Ecological Archives C006-032-A3
Cheng-Yuan Xu, Shaoqing Tang, Mohammad Fatemi, Caroline L. Gross, Mic H. Julien, Caitlin Curtis, and Rieks D. van Klinken . 2015. Population structure and genetic diversity of invasive Phyla canescens: implications for the evolutionary potential. Ecosphere 6:162. http://dx.doi.org/10.1890/es14-00374.1
Appendix C. Supporting information of methods.
Assessing the risk of overestimating Fst
There is a risk that the relatively low number of samples per population (for some South American populations) and microsatellite loci in our study may lead to an overestimation of Fst. We therefore conducted several analyses to test this issue. Each analysis showed our interpretation to be robust:
(1) POWSIM simulation for a sensitivity analysis with a pre-defined Fst = 0.25 (previous work suggest Fst > 0.2 for P. canescens). For South American populations with differentiation corresponding to a pre-defined Fst = 0.25 and an effective population size of 10,000, simulation with our sample size (South American populations before pooling) and loci number resulted in Fst = 0.252 ± 0.002 (95% CI) (POWSIM v4.1, Ryman 2011).
(2) Calculating Fst after excluding all pooled populations. Excluding pooled populations from Fst calculation had limited effect (0.266 for 18 populations with >20 individuals vs. 0.287 for all populations, GenAlex).
(3) Comparing genetic structure genetic structure estimated with microsatellite data and our previous ISSR data. Our microsatellite results are qualitatively consistent with our previous ISSR study on the same system (15 populations with 5 each respectively from Argentina, Australia and France, 30 individuals per population, 0.236 vs 0.210 for among-region differentiation, AMOVA).
(4) Separately estimating Fst for each loci. Fst calculations were not sensitive to loci number in our study. Including all 6 loci gives the same results (0.287) as calculating Fst for each individual loci (0.233 – 0.430) (GenAlex).Therefore, our data set provides reliable estimation of the broad-scale spatial genetic structure of P. canescens.
Putative sources of invasive individuals
Using the third and the fourth level subdivision of South American individuals produced consistent results with that based on the second level subdivision. That is, introduced individuals were primarily linked to central Argentine clusters and Peruvian and Chilean clusters having only limited contribution (data not shown). However, on these levels of South American subdivision, the contribution of many native clusters were not statistically significant (with 95% CI includes 0 in Bayesian assignment or bootstrap support <50% in NJ tree). In this case the source of the invasive cluster cannot be reliably further determined. We therefore concluded that the second level native subdivision was most appropriate for identifying the putative population source of invasive individuals with our data.