(A) Cartoon of the phylogenetic relationships of the nematode species analysed, based on analyses of the nuclear small subunit ribosomal RNA gene. The colours associated with each species are used to decorate the subtrees in (C)–(J). Coloured stars on the subtrees indicate species-specific ‘blooms’ of paralogues. (B) The global tree of nematode Ago represented as an unrooted phylogram. We assume that each subtree is effectively rooted by the other subtrees, although we note that extreme divergence results in the support for structure within some subtrees being marginal. (C) The ALG1/ALG2 subtree, which contains representatives from all the nematode clades surveyed. (D) The RDE1/ERGO1 (endogenous RNAi-deficient Ago 1)/PRG1/PRG2 subtree, which contains no members from the animal parasites of Clade III. (E) The WAGO10/WAGO11/NRDE (nuclear RNAi-defective) subtree, which has wide representation across the Nematoda. (F) A SAGO2 (synthetic secondary siRNA-deficient Ago mutant 2)/PPW (PAZ/PIWI domain-containing) WAGO subtree, which is restricted to Caenorhabditis and P. pacificus. (G) A subtree of WAGO proteins, which lacks any members from Caenorhabditis, Clade IV or Clade I species, but is present in P. pacificus and animal parasites in Clades III and V. (H) The WAGO1/WAGO2/WAGO4/WAGO5 subtree, restricted to Clade IV and V nematode species, with a remarkable bloom of paralogues in P. redivivus. (I) This component of Ago diversity includes C. elegans CSR and a subtree of WAGO proteins restricted to Clade III, IV and V species, with a second paralogue bloom in P. redivivus. (J) The ALG3/ALG4 subtree, in which T. spiralis has a remarkable bloom of 119 distinct Ago proteins. Over 550 distinct Ago proteins (containing PIWI and PAZ domains) were obtained by extensive similarity searching of the NCBI NR protein database, WormBase nematode genome data (http://www.wormbase.org) and NEMBASE4 (http://www.nematodes.org), and aligned using CLUSTAL Omega. The alignment was analysed in MrBayes 3.2.1 using a mixed prior on amino acid evolution model, and run for 1 million generations. After visual inspection in Tracer (http://tree.bio.ed.ac.uk/software/tracer/), the first 500000 generations were discarded as burnin. The input sequences, alignment, MrBayes command block, treefiles and summary phylograms are available on DataDryad (doi:10.5061/dryad.5qs11). For improved clarity, a full-size PDF version of this Figure can be found at http://www.biochemsoctrans.org/bst/041/bst0410881add.htm.