Data from: Surviving ancestors, hard polytomies, and seed plant evolution
Data files
Jun 08, 2026 version files 23.78 MB
-
KlyEmSok.mean.dist
70.44 KB
-
KlyEmSok.mean.dst
21.49 KB
-
KlyEmSok.MLBS.CNet.15plus.nxs
9.38 MB
-
KlyEmSok.MLvsLS.nexml
88.16 KB
-
KlyEmSok.NJBS.CNet.15plus.nxs
8.35 MB
-
KlyEmSok.PBS.CNet.15plus.nxs
5.25 MB
-
README.md
2.97 KB
-
Romerogram.svg
607.17 KB
Abstract
Forty years of phylogenetic analysis have failed to resolve the relationships among the major groups of seed plants. From the fossil record it is clear that more seed plant groups have gone extinct than are extant, though some direct ancestors of today's clades could have survived. Here we reanalyze morphological matrices relevant to seed plants, applying two underutilized visualization tools: Networks for displaying incompatible phylogenetic signal and Romerograms (spindle diagrams) for depicting alternative relationships, suspected budding speciation, and surviving ancestors. Surviving ancestors give rise to hard polytomies, and networks may help pinpoint them as suggested by experiments in which we added simulated ancestors to our matrix. Romerograms — whose reproducible construction we illustrate — offer a heuristic means of representing phylogenetic hypotheses, and we argue that the field's near-exclusive focus on identifying sister groups and representing them in cladograms may have driven phylogenetics into an impasse. Identifying true evolutionary polytomies warrants further methodological research, but broadening the repertoire of data visualization approaches may help overcome the field's entrenched bias toward strictly dichotomous trees.
Dataset DOI: 10.5061/dryad.5qfttdzn2
Description of the data and file structure
This repository includes an editable version of our Romerogram (scalable vector graphic, Romerogram.svg) and selected result files in their native formats. The set of analyses include phylogenetic tree inferences and branch support analyses (nonparametric boostrapping, BS) under three optimality criteria; the tip set comprises extant and extinct seed plant taxa as well as five artificial taxa used in further experiments (EAP = "early angiosperm precursor"; RAFCA = "random angiosperm first common ancestor"; ALCA = "angiosperm last common ancestor"; GFCA and GLCA = "Gnetales first/last common ancestor").
Disclaimer: The underlying matrix is an emended (Amborella added) and curated version of the matrix used by Klymiuk et al. (2022, Am. J. Bot. 199: 966-985, appendix S1) originally published under a CC-BY licence, thus, not included in this repository but elsewhere (linked figshare repository).
Explanation of phylogenetic result files
KlyEmSok.mean.dist -- Neighbor-Net based on mean morphological pairwise (Hamming) distances inferred with SplitsTree. Phangorn library is required for import and visualization in R (Schliep et al., 2017, Methods Ecol. Evol. 8: 1212–1220)
KlyEmSok.mean.dst -- Mean morphological pairwise (Hamming) distance matrix inferred from the character matrix with PAUP* in PHYLIP (.dst) format (input used for SplitsTree4)
KlyEmSok.MLvsLS.nexml -- Tanglegram of the ML tree, inferred with RAxML, with the Neighbor-Joining (NJ) tree fulfilling the least-squares (LS) optimality criterion, inferred with BioNJ algorithm implemented in PAUP*, in NEXML format. Recommended software to open and view: Dendroscope3
.nxs files are 'bootstrap support networks' (BSN; Schliep et al., 2017), filtered consensus networks based on BS pseudoreplicate tree samples, generated with SplitsTree. Edge lengths are proportional to the number of BS pseudoreplicate trees showing the according taxon bipartition (branch; option "count"); taxon bipartitions found in less than 15% of the BS pseudoreplicate trees discarded ("cutoff" = 15).
KlyEmSok.MLBS.CNet.15plus.nxs -- BSN based on 10,000 BS pseudoreplicates inferred with RAxML during ML tree inference.
KlyEmSok.NJBS.CNet.15plus.nxs -- BSN based on 10,000 BS pseudoreplicate NJ/LS trees inferred with PAUP* (cf. PAUP block in KlymiukEmendSokoloff.simple.nex)
KlyEmSok.PBS.CNet.15plus.nxs -- BSN based on 10,000 BS pseudoreplicate inferred with PAUP* under parsimony (cf. PAUP block in KlymiukEmendSokoloff.simple.nex)
