RESEARCH ARTICLE
Mikaela A. J. Bergenius ? Mark I. McCormick
Mark G. Meekan ? D. Ross Robertson
Environmental influences on larval duration, growth
and magnitude of settlement of a coral reef fish
Received: 9 May 2002 / Accepted: 20 December 2004 / Published online: 16 March 2005

 Springer-Verlag 2005
Abstract The in?uence of environmental variables on the
planktonic growth, pelagic larval duration and settle-
ment magnitude was examined for the coral reef sur-
geon?sh Acanthurus chirurgus. Newly settled ?sh were
collected daily from patch reefs in the San Blas Archi-
pelago, Caribbean Panama for 3.5 years. Environmental
in?uences on growth were examined at three di?erent
life history stages: from 0 to 6 days, 7 to 25 days and
from 26 to 50 days after hatching. Larval growth was
correlated, using multiple regression techniques, with a
combination of factors including solar radiation, rain-
fall, and along-shore winds. Depending on the life his-
tory stage, these accounted for 13?38% of the variation
in growth rates when all the months were included in the
analyses. Correlations between environmental variables
and growth also varied among seasons and were stron-
ger in the dry than in the wet season. During the dry
season solar radiation, rainfall and along-shore winds
described 57%, 86% and 74% of the variability in
growth between 0 and 6 days, 7 and 25 days and 26 and
50 days, respectively. During the wet season rainfall,
along-shore winds and temperature only described 38%
of the variability in early growth and 27% of growth just
before settlement. No signi?cant model was found to
describe growth 7?25 days after hatching during the wet
season. Rainfall, solar radiation and along-shore winds
were negatively correlated with growth up to 25 days
after hatching but positively correlated as larvae ap-
proached settlement at a mean age of 52 days. Over 65%
of the variability in pelagic larval duration was
accounted for by a regression model that included solar
radiation and along-shore winds. When data sets from
wet and dry seasons were analysed separately, along-
shore winds accounted for 67% of the change in larval
duration in the dry season, and solar radiation
accounted for 23% of the variation in larval duration in
the wet season. Only 22% of the variability in settlement
intensity could be described by solar radiation and
temperature, when all months of the year were included
in the analysis. Solar radiation and rainfall were
included in a regression model that accounted for 40%
of the variation in numbers of ?sh settling during the dry
season. This study suggests that the levels of solar
radiation, along-shore winds and rainfall during the
early larval life can have important e?ects on the
growth, larval duration and consequently, the settlement
magnitude of marine ?shes. Results also highlight the
need to account for seasonality and ontogeny in studies
of environmental in?uences.
Introduction
Marine ?sh larvae su?er very high mortality (Hjort 1914;
Bailey and Houde 1989; Doherty 1991), which may be
strongly in?uenced by growth rates. Larvae that grow
quickly are the ?rst to attain the sizes necessary to become
juveniles or enter juvenile habitats (Houde 1987). Con-
sequently, fast growing ?sh are exposed to planktonic
predators for a shorter period of time and may be less
vulnerable to somepredators than slowgrowing?shof the
same cohort (Anderson 1988; Cushing 1990).
To date, most of the evidence supporting these con-
cepts (collectively termed the ??growth-mortality
Communicated by G.F. Humphrey, Sydney
M. A. J. Bergenius (&) ? M. I. McCormick
School of Marine Biology and Aquaculture,
James Cook University,
4811 Townsville, Qld, Australia
E-mail: mikaela.bergenius@jcu.edu.au
Tel.: +61-7-47814508
Fax: +61-7-47814099
M. G. Meekan
The Australian Institute of Marine Science,
PO Box 40197,
0811 Casuarina MC, NT, Australia
D. R. Robertson
Smithsonian Tropical Research Institute (Balboa Panama),
Unit 0948, Balboa, Panama APO AA 34002,
USA
Marine Biology (2005) 147: 291?300
DOI 10.1007/s00227-005-1575-z
hypothesis??) originates from studies in temperate re-
gions. However, recent work has also shown that in
tropical reef ?shes larval condition and/or growth are
important determinants of survivorship in the plank-
tonic (Bergenius et al. 2002; Wilson and Meekan 2002;
Meekan et al. 2003), as well as in the post-settlement
stage (Searcy and Sponaugle 2001; Vigliola and Meekan
2002; McCormick and Hoey 2004).
In larval ?shes, the process of growth re?ects the
interaction of an individual?s developmental physiology
with a great range of physical and biological factors.
These may act either directly or indirectly to in?uence
growth rates. For example, water temperature may
determine developmental time, growth, swimming per-
formance (Meekan et al. 2003; Green and Fisher 2004)
and the rate of yolk sac absorption (Fukahara 1990) of
young ?sh. Additionally, some species exhibit an opti-
mal temperature range for development, outside of
which mortality and abnormalities increase (Polo et al.
1991). Physical factors may also interact with other
biological processes to in?uence growth rates. Wind
speed and direction determine small-scale turbulence in
the water column and may therefore indirectly be an
important in?uence on rates at which larval ?sh
encounter and capture prey (Gallego et al. 1996; Dower
et al. 1997; Utne-Palm and Stiansen 2002). Similarly, as
?sh larvae are visual predators, factors such as the
amount of solar radiation may either aid feeding by
increasing the visibility of prey, or reduce survival by
rendering ?sh more visible to other predators (Fortier
et al. 1996). Solar radiation may also reduce larval
survival through the damaging e?ect of ultraviolet
radiation on nucleic acids or through epidermal damage
(Zagarese and Williamson 2001), or indirectly through
its impact on secondary production (Kouwenberg et al.
1999).
The aim of this study was to determine the in?uence
of water temperature, wind speed and direction, rainfall
and solar radiation on planktonic growth, pelagic
duration and settlement intensity of the Caribbean sur-
geon?sh, Acanthurus chirurgus. The results provide new
foci for future studies on processes a?ecting the growth
and survivorship of pelagic larvae in tropical marine
systems.
Materials and methods
Sampling
From January 1984 to January 1988, newly settled
Acanthurus chirurgus were collected daily from two
small patch reefs in Punta de San Blas, on the
Caribbean coast of Panama (934?N, 7858?W). These
reefs were located about 2.5 km apart, in shallow
water and were surrounded by beds of seagrass on the
back-reef of a larger fringing reef. To reduce any
potential for migration of newly settled individuals
reef A (about 1 m deep) was about 25 m from any
other reef, while reef B (about 1.5 m deep) was about
15 m from other areas of coral. A. chirurgus are
strongly site-attached for the ?rst few weeks after
settlement. Newly settled A. chirurgus were collected
using a hand net with the help of the anaesthetic
Quinaldine. Further details and a map are in Rob-
ertson (1992). In the laboratory, total and standard
lengths were recorded and the sagittal otoliths
removed for processing.
Otolith analysis
Robertson (1992) showed that peaks in settlement of
Acanthurus chirurgus occurred during the new-moon
half of the lunar cycle. As relatively few ?sh settled
each day, individuals collected from the patch reefs
were pooled into lunar cohorts from full moon to full
moon to increase sample sizes. Otoliths from a sub-
sample of 30% of the settlers in a lunar cohort, or at
least 15 ?sh, were analysed from each month. In 24 of
43 lunar months less than 15 individuals settled and
therefore the otoliths of all individuals were processed.
Within each month, catches were divided among 1-mm
standard length (SL) size classes and a sub-sample re-
moved in proportion to the abundance of ?sh in each
size class.
Sagittal otoliths were mounted on a glass slide using
thermoplastic cement (Crystalbond) so that the distal
end of the otolith protruded over the edge of the slide.
The otolith was then ground to the nucleus using wet
lapping ?lm (12?0.3 lm). The Crystalbond was reheated
and the polished side of the sagitta was mounted face
down on the slide, so that the rostral end of the otolith
could be ground down to produce a thin transverse
section incorporating the nucleus. Sections were viewed
at 1,100? magni?cation using a compound microscope
linked to a video camera and computer. The width of
each successive increment in otoliths was measured as
the distance between two consecutive opaque zones
using an image analysis system (OPTIMAS). Increments
were always analysed along the longest radius of the
otolith. It was assumed that the ?rst increment closest to
the core was formed at the time of hatching (Thresher
et al. 1989; Wellington and Victor 1989). Sagittal oto-
liths from 11 individuals were analysed three times to
determine the error involved in counts and measure-
ments of increments. Four ?sh with settlement marks
(Wilson and McCormick 1999) on their otoliths were
excluded from the analysis to avoid the confounding
e?ects of immigration of settled ?sh to patch reefs on
our estimates of settlement intensity. Due to the di?-
culties in identifying the details of microstructure at
the otolith margin (Wilson and McCormick 1999)
settlement marks are discernible only a few days after
settlement.
A strong linear relationship between sagittal radius
and standard length of newly settled Acanthurus chir-
urgus together with 25 post-settlement individuals
292
(r2=0.85, P<0.001, n=625) supported the assumption
that there was a proportional relationship between
otolith and somatic growth of young A. chirurgus. Daily
deposition of increments within the otoliths of young
A. chirurgus was validated as follows. Ten A. chirurgus
were collected from patch reefs in Punta de San Blas
using hand nets and the anaesthetic clove oil. These
patch reefs were not the same reefs where daily samples
of ?sh were collected. The ?sh were immediately trans-
ferred to an aquarium and allowed to acclimatise for
several days. Of the 10 individuals, 8 survived this per-
iod and were then placed in an aquarium that contained
a solution of 500 mg/l tetracycline in seawater. The
aquarium was left in complete darkness for 24 h to al-
low the tetracycline to be incorporated into the otoliths
of the ?sh. The ?sh were then removed and left in
another aquarium containing clean aerated seawater for
16 days. At this time, ?sh were sacri?ced and one
sagittal otolith from each individual was processed for
analysis as described above. The sagittal section was
viewed under a high power microscope equipped with a
UV light source. The number of increments following
the ?uorescent tetracycline mark within the otolith was
counted and compared to the number of days ?sh were
kept in aquarium after being marked. The sagittal oto-
lith of six out of eight individuals displayed a ?uorescent
mark. Counts of increments from this mark to the edge
of the otolith closely approximated the number of
days the ?sh were left in an aquarium after their re-
moval from the tetracycline solution (mean=16.7?0.3
days; range=14?18 days), con?rming daily increment
formation.
Environmental variables
As environmental monitoring in the Punta de San Blas
began only in the early 1990s, the physical data used in
this study were obtained from the Galeta Marine Lab-
oratory on Galeta Point, approximately 5 km east of the
entrance to the Panama Canal and 100 km northwest of
Punta de San Blas. A previous study has shown that
wind measured at Punta de San Blas is strongly corre-
lated to wind strength and direction at Galeta Point
(Robertson et al. 1999).
Hourly measurements of sea surface temperature,
rainfall, wind speed and direction and solar radiation
were available from monitoring at Galeta Point (see
http://striweb.si.edu/esp/physical_monitoring/download_
galeta.htm). Daily averages were calculated except for
rainfall. This was recorded as a total (mm/day). Wind
directions were separated into orthogonal components
using cosine and sine functions that were then multiplied
by the average daily wind speed to derive north?south
(cosine) and east?west (sine) vectors. On a large spatial
scale (10 s of km), the north?south component of the
wind resulted in o?shore-onshore winds, while the
east-westerly component generated along-shore winds at
our study sites.
Growth, settlement and pelagic larval duration (PLD)
The ages derived from otolith analyses were used as
estimates of pelagic larval duration (PLD) since ?sh
were collected on the day of settlement. Otolith growth
was used as an estimate of somatic growth thus avoiding
the errors introduced by back-calculation of ?sh size
from otoliths (Chambers and Miller 1995). Otolith
growth was measured over three time periods: from 0 to
6 days, 7 to 25 days and from 26 to 50 days after
hatching. The ?rst of these periods represents the likely
duration of feeding by newly hatched larvae on food
supplied by the yolk sac (Randall 1961). Bergenius et al.
(2002) identi?ed 7?25 days after hatching as a period of
rapid growth of Acanthurus chirurgus that has a major
in?uence on the magnitude of settlement and recruit-
ment of this species. On average, the mean larval dura-
tion of A. chirurgus is 55 days; 50 days after hatching
92% of the individuals are still pelagic (Bergenius 1998).
The ?nal growth period thus encompassed most of the
remainder of the PLD of this species.
As individuals collected during the same lunar month
varied in the length of their PLD, monthly data sets of
?sh settlement were reorganised by the date on which
individual ?sh had hatched before the relationship be-
tween environmental variables and larval growth was
examined. Hatch-dates could easily be determined as
otolith examination provided an estimate of age and the
collection date of each individual was known. By reor-
ganising monthly data sets of ?sh settlement by hatch-
date, individual growth histories were standardised so
that they could be compared with the environmental
conditions experienced by individuals on each day of
their larval lives. Growth, PLD and settlement data in
lunar months with less than two newly settled individ-
uals, after reorganising by hatch-date, were pooled with
data of the succeeding month. From this information, a
mean environmental history and averages of PLD and
growth rates could be calculated for each individual ?sh
and subsequently for each lunar cohort to reduce the
variable nature of individual responses to environmental
signals (Chambers and Leggett 1987). This yielded a
data set of 36 lunar cohorts for which there was su?-
cient data to analyze (12 lunar months in the dry season
and 24 lunar months in the wet).
Environmental variables were also compared with the
magnitude of settlement on a monthly basis after set-
tlement data sets had been reorganised by hatch-date.
Catches were standardised to numbers per 100 m2 and
were log10 transformed in order to reduce the e?ects of
small-scale (10s to 100s of meters) patchiness in settle-
ment (Doherty and Williams 1988).
Comparison of environmental and biological variables
The in?uence of the environmental variables on PLD,
planktonic growth and settlement intensity was exam-
ined using multiple regressions. These were used to ?nd
293
leading indicators for any time series (e.g. growth) from
other time series without lagged e?ects (Davies 2002).
Analyses were run three times. The ?rst included only
data sets from dry seasons, the second only data from
wet seasons and the third analysis the complete data set.
Multiple regression analyses were run using SAS (1987)
software. The ??all possible subsets technique?? was used
to produce an optimum predictive model. For q inde-
pendent variables there are 2q possible models to predict
Y. Out of all possible models the best three for each q
were selected based on maximum R2 (the coe?cient of
multiple determination) and Mallow?s Cp selection sta-
tistics. Mallow?s Cp measures the best possible ?t of a
model based on whether or not the error mean square
contains only random variation (Draper and Smith
1981). The best model was then selected based on the
lowest Cp. As strong correlations among independent
variables can mask or exaggerate outcomes of regression
models, correlation matrices, partial correlations and
tolerance levels were calculated for environmental vari-
ables and examined for evidence of collinearity. When
strong collinearity was detected the environmental var-
iable displaying the strongest correlation to other inde-
pendent variables was excluded from the multiple
regression. In all models, the adjusted coe?cient of
multiple determination (R2adj) was calculated and used
for interpretation rather than the R2, since the former
takes into consideration the number of degrees of free-
dom (Quinn and Keough 2002). Environmental vari-
ables were often log10 transformed to conform to the
assumption of normality made by the analysis. Bivariate
and partial correlations were calculated using the SPSS
statistical package.
As the outcomes of multiple regression analyses are
very susceptible to outliers, Cook?s distances (Cook?s D)
were calculated for each data point. This statistic esti-
mated the amount residuals would change by excluding
a case from computation in a regression analysis. Cases
with a large Cook?s D (i.e. extreme outliers) were re-
moved from the analysis, as these can falsify parameter
estimates of a regression model (Barnett and Lewis
1994).
Results
Environmental seasonality
There are two distinct seasons in Caribbean Panama, a
wet season from mid-April to mid-December and a dry
season for the remainder of the year (Cubit et al. 1989).
Seasonal averages, maxima and minima of water tem-
perature, rainfall, solar radiation, and wind speed re-
corded at Galeta Point for the period of the study are
presented in Table 1. During the wet season from mid-
April to mid-December winds are light and variable in
direction (Fig. 1) and there are heavy rains. Strong on-
shore winds from a northerly direction (Fig. 1), increased
solar radiation and cooler water temperatures characte-
rise the dry season, which encompasses the remainder of
the year.
Patterns of ?sh settlement
Settlement varied from 0 to 660 A. chirugus per 100 m2
of patch reef per lunar month over the 43-month sam-
pling period (Fig. 2). Settlement peaked from November
through to January of each year. Settlement of A. chir-
urgus on patch reefs was correlated (r=0.75) to
recruitment measured on eight large (>300 m2) reefs
spread over a 15 km2 area that were censused during the
week before the new moon, when settlement peaks
(Robertson 1992), indicating that patch reef collections
were a good estimate of larger scale patterns of
recruitment.
Environmental correlates of larval growth
Otolith growth from 0 to 6 days after hatching When the
data sets were pooled among seasons, rainfall and solar
radiation had a weak in?uence on the growth of
Acanthurus chirurgus during the ?rst 6 days of planktonic
life (Table 2). However, when dry and wet seasons were
analysed separately, regression models accounted for
greater amounts of variability in otolith growth in early
larval life. In the dry season, solar radiation accounted for
57% of the variation in growth and these variables were
negatively correlated (Table 2; Rp: 
0.85; Fig. 3a). Dur-
ing the wet season, the model accounted for less (38%) of
the variation in the data set, and rainfall and along-shore
windswere selected by themodel as factors that in?uenced
early growth rates. The partial correlation between rain-
fall and early larval growth was negative (Table 2; Rp:

0.47), while the relationship between along-shore winds
and growth was positive (Table 2; Rp: 0.53). During the
wet season along-shore winds were dominated by winds
from a westerly direction (Fig. 2).
Otolith growth from 7 to 25 days after hatching The
regression model of the complete data set accounted for
38% of the variability in growth of Acanthurus chirurgus
Table 1 Means, maxima and minima of environmental variables
during wet and dry seasons measured at Galeta Point from July
1984 to December 1987. The data were recorded by the Marine
Environmental Science Program of the Smithsonian Tropical Re-
search Institute
Variable Season Mean?SE Maxima Mininima
Temperature
(C)
Wet 28.18?0.02 30.00 25.75
Dry 27.06?0.04 28.60 25.17
Rainfall
(mm day
1)
Wet 10.57?0.68 204.60 0
Dry 1.35?0.28 54.70 0
Solar radiation
(watts m2)
Wet 3918.98?55.91 7234.00 205
Dry 5183.54?81.18 7291.00 819
Wind speed
(ms
1)
Wet 2.55?0.13 3.82 1.77
Dry 5.37?0.20 6.70 4.32
294
between 7 and 25 days after hatching and identi?ed only
solar radiation as a signi?cant in?uence on growth rates
(Table 2; Rp: 
0.58). In the dry season, along-shore
winds and growth were highly correlated (Table 2; R:

0.94; Fig. 3b) and the regression model accounted for
as much as 87% of the variation in growth. While the
along-shore wind component was positively correlated
with growth from 0 to 6 days, it was strongly negatively
correlated with growth from 7 to 25 days after hatching.
No signi?cant relationship was found between the
environmental variables and growth from 7 to 25 days
after hatching during the wet season.
Otolith growth from 26 to 50 days after hatch-
ing Multiple regression analysis of the entire data set
could not identify any combination of environmental
variables that were correlated with growth between 26
and 50 days after hatching (Table 2). However, these
analyses were signi?cant when the data were split into wet
and dry seasons. In the latter season, there was a strong
positive correlation between rainfall and otolith growth
(Table 2; Rp: 0.87; Fig. 3c). Growth at this time was also
correlated with along-shore winds (Table 2,Rp: 0.66) and
together these environmental variables accounted for
74% of the variation in growth. During the wet season,
only 27% of the variability in larval growth could be ac-
counted for by environmental factors (Table 2; rainfall
and temperature; Rp: 0.55 and 
0.49, respectively).
Environmental correlates of planktonic larval duration
As much as 65% of the variability in pelagic larval
duration could be explained by a regression model that
included solar radiation and along-shore winds (Table 2;
Rp solar radiation: 0.64; Rp along-shore wind: 0.26).
When regressions analysed wet and dry season data sets
separately, along-shore winds accounted for 67% of the
variability in PLD during the dry season (Table 2; R:
0.84; Fig. 4a), while solar radiation accounted for 23%
of the variation in larval duration in the wet season
(Table 2; R: 0.51).
Environmental correlates of settlement magnitude
Environmental variables accounted for only 22% of the
variation in numbers of settlers arriving on patch reefs
for the dry and wet seasons combined (Table 2). In this
model, the magnitude of settlement was negatively cor-
related to solar radiation and water temperature (Ta-
ble 2; Rp: 
0.52 and ?0.32, respectively). Analysis of dry
season data produced a model that accounted for a
greater amount of the variation in settlement (40%) and
included the environmental variables of rainfall and
solar radiation. Both were negatively correlated with
settlement patterns (Table 2; Rp rainfall: 
0.63 and Rp
solar radiation: 
0.70; Fig. 4b). Due to the high col-
Fig. 1 Average monthly along-
shore wind (ms
1 ? direction)
and onshore-o?shore wind
(ms
1 ? direction) at Galeta
Point during the study period.
d Represents months in the dry
season and w indicates months
in the wet season
Fig. 2 Acanthurus chirurgus.
Numbers of A. chirurgus
settling per lunar month on two
patch-reefs in Punta de San
Blas in Panama, Caribbean,
from July 1984 to December
1987. Settlers were standardised
to numbers per 100 m2. The
value above each bar is the
number of ?sh sampled for
otolith analysis
295
linearity (R: 0.9) among wind variables and temperature
in the dry season, the latter was excluded from the
analysis. Multiple regression analysis could not identify
a model that accounted for a signi?cant amount of
variation in settlement magnitude during the wet season.
Discussion
The strength and direction of the relationships between
growth and environmental variables di?ered with larval
stage, suggesting that the impact of the environment on
growth of larval Acanthurus chirurgus is dependent on
ontogeny. Moreover, correlations between planktonic
growth, PLD, settlement intensity and environmental
variables were stronger during the dry than the wet
seasons and usually involved di?erent combinations of
environmental variables each season. This suggests that
future studies of tropical ?sh larvae that examine the
in?uence of environmental variables on early life his-
tory characteristics require the data to be analysed
separately for each season, as is the case for temperate
studies.
Growth
Di?erent combinations of the environmental variables
solar radiation, rainfall and along-shore winds ac-
counted for a varying amount (from 13% to 86%) of the
changes in otolith growth of Acanthurus chirurgus from
hatching until the time of settlement. In the ?rst 6 days
Table 2 Acanthurus chirurgus. Summary of results of multiple
regression analyses that compared water temperature (C), rainfall
(mm day
1), solar radiation (watts m
2), onshore-o?shore wind
(ms
1 ? wind direction) and alongshore wind (ms
1 ? wind
direction) with larval growth (otolith growth in microns), pelagic
larval duration (days) and settlement intensity of A. chirurgus.
Larval growth was analysed from 0 to 6, 7 to 25 and 26 to
50 days after hatching. Analyses were repeated on the entire data
set and on data separated into wet and dry seasons. See text for
details of analysis techniques. C(p) = Mallow?s Cp (selection
criteria),n = number of replicates (lunar months) included in the
regression analysis, R2 = coe?cient of multiple determination,R2
adj = adjusted coe?cient of multiple determination, P (F-test) =
signi?cance values of the F-test associated with the ANOVA
computed to test the null hypothesis: all the b?s are 0 (bi=pop-
ulation parameter for the slope of the linear relationship between
a dependent variable and an independent variables), where alpha
was put at 0.05, independent variables = the environmental
variables included in the optimum predictive model selected by
the regression analysis, P (t-test) = signi?cance values of the t-
test computed to test the null hypothesis: bi = 0, of the linear
relationship between a dependent variable and an independent
variables with all the other independent variables partialled out
(i.e. it is a test for the partial correlations), ns = the F or t-test
was not signi?cant and in which the case signi?cance value was
not reported, Rp = partial correlation coe?cient. The symbol 
= when only one variable was selected by the analysis Pearson?s
correlation coe?cient (R) is given. The number of months iden-
ti?ed as outliers and excluded from the analysis is in parentheses
following the number of replicates
Dependent variable Cp R2adj R
2 P (F-test) Independent variables Rp P (t-test)
Growth 0?6 n=35 (1) 2.14 0.13 0.18 0.040 Rainfall 
0.42 0.014
Solar radiation 
0.33 0.049
Growth 0?6 dry season n=11 (2) 4.94 0.57 0.73 0.038 Solar radiation 
0.85 0.004
Onshore-o?shore wind 
0.58 ns
Along-shore wind 0.60 ns
Rainfall 
0.54 ns
Growth 0?6 wet season n=23 (1) 2.77 0.38 0.47 0.007 Rainfall 
0.47 0.033
Solar radiation 
0.35 ns
Along-shore wind 0.53 0.013
Growth 7?25 n=35 (1) 1.93 0.38 0.41 0.000 Solar radiation 
0.58 <0.001
Along-shore wind 
0.30 ns
Growth 7?25 dry season n=10 (2) 
0.5 0.86 0.88 0.000 Alongshore wind 
0.94  <0.001
Growth 7?25 wet season n=24 
0.04 0.01 0.04 ns Solar radiation 
0.19  ns
Growth 26?50 n=35 (1) 4.94 0.15 0.22 ns Temperature 
0.29 ns
Rainfall 0.39 0.026
Solar radiation 0.32 ns
Along-shore wind 
0.23 ns
Growth 26?50 dry season n=12 1.84 0.74 0.79 0.001 Rainfall 0.87 <0.001
Along-shore wind 0.66 0.026
Growth 26?50 wet season n=23 (1) 1.27 0.27 0.33 0.017 Temperature 
0.49 0.020
Rainfall 0.55 0.008
PLD n=35 (1) 0.33 0.65 0.67 0.000 Solar radiation 0.64 0.000
Along-shore wind 0.26 0.016
PLD dry season n=12 -0.75 0.67 0.69 0.001 Along-shore wind 0.84  0.001
PLD wet season n=23 (1) 0.89 0.23 0.26 0.013 Solar radiation 0.51  0.013
Settlers n=36 2.02 0.22 0.27 0.006 Solar radiation 
0.52 0.001
Temperature 
0.32 0.045
Settlers dry season n=12 1.85 0.40 0.51 0.041 Rainfall 
0.63 0.037
Solar radiation 
0.70 0.017
Settlers wet season n=24 2.20 0.13 0.21 ns Rainfall 0.33 ns
Onshore-o?shore wind 0.42 0.045
296
of larval life, growth rates were negatively correlated
with the amount of sunlight during the dry season, a
result that contrasts with the outcomes of previous
studies. Typically, these have found that sunlight has a
positive, albeit indirect e?ect on larval growth (Heath
et al. 1988; Gallego et al. 1996) by warming surface
waters and increasing food production (Cushing 1990;
Heath 1992; 1995). Recent work, however, has shown
that high levels of light may negatively a?ect ?sh larvae.
Ultraviolet radiation may induce direct damage to the
DNA and proteins of eggs and larvae, causing reduced
growth or death (Lesser et al. 2001; Zagarese and Wil-
liamson 2001). Moreover, Boeuf and Le Bail (1999)
suggest that beyond an upper intensity, sunlight may
decrease growth and even be deadly due to negative
impacts on visual development, which in turn will a?ect
feeding activities and prey selection. Our results suggest
that the high levels of sunlight that occur during the dry
season in Punta de San Blas may detrimentally a?ect
growth rates. Alternatively, sunlight may make larvae
more susceptible to predators during early develop-
mental stages, when sensory and locomotory systems are
poorly developed. This may a?ect survivorship or could
force young ?sh deeper in the water column, away from
the depth of optimal light intensities for feeding and
growth (Fortier and Harris 1989).
In addition to sunlight, rainfall was also negatively
correlated with growth of Acanthurus chirugus up to
6 days after hatching, although unlike solar radiation,
this correlation was signi?cant during the wet season.
Fig. 3a?c Acanthurus
chirurgus. a Partial regression
plot of growth (microns) of
A. chirurgus from 0 to 6 days
after hatching and solar
radiation (watts m2) during the
dry season, corrected for the
e?ects of along-shore wind
(ms
1 ? wind direction),
onshore-o?shore wind
(m/s ? wind direction) and
rainfall (mm/day). b Regression
plot of growth (microns) from
7 to 25 days after hatching and
along-shore winds (ms
1 ? wind
direction) during the dry
season. c Partial regression plot
of growth (microns) from 26 to
50 days after hatching and
rainfall (mm/day) during the
dry season, corrected for the
e?ect of along-shore wind
297
Heavy rains in the wet season that decrease salinity in
surface layers may be detrimental to young larvae due to
their small surface to volume ratio and lack of protective
scaling against changes in the osmotic environment. A
turbid upper layer of fresh water may also reduce light
penetration and thus feeding by young ?sh. However,
heavy rains are also associated with increased run-o?
from rivers and nutrients in inshore environments
(D?Croz et al. 1999), potentially resulting in better
feeding conditions for larger, more developed larvae.
This may explain the positive relationship between
rainfall and growth of larvae 26?50 days after hatching
in both the dry and the wet seasons.
There was a complex relationship between growth of
Acanthurus chirurgus larvae and winds. Growth up to
6 days after hatching was positively correlated with
along-shore winds, negatively correlated from 7 to
25 days after hatching and again positively correlated
with along-shore winds after 25 days. While variations
in winds might in?uence growth at local scales by
directing the surface waters and the food sources they
contain away from or along the coast, little evidence
exists to assess this possibility. It has been hypothesised
that winds may a?ect growth in larval ?sh due to their
in?uence on small-scale turbulence. At optimum levels,
turbulence is thought to increase the probability of
encounter between larval ?sh predators and their prey,
thus increasing growth rates and survivorship (Gallego
et al. 1996; Dower et al. 1997). It is possible that the
complex relationship between growth and wind identi-
?ed in this study is actually a re?ection of a change in the
interaction between wind-induced turbulence and
growth throughout ontogeny.
Water temperature was not signi?cantly correlated
with growth rates of larval Acanthurus chirugus in the
?rst 25 days of larval life, although there was a weak
negative correlation between these variables after this
time during the wet season. This result di?ers from
studies in temperate and other tropical regions where
water temperature is thought to be one of the primary
determinants of growth rates (Campana and Hurley
1989; Suthers and Sundby 1993; Meekan et al. 2003). In
tropical NW Australia, Meekan et al. (2003) found that
water temperature explained 30% of the variation in
growth rate of a larval pomacentrid. These contrasting
results may be due to the relatively small temporal
(seasonal, monthly and diurnal) ranges in temperatures
that occur in Punta de San Blas (Table 1).
Pelagic larval duration
Our analysis suggested up to 65% of the variability in
the PLD of Acanthurus chirugus could be accounted for
by a regression model that included the variables of
along-shore winds and solar radiation. When seasons
were analysed separately, along-shore winds were
strongly correlated with PLD in the dry season
(R=0.84), while solar radiation was moderately corre-
lated (R=0.51) with PLD in the wet season. The positive
relationship between along-shore winds and PLD is
surprising, given that such winds were positively corre-
Fig. 4a, b Acanthurus
chirurgus. a The regression
relationship between pelagic
larval duration (PLD; days) and
along-shore winds (ms
1 ? wind
direction) during the dry
season. b Partial regression plot
of number of settlers and solar
radiation (watts m2) during the
dry season, corrected for the
e?ect of rainfall (mm/day)
298
lated to early and late larval growth. Fast growing larvae
might be expected to have shorter larval durations than
slow growing larvae and we found a signi?cant negative
correlation between average growth to settlement and
larval duration (regression analysis, r2=
0.45
P<0.001, n=603). One possibility that might account
for this result is that some compensatory growth oc-
curred during the late larval phase, so that larvae that
were small at earlier stages grew relatively fast at this
time. Such explanations assume that larvae do not typ-
ically delay settlement, and this delay appears to be the
case in this species (Bergenius 1998).
Our results suggest that solar radiation (and with it
UV radiation exposure) may have a negative e?ect on
the development of the pelagic eggs and larvae of
acanthurids, as there was a positive correlation between
solar radiation and PLD of Acanthurus chirurgus during
the dry season. Recent work suggests that the early life
history stages of ?shes and invertebrates are particularly
sensitive to the UV radiation present in natural sunlight
(Lesser et al. 2001; Zagarese and Williamson 2001). The
positive correlation between solar radiation and PLD
was consistent with the negative relationship between
solar radiation and otolith growth up to 25 days, since a
longer larval duration is likely to be a consequence of
decreased growth.
Numbers of settlers per lunar month
Solar radiation and temperature explained 22% of the
variability in numbers of settlers arriving on patch reefs
each lunar month when data were pooled between sea-
sons. When analysed separately, solar radiation and
rainfall were both negatively correlated with settlement
and the regression model described 40% of the vari-
ability in settlement intensity. The e?ect of these vari-
ables on settlement might be explained by their in?uence
on growth rates, given that Bergenius et al. (2002) and
Wilson and Meekan (2002) have demonstrated that
growth during the early larval stages is an important
determinant of the numbers of ?sh surviving the
planktonic phase. As noted previously, high levels of
sunlight may decrease growth rates or increase egg
mortality rates (Frank and Leggett 1981). Similarly, if
heavy rains, which also occur occasionally in the
Caribbean during the dry season, dramatically change
osmotic environments, there might be a resulting in-
crease in mortality of ?sh eggs and newly hatched larvae.
Our study suggests that the environmental variables
of solar radiation, along-shore winds and rainfall may
be used to predict the growth of Acanthurus chirurgus
larvae at di?erent stages of their pelagic life and that the
impact of environmental variables on replenishment
may vary between seasons. To date, most studies of the
e?ects of the environment on life history characteristics
of marine ?shes have examined only one environmental
variable. We suggest that the outcomes of such studies
should be treated with caution and that as many vari-
ables as possible should be included in tests of envi-
ronmental in?uences.
Acknowledgements Statistical advice was gratefully received from
D. Ryan, D. Donald and D. Coomans. We are grateful to B. Green
and two anonymous reviewers who commented on the paper.
Environmental data were provided by the Smithsonian Tropical
Research Institute?s Marine Environmental Science Program. Lo-
gistic support was provided by James Cook University and the
Australian Institute of Marine Science.
References
Anderson JT (1988) A review of size dependent survival during pre-
recruit stages of ?shes in relation to recruitment. J Northwest
Atl Fish Sci 8:55?66
Bailey KM, Houde ED (1989) Predation of eggs and larvae of
marine ?shes and the recruitment problem. Adv Mar Biol 25:1?
83
Barnett V, Lewis T (1994) Outliers in statistical data. Wiley, Eng-
land
Bergenius MAJ (1998) The in?uence of larval growth and stage
duration on settlement variability in a coral reef ?sh. Honours
thesis, Department of Marine Biology, James Cook University,
Townsville
Bergenius MAJ, Meekan MG, Robertson DR, McCormick MI
(2002) Larval growth predicts the recruitment success of a coral
reef ?sh. Oecologia 131:521?525
Boeuf G, Le Bail PY (1999) Does light have an in?uence on ?sh
growth? Aquaculture 177:129?152
Campana SE, Hurley PCF (1989) An age- and temperature-medi-
ated growth model for cod (Gadus morhua) and haddock
(Melanogrammus aegle?nus) larvae in the Gulf of Maine. Can J
Fish Aquat Sci 46:603?613
Chambers RC, Leggett WC (1987) Size and age at metamorphosis
in marine ?shes: an analysis of laboratory-reared winter
?ounder (Pseudopleuronectes americanus) with a review of
variation in other species. Can J Fish Aquat Sci 44:1936?1947
Chambers RC, Miller TJ (1995) Evaluating ?sh growth by means
of otolith increment analysis: special properties of individual
level longitudinal data. pp 155-176 In: Secor DH, Dean JM,
Camapana SE (eds) Recent developments in ?sh otolith re-
search. University of South Carolina Press, South Carolina
Cubit JD, Ca?ey HM, Thompson RC, Windsor DM (1989)
Meteorology and hydrography of a shoaling reef ?at on the
Caribbean coast of Panama. Coral Reefs 8:59?66
Cushing DH (1990) Plankton production and year-class strength in
?sh populations: an update of the match/mismatch hypothesis.
Adv Mar Biol 26:250?293
Cushing DH (1995) Population production and regulation in the
sea: a ?sheries perspective. Cambridge University Press, Cam-
bridge
Davies CA (2002) Statistical methods for the analysis of repeated
measurements. Springer, Berlin Heidelberg New York
D?Croz L, Robertson DR, Martinez JA (1999) Cross-shelf distri-
bution of nutrients, plankton, and ?sh larvae in the San Blas
Archipelago, Caribb Panama Rev Biol Trop 47:203?215
Doherty PJ (1991) Spatial and temporal patterns in recruitment. In:
Sale PF (ed) The ecology of ?shes on coral reefs. Academic, San
Diego, pp 261?293
Doherty PJ, Williams DMcB (1988) The replenishment of coral
reef ?sh populations. Oceanogr Mar Biol Annu Rev 26:487?551
Dower JF, Miller TJ, Leggett WC (1997) The role of microscale
turbulence in the feeding ecology of larval ?sh. Adv Mar Biol
31:170?220
Draper NR, Smith H (1981) Applied regression analysis. 2nd edn.
Wiley, New York
Fortier L, Harris RP (1989) Optimal foraging and density-depen-
dent competition in marine ?sh larvae. Mar Ecol Prog Ser
51:19?33
299
Fortier L, Gilbert M, Ponton D, Ingram G, Robinou B, Legendre
L (1996) Impact of freshwater on a subarctic coastal ecosystem
under seasonal ice (south-eastern Hudson Bay, Canada). III.
Feeding success of marine ?sh larvae. J Mar Syst 7:251?265
Frank KT, Leggett WC (1981) Prediction of egg development and
mortality rates in Capelin (Mallotus villosus) from meteoro-
logical, hydrographic, and biological factors. Can J Fish Aquat
Sci 38:1327?1338
Fukahara O (1990) E?ects of temperature on yolk utilisation, ini-
tial growth, and behaviour of unfed marine ?sh-larvae. Mar
Biol 106:169?174
Gallego A, Heath MR, McKenzie E, Cargill LH (1996) Environ-
mentally induced short-term variability in the growth rates of
larval herring. Mar Ecol Prog Ser 137:11?23
Green BS, Fisher R (2004) Temperature in?uences swimming
speed, growth and larval duration in coral reef ?sh larvae. J Exp
Mar Biol Ecol 299:115?132
Heath MR (1992) Field investigations of the early life stages of
marine ?sh. Adv Mar Biol 28:1?228
Heath MR, Henderson EW, Baird DL (1988) Vertical distribution
of herring larvae in relation to physical mixing and illumina-
tion. Mar Ecol Prog Ser 47:211?228
Hjort J (1914) Fluctuations in the great ?sheries of northern Eur-
ope reviewed in the light of biological research. Rapp PV Reun
Cons Perm Int Explor Mer 20:1?128
Houde ED (1987) Fish early life dynamics and recruitment vari-
ability. Am Fish Soc Symp 2:17?29
Kouwenberg JHM, Browman HI, Runge JA, Cullen JJ, Davis RF,
St-Pierre J-F (1999) Biological weighting of ultraviolet (280?
400 nm) induced mortality in marine zooplankton and ?sh. II.
Calanus ?nmarchicus (Copepoda) eggs. Mar Biol 134:269?284
Lesser MP, Farrell JH, Walker CW (2001) Oxidative stress, DNA
damage and p53 expression in the larvae of Atlantic cod (Gadus
morhua) exposed to ultraviolet (290?400 nm) radiation. J Exp
Biol 204:157?164
McCormick MI, Hoey AS (2004) Larval growth history determines
juvenile growth and survival in a tropical marine ?sh. Oikos
106:225?242
Meekan MG, Carleton JH, McKinnon AD, Flynn K, Furnas
M.(2003) What determines the growth of tropical reef ?sh lar-
vae in the plankton: Food or temperature? Mar Ecol Prog Ser
256:193?204
Polo A, Yufera M, Pascual E (1991) E?ects of temperature on egg
and larval development of Sparus auratus L. Aquaculture
(Amsterdam) 92:367?375
Quinn GP, Keough MJ (2002) Experimental design and data
analysis for biologists. Cambridge University Press, Cambridge,
UK
Randall JE (1961) A contribution to the biology of the convict
surgeon?sh of the Hawaiian Islands, Acanthurus triostegus. Pac
Sci 15:215?272
Robertson DR (1992) Patterns of lunar settlement and early
recruitment in Caribbean reef ?shes at Panama. Mar Biol
114:527?537
Robertson DR, Swearer SE, Kaufmann K, Brothers EB (1999)
Settlement vs. environmental dynamics in a pelagic-swimming
reef ?sh at Caribbean Panama. Ecol Monogr 69:195?218
SAS (1987) SAS statistical guide. Version 6. Edition SAS Institute,
Cary, N.C.
Searcy SP, Sponaugle S (2001) Selective mortality during the larval-
juvenile transition in two coral reef ?shes. Ecology 82:2452?2470
Suthers IM, Sundby S (1993) Dispersal and growth of pelagic
juvenile Arcto-Norwegian cod (Gadus morhua), inferred from
otolith microstructure and water temperature. ICES J Mar Sci
50:261?270
Thresher RE, Colin PL, Bell LJ (1989) Planktonic duration, dis-
tribution and population structure of Western and Central
Paci?c damsel?shes (Pomacentridae). Copeia 1989:420?434
Utne-Palm AC, Stiansen JE (2002) E?ect of larval ontogeny, tur-
bulence and light on prey attack rate and swimming activity in
herring larvae. J Exp Mar Biol Ecol 268:147?170
Vigliola L, Meekan MG (2002) Size at hatching and planktonic
growth determines post-settlement survivorship of a coral reef
?sh. Oecologia 131: 89?93
Wellington GM, Victor BC (1989) Planktonic larval duration of
one hundred species of Paci?c and Atlantic damsel?shes
(Pomacentridae). Mar Biol 101:557?567
Wilson DT, McCormick MI (1999) Microstructure of settlement
marks in the otoliths of tropical reef ?sh. Mar Biol 134:29?41
Wilson DT, Meekan MG (2002) Growth-related advantaged for
survival to the point of replenishment in the coral reef ?sh Steg-
astes partitus (Pomacentridae). Mar Ecol Prog Ser 231:247?260
Zagarese HE, Williamson CE (2001) The implications of solar UV
radiation exposure for ?sh and ?sheries. Fish Fish 2:250?260
300