136 
Fishery Bulletin 106(2) 
ferent water masses that have a constant temperature 
and offer constant food conditions. Water masses, there- 
fore, have the potential to influence increment width by 
affecting growth rates (Sponaugle and Pinkard, 2004) 
or influencing otolith chemistry; the challenge is to 
distinguish between these two mechanisms. 
Materials and methods 
This study was conducted on One Tree Island (OTI) 
in the southern Great Barrier Reef (GBR), Australia. 
This platform reef is closed off from the surrounding 
ocean for 3-5 hours during each low tide (Ludington, 
1979). Thus, the geochemical properties of the lagoon 
water may be distinct from the surrounding ocean 
water (Atema et ah, 2002); this feature may influence 
the increment widths of the otoliths of the fish residing 
in the lagoon. The otoliths of newly settled neon dam- 
selfish ( Pomacentrus coelestis) collected from different 
water masses around OTI (i.e., reef slope ( = oceanic 
water) vs. lagoon water) were examined and the incre- 
ment widths were measured. A series of manipulative 
experiments were conducted with oceanic and lagoon 
water to examine the effects on increment widths of P. 
coelestis otoliths in the absence of potentially confound- 
ing factors, including food and temperature. The incre- 
ment widths derived from these experiments were then 
compared with the results of another study ( Patterson et 
al., 2004a) where otolith chemistry of the same experi- 
mental fish was examined as that used in this study 
to determine if otolith chemistry may affect increment 
width. In addition, scanning electron microscopy (SEM) 
was used to examine otolith microstructure because 
variation in elemental chemistry may potentially alter 
crystal orientation. 
Wild fish 
Newly settled P. coelestis were collected by divers 
using hand nets and clove oil from two sites on the 
reef slope (ocean 1 and 2) and two sites within the 
lagoon (lagoon 1 and 2) at OTI (23°30'S, 152°06'E; 
n=7 per site). The sagittae were removed and one was 
randomly chosen for analysis. Transverse sections 
were prepared by embedding the sagittae in Epofix 
resin (Struers, Milton, Australia) and polishing with 
800-grit-size lapping paper on a grinding wheel and 
then fine-polishing with 3-, am lapping film until the 
primordium was visible and daily rings could be dis- 
tinguished. The daily increments (validated by Flood, 
2000) were counted and measured with a Leica image 
analysis system (Leica Microsystems, North Ryde, 
NSW, Australia). A drop of immersion oil was placed on 
each otolith to enhance its appearance, and an image of 
each otolith was captured and saved at a magnification 
of 400 x with a Leica DC300 digital video camera. The 
distance between each increment on the distal side of 
each otolith was measured with the use of the software 
package Leica IM50. 
Experimental fish 
Experimental procedures were identical to those 
described in Patterson et al. (2004a), but are briefly 
repeated here. Presettlement ( sensu Kingsford et al., 
2002) P. coelestis were collected by using two to four 
moored light traps at several locations within 100 m of 
the reef slope (see Patterson et al., 2004a, Fig. 1) during 
three consecutive austral summer field seasons (January 
and February, 2000-02). Light traps were illuminated 
for three hours each night (21:00-22:00, 00:00-01:00, 
and 03:00-04:00) and were checked on the first high 
tide after sunrise. Samples from the light traps were 
sorted on the boat, and target species were taken to 
the laboratory to be counted and identified. Fish used 
in each experiment were collected over a period of two 
days. Additional individuals were immediately frozen to 
use as control fish (see Crystallography section). 
The experimental design for each year differed only in 
the number of replicate fish placed in each tank at the 
beginning of the experiment (2000, n- 6; 2001, n = 10; 
2002, n = 3; n varied because of among-year differences 
in the availability of fish). Experimental fish were al- 
lowed a minimum of 24 hours to acclimate to experi- 
mental conditions before the start of the experiment. It 
was not possible to measure the standard length (SL) 
of fish before the start of the experiment because this 
process stressed, and in some cases damaged the fish, 
which could have confounded the results. All fish were 
approximately the same size at the beginning of the 
experiment (~15 mm SL). Three replicates of each treat- 
ment (lagoon water vs. ocean water) were set up in 200- 
L tanks and fish were randomly placed in each tank. 
Aerated tanks were kept under shade cloth in a wet 
laboratory and were randomly positioned. Water (from 
lagoon ebb tide plumes and oceanic waters collected 
outside of plumes; see Patterson et al., 2004a, Fig. 1) 
was changed daily (~90%) and debris and potential food 
items were removed with a 30-pm net. The water had 
adequate time (~1 h minimum) to come to a standard 
experimental temperature before being placed in the 
tanks. Natural temperature variations were therefore 
controlled for. Fish were fed twice a day (am and pm) 
with cultured Artemia sp. and wild zooplankton col- 
lected with channel nets or with a 120-pm net during 
horizontal tows. Quantities of food were measured (500 
mL per feeding time) and fish were fed in random order 
at each feeding time to eliminate any potential bias in 
food allocation. The water temperature was recorded at 
each feeding session. No differences were detected in 
water temperature during the experimental treatment 
(mean conditions for all years and treatments ±stan- 
dard error [SE]: am feeding: 26.5°C ±0.09, pm feeding: 
28.2°C ±0.09). For all years of the experiment, there 
were no significant differences in water temperature 
by tank nested in treatment (ANOVA: am: F x 4 =0.09, 
P>0.05; pm: F 14 =0.06, P>0.05) or by treatment (ANO- 
VA: am: F 14 =2.61, P>0.05; pm: F 14 =5.13, P>0.05). Fish 
were maintained for a total of nine days in each experi- 
ment (i.e., nine days per year) after which three fish 
