Phylogeny of the Proteaceae based on atpB and atpB-rbcL intergenic spacer region sequences

Abstract

Parsimony analyses were conducted for 46 genera representing all subfamilies and tribes within Proteaceae using two chloroplast sequences: the gene atpB and the noncoding spacer region between atpB and rbcL. The spacer region was more variable than atpB and provided insertion and deletion data as well as nucleotide substitutions. The atpB and spacer region data sets were highly congruent (as indicated by the partition homogeneity test) and were analysed separately and combined. Both unweighted and weighted character states (3 : 1 correction for transition bias) for the atpB data resulted in very similar strict consensus trees. In addition, the large subfamilies Proteoideae and Grevilleoideae were analysed separately, using appropriate outgroups determined by the analyses with complete sampling. The results from the combination of data were better resolved and supported than the results from each separate data set, although the Grevilleoideae were highly unresolved in all analyses. Most subfamilies in the Proteaceae were essentially monophyletic, but most tribes and subtribes were not. Bellendena is weakly supported as the sister group to all remaining members of the Proteaceae. Monotypic Eidotheoideae is well supported as a member of Proteoideae. Carnarvonioideae and Sphalmioideae are strongly supported as closely allied to the Grevilleoideae, but their positions in relation to this subfamily are unresolved. Other unusual alliances supported by our molecular data are: Isopogon–Adenanthos–Leucadendron–Protea, Petrophile–Aulax, Cardwellia–Euplassa–Gevuina, and Opisthiolepis–Buckinghamia–Grevillea. The tree resulting from the combined data showed limited congruence with morphological characters (flower pairs, stylar pollen presentation, and ovule number). Congruence with chromosome number was minimal, but our tree does support previous hypotheses of multiple aneuploidy and chromosome doubling events. The African and South American genera included in our analysis are dispersed among various clades with taxa from Australia and Asia, suggesting a former Gondwanian distribution for Proteaceae.

A new suprageneric classification of the Proteaceae, with an annotated checklist of genera

“…The genera Protea, Roupala and Floydia were excluded from the coding of their tree for the following reasons. Hoot and Douglas (1998) reported Leucadendron to be more closely related to Protea than to either Adenanthos or Isopogon, contrary to our results (Barker et al 2002, WB&D rbcL). We checked the identity of their DNA sequences of Protea neriifolia by examining a digital image of their voucher specimen and it is definitely a specimen of Leucadendron, although we could not identify it to species.…”

“…Numbered rectangles on the right hand side of the figure refer to named higher taxa recognised here, signified by their numbers in the taxonomic treatment (subfamilies by digits 1-5 in the left-hand column, tribes by numbers 2.1-5.4 in the middle column and subtribes by numbers 4.1.1-5.4.4 in the right-hand column of rectangles). Weston & Barker In chloroplast DNA analyses, Bellendena is either resolved as sister group to the rest of the family ("the rest" receiving 61% bootstrap support, Hoot & Douglas 1998), or left in an unresolved, basal position (WB&D rbcL). Our ITS analysis (WB&D ITS) was rooted at its midpoint, which resulted in Bellendena being resolved as sister group to the Persoonioideae in Fig.…”

“…A supertree analysis using matrix representation and parsimony ('MRP' -BinindaEmonds et al 2002, 2004 was conducted that synthesised the results of published and unpublished molecular phylogenetic analyses of the Proteaceae (Hoot & Douglas 1998, Mast & Givnish 2002, Barker et al 2002 (Analysis 1), WB&D ITS and WB&D rbcL -see above). The terminal taxa were currently recognised genera of Proteaceae.…”

“…Generic membership of several components of the bootstrap consensus tree of Hoot and Douglas (1998) was modified to exclude dubiously or wrongly identified samples. The genera Protea, Roupala and Floydia were excluded from the coding of their tree for the following reasons.…”

“…Some of these analyses have been published while others are still in preparation. Hoot and Douglas (1998) is an analysis of representatives of most genera of Proteaceae, based on an alignment of DNA sequences for the chloroplast loci atpβ and the atpβ-rbcL intergenic spacer. Mast and Givnish (2002: fig.…”