Although being of the same age, experimental groups displayed a marked difference in body mass distribution (po0.01; Fig. 2). In SGS body mass ranged between 73 and 182 g (mean: 119.9576.41 g) whereas it ranged between 194 and 395 g (mean: 288.3714.4 g) in FGS. Mean condition factor (M L3 ) was 1.2970.17 in SGS and 1.62 70.21 in FGS (po0.01). Table 1 summarizes among-strains comparison of the various parameters measured during this experiment. Comparison of the time at which 50% of the population has been removed from the experimental arena (T50) showed that FGS was more tolerant to hypoxia than SGS (T50E260 versus E200 min respectively; Cox F-test: po0.01; Fig. 3a). Moreover, marked intra-strain variability in individual responses to HCT was observed (Fig. 3a). Time to loss of equilibrium indeed ranged between 180 and 410 min for FGS and between 130 and 280 min for SGS. This corresponded to incipient lethal oxygen saturation (ILOS) ranging from 13.4 to 16.7% air sat in FGS and from 14.7 to 18.9% air sat in SGS (Fig. 3c). Overall, SGS was found more tolerant to heat than FGS (T50E400 versusE270 min respectively; Cox F-test: po0.01; Fig. 3b). As for HCT, response to TCT displayed significant, within strain inter-individual variation (Fig. 3b). However, this variability was more marked for the FGS (160 min between the first and the last fish to lose equilibrium) than for the SGS (30 min). This corresponded to incipient upper thermal limit (IULT) ranging from 24.7 to 27.6 1C in FGS and from 28.5 to 29.7 1C in SGS (Fig. 3d). Although fish were allowed a one-week recovery period between consecutive challenges, the possibility of an interaction between performance during HCT and thermal tolerance (TCT) was examined and no significant correlation between ILOS and IULT was found (data not shown). In both strains, active metabolic rate was highly correlated with body mass (Fig. 4a; linear regression, po0.01). On the other hand, SMR was found to increase with body mass in the FGS (linear regression, po0.01) but not in the SGS (linear regression, po0.33). AMR increasing much faster with body mass than SMR, the metabolic scope (MS) increased significantly with mass. Over the whole size range, MS was increased nearly 7 times, 2.2 within SGS and 2.5 within the size range of FGS. Comparison of slopes showed no differences between strains in the slopes of AMR versus body mass and SMR versus body mass relationships (p40.05). Fitting a power model to the overall data set (Fig. 4a; dotted-hatched line) yielded a scaling exponent of 0.86 for SMR and 1.1 for AMR (Table 2). No significant, within strain relationship between Ucrit and body mass was found (Fig. 4b; p40.05). However, significance emerged when the two experimental strains were combined (p¼0.01). The mass of the ventricle, gills, liver and gut displayed significant positive relationships with body mass (Fig. 5; linear regression, po0.01) and slope analysis showed that there was no statistically significant difference among strains (p40.05). Fitting a power model to the data showed that organ-mass-to-bodymass ratios increased as fish got bigger. Mass exponents were quite comparable, ranging from 1.18 for the gills to 1.23 for the gut (Table 2). Analysis of residuals showed that neither organ-to-body-mass ratios, nor metabolic rates (SMR and AMR), nor swimming ability (Ucrit) correlated with performance during environmental tolerance tests (HCT and TCT; data not shown). Conversely, maximal oxygen consumption of permeabilized myofibers (cMO2) was found to be inversely related to heat tolerance (po0.01; Fig. 6). However, no correlation between cMO2 and hypoxia tolerance (ILOS) was found (data not shown).