Population genetic structure, pollen dispersal, and local adaptation in Quercus oleoides forests of Costa Rica

My goal was to understand the evolutionary history of Quercus oleoides in Costa Rica in order to more effectively conserve and possibly restore the region's seasonally dry forest in the future. This study combined analyses of genetic diversity, pollen dispersal, and the growth and survival of various seedling families to provide an integrated evaluation of the response of a critical dry forest species to fragmentation and will help guide management and restoration efforts in the Area de Conservacion Guanacaste (ACG).;The subsequent chapters describe three previously unanswered questions about the past, present, and future status of Q. oleoides in the ACG. In Chapter 1, I characterized the standing genetic diversity of 13 Q. oleoides populations and the geographic structuring of that diversity. I found that Q. oleoides in Costa Rica contained a high level of genetic diversity as well as genetic variation that is geographically structured across the landscape. The degree to which this structuring is due to fragmentation, however, is small in comparison to the genetic structure that has existed prior to fragmentation. This is somewhat counterintuitive due to the expectations provided from population genetic theory that can be applied to fragmented landscapes.;Population genetic variation consists of the sum of all genetic variation among individuals within the population. It can be measured by parameters including allelic richness (A) and expected heterozygosity (He). Allelic richness is the average number of alleles per locus and observed heterozygosity is compared to expected heterozygosity under Hardy-Weinberg equilibrium conditions. Wright's F-statistics are means of describing how genetic diversity is partitioned in a population. High values for FST indicate that subpopulations have very different gene frequencies than the total population. A loss in heterozygosity can occur with inbreeding due to the higher chance that offspring of a mating event between two individuals with the same common ancestor may share the same alleles. One method for quantifying genetic variation within species is to assay highly variable regions of repeated DNA units called microsatellites. Individuals of a population were characterized by the differences in length of 11 of these non-coding genetic units.;Although I observed no significant correlations between genetic distance and geographic distance, flowering time similarity, or environmental similarity in Chapter 1; I analyzed pollen dispersal more rigorously in Chapter 2 in order to better calculate contemporary pollen dispersal distance estimates. It is not unusual for studies of plant populations in fragmented landscapes to report few of the negative consequences predicted by theory, and that is because pollen may actually disperse father in fragmented landscapes. My results from two separate molecular analyses of pollen dispersal distance using 8 of the microsatellite markers from Chapter 1, however, indicated that the average pollen dispersal that resulted in viable offspring predominately occurred over very short distances. Both the paternity exclusion and two-generation methods yielded similarly short dispersal distance estimates. Evidence from the physical trapping of pollen in one location indicated that pollen was capable of moving much farther, however, so the importance of long distance pollen dispersal may rely more on phenology. I observed staminate and pistillate flowering times in 10 sites over two years, but the lack of strong seasonality in flowering obscured any obvious patterns.;The geographic structuring of genetic diversity and the short average pollen dispersal distance provide a sound foundation for testing for local adaptation in Q. oleoides populations. In Chapter 3, I compared the growth and survival of upland and lowland maternal families in their native and foreign environments. The native environment of the populations of families differs most notably in their elevations and the lack of precipitation during the 4-5 month dry season in the lowlands. Seedlings planted in the lowland garden from both populations experienced a much higher level of mortality than seedlings planted in the upland garden, but using the aster models approach for comparing the likelihood of various models of combined growth and survival data, we did not identify evidence for local adaptation. (Abstract shortened by UMI.)...