It has long been suggested that the intergrading morphology and malformed pollen of many North American Crataegus may result from extensive recent hybridization. Hybridization has been inferred on the basis of the intermediacy of hybrid morphology (and, in some cases, flavonoid pattern) relative to the putative parents. Hybridization has been documented between diploid taxa, but little is known about the extent to which polyploids may also be involved. Gametophytic apomixis has been observed in polyploids, so that any cross involving a polyploid has the potential to reproduce itself by agamospermy. We conducted pollination experiments over three years using diploid and tetraploid North American taxa and one diploid European taxon. As in field collections and in previous experiments, very few flowers set seed in all locules and a minimum of one filled pyrene was necessary for fruit maturity. The data are now sufficient to state that in the wide crosses: (1) fruit set can equal that with open pollination (approaching 50%) or with some self-pollen; and (2) seed set is significantly lower than in open-pollinated fruit. This suggests that a somewhat rare event is necessary for successful seed formation in pollinations between plants of different sections and ploidy levels. The mechanism of endosperm formation, the parentage of the embryo (from gamete fusion or parthenogenesis), and the viability of the seed remain to be investigated.

Key words: Crataegus, fruit set, gametophytic apomixis, hybridization, seed set