Countergradient selection has the potential to lead to a
cryptic form of reproductive isolation between populations
evolving in different environments (see also Carroll et al.
2001; Craig and Foote 2001). This differs from the usual
scenario of local adaptation reducing gene flow between populations
in that, with countergradient selection, the optimum
phenotype may be the same in the two environments. Consider,
for example, that a low-drosopterin male guppy who
dispersed downstream over a barrier waterfall as a juvenile,
from a low-carotenoid-availability site to a high-carotenoidavailability
site, would develop orange spots with an abnormally
high carotenoid:drosopterin ratio. Upstream male migrants,
on the other hand, would develop abnormally low
carotenoid:drosopterin ratios. If female guppies indeed prefer
males with orange spots of the normal hue, male interpopulation
migrants would suffer a mating disadvantage, and so
would their hybrid male offspring. This form of incipient
reproductive isolation could not be detected, however, with
standard laboratory mate choice tests. To be valid, mate
choice tests would have to be carried out using true interpopulation
migrants or fish raised on diets designed to precisely
match the carotenoid availability that a migrant would
encounter.
This species, however, may not be a particularly good candidate
for reproductive isolation through countergradient selection.
Guppy populations at opposite extremes of the carotenoid
availability gradient are usually separated from each
other by populations with intermediate levels of carotenoid
availability (G. F. Grether, unpubl. data). Thus, sexual selection
against migrant phenotypes should be relatively weak.
The effects of countergradient selection on reproductive isolation
are likely to be strongest in situations in which the
environmental gradient is steep.