📄 evolutionary biology

Ancient Rapid Radiation Underlies Persistent Phylogenomic Conflict in Early Collembola Diversification

Through a comprehensive phylogenomic analysis of 145 springtail taxa using 1,127 single-copy orthologues, this study resolves the long-standing conflict in Collembola phylogeny by identifying an ancient rapid radiation in the Early Devonian that resulted in persistent phylogenetic discordance, ultimately favoring Poduromorpha as the earliest-diverging lineage among the four extant orders.

Original authors: Cucini, C., Moody, E. R., Cicconardi, F., Montgomery, S. H.

Published 2026-07-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Cucini, C., Moody, E. R., Cicconardi, F., Montgomery, S. H.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine a family reunion for the world's oldest, tiniest soil-dwelling arthropods: the springtails (Collembola). These little critters have been hopping around since the Early Devonian, roughly 407 million years ago, making them some of the very first animals to call land home. But here's the mystery: despite having nearly 9,000 described species (and maybe up to 50,000 in total!), scientists have been stuck in a heated argument for over 20 years about who is related to whom among their four main "orders" (the big family groups). It's like trying to figure out the family tree of a massive, ancient clan where everyone looks a bit different, but the old photo albums are blurry and contradictory.

The Great Detective Work
To solve this, the researchers in this paper acted like digital detectives. They gathered a massive dataset, pulling genetic information from 145 different springtail species (covering 19 families) and looking at 1,127 specific genes. That's the biggest genetic "family photo" ever assembled for these creatures.

But there was a catch. Because they were using fragmented genome data (like trying to solve a puzzle with some pieces missing or broken), they had to build a special filter. Think of it as a high-tech sieve that sifts out "fake cousins" (hidden paralogues) that look like family members but aren't. This new filter cleaned up the data significantly, removing about 31% of the noisy gene matches down to just 17%, making the signal much clearer.

The Big Reveal (and the Big Confusion)
When they ran the numbers, the results were a bit of a rollercoaster.

The "Poduromorpha-First" Theory: When they looked at the amino acid data (the building blocks of proteins), the strongest signal suggested that the Poduromorpha order was the very first to split off from the rest of the family tree. This is a new idea that hadn't been the favorite before, but it kept popping up as the most likely scenario, especially when using complex models that account for how genes change over deep time.
The "Neelipleona-First" Theory: However, when they looked at the raw DNA (nucleotide) data, the story changed. That data consistently pointed to Neelipleona being the first to split off.
The "It's a Mess" Reality: When they used methods that look at how individual genes tell different stories (coalescent methods), the results were all over the place. Some genes said one thing, others said another. The internal branches of the family tree were incredibly short—so short they were almost zero. This suggests that the four main groups didn't split one by one over millions of years; instead, they likely exploded onto the scene in a rapid burst during the Carboniferous period, all within a very short window of time.

What the Paper Rules Out
The researchers didn't just guess; they tested every single possible way the four groups could be related (15 different topologies). They explicitly ruled out the majority of these 15 ideas. Only two specific arrangements (T11 and T4) remained statistically indistinguishable from each other, meaning the data is too fuzzy to say for sure which of those two is the exact truth.

They also ruled out the idea that this confusion is just because they didn't have enough data or enough species. They had the biggest dataset ever, and the confusion persisted. They also ruled out the idea that the weird results were just a simple mistake caused by "long branches" (where fast-evolving species trick the computer). Even when they removed the tricky species or changed how they analyzed the data, the conflict remained.

How Sure Are They?
The paper is very confident about a few things, but cautious about others:

Proven: They are sure that the four main orders are real, distinct groups. They are sure that the confusion is caused by a rapid ancient radiation (a "big bang" of evolution) that happened so fast that the genetic signals got scrambled.
Suggested: They suggest that Poduromorpha is the most likely candidate for the first branch, but they admit that Neelipleona is still a strong contender depending on how you analyze the DNA. They cannot prove which one is definitively first yet.
Measured: They measured the age of the springtail family tree. The main group (crown Collembola) likely started around 407 million years ago (Early Devonian). The four main orders then diversified during the Carboniferous period. Some specific genera (like Folsomia and Isotoma) are surprisingly old, with origins dating back to the Late Cretaceous (over 75 million years ago), meaning some of these "genera" are actually older than many entire families of other animals.

The Bottom Line
This paper doesn't give us a single, perfect family tree with a clear "This is the oldest branch!" label. Instead, it tells us that the springtail family tree is a "soft polytomy"—a moment in deep time where four lines of evolution burst forth almost simultaneously. It's like a fireworks display where four sparks fly off at once; trying to say which one left the fuse first is nearly impossible because they happened too close together.

The study concludes that this confusion isn't a failure of science; it's a biological signature. The fact that we can't resolve the tree perfectly is actually proof that these ancient creatures diversified incredibly fast, leaving a chaotic but fascinating genetic echo that has lasted for over 350 million years. The mystery isn't solved, but the nature of the mystery has finally been understood.

Technical Summary: Ancient Rapid Radiation Underlies Persistent Phylogenomic Conflict in Early Collembola Diversification

Problem Statement
Collembola (springtails) represent one of the oldest extant terrestrial hexapod lineages, with a fossil record extending to the Early Devonian (~400 Ma). Despite their ecological significance and abundance, the phylogenetic relationships among their four extant orders—Entomobryomorpha, Poduromorpha, Symphypleona, and Neelipleona—have remained unresolved for over two decades. Previous studies utilizing morphological data, ribosomal markers, multilocus datasets, mitogenomes, and early phylogenomic approaches have produced conflicting hypotheses regarding inter-order branching orders. The authors posit that this persistent conflict may not stem from insufficient data or taxon sampling, but rather from an ancient rapid radiation that left a complex, conflicting signal across the genome, potentially exacerbated by hidden paralogy in fragmented genomic assemblies.

Methodology
The study presents the most comprehensive phylogenomic analysis of Collembola to date, utilizing a dataset of 1,127 single-copy orthologues sampled from 145 taxa representing all four orders and 19 families.

Data Curation and Filtering: To address the issue of hidden paralogy in BUSCO-derived datasets from fragmented assemblies, the authors developed a novel HMM-based filtering pipeline. This pipeline uses BUSCO profile Hidden Markov Models (HMMs) to screen sequences, identifying bimodal distributions in conditional E-values to distinguish between orthologs and divergent paralogs. Sequences were split into independent partitions (A and B) where supported, and outliers were removed using interquartile-range (IQR) criteria.
Phylogenomic Inference: Analyses were conducted using both concatenation and coalescent-based approaches.
- Concatenation: Maximum-likelihood (ML) trees were inferred using IQ-TREE2 on nucleotide and amino-acid supermatrices. Sensitivity analyses included gene-jackknife replicates, occupancy filtering (50%, 70%, 75%), and amino-acid recoding (Dayhoff-6, SR4) to mitigate compositional heterogeneity. Complex site-heterogeneous models (e.g., LG+C60+F+Γ) were applied to reduced datasets.
- Coalescent: Species trees were inferred using ASTRAL and CASTER-SITE to account for incomplete lineage sorting (ILS). Gene concordance factors (gCF) and site concordance factors (sCF) were calculated to quantify conflict.
Topology Testing: The authors exhaustively tested all 15 possible fully resolved inter-order topologies using likelihood-based tests (AU, KH, SH, ELW, bp-RELL) on both concatenated and gene-wise datasets.
Signal Partitioning: Loci were ranked by their ability to discriminate between topologies ( $\Delta$ logL) to determine if phylogenetic signal was concentrated in a few genes or distributed across the genome.
Divergence Dating: Molecular dating was performed using MCMCTREE under a relaxed clock model. A conservative calibration strategy was employed, using Rhyniella praecursor as a minimum age for crown Collembola without assigning it to specific extant orders, to avoid over-constraining estimates based on uncertain fossil placements.

Key Results

Data Quality: The HMM-based filtering pipeline significantly reduced hidden paralogy, decreasing the median proportion of removed species per locus from 31.3% to 17.3% and improving orthology assignment consistency.
Phylogenetic Topology:
- Concatenation: Amino-acid analyses consistently recovered Poduromorpha as the earliest-diverging lineage, followed by a split separating Entomobryomorpha from a clade containing Symphypleona and Neelipleona (Topology T11). Nucleotide analyses favored a Neelipleona-first arrangement (Topology T4).
- Coalescent: Coalescent methods (ASTRAL, CASTER-SITE) recovered various conflicting backbone topologies depending on data type and occupancy, characterized by extremely short internal branches (0–0.05 coalescent units) and low quartet support.
- Topology Testing: Exhaustive testing of 15 hypotheses reduced the viable candidates to two statistically indistinguishable topologies (T11 and T4). T11 (Poduromorpha-first) was consistently favored across site-homogeneous and site-heterogeneous models in amino-acid datasets.
Nature of Conflict: The conflict is not random. While individual loci often lack sufficient signal to resolve deep nodes, the cumulative signal across hundreds of loci progressively concentrates support on a limited set of hypotheses. The pattern of low quartet support combined with relatively high coalescent units in Poduromorpha suggests a "soft polytomy" resulting from rapid diversification and incomplete lineage sorting, rather than a hard polytomy or simple analytical artifact.
Divergence Times: Crown Collembola originated in the Early Devonian (~~407 Ma), with the diversification of the four extant orders occurring during the Carboniferous (~~346–300 Ma). Symphypleona diversified later, in the Late Triassic (~221 Ma).
Taxonomic Persistence: Many extant genera (e.g., Folsomia, Isotoma) are estimated to be older than many currently recognized families, highlighting exceptional lineage longevity.

Significance and Claims
The paper claims to provide a robust, albeit complex, phylogenomic framework for Collembola that resolves the "unresolved" nature of their deep phylogeny not by finding a single definitive tree, but by characterizing the nature of the conflict. The authors argue that the persistent phylogenetic conflict is a biological signature of an ancient rapid radiation that occurred over a geologically brief interval, leading to pervasive incomplete lineage sorting and reticulate evolutionary history.

Key contributions include:

Methodological Innovation: The development of an HMM-based filtering pipeline that effectively mitigates hidden paralogy in BUSCO datasets derived from fragmented genomes.
Resolution of Uncertainty: Narrowing the uncertainty of Collembola inter-order relationships to two primary hypotheses, with strong evidence favoring Poduromorpha as the earliest-diverging lineage, consistent with some morphological hypotheses but distinct from previous molecular studies.
Evolutionary Insight: Demonstrating that deep phylogenetic conflict in ancient radiations can persist despite genome-scale data, and that this conflict reflects genuine biological history (rapid radiation) rather than merely methodological limitations.
Temporal Context: Establishing that the diversification of springtails coincided with the early assembly of terrestrial soil ecosystems in the Devonian and Carboniferous, reinforcing their role as one of the oldest terrestrial arthropod radiations.

The study concludes that resolving the remaining uncertainty in Collembola phylogeny will require not just larger datasets, but continued advances in evolutionary models capable of extracting signal from complex, ancient radiations.

More like this