Nelson: Fad characteristics and associated fish assemblages 



839 



undisturbed) by 2 (first series vs. second series) by 6 

 (sample date) model and I used a repeated measures 

 ANOVA (repeated on sample date) on the follow- 

 ing dependent variables: assemblage size (total no. 

 of fishes), species richness (S), and HB. I repeated 

 analyses of assemblage-size effects looking at the 

 number of A. troschelii only, and the total number of 

 fishes minus the number of A. troschelii. 



FAD size 



To determine the effect of FAD size on the associ- 

 ated assemblage size and diversity, I compared 

 FAD-associated fish assemblages between triple- 

 size FADs and single FADs. An existing anchored 

 array of eight FADs (two lines parallel to the coast 

 of four FADs each. Fig. 3) was cleared of fishes on 24 

 July 1997. As the fish were removed from the FADs, 

 each FAD was replaced with a fresh (i.e. clean and 

 unfouled) single or triple-size FAD, placed at alter- 

 nating positions. The single FADs were constructed 

 as described above and in Figure 2; the triple-size 

 FADs were identical to the single FADs, except 

 that they consisted of nine, rather than three, 

 purse-seine buoys lashed together and had the 

 effect of nearly tripling the wetted surface area 

 (although inner buoys are less exposed than outer 

 ones) and the volume of the FAD, and of increasing 

 the maximum linear dimension of the FAD by a factor of 

 two. Treatments were not reversed for this or subsequent 

 experiments because sample date appeared to be the major 

 factor determining assemblage size for any species, based 

 on the previous experiment. Note that in each of these 

 experiments, except for the recruit-enrichment experi- 

 ment that used drifting FADs, treatments were assigned 

 uniformly throughout the FAD arrays so that onshore, 

 offshore, or longshore biases in recruitment due to oceano- 

 graphic processes would not confound the results. Fishes 

 at all FADs were counted and identified on three dates (26, 

 28, and 30 July 1997), each observation separated from the 

 next by 48 hours. No fish were collected, with the exception 

 of one balistid, taken from the array on 26 July because 

 it was the first of that species to be observed associated 

 with a FAD. Data were analyzed for experimental effects 

 on total assemblage size, species diversity (S and HB), the 

 number of A. troschelii, and total number of fishes minus 

 the number of A. troschelii. 



Presence of absence of a fouling community 



To determine whether the presence of a fouling community 

 on a floating object affected the associated fish assemblage, 

 I compared FAD-associated assemblage sizes and species 

 richness between fouled and unfouled (control) FADs. Con- 

 trol FADs were scrubbed of all fouling organisms. Fouled 

 FADs had been deployed for a minimum of 14 days (range: 

 14-22 days) in the study area, and had accumulated fouling 

 that completely covered the wetted surface of the FAD with 

 gooseneck barnacles (Lepas sp.), hydroids, and bryozoans. 

 Grapsid crabs and polychaete worms (Amphimone vagans) 



Fish Aggregation Device 

 (FAD) 



steel hoop to 

 distribute model \ 

 aggregation 



to line weight and anchor 



Figure 4 



Treatments for the artificial fish experiment involved suspending 

 artificial fish, lead weights of an equivalent mass and volume, or 

 nothing (control) from a steel hoop lashed beneath the FAD. 



were also intermittent associates of fouled FADs. Control 

 and fouled FADs were deployed on 8 September 1997 in an 

 alternating array of eight buoys, with four FADs per treat- 

 ment (layout and spatial distribution of treatments follow 

 that of the FAD size experiment). All fishes were cleared 

 from FAD positions prior to deploying the FADs, and data 

 collection commenced 24 hours later Data were collected 

 on four consecutive days (9, 10, 11, and 12 September 1997) 

 and analyzed for experimental effects on total assemblage 

 size, species diversity (S and HB), the number of A. tros- 

 chelii, the number of E. bipinnulata, and total number of 

 fishes minus the number of A. troschelii. 



Artificial fish experiment 



I tested the hypothesis that potential recruits would dis- 

 tinguish between FADs with an "assemblage" of artificial 

 fish suspended beneath them, FADs with an "assemblage" 

 of suspended material equal to the artificial fish in size 

 but not resembling fish in appearance, and control FADs 

 without anything suspended beneath them (Fig. 4). I con- 

 structed artificial fish from 31.25-g lead fishing weights. 

 These weights were flattened, tear-drop-shaped objects, 

 painted a dull yellow with black bars to resemble juvenile 

 Abudefduf troschelii and suspended, by using monofila- 

 ment (20 lb. test) and a steel hoop, beneath the "artificial 

 fish FADs" (Fig. 4). I suspended oblong 31.25-g lead fish- 

 ing weights beneath "weighted FADs," and the control 

 FADs had only a steel hoop beneath each (Fig. 4). These 

 FADs were deployed in an anchored array, and the vari- 

 ous treatments were distributed in an alternating pattern 

 throughout the array. FAD positions were cleared of fishes, 



