Figure 3 Comparison of observed SOD differences with those from a simulated perfect clock. The closed circles are observed differences in SOD; The open squares are simulated with a perfect clock (see text for clock parameters). The vertical bars show 1 SD about the mean. Observed and simulated values are for the same time despite their being horizontally offset slightly for clarity.

involved permits one to conclude that what at first appeared to be a very inaccurate clock (1) may have been inaccurate simply because the necessary corrections had not been made. In the same spirit as it was appropriate to correct Zuckerkandl-Pauling differences (1962) by accounting for multiple hits (Margoliash and Smith, 1965) and then correcting that for a finite number of alternative character states (Jukes and Cantor, 1969), it may then be necessary to take into account that the variable positions are not the same in different lineages.

Any estimate of divergence time, where the covarion process has not been taken into account, is in danger of significant error. One might naively believe that, if the differences between species are less than 10%,



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement