Reciprocal style polymorphisms are not easily categorised: the case of heterostyly in Lithodora and Glandora (Boraginaceae).

Interest in reciprocal floral polymorphisms, such as heterostyly, has increased in recent decades because they can be used as suitable model systems to study mechanisms of outbreeding and disassortative mating in plants. Heterostylous plants are characterised by the presence of discrete morphs that differ in sex organ position and in some other ancillary traits. As regards sex organ deployment, different types of polymorphisms have been described, depending on number and type of discrete classes present in populations and degree of reciprocity between them. However, a clear-cut characterisation of stylar polymorphisms does not appear to be the best approach when there is great variability among populations because of continuous variation of some of traits examined. A recent study in Lithodora sensu lato (recently split into two separate genera, Lithodora and Glandora) showed a wide variation in sex organ position across species in the genus, which warrants precise population analysis of stylar polymorphism and its reciprocity. We provide a detailed morphometric analysis of flower sexual traits and include those considered to be ancillary characters. We report a wide variation in these traits in populations of Lithodora s.l. and highlight the subjectivity of the former characterisation of style polymorphism based on visual inspection. Ancillary traits appear repeatedly in Lithodora and Glandora, particularly in the latter. The appearance of these traits seems to be related to greater reciprocity between sexual whorls in Glandora, with the exception of G. prostrata. These results agree with evolutionary steps proposed in the build-up of heterostyly according to some evolutionary models. We also examined variation in polymorphisms in light of current models for evolution of heterostyly, and, more specifically, we sought to verify the prediction that flower traits as a whole (i.e., flower integration) respond to selective pressure to assure the exact location of pollen on the pollinator body. Most reciprocal populations and species, where between-morph pollen transfer is expected to be higher, would show greater integration. Our results confirm this hypothesis.