Hypothesis testing of genetic similarity based on RAPD data using Mantel tests and model matrices

Abstract

Clustering and ordination procedures in multivariate analyses have been widely used to describe patterns of genetic distances. However, in some cases, such as when dealing with Jaccard coefficients based on RAPD data, these techniques may fail to represent genetic distances because of the high dimensionality of the genetic distances caused by stochastic variation in DNA fragments among the units analyzed (species or populations). In this note, we show how Mantel tests can be used to test hypotheses about genetic distances and avoid problems with multivariate data representation. The procedure is illustrated with RAPD data comparing 20 anuran species from the cerrado, in which Jaccard genetic distances were compared to three model matrices linking the species at distinct taxonomic levels.

Distribution of stream macroalgal communities from the mid-western region of Paraná state, southern Brazil: importance of local scale variation

“…3) and ordination procedures failed to produce an adequate representation of the sampling sites based on floristic characteristics. The cophenetic correlation was 0.741, considered by Rodrigues et al (2002) as a relatively low value by standard criterion (usually only values above 0.8 provided a reasonable representation of similarity). The following correlations between Jaccard similarity matrix and model matrices were found: 1) chemical characteristics matrix -r=-0.024, p=0.551; 2) physical characteristics matrixr=0.159, p=0.063 and 3) distance among sampling sites matrix -r=-0.058, p=0.253.…”

Genetic relatedness of Plasmodium falciparum isolates and the origin of allelic diversity at the merozoite surface protein-1 (MSP-1) locus in Brazil and Vietnam

“…In Model II, three levels of similarity were considered: values of 0, 0.5 and 1 were assigned to pair-wise comparisons of isolates with identical, nearly identical and different MSP-1 alleles, respectively. Mantel tests were used to assess the correlation between the microsatellite similarity matrix and each of the MSP-1 model matrices [12]. Coefficients of determination ( r 2 ) were calculated and two-tailed P values were obtained by Monte Carlo simulation with 6,000 permutations performed with the PopTools software.…”

Genetic similarity in a hybrid population of 'Montenegrina' and 'King' mandarins

“…Rodrigues & Diniz-Filho (1998) suggest that the UPGMA clustering method works better in large matrixes, requiring verification by the cophenetic correlation (r). The UPGMA clustering obtained by Rodrigues et al (2002) did not reach a high r, along with other ranking methods.…”