Figure 2.

Swimming performance (U_crit) curves across generations in different stable environments (Experiment 1). The top row of panels shows means at different test temperautres (±s.e.; A,B) and interaction plots of marginal means (±s.e.) for significant interactions (C, D) for females, and the bottom row shows means (E,F) and interaction plots (G,H) for males. There were significant differences in mean U_crit between females (A,B) and males (E,F). In females, U_crit was determined by a three way interaction between generations (F1 = open circles, F2 = grey circles, F3 = black circles), developmental conditions (A: 23 °C development, B: 29 °C development), and test temperature. The interaction plot (E) shows that U_crit was significantly different between developmental treatments in the F1 generation but not in subsequent generations (open bars = 23 °C developmental temperature, filled bars = 29 °C developmental temperature). Additionally, fish developed at 29 °C had greater U_crit than those developed at 23 °C at all test temperatures (D). In males, mean U_crit (E,F) was determined by interactions between generation and developmental temperatures, and between developmental temperature and test temperatures. In males, U_crit differed between developmental treatments in the F2 generation but not in the F1 or F3 generations (G). U_crit was significantly higher in fish from the 23 °C developmental temperature treatment at 26 °C and 36 °C test temperatures (H; open triangles = 23 °C developmental temperature, filled triangles = 29 °C developmental temperature). N = 8–9 fish per treatment group; * p < 0.05, ** p < 0.01, *** p < 0.001.