We assayed two components of performance (development time and survivo
rship), on food medium with and without ethanol, in laboratory populat
ions of Drosophila simulans and D. melanogaster for which ethanol-medi
um was a novel food resource. These assays were done before and after
12 generations of rearing in either one-(regular medium only) or two-r
esource (regular medium and ethanol medium) environments. Initially, D
. simulans was highly susceptible to ethanol, whereas D. melanogaster
was relatively unaffected. After 12 generations in the two-resource en
vironment, D. simulans showed significantly improved mean performance
on ethanol medium; mean performance of D. melanogaster did not signifi
cantly change. Variation among families for both traits was higher on
ethanol medium in D. simulans. Variation in D. melangaster was not sig
nificantly affected by ethanol level, suggesting that resource quality
was more important than novelty per se. In both species, the least va
riation was seen in populations after 12 generations in the two-resour
ce environment. For development time in D. simulans, the decrease in v
ariation was largely due to reduced variation within families, suggest
ing the evolution of canalization. Development time on the two media w
as not negatively correlated. In D. simulans, correlations measured be
fore and after the experiment were not heterogeneous, suggesting that
trade-offs in performance did not block diet expansion. In D. melangas
ter, correlations became significantly less positive after 12 generati
ons in the two-resource environment, supporting the view that correlat
ions between performance on different resources may become less positi
ve over time through selection.