222 



Fishery Bulletin 89(2), 199) 



or environment (Stearns and Crandall 

 1984). Such variability could have impor- 

 tant implications for the baitfishing in- 

 dustry. Juveniles could be caught before 

 they have had a chance to spawn, because 

 the liftnets used in the baitfishery are not 

 size-selective and catch all sizes of En- 

 crasicholina and Spratelloides (see Evans 

 and Nichols 1984). 



The aims of this study were, first, to 

 examine the length at maturity and the 

 spawning seasonality of the six most 

 abundant baitfish species (Encrasicho- 

 lina devisi, E. heterolobus, Spratelloides 

 delicatulus, S. lewisi, S. gracilis, and the 

 apogonid Archamia zosterophora); sec- 

 ondly, to determine whether spawning is 

 random or correlated with environmental cycles; and, 

 thirdly, to assess the effects, if any, of the commercial 

 fishery on these reproductive parameters. 



Methods and materials 



Sampling 



Samples of six species of baitfish were collected from 

 three sites each month for two years. Fish from com- 

 mercial bait catches came from Munda and Tulagi, and 

 fish from an unexploited control site came from Vona 

 Vona, Solomon Islands. Sites and sampling methods 

 are described in Milton et al (1990). Samples were 

 usually collected between 2100 and 2200 hours. Two 

 sites, Munda and Vona Vona, were in enclosed coral 

 reef lagoons with little water movement and were ap- 

 proximately 20 km apart. The other site, Tulagi, con- 

 sisted of a series of protected bays which opened into 

 a narrow channel between two islands and was located 

 over 300km southeast of the other sites. Each month 

 a random sample of over 100 fish of each species from 

 each site was preserved in 4% formaldehyde and taken 

 to the laboratory for analysis. On each sampling occa- 

 sion, the water temperature, cloud cover, moon phase, 

 and wind speed and direction were recorded. At least 

 two 5-minute horizontal plankton tows (~100m 3 ) 

 were made with a 250-jjmm mesh net (0.5 m diameter) 

 1 hour prior to fish collection. The daily rainfall at a 

 village adjacent to each site was recorded throughout 

 the study. 



Laboratory analyses 



Fish were measured (standard length (SL) + 0.5 mm) 

 and weighed (± 0.001 g) and the gonads were removed 

 and weighed ( ± 0.001 g). The gonads of up to 10 females 

 of each species and of any other females with enlarged 



oocytes were randomly subsampled from each site 

 sample each month. Gonads were embedded in paraf- 

 fin wax, sectioned at 9fim, and stained with Ehrlich's 

 haemotoxylin and eosin (McManus and Mowry 1964). 

 Gonad maturation stages were defined following Cyrus 

 and Blaber (1984) and Hunter and Goldberg (1980). 

 Each gonad was staged, based on the relative numbers 

 of cells at each developmental stage (Young et al. 1987; 

 Table 1), and the presence of any postovulatory follicles 

 was noted. Gonosomatic indices (GSI) were calculated 

 as the ratio of wet gonad weight to somatic weight 

 (total weight minus gonad weight) expressed as a 

 percentage. 



Size-at-sexual-maturity was determined from the 

 length at which a fish developed ripe eggs (Table 1: 

 stage 4). Among fish that were only examined macro- 

 scopically, the criterion for sexual maturity was a 

 gonosomatic index greater than the minimum GSI of 

 fish that were shown histologically to have ripe eggs. 



Fish examined histologically were considered to be 

 in spawning condition if more than 85% of the eggs 

 were fully hydrated, running-ripe (Table 1: late stage 

 5). The mean GSI ± 2 standard errors of these fish was 

 calculated for each species at each site. The GSI value 

 of the lower 95% confidence limit of the mean GSI of 

 spawning fish was used as the indicator of spawning 

 among fish examined macroscopically. The proportion 

 in spawning condition in each sample was then calcu- 

 lated from the fraction of the sample (examined both 

 histologically and macroscopically) with a GSI greater 

 than this value. This criterion was used, as it was a con- 

 servative estimate of the real proportion spawning. 



Plankton samples were split in half with a Folsom 

 plankton splitter and one fraction was dried to a con- 

 stant mass to provide an estimate of relative zooplank- 

 ton biomass. The other fraction was sorted and the 

 number of Encrasicholina eggs and larvae (Delsman 

 1931), the number of other eggs, larvae, and potential 



