262 



Fishery Bulletin 103(2) 



and stored an hourly average (Roberts-). Photoperiod 

 data were downloaded from the South African Astro- 

 nomical Observatory database. 3 Pearson Rank correla- 

 tion was used to measure the correlation between GSI 

 and temperature, and GSI and photoperiod trends. 



Results 



Histological examination of the gonads revealed that 

 although juveniles possess both testicular and ovarian 

 tissue simultaneously (i.e., as hermaphrodites) they 

 mature as either a male or female (Table 1) and are 

 therefore late gonochorists (rudimentary hermaphro- 

 dites). Gametogenesis was similar to that described for 

 other late gonochoristic sparids e.g., Pterogymnus lania- 

 rius (Booth and Hecht, 1997). The size at 50% maturity 

 was estimated at 292 and 297 mm FL for females and 

 males, respectively (Fig. 3), and in both cases is equiva- 

 lent to an age of about five years (Brouwer and Griffiths 

 2004). A likelihood ratio test revealed that there was 

 no significant difference between male and female L 50 

 (P>0.5)or or (P>0.1). Complete ( 100% ) maturity for both 

 sexes occurred at 480 mm FL, an age of about 15 years 

 (Brouwer and Griffiths 2004). The sex ratio was 1F:1.3M 

 (n=1776); a chi-square test with Yates' correction factor 

 revealed that this sex ratio was a significant difference 

 from unity (P<0.01). 



Three age-related POF stages were identified within 

 the ovaries of captive spawned carpenter (Fig. 2). Stage- 

 1 POFs (0-6 hours) were very loosely arranged and ap- 

 peared as a long convoluted string with a large clearly 

 defined lumen. The granulosa cells were clearly visible 

 and widely spaced and had clearly visible nuclei (Melo, 

 1994). Stage-2 POFs (7-24 hours) are smaller and more 

 densely packed but still have a visible lumen. The gran- 

 ulosa cells are closely packed and dense. Stage-3 POFs 

 (25-48 hours) are small and densely packed. There is 

 no lumen and the granulosa cells are closely arranged 

 and no longer distinguishable from one another. After 

 48 hours at 16°C, POFs were no longer detectable. 



Mean GSI and the proportions of ripe (stage-5) and 

 ripe, running (stage-6) fish increased in November and 

 remained high until April (Figs. 4 and 5), indicating 

 that carpenter are summer spawners. The presence of 

 early POFs from November to March (sample numbers 

 being too low for April) supported the macroscopically 

 determined spawning season. The monthly spawning 

 fraction did, however, reveal that spawning frequency 

 was highest in January and February when the fish 

 spawned at two-day intervals and lowest in November 

 and April when they were found to spawn every 2-3 

 days (Table 2). 



o 



1B0 230 280 330 380 430 480 530 580 



130 180 230 280 330 380 430 480 530 580 



Fork length (mm) 



Figure 3 



The proportion of mature carpenter (Argyrozona argy- 

 rozona) in length classes sampled in the Tsitsikamma 

 National Park. The curves were fitted with a 2-parameter 

 logistic ogive. 



2 Roberts, M. J. 1999. CD-ROM, Tsitsikamma National Park 

 oceanographic data, version 1.0. Marine and coastal man- 

 agement. Private Bag X2, Rogge Bay 8012, South Africa. 



:l http://www.saao.ac.za [Accessed August 2000], 



Batch fecundity was positively correlated with both 

 fish mass (r=0.71) and fork length (r=0.71). No correla- 

 tion was found between fish length and relative batch 

 fecundity (eggs/fish somatic mass) (Fig. 6). The propor- 

 tion of fish with stage-1 POFs revealed that spawning 

 frequency and length of the spawning season increased 

 with fish length (Table 3). Accounting for size-related 

 patterns in spawning season (Fig. 7) and frequency, we 

 found that annual fecundity increased allometrically 

 with mass (Fig. 8) and age (Table 4). Hydrated egg size 

 was significantly smaller and more variable (average 1.0 

 mm ±0.16) in fish below the length at 100% maturity 

 (480 mm FL) than those above this length (1.1 mm 

 ±0.09) U-test, P<0.005). 



