Larvae of an estuarine grapsid crab Chasmagnathus granulata Dana 1851, from temperate and subtropical regions of South America, were reared in seawater (32 parts per thousand) at five different constant temperatures (12, 15, 18, 21, 24 degrees C). Complete larval development from hatching (Zoea I) to metamorphosis (Crab I) occurred in a range from 15 to 24 degrees C. Highest survival (60% to the first juvenile stage) was observed at 18 degrees C, while all larvae reared at 12 degrees C died before metamorphosis. The duration of development (D) decreased with increasing temperature (T). This relationship is described for all larval stages as a power function (linear regressions after logarithmic transformation of both D and T). The temperature-dependence of the instantaneous developmental rate (D-1) is compared among larval stages and temperatures using the Q(10) coefficient (van't Hoff's equation). Through all four zoeal stages, this index tends to increase during development and to decrease with increasing T (comparing ranges 12-18, 15-21, 18-24 degrees C). In the Megalopa, low Q(10) values were found in the range from 15 to 24 degrees C. In another series of experiments, larvae were reared at constant 18 degrees C, and their dry weight (W) and respiratory response to changes in T were measured in all successive stages during the intermoult period (stage C) of the moulting cycle. Both individual and weight-specific respiration (R, QO(2)) increased exponentially with increasing T. At each temperature, R increased significantly during growth and development through successive larval stages. No significantly different QO(2) values were found in the first three zoeal stages, while a significant decrease with increasing W occurred in the Zoea IV and Megalopa. As in the temperature-dependence of D, the respiratory response to changes in temperature (Q(10)) depends on both the temperature range and the developmental stage, however with different patterns. In the zoeal stages, the respiratory Q(10) was minimum (1.7-2.2) at low temperatures (12-18 degrees C), but maximum (2.2-3.0) at 18-24 degrees C. The Megalopa, in contrast, showed a stronger metabolic response in the lower than in the upper temperature range (Q(10) = 2.8 and 1.7, respectively). We interpret this pattern as an adaptation to a sequence of temperature conditions that should typically be encountered by C. granulata larvae during their ontogenetic migrations: hatching in and subsequent export from shallow estuarine lagoons, zoeal development in coastal marine waters, which are on average cooler, return in the Megalopa stage to warm lagoons. We thus propose that high metabolic sensitivity to changes in temperature may serve as a signal stimulating larval migration, so that the zoeae should tend to leave warm estuaries and lagoons, whereas the Megalopa should avoid remaining in the cooler marine waters and initiate its migration towards shallow coastal lagoons.