Content uploaded by Nicholas K Dulvy
Author content
All content in this area was uploaded by Nicholas K Dulvy on Jan 19, 2018
Content may be subject to copyright.
Scale-dependant control of motile epifaunal
community structure along a coral reef
fishing gradient
Nicholas K. Dulvy *, Rebecca E. Mitchell, Douglas Watson,
Christopher J. Sweeting, Nicholas V.C. Polunin
School of Marine Science and Technology, University of Newcastle,
Newcastle-upon-Tyne, NE1 7RU, UK
Received 15 April 2002; received in revised form 11 July 2002; accepted 18 July 2002
Abstract
Large-scale fishing is mostly conducted using towed gears that reduce the biomass and diversity
of benthic invertebrates. However, it is impossible to differentiate between the physical disturbance
effect of towed gears from the effect of fish predator removal upon benthic invertebrate communities.
Here we explore the impact of fish removal alone on the community structure of small motile coral
reef invertebrates (epifauna) along a subsistence fishing intensity gradient in the Lau group, Fiji. We
deployed settlement plates at three areas in each of six fishing grounds and examined the density and
class richness of the motile epifaunal communities and the associated algal communities in relation
to the structure of fish and benthic communities. Motile epifaunal density was unrelated to fishing
intensity. However, at smaller inter-area scale (0.5 –10 km) motile epifaunal density was negatively
related to plate algal biomass, whereas at the larger inter-fishing-ground scale (4– 180 km) motile
epifaunal density was positively related to the rugosity (substrate complexity) of the surrounding
benthos. The class richness and diversity (Margalef’s d) of motile epifaunal communities were
negatively related to fishing intensity, but unrelated to grazing intensity, rugosity or algal biomass at
either scale. Benthic community structure varied significantly with fishing intensity; hard-coral cover
was lower and turf-algal cover was higher at high fishing pressure. The variation in benthic
community structure was associated with variation in fish community structure, which in turn varied
with fishing intensity. Motile epifaunal community structure upon plates was linked to the structure
of the surrounding benthic community, but was not directly linked to the plate algal community. We
suggest the decline in richness of the motile epifauna community along the fishing gradient is
attributable to either to exploiter-mediated coexistence or the reduction in ‘habitat quality’ of the
0022-0981/02/$ - see front matter D2002 Elsevier Science B.V. All rights reserved.
PII: S 0022-0981(02)00327-1
*
Corresponding author. Present address: Lowestoft Laboratory, Centre for Environment, Fisheries and
Aquaculture Science, NR33 OHT, UK.
E-mail address: n.k.dulvy@cefas.co.uk (N.K. Dulvy).
www.elsevier.com/locate/jembe
Journal of Experimental Marine Biology and Ecology
278 (2002) 1– 29
surrounding benthos. At the large spatial scale substrate complexity is the key determinant of motile
epifaunal density, suggesting predation by fishes plays an important structuring role at this scale.
Assuming that rugosity is inversely related to predation risk then this study represents the first
evidence for spatial-dependence on the top-down (predation) vs. bottom-up (algal biomass) control
of community structure. We argue fisheries exploitation, in the absence of a physical disturbance can
negatively influence motile epifaunal community structure at large spatial scales.
D2002 Elsevier Science B.V. All rights reserved.
Keywords: Caribbean; Intermediate disturbance; Beam trawl; Herbivory; Predation risk; Pollution
1. Introduction
Habitat destructive fishing gears such as bottom trawls modify benthic habitats and
reduce the biomass and diversity of invertebrate communities affecting secondary
production at large spatial scales (Kaiser, 1998; Collie et al., 2000; Kaiser et al., 2000;
Jennings et al., 2001a). However, towed fishing gears have two inseparable impacts on
benthic invertebrate community structure: the physical disturbance effect of towed gears
and the removal of predatory target and nontarget fishes. The physical disturbance of
towed gear results in direct mortality or reduced survivorship of benthic invertebrates
(Kaiser and Spencer, 1995; Collie et al., 1997). Fishes can potentially influence benthic
invertebrate community structure via predation pressure (Whitman and Sebens, 1992;
Greenstreet et al., 1997). Exploitation-mediated changes in fish community structure have
also been inferred to result in variation in the taxonomic composition of temperate soft-
bottom invertebrates (Frid et al., 1999). Fishing is known to reduce the biomass of
predatory fishes (Russ and Alcala, 1989; Jennings et al., 1995; Jennings and Polunin,
1995a,b, 1996, 1997); however, an unresolved issue is whether fish removal alone can
influence benthic invertebrate communities, in the absence of the physical disturbance
effect of fishing gears.
In hard substrate systems, such as coral reefs and kelp beds, fishes exert predatory
control on large motile invertebrates ( > 0.5 mm), such as urchins, molluscs and lobsters,
and smaller motile invertebrates ( < 0.5 mm) (McClanahan, 1994, 1995; McClanahan and
Sala, 1997; Sala et al., 1998; Steneck, 1998; Tegner and Dayton, 2000). The exploitation
and depletion of predatory fishes at the top of food webs has led to the proliferation of
large grazing invertebrates, such as urchins, which have in turn altered algal abundance or
community structure at lower trophic levels. The indirect ecological interaction between
predators and the base of the food web via key intermediary species is known as a trophic
cascade (Kitchell and Carpenter, 1993; Pinnegar et al., 2000). While the role of large
motile invertebrates as cascade intermediaries has been well documented, the potential
ecosystem role of small motile invertebrates ( < 0.5 cm) remains largely unexplored in hard
substrate systems, such as coral reefs.
Grazing motile epifauna potentially has an ecosystem role on coral reefs because they
consume 1% of daily epilithic algal standing crop and between 19% and 31% of daily net
areal production (Klumpp et al., 1988; Klumpp and Polunin, 1989; Polunin and Klumpp,
1992) and can have substantial impacts on algal biomass in areas protected from larger
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–292
herbivores (Brawley and Adey, 1981). Hypotheses on the control of motile epifaunal
density include both top-down and bottom-up possibilities. Top-down hypotheses high-
light the importance of fish predation and the availability of refugia from predation in
structuring motile epifaunal communities (Vadas, 1985; Carpenter, 1986). There are
generally greater densities of motile epifauna in refugia and examples of refugia include
the high local algal biomass of damselfish territories and areas where fish foraging is
limited, such as wave swept areas or small crevices (Bailey-Brock et al., 1980; Lobel, 1980;
Carpenter, 1986; Klumpp et al., 1988; Klumpp and Polunin, 1990). The bottom-up
argument stems from the observation that little energy, typically 10– 20% of algal biomass,
is transferred up through a typical herbivore food web, suggesting there is little scope for
herbivore (e.g. motile epifaunal) control of algal populations (Vadas, 1985). This hypoth-
esis is supported by the positive relationship between algal biomass and the density of
epifaunal groups, e.g. amphipods, copepods, molluscs and polychaetes, which suggests
algal biomass determines epifaunal density (Bailey-Brock et al., 1980; Klumpp et al.,
1988). Experimental increases in algal cover have led to 10-fold increases in motile
epifauna, and polychaete abundance was approximately eight times higher in algal-
dominated areas compared to nearshore areas lacking algae (Bailey-Brock et al., 1980).
Further evidence for bottom-up algal control is evinced by declines in motile epifaunal
abundance as algal density decreases in the austral summer (Klumpp et al., 1987). The
prevailing view is that motile epifaunal density is determined by algal biomass at relatively
small scales ( <3 km), however, it is not known whether this is true at larger spatial scales.
Scale is a pertinent issue in ecology because the knowledge gained about key structuring
processes at one scale cannot necessarily be used to understand or predict patterns at another
scale (Levin, 1992; Willis and Whittaker, 2002). The top-down role of fish predation in
structuring coral reef fish communities appears to be scale-dependent. At small spatial scales
(10–100 m) fish predators strongly influence the structure of prey fish communities (Caley,
1993; Hixon, 1993; Caley and St. John, 1996; Beukers and Jones, 1998), but at large spatial
scales (0.1–5 km) no influence of aggregate predator depletion upon the structure of prey
fish assemblages has been detected (Jennings and Polunin, 1997). In contrast to predation,
fish–algal interactions do not appear to vary with scale in hard substrate systems. The
pattern of fish –algal associations at large spatial scales is consistent with the small-scale
response of fishes to experimental manipulation of algal communities (Williams and
Polunin, 2001; Levin and Hay, 2002; Williams et al., 2002).
Few shelf areas remain where fish are exploited without substantial disturbance to
benthic habitats by mobile benthic fishing gears (Watling and Norse, 1998). One such area
is Fiji, where reef fish are routinely exploited by nondestructive fishing gears, such as
spears and hook and line methods (Jennings and Polunin, 1995b). Fiji offers an almost
unique opportunity to study fishing effects at large spatial scales because mapped marine
tenure systems exist where residents have sole long-term access to defined fishing
grounds (Jennings et al., 2001b). The variation in human population size and coral reef
area among fishing grounds has been successfully used to provide a spatial gradient of
fishing intensity and successful testing of fishing effects hypotheses (Jennings and
Polunin, 1995a,b,c, 1997).
Given the potentially important grazing role of motile epifauna and the role of fish
predation in structuring motile epifaunal communities, it is plausible that (a) they form an
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 3
intermediary role in a trophic cascade, and (b) fishing could disrupt this linkage
between fish communities and benthic community structure by changing the motile
epifaunal community structure. Here we examine whether community structure of motile
epifauna associated with algal communities varies along a fishing intensity gradient
consisting of six fishing grounds (islands) in the Lau Island group, Fiji. We also examine
variation in fish and benthic communities along the fishing intensity gradient to
determine correlates of motile epifaunal community structure. Epifaunal community
structure can either be sampled directly from benthic substrata or by sampling the
community which develops on settlement plates (Vadas, 1985; Carpenter, 1986; Klumpp
et al., 1988). The large spatial scale of this study (0.5–180 km) and small-scale habitat
heterogeneity restricted our study to sampling epifaunal communities that had been left
to develop on coral settlement plates for six months. First, we expect increased density
of motile epifauna at high fishing intensity due to the reduction in predatory fishes.
Second, we expect motile epifaunal density to be linked to either the algal biomass on
settlement plates or reef substrate complexity (rugosity), or a combination of both. In this
case, we assume that rugosity is an indirect measure of epifaunal shelter or refugia
availability. There is the possibility that the sign of the correlation varies with spatial scale,
with one variable explaining density at smaller scale and the other variable explaining
most of the variation at a larger scale. Third, we expect motile epifaunal diversity to
decline along the fishing intensity gradient, as has been found with other benthic
disturbances, such as trawling and pollution (e.g. Warwick and Clarke, 1994, 1995;
Schratzberger and Warwick, 1998).
2. Methods
2.1. Study location and fishing intensity in Fiji
The Lau Island group in the eastern division of Fiji is relatively isolated and only
subject to subsistence levels of agriculture and fisheries (Fig. 1). Each island in this
study constituted a single discrete fishing ground (qoliqoli), where the exclusive
fishing rights of each island’s inhabitants extend from the shoreline to approximately
200 m beyond the outer reef. Fishing grounds were chosen based on the similarity of
outer reef architecture. Fish and benthic surveys were conducted on shallow (7 m chart
datum) leeward (western) outer reefs during three cruises (Apr –May 1999, Sept–Nov
1999, Feb–Mar 2000) at six fishing grounds (Table 1). The barrier reef front of each
fishing ground was divided into areas, each 400 m in length, on the appropriate
marine chart. Fishing grounds of various sizes were sampled in a proportional manner
by randomly selecting one third of all available areas (between 3 and 8). Three areas
were surveyed at the smallest fishing grounds while eight areas were sampled at the
largest fishing ground (Table 1). Sample areas within each fishing ground were
relocated using geographical positioning system, however, we did not attempt to
exactly relocate each replicate site among sampling dates; consequently, variation at
site level among sampling dates is a combination of both seasonal and sample location
variance.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–294
A fishing intensity index for each fishing ground was obtained by dividing the human
population (Anonymous, 1998) by the length of barrier reef front measured from aerial
photographs (scale = 1:50,000; Australian Aerial Mapping 1994, available from the
Department of Land and Surveys, Suva, Fiji). Two fishing grounds, Vuaqava and
Tavunasici, were uninhabited and the effective human population sizes were calculated
from the average number of visiting fishers from the tenure-holding village and the
estimated number of fishing visits per year. The fishing intensity index is strongly
Table 1
Survey details including names of fishing grounds, number of areas surveyed, number of areas from which coral
plates were recovered, number of plates recovered, human population size, coral reef front length and fishing
intensity index
Island
code
Fishing
ground
Number
of areas
surveyed
Number of areas
from which plates
were recovered
Total number
of plates recovered
(coral, ceramic)
Human
population
Reef
front
(km)
Fishing intensity
(population km
reef front
1
)
A Tavunasici 3 3 15, 22 20 7.6 2.6
B Vuaqava 3 3 13, 19 100 15.1 6.6
C Totoya 5 3 13, 18 806 44.7 18.0
D Matuku 6 3 12, 21 854 35.0 24.4
E Moala 8 1 8, 20 1596 60.9 26.2
F Kabara 5 2 5, 11 1012 23.4 43.3
Fig. 1. The fishing grounds (islands) selected for study labelled in alphabetical order of fishing intensity (see
Table 1). Inset shows the location of the Lau islands within Fiji.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 5
positively correlated with the actual fishing activity and fish yield (Jennings and Polunin,
1995c).
2.2. Fish census
Fish density was estimated using SCUBA underwater visual census (UVC) of 7-m
radius point counts. A total of 182 diurnally active reef associated species from 18 families
were censused (Appendix 1). Fishes of >8-cm fork length were censused using six
replicate point counts ( f154 m
2
) haphazardly distributed within each area (Fig. 2) (as
described by Jennings and Polunin, 1995b, 1997; Samoilys and Carlos, 2000). Point
counts have similar power to transect methods at the level of replication used here
(Samoilys and Carlos, 2000). The boundary of the point count was first estimated and
noted relative to reef landmarks, and the radius was confirmed using a tape measure on
completion of each count. The surveyor (NKD) was trained in underwater estimation of a
7-m point count radius distance and radius estimates were accurate to within F5 cm.
Individual fish were counted and fork length estimated to the nearest 1 cm. The surveyor
was trained in fish size estimation to a resolution of 1 cm using objects of fixed sizes
presented at 3-and 7-m distance underwater (Bell et al., 1985; Darwall and Dulvy, 1996).
Fish size estimation was found to be accurate to within F0.7 cm at both distances.
Underwater visibility was >20 m throughout the study and all surveys were conducted in
daylight at least 1 h after sunrise and 1 h before sunset. Mobile species were censused first
followed by territorial and cryptic species. Individual fish entering the point count during
the survey were not recorded. Count time was not standardised because this was dependent
on fish abundance, diversity and habitat complexity. The mean count time was 9.5 min
(range 4–16).
Estimates of fish length were converted to biomass using species-specific length –
weight conversions (Wright and Richards, 1985; Letourneur et al., 1998). If a length –
weight relationship was not available, the relationship for a species of similar morphology
in the same genus was used. Densities from point counts were expressed as g m
2
, which
is equivalent to kg km
2
.
Fig. 2. Hierarchical sampling design and the spatial scale of sampling in the Lau Islands, Fiji.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–296
2.3. Benthic community structure
Digital video footage was recorded at 70 positions haphazardly distributed in a zigzag
manner either side of the 7-m-depth contour, centred on and largely within each UVC
replicate. Each position was approximately 65 F1.6 (mean FSE) cm apart. The camera was
kept at a fixed distance (26 cm) above the substrate using a spacing rod which provided a
22.5 22.5 cm ( f500 cm
2
) quadrat. Video tapes were replayed on a television screen
overlaid with one of five randomly chosen transparent acetate sheets containing 20 randomly
located 1-cm-diameter circles (Williams et al., 2002). The number of circles occupied by
each benthic category was recorded and raised to a percentage. Aggregate categories
recorded in this analysis included: ascidians, ‘bare’ substrate (i.e. imperceptible algal
biomass), blue-green algae, crustose coralline algae, coralline lethal orange disease (CLOD),
hard coral, macroalgae >5 cm in height, Palythoa spp., soft coral, sponges and filamentous
turf algae < 5 cm in height. Analysis began with the first video frame of each UVC site, and
the tape was advanced and randomly paused at the next or next-but-one frame until 30
frames had been analysed for each UVC site. The total area photographically sampled in
each UVC site comprised 15 m
2
, which was f10% of the area of each UVC site.
Rugosity was measured by fitting a 3-m length of small-link chain to the reef surface
perpendicular to the reef crest at the centre of the census area. The corresponding horizontal
distance was measured by tape and the ratio of chain length/horizontal length calculated
(McClanahan and Shafir, 1990). Low rugosity ratios correspond to low surface relief.
2.4. Settlement plates
Five coral plates (12 12 cm) and 10 unglazed ceramic plates (15 15 cm) were
haphazardly placed on the reef surface approximately 30 – 50 cm apart along the 7-m-
depth contour (McClanahan, 1997) in three areas, picked at random, in each of six fishing
grounds during the first cruise and retrieved after 6 months. The coral plates were cut from
cross-section slices of freshly harvested heads of the massive coral Porites lutea using a
commercial rock-cutting saw (McClanahan, 1997). The square coral plates were approx-
imately 1-cm thick with a mean surface area of 144.3 F2.9 cm
2
(mean FSE). On
retrieval, the plates were bagged in situ using ziplock polythene bags and returned to
the surface. On board, formalin was added to achieve a buffered concentration of
approximately 5%, and each bag was sealed and returned to the laboratory in padded
lightproof storage boxes. Motile epifauna were collected by washing each coral plate over
5- and 0.5-mm sieves using filtered seawater; the intermediate motile epifauna fraction
was protein-stained using rose bengal and preserved in 70% ethanol. Care was taken to
minimise the inclusion of motile epifauna from plate underside and edges in the study
samples. Motile epifauna was then identified to varying levels of taxonomic resolution
(suborder-class) and counted. No attempt was made to recover motile epifauna from
ceramic plates.
The percent cover of filamentous turf algae < 5 cm, crustose coralline algae and
macroalgae on each plate was determined by superimposing an acetate sheet of 100
randomly placed circular points onto each settlement plate and counting the number of
points occupied by each algal category. Algal biomass was estimated by sub-sampling
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 7
three randomly placed 1-cm
2
quadrats from each plate (N.V.C. Polunin, unpublished data).
Each quadrat was excavated to a depth of approximately 3 mm, to include any endolithic
algae, and the sample was decalcified with dilute HCL. The sample was filtered onto
predried and preweighed filter papers and oven-dried at 60 jC for 3 days until a constant
weight was achieved (using a Shimadzu Libror AEX-200G balance). The average number
of bite marks cm
1
was used as an indication of grazing pressure. The number of bite
marks and excavations on each plate was counted along three transects randomly placed
across the face of each coral plate. Transects were placed perpendicular to the plate edge.
Parrotfish excavations on the plate edge were excluded as these often overlapped and were
difficult to count accurately.
2.5. Analysis
The unbalanced sampling design required analysis using general linear models. Fish
and motile epifaunal densities were log
10
(x+ 1) transformed; benthic and plate algal
percent cover data were arcsine transformed to achieve normality and homogeneity of
variance. For the analysis of fishing effects, data were aggregated at the lowest hierarchical
level of replication (sites), and then nested at successively larger spatial scales, i.e. UVC
sites < areas < fishing grounds (Fig. 2). For analysis of spatial scale effects, data were
Table 2
The effects of sampling date, fishing intensity and their interaction on mean density (g m
2
) of (a) all censused
fish and (b) fish trophic categories, fish families and percentage cover of benthic categories
(a) All fish (dependant variable) Two-way GLM ANOVA
Fvalue Pvalue
Fishing intensity 10.4 < 0.001 *
Sampling date 2.0 0.14
Interaction 1.3 0.23
(b) Community (dependant variable) Two-way GLM MANOVA Two-way crossed ANOSIM
Fvalue Pvalue Global RPvalue
Fish trophic category
Fishing intensity 4.4 < 0.0001 * 0.25 0.001 *
Sampling date 5.3 < 0.0001 * 0.11 0.04 *
Interaction 1.2 0.08 – –
Fish family
Fishing intensity 4.0 < 0.0001 * 0.15 0.001 *
Sampling date 4.3 < 0.0001 * 0.07 0.043 *
Interaction 1.0 0.35
Benthic category
Fishing intensity 11.4 < 0.0001 * 0.64 0.001 *
Sampling date 4.4 < 0.0001 * 0.2 0.001 *
Interaction 2.4 < 0.0001 * – –
Significance at < 0.05 is represented by an asterisk.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–298
Fig. 3. Boxplots of the variation (%) explained by sampling date, fishing intensity and their interaction, for (A)
fish trophic categories, (B) fish families and (C) benthic categories. The central line represents the median; the
box represents the 50% interquartile range of observations and the whiskers represent the limit of observations
(1.5 interquartile range).
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 9
analysed at two levels; among areas (0.5–10 km) and among fishing grounds (4– 180
km). Two-way parametric (MANOVA, Wilk’s k, as implemented by SAS, SAS
Institute) and nonparametric randomisation (ANOSIM, as implemented by PRIMER
version 5.2.2.) were used to explore the effects of fishing intensity and sampling date
on fish trophic, fish family and benthic community structure. We used multidimensional
scaling (MDS) and ANOSIM was applied to similarity matrices calculated using a
Bray–Curtis similarity measure (Clarke and Warwick, 1994). This combined MAN-
OVA/ANOSIM approach was chosen because although the data were parametric the
number of areas sampled at each fishing ground was not uniform, and MANOVA is
sensitive to unbalanced sampling designs (Johnson and Field, 1993).Plateswere
recovered from three areas at each fishing ground, with the exception of Moala where
coral plates were recovered from only one area. Therefore, subsequent analyses of
motile epifaunal density and richness were restricted to the remaining five fishing
grounds. Cross-correlations of algal plate, motile epifauna plate, benthic and fish
community structures were determined using the RELATE procedure of PRIMER
(Clarke and Warwick, 1994).
3. Results
3.1. Fish density
Total fish density varied significantly with fishing intensity, but not with sampling
date (Table 2a). The mean biomass of fish trophic categories and families varied
significantly with both fishing intensity and sampling date, but the interaction was not
significant (Fig. 3A,B;Table 2b). There were significantly lower densities of all
trophic categories at more heavily fished grounds, apart from territorial omnivores,
which exhibited a dome-shaped response, and omnivores which did not exhibit a
response (Fig. 4;Table 3a). The effect of fishing intensity upon fish families was
generally much weaker compared to trophic categories, however, there were signifi-
cantly lower densities of the parrotfishes at more heavily fished grounds (Fig. 5;Table
3b). There was also significant variation in butterflyfish and goatfish densities among
grounds, but these taxa did not decline along the fishing gradient as would have been
expected.
3.2. Benthic community structure
The benthic community structure varied significantly with fishing intensity and among
sampling dates, but the proportion of variance explained by fishing intensity was on
average 10 times greater than that explained by sampling date (Fig. 3C;Table 2b). There
was also a significant interaction between sampling date and fishing intensity, but this
explained less than half of the variance of fishing intensity alone.
At more heavily fished grounds there was a significantly lower cover of hard corals and
significantly greater cover of turf algae, coralline algae and blue-green algae (Fig. 6;Table
3c). There were also significant differences in the cover of macroalgae and sponges, but
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2910
Fig. 4. Relationships between the total fish biomass, the biomass of fish trophic categories and fishing intensity in
six fishing grounds (meanFstandard error). Trophic categories are arrayed in descending order of biomass.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 11
Table 3
The effects of sampling date, fishing intensity and their interaction on the mean density (g m
2
) of (a) each fish
trophic category, (b) each fish family and (c) percent cover of each benthic category
Community (dependant variable) Two-way GLM ANOVA
Sampling date Fishing intensity Interaction
Fvalue Fvalue Fvalue
(a) Fish trophic categories
Corallivores 0.6 2.4* * 1.4
Herbivores 0.8 5.6* * * 0.6
Invertivores 6.7* * 3.5* * 1.1
Omnivores 1.3 3.3* * 1.7
Piscivores 2.4 9.8* * * 2.0 *
Planktivores 4.1 * 3.6* * 1.2
Territorial omnivores 34.9* * * 15.4* * * 1.1
(b) Fish families
Surgeonfishes (Acanthuridae) 0.1 0.8 1.4
Triggerfishes (Balistidae) 0.01 0.6 1.0
Butterflyfishes (Chaetodontidae) 5.2* * 2.8 * 2.4 *
Porcupinefishes (Diodontidae) 0.6 0.5 0.9
Grunts (Haemulidae) 0.3 2.1 1.3
Chubs (Kyphosidae) 0.6 1.0 0.9
Wrasses (Labridae) 0.2 1.0 1.5
Emperors (Lethrinidae) 0.1 0.4 0.7
Snappers (Lutjanidae) 4.5 * 1.9 3.6* * *
Filefishes (Monacanthidae) 0.6 1.3 1.2
Goatfishes (Mullidae) 0.2 2.7 * 4.7* * *
Threadfin breams (Nemipteridae) 0.5 0.4 0.4
Boxfishes (Ostraciidae) 0.4 0.7 1.7
Parrotfishes (Scaridae) 0.2 2.8 * 2.6 *
Groupers (Serranidae) 1.0 1.4 1.6
Rabbitfishes (Siganidae) 0.4 2.0 1.5
Pufferfishes (Tetraodontidae) 0.2 0.5 1.0
(c) Benthic categories
Ascidians 4.7 * 3.3 * 3.1* *
Bare substrate 0.7 1.2 1.4
Blue-green algae 0.01 3.9* * 2.1 *
Crustose coralline algae 2.0 5.0* * * 1.1
CLOD disease 0.7 1.7 1.3
Hard corals 3.4 * 31.2* * * 8.7* * *
Macroalgae 4.1* 17.9* * * 3.0* *
Palythoa spp. 1.2 3.7* * 3.0* * *
Soft corals 3.7 * 21.0* * * 3.6* * *
Sponges 4.4 * 13.6* * * 3.9* * *
Turf algae 0.4 21.9* * * 7.8* * *
Note that territorial omnivores are solely composed of species from the fish family, Pomacentridae. Significance
is represented by asterisks as follows: * P< 0.05; * * P< 0.01; * * * P< 0.001.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2912
Fig. 5. Relationships between the biomass of fish families and fishing intensity in six fishing grounds
(mean Fstandard error). Families are arrayed in descending order of biomass density.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 13
these were not consistent with a linear fishing effect. There was no significant difference in
either measure of rugosity or reef slope among fishing grounds.
3.3. Plate algal community structure
Turf algae (54%) and coralline algae (31%) dominated both the coral and ceramic
plates. There was little ‘bare’ area (3%), and some fleshy macroalgae were present (11%,
mainly Padina and Lobophora with some Halimeda) on the plates.
There was no significant difference in algal community structure between the different
plate materials (coral vs. ceramic), but there was a significant difference among fishing
Fig. 6. Relationships between the percent reef cover of selected benthic categories and fishing intensity in six
fishing grounds (mean Fstandard error).
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2914
Fig. 7. Relationships between the percent cover of turf algae, coralline algae and macroalgae on settlement plates
and fishing intensity in five fishing grounds (mean Fstandard error). (.) Coral plates; (o) ceramic plates. Note:
endolithic algae was not observed on ceramic plates.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 15
grounds (two-way ANOSIM; plate material Global r= 0.0, P= 0.5, fishing intensity
Global r= 0.18, P= 0.026). From here on, only the results from coral plates are considered.
There was no consistent effect of fishing intensity on plate algal communities; endolithic
Table 4
The effect of fishing intensity on algal community metrics on settlement plates
Algal communities One-way GLM ANOVA
Fvalue Pvalue
Algal biomass (mg cm
2
) 0.3 0.8
Bite marks (no. cm
1
) 0.4 0.8
‘Bare’ (%) 0.6 0.7
Endolithic algae (%) 11.1 0.002* *
Turf algae (%) 1.9 0.2
Crustose coralline algae (%) 0.6 0.7
Macroalgae (%) 19.5 < 0.001* * *
Significance is represented by asterisks as follows: * P< 0.05; * * P< 0.01; * * * P< 0.001.
Fig. 8. Relationships between the total algal biomass and fish grazing intensity (A) across all 14 areas, and (B) at
five fishing grounds. Pvalues were derived from linear regression tests and NS means the test was not significant;
see text for details.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2916
algal cover exhibited a dome-shaped response to fishing intensity; and turf algae and
coralline algae did not differ among fishing grounds and macroalgal cover was greater
only at the most heavily fished ground (Fig. 7;Table 4). There was no significant
difference in algal biomass or the mean number of bite marks cm
1
on plates among
fishing grounds. Algal biomass was positively related to the mean number of bite
marks cm
1
, but only at the largest scale (Fig. 8;F
1,3
= 24.2, P= 0.016). There was
no relationship between algal biomass and rugosity at either area or fishing-ground
scale.
Table 5
The mean density of each taxonomic class of motile epifauna (individuals per 100 cm
2
) at each fishing ground
and the combined percent abundance of each class
Class Fishing ground code (fishing pressure index, persons km reef front
1
)
A (2.6) B (6.6) C (18) D (24.4) F (43.3) Percent
abundance
Malacostraca 64.2 175.3 107.5 91.5 73.9 54.1
Polychaeta 27.5 61.6 35.6 25.9 71.6 23.5
Copepoda 5.6 40.0 9.8 3.4 13.3 7.6
Gastropoda 10.1 16.9 5.3 9.3 11.1 5.6
Nematoda 2.4 8.0 7.0 9.2 6.8 3.5
Echinoidea 22.5 3.6 0.2 0.0 0.0 2.8
Stelleroidea 1.9 2.8 0.6 1.0 2.4 0.9
Bivalva 1.5 2.4 1.9 0.6 1.2 0.8
Turbellaria 4.2 2.1 0.8 0.3 0.0 0.8
Ostracoda 0.6 1.0 0.7 1.3 0.0 0.4
Holothuroidea 0.0 0.3 0.2 0.0 0.0 0.1
Table 6
The effect of fishing intensity on the density (individuals 100 cm
2
) of motile epifaunal classes on settlement
plates
Epifauna, taxonomic class One-way GLM ANOVA
Fvalue Pvalue
Malacostraca 1.7 0.2
Polychaeta 1.1 0.4
Copepoda 4.8 0.024 *
Gastropoda 3.8 0.046 *
Nematoda 0.7 0.6
Echinoidea 5.2 0.019 *
Stelleroidea 1.5 0.3
Bivalva 1.4 0.3
Turbellaria 4.3 0.03 *
Ostracoda 1.8 0.2
Holothuroidea 0.7 0.6
Total combined 2.1 0.2
Significance is represented by asterisks as follows: * P< 0.05; * * P< 0.01; * * * P< 0.001.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 17
3.4. Plate motile epifauna community structure
The density of motile epifauna averaged 189 individuals 100 cm
2
and the most
abundant classes were Malacostraca (54% of individuals), Polychaeta (23.5%), Copepoda
(7.6%) and Gastropoda (5.6%) (Table 5). Several taxonomic classes exhibited significant
differences in density among fishing grounds, but only Echinoidea and Turbellaria
declined systematically with fishing intensity (Tables 5 and 6). Motile epifaunal density
was negatively related to algal biomass at area scale, but unrelated at fishing-ground scale
(Fig. 9A,B;areaR
2
= 0.33, F
1,12
= 7.25, P= 0.02; fishing ground R
2
=0, F
1,12
=0,
P= 0.97). Motile epifaunal density was weakly positively related to reef rugosity at area
scale and strongly related to rugosity at fishing-ground scale (Fig. 9C,D; area scale
R
2
= 0.20, F
1,12
=4.2, P= 0.064; fishing-ground scale quadratic regression R
2
=0.99,
Fig. 9. Relationships between motile epifaunal density and the total algal biomass (A) across all 14 areas, and (B)
at five islands (mean Fstandard error). Relationships between motile epifaunal density and rugosity (C) across all
14 areas, and (D) at five fishing grounds (mean Fstandard error). Hollow data points are used to highlight nearly
overlapping points. Pvalues are from linear regression tests, except for (D) where quadratic regression was used
and NS means the test was not significant; see text for details.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2918
F
2,3
= 724, P< 0.001). When both explanatory variables are considered together, algal
biomass explained more variation (77%) in motile epifaunal density than rugosity (33%) at
the area scale (multiple regression, R
2
= 0.40, F
2,11
= 5.27, P= 0.025). However, at fishing-
ground scale, rugosity explained 99% of the variation in motile epifaunal density. There
Fig. 10. Relationships between (A) density, (B) taxonomic class richness and (C) class diversity of motile
epifauna and fishing intensity (persons km reef front
1
) at five fishing grounds (mean Fstandard error).
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 19
were no relationships between the density of motile epifauna and the fish grazing intensity
on the coral plates at either area or fishing-ground scale.
Motile epifaunal density did not vary systematically along the fishing intensity
gradient (Fig. 10A). However, the class richness and diversity of motile epifaunal
communities were significantly lower at the more heavily fished grounds (Fig. 10B;
richness, F
1,4
= 5.75, P= 0.011; Fig. 10C; Margalef’s d,F
1,4
= 8.09, P= 0.004). The rare
epifaunal classes (Echinoidea, Turbellaria, Ostracoda and Holothuroidea) were absent at
Fig. 11. (A) k-dominance curves for coral plate motile epifauna classes at five fishing grounds. (B) MDS
ordination of the motile epifaunal communities at 15 plate sites at five fishing grounds. Areas are shaded from
light to dark and labelled in alphabetical order corresponding to increasing fishing intensity (see Table 2). These
analyses were based on the proportion of Annelida, Asteroidea, Bivalva, Caprillidea, Copepoda, Crinoidea,
Cumacea, Decopoda, Echinoidea, Gammaridea, Gastropoda, Holothuroidea, Isopoda, Natantia, Nematoda,
Nudibranchia, Ophiuroidea, Ostracoda, Platyhelminths, Polyplacophora, Pycnognida, Tanaidacea at each site.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2920
the most heavily fished ground and only Polychaeta exhibited greater density at the
most heavily fished ground (Table 5). This pattern was reflected in the k-dominance
curves, with the curve representing the lightest fishing intensity lying below and to the
right of the others, and the curves shifting upward and to the left with increasing
fishing intensity, suggesting fewer classes contributed to dominance at higher fishing
intensities (Fig. 11A). The MDS ordination indicated significant grouping of areas
within fishing grounds, indicating a link between motile epifaunal community structure
and fishing intensity (Fig. 11B; ANOSIM Global r= 0.527, P= 0.001). There were no
relationships between class richness or diversity of motile epifauna and fish grazing
intensity, algal biomass on coral plates or benthic rugosity at either area or fishing-
ground scales.
3.5. Correlations among fish, benthic, plate algal and motile epifaunal communities
Motile-epifaunal community structure was strongly correlated with benthic community
structure using cross-correlation of similarity matrices (rho = 0.512, P= 0.001) and weakly
correlated with plate algal community structure (rho = 0.24, P= 0.068). Benthic commun-
ity structure was significantly correlated with fish community structure (fish families,
rho = 0.388, P= 0.002; fish trophic categories, rho = 0.245, P= 0.047). However, fish
community structure, using either families or trophic categories, was not significantly
correlated with either the algal communities or motile epifaunal communities on settlement
plates.
4. Discussion
This study indicates fishing can influence the community structure of small motile
epifauna, in the absence of physical disturbance. There were clear differences in the k-
dominance of these epifaunal communities along the fishing intensity gradient, which
was consistent with lower epifaunal class richness and diversity at grounds with higher
fishing intensities. There was a clear multivariate link between fishing intensity and
community structure of small epifaunal (Fig. 11B), but there was no clear effect of
fishing on total epifaunal density (Fig. 10). However, epifaunal density was determined
by a combination of settlement-plate algal biomass and coral reef rugosity. The relative
importance of algal biomass and rugosity for structuring epifaunal density was scale-
dependent; at smaller spatial scales (among areas, 0.5 – 10 km) algal biomass was more
important, and at larger spatial scales (among fishing grounds, 4 – 180 km) rugosity
was more important. To the best of our knowledge, this study constitutes the first
evidence to suggest that the processes controlling community structure may vary with
spatial scale.
The conclusions of this study are potentially limited by a number of factors; here
we address each in turn. The settlement plates were deployed for a relatively short
period of time, which raises the possibility that algal and epifaunal communities might
not have stabilised, and therefore our findings may not be representative over longer
time scales. However, this is unlikely to be important to our findings because
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 21
communities of small epifauna develop rapidly, and the time to reach short-term
(weeks) stability in algal cover and diversity is approximately 100 days on coral
settlement plates (Martin-Smith, 1994; McClanahan, 1997). Another limitation was that
while there was a multivariate link between fish and other communities, it was not
possible to determine the proximate links between benthic and plate community
structures and particular fish trophic or family groupings. Most of the significant
differences in fish community structure were heavily dependent on the high densities
found in the least fished ground. In order to understand the proximate links between
fish and benthic communities, we suggest greater replication at the fishing-ground scale
is required. The recruitment of epifauna is often cited as having a major structuring role
in benthic communities (e.g. Sutherland, 1981). The variance of fish and benthic
communities associated with sampling date subsumes variation associated with both
recruitment and sampling error. The variance attributable to sampling date of fish and
benthic communities was minor compared to variance attributable to fishing intensity;
we suggest that the recruitment of fish or benthos did not contribute significantly to their
observed community structures over the duration of this study. This raises the possibility
that recruitment may also be less important relative to fishing intensity for structuring
coral reef epifaunal communities. However, we caution that direct measurement of the
effects of fish benthic and epifaunal recruitment would be required to test this
possibility. A key limitation of our study is that an experimental manipulation would
be required to confirm the causal links among rugosity, predation risk and epifaunal
density. However, there are two fundamental constraints to confirming causality at the
large spatial scale studied here. First, manipulating small-scale factors requires caging of
substrata; this would be expensive and logistically difficult to construct and maintain in
the required hierarchical design at such large spatial scales (Vadas, 1985; Raffaelli and
Moller, 2000). Second, large-scale influences, such as fishing and recruitment, cannot be
manipulated experimentally; in this situation the only available test for causality is a
comparative one, as was performed here (Petraitis and Latham, 1999). Finally, we only
examined the role of fishing as a causal factor in determining fish and benthic
community structures; we have not examined the importance of nutrient input or
island-scale oceanographic factors, which may also have the capacity to influence coral
reef communities.
4.1. Factors influencing motile epifaunal density
The density of motile epifauna is known to be a function of either substrate
complexity (rugosity), algal biomass (habitat quality) and recruitment success (Bailey-
Brock et al., 1980; Lobel, 1980; Vadas, 1985; Carpenter, 1986; Klumpp et al., 1988;
Klumpp and Polunin, 1990). The only study which examined the relative importance of
these factors concluded that algal biomass was the major determinant of epifaunal
density at small scales ( < 4 km) (Klumpp et al., 1988). Our data were consistent with
this finding; at the smaller spatial scale algal biomass was the major determinant of
motile epifaunal density. However, at the larger spatial scale rugosity was the most
important determinant, and explained 99% of the variation in the density of small motile
epifauna. Why should rugosity be an important determinant of epifaunal density? We
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2922
suggest that rugosity is a measure of epifaunal shelter availability or refugia from
predation. Rugosity is strongly related to more direct measures of shelter availability
such as the size of holes in the substrate (Friedlander and Parrish, 1998). Predation
generally appears to have an important structuring role on coral reefs; shelter availability
is a major determinant of coral reef fish community structure at large spatial scales (e.g.
Caley and St. John, 1996; Friedlander and Parrish, 1998). Adult fish abundance is
strongly determined by predator density and refuge availability, and the importance of
shelter availability is underscored by the finding that juvenile survivorship of fishes is
enhanced by providing more complex substrata (Beukers and Jones, 1998). Therefore,
rugosity can be thought of as an indirect measure of predation risk or top-down control
upon motile epifauna. If this assumption is valid, then top-down factors were a more
important predictor of the density of small motile epifauna than bottom-up factors such
as algal biomass at large spatial scales. To the best of our knowledge, scale dependence
in the processes structuring the density of community components has not been
described. However, it is becoming clear that diversity of communities is structured
in a scale-dependent manner (Willis and Whittaker, 2002), so the finding of scale-
dependent structuring of density is not unexpected.
If predation risk were a major structuring force of epifaunal communities, then one
would predict that the reduction of fish biomass along the fishing gradient would lead to
higher densities of motile epifauna via prey release. Predatory release of motile epifaunal
prey is a prerequisite for the existence of a potential trophic cascade involving epifauna.
There was no evidence for prey release and epifaunal densities did not vary with fishing
intensity in this study (Fig. 10A); therefore, at these light fishing intensities, it is also
unlikely that epifauna is involved in a trophic cascade. The absence of prey release
suggests the reduction in predator biomass may be insufficient to invoke prey release or a
consequence of the absence of any link between fishing intensity and rugosity at these
light levels of fishing pressure. In more heavily exploited systems, fishing is known to
reduce substrate complexity by facilitating the proliferation of bioeroding herbivorous
urchins (McClanahan and Muthiga, 1988; McClanahan, 1992). The absence of a link
between rugosity and fishing intensity is possibly due to the relatively low density of
grazing urchins in this study, which is a tenth or less (< 1 urchins m
2
) than found in
heavily fished coral reef systems, such as found in the Caribbean and also in East Africa
(c. 10 urchins m
2
)(Carpenter, 1986; Lessios, 1988; McClanahan and Shafir, 1990;
McClanahan and Kurtis, 1991; McClanahan, 1994; McClanahan and Mutere, 1994). The
urchin densities in Fiji are possibly insufficient to measurably reduce substrate complexity
at the comparatively low fishing intensities studied, but the potential for prey release of
epifauna and a trophic cascade role should be borne in mind where fishing pressure is
sufficiently high to influence substrate complexity.
4.2. Factors influencing motile epifaunal richness
Ultimately, understanding the forces structuring motile epifaunal diversity will be
challenging because almost nothing is known of their life histories, dispersal strategies and
niche ecology (Carpenter, 1986; Klumpp et al., 1988). The class richness and diversity of
motile epifauna were independent of the structure of plate algal communities. This
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 23
evidence combined with the known rapid colonisation ( < 2 weeks) suggests the epifaunal
communities on settlement plates reflected the larger benthic epifaunal pool (e.g. Caley
and Schluter, 1997). Assuming this is true, then we can examine three possible
explanations for monotonic decline in epifaunal class richness and diversity along the
fishing intensity gradient, namely differential vulnerability, exploiter-mediated coexistence
and decline in habitat quality. Larger-bodied animal species tend to have lower intrinsic
population growth rates and lower resilience to mortality and would be expected to suffer
most under a disturbance regime (Collie et al., 2000; Frisk et al., 2001; Reynolds et al.,
2001). In these Fijian fishing grounds, increased fishing intensity should be associated
with reduced predation pressure upon motile epifauna, unless there is an extra trophic
group between target fishes and the motile epifauna. Therefore, large-bodied epifauna
should exhibit elevated density at the most heavily fished grounds. However, there is no
clear sign of increased densities of larger epifaunal classes (e.g. malacostracan crustaceans
and gastropods) at the most heavily fished grounds (Table 5), which suggests the
‘differential vulnerability’ mechanism does not underlie the observed reduction in
epifaunal richness and diversity. The second mechanism is exploiter-mediated coexistence,
in which diversity is structured or maintained by frequency-dependent mortality; the most
abundant taxa are predated in proportion to their abundance, facilitating the coexistence of
competitively inferior taxa (Lubchenco, 1978; Menge, 1995). At heavily fished grounds,
the lower density of invertivorous fishes (and potentially lower predation) could have
increased the dominance of competitively superior motile epifaunal taxa at the expense of
less competitive taxa—resulting in a decline in motile epifaunal richness and diversity—
which was consistent with the pattern described here. Another possible explanation for the
change in epifaunal class richness and diversity is the change in the ‘habitat quality’ of the
surrounding benthic community. Coral reef benthic communities are highly heterogeneous
on a very small scale of < 1 m (N.K. Dulvy, unpublished data) and a reduction in patch
heterogeneity of surrounding habitat would be expected to result in lower diversity
(Austen et al., 1998). Hard-coral cover declined and turf-algal cover increased along the
fishing intensity gradient and the variation in benthic community structure was statistically
correlated with the change in fish community structure. However, at present we cannot
explain the causal mechanism(s) underlying this statistical link. The increase in motile
epifaunal habitat (turf algae) on the coral reef and the possible homogenisation of benthos
would be expected to result in a proliferation of primary colonising motile epifaunal
taxa—which is also consistent with the pattern described here.
The spatial variance of abundance and diversity and the underlying structuring
processes are not only of fundamental importance; spatial scaling is also a practical
issue. If we are to understand and manage environmental impacts at large spatial
scales, the analyses of underlying processes must be applied at the appropriate scale.
Inferences drawn from small-scale studies may be erroneous because larger-scale
structuring processes are not necessarily predictable from an understanding of small-
scale processes (Willis and Whittaker, 2002). Finally, management must be feasible at
the appropriate spatial scale. Fortunately, management can directly influence predator
density at large scales through restricting fishing effort or the implementation of no-
take zones, but not algal abundance, except by managing watersheds to reduce nutrient
inputs.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2924
Acknowledgements
We wish to thank Robin South and the staff of the Marine Studies Programme at the
University of the South Pacific for providing facilities and continued support; the Tui
Nayau, chiefs and turaganikoro of all the islands for granting permission for this study; K.
Wynn, C. Glendinning, A. Haddock, R. Ekeheien, C. Gough and B. Vasconcellos for
logistical support and field assistance. We thank D. Stanwell-Smith for help with the
identification of the epifauna. We thank I.D. Williams, T.J.T. Murdoch and also S.
Jennings and an anonymous referee for their useful comments. This study was funded by
the Natural Environment Research Council, UK and the Fisheries Society of the British
Isles. [RW]
Appendix 1
Fish speciesin UVC, maximum lengths (cm) and tropic category (co—corallivore, he—
herbivore, in—invertivore, om—omnivore, pi—piscivore, pl—planktivore, to—terrtorial
omnivore, uf—urchin feeder).
Acanthuridae Acanthurus albipectoralis 33 pl, A. blochii 42, A. guttatus 26 he, A.
lineatus 38 he, A. nigricans 21 he, A. nigricauda 40 he, A. olivaceus 35 he, A. pyroferus 25
he, A. thompsoni 27 pl, A. triostegus 26 he, A. xanthopterus 56 he, Ctenochaetus binotatus
22 he, C. striatus 26 he, C. strigosus 18 he, Naso brevirostoris 60 pl, N. caesius 60 pl, N.
hexacanthus 75 pl, N. lituratus 30 pl, N. tuberosus 60 pl, N. unicornis 70 pl, N. vlamingii
50 pl, Zebrasomsa scopas 20 om, Z. veliferum 40 om.
Balistidae balistapus undulatus 30 uf, Balistoides conspicillum 50 in, B. viridescens 75
uf, Melichthys vidua 35 om, Rinecanthus rectangulus 30 uf, Sufflamen bursa 24 uf, S.
chrysopterus 30 uf, S.fraenatus 38 uf.
Chaetodontidae Chaetodon auriga 23 in, C. baronessa 15 co, C. bennetti 18 co, C.
citrinellus 13 in, C. ephippium 23 in, C. flavirostris 20 C. kleinii 14 in, C. ornatissimus 20
co, C. pelewensis 12 in, C. plebeius 15 in, C. quadrimaculatus 16 co, C. rafflesi 15 in, C.
reticulatus 16 co, C. trifascialis 18 co, C. trifasciatus 15 co, C. ulietenis 15 in, C.
unimaculatus 20 in, C. vagabundus 23 in, Forcipiger flavissimus 22 in, F.longirostris 22
in, Hemitaurichthys polylepis 18 pl, Heniochus acuminatus 25 pl, H. chrysostomus 18 co,
H. monocerus 23 in, H. singularius 25 in, H. varius 19 in.
Diodontidae Diodon hystrix 90 uf.
Haemulidae Plectorhinchus chaetodonoides 72 in, P obscurus 100 uf, P. picus 84 uf.
Kyphoside Kyphosus cinerascens 45 he.
Labridae Anampses caerulopunctatus 42 uf, A. neoguinaicus 15 in, A. twistii 18 in,
Bodianus anthiodes 21 in, B. axillaris 20 in, B. diana 25 in, B. loxozonus 47 in, B.
mesothorax 19 in, Cheilinus chlorourus 45 in, C. fasciatus 38 in, C. oxycephalus 17 in, C.
trilobatus 45 uf, C. undulatus 229 in, Coris aygula 120 uf, C. gaimard 38 uf, Epibulus
insdiator 35 pi, Gomphosus varius 28 in, Halichoeres hortulanus 27 in, H. margaritaceus
13 in, H. marginatus 17 in, Hemigymnus fasciatus 50 uf, H. melapterus 50 uf, Labrichthys
unileatus 16 co, Macrophargodon meleagris 14 in, Oxycheilinus diagrammus 30 pi, O.
unifasciatus 46 uf, Pseudocheilinus hexataenia 7 in, P. octotaenia 12 uf, Stethojulius
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 25
bandanensis 16 uf, Thalassoma amblcephalum 14 in, T. hardwicke 20 in, T. jansenii 20 pi,
T. lutescens 30 uf, T. quinquevittatum 16 uf.
Lenthrinidae Gnathodentex aureolineatus 24, Lethrinus atkinsoni 45 uf, L. erythra-
canthus 70 uf, L.nebulosus 86 uf, L. olivaceaus 100 uf, Montaxis grandoculis 60 uf.
Lutjanidae Aphareus furca 40 pi, Aprion virenscens 100 in, Lutjanus bohar 90 pi, L.
fulviflamma 35 pi, L. fulvus 40 pi, L.kasmira 35 pi, L. monostigma 53 pi, L. russelli 50 pi,
L. semicintus 35 pi, L. vitta 40 pi, Macolor macularis 55 in, M. niger 66 in.
Monacanthidae Aluterus scriptus 110 in, Amanses scopas 16 in, Cantherhines
dumerilii 38 in, C. pardalis 25 in, Oxmonacanthus longirostris 12 co, Pervagor
melanocephalus 10 in.
Mullidae,Mulloidichthys vanicolensis 38 uf, P. bifasciatus 35 in, P. ciliatus 38 in, P.
cyclostomus 50 in, P. miltifasciatus 30 in.
Nemiptridae Scolopsis bilineatus 23 in.
Ostraciidea Ostracion meleagris 16 in.
Pomacentridae Plectroglyphidodon dickii 11 to, P. johnstonianus 9 to, P. lacrymatus
10 to, Pomacentrus bankanesis 9 to, P. vaiuli 9 to, Stegastes fasciolatus 15 to, S. lividus 13
to, S. nigricans 13 to.
Scanridea Cetoscarus bicolor 80 he, Chlorurus frontalis 50 he, Hipposcarus longiceps
60 he, Scarus alptipinis 60 he, S. chameleon 31 he, S. dimidiatus 30 he, S. forsteni 55 he,
S. frenatus 47 he, S. ghobban 75 he, S. globiceps 27 he, S. longipinnis 40 he, S. niger 35
he, S. oviceps 30 he, S.psittacus 30 he, S. rubroviolaceus 70 he, S. schlegeli 38 he, S.
sordidus 40 he, S. spinus 30 he.
Serranidae Anyperodon leucogrammicus 52 pi, Cephalopholis argus 40 pi, C.
leopardus 24 pi, C. urodeta 27 pi, Epinephelus fuscoguttatus 90 pi, E. hexagonatus 26
pi, E howlandi 44 pi, E. maculatus 50 pi, E. polyphekaion 75 pi, Gracila albomarginata
40 pi, Plectropomus areolatus 73 pi, P. laevis 110pi, P. leopardus 70 pi, P. maculatus 70
pi, Variola louti 80 pi.
Siganidae Siganus doliatus 24 he, S. punctatus 40 he, S. stellatus 35 he, S. uspi 24 he.
Tetraodontidae Aronthron mappa 65 in, A. nigropunctatus 33 uf, Canthigaster
valentini 10 he.
References
Anonymous, 1998. 1996 Fiji census of population and housing: general tables. Parliament of Fiji, Suva, Fiji.
Austen, M.C., Widdecombe, S., Villano-Pitacco, N., 1998. Effects of biological disturbance on diversity and
structure of meiobenthic nematode communities. Mar. Ecol., Prog. Ser. 174, 233– 246.
Bailey-Brock, J.H., White, K., Ward, L.A., 1980. Effects of algal turf and depressions as refuges on polychaete
assemblages of a windward reef bench on Enewetak Atoll. Micronesica 16, 43– 54.
Bell, J.D., Craik, G.J.S., Pollard, D.A., Russell, B.C., 1985. Estimating length frequency distributions of large reef
fish underwater. Coral Reefs 4, 41 – 44.
Beukers, J.S., Jones, G.P., 1998. Habitat complexity modifies the impact of piscivores on a coral reef fish
population. Oecologia 114, 50 – 59.
Brawley, S.H., Adey, W.H., 1981. The effect of micrograzers on algal community structure in a coral reef
microcosm. Mar. Biol. 61, 167 – 177.
Caley, M.J., 1993. Predation, recruitment and the dynamics of communities of coral reef fishes. Mar. Biol. 117,
33 – 43.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2926
Caley, M.J., Schluter, D., 1997. The relationship between local and regional diversity. Ecology 78, 70– 80.
Caley, M.J., St. John, J., 1996. Refuge availability structures assemblages of tropical coral reef fishes. J. Anim.
Ecol. 65, 414 – 428.
Carpenter, R.C., 1986. Partitioning herbivory and its effects on coral reef algal communities. Ecol. Monogr. 56,
345 – 363.
Clarke, K.R., Warwick, R.M., 1994. Change in Marine Communities: An Approach to Statistical Analysis and
Interpretation. Natural Environmental Research Council, UK.
Collie, J.S., Escanero, G.A., Valentine, P.C., 1997. Effects of bottom fishing on the benthic megafauna of Georges
Bank. Mar. Ecol., Prog. Ser. 155, 159– 172.
Collie, J.S., Hall, S.J., Kaiser, M.J., Poiner, I.R., 2000. A quantitative analysis of fishing impacts on shelf-sea
benthos. J. Anim. Ecol. 69, 785 – 798.
Darwall, W.R.T., Dulvy, N.K., 1996. An evaluation of the suitability of non-specialist volunteer researchers for
coral reef fish surveys. Mafia Island, Tanzania—a case study. Biol. Conserv. 78, 223– 231.
Frid, C.L.J., Hansson, S., Ragnarsson, S.A., Rijnsdorp, A., Steingrimsson, S.A., 1999. Changing levels of
predation on benthos as a result of exploitation of fish populations. Ambio 28, 578 – 582.
Friedlander, A.M., Parrish, J.D., 1998. Habitat characteristics affecting fish assemblages on a Hawaiian coral reef.
J. Exp. Mar. Biol. Ecol. 224, 1 – 30.
Frisk, M.G., Miller, T.J., Fogarty, M.J., 2001. Estimation of biological parameters in elasmobranch fishes: a
comparative life history study. Can. J. Fish. Aquat. Sci. 58, 969– 981.
Greenstreet, S.P.R., Bryant, A.D., Broekhuizen, N., Hall, S.J., Heath, M.R., 1997. Seasonal variation in the
consumption of food by fish in the North Sea and implications for foodweb dynamics. ICES J. Mar. Sci.
54, 243 – 266.
Hixon, M.A., 1993. Predation, prey refuges, and the structure of coral-reef fish assemblages. Ecol. Monogr. 63,
77 – 101.
Jennings, S., Polunin, N.V.C., 1995a. Comparative size and composition of yield from six Fijian reef fisheries. J.
Fish Biol. 46, 28 – 46.
Jennings, S., Polunin, N.V.C., 1995b. Effects of fishing on the biomass and structure of target reef fish commun-
ities. J. Anim. Ecol. 33, 400 – 412.
Jennings, S., Polunin, N.V.C., 1995c. Relationships between catch and effort in Fijian multispecies reef fisheries
subject to different levels of exploitation. Fish. Manage. Ecol. 2, 89 – 101.
Jennings, S., Polunin, N.V.C., 1996. Effects of fishing effort and catch rate upon the structure and biomass of
Fijian reef fish communities. J. Appl. Ecol. 33, 400–412.
Jennings, S., Polunin, N.V.C., 1997. Impacts of predator depletion by fishing on the biomass and diversity of non-
target reef fish communities. Coral Reefs 16, 71 – 82.
Jennings, S., Grandcourt, E.M., Polunin, N.V.C., 1995. The effects of fishing on the diversity, biomass and
trophic structure of Seychelles’ reef fish communities. Coral Reefs 14, 225 – 235.
Jennings, S., Dinmore, T.A., Duplisea, D.E., Warr, K.J., Lancaster, J.E., 2001a. Trawling disturbance can modify
benthic production processes. J. Anim. Ecol. 70, 459 – 475.
Jennings, S., Kaiser, M.J., Reynolds, J.D., 2001b. Marine Fisheries Ecology. Blackwell, Oxford.
Johnson, C.R., Field, C.A., 1993. Using fixed-effects multivariate analysis of variance in marine biology and
ecology. Oceanogr. Mar. Biol. Ann. Rev. 31, 177–221.
Kaiser, M.J., 1998. Significance of bottom-fishing disturbance. Conserv. Biol. 12, 1230 –1235.
Kaiser, M.J., Spencer, B.E., 1995. Survival of by-catch from a beam trawl. Mar. Ecol., Prog. Ser. 112, 41 –49.
Kaiser, M.J., Ramsay, K., Richardson, C.A., Spence, F.E., Brand, A.R., 2000. Chronic fishing disturbance has
changed shelf sea benthic community structure. J. Anim. Ecol. 69, 494 – 503.
Kitchell, J.F., Carpenter, S.R., 1993. Cascading trophic interactions. In: Carpenter, S.R., Kitchell, J.F. (Eds.), The
Trophic Cascade in Lakes. Cambridge Univ. Press, Cambridge, pp. 1 – 14.
Klumpp, D.W., Polunin, N.V.C., 1989. Partitioning among grazers of food resources within damselfish territories.
J. Exp. Mar. Biol. Ecol. 125, 145 – 169.
Klumpp, D.W., Polunin, N.V.C., 1990. Algal production, grazers and habitat partitioning on a coral reef: positive
correlation between grazing rate and food availability. In: Barnes, M., Gibson, R.N. (Eds.), Trophic Relation-
ships in the Marine Environment. Aberdeen Univ. Press, Aberdeen, pp. 372 – 388.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 27
Klumpp, D.W., McKinnon, A.D., Daniel, P., 1987. Damselfish territories: zones of high productivity on coral
reefs. Mar. Ecol., Prog. Ser. 40, 41– 51.
Klumpp, D.W., McKinnon, A.D., Mundy, C.N., 1988. Motile cryptofauna of a coral reef: abundance, distribution
and trophic potential. Mar. Ecol., Prog. Ser. 45, 95 – 108.
Lessios, H.A., 1988. Mass mortality of Diadema antillarum in the Caribbean: what have we learned? Ann. Rev.
Ecolog. Syst. 19, 371 – 393.
Letourneur, Y., Kulbicki, M., Labrosse, P., 1998. Length– weight relationships of fish from coral reefs and
lagoons of New Caledonia, Southwestern Pacific Ocean. An update. Naga 21, 39 – 46.
Levin, S.A., 1992. The problem of pattern and scale in ecology. Ecology 73, 1943 – 1967.
Levin, P.S., Hay, M.E., 2002. Fish– seaweed association on temperate reefs: do small-scale experiments predict
large-scale patterns. Mar. Ecol., Prog. Ser. 232, 239 – 246.
Lobel, P.S., 1980. Herbivory by damselfishes and their role in coral reef community ecology. Bull. Mar. Sci. 30,
273 – 289.
Lubchenco, J., 1978. Plant species diversity in a marine intertidal community: importance of herbivore food
preference in algal competitive studies. Am. Nat. 112, 23 – 39.
Martin-Smith, K.M., 1994. Short-term dynamics of tropical macroalgal epifauna: patterns and processes in
recolonisation of Sargassum fissifolium. Mar. Ecol., Prog. Ser. 110, 177–185.
McClanahan, T.R., 1992. Resource utilisation, competition, and predation: a model and example from coral reef
grazers. Ecol. Model. 61, 195 – 215.
McClanahan, T.R., 1994. Kenyan coral reef lagoon fish-effects of fishing, substrate complexity, and sea-urchins.
Coral Reefs 13, 231 – 241.
McClanahan, T.R., 1995. A coral reef ecosystem-fisheries model: impacts of fishing intensity and catch selection
on reef structure and processes. Ecol. Model. 80, 1 – 19.
McClanahan, T.R., 1997. Primary succession of coral-reef algae: differing patterns on fished versus unfished
reefs. J. Exp. Mar. Biol. Ecol. 218, 77 – 102.
McClanahan, T.R., Kurtis, J.D., 1991. Population regulation of the rock-boring sea-urchin Echinometra mathaei
(Deblainville). J. Exp. Mar. Biol. Ecol. 147, 121 – 146.
McClanahan, T.R., Mutere, J.C., 1994. Coral and sea urchin assemblage structure and interrelationships in
Kenyan reef lagoons. Hydrobiologia 286, 109 – 124.
McClanahan, T.R., Muthiga, N.A., 1988. Changes in Kenyan coral reef community structure and function due to
exploitation. Hydrobiologia 166, 269 – 276.
McClanahan, T.R., Sala, E., 1997. A Mediterranean rocky-bottom ecosystem fisheries model. Ecol. Model. 104,
145 – 164.
McClanahan, T.R., Shafir, S.H., 1990. Causes and consequences of sea urchin abundance and diversity in Kenyan
coral reef lagoons. Oecologia 83, 362 – 370.
Menge, B.A., 1995. Indirect effects in marine rocky intertidal interaction webs: patterns and importance. Ecol.
Monogr. 65, 21 – 74.
Petraitis, P.S., Latham, R.E., 1999. The importance of scale in testing the origins of alternate community states.
Ecology 80, 429 – 442.
Pinnegar, J.K., Polunin, N.V.C., Francour, P., Badalamenti, F., Chemello, R., Harmelin-Vivien, M.-L., Hereu, B.,
Milazzo, M., Zabala, M., D’Anna, G., Pipitone, C., 2000. Trophic cascades in benthic marine ecosystems:
lessons for fisheries and protected-area management. Environ. Conserv. 27, 179 – 200.
Polunin, N.V.C., Klumpp, D.W., 1992. A trophodynamic model of fish production on a windward reef tract. In:
John, D.M., Hawkins, S.J., Price, J.H. (Eds.), Plant –Animal Interactions in the Marine Benthos. Clarendon
Press, Oxford, pp. 213 – 233.
Raffaelli, D., Moller, H., 2000. Manipulative field experiments in animal ecology: do they promise more than
they can deliver? Adv. Ecol. Res. 30, 299 – 338.
Reynolds, J.D., Jennings, S., Dulvy, N.K., 2001. Life histories of fishes and population responses to exploitation.
In: Reynolds, J.D., Mace, G.M., Redford, K.H., Robinson, J.G. (Eds.), Conservation of Exploited Species.
Cambridge Univ. Press, Cambridge, pp. 147 – 168.
Russ, G.R., Alcala, A.C., 1989. Effects of intense fishing pressure on an assemblage of coral reef fishes. Mar.
Ecol., Prog. Ser. 56, 13 – 27.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–2928
Sala, E., Boudouresque, C.F., Harmelin-Vivien, M., 1998. Fishing, trophic cascades, and the structure of algal
assemblages: evaluation of an old but untested paradigm. Oikos 82, 425–439.
Samoilys, M.A., Carlos, G., 2000. Determining methods of underwater visual census for estimating the abun-
dance of coral reef fishes. Environ. Biol. Fisches 57, 289 – 304.
Schratzberger, M., Warwick, R.M., 1998. Effects of the intensity and frequency of organic enrichment on two
estuarine nematode communities. Mar. Ecol., Prog. Ser. 164, 83 – 94.
Steneck, R.S., 1998. Human influences on coastal ecosystems: does overfishing create trophic cascades. Trends
Ecol. Evol. 13, 429 – 430.
Sutherland, J.P., 1981. The fouling community at Beaufort, North Carolina: a study in stability. Am. Nat. 118,
499 – 519.
Tegner, M.J., Dayton, P.K., 2000. Ecosystem effects of fishing in kelp forest communities. ICES J. Mar. Sci. 57,
579 – 589.
Vadas, R.L., 1985. Herbivory. In: Littler, M.M., Littler, D.S. (Eds.), Handbook of Phycological Methods, Eco-
logical Field Methods: Macroalgae. Cambridge Univ. Press, Cambridge, pp. 531 – 573.
Warwick, R.M., Clarke, K.R., 1994. Relearning the ABC-taxonomic changes and abundance biomass relation-
ships in disturbed benthic communities. Mar. Biol. 118, 739–744.
Warwick, R.M., Clarke, K.R., 1995. New biodiversity measures reveal a decrease in taxonomic distinctness with
increasing stress. Mar. Ecol., Prog. Ser. 129, 301 – 305.
Watling, L., Norse, E.A., 1998. Disturbance of the seabed by mobile fishing gear: a comparison to forest
clearcutting. Conserv. Biol. 12, 1180 –1197.
Whitman, J.D., Sebens, K.P., 1992. Regional variation in fish predation intensity: a historical perspective in the
Gulf of Maine. Oecologia 90, 305 – 315.
Williams, I.D., Polunin, N.V.C., 2001. Large-scale associations between macroalgal cover and grazer biomass on
mid-depth reefs in the Caribbean. Coral Reefs 19, 358 – 366.
Williams, I.D., Polunin, N.V.C., Hendrick, V.J., 2002. Limits to grazing by herbivorous fishes and the effect of
low coral cover on macroalgal abundance on a coral reef in Belize. Mar. Ecol., Prog. Ser. 222, 187–196.
Willis, K.J., Whittaker, R.J., 2002. Species diversity-scale matters. Science 295, 1245 – 1248.
Wright, A., Richards, A.H., 1985. A multispecies fishery associated with coral reefs in the Tigak Islands, Papua
New Guinea. Asian Mar. Biol. 2, 69 – 84.
N.K. Dulvy et al. / J. Exp. Mar. Biol. Ecol. 278 (2002) 1–29 29
