Anderson et al : Current velocity and catch efficiency in sampling settlement-stage larvae of coral reef fisfies 



409 



subsurface and surface channel nets, including only those 

 families that represented I'Tc of the catch of either net for 

 S.E. LSI (r,=0.227, n = \6. P>0.05; Fig. 3A) or for So. Bock 

 (r,=0.088, n=l6, P>0.05; Fig. 3B). In addition, the sizes of 

 larvae (Table 3) caught by light traps and channel nets 

 (but excluding families with <3 individuals collected per 

 sampling device [acanthurids, gobiids, and syngnathids] ) 

 did not differ in the mean (sign test: n = ll, P>0.05), me- 

 dian (« = 11, P>0.05l, or maximum (n = ll, P>0.05) length 

 between sampling devices. For five families that differed 

 in length between sites and that were represented by at 

 least three individuals (Table 3), channel nets caught 

 significantly larger larvae at S.E. LSI than at So. Bock in 

 mean (n=5, P=0.03) and median {n=5. P=0.03) length, but 

 not in maximum length (n=5, P>0.05). 



Current velocity and catch efficiency 



Over a 10-d period in which flow-meter readings were 

 taken simultaneously at both sites, mean daily current 

 velocity was three-times higher at So. Bock than at S.E. 

 LSI (/test: ?=-7.92, df=18, P=0.0001; Fig 4A). Light traps 

 deployed at So. Bock caught on average only 3.1'7f of the 

 total number of fish collected by light traps at both sites, 

 whereas the channel nets positioned at S.E. LSI caught an 

 average of 43.8'7f of the total catch for nets positioned at 

 both sites (Fig. 4B). The mean proportional abundance of 

 all larvae caught by light traps was significantly lower at 

 So. Bock than at S.E. LSI it =-50.3, df=24, P<0.0001) but 

 there was no significant difference in proportional catch 

 for channel nets (^=1.36, df=24, P=0.19). The difference in 

 catch efficiency for each sampling device was also indicated 

 by the opposite trends of abundance of taxa (apogonids, 

 blennioids, labrids, pomacentrids, but not monocanthids) 

 caught in at least nominally greater numbers by light 

 traps at the low-current site (S.E. LSI) compared with the 

 greater catch of these same taxa by channel nets at the 

 high-current site (So. Bock). In addition, and consistent 

 with our hypotheses concerning current velocity, light 

 traps had higher family richness and diversity at S.E. LSI 

 than at So. Bock, whereas the opposite was true for chan- 

 nel nets (Table 2). 



Discussion 



Differential representation of taxa 

 between sampling methods 



Light traps and channel nets differed in the taxonomic 

 composition and the relative abundance of their catch. 

 Expectedly, channel nets had appreciably higher family 

 richness and diversity than did light traps, and this is 

 consistent with our hypothesis that light traps are more 

 selective (also see Choat et al., 1993) because not all larvae 

 exhibit a photopositive response. One of the main differ- 

 ences in the relative abundance of larvae between sam- 

 pling devices in the central Bahamas is that channel nets 

 collect a large proportion of labrids and leptocephalus 

 larvae (Shenker et al., 1993; Thorrold et al., 1994a, 1994b, 

 1994c; Mojica et al., 1995; this study) but light traps do 

 not (this study). Depth-dependent distributions of larvae 

 are also likely to contribute to differences in the relative 

 abundance of taxa because light traps catch larvae from 

 an unknown depth range (over which larvae are attracted 

 to the light) whereas surface and subsurface channel nets 

 operate at discrete depths of 0-1 m and 2-4 m, respec- 

 tively. Moreover, not only can surface and subsurface 

 channel nets differ in their catch (Fig. 3), the relative 

 abundance of particular taxa between surface and subsur- 

 face nets can switch over time (Thorrold et al., 1994c). 



The similar sizes of larvae from families caught by both 

 light traps and channel nets are somewhat inconsistent 

 with the findings by Choat et al. (1993). who found that 

 larvae caught by light traps were larger than those caught 

 by towed nets and seines. They attributed this differerence 

 in size to a stronger photopositive response by larger pe- 

 lagic larvae. The smaller size of larvae that they caught in 

 towed nets and seines may also indicate that larger larvae 

 may better sense the presence of these nets and avoid cap- 

 ture. Channel nets may decrease net avoidance by larvae 

 because their stationary position may lessen water distur- 

 bance and hence detection of the net by incoming larvae, or 

 possibly because other larval behaviors are exhibited. 



Differential representation of taxa between sites 



The greatest difference between light traps and channel 

 nets was in their relative catch between sites. Light traps 

 were ineffective in collecting larvae at So. Bock, whereas 

 channel nets collected a lower but not significantly differ- 

 ent proportion of larvae at S.E. LSI. This result cannot 

 be explained simply by differences in the abundance of 

 taxa between sites; the same pattern was observed for 

 taxa sufficiently represented by both sampling devices 

 (apogonids, blennioids, labrids, pomacentrids). These sites 

 did differ substantially in mean current velocity, with 

 average flow rates at So. Bock three times higher than 

 at S.E. LSI. Shenker et al. (1993) observed that larvae 

 collected by channel nets were significantly more abun- 

 dant at a sampling station that also had greater current 

 velocity, and this finding is consistent with our results. 

 Similarly, Thorrold (1992) observed that light traps that 

 were allowed to drift with water masses collected signifi- 



