Nelson: Fad characteristics and associated fish assemblages 



841 



the FADs, entered the water from the boat a minimum of 

 10 m from each FAD. 



The FADs were deployed from an inflatable boat at 50-m 

 intervals in a roughly linear array, and checked at hourly 

 intervals. The FADs did not maintain their initial spatial 

 arrangement, but I did not move any FAD once the drift 

 began unless FAD-to-FAD distance had been reduced to 

 less than 10 m. In this instance, I moved one or more FADs 

 to a minimum FAD-to-FAD distance of 50 m after checking 

 for any FAD-associated fishes. In none of these instances 

 were any FAD-associated fishes observed. I monitored the 

 drift for four hours; deteriorating weather and fading light, 

 however, did not permit additional observations. 



I used linear regression to test the hypothesis that the 

 number of FAD-associated fishes changed over time for 

 the enriched FADs and for the nonenriched FADs. I used 

 a ^-test to compare the slopes of the two regression models 

 and to test the hypothesis that the treatments accumulated 

 fish at different rates. 



Results 



Twenty-six species of fishes from 16 families were recorded, 

 including species associated with reef, soft bottom, and 

 coastal pelagic habitats as adults (Table 1). Only juvenile 

 specimens were observed clearly associated with FADs, 

 with the exception of Ahiterus scriptus and Lobotes paci- 

 ficus, of which both juvenile and adult forms were observed 

 in close, continuous proximity to the FADs. Two needlefish 

 species (Tylosaurus acus pacificus and T. crocodilus fodia- 

 tor) appeared occasionally in close proximity to the FADs, 

 but they were not clearly associated with the FADs. An 

 adult Lobotes pacificus (tripletail) was observed once and 

 a single adult Al uterus scriptus (scrawled filefish) were 

 observed on three separate instances. Horizontal under- 

 water visibility averaged 13.4 m (±1.7 SE) for all sampling 

 days combined. 



Juvenile sergeant major damselfish (Abudefduf trosche- 

 lii) were the dominant species by frequency of occurrence 

 and numerical abundance (Table 1) for all experiments. 

 The damselfish was followed in rank overall by juvenile 

 rainbow runner (Elagatis bipinnulata), although this spe- 

 cies was observed with the FADs only during the fouling 

 and model fish experiments. Juvenile threadfin {Polydac- 

 tylus approximans), mullet (Mugil sp.), and yellow snapper 

 (Lutjanus argentiuentris) were equally frequent but dif- 

 fered slightly in abundance (P approximans>Mugil sp.>L. 

 argentiventns; Table 1). The latter pattern was consistent 

 across all experiments. Specimens from a suite of juvenile 

 carangids (excluding E. bipinnulata) were also observed 

 frequently. 



Fish-removal experiments 



Sample date, series, and treatment combined to have a 

 significant effect on A. troschelii abundance (three way 

 interaction, P=0.03), but there was no clear pattern; the 

 remaining species (combined species less numbers of A. 

 troschelii) were influenced by sample date (date by series 



30 

 25 

 20 

 IS 

 10 



LU 

 U) 



A 



A Abudefduf troschelii only 



□ observed 



□ removed -j- 



1 



^ 





8-1 



0) 



1 r 



3-Jul 4-Jiil 5-Jul 6-Jul 7-Jul 8-]ul 9-Jul 10-|ul 



B combined spp. - A. trosclielii 



T 



MA 



i 



I 



3-Jul 4-Jiil 5-Jul 6-Jul 7-Jiil 8-JliI 9-Jul 10-]ul 



Figure 5 



Repeated fish-removal effects (disturbed vs. undisturbed 

 [observed only]) on aggregation size (mean no. of fishes) for 

 Abudefduf troschelii alone (A) and for all species combined 

 less A. troschelii (B). See Tables 3-5 for sample sizes and 

 ANOVA results for assemblage size and diversity measures. 



interaction, P<0.01) but not by treatment (P=0.73, Table 2, 

 Fig. 5). Measures of diversity varied between series (series: 

 S, P<0.01; HB, P=0. 01) but were unaffected by treatment. 

 Thus, fish removal or fish disturbance may contribute to 

 assemblage sizes for individual species (e.g. A. troschelii), 

 but, in the present study, the total number of combined 

 species was unaffected. 



FAD size 



Abudefduf troschelii was strongly affected by a combina- 

 tion of treatment and sample date (date by treatment 

 interaction, P=0.03, Table 3, Fig. 6). Results from the 

 remaining species combined were comparable with larger 

 total numbers at the larger FADs, although not statisti- 

 cally significant (treatment, P=0.07). Although both mea- 

 sures of diversity (S and HB) suggested that the treatment 

 may have had a positive effect on diversity (S, treatment, 

 P=0.02), species richness was positively correlated with 

 sample size. HB, a diversity measure comparatively unaf- 



