We investigated the effects of forest fragmentation on reproduction of a common understory perennial, Trillium camschatcense, in Hokkaido, Japan. We measured seed production of 12 (1998) and 14 (1999) populations, and we explored the relationships of seed production to the number of flowering plants (  population size) and to the surrounding landscape condition (  landscape type). We discriminated between two landscape types, isolated and continuous, to describe the quantity and distribution of forests around T. camschatcense populations. Population size may be correlated with attractiveness to pollinators and the availability of compatible mates, whereas landscape condition can affect the abundance of pollinators. The number of seeds per flower was related to population size in 1999, but not in 1998. In both years, small populations of <50 flowering plants produced few seeds. Although the effect of landscape type was not significant, continuous populations with abundant forest series in their neighborhood tended to produce a higher number of seeds than isolated populations located in small, isolated forest. We attribute the decrease in seed production to pollen limitation because the stigmatic pollen load had a positive relationship to seed production and addition of pollen by hand increased seed production. Variation in population structure (   four stage classes) among populations was better explained by population size than number of seeds produced, suggesting the relative importance of population size for the demography of T. camschatcense. Larger populations had a higher proportion of seedlings, possibly because large populations had fewer edge effects, which reduce seedling recruitment and survival. Small populations with 220 flowering plants or less showed almost no seedling recruitment. The data on seed production and population structure suggest that continuous populations with at least 1000 flowering plants may be required to avoid the effects of fragmentation.

Trillium camschatcense, a long-lived common woodland herb, has been experiencing intensive habitat fragmentation over the last century in eastern Hokkaido, Japan. We examined the genetic diversity and population genetic structure of 12 fragmented populations with different population sizes using allozyme electrophoresis. The percentage of polymorphic loci and mean number of alleles per locus were positively related to population size, probably due to the stochastic loss of rare alleles (frequency of q <0.1) in small populations. Populations with 350 flowering plants or fewer had lost almost all of their rare alleles. While the heterozygosity and inbreeding coefficient were not related to population size, some small populations showed relatively high inbreeding coefficients. In spite of the low genetic differentiation among overall populations (F ST =0.130), local population structuring was recognized between the two geographically discontinuous population groups. Within groups, sufficient historical gene flow was inferred, whereas a low dispersal ability of this species and geographical separation could produce apparent differentiation between groups.

Genetic diversity within species is a potentially important, but poorly studied, determinant of plant community dynamics. Here we report experiments testing the influence of genotype identity and genotypic diversity both on the invasibility of a foundation, matrix-forming species (Kentucky bluegrass, Poa pratensis), and on the invasiveness of a colonizing species (dandelion, Taraxacum officinale). Genotypes of Kentucky bluegrass in monoculture showed significant variation in productivity and resistance to dandelion invasion, but the productivity and invasion resistance of genotypic mixtures were not significantly different from those of genotypic monocultures. Indirect evidence suggested temporal shifts in the genotypic composition of mixtures. Dandelion genotypes in monoculture showed striking and significant variation in productivity and seed production, but there was no significant tendency for these variables in mixtures to deviate from null expectations based on monocultures. However, productivity and seed production of dandelion mixtures were consistently greater than those of the two least productive genotypes, and statistically indistinguishable from those of the three most productive genotypes, suggesting the possibility of greater invasiveness of genotypically diverse populations in the long run due to dominance by highly productive genotypes. In both experiments, the identity of genotypes was far more important than genetic diversity per se.

No abstract

No abstract