Population Genetics of Tropical Trees

Tropical tree breeding patterns – We use polymorphic allozyme and microsatellite marker loci to examine breeding patterns within populations of several tropical tree species. Our research indicates that the majority of the species analyzed (about 20 species to date) are outcrossing. We have also shown that the density of flowering individuals has a distinct impact on the genetic composition of pollen received by each tree; species with low densities of flowering individuals have greater heterogeneity in pollen allele frequencies than species that occur and flower at higher densities. By using multi-locus analyses of large progeny arrays (i.e., paternity analyses) several breeding structure parameters can be determined; the proportion of long-distance gene movement, pollen movement patterns within populations, and the relative success of individuals as pollen donors. Our results indicate that male reproductive success varies greatly among individuals. Pollen movement in most of the species examined is extensive with 50% or more of the pollen traveling 500 m or more.

We have also taken advantage of the unique breeding systems of figs (Ficus), mimosoid legumes and orchids (all produce fruits in which the seeds are full-sibs i.e., there is a single pollen donor) to estimate the number of individuals contributing pollen to an individual tree. We have found that the number and identity of the pollen donors varies greatly among maternal plants and among years for individual plants. This “full-sib” paternity analysis provides a level of detail of the breeding system that is unavailable for any other analyses. We are continuing these analyses to better understand spatial and temporal variation in the breeding patterns of these species.

Tropical forest fragmentation – As has been well documented, tropical forests are being destroyed at an alarming rate. Such landscape level destruction often produces fragments of the original forest that vary considerably in their size and in the degree of physical isolation. Much of my current research is focused on the genetic consequences of tropical forest fragmentation. The majority of this research occurs in the tropical dry forest of Guanacaste Province in Costa Rica. The chief study species is Enterolobium cyclocarpum, the Guanacaste tree. Enterolobium cyclocarpum is an insect-pollinated mimosoid legume which is a dominant tree in the drier regions of Central America. We use full-sib analyses to compare rates of pollen movement in heavily disturbed and relatively undisturbed forests.

We are also investigating the role that isolated individuals play in the breeding populations of this species. Our preliminary results show that, on average, 75% of the effective pollen is the result of long-distance (>500m) dispersal by its nocturnal insect pollinators. Furthermore, it appears that trees in isolated fragments receive pollen from as many or perhaps more pollen donors than trees in undisturbed forest. Isolated trees, somewhat surprisingly, receive pollen from a greater number of pollen donors than trees that grow in less isolated situations indicating that these isolated trees may play a significant role in the breeding populations of this species. Recently we have begun to investigate how the flowering phenology of individual trees is associated with the patterns of mating we have documented genetically.

Dr. Peter Smouse (Rutgers University) and I are applying an analytical procedure termed Two-Gener to analyze the breeding system and gene flow in several E. cyclocarpum populations. The Two-Gener approach, which was developed by Dr. Smouse, allows one to obtain estimates of the effective number of pollen parents and pollen movement distances without performing a complete paternity analysis. This approach allows us to partition the data to examine the effects of habitat type, distance to the nearest neighbor, and individual tree phenology on these breeding structure parameters. These results are being compared directly to those from the full-sib analyses discussed above. The Two-Gener analysis will also be applied to an expanded sampling scheme of A. julibrissin in the Athens, Georgia area. Seed collections are being made from 50+ clusters of three trees each. This will allow us to determine how much of the tree-to-tree variation in the pollen donor pool occurs at a local scale (<50m) or whether most of the variation in the pollen donor pool occurs at a larger spatial scale (> 1km).