6. Phylogenetic Analyses of Large Data Sets
101
and Crane 1997), and increased bootstrap support for the clades recovered (Mishler
et al. 1998). Compartmentalization may be a particularly attractive option for analyses
of large data sets that span broad phylogenetic distances, because it allows the use
of clade-specific characters and homology assessments within compartments in
addition to global characters that apply across compartments.
8 Recommendations
We recommend a combination of the approaches discussed here when working
with large data sets at any taxonomic level. Quick search approaches, such as the
parsimony jackknife or fast bootstrap can be applied to any large data set. These
will quickly retrieve the well-supported clades and concomitantly illustrate those
areas of the topology in which confidence is weak. Where critical portions of the
topology are not well supported, investigators are advised to collect more data (additional characters) and perhaps add taxa, rather than conduct lengthy parsimony
searches of a data set in the hope of finding shorter trees (e.g., Rice et al. 1997).
Adding taxa can be particularly important in breaking up long branches (Felsenstein
1978b). Adding characters can increase resolution and internal support for clades,
and shorten run times. Parsimony searches should be conducted taking advantage
of advances in software, such as the RATCHET, which more rapidly recovers short
trees, enabling the investigator to explore larger areas of tree space in a given time.
Tree combination and grafting methods are not a substitute for the analysis of large
data sets. In many cases, however, replacing the exemplars for a given with a topology for that clade that represents more thorough taxon and character sampling will
be both useful and informative.
References
Bremer K (1988) The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42:795-803
Chase MW, Albert VA (1998) A perspective on the contribution of plastid rbcL sequences
to angiosperm phylogenetics. In: Soltis DE, Soltis PS, and Doyle, 11 (Eds) Molecular
systematics of plants II. Chapman and Hall, New York, pp 488-507
Chase MW, Cox VA (1998) Gene sequences, collaboration and analysis of large data sets.
Aust Syst Bot 11:215-229
Chase MW, Soltis DE, Olmstead RG, Morgan D, Les DH, Mishler BD, Duvall MR, Price
RA, Hills HG, Qiu Y-L, Kron KA, Retig JH, Conti E, Palmer JD, Manhart JR, Sytsma
KJ, Michaels Hl, Kress WJ, Karol KG, Clark WD, Hedren M, Gaut BS, Jansen RK,
Kim K-J, Wimpee CF, Smith JF, Fumier GR, Strauss SH, Xiang Q-Y, Plunkett GM,
Soltis PS, Swensen SM, Williams SE, Gadek PA, Quinn CJ, Eguiarte LE, Golenberg
E, Learn GH, Jr., Graham SW, Barrett SC, Dayanandan S, Albert VA (1993)
Phylogenetics of seed plants: An analysis of nucleotide sequences from the plastid gene
rbcL. Ann Mo Bot Gard 80:628-580
101
and Crane 1997), and increased bootstrap support for the clades recovered (Mishler
et al. 1998). Compartmentalization may be a particularly attractive option for analyses
of large data sets that span broad phylogenetic distances, because it allows the use
of clade-specific characters and homology assessments within compartments in
addition to global characters that apply across compartments.
8 Recommendations
We recommend a combination of the approaches discussed here when working
with large data sets at any taxonomic level. Quick search approaches, such as the
parsimony jackknife or fast bootstrap can be applied to any large data set. These
will quickly retrieve the well-supported clades and concomitantly illustrate those
areas of the topology in which confidence is weak. Where critical portions of the
topology are not well supported, investigators are advised to collect more data (additional characters) and perhaps add taxa, rather than conduct lengthy parsimony
searches of a data set in the hope of finding shorter trees (e.g., Rice et al. 1997).
Adding taxa can be particularly important in breaking up long branches (Felsenstein
1978b). Adding characters can increase resolution and internal support for clades,
and shorten run times. Parsimony searches should be conducted taking advantage
of advances in software, such as the RATCHET, which more rapidly recovers short
trees, enabling the investigator to explore larger areas of tree space in a given time.
Tree combination and grafting methods are not a substitute for the analysis of large
data sets. In many cases, however, replacing the exemplars for a given with a topology for that clade that represents more thorough taxon and character sampling will
be both useful and informative.
References
Bremer K (1988) The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42:795-803
Chase MW, Albert VA (1998) A perspective on the contribution of plastid rbcL sequences
to angiosperm phylogenetics. In: Soltis DE, Soltis PS, and Doyle, 11 (Eds) Molecular
systematics of plants II. Chapman and Hall, New York, pp 488-507
Chase MW, Cox VA (1998) Gene sequences, collaboration and analysis of large data sets.
Aust Syst Bot 11:215-229
Chase MW, Soltis DE, Olmstead RG, Morgan D, Les DH, Mishler BD, Duvall MR, Price
RA, Hills HG, Qiu Y-L, Kron KA, Retig JH, Conti E, Palmer JD, Manhart JR, Sytsma
KJ, Michaels Hl, Kress WJ, Karol KG, Clark WD, Hedren M, Gaut BS, Jansen RK,
Kim K-J, Wimpee CF, Smith JF, Fumier GR, Strauss SH, Xiang Q-Y, Plunkett GM,
Soltis PS, Swensen SM, Williams SE, Gadek PA, Quinn CJ, Eguiarte LE, Golenberg
E, Learn GH, Jr., Graham SW, Barrett SC, Dayanandan S, Albert VA (1993)
Phylogenetics of seed plants: An analysis of nucleotide sequences from the plastid gene
rbcL. Ann Mo Bot Gard 80:628-580