CLEISTOGAMY LIMITS REDUCTION IN SEED SET IN DANTHONIA-COMPRESSA (POACEAE) CAUSED BY ALLELOPATHIC POLLEN FROM PHLEUM-PRATENSE (POACEAE)

Allelopathic pollen of Phleum pratense (Poaceae) has been shown to red uce pollen germination and seed set of heterospecific Poaceae. Danthon ia compressa has a mixed breeding system, i.e., both cleistogamous and chasmogamous florets. Previous studies revealed that extracts from 1- 5 pollen grains of P. pratense reduce pollen germination and seed set although direct field evidence was lacking. Therefore, we tested wheth er the presence or absence of pollen from P. pratense affected seed se t in D. compressa. In 1991, inflorescences of P. pratense were clipped to reduce pollen dispersal in an abandoned hayfield. In an adjacent a bandoned hayfield, P. pratense flowered unimpeded. Comparing the clipp ed versus unclipped fields, the mean number of pollen grains of P. pra tense/stigma of D. compressa decreased from 7.6 to 0.19 while mean see d set/culm increased from 50% to 76.9%. All of the increased seed set was related to increased seed set in chasmogamous florets. In 1992, wh en P. pratense was allowed to flower in both fields, mean seed set/cul m was less than 50%. In a third population of D. compressa, at a site where P. pratense was not present, mean seed set/culm was over 76% in both years. Cleistogamy in D. compressa maintained a minimum seed set of 41.9% in all fields. Hence, there is an effective barrier against a llelopathic pollen of P. pratense. The implications of loss of chasmog amously produced seed are unclear. There were no differences in mean s eed mass or size between chasmogamous and cleistogamous florets. Hence fitness may be unaffected. Although loss of chasmogamous seeds may re sult in increased sibling competition in cleistogamously produced seed lings, this hypothesis has yet td be tested. The relative numbers of c hasmogamous and cleistogamous florets did not change in 1991-1992, alt hough there were more chasmogamous florets produced by D. compressa at the site where P. pratense did not exist. Tests of whether pollen all elopathy acts as a selection pressure on the relative numbers of chasm ogamous and cleistogamous florets in D. compressa will require long-te rm field studies.