When do species-tree and concatenated estimates disagree? An empirical analysis with higher-level scincid lizard phylogeny

Shea M. Lambert, Tod W. Reeder, John J Wiens

Research output: Contribution to journalArticle

39 Citations (Scopus)

Abstract

Simulation studies suggest that coalescent-based species-tree methods are generally more accurate than concatenated analyses. However, these species-tree methods remain impractical for many large datasets. Thus, a critical but unresolved issue is when and why concatenated and coalescent species-tree estimates will differ. We predict such differences for branches in concatenated trees that are short, weakly supported, and have conflicting gene trees. We test these predictions in Scincidae, the largest lizard family, with data from 10 nuclear genes for 17 ingroup taxa and 44 genes for 12 taxa. We support our initial predictions, and. suggest that simply considering uncertainty in concatenated trees may sometimes encompass the differences between these methods. We also found that relaxed-clock concatenated trees can be surprisingly similar to the species-tree estimate. Remarkably, the coalescent species-tree estimates had slightly lower support values when based on many more genes (44 vs. 10) and a small (~30%) reduction in taxon sampling. Thus, taxon sampling may be more important than gene sampling when applying species-tree methods to deep phylogenetic questions. Finally, our coalescent species-tree estimates tentatively support division of Scincidae into three monophyletic subfamilies, a result otherwise found only in concatenated analyses with extensive species sampling.

Original languageEnglish (US)
Pages (from-to)146-155
Number of pages10
JournalMolecular Phylogenetics and Evolution
Volume82
Issue numberPA
DOIs
StatePublished - Jan 1 2015

Fingerprint

Lizards
Phylogeny
empirical analysis
lizard
lizards
phylogeny
Genes
gene
Scincidae
sampling
genes
Uncertainty
prediction
methodology
uncertainty
phylogenetics

Keywords

  • Concatenated analysis
  • Phylogenetic methods
  • Reptile
  • Scincidae
  • Species-tree analysis
  • Taxon sampling

ASJC Scopus subject areas

  • Ecology, Evolution, Behavior and Systematics
  • Molecular Biology
  • Genetics
  • Medicine(all)

Cite this

When do species-tree and concatenated estimates disagree? An empirical analysis with higher-level scincid lizard phylogeny. / Lambert, Shea M.; Reeder, Tod W.; Wiens, John J.

In: Molecular Phylogenetics and Evolution, Vol. 82, No. PA, 01.01.2015, p. 146-155.

Research output: Contribution to journalArticle

@article{a0cd7645c2154248b88fb807d4ca904c,
title = "When do species-tree and concatenated estimates disagree? An empirical analysis with higher-level scincid lizard phylogeny",
abstract = "Simulation studies suggest that coalescent-based species-tree methods are generally more accurate than concatenated analyses. However, these species-tree methods remain impractical for many large datasets. Thus, a critical but unresolved issue is when and why concatenated and coalescent species-tree estimates will differ. We predict such differences for branches in concatenated trees that are short, weakly supported, and have conflicting gene trees. We test these predictions in Scincidae, the largest lizard family, with data from 10 nuclear genes for 17 ingroup taxa and 44 genes for 12 taxa. We support our initial predictions, and. suggest that simply considering uncertainty in concatenated trees may sometimes encompass the differences between these methods. We also found that relaxed-clock concatenated trees can be surprisingly similar to the species-tree estimate. Remarkably, the coalescent species-tree estimates had slightly lower support values when based on many more genes (44 vs. 10) and a small (~30{\%}) reduction in taxon sampling. Thus, taxon sampling may be more important than gene sampling when applying species-tree methods to deep phylogenetic questions. Finally, our coalescent species-tree estimates tentatively support division of Scincidae into three monophyletic subfamilies, a result otherwise found only in concatenated analyses with extensive species sampling.",
keywords = "Concatenated analysis, Phylogenetic methods, Reptile, Scincidae, Species-tree analysis, Taxon sampling",
author = "Lambert, {Shea M.} and Reeder, {Tod W.} and Wiens, {John J}",
year = "2015",
month = "1",
day = "1",
doi = "10.1016/j.ympev.2014.10.004",
language = "English (US)",
volume = "82",
pages = "146--155",
journal = "Molecular Phylogenetics and Evolution",
issn = "1055-7903",
publisher = "Academic Press Inc.",
number = "PA",

}

TY - JOUR

T1 - When do species-tree and concatenated estimates disagree? An empirical analysis with higher-level scincid lizard phylogeny

AU - Lambert, Shea M.

AU - Reeder, Tod W.

AU - Wiens, John J

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Simulation studies suggest that coalescent-based species-tree methods are generally more accurate than concatenated analyses. However, these species-tree methods remain impractical for many large datasets. Thus, a critical but unresolved issue is when and why concatenated and coalescent species-tree estimates will differ. We predict such differences for branches in concatenated trees that are short, weakly supported, and have conflicting gene trees. We test these predictions in Scincidae, the largest lizard family, with data from 10 nuclear genes for 17 ingroup taxa and 44 genes for 12 taxa. We support our initial predictions, and. suggest that simply considering uncertainty in concatenated trees may sometimes encompass the differences between these methods. We also found that relaxed-clock concatenated trees can be surprisingly similar to the species-tree estimate. Remarkably, the coalescent species-tree estimates had slightly lower support values when based on many more genes (44 vs. 10) and a small (~30%) reduction in taxon sampling. Thus, taxon sampling may be more important than gene sampling when applying species-tree methods to deep phylogenetic questions. Finally, our coalescent species-tree estimates tentatively support division of Scincidae into three monophyletic subfamilies, a result otherwise found only in concatenated analyses with extensive species sampling.

AB - Simulation studies suggest that coalescent-based species-tree methods are generally more accurate than concatenated analyses. However, these species-tree methods remain impractical for many large datasets. Thus, a critical but unresolved issue is when and why concatenated and coalescent species-tree estimates will differ. We predict such differences for branches in concatenated trees that are short, weakly supported, and have conflicting gene trees. We test these predictions in Scincidae, the largest lizard family, with data from 10 nuclear genes for 17 ingroup taxa and 44 genes for 12 taxa. We support our initial predictions, and. suggest that simply considering uncertainty in concatenated trees may sometimes encompass the differences between these methods. We also found that relaxed-clock concatenated trees can be surprisingly similar to the species-tree estimate. Remarkably, the coalescent species-tree estimates had slightly lower support values when based on many more genes (44 vs. 10) and a small (~30%) reduction in taxon sampling. Thus, taxon sampling may be more important than gene sampling when applying species-tree methods to deep phylogenetic questions. Finally, our coalescent species-tree estimates tentatively support division of Scincidae into three monophyletic subfamilies, a result otherwise found only in concatenated analyses with extensive species sampling.

KW - Concatenated analysis

KW - Phylogenetic methods

KW - Reptile

KW - Scincidae

KW - Species-tree analysis

KW - Taxon sampling

UR - http://www.scopus.com/inward/record.url?scp=84909979108&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84909979108&partnerID=8YFLogxK

U2 - 10.1016/j.ympev.2014.10.004

DO - 10.1016/j.ympev.2014.10.004

M3 - Article

C2 - 25315885

AN - SCOPUS:84909979108

VL - 82

SP - 146

EP - 155

JO - Molecular Phylogenetics and Evolution

JF - Molecular Phylogenetics and Evolution

SN - 1055-7903

IS - PA

ER -