GENETIC CONSTRAINTS ON THE INDEPENDENT EVOLUTION OF MALE AND FEMALE REPRODUCTIVE CHARACTERS IN THE TRISTYLOUS PLANT LYTHRUM-SALICARIA

Here we test whether the potential exists for the independent evolutio n of allocation to male, female, and attractive functions within a flo wer. We employed half-sib and parent-offspring regression methods in t he tristylous plant Lythrum salicaria to determine whether there is ad ditive genetic variation for characters important to male and female r eproductive success and whether genetic correlations could constrain t he independent evolution of male and female function. Although signifi cance levels were not consistent among morph types or between populati ons, there were significant narrow-sense heritabilities for several tr aits including stamen mass, pistil mass, perianth mass, petal length, and calyx length. Traits that might be under strong stabilizing select ion to promote specific pollen transfer, such as stamen and style leng ths, had little heritable variation. In the majority of cases in which heritable variation was present, there were positive genetic correlat ions among floral traits. A strong positive genetic correlation appear ed between stamen and pistil mass in the short-styled morph from one o f the populations studied. This suggests that selection might not be a ble to act independently on biomass allocation to male and female flow er parts. No evidence of negative genetic correlations appeared that w ould suggest trade-offs and that could augment a selection response to wards sexual specialization. The observed positive correlations could be explained if we consider the ''functional architecture'' that under lies the covariance structure. If there is more covariance generated b y pleiotropic loci controlling overall flower size than at loci contro lling male versus female allocation, it could result in the observed p ositive covariance. At the phenotypic level, we did find significant n egative partial correlations between male and female traits when flowe r size was controlled, but these trade-offs were among rather than wit hin morphs.