Amplified fragments were scored as presence or absence of bands, and a binary matrix of RAPD phenotypes was assembled. Only polymorphic loci were used. The data analysis were further restricted to bands whose observed frequencies in each population were less than 1 – (3/N), where N is the number of plants analysed [14] to avoid significant bias in estimates of population genetic parameters. Non-random associations between pairs of loci were investigated using the Spearman rank correlation, with the SPSS 12.0 computer package. Intrapopulational genetic diversity was assessed as the proportion of polymorphic loci (P, using the 95% criterion), mean number (N) and effective number (Ne) of alleles per locus, Nei's gene diversity [15] (He, that was adopted assuming the populations to be in Hardy-Weinberg equilibrium, although we were not able to investigate this since dominant markers were used), and Shannon's information index. The latter can be considered a measure of phenotypic diversity (I = −Σpilog2pi, where pi is the frequency of presence or absence of a given RAPD fragment; [16]), assuming that populations are not in Hardy-Weinberg equilibrium. This index is frequently used in RAPD analysis because it is insensitive to bias that may be introduced into data owing to undetectable heterozygosity [17]. Calculations were performed using the GENALEX 6.5 software package [18].
The distribution of genetic diversity within and among populations was assessed using Nei's genetic differentiation degree (GST) [19]. Analysis of Molecular Variance (AMOVA), based on squared Euclidean distances between all pairs of RAPD phenotypes [20], was employed using the Arlequin software [21]. The AMOVA procedure was performed in order to further partition the total genetic variation among taxa, among population within subspecies and within populations, and to compute a pairwise population FST value matrix according to Weir and Cockerham [22]. The statistical significance of the covariance components was estimated by nonparametric randomisation tests using 10000 permutations. The null distribution of pairwise FST values under the hypothesis of no differences between the populations was also tested by using a permutational approach (10000 replicates).
Cluster analysis was performed on pairwise FST distances using the Unweighted Pair Group Method with Arithmetic Averages (UPGMA; [23]) with the SAHN program in NTSYS-pc 2.10j[24]. A cophenetic value matrix (COPH in NTSYS) was produced from the dendrogram and compared with the genetic distance matrix by using the MXCOMP program in NTSYS to estimate the goodness of fit of the cluster analysis. Principal coordinate analysis (PCoA) was also performed, based on pairwise FST distances matrix (DCENTER and EIGEN procedures in NTSYS), to better understand genetic relationships among populations.
The genetic structure of the populations was also explored using Bayesian clustering, performed using the software Structure [25]. The program uses a Markov chain Monte Carlo (MCMC) algorithm to cluster individuals into populations on the basis of multilocus genotypic data. Individual multilocus genotypes are first assigned probabilistically to genetic clusters (K) without considering sampling origins. Admixed or hybrid individuals can be identified as they will have a fraction of their alleles derived from each genetic cluster. Posterior probabilities of K were calculated from the means of 20 runs for each value of K (from 1 to 6), and the optimum K determined using the method of Evanno et al. [26].
Amplified fragments were scored as presence or absence of bands, and a binary matrix of RAPD phenotypes was assembled. Only polymorphic loci were used. The data analysis were further restricted to bands whose observed frequencies in each population were less than 1 – (3/N), where N is the number of plants analysed [14] to avoid significant bias in estimates of population genetic parameters. Non-random associations between pairs of loci were investigated using the Spearman rank correlation, with the SPSS 12.0 computer package. Intrapopulational genetic diversity was assessed as the proportion of polymorphic loci (P, using the 95% criterion), mean number (N) and effective number (Ne) of alleles per locus, Nei's gene diversity [15] (He, that was adopted assuming the populations to be in Hardy-Weinberg equilibrium, although we were not able to investigate this since dominant markers were used), and Shannon's information index. The latter can be considered a measure of phenotypic diversity (I = −Σpilog2pi, where pi is the frequency of presence or absence of a given RAPD fragment; [16]), assuming that populations are not in Hardy-Weinberg equilibrium. This index is frequently used in RAPD analysis because it is insensitive to bias that may be introduced into data owing to undetectable heterozygosity [17]. Calculations were performed using the GENALEX 6.5 software package [18].The distribution of genetic diversity within and among populations was assessed using Nei's genetic differentiation degree (GST) [19]. Analysis of Molecular Variance (AMOVA), based on squared Euclidean distances between all pairs of RAPD phenotypes [20], was employed using the Arlequin software [21]. The AMOVA procedure was performed in order to further partition the total genetic variation among taxa, among population within subspecies and within populations, and to compute a pairwise population FST value matrix according to Weir and Cockerham [22]. The statistical significance of the covariance components was estimated by nonparametric randomisation tests using 10000 permutations. The null distribution of pairwise FST values under the hypothesis of no differences between the populations was also tested by using a permutational approach (10000 replicates).Cluster analysis was performed on pairwise FST distances using the Unweighted Pair Group Method with Arithmetic Averages (UPGMA; [23]) with the SAHN program in NTSYS-pc 2.10j[24]. A cophenetic value matrix (COPH in NTSYS) was produced from the dendrogram and compared with the genetic distance matrix by using the MXCOMP program in NTSYS to estimate the goodness of fit of the cluster analysis. Principal coordinate analysis (PCoA) was also performed, based on pairwise FST distances matrix (DCENTER and EIGEN procedures in NTSYS), to better understand genetic relationships among populations.The genetic structure of the populations was also explored using Bayesian clustering, performed using the software Structure [25]. The program uses a Markov chain Monte Carlo (MCMC) algorithm to cluster individuals into populations on the basis of multilocus genotypic data. Individual multilocus genotypes are first assigned probabilistically to genetic clusters (K) without considering sampling origins. Admixed or hybrid individuals can be identified as they will have a fraction of their alleles derived from each genetic cluster. Posterior probabilities of K were calculated from the means of 20 runs for each value of K (from 1 to 6), and the optimum K determined using the method of Evanno et al. [26].
การแปล กรุณารอสักครู่..
