Phylogenetic informativeness reconciles ray-finned fish molecular divergence times

Abstract

BackgroundDiscordance among individual molecular age estimates, or between molecular age estimates and the fossil record, is observed in many clades across the Tree of Life. This discordance is attributed to a variety of variables including calibration age uncertainty, calibration placement, nucleotide substitution rate heterogeneity, or the specified molecular clock model. However, the impact of changes in phylogenetic informativeness of individual genes over time on phylogenetic inferences is rarely analyzed. Using nuclear and mitochondrial sequence data for ray-finned fishes (Actinopterygii) as an example, we extend the utility of phylogenetic informativeness profiles to predict the time intervals when nucleotide substitution saturation results in discordance among molecular ages estimated.ResultsWe demonstrate that even with identical calibration regimes and molecular clock methods, mitochondrial based molecular age estimates are systematically older than those estimated from nuclear sequences. This discordance is most severe for highly nested nodes corresponding to more recent (i.e., Jurassic-Recent) divergences. By removing data deemed saturated, we reconcile the competing age estimates and highlight that the older mtDNA based ages were driven by nucleotide saturation.ConclusionsHomoplasious site patterns in a DNA sequence alignment can systematically bias molecular divergence time estimates. Our study demonstrates that PI profiles can provide a non-arbitrary criterion for data exclusion to mitigate the influence of homoplasy on time calibrated branch length estimates. Analyses of actinopterygian molecular clocks demonstrate that scrutiny of the time scale on which sequence data is informative is a fundamental, but generally overlooked, step in molecular divergence time estimation.

Biogeography and divergent patterns of body size disparification in North American minnows

“…This was a period of global cooling and lineage turnover when other North American fishes also began to radiate, supported by fossil evidence (Cavender, 1986(Cavender, , 1998Near and Koppelman, 2009) and divergence dating studies (Centrarchidae; Near et al, 2005Near et al, , 2011. This date is far younger than that inferred by a recent biogeographical reconstruction for the age of North American minnows (~68.2 Ma; Imoto et al, 2012), but that estimate was based on complete mitogenomic data, which is prone to saturation that may inflate divergence dates (Dornburg et al, 2014).…”

Towards a uniform nomenclature for ground squirrels: the status of the Holarctic chipmunks

“…Cytb has frequently been used in mammals to assess phylogenetic relationships above the family level (e.g., Agnarsson et al 2011) and has been shown to be surprisingly effective at these deeper nodes (Tobe et al 2010). As noted earlier, although saturation may introduce systemic bias in age estimates (Hugall et al 2007, Dornburg et al 2014, such bias should influence both the chipmunk and Holarctic ground squirrel clades equally as long as no calibrating fossils are incorporated a priori. At present, the best available dataset (Cytb) provides a compelling and unambiguous argument that the three chipmunk lineages exhibit genus-level genetic differentiation.…”

“…Systemic bias in molecular age estimates has been demonstrated in numerous studies (Ho and Jermiin 2004, Jansa et al 2006, Norris et al 2015. Deep branches may be underestimated relative to more recent branches (Ho and Larson 2006), especially in situations where fast-evolving genes have become saturated (Hugall et al 2007, Dornburg et al 2014). If such systemic bias is present, it may affect both chipmunks and Holarctic ground squirrels equally, but the bias would be corrected only for the Holarctic ground squirrels, thanks to the presence of a calibrating fossil.…”

Molecular phylogeny of bark and ambrosia beetles (Curculionidae: Scolytinae) based on 18 molecular markers

“…Rates are not always inversely correlated with phylogenetic resolution and clade support [80] and only the implementation in large taxonomic samples represents the ultimate test of a phylogenetic marker performance. Our gene classification based on phylogenetic utility that was assessed according to clade congruence and phylogenetic informativeness (PI) must therefore be taken as a preliminary proxy for a marker's phylogenetic signal [81,82,83]. It will be particularly interesting to observe the contribution of Iap2 in a larger data set given its much higher PI compared with other markers.…”