Genetic differences among populations of Atlantic silverside (Menidia menidia) are hypothesized to be evolutionary responses to intense, size-selective winter mortality at high latitudes. Three experiments were conducted to test features of winter mortality. In the first experiment, we varied size and whether food was provided or withheld; temperatures were permitted to follow ambient (New York) wintertime fluctuations. Mortality and depletion of energy reserves were more rapid in the units receiving no food. Small fish died before larger fish in these units, but not in the units receiving food. Energy depletion of fish in the no-food treatment resembled that of fish in the wild. In the second experiment, we varied size and population of origin, representing high-latitude (Nova Scotia), midlatitude (New York), and low-latitude (South Carolina) populations. These fish were provided food and showed minimal depletion of energy reserves, but mortality rates were high when water temperatures were low. Mortality did not vary with size in New York and South Carolina fish, but was highest in intermediate-size fish from Nova Scotia. There was a pronounced population difference in survival rate (Nova Scotia > New York > South Carolina). In the third experiment, food was withheld and extreme low temperatures were moderated. Energy depletion was rapid and small fish died before large fish. We conclude that populations in seasonal environments are likely to be subject to size-selective winter mortality when energy reserves are depleted and that juvenile growth rates have evolved in response to this selection pressure. In addition, high-latitude populations have evolved greater tolerance to other winter stresses associated with low temperatures.

Field observations of a variety of fish species at different latitudes suggest that there is an inverse (countergradient) relationship between growth rate within the first growing season and the length of the growing season. In this study, larvae of striped bass Morone saxatilis from four different latitudes-Hudson River, New York (NY) (41 degrees N); Nanticoke and Patuxent rivers, Maryland (MD) (38 degrees N); Santee-Cooper river system, South Carolina (SC) (33 degrees N); Apalachicola River, Florida (FL) (30 degrees N)-were raised in captivity under identical conditions to determine if differences in growth could be detected among the populations. Larvae from two mothers per latitude were reared at two temperatures (17 degrees C and 21 degrees C) with replication. The general result was that MD and NY larvae exhibited the highest growth rates and SC larvae grew the slowest. At 17 degrees C, a significant difference in standard length growth rate was detected between MD larvae, the fastest growing population, and SC larvae, the slowest growing population. At 21 degrees C, the rank order for growth in standard length was MD = NY > FL = SC. No significant differences in dry weight growth were found at 17 degrees C. At 21 degrees C, a significant difference in dry weight growth rate was detected between NY larvae-the fastest growing population-and SC larvae-the slowest growing population. Significant differences among the populations also were found in egg weight and egg oil volume. However, these maternal effects did not exert a statistically significant influence on growth rate. Results indicate the existence of a countergradient relationship between larval growth rate and the length of the growing season among striped bass from different latitudes.

Common garden experiments were undertaken to test the hypothesis that the genetic capacity for growth of striped bass (Morone saxatilis) varies inversely with length of the growing season across a latitudinal gradient (i.e., countergradient variation (CnGV) in growth). Newly hatched larvae were obtained from six native anadromous stocks spanning most of the natural range of striped bass (Florida to Nova Scotia). Growth experiments were conducted under conditions of unlimited food at three temperatures (17, 21, 28°C) and commenced after larvae had exhausted maternal energy resources (i.e., yolk and oil) and had begun to metamorphose. Mixed-model nested ANOVA demonstrated that length and dry weight growth differed significantly among latitudes in most comparisons. South Carolina fish ranked consistently as the slowest growing group in virtually all comparisons, but the rank order of the other latitudes differed among trials. North Carolina and Gulf of Mexico fish generally had intermediate rates of growth whereas New York, Maryland, and Nova Scotia fish generally had the highest growth. Overall, the average growth rate of progeny from the 28 mothers tested in this study had a strong positive correlation with latitude of origin, strongly indicating CnGV in growth.

What happens when a population with environmental sex determination (ESD) experiences a change to an extreme environment that causes a highly unbalanced sex ratio? Theory predicts that frequency-dependent selection would increase the proportion of the minority sex and decrease the level of ESD in subsequent generations. We empirically modeled this process by maintaining five laboratory populations of a fish with temperature-dependent sex determination (the Atlantic silverside, Menidia menidia) in extreme constant temperature environments that caused highly skewed sex ratios to occur initially. Increases in the minority sex consistently occurred from one generation to the next across all five populations, first establishing and then maintaining a balanced sex ratio until termination of the experiment at 8 to 10 generations. The extent to which the level of ESD changed as balanced sex ratios evolved, however, was not consistent. Two populations that experienced high temperatures each generation displayed a loss of ESD, and in one of these ESD was virtually eliminated. This suggests that temperature-insensitive, sex-determining genes were being selected. In populations maintained in low temperature environments, however, the level of ESD did not decline. Instead, the response of sex ratio to temperature was adjusted upward or downward, perhaps by selection of sex-determining genes sensitive to higher (or lower) temperatures. The two different outcomes at low versus high temperatures occurred independent of the geographic origin of the founding population. Our results demonstrate that ESD is capable of evolving in response to selection.