A Phylogeny of the Monocots, as Inferred from rbcL and atpA Sequence Variation, and a Comparison of Methods for Calculating Jackknife and Bootstrap Values

A phylogenetic analysis of the monocots was conducted on the basis of nucleotide sequence variation in two genes (atpA, encoded in the mitochondrial genome, and rbcL, encoded in the plastid genome). The taxon sample of 218 angiosperm terminals included 177 monocots and 41 dicots. Among the major results of the analysis are the resolution of a clade comprising four magnoliid lineages (Canellales, Piperales, Magnoliales, and Laurales) as sister of the monocots, with the deepest branch within the monocots between a clade consisting of Araceae, Tofieldiaceae, Acorus, and Alismatales, and a clade that includes all other monocots. Nartheciaceae are placed as the sister of Pandanales, and Corsiaceae as the sister of Liliales. The Triuridaceae, represented by three genera, including Lacandonia, are resolved as monophyletic and placed in a range of positions, generally within Pandanales. Dasypogonaceae and Arecaceae diverge sequentially from a clade that includes all other commelinid taxa, and within the latter group Poales s. lat. are sister of a clade in which Zingiberales and Commelinales are sisters. Within Poales s. lat., Trithuria (Hydatellaceae) and Mayaca appear to be closely related to some or all elements of Xyridaceae. A comparison was conducted of jackknife and bootstrap values, as computed using strict-consensus (SC) and frequency-within-replicates (FWR) approaches. Jackknife values tend to be higher than bootstrap values, and for each of these methods support values obtained with the FWR approach tend to exceed those obtained with the SC approach. The monocots, an angiosperm group with about 100 families (Kubitzki 1998a, b), have been the subject of numerous phylogenetic analyses (Wilson and Morrison 2000, and citations therein), variously focused on particular subgroups or on the overall phylogenetic structure of the group as a whole. An early and critical set of contributions in this area was generated by Rolf Dahlgren and colleagues (Dahlgren and Clifford 1982; Dahlgren and Rasmussen 1983; Dahlgren et al. 1985), who assembled a comprehensive suite of structural characters for the monocots, surveyed the group for variation in these characters, and applied formal cladistic logic to the analysis of relationships among the monocots using these characters. After the last of these works was published, nucleotide sequence variation in rbcL was brought to bear on the problem of relationships in the monocots (Duvall et al. 1993a, b) and in the angiosperms as a whole (Chase et al. 1993), including a broad sampling of monocots. Many additional studies along these lines have since been conducted (cited below), as various investigators have sampled particular lineages in greater depth, and have generated DNA sequence data sets based on additional genes plus combined matrices that included structural and sequence characters. In a recent review and new analysis, Chase et al. (2000) described a mixed state of affairs. Although several major lineages within the monocots have been identified, with varying degrees of confidence, and some relationships among these groups appear to have been well established, several areas of instability remain. Chase et al. (2000) analyzed relationships among 126 monocot terminals, which had been sampled for nucleotide sequences in three genes (rbcL and atpB from the plastid genome, and 18S ribosomal DNA from the nuclear genome), and they framed much of their discussion in terms of relationships within and among a set of 12 major lineages. While there was robust support for the monophyly of most of these groups, most relationships among them had only weak 468 [Volume 29 SYSTEMATIC BOTANY to moderate support. In the present analysis we examine relationships among 177 monocot taxa, on the basis of nucleotide sequences of two genes, rbcL from the plastid genome and atpA from the mitochondrial genome. To accommodate differences between the results reported here and those reported previously, higher-level relationships are discussed in terms of 15 principal lineages that collectively include all mono-