866 



Fishery Bulletin 98(4) 



ducted immediately after introduction and once every 4 h 

 over the following 24 h. For 10 mm, 15 mm, 21 mm, and 28 

 mm fish, observations were also made 36 h after introduc- 

 tion. During each observation period, the number offish in 

 each section of the tank was counted. At night time (20:00, 

 00:00 and 04:00 h), infrared illumination aided observa- 

 tion. At each observation, counts were repeated five times 

 for each tank at approximately 5-min intervals. The aver- 

 age number of fish in each section with a flotsam was 

 calculated and then divided by the total number of fish 

 (n=10). This value, defined as the association ratio, would 

 be 1.0 if fish associated strongly to the object and would be 

 0.25 with random movement or distribution. In the control 

 treatment, the number offish in the southeast, southwest, 

 or northwest quadrants was counted from the triplicated 

 tanks, respectively. The average value of these three tanks 

 was used as a control. Total lengths of all the fish were 

 measured after the experiment. 



Association ratios were compared for each type of flot- 

 sam for each time of observation. Because association 

 ratios deviated from normal distribution, we applied arc- 

 sine transformation (p'=arcsin\^; Zar, 1996). The resultant 

 data showed nearly normal distribution. One-way ANOVA 

 was then applied followed by Scheffe's test as a post hoc- 

 test (Zar, 1996; Abacus Concepts, Inc., 1992). We consid- 

 ered an association ratio higher than the control (P<0.05, 

 Scheffe's test) to be evidence of association behavior with 

 that flotsam. 



Results 



In 5.5-mm, 8.8-mm, and 10-mm fish, no association behav- 

 ior was observed with any types of flotsam, and the average 

 association ratios were always near 0.25 ( Fig. 2, A-C ). At 12 

 mm, the first significant association behavior was observed 

 both with transparent and gray objects (Fig. 2D; P<0.05). 

 Association ratios did not differ significantly between these 

 two object types at this stage (P>0.2). At 15 mm, associa- 

 tion behavior was not obvious, perhaps because the fish 

 swam around in the tank restlessly in this particular trial, 

 both in test and control groups. At 20 mm, association 

 was again obvious and the association ratio to the gray 

 pipe tended to be higher than that to the transparent pipe 

 (P=0.013 at 20:00 h on the second day. Fig. 2F). Association 

 behavior was especially strong at night, and some individ- 

 uals almost touched the transparent and gi'ay pipes (Fig. 

 3). In most cases it took at least 12 h for fish to show an 

 association behavior in the experimental tank, but 21-mm 

 fish showed an association with gray objects immediately 

 after they were introduced to the tank. At 28 mm, an asso- 

 ciation with transparent and gray pipes was positive, but 

 the association ratio tended to be slightly less than fish at 

 20 mm and 21 mm. No association behavior with shadows 

 was observed at any life-history stage of the fish. 



Association ratios after 12 h and 24 h in the experi- 

 ment (12:00 h and 20:00 h on the second day) were plot- 

 ted against the total length of each fish (Fig. 4i, showing 

 that striped jack begin association behavior at 12 mm in 

 TL with both transparent and gray objects. 



Discussion 



Accurate observation of association behavior can suffer 

 from biases attributed to other behavior and taxis. For 

 example, striped jack show strong phototaxis from 3.5 

 mm TL and rheotaxis from 4.5 mm TL (Masuda and Tsu- 

 kamoto, 1996). In the present study these potential con- 

 founding factors were prevented by illuminating the tanks 

 strongly from above and by using tanks without currents. 

 This procedure was apparently successful because most 

 fish smaller than 10 mm distributed themselves about 

 equally in the experimental tanks (Fig. 2, A-C). Striped 

 jack also exhibit schooling behavior from about 16 mm 

 (Masuda and Tsukamoto, 1998b) which may have con- 

 tributed to the fluctuating association ratio values with 

 15-mm or larger fish in the control tanks (Fig. 2, E and F). 



Small tank size (30 liters) used in this experiment may 

 have affected the results because the fish may have con- 

 sidered a tank wall as a structure with which to associate. 

 However, fish apparently preferred the pipe to the tank 

 wall and associated closely with it (Fig. 3). Strong aeration 

 outside tanks in the water bath might have minimized 

 their association to the tank wall. The limited swimming 

 ability and shortsightedness of the larvae and juveniles 

 also helped to reduce the effect of the tank wall. 



Our results indicate that striped jack first started to 

 show association behavior at 12 mm TL. Hunter and 

 Mitchell (1967) reported six species of carangid fish juve- 

 niles associated with flotsam offshore Costa Rica, and their 

 smallest sizes (given as standard lengths) were as follows: 

 Caranx caballus: 9 mm; C. hippos: 16 mm; Decapterus sp.: 

 17 mm; Elagatis bipinnulatus: 11 mm; Selar crumenoph- 

 thalmus: 15 mm; and Seriola sp.: 10 mm. From observa- 

 tion of hatchery tanks, Tachihara et al. (1993) reported 

 that artificially reared gi'eater amberjack, Senola dumer- 

 ili, larger than 11 mm standard length associate with 

 objects. The smallest size of an flotsam-associating indi- 

 vidual in natural waters may differ depending on species, 

 area, season, and sampling methods. The smallest size ( 12 

 mm in TL) to associate with flotsam in our study was 

 therefore consistent with field and hatchery observations 

 of other carangid species. 



We hypothesized that mechanical and visual stimuli 

 may be cues for association behavior No association with 

 shadow flotsam was observed at any fish stages in our 

 study, suggesting that shade is too weak a stimulus to 

 evoke association behavior At 12 mm, association behav- 

 ior was obsei-ved to both transparent and gray flotsam, 

 and in fish larger than 20 mm, association to gray flotsam 

 was stronger This finding suggests that in early juvenile 

 stages, a mechanosensory system, as well as vision, may 

 be involved in association behavior Striped jack showed 

 stronger association behavior in darkness than in light — a 

 finding that seems to conflict with Gooding and Magnuson 

 (1967) who found that residents of an experimental float- 

 ing raft accumulate more rapidly by day than by night. 

 In nature striped jack may find flotsam by vision during 

 the daytime and mauitam an association with it by a 

 mechanosensory system during night. Floating structures 

 may sei-ve as feeding places, shelters from predation, and 



