hemiplasy
Hahn MW & Nakhleh L 2016 Irrational exuberance for resolved species trees. Evolution 70:7-17.
- the mapping of traits onto trees generally uses only a single representation of the species tree, ignoring variation in the gene trees used to construct it
- standard methods for character mapping will incorrectly infer the number of times a trait has evolved
- this phenomenon, dubbed "hemiplasy," poses many problems in analyses of character evolution
- in mapping all sorts of characters—behaviors, morphologies, nucleotide substitutions—onto a single tree, we are in effect making the assumption that the constituent gene-trees match the species tree
- when they do not, we are implicitly forcing the gene trees to fit into the species tree, and consequently are forcing character-state transition to occur on the species tree
- the incorrect inferences often include finding convergent evolution when none has occurred
- we use hemiplasy to mean any incorrect inference about character-state evolution caused by gene-tree discordance, regardless of the cause of discordance
- the purpose of this expanded definition is to stress that there are multiple processes that lead to the incorrect inference that alleles identical-by-descent in different species have multiple origins, and that all of these processes lead to similar problems
- the use of a single tree for carrying out comparative analyses will lead to incorrect inferences about character evolution because this fixed-tree representation is only an average of all the relationships for each gene and character
- high bootstrap support is often used to justify the statement that species trees are well resolved
- bootstrap values are almost completely uninformative when dealing with genome-scale data, especially about the presence of discordance in a dataset
- even with a small proportion of discordant trees, hemiplasy may be the best explanation for paraphyletic distributions of character states
- most forms of selection will actually act to increase the fraction of concordant trees by reducing the time to coalescence
- a species tree can be inferred and ancestral states reconstructed
- it seems highly likely that many traits will not follow this tree
- an alternative approach to understanding trait evolution will be needed
- under hemiplasy the nucleotide changes will be the same in all lineages (because there is truly only one substitution)
- under convergence the changes can be the same or different
- if the same functional alleles are found together with the same nonfunctional (e.g., synonymous) alleles in separate lineages, the results are likely due to hemiplasy and not selection-driven convergence
- the problem of hemiplasy raises multiple issues for the use of comparative methods to understand the evolution of traits on trees
- a major goal of recent phylogenomic studies specifically appears to be the resolution of tangled relationships
- we may need to reconsider this goal
- or at least to always remember that not all resolved relationships are without incongruence