Density-Dependent Natural Selection and Trade-Offs in Life History Traits

Theories of density-dependent natural selection state that at extreme population densities evolution produces alternative life histories due to trade-offs. The trade-offs are presumed to arise because those genotypes with highest fitness at high population densities will not also have high fitness at low density and vice-versa. These predictions were tested by taking samples from six populations of Drosophila melanogaster kept at low population densities (r-populations) for nearly 200 generations and placing them in crowded cultures (K-populations). After 25 generations in the crowded cultures, the derived K-populations showed growth rate and productivity that at high densities were elevated relative to the controls, but at low density were depressed. The role of cytoplasmic domains in the determination of the fates of ectodermal and mesodermal cells has been investigated in leech embryos. When yolk-deficient cyto-plasm (teloplasm) was extruded from the animal pole of the zygote, the ectodermal precursor blastomere was converted to a mesodermal fate. This change of fate can be prevented by replacement of the extruded animal teloplasm with teloplasm from the vegetal pole. The fate of the mesodermal precursor blastomere was unaffected by teloplasm extrusion or rearrangement. These results demonstrate that ectodermal and mesodermal determination of fate involves a binary decision dependent on the position of teloplasm along the animal-vegetal axis.


Density-Dependent Natural Selection and
Trade-Offs in Life History Traits LAURENCE D. MUELLER, PINGZHONG Guo,* FRANcisco J. AYALA Theories of density-dependent natural selection state that at extreme population densities evolution produces alternative life histories due to trade-offs. The trade-offs are presumed to arise because those genotypes with highest fitness at high population densities will not also have high fitness at low density and vice-versa. These predictions were tested by taking samples from six populations of Drosophila melanogaster kept at low population densities (r-populations) for nearly 200 generations and placing them in crowded cultures (K-populations). After 25 generations in the crowded cultures, the derived K-populations showed growth rate and productivity that at high densities were elevated relative to the controls, but at low density were depressed. O NE OF THE FIRST SUCCESSFUL combinations of theory from ecology and evolution was the theory of density-dependent natural selection, often called rand K-selection, where r and K refer to low-and high-density conditions, respectively (1). The initial models combined theoretical models of population growth dynamics with single-locus population genetic models in order to describe evolution in environments that differ with respect to population density (2); more complex elaborations of these early models were advanced later (3). These theories and models have assumed a central role in the theory of evolutionary ecology (4). A crucial assumption of these theories has been that genotypes selected for sustained reproduction and survival at high population densities are not likely to do as well at low densities; likewise organisms capable of rapid reproduction under low levels of crowding may not reproduce as well under crowded conditions. It has been, however, difficult to demonstrate empirically the postulated trade-offs (5).
We have previously shown (6) that three populations of Drosophila melanogaster kept at low density (r-populations) had, after eight generations, higher rates of population growth when tested at low densities than three populations kept at high density (Kpopulations), whereas the opposite was the case for growth rates tested at high population densities. The interpretation and significance of these results were complicated by three issues. (i) Given that both the lowand high-density environments were novel for these populations, it remains possible that the differences observed between the rand K-populations were not due to changes in both populations but rather that only one population had evolved whereas the other retained the attributes of the founder population. (ii) The differences observed between r-and K-populations with respect to growth rates and total productivity at low density were only marginally significant. (iii) No additional tests have been carried out to verify these results; indeed one study of mosquitoes from natural populations did not show any trade-off in population growth rates (7).
We describe an experiment designed to overcome these problems, the results of which confirm that trade-offs do occur in the evolution by density-dependent natural selection. We test two types of high density populations (rKand rx rK-populations), both derived from r-populations now transferred to the K-environment; the controls are two types of low density populations (rand rxr-populations) (Fig. 1). The rxrpopulations were created to introduce genetic variation into replicate sets of low density populations. The rK-populations were created from samples of each r-population and had been maintained at high densities for 25 generations prior to this experiment. In a similar fashion the rx rKpopulations are samples from the rx r-populations that have been kept at high densities.
Each of the four types of populations consisted of three replicates. During the course of this experiment the rand r x rpopulations are not expected to change significantly, given that they had been previously kept for nearly 200 generations at the same low density as now. However, the rKand rxrK-populations have been transferred from the low density r-environment to the high density K-environment. Hence, differences that arise between them and the r and rx r controls may be attributed to adaptation to the new high-density environments.
After the three rKand three rxrK-populations had undergone 25 generations of natural selection in their new environments, we measured rates of population growth and net productivity in each of these six populations and the six controls, at three different densities (10, 750, and 1000 adults) following the methods described in (6). We determined the rates of population growth (6) and the net productivity (8), two differ-ent measures of population fitness that can be estimated from the same experimental data (Table 1 and Figs. 2 and 3). Rate of population growth is the measure of fitness most closely tied to the theories of density- Adult density adults and progeny produced. Hence, both these quantities are used in these estimates of net productivity although other investigators have concentrated on just progeny production (8). For each population, six replicate experiments were conducted at density 10, and three at each of the two higher densities. The graph shows the difference in net productivity between the mean of the three K-populations (either rK or rx rK) and the three controls (r or rx r). Only the plus or minus portion of the 95% confidence interval is shown when one of the bars would extend beyond the graph. The null hypothesis is that the K-populations do not have lower productivity than the controls in the low-density tests and higher in the high-density tests. The tests reject this null hypothesis, thereby supporting our working hypothesis that trade-offs in life history evolve by natural selection in response to population density. The asterisk indicates that the difference is statistically significant according to the Mann-Whitney U test (P = 0.05). The ANOVA gave significant population-by-density interactions for r versus rK (P = 0.002), and for rxr versus rxrK (P < 0.001). The difference in the ordinate is in numbers of individuals. dependent natural selection, but it cannot be estimated as accurately as can net productivity. Trade-offs should produce a significant interaction between density and population in an analysis of variance (ANOVA). With the sample sizes used in this study the power of the typical ANOVA is quite low, so we have also tested the differences in growth rate and productivity by the Mann-VVhitney U test.
The differences between the rx rK and the rxr-populations in net productivity are statistically significant at all densities (Fig. 2). The expected trade-offs are evident. At low density (ten adults per culture) the net production of progeny is greater in the rx rpopulations than in the rxrK-populations, whereas the reverse occurs at the two high densities. Similar trade-offs in the net productivity are apparent between the rand rK-populations, but because of high withintreatment variability only the difference at the highest density is statistically significant.
The expected trade-offs also occur with respect to population growth rate (Fig. 3). The differences between the. rx rK-and r x rpopulations are each statistically significant by at least one test. The rK-populations show a significantly higher average growth rate than their r controls at the highest density, but no significant differences occur at the two other densities (Fig. 3).
The growth rate data of the r-and rK-  Fig. 2. Growth rate takes into account the time when progeny are produced (6) in addition to just the number of survivors and offspring (which are the components of net productivity). Thus, progeny that emerge during the second week of the experiment will have a greater impact on the growth rate than progeny that emerge during the fourth week. For the computation of net productivity the progeny from the second and fourth weeks have equal effects. The symbols indicate statistical significance as described for Fig. 2. The ANOVA, on log-transformed data, indicates no significant population by density interaction for r versus rK but a significant interaction for rXr versus rXrK (P = 0.003). These experiments, like the tests described in Fig. 2, corroborate the evolution of life historyv trade-offs in response to population density. The 95% confidence interval is shown for differences in growth rate.