Data resolution: a jackknife procedure for determining the consistency of molecular marker datasets

The results of genetic diversity studies using molecular markers not only depend on the biology of the studied objects but also on the quality of the marker data. Poor data quality may hamper the correct answering of biological questions. A new statistic is proposed to estimate the quality of a marker data set with regard to its ability to describe the structure of the biological material under study. This statistic is called data resolution (DR). It is calculated by splitting a marker data set at random into two sets each with half the number of markers. In each set, similarities between all pairs of objects are calculated. Subsequently, the similarities obtained for the two sets are correlated. This process is repeated a large number of times. The average of the correlation coefficients obtained in this way is the DR of the dataset. In the present paper, the DR statistic is applied to four studies involving amplified fragment length polymorphism as well as micro-satellite markers. In addition, some properties and possible applications of DR are discussed, including the prediction of the added value of scoring additional markers, and the determination of which similarity measure is, apart from genetical considerations, most appropriate for analyzing the data