Effective seed harvesting strategies for the ex situ genetic diversity conservation of rare tropical tree populations

Abstract

Tropical rain forests harbor a high diversity of tree species, a large portion of which are considered rare and threatened by anthropogenic factors such as land-use change. Addressing the global need for ex situ rescue populations of such species, we investigated whether certain methods of wild-source seed collection harvest greater amounts of genetic variation from the source population while minimizing the costs expended. We used an individual-based computer program (NEWGARDEN) to conduct simulation trials examining which aspects of seed procurement provided the most cost-effective methods for maximal rare allele capture for single isolated populations of rare tropical tree species. Aspects examined included the geometric pattern of harvesting, location of seed collections within the spatial distribution of the population (central vs. peripheral), geometric partitioning of sampling regions (contiguous vs. subdivided), gene dispersal distance, linear travelling costs, and ratio of seeds collected to plants sampled. Results demonstrated that ratio of seeds collected to plants sampled had the highest effect on genetic diversity harvested relative to sampling location or geometric pattern of collection, explaining 82 and 79% of the variance in alleles harvested, respectively. Collecting from subdivided peripheral sampling regions and populations having long distance gene dispersal led to increases of up to 39 and 51% in the amount of genetic variation conserved respectively. These results suggest that ex situ conservation efforts for rare tropical trees can benefit by collecting fewer seeds from each of a larger number of trees, sampling from peripheral subdivided regions of the population rather than in a contiguous centrally located one under the transect or nearest neighbor strategies, and adjusting harvesting based on the dispersal biology of the source population. This study underscores the need for more information regarding the dispersal characteristics of the target species and demonstrates the value of predictive modeling for planning seed collection projects.