M. Stanton et Hj. Young, SELECTING FOR FLORAL CHARACTER ASSOCIATIONS IN WILD RADISH, RAPHANUS-SATIVUS L, Journal of evolutionary biology, 7(3), 1994, pp. 271-285
ESS models of reproductive allocation have been used extensively to ex
plain patterns of floral diversity in angiosperms. These theoretical e
xplorations assume that proportional allocation to pollen, ovules, and
seeds, as well as to secondary features such as showy petals and nect
ar rewards, can evolve independently within the limits set by total re
source availability. In populations of California wild radish, we have
shown previously that petal size, a strong determinant of visitation
by honey bee pollinators, is positively correlated with both pollen an
d nectar production, but not with ovule or seed number per flower. The
se phenotypic associations may reflect selection, environmental correl
ation, and/or genetic constraint. By exerting selection on the petal s
ize:pollen number ratio over two generations, we eliminated the positi
ve correlation between petal size and pollen production, with both cha
racters showing significant change after a single selection episode. O
nce these two floral traits became uncoupled, nectar sugar production
was significantly correlated only with petal size. Our results suggest
that natural selection could readily alter reproductive allocation in
these flowers, and that the phenotypic correlations observed in natur
e may be maintained by selection for effective reproductive phenotypes
.