Evolvability and nonevolvability of allometric slopes.

The science of morphological allometry centers around relationships between the size of organismal structures and overall body size. Morphological allometry usually is concerned with two parameters, the slope of log–log plots of trait parameters on body mass or size (reflecting how trait size varies with body size), and the intercept of such plots (providing measures of body-size–independent trait parameters). Morphological allometry is a central area of study for organismal and evolutionary biologists, because body size explains a large proportion of the variance in size of most body structures, and because we still lack a broad mechanistic understanding of the processes that produce allometric slopes. One of the debated themes of allometric research has been whether allometric slopes are determined by natural selection or represent constraints on biological variation. In PNAS, Bolstad et al. (1) address this question directly by testing whether the allometric slope of wing vein length on wing area of Drosophila melanogaster could be changed by artificial selection (yes, it was changed), and also show that after selection on allometric slopes, but not their intercepts, populations returned rapidly to control values when artificial selection ceased (Fig. 1). Bolstad et al. conclude that the rapid return of allometric slope to control values provides evidence for a pleiotropic constraint on the slope; that is, the genetic, developmental, and functional bases for different traits are not independent, so that selection on one allometric relationship affects others and overall fitness. They state “. . . this hypothesis could provide a general explanation for the striking evolutionary conservatism of allometric power laws, while still leaving open the possibility for allometries to be optimized by natural selection over long time scales through compensatory mutations” (1). Allometry can be studied during ontogeny, within populations for individuals at a specific developmental stage (static allometry) and across populations or species (evolutionary allometry). Understanding the relationships across these different scales is critical for development of unifying theories that integrate between proximate mechanisms and our understanding of evolutionary allometries because, to some extent, allometries at higher levels of organization depend on allometries at the lower level (2, 3). Bolstad et al. (1) first compared the interspecific and within-species allometric relationships for 111 Drosophilid species. Across and within species, as wings become larger, the L2 wing vein intersects the leading edge of the wing more distally; that is, log–log plots of the L2 vein length on the square root of wing area have slopes >1. For reasons that are unclear, this trend (and the slope of the log–log plot) is greater for interspecific comparisons than within populations. Evolutionary changes in the allometric slope were less than 2.8% per million y, showing that although the slope can evolve, there is strong conservation of these relationships. Artificial selection for relatively long or short wing veins at a given body size was able to change the intercept of the static allometric lines by 6–9% in only seven generations. Importantly, these changes were relatively stable after selection stopped, with only a 15% reversion toward control values over 16 generations, suggesting that wing shape was not under strong selection in these cultures. Artificial selection on the allometric slope was also successful, with “up-selection” increasing the slope to a value nearly 30% above control values, and “down-selection” decreasing the slope to a value about 75% of the control slope. These selected populations had allometric slopes similar to the most extreme Drosophilid species. However, unlike for the populations that had been