And spent three hours running BEAST and the MCC tree insists my arid-adapted beetles are sister to a clade of marine gastropods — convergent traits and long-branch attraction teaming up like pranksters, or did I botch the partitioning by gene? Anyone else have a dataset where homoplasy turns your careful phylogeny into improv comedy?
And i had a “long-branch attraction” mess in BEAST, and what fixed it was dropping 3rd codon positions before the run; after that the MCC stopped trying to marry my arid-adapted beetles to gastropods. If the saturation plots look nasty, give that one rerun even if it costs another three hours — were you partitioning by codon position or just by gene?
Before another 3-hour BEAST run, try IQ-TREE’s PMSF/C60 to blunt LBA: https://iqtree.org. Also recheck outgroup choice.
Sounds like composition might be meddling — run a quick SymTest/BaCoCa check, and if partitions are wonky try RY- or Dayhoff-6 recoding, or even CAT-GTR in PhyloBayes (http://www.phylobayes.org) to calm the “long-branch attraction” jitters. I’ve also seen a single rogue taxon flip the MCC until removed based on leaf stability. Any obvious GC-bias differences between the desert beetles and the sea snails, or is that just a mirage?
And i’d try TreeShrink for rogue taxa before another “three hours”: GitHub - uym2/TreeShrink: Implementation of the TreeShrink problem — did the “MCC tree insists” pairing vanish?