Does the Pollinator Abundance Hypothesis Explain Geographic Variation in the Breeding System ofPachycereus pringlei?

Abstract

Previous studies have shown that the nectar-feeding bat, Leptonycteris curasoae, is the major pollinator of Pachycereus pringlei (cardon), a columnar cactus whose populations are either gynodioecious or trioecious in the Sonoran Desert. On the basis of evidence of pollinator limitation in females and low flower visitation rates, a hypothesis has been proposed to explain geographic variation in the breeding system of P. pringlei. The pollinator abundance hypothesis predicts high frequencies of males and females in areas where L. curasoae is abundant and high frequency of hermaphrodites in areas where this bat is uncommon. We test this hypothesis by describing the following: first, the relationship between sex frequencies and bat abundance as revealed by nocturnal pollen deposition on stigmas of female plants and, second, the extent of pollinator limitation for fruit set in females and hermaphrodites in four populations of P. pringlei from coastal Sonora, Mexico. We also describe geographic variation in the relative fitness of females compared with hermaphrodites and outcrossing rates and inbreeding depression of self-compatible hermaphrodites in these populations. Some of these parameters should vary with bat abundance according to this hypothesis. Sex-frequency distributions showed a clear latitudinal pattern, with gynodioecious populations in the southern range and trioecious populations in the northern range. Large differences in nocturnal pollen deposition in females and abundance of L. curasoae were detected among populations but not as predicted by the pollinator abundance hypothesis. Fruit set was greater in females than in hermaphrodites in different pollination treatments, and no evidence of pollinator limitation was detected in either sex class. In general, the relative fitness of females was greater than hermaphrodites in most populations. Outcrossing rates in hermaphrodites were close to 1.0 in each population. Field and experimental data indicated no evidence of inbreeding depression. Our results do not support the hypothesis that the sex-frequency distribution of P. pringlei is closely correlated with the abundance of L. curasoae but do demonstrate that females outperform hermaphrodites in gynodioecious and trioecious populations. Studies of the distribution of mtDNA variants are needed to understand geographic variation in the breeding system of P. pringlei.