200 



Fishery Bulletin 91(2). 1993 



the larger individuals of most families. Analysis of the 

 catch by method (Tables 1,2) suggests that the appar- 

 ent selectivity of the light-trap reflects size-specific 

 rather than taxonomic biases. The absence of certain 

 taxa from the light-trap during the sampling period 

 may mean that few large individuals were in the sam- 

 pling area. Table 3 shows that, with the exception of 

 bothids, schindleriids and carangids, taxa not caught 

 by the light-trap were represented by relatively small 

 individuals in the catch by other methods. Whether 

 large carangids were present in more than trivial num- 

 bers is unclear. A single 57.4 mm carangid was taken 

 by the light-seine, but the next-largest carangid taken 

 by other methods was 14.2mm. The question of selec- 

 tivity by light-traps must be resolved by more compre- 

 hensive sampling. 



The light-seine and Tucker trawls captured most of 

 the abundant families in all sizes. The neuston net 

 and purse-seine captured the same abundant taxa, with 

 size-ranges similar to one another. The exceptions were 

 mullids, microdesmids, gobiids, and atherinids, for 

 which the neuston net captured larger individuals. For 

 the mullids and microdesmids, size distributions pro- 

 duced by the two methods overlapped slightly. 



Catches by all methods were dominated by a few 

 abundant families of fishes. The first five most- 



abundant families listed in Table 2 accounted for 80% 

 or more of the catch by all methods. The Tucker trawl 

 was the most equitable in terms of abundance distri- 

 butions, and the light-trap the least. However, the rank 

 order of abundant families was not the same for all 

 methods (Fig. 2). The dominant families for all towed 

 nets and the purse-seine were gobiids and apogonids. 

 For light-trap and light-seine the dominant families 

 were pomacentrids, followed by gobiids. Small apo- 

 gonids, although consistently abundant in net samples, 

 were not captured by light-aggregation devices. In light- 

 trap catches, a single family — the Pomacentridae — 

 accounted for 93% of individuals collected. 



For most collecting methods, there was a high de- 

 gree of consistency among samples. Results of PCA 

 (Fig. 3) showed that samples taken by light-trap were 



0) 



o 



c 



D 

 X) 



c 

 < 



o 



Q. 



c 



D 



Bongo Net 

 n=43417 



4 ■*! 



Purse Seine 

 n = 812 



■6k 



.4 



.2 



ttCb 



I 2 3 4 5 6 7 8 9 10 



I 2 15 4 10 8 3 6 7 16 



Neuston Net 

 n=2418 



Light Seine 

 n=2707 



Etk 



I 2 II 3 4 7 6 5 10 



10 12114178673 



Tucker Trawl 

 n = 723 



rm^ 



2 12 10 4 II 9 13 5 14 







Light Trap 

 n=7624 



10 I 17 8 18 6 19 II 20 21 



Family 

 Figure 2 



Mean proportional abundance i±l SE, vertical axis, shown only 

 upward) and ranked taxonomic categories of fishes (clupeoids 

 excluded) collected by six sampling methods off Lizard Island, 

 Great Barrier Reef on 3, 5. and 6 December 1986. Other sample 

 data are given in Table 1. Key to taxa: 1 Gobiidae, 

 2 Apogonidae, 3 Pinguepididae, 4 Lutjanidae. 5 Carangidae, 

 6 Nemipteridae. 7 Callionymidae. 8 Labridae, 9 Monocanthidae, 

 10 Pomacentridae, 11 Atherinidae, 12 Schindieriidae, 

 13 Ephippididae, 14 Bothidae, 15 Scaridae, 16 Microdesmidae, 

 17 Mullidae, 18 Lethrinidae, 19 Synodontidae, 20 Scnmbridae, 

 21 Blenniidae. 



