Population-level variation in the expression of heterostyly in three species of Rubiaceae: does reciprocal placement of anthers and stigmas characterize heterostyly?

Heterostyly (i.e., reciprocal placement of anthers and stigmas between two or three floral morphs) is hypothesized to enhance outcrossing and reduce selfing. However, few studies have documented reciprocity among individual plants; instead, mean anther and stigma heights for floral morphs are usually reported, masking interindividual variation. We measured eight floral dimensions for individuals in five populations of three heterostylous Rubiaceae. The three methods used to quantify reciprocity yielded different conclusions regarding the degree to which populations conformed to expectations for heterostylous plants. Only Psychotria poeppigiana had stigma and, to a lesser degree, anther heights in discrete classes. Variation among plants of Bouvardia ternifolia and Psychotria chiapensis yielded a continuum of anther and stigma heights across populations. Comparison of distances between stigma and anthers indicated that only flowers of B. ternifolia had, as expected, a constant value for this distance. Finally, regression relationships between anther and stigma heights and corolla length showed that only in one population each of B. ternifolia and P. poeppigiana, and in P. chiapensis, was distance between anthers and stigmas the same across the range of corolla sizes for both floral morphs. Variation among these species in expression of heterostyly was not clearly linked to phylogenetic relationship or pollinator syndromes. Two approach herkogamous (AH) species were studied for comparison. Flowers of Psychotria brachiata were consistently AH, but flowers of P. pittieri were highly variable. Determining fitness consequences of population-level variation in sexual systems requires studies linking floral morphology to pollinator behavior and pollen transfer.