Abstract Speciation proceeds independently across lineages, generating varying levels of genetic divergence that are difficult to translate into discrete species boundaries. Widely used empirical heuristics such as genetic distance thresholds, DNA barcoding gaps, and measures of gene tree exclusivity provide practical tools for species delimitation, but their reliance on arbitrary cutoffs can strongly bias the number of delimited species. Here, we evaluate a coalescent-based alternative, the genealogical divergence index (gdi), which uses Bayesian estimates of divergence times, population sizes, and migration rates to explicitly quantify genealogical independence along the speciation continuum. We explore the utility of gdi by applying it to genomic data from 76 species and 113 populations of North American lizards. We test whether gdi reliably distinguishes populations from recognized species, if a biologically meaningful tipping point exists, and how empirical patterns align with commonly used species delimitation thresholds. We show that while gdi captures key features of the population-species continuum, it inconsistently separates populations from species. At the recommended species-level threshold of gdi = 0.7, the predicted probability of species status is approximately 0.5 in our test data. Among the population parameters examined, unscaled divergence (measured in substitutions per site) is the strongest predictor of species status, followed by migration rate. By conducting a quantitative review of the literature, we show that the pattern of continuous gdi variation across populations and species holds true in a more diverse set of taxa. Our results highlight the importance of considering gdi as part of a multi-evidence and reference-based framework for species delimitation, rather than a binary decision metric.
Phylogenetic analyses often require the summarization of multiple trees, e.g., in Bayesian analyses to obtain the centroid of the posterior distribution of trees, or to determine the consensus of a set of bootstrap trees. The majority-rule consensus tree is the most commonly used. It is easy to compute and minimizes the sum of Robinson-Foulds (RF) distances to the input trees. In mathematical terms, the majority-rule consensus tree is the median of the input trees with respect to the RF distance. However, due to the coarse nature of RF distance, which only considers whether two branches induce exactly the same bipartition of the taxa or not, highly unresolved trees can be produced when the phylogenetic signal is low. To overcome this limitation, we propose using median trees with respect to finer-grained dissimilarity measures between trees. These measures include a quartet distance between tree topologies, and transfer distances, which quantify the similarity between bipartitions, in contrast to the presence/absence view of RF. We describe fast heuristic consensus algorithms for transfer-based tree dissimilarities, capable of efficiently processing trees with thousands of taxa. Through evaluations on simulated datasets in both Bayesian and bootstrapping maximum-likelihood frameworks, our results show that our methods improve consensus tree resolution in scenarios with low to moderate phylogenetic signal, while providing better or comparable dissimilarities to the true phylogeny. Applying our methods to mammals, Ebola virus, and a large HIV dataset of over nine thousand taxa confirms the improvement with real data. These results demonstrate the usefulness of our new consensus tree methods for analyzing the large datasets that are available today. Our software, PhyloCRISP, is available from https://github.com/yukiregista/PhyloCRISP. ### Competing Interest Statement The authors have declared no competing interest. JSPS KAKENHI, 25K24364, 22KJ1131, JPMJFR2135 Paris Artificial Intelligence Research Institute (PRAIRIE), ANR-19-P3IA-0001
Species identification in taxonomically complex plant groups is frequently limited by the inadequacy of organellar markers, whose phylogenetic signal is disrupted by cytonuclear discordance and chloroplast capture. Using the taxonomically complex genus Populus as a model, we evaluated an assembly-free k-mer workflow against a curated SNP reference benchmark. Whole-genome resequencing data from 235 Populus individuals were curated to a 202-individual, 34-species reference dataset in which all retained species are strictly monophyletic in a genome-wide SNP analysis. Independent maximum likelihood analyses further confirmed that the 31 non-hybrid backbone species each maintained high-support monophyly, while taxa of documented reticulate origin showed placement patterns consistent with their reticulate histories. ABBA-BABA D-statistics detected widespread residual allele sharing within the backbone, though the strongest signals did not correspond to the species pairs responsible for the few k-mer identification failures. Against this benchmark, complete plastomes showed limited resolution, recovering only 3.0% species monophyly and 71.1% nearest-neighbor assignment. The optimized k-mer workflow, operating directly on raw reads without assembly or alignment, recovered 91.2% species monophyly, 99.0% nearest-neighbor assignment, and 98.0% group-average assignment. K-mer length was the primary accuracy-controlling parameter, with k = 31 falling within a stable accuracy plateau. Distance-based metrics reached near-saturation at 0.2× sequencing depth, indicating that low-coverage genome skimming can support scalable nuclear genome-based identification with standard computational resources. K-mer distance heatmaps also flagged unusual genomic affinities in hybrid-origin and outlier samples, providing a rapid screen for subsequent population genomic analyses. These results support assembly-free k-mer distances as an efficient tool for reference-based species identification and sample screening in complex plant groups, with residual limitations concentrated near recently diverged species boundaries. Model-based phylogenomic, coalescent, and network analyses remain necessary for resolving deeper species relationships and detailed introgression histories.
Dietary adaptations are fundamental drivers of evolutionary innovations. Still, the morphological and molecular mechanisms underlying these transitions remain poorly understood. True bugs (Heteroptera), one of the most diverse clades of hemimetabolous insects, have evolved a vast array of trophic niches. They exhibit remarkable structural and functional diversity in their salivary glands, providing an ideal system for investigating how dietary variation shapes evolutionary trajectories. Using morphometric, proteo-transcriptomic, and comparative phylogenetic approaches, we examined the evolutionary patterns of salivary glands across 220 hemipteran species, encompassing all seven infraorders and six dietary categories of Heteroptera. Our results reveal strong correlations between dietary shifts, salivary gland size and allometric scaling. Ancestral state estimation analysis demonstrates multiple evolutionary transitions in salivary gland morphology, highlighting structural adaptations associated with different diets. Furthermore, we found that distantly related species with similar diets share dominant salivary protein groups, implying convergent molecular solutions to shared feeding challenges. Notably, we identify venom protein family 2 as a key adaptation. This family originated through lineage-specific co-option of an ancestral protein into heteropteran salivary systems, followed by extensive expansion that facilitated dietary diversification. These findings provide novel insights into the morphological, physiological, and molecular adaptations of salivary glands, emphasizing the dynamic interplay between feeding ecology and macroevolution.
The eukaryotic genome has been described as a collection of different phylogenetic histories. For most phylogenomic analyses the primary goal is to identify the species tree, the singular history that underlies and shapes the "gene trees" of individual loci. Discordance among gene trees and with the species tree is expected due to deep coalescence/lineage sorting, while also resulting from various technical causes (e.g., long branch attraction, pseudo-orthology), or, of greater interest, by introgression and horizontal transfer. Where do competing phylogenetic signals reside in gene tree topology space-that part of tree space occupied by the gene trees reconstructed for a particular dataset of taxa and genetic loci? We explored this question in the small (~30 species) leguminous plant genus, Glycine, which has extensive genomic resources due to the inclusion of cultivated soybean (G. max). Glycine genomes are highly duplicated due to relatively recent (~10 million years) ancestral polyploidy and have extensive nuclear-cytoplasmic discordance. We explored Glycine gene tree topology space using a set of 2389 nuclear genes and 61 representative accessions selected from a 570-taxon x 100 gene concatenation supergene tree, reconstructing gene trees for all nuclear loci and from complete plastid genomes and partial mitochondrial genomes. Species trees (ASTRAL) and maximum likelihood (ML) concatenation trees were congruent with one another but were discordant with organellar genome trees, which were incongruent with one another. Individual loci all had unique topologies for the 61-taxon dataset and for a reduced dataset of 27 taxa. No locus tracked either the species tree or the plastome topology in the resulting "flat" gene tree topology space of either dataset, nor did clustering identify any regional differentiation of gene tree topology space populated by loci with similar topologies. Only when the dataset was reduced to six Glycine species, chosen because they have complete genome sequences, and an outgroup was a topological landscape produced in which most loci tracked the species tree topology, with secondary peaks that included, most prominently, the discordant plastome topology. There was no evidence of pseudo-orthology in this landscape, and synteny-based assessment of thousands of loci across these six genomes identified few candidate pseudo-orthologs. Thus, while it is true that the Glycine genome is indeed a collection of different historical signals, those signals are complex and exist at the level of clades within trees rather than as entire gene trees. Although phylogenomic methods can reconstruct the species tree from signals scattered among many loci, even loci with very low resolution, other biologically relevant signals are much more difficult to localize without an explicit starting hypothesis.
