Notes on Hart & Grosberg (2009): Caterpillars did not evolve from onychophorans by hybridogenesis

Context and Aim

• Article by Hart MW and Grosberg RK (PNAS, 2009) responding to Williamson (2009) claim that caterpillar larvae evolved from onychophorans via hybridization and that “larval transfer” leaves detectable genomic signatures.
• Purpose: use existing literature data to test Williamson’s predictions about genomes (size, content, phylogeny) and to argue against hybridization/larval transfer between velvet worms (Onychophora) and insects.
• Core claim: hybridogenesis between distantly related animals does not explain patterns of larval evolution; genomic data strongly contradict the proposed onychophoran–insect larval transfer.
• Key topics addressed: genome size (C value), genome content (gene counts), phylogenetic relationships, and empirical tests of hybridization scenarios.

Williamson’s hypothesis and testable predictions

• Williamson’s larval transfer hypothesis: basic larval types originated as adults of different lineages; larvae were transferred to other lineages via hybridization, yielding insect caterpillars and related larval forms.
• Predicted genomic patterns from his hypothesis:
  • Rhizocephala (parasitic barnacles) would show mixed genomes (three genomes: one adult-like and two larval-like) if they were not true arthropods but acquired arthropod larvae by hybridization.
  • Insects with caterpillar-like larvae (holometabolans) would show genomic signatures reflecting donor onychophoran genomes (e.g., larger base-pair content) relative to non-larval insects.
  • Onychophoran genomes would be smaller than holometabolous insect genomes if they were donors of larval traits.
  • In tunicates/larvaceans (urochordates), predictions would mirror the idea that larval forms could be transferred across distant groups via hybridization.
  • A comprehensive genome-wide test would reveal close phylogenetic signals between onychophoran genes and holometabolous insect genes (orthologue transfer), inconsistent with a single, clean arthropod lineage signal.
• Hart & Grosberg emphasize that many of Williamson’s testable corollaries are already addressed in the literature and that data widely contradict them.

Rhizocephalans Have Small Genomes and Did Not Acquire Larvae by Hybridization with Other Crustaceans

• What Rhizocephala are:
  • A clade of highly modified barnacles that are endoparasites of crustaceans; adults lack typical barnacle features (digestive tract, segmentation, exoskeleton plates).
  • Their larval stage (cypris) resembles cyprid larvae of other Cirripedia and initiates infection by recruiting to hosts.
• Williamson’s alternate hypothesis for Rhizocephala: they acquired a barnacle larval form by hybrid transfer, implying mixed, unusually large genomes with separate donor larval genomes.
• Molecular phylogenetic evidence:
  • Rhizocephalans are a clade within barnacles; phylogenies based on genes expressed in both larval and adult stages place Rhizocephala with Thoracica (acorn and stalked barnacles).
  • The sister group to Rhizocephala is Acrothoracica, not a genome with separate larval recipient lineages.
• Genome size evidence contradicts hybrid-genome expectations:
  • Feulgen/flow cytometry data across many animals provide genome size (C value) estimates. For sacculinid rhizocephalans, C ≈ $0.67 ext{ pg}$, which is smaller than most barnacles (range $0.74–2.60 ext{ pg}$) and much smaller than would be expected if an “adult genome plus two independent larval genomes” scenario existed.
• Conclusion for this section:
  • The Rhizocephala are true barnacles with small genomes, not a mosaic of three independent crustacean genomes. This directly contradicts Williamson’s prediction that their genomes should reflect a hybrid acquisition of larval genomes.

Holometabolous Insects Have Small Genomes Relative to Insects Without Larvae

• Context: major insect lineages vary in life history complexity; Williamson argued that exopterygote insects lacking larvae would have smaller genomes than holometabolous insects with larvae.
• What the data show:
  • Genome sizes across insect orders are highly variable and do not support a simple correlation where larval life cycles predict smaller genomes.
  • Orders lacking caterpillar-like larvae (e.g., Blattodea, Heteroptera, Odonata, Orthoptera, Phasmatodea) show wide C-value ranges: $C ext{ ranges from } 0.18 ext{ to }16.93 ext{ pg}$ across these groups (data summarized from genome size databases).
  • Holometabolous groups (Coleoptera, Diptera, Hymenoptera, Lepidoptera) also span broad genome sizes but include substantially smaller gene counts than some non-larval groups.
• Gene content data for model holometabolous insects (five species) show relatively few genes, contrary to a prediction that hybridization would greatly inflate gene content in holometabolans:
  • Gene counts range from $10{,}157$ (Apis mellifera) to $16{,}404$ (Tribolium castaneum).
  • All five holometabolous genomes have fewer genes than complete genomes of some vertebrates/birds/mammals, and even fewer than some crustaceans and other arthropods with varying degrees of larval forms.
  • Some non-holometabolous insects and other arthropods have higher gene counts (e.g., Ixodes scapularis: $20{,}467$ genes; Daphnia pulex: $30{,}939$ genes; Acyrthosiphon pisum: $12{,}082$ genes from EST data).
• Significance:
  • The notion that holometabolous insects should have notably larger gene inventories due to hybrid transfer of larval programs is not supported; holometabolous genomes do not show the expected pattern of exchange with an onychophoran donor genome.

Holometabolous Insect Genomes Do Not Contain an Onychophoran-Like Genome Acquired by Hybridization

• Phylogenetic tests of orthologue relationships question Williamson’s prediction that algebraic contrasts would reveal episodic transfers from onychophorans to holometabolans:
  • Roeding et al. (2009) sampled 149 genes (11,168 codons) from a scorpion genome and compared to orthologues from 2 onychophorans and 17 insects plus 12 crustaceans and 21 other taxa.
  • Result: no evidence of close similarities between onychophoran genes and holometabolan insect genes; instead, all insect genes form a clade closely related to branchiopod crustaceans with Onychophora as sister to the crown group Arthropoda.
• Implication for larval evolution:
  • If insects were highly derived crustaceans with nauplius larvae and gained a holometabolous larva by hybrid transfer, one would expect a gene history revealing onychophoran–holometabolan signalings; Roeding et al. found none of that pattern.
  • The phylogeny implies the nauplius lost in the lineage leading to insects, followed by a specialized holometabolous larval form, not a cross-lineage transfer of larval programs via hybridization.

Onychophorans Have Genomes Larger Than Those of Holometabolous Insects and Did Not Hybridize with Insects

• Williamson’s prediction: onychophoran genomes should be smaller than holometabolous insect genomes if they were donors of larval features.
• Empirical data:
  • Velvet worm genomes: Epiperipatus biolleyi ≈ $C = 4.43 ext{ pg}$; Euperipatoides kanangrensis ≈ $C = 6.88 ext{ pg}$.
  • These values are larger than most holometabolous insect genomes and larger than the ranges observed in many non-holometabolous insects.
• Conclusion:
  • Onychophorans have genomes larger than holometabolous insects, contradicting Williamson’s prediction that they would be smaller and thus more likely donors of larval traits.

Tunicates Have Genome Sizes Similar to Larvaceans and Did Not Acquire Tadpole Larvae by Hybridization with Larvaceans

• Williamson’s reference to tunicate/larvacean comparisons suggested a potential parallel to hybridization of larval forms.
• Genomic data across four urochordates show multiple C-values:
  • Ciona intestinalis (tunicate): C igl(= ext{genome size}igr)
    ightarrow 0.20 ext{ pg}
  • Ascidia atra: $C = 0.16 ext{ pg}$
  • Phallusia mammillata: $C = 0.06 ext{ pg}$
  • Oikopleura dioica (larvacean): $C = 0.07 ext{ pg}$
• The Oikopleura genome is approximately one-third the size of the tunicate genome (i.e., $rac{0.07}{0.20} <br /> oughly frac{1}{3}$).
• Why this matters: Williamson cited only part of the tunicate data to support a hybridization claim; including all data shows no simple pattern supporting larval transfer via tunicate–larvacean hybrids.

Tests of Larval Transfer via Genetic Analysis of Experimental Hybrids Have Rejected Its Predictions

• Williamson urged genomic testing of hybrids (i) between larval-bearing animals and relatives without larvae, and (ii) among related taxa that could share donor/recipient genomes.
• A documented laboratory hybridization case (tunicate eggs × sea urchin sperm) was analyzed by Hart (1996) for maternity in a tunicate–sea urchin cross:
  • Maternal mitochondria and biparental nuclear genome analysis showed only sea urchin genes and genomes; no tunicate genes were detected.
  • Williamson later dismissed these results or reframed them, but the core empirical finding stands: no evidence of tunicate genomic material in the hybrid offspring.
• Subsequent reviews and writings by Williamson did not overturn this empirical result; Hart and Grosberg argue that this, along with other data, falsifies the hybridization/larval-transfer hypothesis.
• Broader interpretation:
  • While lateral gene transfer and hybridization occur in some lineages, there is no robust empirical support that larval metamorphosis in animal life cycles arose through such genome transfers between distant clades (e.g., velvet worms and insects).

Synthesis: What the data collectively show

• Across genomes and phylogenies, there is no consistent pattern supporting hybrid transfer as a driver of larval forms or metamorphosis in insects or other major clades.
• Key empirical points:
  • Rhizocephala are true barnacles with unexpectedly small genomes, not genomes composed of multiple donor lineages.
  • Holometabolous genomes do not show the predicted genomic signatures of donor onychophoran genes; holometabolous insect gene content is relatively modest and overlaps with non-larval groups in unpredictable ways.
  • Onychophorans have genome sizes that exceed many holometabolous insects, contradicting a simple donor-to-recipient genome-size expectation.
  • Phylogenetic analyses place onychophoran genes as sister to all arthropods, not as donor-derived sequences to holometabolous insects.
  • Tunicate/larvacean genome data do not support a simple larval-transfer scenario either; the relative sizes and relationships do not align with Williamson’s claims.
  • Direct experimental hybridization tests have failed to reveal hybrid genomes containing the donor larval signatures Williamson proposed.
• Overall conclusion of Hart & Grosberg:
  • There is no credible genomic or phylogenetic evidence to support the hypothesis that caterpillar-like larvae in insects originated via hybridization with velvet worms or other distant clades.
  • Hybridization and lateral gene transfer are important evolutionary mechanisms in some contexts, but they do not explain the origin and loss of animal larval forms as proposed by Williamson.

Implications for understanding larval evolution and methodological lessons

• Larval evolution should be understood in terms of natural selection, developmental constraints, and historical contingency, rather than as a byproduct of inter-lineage genome transfers.
• The study highlights the importance of broad, data-rich tests across multiple lines of evidence (genome sizes, gene content, phylogeny, and experimental hybrids) when evaluating controversial evolutionary hypotheses.
• It serves as a reminder that correlations between morphological similarity (e.g., larval forms) across distant taxa do not by themselves imply histories of hybridization or genome mixing; robust phylogenomic evidence is essential.

Concluding perspective

• Hart and Grosberg conclude that Williamson’s larval transfer hypothesis is falsified by available data: genome sizes, gene counts, and phylogenetic relationships do not support the proposed interphyletic hybridization scenario between velvet worms and insects.
• They acknowledge the importance of hybridization and lateral gene transfer in evolution in general, but maintain there is no current evidence for a role in the evolution and loss of animal larvae as proposed.

Key references cited in this discussion (selected)

• Roeding F, et al. (2009) A 454 sequencing approach for large-scale phylogenomic analysis of the common emperor scorpion (Pandinus imperator). Mol Phylogenet Evol.
• The Honeybee Genome Sequencing Consortium (2006) Insights into social insects from the genome of the honeybee Apis mellifera. Nature.
• Tribolium Genome Sequencing Consortium (2008) The genome of the model beetle Tribolium castaneum. Nature.
• Xia QY, et al. (2008) The genome of Bombyx mori. Insect Biochem Mol Biol.
• Hart MW (1996) Testing cold fusion of phyla: Maternity in a tunicate–sea urchin hybrid determined from DNA comparisons. Evolution.
• Williamson DI (1992, 2009) Earlier works proposing larval transfer and its refutations.