Choat et al.: Comparison of ichthyoplankton sampling methods 



20! 



distinct from net samples, and that samples taken 

 by light-seine were intermediate between net and 

 light-trap samples. Tucker trawl samples were almost 

 completely distinct from bongo, neuston, and seine 

 net samples. Bongo net samples formed a more 

 discrete group than did the neuston and seine net 

 samples. 



The data sets for size analysis were heterogenous. 

 Therefore, we attempted only to test for differences in 

 density among methods within selected size-classes 

 using single-factor ANOVA(df 3,39; p<0.05). The power 

 of these tests to detect differences among methods was 

 low. For apogonids, gobiids, lutjanids, and poma- 

 centrids, there were sufficient numbers for statistical 

 comparisons across the first three size-classes (i.e., 

 <6mm, Fig. 4). For all four families, density estimates 

 provided by the bongo net were as high as, and in 

 many cases higher than, those provided by the other 

 nets. The Tucker trawl provided the lowest density 

 estimates. 



For the larger sizes (>6mm), low or zero catches in 

 some size-classes precluded statistical tests in most 

 cases. We compared the Tucker trawl, which is de- 



signed to capture such large stages with the bongo 

 net. The few tests that were possible show that in no 

 instance did the Tucker trawl provide higher density 

 estimates than the Bongo net (Fig. 4). 



Two taxa, pomacentrids and gobiids, were sufficiently 

 abundant to allow for comparisons of density by 2 mm 

 size-classes between the aggregation devices. For 

 pomacentrids we tested the 7-15 mm size-classes. 

 Light-traps caught significantly higher numbers of 

 pomacentrids in the 7, 9, and 11mm size-classes than 

 the light-seines (Fig. 4B). The two aggregation devices 

 provided similar estimates of numbers for the 13 and 

 15mm size-classes (Fig. 4B). The difference in overall 

 density for pomacentrids sampled by light-traps and 

 light-seines is due to the greater number of poma- 

 centrids in the 7, 9, and 11mm size-classes in the 

 light-trap catches. Pomacentrid larvae >14mm were 

 collected by the light-seine on one night only. 



Although we did not statistically test the gobiid data, 

 the light-seine appeared to collect greater numbers of 

 smaller (<4mm), and the light-trap greater numbers 

 of larger (>8 mm), individuals (Fig. 4B). The light-seine 

 collected few gobiids >6 mm and the light-trap almost 

 no gobiids <6 mm. Sizes of apogonid and lutjanid fishes 

 sampled by the light-seine were similar to those of the 

 purse-seine (Fig. 4C). No lutjanids and only four 

 apogonids were collected by the light-traps. 



Results of pooled samples from three nights for eight 

 taxa (Materials and methods) by the different nets (Fig. 

 5) reflect both entry of fish into nets and subsequent 

 extrusion. Most of the fishes taken by all nets were 

 small (Table 2, Fig. 4). Bongo nets consistently provided 

 the highest estimates of density of small fishes, espe- 

 cially gobiids, apogonids, lutjanids, labrids, and 

 lethrinids. This reflects both the low-avoidance and high- 

 retention properties of this fine-mesh net. The purse- 

 seine filtered only small volumes of water, but provided 

 high estimates of density, especially for gobiids, 

 apogonids, and lutjanids (Fig. 4). Extrusion is probably 

 minimal, due to the passive mode of filtering and the 

 very fine mesh of this seine. Neuston nets provided low 

 estimates of density for all families except two that 

 concentrate in the surface layer — atherinids and mullids 

 (Leis 1991a). Density estimates from the Tucker trawl 

 were low for all families, most probably due to the loss 

 of smaller larvae through its large mesh. Both atherinids 

 and mullids, which attained large size (Table 2), were 

 also poorly represented in Tucker trawl catches, possi- 

 bly because the Tucker trawl did not sample the 

 neustonic habitat of these taxa. 



For aggregation devices, we compared densities of 

 the important families identified by PCA, excepting 

 apogonids and lutjanids which were rare or absent 

 from light-traps. Light-traps collect mainly large indi- 

 viduals, so the samples were subdivided by size 



