TREE vol. 13, no. 9 September 1998 361 most of the growing season but, following certain environmental cues, amictic females produce mictic daughters. These mictic females produce haploid eggs meiotically that develop into either males or, if fertilized, resting eggs. Mictic females can only be fertilized within a few hours of birth 3 . Therefore, in Vollrath’s example a Brachionus rubens female does not fit Fisher’s theory of sex allocation in a simple way because she does not invest half of her reproductive resources into males and half into females. A rotifer female produces either male or female offspring, but not combinations of both. An amictic female produces only daughters, whereas a mictic female produces either males or resting eggs. As a result, the ratio of the investment in males and amictic females has nothing to do with sex allocation theory. Sex allocation should be judged on the basis of the ratio of male-producing mictic females to resting-egg-producing mictic females. An example of the appropriate application of sex-ratio theory to the sexual phase of the rotifer life cycle is presented by Aparici et al. 4 They showed that sex-ratio theory predicts that half of the mictic females are male producing and half resting-egg producing. The equal sex-allocation principle implies, in this case, that half of the mictic females have a female sexual role, receiving sperm and producing resting eggs, and half have a male role of producing sperm. Consequently, Fisher’s sex-ratio principle is applicable to rotifers, but its proper application requires a thorough understanding of the life cycle. So the question remains: why are male rotifers dwarf? At present we cannot provide a definitive answer, but there are several features of rotifers relevant to explaining this phenomenon. Rotifer males are not only dwarf, they are also haploid and have reduced morphology (e.g. they do not feed). Their development and maturation are fast, they swim quickly and they have a short life span. Furthermore, egg size is around 30% of the adult rotifer size, which suggests that the amount of resources a mother allocates to each egg is a substantial investment. Dwarf rotifer males might be, therefore, a consequence of selection on mothers to produce rapidly as many males as possible. Manuel Serra Dept of Microbiology and Ecology, University of Valencia, E46100-Burjassot, Spain (manuel.serra@uv.es) Terry W. Snell School of Biology, Georgia Institute of Technology, Atlanta, GA 30332, USA (terry.snell@biology.gatech.edu) References 1 Vollrath, F. (1998) Trends Ecol. Evol. 13, 159 –163 2 Nogrady, T., Wallace, R.L. and Snell, T.W. (1993) Rotifera, Volume 1: Biology, Ecology and Systematics, SPB Academic Publishing bv 3 Snell, T.W. and Childress, M.J. (1987) Int. J. Invertebr. Reprod. Dev. 12, 103 –110 4 Aparici, E., Carmona, M.J. and Serra, M. Am. Nat. (in press) T he scarcity of direct information about the history of life has long frustrated evolutionary biologists. To understand evolution, it is necessary to know not only the character states of living organisms, but also of their ancestors. Although the fossil record is replete with examples of evolutionary transformations, adequate fossils are not available for many taxa and character types. An increasingly popular alternative is to infer ancestral character states by mapping the character states of living organisms onto phylogenies using the method of maximum parsimony 1,2 . For example, if all members of a monophyletic group have hair, it is parsimonious to con- clude that their last common ancestor was also hairy. In more complex cases, an- cestral character states are reconstructed using parsimony algorithms 3,4 (Box 1). A decade ago, Coddington’s landmark paper 5 used ancestral character states as a basis for testing hypotheses about adap- tation in many cases, including the evolu- tion of spider webs. Coddington argued that many adaptationist hypotheses be- come meaningless without information about the order and timing of character state changes. Soon after, Donoghue 6 ar- gued, in a paper on seed plant evolution, that explicit character state reconstruc- tions are often the only source of infor- mation about important issues, such as the number of times a character state has arisen independently. Although Codding- ton and Donoghue made relatively conser- vative inferences, later studies proposed a plethora of evolutionary hypotheses to be tested using ancestral character state re- constructions, reviewed in Phylogeny, Ecol- ogy and Behavior 7 . This widely cited book – together with advances in parsimony algorithms 4 and the publication of user- friendly computer programs (e.g. MacClade 3.0; Ref. 2) – has established a central role for ancestral state reconstruction in mod- ern evolutionary biology. Parsimony reconstructions are appeal- ing and intuitively satisfying, and their authority is usually accentuated on the tree diagram by unambiguous bold lines and shadings. In large part, the recent explosion of interest in phylogenetic information has been driven by workers in many disci- plines who are eager to map their charac- ters of interest onto newly constructed phylogenies. Some especially innovative applications of ancestral state reconstruc- tion include exploring the catalytic prop- erties of ancestral proteins 8 and observing the response of living species to ancestral mating calls 9 . Copyright © 1998, Elsevier Science Ltd. All rights reserved. 0169-5347/98/$19.00 PII: S0169-5347(98)01382-2 PERSPECTIVES Reconstructing ancestral character states: a critical reappraisal Clifford W. Cunningham Kevin E. Omland Todd H. Oakley Using parsimony to reconstruct ancestral character states on a phylogenetic tree has become a popular method for testing ecological and evolutionary hypotheses. Despite its popularity, the assumptions and uncertainties of reconstructing the ancestral states of a single character have received less attention than the much less challenging endeavor of reconstructing phylogenetic trees from many characters. Recent research suggests that parsimony reconstructions are often sensitive to violations of the almost universal assumption of equal probabilities of gains and losses. In addition, maximum likelihood has been developed as an alternative to parsimony reconstruction, and has also revealed a surprising amount of uncertainty in ancestral reconstructions. Cliff Cunningham and Todd Oakley are members of the Evolution, Ecology and Organismal Biology Group and the Zoology Dept, Duke University, Durham, NC 27708-0325, USA (cliff@acpub.duke.edu; tho@acpub.duke.edu); Kevin Omland is at the Bell Museum of Natural History and Dept of Ecology, Evolution and Behavior, University of Minnesota St Paul, MN 55108, USA (komland@biosci.cbs.umn.edu).