skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: EVOLUTIONARY TREES FROM GENE FREQUENCIES AND QUANTITATIVE CHARACTERS: FINDING MAXIMUM LIKELIHOOD ESTIMATES

Authors:
 [1]
  1. Department of Genetics, University of Washington, Seattle Washington 98195
Publication Date:
Sponsoring Org.:
USDOE
OSTI Identifier:
1401556
Resource Type:
Journal Article: Publisher's Accepted Manuscript
Journal Name:
Evolution
Additional Journal Information:
Journal Volume: 35; Journal Issue: 6; Related Information: CHORUS Timestamp: 2017-10-20 17:19:09; Journal ID: ISSN 0014-3820
Publisher:
Wiley-Blackwell
Country of Publication:
United States
Language:
English

Citation Formats

Felsenstein, Joseph. EVOLUTIONARY TREES FROM GENE FREQUENCIES AND QUANTITATIVE CHARACTERS: FINDING MAXIMUM LIKELIHOOD ESTIMATES. United States: N. p., 2017. Web. doi:10.1111/j.1558-5646.1981.tb04991.x.
Felsenstein, Joseph. EVOLUTIONARY TREES FROM GENE FREQUENCIES AND QUANTITATIVE CHARACTERS: FINDING MAXIMUM LIKELIHOOD ESTIMATES. United States. doi:10.1111/j.1558-5646.1981.tb04991.x.
Felsenstein, Joseph. Wed . "EVOLUTIONARY TREES FROM GENE FREQUENCIES AND QUANTITATIVE CHARACTERS: FINDING MAXIMUM LIKELIHOOD ESTIMATES". United States. doi:10.1111/j.1558-5646.1981.tb04991.x.
@article{osti_1401556,
title = {EVOLUTIONARY TREES FROM GENE FREQUENCIES AND QUANTITATIVE CHARACTERS: FINDING MAXIMUM LIKELIHOOD ESTIMATES},
author = {Felsenstein, Joseph},
abstractNote = {},
doi = {10.1111/j.1558-5646.1981.tb04991.x},
journal = {Evolution},
number = 6,
volume = 35,
place = {United States},
year = {Wed May 31 00:00:00 EDT 2017},
month = {Wed May 31 00:00:00 EDT 2017}
}

Journal Article:
Free Publicly Available Full Text
This content will become publicly available on May 31, 2018
Publisher's Accepted Manuscript

Save / Share:
  • The assumptions involved in maximum likelihood estimation of evolutionary trees from quantitative character data are described. A strict maximum likelihood method applied to the case of two populations encounters singularities in the likelihood surface, and even when restrictions are placed on the parameters to avoid this the resulting estimate is converged to the wrong value as more characters are considered. These problems arise because new nuisance parameters are introduced every time a new character is added. If the data are assumed to consist only of the differences between population phenotypes, and a maximum likelihood solution based on this transformed datamore » is found, this restricted maximum likelihood (REML) method behaves well. Two computational techniques, the pruning algorithm and the pulley principle, are described which allow rapid computation of the restricted likelihood. They allow construction of an iterative procedure for finding the maximum of the restricted likelihood within a given tree topology. Combined with an algorithm for searching among similar tree topologies, this allows construction of a computer program to find the REML estimate of the tree. The fact that the REML estimate is a maximum likelihood estimate obtained from transformed data allows use of likelihood-ratio testing of evolutionary hypotheses. Constancy of evolutionary rate per unit time can be tested, but it appears that discrimination between gradualist and punctuated-equilibrium hypotheses will require fossil data.« less
  • It is shown that it is very much more difficult to show a detectably significant difference between gene frequencies than between the means of metric characters in comparisons of populations and species, when the loci influencing the trait are equally differentiated. This result is independent of gene frequencies or of the variances of the metric character. In this case, the number of loci segregating for the character must be less than the reciprocal of the heritability for there to be equal statistical power of detection. In contrast, when there is a large variance among loci in their differentiation between populations,more » it will be much more difficult to detect metric differences than gene frequency differences. Evolutionists are cautioned against drawing unwarranted inferences from the observation that populations or species that are highly differentiated in metric characters appear to be less so for gene freqencies, or vice versa.« less
  • The application of maximum likelihood techniques to the estimation of evolutionary trees from nucleic acid sequence data is discussed. A computationally feasible method for finding such maximum likelihood estimates is developed, and a computer program is available. This method has advantages over the traditional parsimony algorithms, which can give misleading results if rates of evolution differ in different lineages. It also allows the testing of hypotheses about the constancy of evolutionary rates by likelihood ratio tests, and gives rough indication of the error of the estimate of the tree.