E nvironm ental B iology of Fishes 49: 239–246, 1997.
1997 K luwer A cadem ic Publishers. Printed in the N etherlands.
The natural history of a monogamous coral-reef fish, Valenciennea strigata
(Gobiidae): 1. abundance, growth, survival and predation
R obert H . R eavis
M useum of Vertebrate Z oology and D epartm ent of Integrative B iology, University of California, B erk eley,
CA 94720, U.S.A .
Current address: D epartm ent of L ife Sciences, A riz ona State University West, 4701 W. T hunderbird R d.,
Phoenix, A Z 85069–7100, U.S.A .
R eceived 27.7.1995
A ccepted 27.5.1996
Key words: population ecology, equilibrium/nonequilibrium processes, recruitment, monogamy, mating
system, mate guarding
Synopsis
The population dynamics of a monogamous coral-reef fish were examined to test hypotheses of recruitment
limitation, predation, and postrecruitment processes, and to determine their affects on the mating system.
Valenciennea strigata are monogamous gobies that live in sand and rubble zones throughout the Indo-Pacific.
Seasonal abundance was recorded in the summer and winter over 2.5 years. A subset of this population was
tagged (n = 256) and followed to determine mortality and mobility. Valenciennea strigata were more abundant
in summer than in winter, suggesting that a pulse of recruitment in the spring set the maximum population
density. G rowth rates derived from tagged fish support the hypothesis that recruitment peaked in the spring.
Tagged fish experienced 88% mortality within six months; the annual mortality rate approached 100% . E vidence of predation, antipredatory behavior and strong site fidelity implicate predation as the primary source
of mortality. Competition for space was not observed between adults, but may affect settlement and recruitment. D espite the lack of adult competition for space, both sexes guarded their mates and courted individuals
of the opposite sex. Thus, although population size appears to be determined by nonequilibrium processes,
the mating system is affected by competition for mates. Successful mate guarding by both sexes enforced
monogamy.
Introduction
E lton’s (1927) model of community ecology suggests that most communities are structured by equilibrium processes. A lthough this model has dominated ecological theory, recent evidence supports a
nonequilibrium view of the natural world (reviewed by R eice 1994). Studies of coral-reef fishes
have been instrumental in this change of perspective (e.g. the lottery hypothesis, Sale 1978; the
recruitment-limitation hypothesis, D oherty 1983).
These and other studies suggested that populations
of coral-reef fishes are determined primarily by
oceanic processes that affect recruitment, while
equilibrium processes have little effect (D oherty
1991, Victor 1991). Similarly, predation can maintain
populations at levels that minimize competition
(H ixon 1991).
A lthough the emerging paradigm suggests that
coral-reef fish populations are limited by nonequi-
240
Figure 1. Study sites on Moorea Island.
librium factors, equilibrium processes may still affect these fishes (Jones 1991). In particular, competition for mates can occur even at low densities, especially if resources are clumped, affecting both reproductive output and the mating system (E mlen &
O ring 1977). Thus, studying the mating systems of
coral-reef fishes should provide evidence useful in
distinguishing postrecruitment processes. For example, monogamy in coral-reef fishes has been explained by both equilibrium and nonequilibrium
models (reviewed by Barlow 1984, Barlow 1986).
Butterflyfishes appear to be monogamous because
the competition for coral prevents males from sequestering multiple mates (H ourigan 1987). Conversely, a tilefish in the R ed Sea exists at such low
densities that mates remain together because encounters with potential mates are rare (Clark &
Pohle 1992).
Valenciennea strigata (Broussonet) is one of the
fish whose behavior was reviewed by Barlow. Little
additional information had been published on V.
strigata or their relatives until the recent review of
the genus by H oese & Larson (1994). They described 15 species, ecologically similar in their use of
sand and rubble habitats. Monogamy appears common to the genus; conspecifics are typically observed in pairs sharing a burrow that they construct
themselves (e.g. Barlow 1984, H oese & Larson
1994).
H ere I examine the population dynamics of V.
strigata to determine whether their population is influenced primarily by equilibrium or nonequilibrium factors. My companion paper (R eavis 1997)
addresses the behavior and mating system of the
same population. Together, these data suggest that
the population is limited by nonequilibrium processes, and that the low population density allows
both sexes to successfully guard a monogamous
mate.
241
Methods
I followed members of a population of V. strigata
over 2.5 years to determine the changes in population and the potential factors that influence these
changes. Individually marked fish were observed
daily to determine their survival and movement.
R ecaptures of marked fish provided growth data.
Finally, I observed the burrows of V. strigata, their
anti-predator function, and territorial interactions.
R esearch was conducted along the north shore of
Moorea, Society Islands, French Polynesia. A ll
work was done using snorkel gear over the course of
five seasons: 27 D ecember 1991–17 A pril 1992; 24
June–15 August 1992; 20 January–18 A pril 1993; 23
June–6 July 1993, and 11–20 January 1994, for a total
of 1046 research hours in the water.
Work was concentrated at three sites: the Bali
H ai (BH ), in the shallow backreef offshore of the
H otel Bali H ai; the Moorea Lagon (ML), an area of
shallow fringing reef that runs continuously into the
back reef, beginning 100 m offshore and just east of
the H otel Moorea Lagon, and the West Backreef
(WB), a deeper backreef area 150–200 m seaward
of the channel west of Cook’s Bay (Figure 1). These
sites are similar in size: BH = 13 500 m 2, ML =
13 400 m 2, and WB = 14 800 m 2. They varied in
depth between 1.5–4 m over a predominantly sand
and rubble bottom with scattered coral heads. Sites
were marked with floating bottles, and weighted
plastic flags were placed near at least one burrow
per territory. I measured the distance between these
flags to construct maps of the study sites.
The abundance of V. strigata at each site was recorded by swimming over the site in tandem with a
dive buddy in 15-wide strips (after H elfman 1983).
A ll V. strigata were counted and identified as adults
or juveniles, and single or paired. The accuracy of
these surveys was immediately verified by a search
for tagged fish. A ll abundance surveys agreed exactly with the tagged fish surveys. A bundance was
measured in the summer and winter of 1992 and
1993, and the summer of 1994. H owever, in the first
season (summer 1992) abundance was not measured until early fall. Because the population declined from summer to fall, I have substituted data
taken from daily surveys made earlier in the sum-
mer. These surveys focused on tagged fish and
probably underestimated the actual population
size.
A t the start of each of the first two seasons, and
opportunistically thereafter, I tagged most of the V.
strigata at each of the study sites. In the summer of
1993 tagging was concentrated only in certain areas
of the study sites because of the high density of fish
that season. Fish were captured by chasing them into their burrows, covering the entrances with nets,
and then injecting a solution of quinaldine into the
burrows. Most fish swam into the nets immediately
after the quinaldine was applied; others became anesthetized and were collected by digging into the
burrow. Pairs often went into a burrow together,
and both were usually captured on the same day.
Captured fish were placed into a plastic bag for
handling and brought to a boat where I sexed them
by genital papilla (214 of 256 tagged fish), measured
their standard length (SL) to the nearest mm, and
tagged them. Male genital papillae taper sharply
from the base and become pointed at the distal end;
female genital papillae are broad throughout (this
sexual pattern is common in gobies, Miller 1984).
Sex was later verified by behavior. I arbitrarily remeasured some fish at the original time of capture;
all measurements agreed within 0–3 mm.
A t each study site, individuals received a unique
combination of colored dyes injected under the
skin. These marks remained clear throughout the
study. Finally, each fish was scored as either mated
or single. Pairs seen or captured together were assumed to be mates.
To determine growth rates I recaptured and remeasured 22 tagged fish. The longest lived female
was remeasured in three seasons. The analysis of
growth rates presented here treats each growth interval for this female as a separate datum, providing
a total of 24 growth rates. A lternative analyses were
not significantly different.
I surveyed all sites daily to relocate each tagged
fish during the summer and winter of 1992 and summer 1993 (27–89 days per season). Natural history
and behavior were recorded opportunistically
throughout the course of the study (for behavioral
data see R eavis 1997). I also observed the burrows
created by V. strigata, and noted evidence of preda-
242
Figure 2. Mean summer abundance of V. strigata is greater than
mean winter abundance (one-way A NOVA , p < 0.05; n = 3 study
sites, error bars = 1 SE ). * Summer 1992 surveys were conducted
late in the season and underestimate mid-summer abundance.
Figure 3. Survival of tagged V. strigata. Starred bars represent the
number of fish tagged that season. Plain bars represent the number of tagged fish of that cohort resighted in subsequent surveys.
# No tagged fish were found in January 1994.
tion. Finally, I determined the diel pattern of activity from paired dusk-dawn observations at each site
in the summer and winter of 1992.
differ by sex (A NOVA , p > 0.10). H owever, pairs
were positively assorted for size (simple regression,
r 2 = 0.67, p < 0.01, n = 91), and males were 5.0 mm
longer than their mates (paired t-test, p < 0.01, n =
91).
D uring daily surveys, I usually found tagged fish
on their original territories and with the same mate
(for mate fidelity see R eavis 1997). Fish that
changed territories over the course of the study typically moved to a neighboring territory after the loss
of a mate. The longest recorded movements were 80
m (3 of 256 fish). H owever, the majority of tagged
fish disappeared in the same season that they were
tagged (Figure 3). O nly 31 (12.1% ) survived from
one season to the next (5–7 months); two of these
females survived at least one year, and one was seen
17 months after tagging. No tagged fish could be
found in the summer of 1994, 24 months after the
start of tagging. Survivorship did not vary by sex; 15
of 106 males and 13 of 108 females survived until a
second season (chi-square, p > 0.50).
A lthough some fish disappeared immediately after tagging, tags did not generally affect survival. O f
the 245 fish counted in the summer 1993 surveys, 57
(23.2% ) had been tagged. E ight tagged fish survived to the winter survey, out of a total of 42 fish in
that survey (19% ; chi-square, p > 0.50).
G rowth rates decelerated with increasing size
(Figure 4). From these data it could not be determined whether this change in rate was linear (sim-
Results and discussion
Population dynam ics
Fish abundance did not differ among the three
study sites over the course of the study (A NOVA ,
p > 0.10). A cross all sites, V. strigata were 4.7 times
more abundant in summer than winter (A NOVA ,
p < 0.05, Figure 2). This difference was primarily
due to the change in mean (± SD ) adult abundance
between seasons: summer = 68.6 (± 43.9) adults per
site, winter = 11.5 (± 7.1) adults per site (A NOVA , p
< 0.05). Juvenile abundance did not differ significantly between seasons: summer = 8.9 (± 15.0) juveniles per site, winter = 5.0 (± 7.6) juveniles per site
(A NOVA , p > 0.50). A dults occurred only in pairs
or as singletons, except during agonistic interactions. In the summer, 10.8% of fish were singletons;
19.2% were alone in the winter (chi-square, p <
0.05).
The mean (± SD ) SL of all tagged fish = 110.0 mm
(± 25.4, range: 30–152, n = 256). O f these, 214 were
of known sex. Mean male SL = 118.4 mm (± 15.4;
range: 70–152, n = 106); mean female SL = 114.8 mm
(± 17.1, range: 59–144, n = 108). O verall, size did not
243
Figure 4. G rowth rates of V. strigata decelerate with increasing
body size (polynomial regression, r 2 = 0.387, p < 0.05; n = 24).
ple regression, r 2 = 0.365, p < 0.01) or exponential
(polynomial regression, r 2 = 0.387, p < 0.01). The
largest growth rates occurred in small adults over
short intervals. These small but reproductively active individuals grew up to 26 mm in 45 days, an increase of 25% . Males grew at a mean rate of 0.16
(± 0.15) mm per day (n = 12); females grew at a
mean rate of 0.11 (± 0.09) mm per day (n = 12; A NO VA , p > 0.10).
A bundance and survival data suggest that V. strigata recruit primarily in the spring with few individuals surviving a full year. A lternatively, higher summer abundance could be due to adult immigration
into study sites in the spring and emigration from
study sites in the fall. D ata from the tagged fish suggest that emigration from study sites was unlikely.
Most fish remained on their original territories, and
none were found greater than 80 m from where they
were tagged, despite my searches over the entire
study sites (13 400–14 800 m 2). Thus, fish that disappeared probably died. Similarly, the site fidelity of
tagged fish suggests that increased summer densities were due to the settlement of juveniles, not immigration of untagged adults.
A lthough V. strigata were not observed in the
spring, recruitment of other fishes in Moorea peaks
in spring and summer (Planes et al. 1993). A dditionally, size at settlement and growth data support the
hypothesis that most of the summer population
recruited in the spring. The smallest V. strigata
tagged was 30 mm SL and the smallest untagged
fish were estimated to be 25 mm SL. Thus, settlement occurs at about 25 mm SL. This size at settlement is large for gobies, although larger presettlement juveniles have been recorded in this family
(Leis & R ennis 1983).
If recruitment in the spring accounts for most of
the summer population, young V. strigata grow
about 75 mm in approximately 100 days (0.75 mm
per day) to reach the 100 mm SL commonly seen in
January. This rate is consistent with the growth
curves derived here. The predicted growth rate of a
25 mm SL fish is either 0.62 mm per day (linear
model) or 1.31 mm per day (exponential model).
These growth rates vary by a factor of two and come
from measurements of adults, therefore they provide only a rough estimate of juvenile growth rates.
H owever, most fishes do experience increasing
growth rates (exponential model) prior to reaching
reproductive age (R icker 1975), and one reproductive male (102 mm SL) grew at a rate of 0.58 mm per
day. Similarly, Thresher (1983) found growth rates
approaching 0.70 mm per day in juvenile A canthochrom is polyacanthus, a monogamous damselfish.
Thus, it seems reasonable that juveniles recruited in
the spring could reach 100 mm SL by the end of January when most fish were tagged.
Mortality did not differ by sex in V. strigata. This
pattern is common in monogamous coral-reef fishes (e.g. A . polyacanthus, Thresher 1983; Paragobiodon echinocephalus, Kuwamura et al. 1994), and
other monogamous vertebrates (e.g. Promislow
1992). By contrast, polygamy is associated with differential survival between the sexes (Promislow
1992).
Predation and territoriality
Potential predators were common. Muraenids, serranids and octopuses were seen in and around the
burrows of V. strigata. Those burrows were avoided
by V. strigata. Several burrows appeared dug up by
octopuses. The resident V. strigata were usually missing on days when burrows were found opened, and
some never returned. I only witnessed one act of
predation, when I chased a male off his territory and
he was swallowed by a serranid (E pinephelus m erra) stationed beneath a coral head. O ther fish were
244
seen with wounds on their bodies suggesting predation attempts.
Burrows were dug under coral pavement and
rubble. They were used as both a refuge and a nest
site. Fish maintained several burrows per territory
(range: 4–10) and remained in close proximity to a
burrow. Burrows typically had two or more entrances, but only one remained open. The other entrance was covered by coral rubble, sand and algae.
When capturing fish in their burrows, they were as
likely to go out a closed entrance as an open entrance. Many burrows were maintained over multiple seasons, by either the original or subsequent
inhabitants. D espite this investment in burrows, interactions between territory residents and intruders
appeared to be over mates, not territories. Fish of
both sexes attacked intruders of the same sex, but
courted intruders of the opposite sex.
Juveniles also maintained burrows and defended
mates. H owever, in early summer 1993, when fish
density was the highest recorded, multiple fish were
found in close proximity at one site. I observed
many agonistic bouts at this time with associated
changes in mates and burrow use.
Territories were not defended interspecifically.
O ther fishes used V. strigata burrows, particularly
small gobies (e.g. G natholepis anjerensis) and juveniles of other families (e.g. chaetodontids, labroids,
and mullids). Valenciennea strigata rarely reacted to
those fishes, except to chase them briefly. Many
grazers and browsers fed in V. strigata territories, as
well as goatfishes that probed within the substrate.
Valenciennea strigata were diurnally active. Pairs
returned to their burrows at least one hour before
dark and typically closed the entrance behind them
with algae (see also H iatt & Strasburg 1960). Most
diurnal fishes were still active at this time (e.g.
cleaner wrasses, L abroides spp). Pairs emerged
from their burrows more than an hour after sunrise,
when other fishes were already active. They were
never seen out of their burrows at night.
A lthough measuring predation is difficult, predation may have strong effects on populations of coral-reef fishes (reviewed by H ixon 1991). Sweatman
(1984) estimated that a single species of lizardfish
caused 65% annual mortality in a prey population.
Thresher (1983) found 20–75% mortality in juve-
nile A . polyacanthus in their first 30 days, and adult
mortality was higher in the presence of a serranid.
Predation probably accounts for the significant
drop in abundance of V. strigata from summer to
winter, and the high rate of loss in tagged fish
throughout the year. A lternatively, fish may have
died due to age or parasites, or emigrated from
study sites. H owever, some fish lived over a year,
there were no indications of parasites externally or
internally (R eavis unpublished data) and tagged
fish showed strong site fidelity. These data do not
support the alternative hypotheses. A dditionally,
predators and evidence of predation attempts were
common (e.g. wounded fish, opened burrows). Besides the predators I observed, R . Caldwell (personal communication) witnessed a mantis shrimp,
Lysiosquilla sulcata, capture V. strigata near my
study sites. A t least two species of piscivorous mantis shrimp co-occur with V. strigata in Moorea and
regularly prey on gobioid fishes (R . Caldwell personal communication).
Valenciennea strigata also exhibited several
forms of anti-predator behavior. They maintained
multiple burrows and generally remained in close
proximity to a burrow (quantified by R eavis 1997).
The extra burrow entrances provided an escape
route from predators. A lthough these additional
entrances could also facilitate water movement
through burrows for aerating eggs, all burrows had
these extra entrances, but only one burrow was
used at any time for egg care. The diel-activity pattern of V. strigata also suggests predator avoidance;
they limited their twilight exposure to predators
more than other fishes (e.g. wrasses, H elfman 1993).
Finally, their strong site fidelity may be selected for
by predators. R emaining on a known territory ensures knowledge of the location of refugia (i.e. burrows, H inde 1956).
E quilibrium or nonequilibrium ? effects on the
m ating system
Juvenile settlement and recruitment were not addressed directly in this study; however, competition
between juveniles was observed in the early summer of 1993. Thus, competition for space may affect
recruitment in V. strigata. Predation apparently re-
245
duces adult competition over resources. The dramatic drop in seasonal abundance of V. strigata suggests that equilibrium processes on the population
are negligible in the fall and winter. E ven in the
summer, all adults maintained territories with multiple burrows. Burrows provide two resources that
limit other populations of fishes: nest sites (Breder
& R osen 1966) and refugia (Smith & Tyler 1975,
Shulman 1984). The abundance of burrows suggests
they are not limiting for V. strigata during most of
the year. A dditionally, many territories were bordered by unused habitat indistinguishable from territories (R eavis & Barlow unpublished data). Finally, territory intrusions occurred infrequently, and
these intrusions appeared to be over mates rather
than resources (see also R eavis 1997).
E mlen & O ring’s (1977) model of mating systems
suggested that the distribution of limiting resources
determines the distribution of females, and affects
the ability of males to encounter and monopolize
females. For V. strigata, however, resources appeared to be abundant, and these fish were broadly
distributed across my study sites. Further, the high
mortality and high fecundity (R eavis 1997) in these
fish suggest that they may be relatively r-selected.
Thus, E mlen & O ring’s (p. 215, 1977) model may not
be applicable.
Predation also affects the distribution of animals
and their mating systems (e.g. Jarman 1974). H ere,
predation probably selects for site fidelity, and limits the potential pool of mates to those within a radius of 80 m. A low probability of finding potential
mates can select for strong pair bonds and monogamy (e.g. Clark & Pohle 1992). A lthough many pairs
of V. strigata persisted over a study season, intruders
of the opposite sex were courted (see also R eavis
1997). The relatively weak pair bonds exhibited
suggest that a lack of potential mates does not explain monogamy in these fish.
Proximally, both sexes enforced monogamy by
mate guarding. Mate guarding should be selectively
advantageous when the costs are low relative to the
benefits. Mate-guarding costs for V. strigata appear
to be relatively low. The adult stock was determined
by nonequilibrium factors, suggesting that resources are abundant and not monopolizable by a
few males. In particular for gobioid fishes, an abun-
dance of burrows allows all males to defend a nest
site (Barlow 1986). If all males can breed, the cost of
mate guarding for females is reduced, and a guarding female becomes herself easily guarded because
of her proximity to the male. A ll male V. strigata in
this study maintained burrows (i.e. nest sites).
Moreover, R eavis (1997) found that both sexes prefer big mates and that big mates may provide
greater benefits.
Thus, monogamy in V. strigata appears to be affected by both the nonequilibrium processes that
limit the population, the resulting opportunities for
mate guarding by both sexes, and competition for
large mates. Because many populations of coralreef fishes appear to be recruit-limited, similar examples of monogamy may be found in these fishes.
A cknowledgements
I am indebted to Leilani E . Wright for her enthusiasm and hard work in the field, including hundreds
of hours watching V. strigata. E van R aymond also
participated in much of the water work. Thanks to
G eorge W. Barlow who first introduced me to this
problem and provided advice and encouragement
along the way. I am also grateful to R . Caldwell, V.
R esh, R . Swenson, C. St. Mary, R . Coleman and two
anonymous reviewers for their helpful comments
made on earlier drafts of this manuscript. This work
was supported by an NSF Fellowship, two Sigma Xi
grants, a Lerner-G ray grant, and the Museum of
Vertebrate Z oology and D epartment of Integrative
Biology, U niversity of California, Berkeley. The R ichard G ump South Pacific Biological R esearch Station provided additional support and assistance;
thanks to the director W. Loher, managers R . Steger
and F. Murphy, and to the many others I had the
pleasure to work with, particularly M. G leason.
This is contribution number 33 of the G ump Biological R esearch Station. Thanks to the government
and people of French Polynesia for their gracious
hospitality. A lec and A nne Langbridge, and John
and Tehea McInnis provided occasional living
space and diversion from field work, Mahu’uru’uru
R oa!
246
References cited
Barlow, G .W. 1984. Patterns of monogamy among teleost fishes.
A rch. FischWiss. 35: 75–123.
Barlow, G .W. 1986. A comparison of monogamy among freshwater and coral-reef fishes. pp. 767–775. In: Second Int. Conf.
on Indo-Pacific Fishes, Tokyo.
Breder, C.M. & D.E . R osen. 1966. Modes of reproduction in fishes. Natural H istory Press, G arden City. 941 pp.
Clark, E . & J.F. Pohle. 1992. Monogamy in the tilefish, M alacanthus latovittatus, compared with polygyny in related species.
R esearch and E xploration 8: 276–295.
D oherty, P.J. 1983. Tropical territorial damselfishes: is density
limited by aggression or recruitment? E cology 64: 249–261.
D oherty, P.J. 1991. Spatial and temporal patterns in recruitment.
pp. 261–293. In: P.F. Sale (ed.) The E cology of Fishes on Coral
R eefs, A cademic Press, San D iego.
E lton, C. 1927. A nimal ecology. Sidgwick and Jackson, London.
207 pp.
E mlen, S.T. & L.W. O ring. 1977. E cology, sexual selection, and
the evolution of mating systems. Science 197: 215–223.
H elfman, G .S. 1983. U nderwater methods. pp. 349–370. In: L.A .
Nielson & D.L. Johnson (ed.) Fisheries Techniques, A merican
Fisheries Society, Bethesda.
H elfman, G .S. 1993. Fish behavior by day, night and twilight. pp.
479–512. In: T.J. Pitcher (ed.) Behaviour of Teleost Fishes,
Chapman and H all, London.
H iatt, R .W. & D.W. Strasburg. 1960. E cological relationships of
the fish fauna on coral reefs of the Marshall Islands. E col.
Monogr. 30: 65–127.
H inde, R .A . 1956. The biological significance of the territories of
birds. Ibis 98: 340–369.
H ixon, M.A . 1991. Predation as a process structuring coral reef
fish communities. pp. 475–508. In: P.F. Sale (ed.) The E cology
of Fishes on Coral R eefs, A cademic Press, San D iego.
H oese, D.F. & H .K. Larson. 1994. R evision of the Indo-Pacific
gobiid genus Valenciennea: with descriptions of seven new
species. Indo-Pacific Fishes: 23, Bishop Museum Press, H onolulu. 71 pp.
H ourigan, T.F. 1987. The behavioral ecology of three species of
butterflyfishes (Family Chaetodontidae). D octoral D issertation, U niversity of H awaii, Manoa. 496 pp.
H udson, R .C.L. 1977. Preliminary observations on the behaviour of the gobiid fish Signigobius biocellatus H oese and A llen, with particular reference to its burrowing behaviour. Z .
Tierpsychol. 43: 214–220.
Jarman, P.J. 1974. The social organisation of antelope in relation
to their ecology. Behaviour 48: 215–267.
Jones, G .P. 1991. Postrecruitment processes in the ecology of coral reef populations: a multifactorial perspective. pp. 294–328.
In: P.F. Sale (ed.) The E cology of Fishes on Coral R eefs, A cademic Press, San D iego.
Kuwamura, T., Y. Yogo & Y. Nakashima. 1994. Population dynamics of goby Paragobiodon echinocephalus and host coral
Stylophora pistillata. Mar. E col. Prog. Ser. 103: 17–23.
Leis, J.M. & D.S. R ennis. 1983. The larvae of Indo-Pacific coral
reef fishes. New South Wales U niversity Press, Sydney. 269 pp.
Miller, P.J. 1984. The tokology of gobioid fishes. pp. 119–153. In:
G .W. Potts & R .J. Wootton (ed.) Fish R eproduction: Strategies and Tactics, A cademic Press, London.
Planes, S., A . Levefre, P. Legendre & R . G alzin. 1993. Spatiotemporal variability in fish recruitment to a coral reef (Moorea, French Polynesia). Coral R eefs 12: 105–113.
Promislow, D.E .L. 1992. Costs of sexual selection in natural populations of mammals. Proc. of R oyal Soc. (London), Biological Series 247: 203–210.
R eavis, R .H . 1997. The natural history of a monogamous coralreef fish, Valenciennea strigata (G obiidae): 2. behavior, mate
fidelity and reproductive success. E nv. Biol. Fish. (in press).
R eice, S.R . 1994. Nonequilibrium determinants of biological
community structure. A mer. Sci. 82: 424–435.
R icker, W.E . 1975. Computation and interpretation of biological
statistics of fish populations. Bull. Fish. R es. Board Can. 191.
382 pp.
St. John, J., G .P. Jones & P.F. Sale. 1988. D istribution and abundance of soft-sediment meiofauna and a predatory goby in a
coral reef lagoon. Coral R eefs 8: 51–57.
Sale, P.F. 1978. Coexistence of coral reef fishes: a lottery for living
space. E nv. Biol. Fish. 3: 85–102.
Shulman, M.J. 1984. R esource limitation and recruitment patterns in a coral reef fish assemblage. J. E xp. Mar. Biol. E col. 74:
85–109.
Smith, C.L. & J.C. Tyler. 1975. Succession and stability in fish
communities of dome-shaped patch reefs in the West Indies.
A mer. Mus. Novitiates 2572: 1–18.
Sweatman, H .P.A . 1984. A field study of the predatory behavior
and feeding rate of a piscivorous coral reef fish, the lizardfish
Synodus englem ani. Copeia 1984: 187–193.
Thresher, R .E . 1983. H abitat effects on reproductive success in
the coral reef fish, A canthochrom is polyacanthus (Pomacentridae). E cology 64: 1184–1199.
Trivers, R .L. 1972. Parental investment and sexual selection. pp.
136–179. In: B. Campbell (ed.) Sexual Selection and the D escent of Man, A ldine Press, Chicago.
Victor, B.C. 1991. Settlement strategies and biogeography of reef
fishes. pp. 231–260. In: P.F. Sale (ed.) The E cology of Fishes on
Coral R eefs, A cademic Press, San D iego.