Estimation of outcrossing rates at hierarchical levels of fruits, individuals, populations and species in Magnolia stellata.

In plant species with mixed mating systems, differences in diverse factors-including their pollination system, flowering phenology, life form and susceptibility to inbreeding depression-cause variation in outcrossing rates among fruits within individuals, among individuals within populations and among populations within species. To quantify this hierarchical variation, we examined outcrossing rates at the seed stage in five populations of Magnolia stellata, a self-compatible, insect-pollinated and protogynous tree species. For this purpose, we sampled 1498 seeds within 204 fruits obtained from 56 individuals of the five populations, determined genotypes of the sampled seeds and maternal individuals at six polymorphic microsatellite loci, then estimated outcrossing rates and their variance components at four hierarchical levels (fruits, individuals, populations and species) using a nested analysis of variance-type linear model with a Bayesian approach. The species-level outcrossing rate was 0.730 (95% credible interval, 0.595-0.842), indicating that this species has a mixed mating system. Outcrossing rates were not significantly different among populations, but were significantly different among individuals within populations. Variance components at the levels of individual and fruit were statistically supported and were highest for the former. Thus, factors influencing outcrossing rates at the individual level, such as differences in flowering phenology and early-stage inbreeding depression, appear to have important effects within these M. stellata populations, but not among them. The method of hierarchically estimating outcrossing rates using a Bayesian approach, as applied in this study, is compared with conventional methods for estimating outcrossing rates, and the statistical properties of the Bayesian approach are discussed.

Genetic variation and differentiation in populations of a threatened tree, Magnolia stellata: factors influencing the level of within-population genetic variation.

Genetic variation and differentiation in Magnolia stellata were studied in 20 populations distributed across most of the species' range using 10 microsatellite markers, and the factors influencing their levels of within-population genetic variation were examined. Generally, populations distributed intermittently from southern Gifu to central Aichi Prefectures showed substantially higher levels of genetic variation (exceptions included populations located at unusually high altitude sites or western and southern edges of the range) than more isolated populations on the Atsumi Peninsula of southern Aichi Prefecture and in northern Mie Prefecture. Significant isolation-by-distance patterns were detected in genetic differentiation among the studied populations, and a neighbor-joining tree based on D(A) distances among the populations reflected well the geographical positions of the populations. The level of within-population genetic variation was significantly influenced not only by the size of the populations (represented by the number of reproductive individuals) but also by their degree of isolation (represented by the number of populations within a radius of 0.5 km around them). Therefore, areas within radii of 0.5 km may encompass M. stellata metapopulations, in which gene flow may usually occur. We suggest that this area may be a suitable standard for constructing conservation units for the species.

Size distribution and genetic structure in relation to clonal growth within a population of Magnolia tomentosa Thunb. (Magnoliaceae).

To establish a baseline for conservation of a threatened clonal tree, Magnolia tomentosa, we investigated size distribution and genetic structure within a population, using six microsatellite markers. Within the study site, 1044 living ramets (stems) were distinguished into 175 genets (individuals). The mean number of ramets per genet was 5.97, and 76% of all genets had multiple ramets. Genets, which apparently produced new ramets through sprouting and layering, were generally composed of several large ramets and many small ramets. Spatial autocorrelation analysis of microsatellite alleles revealed positive autocorrelation over short distances for both ramets and genets. The Moran's I-value of ramets in the shortest distance class was 3.8 times larger than that of genets, reflecting the effect of clonal growth. To analyse the size-class differences in genetic structure, the 175 genets were separated into two size classes, small and large. The correlogram for the small genets exhibited positive spatial autocorrelation in the shortest distance class, but this was not the case for the correlogram for the large genets, indicating that genetic structure is weakened or lost through self-thinning as the genets grow. The FIS value over all loci for the small genets was positive and deviated significantly from zero, while the corresponding value for the large genets was close to zero. The excess homozygotes in the small genets may be the result of genetic substructuring and/or inbreeding, and the reduction in homozygote frequency from the small to large genets may be because of loss of genetic structure and/or inbreeding depression.