DNA hybridization as a guide to phylogenies: raw data in muroid rodents

A strong criticism of raw data analysis in DNA/DNA hybridization studies (Marks et al., 1988; Sarich et al., 1989) states that a widely used statistic, T50H, can in some cases lead to spurious conclusions, as exemplified by some hominoid and bird DNA/DNA results found in Sibley and Ahlquist (1987a, b). To test this criticism and compare the phylogenetic information deduced from several alternative methods of analysis, I have re-analysed the melting curves of more than 150 hybrid comparisons among muroid rodents, including almost all of my previously published data (Catzeflis et al. 1987). Four major statistics have been compared: Mode, Tm, T50H and NPH (for definitions of terms: see Sibley and Ahlquist 1981). Although the T50H statistic was found to be more variable when compared to Mode and Tm, confirming the claims of Sarich et al. (1989), it nevertheless appeared to be, under certain technical conditions, the method of choice for analyzing distantly related hybrids. The Mode and Tm appeared to be the best indexes of genomic dissimilarity for analyzing more similar genomes. The use of Tm is not very useful for comparing dissimilar genomes, as its range of efficient use was shown to be much more limited towards the lower temperatures (i.e., -larger delta-values) than for the Mode. Mode was found to be the statistic of choice because of i) its large range of application (up to delta-Mode values of 18-20 degrees C), ii) its reproducibility, and iii) its consistency between experiments and in reciprocal comparisons. For the data set on which our earlier conclusions concerning the very rapid rate of DNA change in muroid rodents (Catzeflis et al., 1987) were based, it appears that Mode, Tm and T50H do yield similar conclusions as far as the evolutionary tree of the muroid rodents is concerned. The fourth index, which is the Normalized Percent of Hybridization, appears to be strongly affected by experimental conditions, and should be used with caution.