Abstract:

Floral traits are predicted to diverge in a species due to different selective forces acting along ecological gradients. As elevation and latitude gradients are always indicative of drastic variation in plant communities, elevation and latitude models should indicate how natural selection has shaped some floral traits and/or how floral characteristics have adapted to local or regional conditions. The measurement of floral traits among multiple populations can be used to observe floral trait divergence along geographic gradients. Primula poissonii, a distylous species, is widely distributed in the core areas of the Hengduan Mountains in southwest China. Distyly plays an important role in inter-morph cross pollination in P. poissonii. In this study, we investigated 16 populations of P. poissonii in Yunnan and Sichuan provinces. We measured eight floral traits (corolla, androecium and gynoecium) in short- and long-styled morphs of each population. Sexual organ reciprocity was estimated by calculating the spatial distribution overlap between the height of sexual organs. The spatial distribution of sexual organs was fitted by using the maximum likelihood method with a normal distribution. Measured character and sexual organ reciprocity were tested to examine the floral trait variation along geographic gradients by using linear regression models. Results showed that floral tube opening width and anther height in both morphs and stigma height in the short-styled morph displayed a positive correlation with increasing elevation; the petal length in long-styled morph flowers, the distance between the stigma and tube opening in long-styled morph, the floral tube length of short style morph, and corolla width in both morphs were decreased along elevational gradients. However, petal length of the short-styled morph, the distance between the highest sexual organs and tube opening in short-styled morph, the distance between anther and stigma in both morphs, and the floral tube length and stigma height in long-styled morph did not correlate with elevation. Furthermore, floral traits except for stigma height of short-styled morph were significantly negatively correlated with latitude gradients. In addition, floral traits vary along a longitudinal gradient. Differences in selective forces may contribute to the geographic mosaic of floral trait variation and sexual reciprocity may be maintained while the other traits differential. We prefer to conclude that floral trait divergence in P. poissonii is driven by flower visitors.

Key words: pollinator-mediated selection, reciprocal herkogamy, geographic gradients, Primula poissonii, floral trait divergence