Southern Hemisphere in Apiaceae origin and diversification. Appendix S2. Page 1
SUPPORTING INFORMATION
The role of the Southern Hemisphere in the evolutionary history of Apiaceae, a
mostly north temperate plant family
Carolina I. Calviño, Federico E. Teruel and Stephen R. Downie
Journal of Biogeography
Appendix S2. Characteristics of the alignments and maximum likelihood trees resulting
from phylogenetic analyses of the (a) Apiaceae rps16 intron data set, (b) distal apioids
nrDNA ITS data set, and (c) Annesorhizeae rps16 intron plus nrDNA ITS data set. Each
of these trees was highly consistent with those inferred using maximum parsimony and
Bayesian analyses; topological differences between the maximum likelihood tree and
maximum parsimony and/or Bayesian trees are indicated by asterisks. Maximum
likelihood and maximum parsimony bootstrap support values are indicated above
branches (left and right, respectively). Posterior probability values, expressed as
percentages, are indicated below branches. The names of tribes and other major clades
are based on Calviño & Downie (2007) and Downie et al. (2010). Malagasy umbellifers
are highlighted in gray.
Southern Hemisphere in Apiaceae origin and diversification. Appendix S2. Page 2
(a)
Alignment of the 129 rps16 intron sequences resulted in a matrix of 1338
positions, of which 377 were excluded from the analysis because of alignment
Southern Hemisphere in Apiaceae origin and diversification. Appendix S2. Page 3
ambiguities. Of the remaining positions, 322 were parsimony informative, 169 were
autapomorphic, and 470 were not variable. In addition, 48 parsimony informative gaps of
1-22 bp in size were inferred. Maximum parsimony (MP) analysis of the unambiguously
aligned positions resulted in the preset maximum tree limit of 20,000 trees, each of 1306
steps (consistency indexes = 0.5536 and 0.4738, with and without uninformative
characters, respectively; retention index (RI) = 0.8445). Repeating the analysis with the
48 scored gaps also resulted in the preset limit of 20,000 trees, each of 1367 steps
(consistency indexes = 0.5640 and 0.4902, with and without uninformative characters,
respectively; RI = 0.8529). The topology of the strict consensus of these trees was
congruent to the strict consensus tree obtained when gaps were not included, but the
inclusion of gap characters increased resolution within tribes Apieae, Tordylieae,
Scandiceae, Heteromorpheae, and Annesorhizeae. The Bayesian analysis was stopped
at 1,880,000 generations because at that point the average standard deviation of the
split frequencies between the two runs dropped to less than 0.01, indicating
convergence in topologies. The first 4700 trees of each run were discarded as “burn in”
and a majority-rule consensus tree that summarizes topology and branch length
information was calculated based upon the remaining 28,200 trees.
In the MP and Bayesian trees, the sister relationship of tribe Heteromorpheae
and the rest of the Apioideae collapses and the four members of tribe Pleurospermeae
constitute three unresolved lineages in a polytomy with the clade of Diplolophium
somaliense through Pimpinelleae. Other differences between these trees include slightly
less resolution within most tribes in the MP and/or Bayesian trees. Despite these minor
differences, all phylogenetic analyses estimated the same major clades as identified in
previous studies using ITS or other chloroplast markers (Calviño & Downie, 2007;
Nicolas & Plunkett, 2009; Downie et al., 2010; Magee et al., 2010).
Southern Hemisphere in Apiaceae origin and diversification. Appendix S2. Page 4
(b)
Southern Hemisphere in Apiaceae origin and diversification. Appendix S2. Page 5
Alignment of the 99 ITS sequences comprising the distal apioids data set
resulted in a matrix of 647 positions, of which 331 were parsimony informative, 73 were
autapomorphic, and 243 were not variable. MP analysis of these aligned positions
resulted in 1996 trees, each of 2090 steps (consistency indexes = 0.3608 and 0.3333,
with and without uninformative characters, respectively; RI = 0.8445). The Bayesian
analysis was stopped at 2,397,400 generations because at that point the average
standard deviation of the split frequencies between the two runs dropped to less than
0.01. The first 5000 trees of each run were discarded as “burn in” and a majority-rule
consensus tree that summarizes topology and branch length information was calculated
based upon the remaining 37,948 trees. Differences among the trees correspond to the
collapse of a few nodes indicated by asterisks in figure b.
(c)
Alignment of the 19 ITS and rps16 intron sequences obtained for the
Annesorhizeae data set resulted in a matrix of 1629 positions, of which 106 were
Southern Hemisphere in Apiaceae origin and diversification. Appendix S2. Page 6
excluded because of alignment ambiguities. Of the remaining 1523 aligned positions,
155 were parsimony informative, 140 were autapomorphic, and 1228 were not variable.
MP analysis of all unambiguously aligned positions resulted in six equally most
parsimonious trees, each of 475 steps (consistency indexes = 0.7558 and 0.6431, with
and without uninformative characters, respectively; RI = 0.7666). The Bayesian analyses
showed MCMC convergence for all parameters in the best-fit model (PSRF reached 1
for all parameters) by 5 million generations. The first 12,500 trees of each run were
discarded as “burn in” and a majority-rule consensus tree that summarizes topology and
branch length information from both runs was calculated based upon the remaining
75,000 trees.
All phylogenetic analyses estimated two major and well supported clades: one
that comprises the genera Annesorhiza, Chamarea, and Itasina from Southern Africa
(100% maximum likelihood bootstrap (MLBS), maximum parsimony bootstrap (MPBS),
Bayesian posterior probability (PP)), and the other that includes the Southern African
Ezosciadium capense sister to Astydamia latifolia from the Canary Islands and
Molopospermum peloponessiacum from the Pyrenees, Massif Central, and southern
Alps (94% MLBS, 75% MPBS, 95% PP).