Davis and West: Reproductive biology of Lutjanus vittus from North West Shelf of Australia 



227 



were washed in freshwater, counted, dried for at least 

 24 h at 60°C, and weighed to the nearest 0. 1 mg. 



Size-at-maturity <L o50 ) was denned as the smallest 

 length at which half of the fish had yolked or ripe- 

 stage ovaries. Weighted least-squares nonlinear regres- 

 sion was used to select the model (Gompertz curve) 

 that best described the relationship between propor- 

 tion of mature fish and fish length. The simplest model 

 was chosen in which higher parameter models did not 

 significantly improve the fit using the F-ratio statistic 

 (Schnute 1981). 



RGI's were log-transformed to stabilize variances, 

 and a 2x2 factorial ANOVA was used to test for differ- 

 ences in RGI's between samples collected in August 

 and October and between consecutive years. Differ- 

 ences in RGI over the four periods sampled during the 

 main spawning season (September 1982-April 1983) 

 were tested using ANOVA, and the period effect was 

 partitioned into a linear component and deviations from 

 a linear component to test for trends. 



The log-likelihood ratio x 2 was used to test for inde- 

 pendence in contingency-table data (frequency offish 

 with and without postovulatory follicles by time of day, 

 month, etc.). Cells with expected frequencies <5 were 

 excluded from statistical tests. Repeated-measures 

 ANOVA was used to test for differences in the num- 

 bers of eggs/mg tissue taken from the inner and outer 

 walls, and from anterior, middle, and posterior parts 

 of the ovary of five fish. The relationship between log 

 fecundity and log length at different times of the spawn- 

 ing season was examined using ANCOVA. 



Results 



Maturation of ovaries 



Size-frequency and stage of whole oocytes in ovaries 

 representing the developmental sequence of matura- 

 tion (Fig. 3) indicated that the size of oocytes is corre- 

 lated with their stage of development but there is con- 

 siderable overlap between stages. The pattern of oocyte 

 development is asynchronous. Apart from the infer- 

 ence that L. vittus is a multiple spawner, the number 

 of spawnings cannot be inferred from modes in the 

 size-frequency distribution (Hunter & Goldberg 1980). 



Size-at-maturity 



Ovaries were considered mature if their most advanced 

 oocytes were yolked or ripe, as judged by the appear- 

 ance of the whole oocyte. To avoid including mature 

 fish that had regressed to an unyolked stage towards 

 the end of the spawning season, only data taken be- 

 tween September and February were used. The small- 



est ripe female observed was 142 mm long (Fig. 4). All 

 fish >220mm were mature (Fig. 5). The Gompertz curve 

 was fit to the proportion of mature females by 10 mm 

 length groupings: 



Y(t)=yj> 



-e-gll - <„' 



(ll 



where Y(t) = proportion mature at length t, and y„, g, 

 and t„ are parameters. This provided the best fit with 

 the least number of parameters. The length at which 

 50% of females were mature (L 050 ) was calculated to 

 be 154mm. L 050 is reached at age-1 (Davis & West 

 1992), and all females at age-3 and older were mature 

 (Fig. 5). 



Spawning season 



Only data on fish >200 mm were used to describe sea- 

 sonal changes. The main spawning season appears to 

 be September-April (Fig. 6). The presence offish with 

 hydrating oocytes throughout the year indicates the 

 spawning season is protracted, although only four fish 

 with hydrated oocytes were collected in June and Au- 

 gust 1983. Some mature fish had only unyolked oo- 

 cytes in their ovaries during April-August (Fig. 6). 



The mean relative gonadal index (RGI) peaked in 

 September-October and then declined gradually, reach- 

 ing its lowest value in June (Fig. 6). There was a 

 marked and significant increase in mean RGI during 

 August-October (2x2 factorial ANOVA, month effect, 

 F=309, df 1,339, p<0.001), with the same increase in 

 both years (interaction effect not significant: F=2.4, 

 p>0.1), and there was a weak but significant differ- 

 ence between years (F=4.3, p<0.05). There were sig- 

 nificant differences in RGI between the four periods 

 sampled during the main spawning season, Septem- 

 ber 1982-April 1983 (ANOVA, F=50.5, df 3,706, 

 p<0.001), predominantly due to a linear component 

 (F=130.1, df 1,706, p<0.001) caused by a decline in 

 RGI over this period (slope -0.236, SE 0.021). On the 

 other hand, there was no significant difference in pro- 

 portion of ripe fish throughout this period (likelihood 

 ratio x 2 = 6.3, df 3, p=0.10), nor any indication of a 

 decline in the proportion. This suggests that spawning 

 activity remained at about the same level during the 

 main spawning period but that gonads of individual 

 fish were depleted by successive spawnings. 



Lunar periodicity in spawning activity 



Ovaries of mature fish (rc=374) sampled in November- 

 December 1982 were examined histologically for evi- 

 dence of spawning activity. Data were grouped into 1 d 

 periods according to moon age. Two measures of spawn- 

 ing were used: proportion of fish with postovulatory 



