The influence of population density on pollinator movements and outcrossing rates was studied in experimental arrays of Mimulus ringens (square-stemmed monkeyflower), a wetland perennial species with a mixed-mating system. Each population was composed of genets with unique multilocus combinations of homozygous genotypes, facilitating determination of outcrossing rates through paternity exclusion. Replicate arrays were cloned from the same set of genets to minimize differences in floral and vegetative morphology among density treatments. Two arrays were planted at each of three spacings typical of the range of densities found in natural M. ringens populations. Both the proportion of pollinator flights between plants and the frequency of outcrossing were significantly greater at high density. These results suggest that ecological and demographic factors, such as population density, can significantly influence levels of inbreeding in species with mixedmating systems.Keywords: density, experimental arrays, geitonogamy, outcrossing rate, paternity analysis, pollinator movements. IntroductionIn several species of animal-pollinated flowering plants, outcrossing rates vary dramatically among populations (Harding & Barnes, 1977;Valdeyron et a!., 1977;Ellstrand et al., 1978;Schoen, 1982;Schemske & Lande, 1985; Ritland & Ganders, 1987a;Waller & Knight, 1989; Barrett & Husband, 1990;Dole, 1991;Godt & Hamrick, 1991). This variation is often attributed to ecological factors, such as population density, that influence pollinator movements (Aide, 1986;Waller, 1986;Brown et al., 1989;Watkins & Levin, 1990;Murawski & Hamrick, 1991;Lloyd & Schoen, 1992). Despite theoretical predictions that outcrossing rates will be positively correlated with population density (Bateman, 1956;Handel, 1983;Brown et a!. 1989), the relationship between these variables remains poorly understood.Most studies documenting intraspecific variation in plant mating systems have involved comparisons of natural populations. However, in nature, population density is often confounded with other demographic variables such as plant size, population shape and population size (Barrett & Eckert, 1990;Kunin, 1993;Karron, 1995). In addition, differences among populations in genetically determined floral traits influencing *Correspondence the mating system (Campbell & Abbott, 1976;Schoen, 1982; Marshall & Abbott, 1984; Wyatt, 1984a,b;Holtsford & Ellstrand, 1989 Barrett & Husband, 1990;Clegg & Epperson, 1988;Morgan & Barrett, 1990) may further limit the utility of descriptive studies correlating population density with outcrossing rate.The influence of density on plant mating systems is well suited to experimental study. Spacing between plants is readily manipulated in population arrays, and recent experiments have examined the effects of density on patterns of pollinator movement and seed set in self-incompatible species (Kunin, 1992(Kunin, , 1993. Artificial arrays have also been used to determine outcrossing rates of distinct floral morphs in polymorphic populations...

Anther-stigma separation (herkogamy) is highly variable within populations of Mimulus ringens, a bumblebee-pollinated perennial herb with a mixed-mating system. The relationship between this floral trait and individual outcrossing rates was studied in two experimental populations composed of genets with unique multilocus combinations of homozygous genotypes. This facilitated determination of individual outcrossing rates through unambiguous assignment of paternity to all 1560 sampled progeny. In each population there was significant heterogeneity among maternal families in frequencies of self and outcross progeny. Individual outcrossing rates were positively correlated with anther-stigma separation.

Pollinator movements and pollen-mediated gene dispersal were quantified in experimental populations of square-stemmed monkeyflower (Mimulus ringens), a wetland perennial herb with a mixed-mating system. Each population consisted of genets with unique multilocus combinations of homozygous genotypes, facilitating assignment of paternity to all sampled seeds. Replicate arrays were planted at each of three spacings spanning the range of densities typically observed in natural M. ringens populations. In all six arrays the distribution of gene dispersal distances differed significantly from the distribution of pollinator flight movements. The mean gene dispersal distance was 1.46 times as far as the mean pollinator flight distance. These differences were probably caused by pollen carryover as bumblebees visited up to 12 M. ringens plants on a single foraging trip. Although gene movements exceeded pollinator flight distances, estimates of neighbourhood size were consistently low, ranging from 1.66 to 5.53. Therefore, high levels of random local genetic differentiation are likely in this species.