342 
Fishery Bulletin 117(4) 
by using analysis of variance with the Tukey’s honestly 
significant difference (HSD) test (Table 1). The Kn was cal¬ 
culated with the following equation: 
Kn = 
( 2 ) 
where W = weight (in grams) and 
W = the fish weight (in grams) predicted from the 
least squares regression of log W on the log of 
the fish TL. 
Measures of SMR, MMR, AMS, and exhaustion time 
were log transformed to meet assumptions of normality of 
residuals and homogeneity of variance, which were veri¬ 
fied by using Shapiro-Wilk and Levene tests, respectively. 
Separate linear models that use standard least squares 
were used to test the effects of treatment temperature, 
DO concentration, weight, Kn, and acclimation time of 
fish on SMR, MMR, AMS, and exhaustion time. A similar 
set of linear models were also generated with the P0 2 in 
place of DO concentration; this second set of models was 
used to examine the effect of uneven P0 2 across combina¬ 
tions of temperature and dissolved oxygen on the metab¬ 
olism and exhaustion time. This entire analysis was 
repeated except the SMR, MMR, and AMS values were 
not adjusted for weight (i.e., values were not divided by 
fish weight; therefore, they were absolute, or raw, meta¬ 
bolic rates in milligrams per hour). This second round of 
analysis with unadjusted values was done to understand 
differences in the effects of temperature or DO concen¬ 
tration due to estimating SMR, MMR, and AMS as ratios 
(with weight in the denominator) rather than using raw 
metabolic rates. For all analyses, Tukey’s HSD tests were 
used where appropriate to assess differences among 
treatments. For all statistical tests, the significance level 
or alpha level was set at 0.05. 
