Pancrustacea Metamorphosis Gene Families

Bottom line: metamorphosis evolved four separate times in Pancrustacea (insects,
decapods and krill, copepods, barnacles), and a 2026 phylogenomic preprint (Campli et al.)
finds that each origin recruited different gene families that converge on the same
developmental functions, with a small core of 15 families showing adaptive expansion. We took
the eight Drosophila reference genes the paper names for those families (kni, hairy,
klg, trn, caps, kek1, krz, insc) and reviewed every existing GO annotation on
them, because almost all functional knowledge for these families comes from the fly and would
be the source of any transfer to crustacean orthologues. All eight reviews are done and
validate: 183 existing annotations adjudicated (108 ACCEPT, 46 KEEP_AS_NON_CORE, 24 MODIFY,
1 REMOVE, 1 MARK_AS_OVER_ANNOTATED, 3 UNDECIDED) plus 4 proposed NEW terms. The main
corrections were redirecting propagated receptor terms on adhesion molecules (trn, klg)
and on the ligand-less orphan receptor knirps to what the proteins actually do, and
resolving every bare protein binding row to its named partner. The optional next step, an
ecdysteroid-regulation module, has not been started.

Source

Campli G, Chipman AD, Robinson-Rechavi M, Waterhouse RM.
Convergent gene family evolution underpins repeated transitions to
metamorphic development across Pancrustacea.
bioRxiv (2026), posted July 26, 2026.
doi: 10.64898/2026.05.06.723392
(preprint, not peer reviewed; CC-BY 4.0).

This page is a reviewer's digest of that preprint plus a curation to-do list.
It is not a reproduction of the paper.

What the paper does

The study asks whether the repeated evolution of metamorphosis —
a post-embryonic, moult-mediated life-stage progression to adulthood marked by
major morphological and ecological change — leaves a shared genomic signature.
It assembles a phylogenomic dataset of 54 species across 26 pancrustacean
orders (median Arthropoda BUSCO completeness 95.8%) and time-calibrates a
species tree spanning ~500 My. Four clades are treated as independent
evolutionary replicates of a transition to metamorphic development
("Metamorphosis LCAs"):

Orthologous groups (OGs) were delineated at the Pancrustacea LCA with
OrthoLoger/OrthoDB (42,841 OGs, 782,987 genes, 73% of input), phyletic ages
assigned, and gene gains/losses reconstructed with CAFE v5 on ancient,
widespread OGs (≥85% species). GO enrichment (GenBank/RefSeq/FlyBase/InterPro
annotations) was contrasted between Metamorphosis LCAs, Sister LCAs, and deeper
ancestral nodes, and OU/BM evolutionary models tested for lineage-specific
adaptive expansions.

Key findings a curator should carry

1. More births, more expansions at metamorphic origins. Metamorphosis LCAs
   show elevated gene-family births (8.7%, 3,722 OGs vs 2.6%, 1,112 at Sister
   LCAs) and expansions (2,078 OGs vs 843 at Sisters / 442 at Deep nodes).
   Emergent and expanding families are more sequence-constrained (lower
   divergence) than at Sister LCAs.
2. Convergent functions, divergent genes. Of 100 GO terms enriched among
   families expanding at any Metamorphosis LCA, 60 are shared by all four
   (28 semantic clusters) — yet the expansions mostly involve distinct
   genes, not parallel expansion of the same family. Functional convergence
   is achieved through different genetic trajectories.
3. The convergent functions cluster on: central/peripheral nervous-system
   development and neurogenesis; epithelial and cuticle morphogenesis
   ("chitin-based cuticle development"); developmental maturation; neuropeptide
   signalling and regulation of autophagy; compound-eye/photoreceptor
   development; segmentation; and immune activation. These are all plausibly
   tied to the morphological reorganisation and adaptive-landscape shift of
   metamorphosis.
4. A small adaptive core. Of 528 shared-enrichment expanding families, only
   15 (3%) show OU two-optima signatures of lineage-specific adaptive
   expansion (largest optimum in the metamorphic lineages), annotated to
   nervous-system development, chitin-based cuticle development, dorsal closure,
   head involution, imaginal-disc-derived wing-vein/chaeta/salivary-gland
   morphogenesis, and MAPK-signalling regulation.
5. Reframing moulting. The authors argue the ancestral moulting programme is
   an evolutionarily flexible developmental substrate whose repeated
   modification enabled complex multi-phasic life histories — echoing (but
   distinct from) toolkit overlap seen in arthropod terrestrialisation studies.

Curation caveat (stated by the authors, important here): nearly all
functional knowledge for the named families comes from model insects
(chiefly Drosophila). Copy number is dynamic across lineages, and
pleiotropy, post-duplication regulatory rewiring, and co-option/exaptation mean
insect-derived functions should not be assumed to transfer to crustacean
orthologues. Any review should keep organism-of-evidence explicit and resist
propagating Drosophila function onto uncharacterised paralogues.

Candidate gene families for review

The paper singles out these families as showing adaptive, lineage-specific
expansion at metamorphic origins. Named genes below are the Drosophila
reference members — the natural entry points for a GO-annotation review, none
of which is yet reviewed in this corpus. "Family N" is the paper's numbering.

deadpan (dpn) is also mentioned alongside knirps/hairy in insect neural
development but was not called out as an adaptively expanding family.

Suggested review order: the two transcription factors knirps and hairy
are the best-characterised and most cross-lineage-informative starting points;
the adhesion/receptor families (klingon, tartan/capricious, kekkon) form
a coherent "neuronal wiring & disc morphogenesis" second batch; Kurtz and
inscuteable round out the signalling/asymmetric-division angle.

To start a review for any of these:

just fetch-gene DROME kni      # then deep research + notes, then the ai-review.yaml

Open questions this paper raises for curation

• Do the Drosophila GO annotations for these families over-attribute
  insect-specific developmental roles (e.g. imaginal-disc terms) that cannot
  hold in crustacean orthologues? This is a candidate over-annotation pattern.
• For pleiotropic pair-rule/gap TFs (hairy, knirps), are segmentation,
  neurogenesis, and ecdysteroid-regulation roles all separately supported, or is
  one propagated from another by IEA/IBA?
• Which of the convergent GO clusters ("chitin-based cuticle development",
  "neuropeptide signalling pathway", "regulation of autophagy") are backed by
  experimental evidence in Drosophila vs inferred electronically?

Status

IN_PROGRESS. All eight Drosophila reference genes named in the paper have
now been reviewed. The two transcription factors:

• knirps (kni, P10734) — 29 annotations adjudicated (17 ACCEPT, 8 MODIFY,
  3 KEEP_AS_NON_CORE, 1 REMOVE). Notable: removed an
  over-propagated intracellular receptor signaling pathway term (knirps is a
  ligand-independent orphan NR that lost its ligand-binding domain) and
  redirected the IBA nuclear receptor activity / estrogen response element binding terms and the bare protein binding IPIs to informative repressor /
  corepressor-binding terms.
• hairy (h, P14003) — 45 annotations adjudicated (28 ACCEPT, 9 MODIFY, 6
  KEEP_AS_NON_CORE, 1 UNDECIDED, 1 MARK_AS_OVER_ANNOTATED). Notable:
  resolved every protein binding IPI to its curated WITH/FROM partner — STUbL
  (Topors/Degringolade) ligase binding, Groucho/CtBP corepressor binding, and
  DNA-binding transcription factor binding for the Ultrabithorax interaction — and
  flagged a distal membrane organization term as over-annotation of a nuclear
  repressor.

The adhesion/receptor and signalling/asymmetric-division candidates are now also
reviewed:

• klingon (klg, Q9VCT4) — 19 annotations. GPI-anchored IgSF adhesion
  molecule; core homophilic-adhesion and R7 photoreceptor roles accepted, distal
  ethanol/long-term-memory behaviours kept non-core, bare protein binding
  redirected to cell adhesion molecule binding (WITH/FROM partner cDIP), and the
  IBA axon guidance receptor activity corrected to cell-cell adhesion mediator activity (as in trn), since a GPI-anchored protein cannot itself transduce a
  guidance signal.
• tartan (trn, M9PFH7) — 6 annotations. LRR adhesion molecule; IBA
  signaling receptor activity corrected to cell-cell adhesion mediator activity (propagated from a TLR-containing family).
• capricious (caps, A0A0S0WP14) — 11 annotations. LRR adhesion molecule
  (trn paralog); core homophilic-adhesion / axon-target-recognition accepted,
  tissue-specific contexts (some the cited papers found caps dispensable for)
  kept non-core.
• kekkon-1 (kek1, Q9VK54) — 22 annotations. Dedicated negative-feedback
  inhibitor of EGFR signalling; the EGF-receptor-binding / receptor-inhibitor /
  neg-reg-EGFR core is strongly experimentally supported and accepted.
• Kurtz (krz, Q9V393) — 21 annotations. Non-visual β-arrestin; core
  GPCR-binding/internalization adaptor accepted, the many pleiotropic
  signalling-attenuation roles (MAPK/Toll/Hedgehog/Notch) kept non-core.
• inscuteable (insc, Q9W2R4) — 30 annotations. Cytoskeletal
  spindle-orientation adaptor; establishment of mitotic spindle localization
  corrected to …orientation; two mis-cited/unverifiable references flagged in
  reference_review.

All eight reviews validate clean, with every supporting_text quote
independently confirmed verbatim against the cached literature. Each bare
protein binding IPI was resolved through the GOA WITH/FROM partner to a
specific binding term. The full candidate set named in the paper is now reviewed;
next steps are the GO-CAM/module angle (e.g. an ecdysteroid-biosynthesis-
regulation module) and, optionally, the secondary mentions (deadpan).

Data-provenance note. For klingon and inscuteable, fetch-gene first
resolved sparse TrEMBL accessions (3 and 1 annotations); the reviews use the
annotation-rich accessions Q9VCT4 and Q9W2R4 instead.

Slides

• Slides (Marp source: PANCRUSTACEA_METAMORPHOSIS-slides.md) — AI generated