Fuiman and Ottey: Temperature effects on behavior of young Saaenops ocellatus 



25 



since transfer. Each behavioral measure (activity, pause 

 frequency, pause duration) was considered the 'trials' 

 factor in a separate analysis. Size-class and tempera- 

 ture treatment were grouping factors. Activity was re- 

 stricted, by definition, to values between and 300 s. 

 Data were expressed as a percentage of the total time- 

 period, and an angular (arcsine) transformation was 

 applied to satisfy the statistical requirement of 

 MAN OVA for normality (Snedecor & Cochran 1967). 

 Pause frequency and duration were not transformed. 



Further analyses focused on three questions: (1) How 

 does behavior differ at the two acclimation tempera- 

 tures? (2) What is the immediate effect of a transfer of 

 5°C on behavior (relative to fish maintained at accli- 

 mation temperatures)? and (3) When does behavior 

 stabilize after such a transfer? These questions were 

 addressed by repeated-measures MANOVA on pairs of 

 treatments within size-classes. Since temporal trends 

 in variables were of greater interest than the mere 

 presence of significant differences among the time-pe- 

 riods, we examined first (linear)- through fourth (quar- 

 tic)-order polynomial trends over time with univariate 

 F statistics (Wilkinson 1990), in addition to testing for 

 significant differences due to temperature treatments 

 within each time-period. 



By design, fish transferred upward or downward ex- 

 perienced a temperature change, but control fish did 

 not. However, behavior of all fish could be expected to 

 vary with time, due to effects of handling, hunger, or 

 circadian rhythms. Effects of handling should dimin- 

 ish with time since transfer, but hunger should in- 

 crease over time and have a stronger influence on be- 

 havior of smaller fish. Therefore, behavior of fish in 

 transfers was compared with that of control fish from 

 the same acclimation temperature to answer questions 

 (2) and (3). Specifically, upward transfers were com- 

 pared with the low controls, and downward transfers 

 with the high controls, to examine changes in behav- 

 ior relative to acclimation levels. Figures 2, 4, and 6 

 depict the effects of thermal transfers as differences 

 between means of six fish in the transfer and control 

 treatments, but statistical tests were based on results 

 for individual fish. 



Results 



Spontaneous behavior was composed of conspicuous 

 periods of swimming activity interspersed with pauses, 

 which were sometimes quite long. Behavior of medium 

 and large fish was grossly similar, their transitions 

 from active swimming to pausing were gradual, in part 

 because of passive coasting. At the larger sizes, swim- 

 ming involved forward movements generated by the 

 caudal and pectoral fins and complex maneuvers us- 



ing sculling motions of the pectoral fins alone. Move- 

 ments of small fish were less fluid. Their small size 

 and low velocities prevented them from appreciable 

 coasting, so their motion stopped abruptly when pro- 

 pulsive strokes of the caudal region ceased. 



Activity 



The proportion of time in active swimming was high 

 for all sizes, usually >65%. However, there were sig- 

 nificant differences in activity among the three size- 

 classes, among the four temperature treatments, and 

 within individuals over the course of the experiments. 

 When the four temperature treatments were exam- 

 ined separately, significant differences among size- 

 classes were found only for downward transfers and 

 high controls (Table 1). 



Activity varied significantly with time during ex- 

 periments within most temperature treatments for 

 small and medium fish (Table 1). Large fish showed no 

 significant changes in activity with time in any of the 

 treatments because variability among individuals was 

 greater than at smaller sizes. Differences in activity 

 over time constituted temporal trends for small fish in 

 all treatments and medium fish in downward trans- 

 fers. These trends were usually linear, but several 

 higher-order polynomials were significant for down- 

 ward transfers (Table 1). In subsequent comparisons 

 of the different temperature treatments, we examined 

 each size-class separately because of the highly sig- 

 nificant differences among sizes. 



Controls Activity was generally greater at 26° C for 

 all sizes offish (Fig. 1). This difference was significant 

 for small and medium fish, but the attained signifi- 

 cance, P, for large fish was slightly beyond the crite- 

 rion of a=0.05 (Table 1). 



Mean activity followed a monotonic trend with time 

 at both temperatures for all sizes (Fig. 1). Small fish 

 were most active immediately after transfer, while 

 medium fish were least active at that time. Activity 

 levels generally became stable within approximately 

 4h. Mean differences (±SD) between the two control 

 levels after 4h were 40.0 (±12.9), 49.1 (±26.6), and 

 45.7 (±41.5) s of swimming/5 min for small, medium, 

 and large fish, respectively. Thus, there was remark- 

 able similarity in the effect of acclimation tempera- 

 ture on mean activity, but variability increased 

 steadily with size. 



Significant differences between acclimation tempera- 

 tures were common during the first 2 h of the trials, 

 and rare thereafter (Fig. 1), suggesting that the effect 

 of handling on activity and the rate of recovery were 

 temperature-related. However, these effects of tempera- 

 ture were not consistent. Activity of small fish stabi- 



