196 



Fishery Bulletin 91|2). 1993 



of nets and light-traps (Gregory & Powles 1988), nets 

 and plankton pumps (Brander & Thompson 1989), and 

 towed nets and purse-seines (Kingsford & Choat 1985), 

 but have dealt with the less-diverse fauna of temper- 

 ate waters. 



The purpose of this study was to compare several 

 types of towed and seine nets and an automated light- 

 trap (Doherty 1987) in terms of taxa, numbers, and 

 sizes of larvae and pelagic juveniles of coral reef fishes 

 captured. These methods represent the range of sam- 

 pling devices currently used to collect larval and pe- 

 lagic juvenile fishes. For the towed nets, we used di- 

 mensions and mesh size normally employed to sample 

 larval and pelagic juvenile fishes. We used designs of 

 purse-seine and light-trap which had been subject to 

 thorough field testing (Kingsford & Choat 1985 and 

 1986, Kingsford et al. 1991, Doherty 1987). For each 

 sampling device we obtained the following informa- 

 tion: ( 1 ) Taxonomic composition of samples at the level 

 of family; (2) patterns of density and size structure in 

 selected taxa; and (3) temporal patterns in the density 

 of selected taxa over short time-periods. The program 

 also provided information on the logistic constraints 

 associated with each sampling method. 



Our findings will be useful to those designing sam- 

 pling programs for larval and pelagic juvenile stages 

 of demersal fishes in tropical and other areas, and 

 should have some generality because the taxa sampled 

 included a wide variety of body shapes and swimming 

 capabilities. Among the taxa studied are families of 

 great importance in coral reef ecosystems as adults 

 (Apogonidae, Atherinidae, Callionymidae, Gobiidae, 

 Labridae, Pomacentridae), and several are also impor- 

 tant in commercial, sport, or subsistence fisheries 

 throughout the tropics (Carangidae, Lethrinidae, 

 Lutjanidae, Mullidae, Nemipteridae, Platycephalidae, 

 Scaridae). All are abundant in ichthyoplankton sam- 

 ples in tropical coastal areas, especially in the Indo- 

 Pacific. 



Materials and methods 



Sampling and identification procedures 



We sampled at 150-600 m off the fringing reefs at 

 Watsons Bay on the NW side of Lizard Island in the 

 lagoon of the northern Great Barrier Reef, Australia 

 (145°26'E, 14°40'S). Water depth was 20-30 m over a 

 sandy bottom (Fig. 1). This site was chosen for its 

 proximity to the logistic support offered by the Lizard 

 Island Research Station, a base for much work on the 

 pelagic phase of coral reef fishes (Leis 1991b). Also, it 

 offered relatively sheltered conditions from the 15- 

 25 kn southeasterly winds present during the sampling 



LIZARD ISLAND 



Prevailing Trade Wind 



>^ 



500m. 



Palfrey Island 



Figure 1 



Lizard Island, Great Barrier Reef, Australia, showing loca- 

 tion of study area and position of sampling sites for light- 

 traps, towed nets, and purse-seines at Watsons Bay. Coral 

 reefs are shown as broken lines. Lizard Island ( 145 : 26'E. 

 14°40'S) is located 30 km off the eastern coast of mainland 

 Australia. 



period. This was particularly important for the conti- 

 nuity of sampling over a number of nights. 



We sampled on the nights of 2, 3, 5, 6, and 7 Decem- 

 ber 1986, starting at a minimum of 1.25 h after sun- 

 set. Sampling never continued past 0200 h. New moon 

 was on 2 December 1986. Nocturnal sampling reduces 

 potential bias due to vertical distribution because 

 ichthyoplankton show little vertical stratification at 

 night in the study area (Leis 1986, 1991a). In addi- 

 tion, the nets should operate at peak efficiency at night 

 due to lessened visual avoidance. Finally, the aggrega- 

 tion devices are effective only at night because they 

 depend on self-generated light to attract fishes. 



We concentrated our analyses on data from 3, 5, and 

 6 December because we were able to take and process 

 all planned samples from all gears only on these nights. 

 For some gears, it was possible to examine temporal 

 trends over the full sampling period. 



Six different sampling devices were deployed each 

 night. Three nets were towed from the 14 m catama- 



