564 



Fishery Bulletin 103(4) 



Adult fish were injected in the coelomic cavity with 0.05 

 g/mL tetracycline in sterile saline solution at concentra- 

 tions equivalent to 0.05 g/kg body weight (McFarlane 

 and Beamish, 1987). The approximate weight of each 

 individual was estimated from the relationship between 

 weight and SL. Juveniles were mass marked by immer- 

 sion in a tetracycline solution (concentration: 0.5g/L) 

 in seawater for 12 hours (overnight). The tetracycline 

 generally forms a very effective time-marker in oto- 

 liths; it fluoresces when viewed under ultraviolet light 

 (Geffen, 1992). 



The experiment commenced in May 2002 and fish 

 were sacrificed after six months, one year (June 2003), 

 and one-and-a-half years (November 2003). Ten fish had 

 readable otoliths for which validation was attempted. 

 Otolith sections were viewed with a compound micro- 

 scope and incident ultraviolet light in a darkened room. 

 When a fluorescing tetracycline band was identified, 

 its position in relation to the edge was measured. The 

 section was then examined under reflected white light 

 and measurements of increment widths and marginal 

 increments were recorded. Known time at liberty, ex- 

 pressed as a proportion of one year, was then compared 

 with estimated time at liberty by using the growth of 

 the otoliths. If estimated time at liberty equalled ac- 

 tual time at liberty, it supported the hypothesis that 

 opaque increments were deposited annually. Juveniles 

 and adults were collected on each occasion to determine 

 whether increments were deposited annually, early and 

 late in life. 



The length of time for increment formation was also 

 estimated by calculating the number of days after tetra- 

 cycline treatment. The number of days after treatment 

 was estimated by comparing the position of the tetra- 

 cycline mark with that of the last (marginal) opaque 

 increment and the width of a full annual increment 

 with the following formula: 



L,=L„[l-e- A '"-'»'], 



Number of days after treatment 



TE-MI 

 IW 



x365. 



where TE = otolith growth after treatment; 

 MI = the marginal increment; and 

 IW = the final full increment width. 1 



where L a = the asymptote of the growth curve (average 



maximum length); 

 L, = length at age t\ 

 K = the rate at which the growth curve 



approaches the asymptote (Lj\ 

 t = age of fish in years; 

 t = the theoretical origin of the growth curve 



(i.e., the hypothetical age of the fish when 



it has no length); and 

 e = the base of the natural logarithm. 



Differences in growth curves for A polyacanthus from 

 each reef sampled were visualized by using the tech- 

 nique of Kimura (1980), where 95% confidence ellipses 

 were generated around the parameter estimates of K and 

 L x . Confidence ellipses that did not overlap indicated dif- 

 ferences in growth parameters and enabled the pooling 

 of data from sites within reefs at each distance stratum. 

 The parameter t was constrained to minus 0.05 to take 

 into account the approximate size of A. polyacanthus at 

 hatching (5 mm: Kavanagh, 1998, 2000). 



Mortality 



The instantaneous rate of mortality (Z) was calculated 

 by using log-linear regression analyses of age-frequency 

 data sets for A. polyacanthus populations from each 

 reef (Pauly, 1984). With this method, recruitment was 

 assumed to be consistent over time at each reef. The 

 natural logarithm of the number of fish sampled from 

 each age class was compared with their corresponding 

 age. Year classes to the left of the age-frequency mode 

 were excluded from the analysis because our sampling 

 technique was biased against small A. polyacanthus. 

 Fish greater than 60 mm were collected. The slope of 

 the regression line between year classes estimated the 

 instantaneous mortality rate (Z): 



Z = F +M, 



where F = fishing mortality; and 



M = natural mortality (Gust et al. 



2002). 



Growth 



It was hypothesized that patterns of growth would 

 vary with distance from the coast. Growth rates were 

 described by using von Bertalanffy growth functions 

 that provided the best fit to size-at-age data when com- 

 pared with estimates of the Schnute growth function 

 (Schnute, 1981). The von Bertalanffy expression for 

 length at age t (L t ), as a function of time is 



1 We assumed similar IWs for fish older than 3 years. For 

 fish 3 years or younger the IW was calculated as an average 

 from all experimental fish. 



Because there is no fishery for A. polyacanthus on the 

 GBR, F equals zero and therefore Z estimates natural 

 mortality only. Annual survival rate estimates were then 

 calculated according to the equation S = e~ z (Ricker, 

 1975). Comparisons of the slopes of age-frequency rela- 

 tionships (for estimates of Z) were made by using analy- 

 sis of covariance (ANCOVA) according to the procedures 

 of Zar (1999). Data from each site were pooled for each 

 reef because in many cases sample sizes were too small 

 to provide reliable estimates of mortality at the site 

 level. Similarities in mortality rates among replicate 

 reefs within distance strata allowed us to pool data at 

 the strata level so that comparisons of mortality between 

 shelf positions could be made. 



