. J Exp Mar Biol Ecol 60: Mediterranean. Mar Ecol Prog Ser 140:71–81
129–162 Sala E, Boudouresque CF, Harmelin-Vivien M (1998) Fishing,
Cole RG (1994) Abundance, size structure, and diver-oriented trophic cascades, and the structure of algal assemblages:
behaviour of three large benthic carnivorous fishes in a marine evaluation of an old but untested paradigm. Oikos 82:425–439
reserve in northeastern New Zealand. Biol Conserv 70:93–99 Scheibling RE (1996) The role of predation in regulating sea
Cole RG, Haggitt TR (2001) Dietary preferences of Evechinus urchin populations in eastern Canada. Oceanol Acta 19:421–430
chloroticus and the persistence of the fucalean macroalga Schiel DR (1990) Macroalgal assemblages in New Zealand: structure,
Carpophyllum flexuosum on coralline-dominated areas in interactions and demography. Hydrobiologia 192:59–76
142
Schiel DR, Foster MS (1986) The structure of subtidal algal stands Wharton WG, Mann KH (1981) Relationship between destructive
in temperate waters. Oceanogr Mar Biol Annu Rev 24:265–307 grazing by the sea urchin, Strongylocentrotus droebachiensis,
Schmitz OJ, Hamback PA, Beckerman AP (2000) Trophic cascades and the abundance of American Lobster, Homarus americanus,
in terrestrial systems: a review of the effects of carnivore on the Atlantic Coast of Nova Scotia. Can J Fish Aquat Sci 38:
removals on plants. Am Nat 155:141–153 1339–1349
Steinberg PD, Estes JA, Winter FC (1995) Evolutionary conse- Willis TJ (2001) Marine reserve effects on snapper (Pagrus auratus:
quences of food chain length in kelp forest communities. Proc Sparidae) in northern New Zealand. PhD thesis. University of
Natl Acad Sci USA 92:8145–8148 Auckland, New Zealand
Tegner MJ, Dayton PK (2000) Ecosystem effects of fishing in kelp Willis TJ, Millar RB (2001) Modified hooks reduce incidental
forest communities. ICES J Mar Sci 57:579–589 mortality of snapper (Pagrus auratus: Sparidae) in the New
Tegner MJ, Levin LA (1983) Spiny lobsters and sea urchins: analysis Zealand commercial longline fishery. ICES J Mar Sci 58:
of a predator-prey interaction. J Exp Mar Biol Ecol 73:125–150 830–841
Vadas RL, Steneck RS (1995) Overfishing and inferences in kelp-sea Willis TJ, Millar RB, Babcock RC (2000) Detection of spatial
urchin interactions. In: Skjoldal HR, Hopkins C, Erickstad variability in relative density of fishes: comparison of visual
KE, Leinaas HP (eds) Ecology of fjords and coastal waters. census, angling, and baited underwater video. Mar Ecol Prog
Elsevier, Amsterdam, pp 509–524 Ser 198:249–260