MARINE ECOLOGY PROGRESS SERIES
  Vol. 189: 125-134, 1999                                     Published November 26
                         Mar Ecol Prog Ser




    Changes in community structure in temperate
           marine reserves
Russell C. Babcock*, Shane Kelly, Nick T. Shears, Jarrod W. Walker, Trevor J. Willis
          Leigh Marine Laboratory, University of Auckland. PO Box 349, Warkworth. New Zealand




       ABSTRACT. 'No-take' marine reserves provide a valuable tool for managing marine resources as well
       as for providing relatively undisturbed habitat with which to assess modifications to ecosystems. We
       studied 2 marine reserves in northeastern New Zealand, the Leigh Marine Reserve (established 1975)
       and Tawharanui Marine Park (established 1982) in order to assess whether changes in protected pre-
       dator populations had resulted in other indirect changes to grazers and consequently to algal abun-
       dance. Estimates of abundance of the most common demersal predatory fish Pagrus auratus indicated
       that adults of this species (i.e.large enough to prey upon urchins) were at least 5.75 and 8.70 times
       more abundant inside reserves than in adjacent unprotected areas. Overall, P auratus were also much
       larger inside reserves with mean total lengths of 316 mm compared with 186 mm in fished areas. The
       spiny lobster Jasus edwardsii displayed similar trends, and was approximately 1.6 to 3.7 times more
       abundant inside the reserves than outside. Lobsters within the reserves had a mean carapace length of
       109.9 mm, compared with 93.5 mm outside the reserves. In one of the reserves, densities of the sea
       urchin Evechinus chloroticus had declined from 4.9 to 1.4 m-2 since 1978 in areas formerly dominated
       by it. Consequently, kelp forests were more extensive in 1998 than they were at the time of reserve
       creation. Urchin-dominated barrens occupied only 14 % of available reef substratum in reserves as
       opposed to 40"4 in unprotected areas. These changes in community structure, which have persisted
       since at least 1994, demonstrate not only higher trophic complexity than anticipated in Australasian
       ecosystems but also increased primary and secondary productivity in marine reserves as a conse-
       quence of protection. Trends inside reserves indicate large-scale reduction of benthic primary produc-
       tion as an indirect result of fishing activity in unprotected areas.

       KEY WORDS- Marine reserves . Habitat change . Trophic structure - Fishing effects . Urchins . Kelp .
       Fish . Spiny lobsters




           INTRODUCTION                 indirect response by marine ecosystems to renewed
                                 abundance of these species. Algal succession on trop-
  'No-take' marine reserves are increasingly being        ical coral reefs has been shown to vary according to
promoted as a means of managing coastal resources.        types of grazing pressure in marine reserves, with
This is true not only with respect to restoring or pro-      more diverse and abundant algal turfs in mar-
tecting exploited species, but also as a way of pro-       ine reserves primarily grazed by herbivorous fish
tecting marine ecosystems from the indirect effects of      (McClanahan 1997). Intertidal comnlunities in Chile
fishing (Agardy 1994, Botsford et al. 1997, Allison et      have responded to protection from harvesting with
al. 1998, Pauly et al. 1998). It has been demonstrated      marked structural changes (Duran & Castilla 1989). In
that exploited species are more abundant within pro-       the absence of the predatory gastropod Concholepas
tected areas than outside of them (Bennett & Attwood       choncholepas, mussels dominated the intertidal zone,
1991, Rakitin & Kramer 1996, Russ & Alcala 1996,         but inside a reserve, predation of mussels by C.
Millar & Willis 1999), but much less is known of the       choncholepas allowed quite a different community
                                 dominated by barnacles and algae to develop. Similar
                                 information from subtidal temperate systems is sparse,
                                 but what little is known about indirect responses of

0 Inter-Research 1999
Resale of full article not permitted
                     Mar Ecol Prog Ser 189. 125-134, 1999



ecosystems to predator increases has had a consider-               METHODS
able influence on marine ecological theory. Perhaps
the best known example of this is the interaction       Study areas. The study areas were the 2 oldest
between sea otters, sea urchins and kelp (Estes &      marine reserves in New Zealand (Fig. l ) , the Cape
Palmisano 1974, Duggins 1980).               Rodney to Okakari Point Marine Reserve (hereafter
  The importance of the 3-tiered sea otter-urchin-kelp   the Leigh Marine Reserve, 549.16 ha, established
trophic cascade was demonstrated after sea otters      1975) and Tawharanui Marine Park 15 km to the south
were all but wiped out by harvesting for their fur,     (350 ha, established 1982).Non-reserve reference sites
allowing their prey, sea urchins, to overgraze kelps    were located on similar areas o coast adjacent to the
                                               f
and dominate many benthic ecosystems. After the re-     reserves. Sampling effort was spread evenly across
population of areas by otters, kelp and its associated   both reserve and non-reserve areas by dividing them
communities became much more abundant. Recent        into a number of sites and then sampling haphazardly
declines in otter populations in the Aleutian Islands    within them. This was done to avoid biases that might
have confirmed their importance in that ecosystem      be associated with human activities in the Leigh
(Estes et al. 1998). Similar processes, involving fish   marine reserve (e.g. fish feeding, cf. Cole 1994). The
predators, are thought to be present in Mediterranean    fact that fish feeding is not common at Tawharanui
benthic communities (Sala et al. 1998) although other    further reduces the chance o any such biases affecting
                                             f
physical and ecological factors are also likely to be    the result. Abundances o predators such as demersal
                                            f
important. Trophic cascades involving predation by     fish (primarily the sparid snapper Pagrus auratus) and
fish and lobsters on urchins have been proposed to     spiny lobster Jasus edwardsii, and differences in ben-
exist in North Atlantic subtidal ecosystems, but remain   thic community structure were compared between
controvers~al (Chapman & Johnson 1990,Elner & Vadas     reserve and non-reserve areas o coast. Snapper and
                                                f
1990). Furthermore, it has been suggested that in      the spiny lobster are heavily targeted by commercial
southern hemisphere subtidal habitats a much simpler    and recreational line and trap fishing; however the
2-tiered trophic cascade exists (Steinberg et al. 1995),  New Zealand urchin Evechinus chloroticus is not har-
in which predation on urchins is not intense enough to   vested in significant numbers in this region. Popula-
control grazer populations (Andrew & Choat 1982,      tions of algae and echinoid grazers at Leigh in 1994
Andrew & MacDiarmid 1991).                 and 1996 were also compared with data collected in
  Much of the early subtidal marine ecological work in   1977-78, shortly after the area was protected.
north.eastern New Zealand has been carried out in the     Fish abundance. Estimates of predatory fish abun-
Leigh Marine Reserve (Choat & Schiel 1982, Schiel     dance and size were made during October and
1982, Creese & Jeffs 1993). Surveys at the time of     November 1997 using remotely deployed baited video
reserve establishment described 3 major habitat types.   stations. A vertically oriented video camera was
These were the shallow (0 to 4 m) 'mixed algae' zone,   mounted over an enclosed bait consisting of 4 whole
urchin barrens or 'rock-flats' dominated by crustose    pilchards Sardinops neopilchardus with 1 additional
coralline algae and maintained by the common sea      pilchard tied externally to the bait container. The stand
urchin Evechinus chloroticus (4 to 10 m), and deeper    was deployed on sandy substratum adjacent to reef
(10 to 20+ m) kelp forests dominated by the laminanan   areas at depths of 16 to 24 m. Fish responding to the
Ecklonia radiata (Schiel 1990).              bait were filmed during 30 min deployments. The max-
  We asked the following questions: (1) Has the abun-   imum number of fish observed during each deploy-
dance of predators in protected areas increased?      ment was determined by viewing the video recording,
(2) Has the abundance of invertebrate grazers de-     and this value was then used as an index of abundance
creased? (3) Are changes in grazer abundance caused     (Willis et al. unpubl.). The fish present in the field of
by predator abundance? (4) Has the abundance of       view at the time of maximum abundance were mea-
macroalgae increased? (5) Are changes in the abun-     sured by calibrating their images using marks on the
dance of macroalgae caused by changes in grazer       video frame. This method provided estimates of total
abundance? Increased density and size o predators in
                      f        length. The accuracy o length estimates was assessed
                                          f
areas protected from fishing should result in high     at + c20 mm using plastic model fish of known sizes,
levels of predation on urchins, and reduced grazing     deployed under the video stand in situ. Four replicate
should result in more extensive and denser kelp popu-    deployments were made in each of 12 sites (each rep-
lation~ (Schiel 1990).In addition to demonstrating indi-  resenting approximately 500 m o coastline) at Leigh
                                                f
rect ecological effects as a result of marine reserve    and 6 sites at Tawharanui, a total of 72 deployments or
status, the presence of such effects would represent    36 h of video. There were equal numbers of reserve
evidence for the existence of a 3-tiered trophic cascade  and non-reserve sites at both areas. Gut content data
in an Australasian temperate reef ecosystem.        were obtained for fish from the Leigh Marine Reserve
                  Babcock et al.: Habitat change in marlne reserves                127




that suffered mortality incidental to a separate concur-   classifications were made on the following basis; kelp
rent project (Millar & Willis 1999).             forest: > 3 adult phaeophytes m-', e.g. Ecklonia radi-
  Spiny lobster abundance. Lobster surveys were con-     ata or Carpophyllum flexuosum; rock-flats: > 50 %
ducted at 2 sites within and 2 sites outside each of the   cover of crustose coralline algae and < 3 adult phaeo-
reserves. Sites at Leigh and Little Barrier Island were    phytes m"; shallow mixed fucoid zone: >20% algal
sampled in April, and at Tawharanui and Kawau         cover or > 3 adult phaeophytes m-2 at depths less than
Islands in October 1995 (Fig. 1).At each site 5 haphaz-    4 m; turf flats: >50% cover of turf-forming red or
ardly placed 50 X l 0 m transects were surveyed, giving    green algae with large phaeophytes < 3 m-2; bare
a total of 80 transects covering 4 ha of sea floor. The    rock: >50% of surfaces devoid of any algae. Data
choice of this transect size was made based on the      were analysed as the proportions of each habitat over
study by MacDiarmid (1991). He compared the preci-      the entire transect.
sion of 3 different sized transects; 10 m X 10 m, 25 m X    Temporal change in the density of kelp and urchins
10 m and 50 m X 10 m, in estimating lobster abundance     in the Leigh Marine Reserve. Measurements of the
within the Leigh Marine Reserve and found it to be      abundance and size of kelp and urchins made in pre-
similar in all cases. The larger transect size used was    scribed areas of the Leigh reserve in 1977-78 ('perma-
therefore selected to reduce the likelihood of obtaining   nent quadrats' Ayling 1978) provided a direct basis for
zero counts in low abundance areas. Two depths ( < l 0    assessing temporal change. These data were obtained
and 10 to 25 m) were sampled at each site to allow for    in a range of habitat types using haphazard 1 m2
seasonal changes in the depth distribution of lob-
sters (MacDiarmid 1991). Sampling was limited
to suitable lobster habitat such as broken boulder
areas and/or fractured reef (MacDiarmid 1991).
All sites were dominated by laminarian and
fucalean kelp forests and urchin Evechjnus
chloroticus zones at shallower depths.
  Within each transect the size of every lobster
was recorded. A visual size estimation method
was used (MacDiarmid 1991), with divers esti-
mating carapace length (CL) to within 5 mm
without capturing or handling individual lob-
sters. This level of accuracy was achieved
through a series of calibration dives where the                        Tawharanui Marine
size of individual lobsters was first estimated by
each censor, after which each lobster was caught
by hand and measured with vernier calipers to
obtain a true length measurement.
 Algal abundance. To test whether urchin-
dominated rock-flats (where kelp had been
removed by grazing) were less extensive inside
protected areas we conducted a series of mea-
surements on 65 transects at 21 sites in and
around the 2 reserves in December 1997. The
sites used were the same as those for the esti-
mates of fish abundance and an additional
3 transects were laid at Kawau Island (Fig. l ) ,
2 km south of Tawharanui. Transects were 1 m
wide and perpendicular to the shore, extending
from 0 to 10 m below chart datum. Three to 4
transects, marked by a weighted line, were
located haphazardly within sampling sites.
Habltat type under the transect was classified by
visual estimate at 1 m intervals as either shallow
                            Fig. 1. Map of the study sites in northeastern New Zealand. Dashed
mixed fucoid, urchin-dominated rock-flats, kelp    line: Marine Reserve boundary Numbers indicate sampling sites
forest, articulated coralline turf flat, bare rock or  used in fish (1 to 18) and benthic transect (1 to 21) sampling.
sand. Depth was also recorded. Visual habitat          Shaded areas at sea are spiny lobster samphng sites
128                     Mar Ecol Prog Ser 189: 125-134, 1999




quadrats within 26 defined 100 m2 areas. The same             Primary productivity. Estimates o primary produc-
                                                        f
measurements were repeated as close as possible to 9          tivity were calculated based on habitat maps (Ayling
of these areas in 1994 and 14 of them in 1996, using the        1978) and comparisons between our 1996 data and the
detailed maps provided by Ayling (1978). All of these          data provided by Ayling (1978) for the density and
areas were classified by Ayling (1978) as either kelp          population structure of Ecklonia radiata in kelp forest,
forest or rock-flats. Samples from 1994 and 1996 were          rock-flats and shallow mixed algal habitats. Each area
estimated to be within a 20 m radius of the original          was classified according to the habitat type in which it
1978 areas, and were in the same depth range as             fell during the 1978 survey, and productivity values
Ayling's areas, as indicated on his habitat maps            (g dry weight m-2 yr-l) were calculated. For each of
(Ayling 1978).In each of these areas, a total of sixteen        these habitat types the primary productivity of E. radi-
l m2 quadrats were measured to determine the density          ata sporophytes was estimated according to Novacek
and population structure o Ecklonia radiata and the
              f                      (1980), who obtained size-specific productivity mea-
density of Evechinus chloroticus. The size of E. radiata        sures for individual plants. E. radiata populations were
was measured as stipe length and basal stipe diameter,         divided into 4 size classes: I: basal stipe diameter
though only stipe length was used in subsequent pro-          c10 mm; 11: basal stipe diameter 10 to 16 mm; 111: basal
ductivity estimates. The size of E. chloroticus was mea-        stipe diameter > l 6 mm, stipe length <700 mm; IV:
sured as test diameter.                         basal stipe diameter > l 6 mm, stipe length >?00 mm.


rable 1. Change in density of kelp Ecklonia radiata and urchins Evechinus chloroticus at sites in the Leigh Marine Reserve.
Means and SE for 16 quadrats. Numbering for sites listed under rock-flat and kelp forest habitat correspond to 'permanent
quadrat' numbers of Ayling (1978). Bold: number of simulations (of 100) in which the 1994 or 1996 populations were found to be
significantly different to populations in 1978 ( a = 0.05) 'Change in abundance of fucoid algae similar in magnitude to that of
             E. radiata (Shears & Babcock unpubl. data). Blank cells were not sampled in 1994

Habitat type in           E. radiata m-'                       E. chloroticus m-*
1978           1978      1994      1996             1978      1994      1996

Rock-flats
1            0.6(0.5)            2(1.1)             5.3(1)             l(2.6)
                              11'                              81
2             3(0.3)            6.4(2.8)            2.1(0.5)             2(2.4)
                              99'                              6
3            0.5(0.3)           8.9(1.?)            3.7(0.7)            2.3(2.9)
                              99'                              40
4            0.8(0.4)    1.8(1.6)    5.6(1.9)            9.6(1.5)    1.4(1.6)     3.3(5)
                      100      98'                      90       60
6              0      11.3(8.6)   13.2(4.9)           3.8(0.6~    l(1.6)       0
                      100       100                              100
7              o      15.4(9)    13.4(6.9)            4(o.a)     0.5(1.a)      0.8
                      100      100                              100
9              0     24.3(15.1)   13.3(5.6)            2.5(0.3)   0.07(0.25)      0
                      100      100                              100
1.5           2.9(1..4)          11 3(2.3)            4.7(0.7)            1.2(0.9)
                             100                              85
23             0              0.13              3                0
                              10                              100
22             0             19.9(6.3)           10.3(1.2)            3.4(0.6)
                             100
Grand mean        0.8(1.1)   13.2(9.3)    9.4(5.9)           4.9(2.7)             1.4(1.3)
Kelp forest
16            17.5(2)    6.3(0.7)   11 l ( 0 . 9 )          0.4       0
                              87                      100
17           18.3(3.4)    3.9(0.8]    12.7(1)             0.3       0
                             44                      100
18           25.5(2.1)    2.4(0.7)    9.5(1.5)            0.09       o
                             100                      100
19           17.1.(1.91   4.7(1.2)    9 g(0.9)            0.7       0
                              91                      100
Grand mean       19.6(3.9)   4.3(1.6)    10.8(1.4)           0.38(0.13)     0
                    Babcock et al.: Habitat change in marine reserves                     129




Table 2. Habitat change and primary production at the Leigh reserve, 1978 to 1996. Total area (549.16 ha) and proportional rep-
resentation of subtidal rocky habitat within the reserve were estimated from Ayling (1978). Primary production estimates were
based on our 1996 density and size-frequency data for Ecklonia radiata populations and published estimates of size-specific
    primary productivity for E. radiata (Novaczek 1980) and temperate crustose coralline algae (Littler & Arnold 1982)


Habitat type                 1978                          1996
                % Total area    Primary production         % Total area    Primary production
                          (tonnes dry wt yr.')                  (tonnes dry wt yr-'1

Shallow mixed algae         22.2         67.6              34.7         105.8
Rock-flats             31.4         6.8               3.2          0.7
Ecklonia radiata forest       29.8        203.6              45.5         333.2
Total                83.4        278.0              83.4         439.6




Using published values for primary productivity of         and, at least in theory, are not subject to fishing mor-
temperate crustose coralline algae (CCA) (Littler &        tality. Plots of the residuals against predicted or fitted
Arnold 1982), the productivity of CCA-dominated rock        values and a normal quantile-quantile, or qqplot, of the
flats was also calculated on a per-unit area basis. These
data were used to produce estimates of primary pro-
ductivity for each of the main habitat types in the Leigh                   Pagrus auratus
                                          1
Marine Reserve based on their estimated projected
area1 extent (Table 2 ) . The estimates of area are there-
fore conservative and assume a flat substratun~.   Data
for each prescribed area in 1978 and 1996 were aver-
aged for each habitat type and used to calculate pri-
mary productivity and change in total productivity of
the main reserve habitats.

                                               Jasus edwardsii
              RESULTS

           Fish abundance

 Pagrus auratus was the most abundant demersal
species, making up 54 % of the demersal fish observed
in video drops. At both Tawharanui and Leigh P, aura-
tus were recorded at significantly greater abundances
within reserves than outside (Fig. 2). Conservative
estimates of the magnitude of difference between
reserve and non-reserve areas were made by compar-
ing the lower 95% confidence bounds of the reserve
estimate with the upper 95 % confidence bound of the
non-reserve estimate. Using this method, Tawharanui
and Leigh P. auratus (> minimum legal size) abun-
dances were at least 5.75 and 8.70 times greater within
reserves, respectively, than outside. Abundance of P.                    Reserve Non reserve
auratus > minimum legal size (270 mm fork length)
                                  Fig. 2. Effects of no-take reserves. Predator abundance (snap-
was formally analysed using a mixed model ANOVA          per Pagrus auratus, spiny lobster Jasus edwardsii) and pro-
with Area (i.e. Leigh or Tawharanui) and Status (i.e.       portional habitat distribution inside and outside reserves.
reserve or fished areas) treated as fixed factors, and       Data are means ( + 9 5 % CI). Open bars: Leigh Marine
Site a nested factor within area and status. Sub-legal P.     Reserve; solid bars: Tawharanui Marine Park. Max snapper
                                  data for snapper refer only to fish greater than 270 mm mini-
auratus were not analysed because they do not feed on       mum legal length (Max snapper = maximum number of snap-
Evechinus chloroticus (Kingett & Choat 1981), vary         per in the camera field during one 30 min deployn~ent.  See
seasonally regardless of reserve status (Francis 1995)               text for full description of index)
 130                     Mar Ecol Prog Ser 189:  125-134, 1999




residuals indicated that the assumptions of equality of     action terms that included Depth were significant due
variances and normality were met. Differences with        to the seasonal movement of lobsters between depth
Status were statistically significant (F,,,, = 11.16, p <    strata (see MacDiarmid 1991) and the fact that Leigh
0.01), whereas the interaction term (Fl,14 2.58, p >
                        =         (April)and Tawharanui (October)were sampled at dif-
0.1) was not. Fish abundance did not vary significantly     ferent times of year. The mean density for the non-
between areas (F,,,, = 3.01, p > 0.1) but among-site       reserve sites was 174 + 38.8 (95% CI) lobsters ha-'
variability was high     = 4.58, p < 0.05).         compared with 455 + 60.6 (95% CI) in the protected
 Size of Pagrus auratus inside reserves was much        sites.
greater than outside them, with a mean total length in       Likewise, the mean size of lobsters was significantly
reserves of 327 + 9.5 mm (95% Cl) for 147 fish, com-       larger in reserves compared to unprotected sites. Lob-
pared with 191 i 7.6 mm (95% Cl) for 67 fish in non-       sters within the reserves (n = 909) had an adjusted
reserve areas. The size of all P. auratus was assessed      mean carapace length of 109.9 + 1.7 mm (95% Cl),
using a 3-way ANOVA with Area, Status and Depth         compared with 93.5 + 2.8 mm (95% Cl) outside the
treated as fixed factors. Diagnostic plots of ln-trans-     reserves (n=348).A 3-way ANOVA was used to exam-
formed data indicated that the model assumptions         ine the effect o protection on lobster size with Status,
                                             f
were met. No significant differences were found with       Area and Depth treated as fixed factors. Plots of the
Area     = 1.1, P > 0.05), Depth    = 0.5, P = 0.88)   residuals against predicted or fitted values and a
or interaction terms, but Status was statistically signifi-   normal quantile-quantile, or qqplot, of the residuals
cant     = 34.3, p < 0.01).                 indicated that the assumptions of equality of variances
                                 and normality were met. Significant differences were
                                 found between Area              = 17.63, p < 0.01), Depth
        Fish predation on urchins            ( F l , , 8 = 2 . 9 9 , p < 0 . 0 5 ) a n d S t a t u(F1,8=34.69,p<0.01).
                                                           s
                                 Jasus edwardsii is known to feed on Evechinus
 Analysis of gut contents of Pagrus auratus caught on      chloroticus (Andrew & MacDiarmid 1991) and dis-
coastal reefs in the area indicated that Evechinus        carded urchin tests are common at lobster den sites
chloroticus form a conspicuous part of its diet. Analysis    (authors' pers. obs).
of gut contents of fish from the Leigh reserve showed
that l l of 68 P. auratus as well as 3 of 6 Parapercis colias
had recently fed upon E. chloroticus. The smallest fish                 Algal abundance
found to have fed on E. chloroticus was 299 mm and the
largest 665 mm fork length. Sizes of intact or partially      The mean proportion o the substratum occupied by
                                              f
intact tests taken o P, auratus were 6, 8, 9, 15, 33, and
           f                     urchin-dominated rock-flat habitat at depths < 10 m
-60 mm. Based on the number of urchin pyramids         differed significantly between reserve and non-
found, there were at least 25 other urchins in the guts of   reserve areas, averaging 39 & ? % (95% CI) outside
these fish. Test diameter of these urchins was estimated    reserves, and 13 + 5.9% (95% CI) inside reserves
using regression equations relating test diameter to      (Fig. 2). Proportional cover of rock-flat habitat was arc-
pyramid length (Kerrigan 1987).The average size of all     sine-transformed and examined using a mixed model
urchins taken was 26.1 rt 8.28 mm (95% CI) There was      ANOVA with Area (i.e. Leigh or Tawharanui) and Sta-
also one 5 mm intact test found in a P, colias.         tus (i.e.reserve or fished areas) treated as fixed factors,
                                and Site (nested within area and status) as a random
                                factor. Diagnostic plots indicated that the model
        Spiny lobster abundance             assumptions were met. Differences with Status were
                                statistically significant (Flsl7 13.21, p < 0.01), whereas
                                                =
  Mean densities of Jasus edwardsii were also sig-       those due to the interaction term (F1,,, 3.4, p = 0.08)
                                                      =
nificantly greater within the Tawharanui and Leigh       were not. Rock-flat coverage did not differ signifi-
reserves compared to their respective unprotected        cantly between Areas (F,,, = 0.86, p = 0.37) but there
sites (Fig. 2). Abundance was examined with a mixed       was significant patchiness at the Site level       =
model ANOVA with Status, Area and Depth treated as       6.05, p < 0.01).
fixed factors and Site as a random factor. Counts were
ln(x + 1) transformed prior to analysis and diagnostic
plots indicated that the model assumptions were met.       Temporal change in the density of kelp and urchins
Differences in Status and Site were found to be signif-          in the Leigh Marine Reserve
icant (Fln8 25.26. p < 0.01 and
      =              = 4.92, p < 0.01
respectively) but Depth and Area were not     = 0.37,    By 1996, 8 of 10 areas that were initially classified
p = 0.69 and F1,8 0.15, p = 0.54 respectively). All inter-
         =                       as urchin-dominated rock-flats were re-classified as
                   Babcock et al.: Habitat change in marine reserves                    131




either kelp forest or shallow mixed fucoid habitat due
to increased densities of Ecklonia radiata (Fig. 3 ,
Table l),and/or large fucoid algae (mainly Carpophyl-
l u m plumosum and C. flexuosum). Mean densities of E.
radiata were 0.8 * 1.1 m-2 SE in 1976, but for the same
10 sites in 1996, mean densities increased to 9.4 *
5.9 m-2 SE. Four of these former rock-flat areas have
been kelp forest since at least 1994 (Table 1). There
have been corresponding declines in the overall abun-
                      *
dance of Evechinus chloroticus from 4.9 2.7 m-2 SE to
  *
1.4 1.3 m-' SE at the same sites (Fig. 3, Table 1).
While E, ckloroticus was still found at most o the for-
                         f
mer rock-flat sites in 1994 and 1996, i.t was present at
much lower densities than in 1978 (Table l ) , and
changes in density of both urchins and algae were sta-
tistically significant. Mean densities for each o the 10
                          f
sites originally classified as rock-flats were used to
compile a data set that could be used to compare habi-
tat changes between 1978 and 1996. Data for E. radiata
and E. chloroticus were compared using a l-way
ANOVA which indicated significant differences in E.
chloroticus density (Flsl812.6,p 0.01), as well as the
              =
density of E. radiata populations (F],1819.9, p < 0.01).
                     =
  In addition to these comparisons using means for
each site, densities from 1978 (available only as mean
                                Fig. 3 . Habitat change at Leigh 1978 to 1996. (A) Sites classi-
and SE, Ayling 1978) were compared with data from
                                fied a s urchin-grazed rock-flats in 1978, (B) sites classified a s
1996 for each individual site. This was done using       kelp forests in 1978 (Ayling 1978). Data a r e means (+95% CI).
Monte Carlo sirnulations (POWER.Borenstein & Cohen       Open bars: Ecklonia radiata; solid bars: Evechinus chlor-
1988), assuming normally distributed error structures                    oticus
based on data from the 1996 surveys, for which we had
full data sets. For both Evechinus chloroticus and Eck-
lonia radiata 100 simulations were assessed at a < 0.05.    area. We designated the changed areas < 6 m as shal-
For E. chloroticus the probability o a statistically signif-
                  f             low mixed algae and those deeper than 6 m as kelp
icant decrease in density was >0.8 at 7 of the 10 sites.    forest. Both o these habitat types expanded apprecia-
                                       f
There was an increase in density o E. radiata at 8 o the
                  f         f    bly over the period 1978 to 1996, at the expense of
10 sites; however a decrease in density of E. chloroticus   rock-flat habitat. Consequently the total estimated pri-
did not always correspond directly to an increase in E.    mary productivity of each o these habitats increased
                                              f
radiata (e.g. quadrats 1 and 23, Table 1).None of the 4    by 57% for shallow mixed algal areas and 64% for
kelp forest sites had changed community structure       kelp forest. Overall, the total primary productivity of
noticeably. Nevertheless there was a decrease in abun-     the rocky reef habitats we examined increased by 58 %
dance of kelp at 3 of the 4 kelp forest sites (Table 1).    from 278 to 439 t dry weight yr-' (Table 2).
While urchins were present at low densities in kelp for-
est sites in 1978, they were completely absent from kelp
forest sites during our study.                              DISCUSSION

                                 Greater abundances of Jasus edwardsii that we
         Primary productivity             found inside reserves are consistent with earlier stu-
                                dies (Cole et al. 1990, MacDiarmid & Breen 1993).We
  Between 1978 and 1996, 80% of rock-flat habitat       also found much larger increases in the relative abun-
sites changed to kelp forest or shallow mixed algal      dance of Pagrus auratus than have previously been
habitat, based on changes in density of Ecklonia radi-     suggested (Cole et al. 1990),largely because we used a
ata at prescribed areas in the Leigh Marine Reserve.      method of censusing fish that overcame some o the f
Consequently we estimate that the total area of rock-     biases that confounded the visual survey results o   f
flats in the reserve has shrunk by an order of magni-     Cole et al. (1990).These increases are consistent with a
tude from 31.4 to 3.2 % of the total available rocky reef   release from fishing pressure.
 132                   Mar Ecol Prog Ser




  Both Jasus edwardsii and Pagrus auratus are known      the shift in community structure we have observed is
to feed on the dominant grazer in New Zealand ben-       not as pronounced as reported in southeastern Alaska,
thic reef communities, the echinoid Evechinus chlor-      it is similar to the situation found in the western Aleu-
 oticus. Higher E, chloroticusmortality within the Leigh    tians (Estes & Duggins 1995), where high rates of
reserve, relative to non-reserve areas, has been        urchin recruitment appear to reduce the rate and
inferred by Cole & Keuskamp (1998). Urchins trans-       degree of habitat transition.
planted to the reserve were reduced by between 10         Our results also suggest how no-take manne reserves
and 90 % after 6 d , while over the same period > 80 % of   can change benchmarks for environmental and fish-
urchins transplanted to areas outside the reserve       eries management. The recovery of kelp populations
remained (Cole & Keuskamp 1998). Furthermore E.        subsequent to protection indicates that fishing activities
chloroticus populations in marine reserves tended to      on New Zealand's northeastern coast have had ecolog-
have bimodal population structures (Cole & Keuskamp      ical impacts far beyond the target species. These
1998, Figs. 3 & 6). The size of urchins we found in      effects, though indirect, are just as striking as those
the guts of snapper from the Leigh reserve (26.1 +      caused by fishing practices such as trawling, which
8.28 mm) fell neatly between these 2 modes. Measure-     remove or destroy conspicuous components of fished
ments at a number of sites in the Leigh Marine Reserve    habitats (Rothschild et al. 1994, Dayton et al. 1995,
showed that densities of E. chloroticus have declined     Watling & Norse 1998). Perhaps more importantly in
since 1978.                          the case of kelp forests, these conspicuous components
  The fact that proportional cover of urchin-grazed     are also major primary producers (Mann 1973),whose
rock-flat habitats in the 2 marine reserves studled is    contribution to detrital food webs is central to ecosys-
significantly lower than in unprotected areas is likely    tem function and diversity (Duggins 1980).Using habi-
a n indirect consequence of this increase in predator     tat maps and surveys of the Leigh reserve made in 1978
abundance and reduced grazer density. Experimental       (Ayling 1978) and 1996, we have estimated that
removals of Evechinus chloroticus (Andrew & Choat       macroalgal primary productivity is approximately 58 %
1982) have shown that reduced grazer densities will      greater within the Leigh reserve than it was 20 yr ago.
lead to increased macroalgal cover, including Ecklonia    The corollary of this is that benthic primary productivity
radiata. In one of the reserves, E. radiata forest has    in areas outside of reserves is much lower than it was
expanded dramatically over a 20 yr period into areas     prior to intensive fishing. It is not yet clear how such
that were formerly urchin-grazed rock-flats. In con-     fishing-mediated reductions in primary productivity
trast, all areas that were classified as kelp forest in    flow through to different trophic levels, but data from
1978 remained stable as kelp forest in 1996 (Fig. 3).     kelp forests in the northeast Pacific indicate that their
This was despite substantial episodic mortalities of E.    contribution could be substantial (Duggins et al. 1984).
radiata during that time (Cole & Babcock 1996, Easton     Kelp forests in northeastern New Zealand are known to
et al. 1997). We propose that higher predation upon      have far higher rates of secondary production than do
and mortality of the urchin E. chloroticus inside the     rock-flat habitats (Taylor 1998),supporting the sugges-
Leigh reserve is the cause of the observed changes in     tion that these effects are likely to be broadly felt
community structure within the Leigh and Tawharanui      throughout the ecosystem.
Marine Reserves.                         These observations indicate that the efficient man-
  Indirect ecological changes such as those we have      agement of coastal fisheries can no longer ignore the
observed have interesting implications because it has     consequences of fishing on the wider ecosystem, and
been suggested that Australasian kelp ecosystems        attempts are now being made to understand benthic
have always lacked 'keystone' predators (Andrew 81      ecosystems in terms of multi-specles fisheries models
Choat 1982, Andrew & MacDiarmid 1991).As a conse-       (McClanahan & Sala 1997). No-take marine reserves
quence of this conclusion it has been hypothesised that    represent a large scale ecological tool that can play an
Australasian reefs have been dominated by urchins,      important part in the development of such models.
and algae there have developed heavier chemical        Unfished areas provide a way of assessing ecosystem
defenses than their northern hemisphere counterparts     function and avoiding the 'sliding baseline' phenome-
(Steinberg et al. 1995). Further evaluation of these     non, in which expectations of what is natural are
ideas is required in light of our data which suggest that   reduced because many of the original components of
predation can play a key regulatory role in New        the system are reduced or absent (Dayton et al. 1998).
Zealand kelp ecosystems. The protection and reintro-     It has been suggested that the results of ecological
duction of sea otters in parts of their former range in    studies conducted in manne reserves cannot be
the northeastern Pacific (Estes & Palmisano 1974, Dug-    extrapolated to other, fished, areas (Cole & Keuskamp
gins 1980) present a situation analogous to the one we    1998).We suggest that a more useful perspective is not
have described in northeastern New Zealand. While       to extrapolate and use ideas based on exploited areas
                    Babcock et al.: Habitat change In marine reserves




as the basis for understanding the fundamentals o   f       marine reserves. In: Battershill CN, Schiel DR. Jones GP,
benthic ecosystem function. By establishing knowl-          Creese RG. MacDiarmid AB (eds) Proceedings of the 2nd
                                   International Temperate Reef Symposium. NIWA Marine,
edge of ecosystem function based on both protected
                                   Wellington, p 15-45
and exploited areas, ecology will be more likely to con-     Dayton PK, Thrush SF, Agardy MT, Hofman RJ (1995) Envi-
tribute to the judicious use of marine ecosystems and        ronmental effects of marine fishing. Aquat Conserv: Mar
the preservation of their many attributes.              Freshw Ecosyst 5:205-232
                                 Dayton PK Tegner MJ, Edwards PB, Riser KL (1998) Sliding
                                   baselines, ghosts, and reduced expectations in kelp forest
                                   communities. Ecol Appl 8:309-322
Acknowledgements. Thanks to G Branch, R. Creese, J . Estes,   Duggins DO (1980) Kelp beds and sea otters: a n experimental
D. Schiel, P. Steinberg, M. Tegner and 3 anonymous review-      approach. Ecology 6:447-453
ers for cornrnents on the manuscript M. Birch, B. Doak, S.    Duggins DO, Simenstad SA, Estes J'A (1984) Magnificat~ono   f
Lyons and L. Munro assisted with data collection. This work     secondary production by kelp detritus in coastal marine
was funded in part by the New Zealand Dept. of Conserva-       ecosystems. Science 245170-173
tion, Investigation Nos. 1946, 1535 and 2174.          Duran LR, Castilla J C (1989) Variation a n d persistence of the
                                   middle rocky intertidal community of central Chile, with
                                   a n d without human harvesting. Mar Biol 103:555-562
          LITERATURE CITED               Easton LM, Lewis GD, Pearson MN (1997) Virus-like particles
                                   associated with dieback symptoms in the brown alga
                                   Ecklonia radiata. Dis Aquat O r g 30:21?-222
Agardy MT (1994) Advances in marine conservation: the role    Elner RW, Vadas RL (1990) Inference in ecology: the sea
  of marine protected areas. Trends Ecol Evol9:267         urchin phenomenon in the northwestern Atlantic. Am Nat
Allison GW, Lubchenco J , Carr MH (1998) Marine reserves       136:108-125
  are necessary but not sufficient for marine conservation.  Estes JA, Duggins DO (1995) Sea otters and kelp forests in
  Ecol Appl 8:579-592                       Alaska: generality and variation in a community ecologi-
Andrew NL, Choat J H (1982) The influence of predation and      cal paradigm Ecol Monogr 65:75-100
  conspecific adults on the abundance of juvenile Evechinus  Estes JA, Palmlsano J F (1974) Sea otters: their role in struc-
  chloroticus (Echinoidea: Echnodermata). Oecologia 54:      turing nearshore communities. Science 185:1058-1060
  80-87                            Estes J . Tinker MT, Williams TM, Doak DF (1998) K~ller  whale
Andrew NL. MacDiarmid AB (1991) Inter-relations between       predation on sea otters linking oceanic and nearshore sys-
  sea urchins and spiny lobsters in northeastern New        tems. Science 282:473-476
  Zealand. Mar Ecol Prog Ser 70:211-222            Francis MP (1995) Spatial and seasonal variation in the
Ayling AM (1978) Cape Rodney to Okakari Point Marine         abundance of juvenile snapper (Pagrus auratus) in the
  Reserve Survey. Leigh Marine Laboratory Bulletin 1,       northwestern Hauraki Gulf. NZ J Mar Freshw Res 29:
  Leigh Marine Laboratory, Auckland                565-579
Bennett BA, Attwood CG (1991) Evidence for recovery of a     Kerrigan B (1987) Abundance patterns of intertidal and subti-
  surf-zone fish assemblage following the establishment of a    dal populations of Evechinus chloroticus. MSc thesis, Uni-
  marine reserve on the southern coast of South Africa. Mar    versity of Auckland
  Ecol Prog Ser 75:173-181                   Kingett PD, Choat J H (1981) Analysis of density and distribu-
Borenstein M, Cohen J (1988) Statistical power analysis: a      tion patterns In Chrysophrys auratus (Pisces. Sparidae)
  computer program. Lawrence Erlbaum Ass Inc, Hillsdale,      within a reef environment: an experimental approach
  NJ                                Mar Ecol Prog Ser 5:283-290
Botsford LW, Castilla JC, Peterson CH (1997) The manage-     Littler MM, Arnold KE (1982) Primary productivity of manne
  ment of marine ecosystems. Science 277.509-515          macroalgal functional-form groups from south\vestern
Chapman ARO, Johnson CR (1990) Disturbance and organi-        North America. J Phycol 18:307-311
  zation of macroalgal assemblages in the northwest      MacDiarmid AB (1991) Seasonal changes in depth distribu-
  Atlantic. Hydrobiologia 192:7?-121                tion, sex ratio and size-frequency of spiny lobster Jasus
Choat JH, Schiel DR (1982) Patterns of distribution and abun-    edwardsii on a coastal reef in northern New Zealand. Mar
  dance of large brown algae and invertebrate herbivores in    Ecol Prog Ser 70:129-141
  subtidal regions of northern New Zealand. J Exp Mar Biol   MacDiarmid AB, Breen PA (1993) Spiny lobster population
  Eco160:129-162                          change in a marine reserve. In: Battershill CN, Schiel DR.
Cole RG (1994) Abundance, size structure, and diver-oriented     Jones GP, Creese RG, MacDiarmid AB (eds) Proceedings
  behaviour of three large benthic carnivorous fishes in a     of the 2nd International Temperate Reef Symposium.
  marine reserve in northeastern New Zealand. Biol Con-      NIWA Marine. Wellington, p 47-56
  serv 70:93-99                        Mann KH (1973) Seaweeds: their productivity and strategy for
Cole RG, Babcock RC (1996) Mass mortality of a dominant       growth. Science 182:975-981
  kelp (Laminariales) at Goat Island, north-eastern New    McClanahan TR (1997) Primary succession of coral-reef
  Zealand. Mar Freshw Res 47:907-911                algae: differing patterns on fished versus unfished reefs.
Cole RG, Keuskamp D (1998) Indirect effects of protection      J Exp Mar Biol Ecol218:77-102
  from exploitation: patterns from populations of Evechinus  McClanahan TR, Sala E (1997) A Mediterranean rocky-bot-
  chloroticus (Echinoidea) in northeastern New Zealand.      tom ecosystem fisher1.e~ model. Ecol Model 104:145-164
  Mar Ecol Prog Ser 1?3:215-226                Millar RB, Willis TJ (1999) Estimating the relative density of
Cole RG, Ayling AM, Creese RG (1990) Effects of marine        snapper in and around a marine reserve using a log-linear
  reserve protection at Goat Island, northern New Zealand.     mixed effects model. Aust N Z J Stat 41:383-394
  NZ J Mar Freshw Res 24:19?-210                Novaczek 1 (1980) T h e development and phenology of Eck-
Creese RG, Jeffs (1993) Biological research in New Zealand      lonia radiata. PhD thesis, University of Auckland
                         Mar Ecol Prog Ser 189: 125-134, 1999




Pauly D, Christensen V, Dalsgaard, J , Froese R, Torres F        chloroticus, and the removal of plants from subtidal algal
  (1998) Fishing down marine food webs. Science 279:          stands in northern New Zealand. Oecologia 54:379-388
  860-863                              Schiel DR (1990) Macroalgal assemblages in New Zealand:
Rakitin A, Kramer DL (1996) Effect of a marine reserve on the      structure, interactions and demography. Hydrobiologia
  distribution of coral reef fishes in Barbados. Mar Ecol Prog     19259-76
  Ser 131:97-113                          Steinberg PD, Estes JA, Winter FC (1995) Evolutionary conse-
Rothschild BJ, Ault JS, Goulletquer P, Heral M (1994)Decline      quences of food chain length in kelp forest communities.
  of the Chesapeake Bay oyster population: a century of        Proc Natl Acad Sci USA 92:8145-8148
  habitat destruction and overfishing. Mar Ecol Prog Ser      Taylor RB (1998)Density, biomass and productivity of animals
  11 1:29-39                              in four subtidal rocky reef habitats: the importance of
Russ GR, Alcala AC (1996) Marine reserves: rates and pat-        small mobile invertebrates. Mar Ecol Prog Ser 172:37-51
  terns of recovery and decline of large predatory fish. Ecol    Watling L, Norse EA (1998) Disturbance of the seabed by
  Appl6:947-961                            mobile fishing gear: a comparison to forest clearcutting.
Sala E, Boudouresque CF, Harmelin-Vivien M (1998) Fishing,       Consew Biol 12:1180-1197
  trophic cascades, and the structure of algal assemblages:     Willis TJ, Millar RB, Babcock RC (unpublished) Detection of
  evaluation of a n old but untested paradigm. Oikos 82:        spatial variability in relative density of fishes: comparison
  425-439                               of visual census, angling, and baited unterwater video.
Schiel DR (1982) Selective feeding by the echinoid Evechinus      Mar Ecol Prog Ser

Editorial responsibility: Otto Kinne (Editor),            Submitted: March 3, 1999; Accepted: June 16, 1999
OldendorfLuhe, Germany                        Proofs received from author(s): November 9, 1999