Choat et al Comparison of ichthyoplankton sampling methods 



197 



ran RV Sunbird at lm/s along a fixed 1km path. The 

 towed nets were fitted with flowmeters and were 

 washed with pumped seawater. Details of each collec- 

 tion device are as follows. 



1 A neuston net of mouth dimensions 1.0x0.3 m with 

 0.5 mm mesh was rigged to sample water between the 

 bows of the catamaran. Typically, the net sampled to a 

 depth 0.1m and filtered 187-312 nvVtow. Four tows 

 were taken per night. 



2 A bongo net (McGowan & Brown 1966) of 0.85 m 

 mouth diameter per side, and with 0.5 mm mesh, was 

 towed from an "A'-frame at the stern. The RV Sun- 

 bird draws lm, and the net was towed so its top was 

 lm below surface and on the vessel's centerline in wa- 

 ter which had not been disturbed by the passage of its 

 twin hulls. The volume of water filtered for each side 

 of the net was 498-673 m 3 tow. Samples from only the 

 port-side net were analyzed. Four tows were taken per 

 night. 



3 A Tucker trawl (Tucker 1951) with nominal mouth 

 dimensions of 2x2 m and of 3mm mesh was towed in 

 the same position as the bongo net. At a towing speed 

 of 1 m/s, a diver estimated that the bottom bar of the 

 net trailed the top bar by -0.5 m, so the effective mouth 

 area was -3.8 m 2 . Between 3240 and 4570 m 3 of water 

 were filtered per tow. Four tows were taken per night. 

 Both the bongo net and the Tucker trawl used the 

 same depressor. 



Time constraints and the logistics of rigging and 

 deploying each net precluded randomising the order of 

 bongo and Tucker trawl tows, so they were taken in 

 blocks of four, with the order alternating from one 

 night to the next. Neuston net samples were taken 

 during the Tucker trawl tows. 



4 A plankton mesh purse-seine of 14x2 m (Kingsford 

 & Choat 1985) of 0.28 mm mesh was used to take 

 samples of -32 m 3 each. This estimate was based on 

 the ideal cylinder of water enclosed by the net at the 

 beginning of pursing and made no allowance for herd- 

 ing of fishes during deployment or loss during pursing. 

 There was no estimate of variation in the volume en- 

 closed by the net sets. The net was deployed from a 



4 m dinghy adjacent to the northern end of the tow 

 path (Fig. 1). Wind conditions precluded effective de- 

 ployment of this net at greater distances offshore. Two 

 to four samples were taken per night. 



5 Two automated light-traps (Doherty 1987) were de- 

 ployed from an anchored boat adjacent to the center of 

 the tow path and -700 m from the purse-seine site. 

 Traps were positioned at ~10m apart. Entries into the 

 trap were at 0.5-1 m below surface. The second trap 

 began to sample 30 min after the first, and both traps 

 sampled for hourly intervals, resulting in continuous 

 sampling in overlapping, 1 h segments. The trap de- 

 ployment was staggered to allow for clearing and pro- 



cessing of each trap after the 1 h fishing period. Eight 

 to nine 1 h light-trap samples were taken per night. 

 6 A battery-powered fluorescent light source identi- 

 cal to that in the trap (Doherty 1987) was deployed 

 from a second boat anchored at the purse-seine site. 

 After 1 h in the water, the light was set adrift and the 

 water around it immediately sampled by the same 

 purse-seine used in (4) above. Our estimates of what 

 was attracted to the light included only those indi- 

 viduals that were within -2 m (i.e., radius of the seine 

 at pursing) of the light at the time of seining. Four to 

 five light-seine samples were taken per night. Purse- 

 seine (no light, (4) above) and light-seine samples were 

 interspersed during the night. 



Our goal was to sample simultaneously using six 

 methods in the same location over several nights, so 

 as to avoid confounding comparisons of methods with 

 temporal or spatial variation. The purse-seine, light- 

 seine, and light-trap samples were taken throughout 

 the nightly sampling period. At the same time, the RV 

 Sunbird sampled with the towed nets. Logistic prob- 

 lems required two compromises in this program. Bongo 

 tows and Tucker trawl tows (and simultaneous neus- 

 ton tows) were done in sequential blocks of four each 

 night as discussed in (3) above. The purse-seine and 

 light-trap samples were taken 700 m apart because it 

 was not possible to duplicate these devices and thus 

 randomize their positions. The RV Sunbird tow track 

 covered the area between these two. 



Fishes from the towed nets, purse-seines, and light- 

 seines were immediately fixed in 10% formalin seawa- 

 ter. Samples from the light-traps were maintained alive 

 until returned to the Research Station where they were 

 subsequently fixed in 100% ethanol or 10% formalin 

 seawater. All fish were transferred to 70% ethanol for 

 at least a month prior to measurement. 



For light-traps and light-seines, density is expressed 

 as number per sample. Catches from the towed net 

 and purse-seine collections were standardized to the 

 number of fishes/1000 m 3 on the basis of flowmeter 

 records or purse-seine geometry. 



All fishes were removed from samples and identified 

 to family following Leis & Rennis (1983) and Leis & 

 Trnski ( 1989). Standard lengths were measured to the 

 nearest 0. 1 mm using a Bioquant software package that 

 allows for measurement of enlarged camera lucida im- 

 ages offish and accommodates curvature of specimens. 

 The accuracy of electronic measurement was monitored 

 by measuring subsamples manually with calipers and 

 eye-piece micrometers. In a few samples with very large 

 numbers of certain taxa such as gobiids, the catch was 

 subsampled and a minimum of 10% of the sample mea- 

 sured. For some analyses, fishes were divided into small 

 (<6mm) and large (>6mm) size-groups. This was done 

 because, on the basis of results reported here, the light- 



