Simulation Study Of Impact Of Landscape Features And Home-Range Size On Genetic Variation Based On The Optimal Statistical Method

Landscape genetics is the amalgamation of landscape ecology and population genetics. Results from such studies may have great implications for biodiversity conservation and reserve management. Individual-based landscape genetic methods have become more and more popular for detecting how fine-scale landscape influences gene flow. Gene flow and landscape features are often correlated with geography. One method of partial statistics, Mantel tests, often correlate genetic differentiation and landscape variables by controlling the effects of geography. Comparing correlations of ecological and genetic distances calculated among individuals of a species is a common used method in landscape genetic. However, animals may move through their home range in some locations, which leads to some degree of sampled individual’s spatial uncertainty. In this study, we use simulated landscapes to evaluate the performance of partial Mantel tests and ordination method for detecting isolation by landscape resistance. We described the effects of suitable habitat proportion and fragmentation on metrics of accuracy (frequency of correct results) for each using individual-based gene flow simulations. We also evaluate the ways that spatial uncertainty, landscape characteristics, and genetic stochasticity interact to influence the strength and variability of conclusions about landscape genetics relationships. We used spatially explicit, individual-based simulation modeling to generate different landscapes, varying in resistance, aggregation, and structural connectivity. We created true and alternate locations for 500 individuals, calculated ecological distances (least-cost paths), and simulated genetic distances among individuals. We simulated genotypes at 10 neutral loci and investigated whether the effect could be detected in partial Mantel test while controlling for the effects of isolation-by-distance. Mantel tests exhibited higher accuracy in landscape genetic simulation with low habitat amount, and the accuracy increased with increasing fragment. Ordination methods were effective for detecting isolation by resistance in different suitable habitat amount. A combination of statistical tests can provide higher accuracy for further landscape genetic simulation study. Spatial Uncertainty, resistance, aggregation and cohesion all contributed to genetic differentiation. The effect of spatial uncertainty changed with varied aggregation, and when resistance was larger, landscape-genetics relationships was nearly not influenced by other variables. Larger correlations occurred with low aggregation and low levels of cohesion and genetic stochasticity was influenced by landscape characteristics.