molecular phylogenetics of fouquieriaceae: evidence from nuclear ...

American Journal of Botany 86(4): 578–589. 1999.
MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE:
EVIDENCE FROM NUCLEAR RDNA ITS STUDIES 1
LISA M. SCHULTHEIS 2 AND BRUCE G. BALDWIN
Department of Integrative Biology and Jepson Herbarium, University of California, Berkeley, California 94720
Molecular sequence data from the 18S–26S rDNA internal transcribed spacer (ITS) region support the monophyly of
Fouquieria sensu lato (Fouquieriaceae) and the three subgenera (subg. Fouquieria, subg. Bronnia, subg. Idria) previously
recognized within it. Resolution within subg. Fouquieria differs somewhat between parsimony and maximum likelihood
(ML) trees. Section Fouquieria and sect. Ocotilla within subg. Fouquieria are not well supported as monophyletic groups.
Uncertainty regarding placement of the root within Fouquieriaceae makes discussion of character evolution within the family
difficult. Three root positions are consistent with rate-constant evolution of ITS sequences: (1) along the branch to subg.
Idria, (2) along the branch to subg. Bronnia, and (3) along the branch to subg. Fouquieria. The first root position listed is
equivalent to an outgroup rooting. The third root position listed is equivalent to a midpoint rooting. Of the three root
positions above, only the third is along a branch that may be sufficiently long to act as a long-branch attractor. The first
two root positions would result in character reconstruction suggesting that succulent growth forms and white floral pigmentation
are ancestral within the family, with shifts to woody growth forms and to red floral pigmentation. The third root
position results in equivocal reconstruction of the ancestral growth form, equivocal reconstruction of ancestral floral pigmentation
in parsimony trees, and a suggestion of white floral pigmentation as ancestral in ML trees. Two previous hypotheses
of polyploid origins are compatible with the molecular data presented here: (1) origin of the tetraploid F. diguetii
from F. macdougalii, and (2) allopolyploid origin of the hexaploid F. burragei from the tetraploid F. diguetii and a diploid
species similar to F. splendens. Direct descent of the hexaploid F. columnaris from the subg. Bronnia lineage is not supported
by our data. An amphiploid origin of F. columnaris involving a member of the subg. Bronnia lineage and an extinct taxon
outside subg. Bronnia, however, cannot be ruled out.
Key words: boojum; Fouquieria; Fouquieriaceae; ITS; long-branch attraction; ocotillo; rooting.
Fouquieriaceae is a small and well-diagnosed family
of flowering plants distributed throughout the warm deserts
of Mexico and the southwestern United States.
Unique features of Fouquieriaceae include placentation
that changes from parietal to axile during fruit development,
decurrent spines formed from the petioles of primary
leaves, and an anastomosing network of cortical
water storage tissue (Scott, 1932; Henrickson, 1969a,
1972). Relationships within the family have been examined
most thoroughly by Henrickson (1972) using morphological,
cytological, and ecological data. From this
work, Henrickson proposed the recognition of one genus,
Fouquieria (including Idria), three subgenera (Fouquieria,
Bronnia, and Idria), and 11 species in Fouquieriaceae
(see Table 1). We generated phylogenetic hypotheses for
1 Manuscript received 10 March 1998; revision accepted 10 September
1998.
The authors thank Sarah Vetault and Dr. Michael Donoghue (Harvard
University) for providing DNA samples; Dr. Mark Porter (Rancho Santa
Ana Botanic Garden), Staci Markos (U. C. Berkeley), and Lena Hileman
(Harvard University) for providing DNA sequences; and Bob Perill
(Boojum Unlimited, Tucson, AZ), Dr. Richard Felger (Drylands Institute,
Tucson, AZ), and Cathy Babcock (Desert Botanic Garden, Phoenix,
AZ) for additional plant material. We thank Dr. John Huelsenbeck
(University of Rochester) for his extensive help and advice regarding
molecular sequence simulations and use of his Siminator program, and
Dr. David Swofford (Smithsonian Institution) for granting us permission
to use and publish results from a test version of PAUP* 4.0. We also
thank Staci Markos, Dr. Mark Porter, and two anonymous reviewers for
helpful comments on the manuscript. This work was supported in part
by funding from the A. W. Mellon Foundation through the Botany Department
at Duke University and the Lawrence R. Heckard Fund of the
Jepson Herbarium at U. C. Berkeley.
2 Author for correspondence.
578
Fouquieriaceae based on molecular data from the 18S–
26S rDNA internal transcribed spacer (ITS) region. These
trees were used to evaluate the current taxonomy of the
group, as proposed by Henrickson and to examine hypotheses
of character evolution.
Henrickson’s (1972) subgenera of Fouquieriaceae
roughly correspond to growth forms. Subgenus Idria and
subg. Bronnia comprise the succulent taxa; subg. Fouquieria
contains the woody taxa. The succulent species
achieve their growth form by an increase in parenchymatous
water storage tissue within the secondary xylem
and, in F. columnaris, the pith (Humphrey, 1935; Henrickson,
1969c). Subgenus Idria contains F. columnaris
(boojum tree), a columnar plant with a succulent trunk
reaching a height of up to 40 feet (Humphrey, 1935).
Subgenus Bronnia contains F. fasciculata and F. purpusii,
both of which have basally succulent trunks. Subgenus
Fouquieria contains eight woody (or dendroid)
species, split into sect. Fouquieria and sect. Ocotilla. The
six species of sect. Fouquieria bear distinct trunks and
branching stems. The two species of sect. Ocotilla have
a reduced trunk and vertical unbranched stems. In his
comparative anatomical study of F. columnaris and F.
splendens, Humphrey (1935) suggested that members of
Fouquieriaceae cannot be readily classified as shrubs or
as stem-succulents, the two predominant types of xerophytic
plants he recognized within the southwestern
North American deserts. The shrub forms of subg. Fouquieria
possess a degree of cortical succulence, and the
succulent forms are not succulent throughout, even in F.
columnaris, which possesses nonsucculent branches.
Growth forms in the family are intermediate between typical
xeromorphic types.

April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE
579
TABLE 1. Classification of Fouquieriaceae, following Henrickson
(1972).
Fouquieriaceae A.P. de Candolle
Fouquieria H.B.K.
Subgenus Fouquieria
Section Fouquieria
F. leonilae F. Miranda
F. ochoterenae F. Miranda
F. macdougalii Nash
F. diguetii (van Tiegh.) I.M. Johnston
F. burragei Rose
F. formosa H.B.K.
Subgenus Fouquieria
Section Ocotilla Henrickson
F. splendens Engelm. in Wisliz.
ssp. splendens
ssp. campanulata (Nash) Henrickson
var. campanulata
var. albiflora Henrickson
ssp. breviflora Henrickson
F. shrevei I.M. Johnston
Subgenus Bronnia (H.B.K.) Henrickson
F. fasciculata (Willd. ex Roem. et Schult.) Nash
F. purpusii T.S. Brandegee
Subgenus Idria (Kellogg) Henrickson
F. columnaris (Kellogg) Kellogg ex Curran
The base chromosome number in Fouquieriaceae is x
12 (Henrickson, 1972). Both Fouquieria columnaris
and F. burragei are hexaploids (n 36), F. diguetii is a
tetraploid (n 24), and the remaining species are diploid
(n 12) (chromosome counts for F. formosa are lacking).
Henrickson (1972) suggested that F. burragei is part
of a polyploid series including F. macdougalii and the
tetraploid F. diguetii. He suggested that F. diguetii may
have descended from ‘‘F. macdougalii stock’’ and that
F. burragei may have been of amphiploid origin with
possible parents including F. diguetii and ‘‘some presumably
now extinct diploid species having numerous stamens,
possibly short, white corollas and spicate or racemose
inflorescences’’ (Henrickson, 1972, p. 499). Similarly,
a species from the F. purpusii and F. fasciculata
lineage (subg. Bronnia) may have served as a parent in
an amphiploid derivation of F. columnaris (subg. Idria)
(Henrickson, 1972).
With regard to floral characteristics, just over half of
Fouquieriaceae species possess stereotypical hummingbird-pollinated
flowers, with tubular red corollas containing
nectar at the base. All red-flowered species are in
subg. Fouquieria. Also in subg. Fouquieria are F. burragei
and F. shrevei, both of which have corollas that are
pink in bud and white to pink at maturity (Henrickson,
1972). Members of subg. Bronnia and subg. Idria have
corollas that are cream yellow to white throughout their
development (Henrickson, 1972). Members of these two
subgenera, as well as F. shrevei, lack floral anthocyanins
(Scogin, 1977). In addition to floral pigmentation, flowers
of Fouquieria taxa differ in the number of stamens, the
degree of corolla limb reflexion, and their pattern of arrangement
in inflorescences.
Fouquieriaceae has been considered closely related to
various families and orders since its establishment by De
Candolle in 1828. De Candolle suggested an affinity with
Portulacaceae, as did Humboldt, Bonpland, and Kunth
when they first described the genus Fouquieria in 1823
(for expanded discussion and references see Henrickson
[1967, 1972]). The hypothesized affinity of Fouquieriaceae
with Portulacaceae is untenable because Caryophyllales
is diagnosed by numerous characteristics (Mabry,
1977) lacking in Fouquieriaceae. The type specimen
for Fouquieria was originally placed in Cantua (Polemoniaceae)
by Roemer and Schultes (1819). An affinity
with Polemoniaceae was also suggested by Nash (1903),
in the first treatment of Fouquieriaceae, and by Henrickson
(1967, based on pollen comparisons), the author of
the most recent treatment. Additional suggestions for the
position of Fouquieriaceae have included alignment with
Tamaricales (e.g., Hutchinson, 1926; Takhtajan, 1969),
Violales (e.g., Cronquist, 1981), Ebenales (e.g., Bessey,
1915; Morton et al., 1996), Solanales (e.g., Thorne, 1969;
Scogin, 1977; both included Polemoniaceae within Solanales),
and Ericales (e.g., Dahlgren, Jensen, and Nielsen,
1976; Jensen and Nielson, 1982; Hufford, 1992;
Olmstead et al., 1993). The Ericales clade in which Fouquieriaceae
has been placed in molecular trees is broadly
circumscribed, including Ebenales taxa among others
(Olmstead et al., 1993). Anderberg (1992) placed Fouquieriaceae
at the base of Asteridae but noted that the
family could shift closer to the Ericales clade with increased
taxon sampling. The general pattern that has
emerged from various phylogenetic analyses is placement
of Fouquieriaceae at the base of Asteridae, perhaps within
a broadly circumscribed ericalean grouping.
MATERIALS AND METHODS
Total DNA was isolated from 24 specimens (Table 2), following a
minor modification of the CTAB protocol of Doyle and Doyle (1987),
and purified on CsCl 2 gradients. Two chloroform/isoamyl alcohol extractions
and two ethanol precipitations (following isopropanol precipitation)
were added to Doyle and Doyle’s method. Leaf material was
dried and stored in silica gel desiccant prior to DNA isolation. Most
DNAs from Fouquieriaceae taxa were generously provided by Sarah
Vetault (University of Arizona) and Michael Donoghue (Harvard University),
with subsequent collections obtained to provide additional representation
within each species, when possible. Outgroup taxa were chosen
to represent the various orders with which Fouquieriaceae have been
aligned. One ingroup sequence was provided by J. Mark Porter (Rancho
Santa Ana Botanic Garden), and a total of nine outgroup sequences
were provided by J. Mark Porter, Staci Markos (U. C. Berkeley), and
Lena Hileman (Harvard University), as noted in Table 2.
Single-stranded DNAs of ITS 1 and ITS 2 were generated, purified,
and sequenced following Baldwin (1992). For some taxa, double-stranded
DNAs were generated using a GeneAmp 9600 with the following
conditions: initial denaturation (97C, 1 min), followed by 40 cycles of
denaturation (97C, 10 s), annealing (48C, 30 s), and extension (72C,
20 s increasing 4 s with each cycle), and concluding with a final extension
(72C, 7 min). The polymerase chain reactions (PCR) contained
the following components: 5.0 L 10X PCR buffer II (Perkin Elmer,
Foster City, California), 2.5 mmol/L MgCl 2, 1.0 mmol/L dNTPs, 2.5
L glycerol, 0.5 mol/L ITS-I primer (5’-GTCCACTGAACCTTAT-
CATTTAG-3’; designed by L. E. Urbatsch, Louisiana State University),
0.5 mol/L ITS4 primer (White et al., 1990), 1.0 unit AmpliTaq DNA
polymerase (Perkin Elmer, Foster City, California), and 1–10 ng DNA,
to a total volume of 50 L. PCR products were cleaned using either
Ultrafree-MC 100000 NMWL polysulfone membrane filter units (Millipore
Corporation, Bedford, Massachusetts), or Wizard PCR Preps
DNA Purification System (Promega, Madison, Wisconsin), following
manufacturer’s instructions. Sequencing reactions were performed with

580 AMERICAN JOURNAL OF BOTANY
[Vol. 86
TABLE 2. Source of taxa used. a
Taxon Source
Family Cyrillaceae
Cyrilla racemiflora L. Schultheis 2–94; DUKE
Family Ericaceae
Arctostaphylos nummularia A. Gray
A. uva-ursi (L.) Sprengel
Arbutus menziesii Pursh
Lyonia lucida (Lam.) K. Koch
Vaccinium fuscatum Ait.
Family Fouquieriaceae
Fouquieria burragei Rose (1) e
F. burragei (2)
F. columnaris (Kellogg) Kellogg ex Curran (1) e
F. columnaris (2)
F. diguetti (van Teigh.) I.M. Johnston (1) e
F. diguetti (2)
F. fasciculata (Willd. ex Roem. et Schult.) Nash (1)
F. fasciculata (2) e
F. fasciculata (3)
F. formosa H.B.K. e
F. leonilae F. Miranda (1) e
F. leonilae (2)
F. macdougalii Nash (1) e
F. macdougalii (2)
F. ochoterenae F. Miranda e
F. purpusii T.S. Brandegee e
F. shrevei I.M. Johnston (1) e
F. shrevei (2)
F. splendens Engelm. in Wisliz. (1) e
F. splendens (2)
Family Polemoniaceae
Acanthogilia gloriosa (Brandeg.) Day & Moran
Gen et sp. nov. (Porter, in prep)
Bonplandia geminiflora Cav.
Cantua quericifolia Cav.
Cobaea baiurita Standley
Loeselia glandulosa (Cav.) Don.
V. T. Parker & M. Vasey 0398; SFSU b
V. T. Parker & M. Vasey 0046; SFSU b
University of California, Berkeley Botanic Garden c
Schultheis 1–94; DUKE
Schultheis, n.v. d
Perill f , p.v. g
DBG h 1993-0948 (Henrickson 9113)
C. T. Mason 136413 i
DBG 1939–0072 (G. Lindsey & R. Hoard), p.v.
Perill, p.v.
DBG 1939–0074 (G. Lindsey sn), p.v.
DBG 1977–1444, p.v.
Perill, p.v.
DBG 1974–0221, p.v.
MMBG h 66–035 (Henrickson 2108)
MMBG 66–036 (Henrickson 2164; RSA)
DBG 1993001501 (Henrickson 4236; RSA)
Perill, p.v.
DBG 1965–7944, p.v.
Perill, p.v.
Perill, p.v.
R. Felger j , p.v.
DBG 1974–0154 (R. Engard sn), p.v.
n.v.
J.M. Porter sn; SJNM k
J.M. Porter & K.D. Heil 7987; SJNM k
J.M. Porter & K. Heil 7991; SJNM k (Gilia scabra Brandegee;
see Porter, 1996)
R. Patterson sn; RSA k
R. Patterson sn; RSA k
O. Clarke 293; AZ k
J.M. Porter & C. Campbell 9231; AZ, SJNM k
Family Symplocaceae
Symplocus tinctoria L Hérit Schultheis 7–94; DUKE
Family Tamaricaceae
Tamarix L. sp. Schultheis 19–94; DUKE
a Sequences from all samples are available in GenBank (accession numbers GBANAF084314–GBANAF084361 [the prefix GBAN links the online
version of AJB with GenBank and is not part of the actual GenBank accession number]) with the exception of sequences provided by others,
as noted. The sequence alignment is available from L. Schultheis upon request.
b Sequence provided by S. Markos, University of California at Berkeley (see Markos, 1995).
c Sequence provided by L. Hileman, Harvard University (GenBank accession number GBANAF086828).
d n.v. no voucher
e Genomic DNA received from S. Vetault (University of Arizona) and M. Donoghue (Harvard University), who received material from the source
indicated.
f Perill Material received from Bob Perill, Boojum Unlimited, Tucson, AZ 85743.
g p.v. photo voucher, deposited at UC.
h The botanic garden accession number is given, with a collector and/or collection number and place of deposition following in parentheses if
that information is available. DBG Desert Botanical Gardens of Arizona, Phoenix, AZ. MMBG Mildred Mathias Botanical Garden, Los
Angeles, CA.
i Material taken from the University of Arizona cactus garden. The cited collection, deposited at ARIZ, was also taken from the cactus garden.
j Material received from Dr. Richard Felger, Drylands Institute, Tucson, AZ
k Sequence provided by J.M. Porter, Rancho Santa Ana Botanic Garden (see Porter, 1996).
an Applied Biosystems, Inc. (ABI) PRISM Dye Terminator Cycle Sequencing
Ready Reaction Kit (Foster City, California). Sequencing
products were cleaned with Centri-sep spin columns (Princeton Separations,
Adelphia, New Jersey), and electrophoresed on 4% polyacrylamide
gels using an ABI 377 automated sequencer. Sequences were
visualized using ABI Sequence Navigator software.
Alignment of ITS 1 and ITS 2 sequences was conducted separately.
Sequences from the 5.8S region were not included. Visual alignment of
sequences within Fouquieriaceae was achieved readily. Sequence alignment
with the outgroups was attempted using Clustal V (Higgins, 1994)
with fixed and floating gap penalties of ten. Sequence divergence between
Fouquieriaceae and potential outgroups was substantial, ranging

April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE
581
TABLE 3. Results of test for rate constancy of ITS sequence evolution by nucleotide substitution in Fouquieria with placement of the root in all
possible positions. a
FC f
FC,FFA,FP
FFA,FP
FP
FFA
FF
FO
FL
FB(2)
FS(2)
FSH
FM
FB
FD
FF,FP,FFA,FC
Root b With clock c Without clock 2lnLR d Reject clock e
FC,FFA,FP,FO,FL,FF
FL,FO
FS(2),FSH
FD,FB
1121.91562
1121.59848
1121.91924
1139.11882
1139.25322
1131.54298
1139.42992
1139.41329
1141.49784
1142.50684
1142.54318
1140.37033
1156.27012
1156.24244
1131.54288
1136.29718
1136.28829
1141.50706
1140.33975
1118.27553 7.28
6.65
7.29
41.69
a The tree topology used was derived from a branch-and-bound search of the listed samples with the parsimony criterion (analysis 5 of Materials
and Methods).
b Root the root is placed along the branch between the sample(s) listed and the remaining samples.
c With and without clock ln likelihood values with and without enforcement of a molecular clock.
d LR likelihood ratio; the difference between ln likelihood values with and without enforcement of a clock.
e Rejection of the clock is based on chi-square values (df 10, P 0.05, CV 18.31)
f FC Fouquieria columnaris; FFAF. fasciculata; FPF. purpusii; FFF. formosa; FO F. ochoterenae; FL F. leonilae; FS F.
splendens; FSH F. shrevei; FMF. macdougalii; FBF. burragei; FDF. diguetii. All sequences were from sample (1) as listed in Table
2, unless otherwise indicated.
from 33.8 to 50.8%, as was divergence between potential outgroups.
Because of these high divergences and alignment difficulties, the full
data set was not used. Instead, a data file was created retaining all
sequence information within Fouquieriaceae but with all ambiguously
aligned regions within the outgroup sequences coded as missing data
(‘‘?’’). This is similar to the approach taken by Bruns et al. (1992) to
enable retention of sequence information for those sequences with unambiguous
alignments in otherwise problematic regions.
Phylogenetic analyses were carried out using PAUP version 3.1
(Swofford, 1993) and test versions d55-d59 of PAUP* 4.0 (with permission,
D. L. Swofford, Smithsonian Institution, personal communication).
Multiple analyses were undertaken, both with and without inclusion
of outgroup taxa.
Analysis 1 of Fouquieriaceae taxa (without outgroup sequences) employed
a branch-and-bound search with the parsimony criterion to find
all minimum-length (unrooted) trees. Clade support was assessed with
both bootstrap and decay analyses. The bootstrap employed 100 replicates
with branch-and-bound searches. Decay analysis employed a
branch-and-bound search saving all trees up to six steps longer than the
most parsimonious tree. The trees were filtered for progressively shorter
tree lengths and a strict consensus tree was calculated at each tree
length.
Analysis 2 included all taxa but with partial sequence data, the ambiguously
aligned regions in outgroup sequences being coded as missing
data. Complete heuristic searches were conducted with 100 replicates
of random taxon addition, TBR branch swapping, and MULPARS
in effect to find all minimum-length trees and to obtain an outgroup
rooting of Fouquieriaceae. Clade support was assessed using the ‘‘fast
heuristic’’ bootstrap algorithm available in PAUP*. This algorithm employs
less thorough searches than the standard bootstrap algorithm, but
provides a conservative estimate of bootstrap values (M. J. Sanderson,
U. C. Davis, personal communication).
Analysis 3 employed maximum-likelihood (ML) analysis of ingroup
taxa under the HKY85 model of sequence evolution, both with and
41.95
26.53
42.31
42.28
46.44
48.46
48.53
44.19
75.99
75.93
26.53
36.04
36.02
46.46
44.13
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
without enforcement of a molecular clock to determine whether similar
topologies would be obtained (consistent with rate constancy of ITS
evolution) and to compare with results of the parsimony analyses. The
12 trees from analysis 1 were used as the starting trees for a heuristic
search with TBR branch swapping. Base frequencies were empirically
assessed. The proportion of invariant sites was left at the default setting
of zero. Transition: transversion () parameters and shape () parameters
were estimated on tree number 1 from analysis 1 (6.390220 and
0.193432, respectively), with rate heterogeneity across sites following
a gamma distribution with five rate categories.
Analysis 4 was conducted with ingroup taxa using a parsimony criterion,
but with various weighting schemes, in consideration of the high
transition: transversion () ratio estimated on tree number 1 from analysis
1. Heuristic searches with 20 replicates of random taxon addition,
TBR branch swapping, and MULPARS in effect were conducted. The
transversion:transition weighting schemes employed were 18:10, 2:1,
25:10, 27:10, and 3:1. A weighting scheme of 18:10 is recommended
by Albert and Mishler (1992) for an estimated transition:transversion
() ratio of 6.4 and lambda value of 0.1.
Analysis 5 determined the likelihood values both with and without
enforcement of a molecular clock for one of two trees generated with
a branch-and-bound search of ingroup taxa under the parsimony criterion
where the root placement within the ingroup was forced to fall at
each of all possible branches. Fewer ingroup sequences were included
in this analysis in order to minimize the number of necessary calculations
[FC(1), FFA(1), FP(1), FL(1), FO(1), FM(1), FD(1), FB(1), FB(2),
FS(2), FSH(1), FF; see Table 3 for abbreviations]. Maximum-likelihood
estimates of the transition:transversion () ratio (5.756055) and the
shape parameter (0.147754) were generated from the same unrooted
topology, with rate heterogeneity following a gamma distribution with
five rate categories. Base frequencies were empirically assessed. Evolutionary
rate heterogeneity across lineages was assessed using a global
likelihood-ratio (LR) test (Felsenstein, 1988; Huelsenbeck and Rannala,
1997).

582 AMERICAN JOURNAL OF BOTANY
[Vol. 86
Analysis 6 used simulated data sets to examine the possibility that
root placement within Fouquieriaceae was due to long-branch attraction
(Huelsenbeck, 1997; see Fig. 4 for a phylogram). One hundred data sets
for each of 20 possible root placements within Fouquieriaceae were
simulated using the Siminator program provided by J. P. Huelsenbeck
(University of Rochester). The ingroup topology of input trees was
equivalent to that of tree number 1 in the parsimony analysis of only
ingroup taxa (analysis 1). Nucleotide frequencies and size of each simulated
data set were kept identical to those found in the data set from
analysis 2. Transition:transversion () ratios, shape parameters, and
branch lengths were estimated in PAUP* using a maximum-likelihood
criterion for each of the input trees on which the simulated data sets
were modeled. Heuristic searches of all simulated data sets were conducted
with simple addition of sequences and tree-bisection-reconnection
(TBR) branch swapping under a parsimony criterion. Root placements
within Fouquieriaceae in the resulting trees were tallied by filtering
trees with backbone constraints corresponding to each of 20 possible
root placements.
MacClade version 3.0 was used to explore character evolution (Maddison
and Maddison, 1992). All characters and character state changes
were equally weighted, and all most parsimonious states were reconstructed
at each node.
RESULTS
Analysis 1, of Fouquieriaceae taxa only, resulted in 12
minimum-length trees of 85 steps (CI 0.82; 0.78 without
uninformative characters). Fifty-two out of 487 characters
(postalignment) are potentially informative for parsimony
analysis (26/260 in ITS 1, 26/227 in ITS 2). Absolute
lengths for Fouquieriaceae sequences range from
254 to 260 base pairs (bp) in ITS1 and from 225 to 226
in ITS2. Bootstrap and decay values show strong support
along the branches separating each of the three groups
corresponding to Henrickson’s three subgenera (subg. Idria,
subg. Bronnia, and subg. Fouquieria) (Fig. 1). All
but one set of intraspecific samples also form groups with
strong support. Fouquieria burragei is the only exception,
with one sample falling in the F. diguetii clade, with
robust support, and the other sample falling with weak
support in the sect. Ocotilla clade, with F. splendens and
F. shrevei. Relationships within subg. Fouquieria are not
well resolved. Fouquieria formosa falls sister to the remaining
members of subg. Fouquieria, within which F.
leonilae, F. ochoterenae, and a clade comprising F. diguetii,
F. burragei, F. macdougalii, F.splendens, and F.
shrevei form a trichotomy.
Analysis 2, including the alignable portions of the outgroup
sequences, yielded 240 minimum-length trees of
504 steps (CI 0.71; 0.62 without uninformative characters).
Out of 529 characters (postalignment), 168 are
potentially informative for parsimony analysis (84 from
ITS1; 84 from ITS2). The root for Fouquieriaceae falls
between the two succulent subgenera, with subg. Idria
sister to subg. Bronnia plus subg. Fouquieria. The strict
consensus of all minimum-length trees (Fig. 2) has less
resolution than does the strict consensus tree from analysis
of ingroup taxa alone (Fig. 1). ‘‘Fast heuristic’’ bootstrap
values indicated strong support for subg. Idria and
subg. Bronnia (96 and 92%, respectively) and modest
support for subg. Fouquieria (64%). Aside from intraspecific
groupings, all but one of which are well supported,
there is no resolution within subg. Fouquieria.
Samples of F. burragei again are the exception to intra-
specific groupings, with one sample placed in the F. diguetii
clade and the other placed in an unresolved position.
Only 54 unique ingroup topologies are present
among the 240 trees. Of these 54 topologies, 12 are compatible
with the strict consensus of the 12 parsimony trees
from analysis 1 (Fig. 1) and six are compatible with the
strict consensus of 54 maximum-likelihood trees from
analysis 3 (Fig. 3). The remaining 36 trees have a clade
or grade at the base of subg. Fouquieria consisting either
of F. formosa, F. leonilae, and F. ochoterenae (12 trees)
or of F. burragei (2), F. splendens, and F. shrevei (24
trees). The length of these ingroup topologies after pruning
outgroup taxa is 85 steps for the 12 trees compatible
with the results of analysis 1, and 86 steps for the remaining
42 trees.
Analysis 3 produced 54 maximum-likelihood (ML)
trees without enforcement of a molecular clock (-ln likelihood
1151.36, Fig. 3) and two trees with enforcement
of a molecular clock (-ln likelihood 1156.04). The two
trees obtained from ML analysis of the data set with enforced
constancy of evolutionary rate are among the 54
trees obtained from ML analysis without a clock constraint.
If the rooting indicated in the two clock-enforced
trees is accepted (Fig. 4), then a molecular clock cannot
be rejected at the conventional alpha 0.05 level
(2lnLR 9.36, df 18, P 0.95). The ingroup topology
of the strict consensus of all 54 trees (Fig. 3)
differs from that found in the parsimony-based analysis
of ingroup taxa alone (Fig. 1) but is compatible with a
subset of ingroup topologies found in the parsimonybased
analysis with the inclusion of outgroups (Fig. 2).
The positions of the group composed of F. formosa, F.
leonilae, and F. ochoterenae and the group comprising
F. splendens, F. shrevei and one specimen of F. burragei
are reversed relative to the parsimony-based analysis of
ingroup taxa alone. Using the strict consensus of the 54
ML trees as a backbone constraint for a branch-andbound
parsimony analysis reveals that the ML trees are
only one step longer than the maximum parsimony trees
(86 vs. 85 steps, respectively). This length difference is
insignificant as assessed by the ‘‘compare-2’’ T-PTP test
with 100 replicates of branch-and-bound searches, implemented
in PAUP* (P 0.4). (When described under a
likelihood criterion, the maximum parsimony trees have
-ln likelihood values ranging from 1155.07 to 1156.04.)
In analysis 4, three of the five transversion:transition
weighting schemes employed resulted in the same 12 tree
topologies. The strict consensus of these 12 trees (length
1040 for 25:10 weighting, length 1064 for 27:10 weighting,
and length 110 for 3:1 weighting; all are length 86
under unweighted parsimony) is equivalent to the strict
consensus of the 54 ML trees from analysis 3 (Fig. 3),
except that the weighted-parsimony tree contains less resolution
in the Fouquieria formosa, F. leonilae, and F.
ochoterenae clade, with the three taxa forming a trichotomy.
The search with a 2:1 weighting scheme resulted in
24 trees of length 98, the strict consensus of which lacked
resolution within subg. Fouquieria beyond intraspecific
groupings and the placement of F. burragei (1) with F.
diguetii sequences. Twelve of these 24 trees (length 85
under unweighted parsimony) are compatible with the
strict consensus of trees from the unweighted parsimony
analysis (Fig. 1), six (length 86 under unweighted par-

April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE
583
Fig. 1. The strict consensus of 12 minimum-length unrooted trees of 85 steps (CI 0.82; 0.78 without uninformative characters), resulting
from a branch-and-bound search of ingroup taxa only with the parsimony criterion (analysis 1 in Materials and Methods). Numbers above the
branches indicate the percentage of trees in which the clade appears in 100 bootstrap replicates followed by decay indices (in parentheses). Subgenera
indicated are those recognized by Henrickson (1972).
simony) are compatible with the strict consensus of the
ML trees (Fig. 3), and the remaining six (length 86 under
unweighted parsimony) are compatible with the strict
consensus of the ML trees except that the positions of
Fouquieria formosa and F. ochoterenae are reversed. The
search with an 18:10 weighting scheme resulted in 12
trees of length 954 (length 85 under unweighted parsimony),
the strict consensus of which is equivalent to that
of trees from the unweighted parsimony search (Fig. 1).
Analysis 5 results (Table 3) indicate that clock-like
evolution can be rejected for all root placements within
Fouquieriaceae except (1) between the succulent and
woody taxa (equivalent to the midpoint rooting and to
the rooting seen in Fig. 4), (2) between subg. Idria and
the remaining taxa (equivalent to the outgroup rooting
seen in Fig. 2), or (3) between subg. Bronnia and the
remaining taxa (Table 3).
Analysis 6 suggests that neither the branch between
subg. Idria and remaining taxa, nor between subg. Bronnia
and remaining taxa is sufficiently long to act as a
long-branch attractor. Of the 31 604 trees generated by
analysis of 2000 simulated data sets (100 for each of 20
possible root placements), only in 5% was the root placed
between subg. Idria and the remaining taxa, and in only
6% between subg. Bronnia and the remaining taxa (Table
4). The root was placed between the succulent and woody
taxa in 15% of the trees, indicating that the branch separating
the two groups may indeed act as a long-branch
attractor.
Ingroup ITS topologies in Fouquieriaceae differ de-

584 AMERICAN JOURNAL OF BOTANY
[Vol. 86
Fig. 2. The strict consensus of 240 minimum-length trees of 504 steps (CI 0.71; 0.62 without uninformative characters), resulting from a
heuristic search with 100 replicates of random sequence addition and with the parsimony criterion (analysis 2 in Materials and Methods). Sequences
of ingroup taxa (Fouquieria ssp.) are included in their entirety, while only the alignable portions of outgroup taxa are included, with the unalignable
sites coded as missing data. Numbers above branches indicate the percentage of trees in which the clade appears in a ‘‘fast heuristic’’ bootstrap,
as implemented in PAUP*. Subgenera indicated are those recognized by Henrickson (1972).
pending on the type of analysis conducted (parsimony,
weighted parsimony, maximum likelihood), and on
whether outgroup taxa are included. The trees produced
by weighted parsimony with 25:10, 27:10, and 3:1 transversion:
transition weighting and maximum-likelihood
(ML) analyses are compatible (Fig. 3); future reference
to ML trees in the Discussion also includes these weighted
parsimony trees, while reference to parsimony trees
refers to the unweighted parsimony trees. The trees produced
by parsimony analysis including outgroup taxa
(Fig. 2) contain trees compatible with both parsimony
(Fig. 1) and ML (Fig. 3) analyses of ingroup taxa alone.
When described under a parsimony criterion, the ML
trees are only one step longer than the most parsimonious
trees (86 vs. 85 steps), an insignificant difference (P
0.4, ‘‘compare-2’’ T-TPT test). Differences between the
parsimony and ML trees involve clades with weak support,
as estimated by bootstrap values and decay indices
(on the parsimony trees).
DISCUSSION
Monophyly of subgenera—The three subgenera recognized
by Henrickson (1972) are supported as mono-
phyletic groups based on the ITS data, irrespective of tree
reconstruction criterion (parsimony or ML) or rooting
method (outgroup, midpoint, or clock-consistent approaches).
Within subg. Fouquieria, neither sect. Fouquieria
nor sect. Ocotilla are well-supported monophyletic
groups in the molecular trees. Those taxa corresponding
to sect. Fouquieria are nested within those that
correspond to sect. Ocotilla in the ML trees (Figs. 3–4).
The reverse is the case in the maximum parsimony topologies
(Fig. 1).
Uncertainty in placement of the root in Fouquieriaceae
dictates that if the genus Idria, containing I. (Fouquieria)
columnaris, is recognized (e.g., Wiggins, 1980) then the
genus Bronnia should also be reinstated for F. purpusii
and F. fasciculata to ensure monophyly of Fouquieria
sensu stricto (s.s.). The issue of whether to recognize one
or three genera in Fouquieriaceae is, however, purely a
subjective question of rank (subgenus vs. genus) (see
Cantino, Olmstead, and Wagstaff, 1997).
In his 1972 treatment of Fouquieriaceae, Henrickson
presented a ‘‘putatively phylogenetic system of relationship’’
in which he identified four main groups of taxa,
corresponding to the subgenera and sections in his tax-

April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE
585
Figs. 3–4. Trees resulting from maximum-likelihood analyses of ingroup taxa. 3. (Top) The strict consensus of 54 unrooted trees of -ln
likelihood 1151.36, resulting from a heuristic search of ingroup taxa with the maximum-likelihood criterion and without enforcement of a
molecular clock (analysis 3 of Materials and Methods). Subgenera are those recognized by Henrickson (1972). This tree is equivalent to the strict
consensus of trees produced under parsimony analysis of ingroup taxa with certain differential weightings of transitions vs. transversions, except
that Fouquieria formosa, F. leonilae, and F. ochoterenae form a trichotomy in the parsimony consensus tree (analysis four of Materials and
Methods). 4. (Bottom) A phylogram depiction of the first of two trees of -ln likelihood 1156.04, resulting from a heuristic search of ingroup
taxa with the maximum-likelihood criterion and enforcement of a molecular clock. The two trees produced in this search are among the 54 produced
in the search without enforcement of a molecular clock. The root depicted is equivalent to a midpoint rooting.
onomic scheme (Table 1). The four groups Henrickson
identified were based on cytological data, anatomical
data, and a phenetic analysis of 71 ecological, vegetative,
and reproductive characters. He identified Fouquieria
leonilae and F. ochoterenae as most representative of ancestral
conditions within the family, citing in particular
the following characters as primitive: ‘‘diploid, . . . dorsiventral
leaves, initial single traces to each sepal, ten stamens,
. . . 6(-12) ovules per ovary.’’ Other species within
the family possess dorsiventral or isolateral leaves, a
higher but variable number of sepal traces, ten or more
stamens, and a higher but variable number of ovules.
Henrickson’s proposition that Fouquieria leonilae and F.
ochoterenae possess the greatest number of plesiomorphic
characteristics within the family (1972) is most consistent
with the parsimony trees (Fig. 1), assuming that
out of the three most reasonable root placements, the root
is placed along the branch between subg. Fouquieria and
the remaining taxa. From the lineage represented by F.
leonilae and F. ochoterenae, Henrickson suggested the
derivation of three additional lineages: (1) F. formosa,
(2) the putatively polyploid series of F. macdougalii, F.
diguetii, and F. burragei, and (3) the ocotillos, F. splendens
and F. shrevei. Fouquieria leonilae and F. ochoterenae,
together with F. formosa and the polyploid series,
constitute the dendroid subg. Fouquieria sect. Fouquier-

586 AMERICAN JOURNAL OF BOTANY
[Vol. 86
TABLE 4. Percentage of trees from parsimony analysis of 2000 simulated
data sets (100 for each of 20 examined root positions; see
Materials and Methods) in which the root is placed between the
taxa listed and the remaining taxa.
Taxa %
FB a
FB(2)
FC
FC,FFA,FP,FO,FL,FF
FD
FDFB
FFA
FFA,FP
FF
FL
FL,FO
FO
FS,FSH
FSH
FC,FFA,FP
FP
FS
FM
FF,FC,FP,FFA
FSH,FS,FB(2)
a Refer to Table 3 for abbreviations.
0.0
7.0
5.0
1.0
0.0
8.0
1.0
6.0
5.0
6.0
2.0
7.0
0.7
6.0
15.0
2.0
11.0
6.0
1.0
1.0
ia. Fouquieria splendens and F. shrevei form subg. Fouquieria
sect. Ocotilla. The remaining two groups Henrickson
recognized are the succulent subg. Idria (F. columnaris)
and subg. Bronnia (F. fasciculata and F. purpusii).
The four groups Henrickson identified (1972) are not,
he noted, of equal phenetic distinctiveness relative to
each other. The two succulent subgenera are more distinct
phenetically from each other and from the woody taxa
than are the two woody sections from each other. This
impression is supported by the clock-like molecular trees
in which the three subgenera are on long branches relative
to branches within the subgenera (Fig. 4). Henrickson
pointed out that the succulent subg. Bronnia is phenetically
intermediate between subg. Idria and woody
subg. Fouquieria. His dendrograms indicate a closer phenetic
similarity of subg. Bronnia with subg. Fouquieria
using ecological and vegetative characters, and a closer
similarity with subg. Idria using reproductive characters.
Within subg. Fouquieria, Henrickson noted that F. burragei
is intermediate between sect. Ocotilla and sect.
Fouquieria. This is perhaps a reflection of the hypothesized
parental contribution to F. burragei from both sections
within subg. Fouquieria, a hypothesis supported by
our molecular data.
Outgroups and rooting—Rooting Fouquieriaceae using
the outgroup method was problematic due to the levels
of sequence divergence between Fouquieriaceae and
its potential outgroups. Outgroup sampling in this study
was not aimed at determining the placement of Fouquieriaceae
relative to other angiosperm families, but was
guided by other higher level phylogenetic analyses which
included Fouquieriaceae (Anderberg, 1992; Hufford,
1992; Olmstead et al., 1993). Results of analyses including
only those portions of the outgroup sequences that
were alignable suggest that an ericalean grouping is sister
to Fouquieriaceae. This clade, however, has weak support.
This result is in agreement with interpretations of a
position of Fouquieriaceae close to Ericales (e.g., Anderberg,
1992; Olmstead et al., 1993). The tree shown (Fig.
2) was rooted using Tamarix based on recent evidence
that Tamaricaceae falls close to Caryophyllales, well outside
of Asteridae sensu lato (Lledó et al., 1998).
Using only the alignable portions of the outgroup sequences
placed the root in Fouquieriaceae between Fouquieria
subg. Idria (F. columnaris) and the remaining
species (Fig. 2). A midpoint rooting results in a basal
dichotomy between subg. Fouquieria and both subg. Idria
and subg. Bronnia (equivalent to the root seen in Fig.
4). Both of these root placements as well as a root placement
along the subg. Bronnia branch are consistent with
clock-like evolution of ITS sequences (Table 3). Results
from data sets simulated with various root positions suggest
that neither the subg. Idria branch nor the subg.
Bronnia branch act as long-branch attractors, whereas the
branch between the succulent and woody taxa may attract
other long branches (Table 4). The root placement in trees
resulting from the analyses including outgroup taxa (Fig.
2, analysis 2) does not, therefore, appear to be a result
of long-branch attraction. Because the root placement
within Fouquieriaceae is uncertain, the implications of all
reasonable root positions will be addressed in discussion
of character evolution.
Succulent and woody habits—Whether the ancestral
condition in Fouquieriaceae is succulent or woody is
equivocal, in part due to uncertainty about rooting of the
ingroup tree topology and identity of the closest relatives
of the family. If subg. Idria and subg. Bronnia fall on
different sides of the basal dichotomy in Fouquieriaceae
(e.g., a root as in Fig. 2) then succulence may well be
the ancestral condition in the family, assuming monophyly
of subg. Fouquieria. This would indicate a reversal
to the woody condition in subg. Fouquieria. A transition
from a succulent habit to a woody habit would be unusual,
if not unique. In Cactaceae, the woody Pereskioideae
appear to be basal (Hershkovitz and Zimmer, 1997,
citing others), with succulence perhaps achieved via a
neotenic reduction of wood developmental rates (Altesor,
Silva, and Ezcurra, 1994). Sufficient phylogenetic information
is lacking to assess the direction of life-form evolution
in other groups containing both succulent and
woody members (e.g., Euphorbiaceae, Asclepiadaceae,
and Asteraceae). If the root within Fouquieriaceae falls
as suggested above, a woody or an intermediate ancestral
condition rather than a succulent condition would dictate
independent elaboration of succulence in the three subgenera
of Fouquieria: cortical succulence in subg. Fouquieria
and stem succulence in both subg. Bronnia and
subg. Idria. The woody species of subgenus Fouquieria
possess a cortical network of water storage tissue (Scott,
1932) that is slightly more developed than that of the
stem succulents (Humphrey, 1935; Henrickson, 1969c).
Succulence in both subg. Idria and subg. Bronnia is due
to parenchymatous water storage cells within the xylem
(Humphrey, 1935; Henrickson, 1969c), originating from
both cambial division and division of pre-existing parenchymatous
cells (Henrickson, 1969c). Succulent tissue

April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE
587
is most extensive in F. columnaris (subg. Idria), which
possesses a succulent pith, less extensive in F. purpusii,
and least extensive in F. fasciculata (Henrickson, 1969b,
c, 1972). If subg. Bronnia is sister to subg. Idria (i.e., a
root as in Fig. 4), this would suggest a single origin of
stem succulence in Fouquieriaceae and an equivocal ancestral
habit in the family.
Determining the direction of evolution between ocotillo
and dendroid growth forms relies on resolution of
relationships within subg. Fouquieria. Parsimony analysis
indicates paraphyly of sect. Fouquieria and derivation
of the ocotillo growth form from dendroid forms (Fig.
1). However, ML analyses indicate paraphyly of sect.
Ocotilla and derivation of dendroid growth forms from
ocotillo forms (Fig. 3). More data on relationships among
the members of subg. Fouquieria are needed to resolve
life-form evolution in the group.
Polyploidy—Polyploids within Fouquieria include the
tetraploid F. diguetii (n 24) and the hexaploids F. burragei
(n 36) and F. columnaris (n 36). Henrickson
(1972) suggested that F. diguetii may have descended
directly from ‘‘F. macdougalii stock’’ and that F. burragei
is perhaps an amphiploid derivative of F. diguetii
and another, possibly extinct taxon. Fouquieria macdougalii
and F. diguetii share similar floral and inflorescence
structure (Henrickson, 1969b, 1972). Fouquieria macdougalii
(n 12) is distributed in portions of the Sonoran
desert and adjacent tropical deciduous and thorn forests
in mainland Mexico (Henrickson 1969b, 1972). Fouquieria
diguetii is found in Baja California, but overlaps
with the range of F. macdougalii around Guaymas, in
mainland Mexico (Henrickson, 1969b, 1972). The molecular
data shown here indicate a close relationship between
F. diguetii and F. macdougallii that is consistent
with (but not directly supportive of) Henrickson’s interpretation
of a progenitor–derivative relationship between
F. macdougalii and F. diguetii.
Fouquieria burragei is vegetatively very similar to F.
macdougalii and, especially, F. diguetii, but is very different
in floral characteristics from the two species, with
more open flowers, more numerous stamens, pink to
white corolla pigmentation (vs. red), and more elongate
inflorescences (Henrickson, 1969b, 1972). Samples of F.
burragei fall in two places in the molecular trees: one
with F. diguetii in the parsimony and ML trees and the
other with the ocotillo clade (F. splendens and F. shrevei)
in the parsimony trees or within the ocotillo grade in the
ML trees. Different phylogenetic placements of F. burragei
ITS sequences might be the result of bidirectional
concerted evolution (Wendel, Schnabel, and Seelanen,
1995), i.e., different populations of F. burragei may have
become fixed for different ITS repeat types inherited
from the two parental taxa involved in its putative allopolyploid
origin. This interpretation is consistent with
Henrickson’s (1972) hypothesis that F. diguetii served as
one parent to F. burragei, while a species similar to a
white-flowered F. splendens served as the other. Fouquieria
burragei is distributed along the eastern coast of
lower Baja California (Henrickson, 1969b, 1972), overlapping
in range with F. diguetii. The current range of
F. splendens overlaps with that of F. diguetii on the Baja
peninsula, but does not reach quite as far south as the
northern known range limit of F. burragei (Henrickson,
1969b, 1972). Different subspecies of F. splendens differ
in part in floral pigmentation, ranging from red to creamwhite,
but none of the white forms of F. splendens are
known from Baja.
Henrickson’s (1972) suggestion that a representative
from the Fouquieria purpusii and F. fasciculata lineage
may have served as a parent in the amphiploid derivation
of F. columnaris is neither supported nor refuted here,
but can be reconciled with our results most easily if the
ingroup root falls between the woody and succulent species.
Fouquieria purpusii and F. fasciculata (subg. Bronnia)
share with F. columnaris (subg. Idria) both vegetative
and floral features, including succulent trunks and
white to cream-white decandrous flowers (Henrickson,
1972). Fouquieria columnaris is found in central Baja
California and as a small population at Punta Cirio in
Sonora, Mexico (Henrickson, 1969b, 1972; Humphrey
and Marx, 1980); Fouquieria fasciculata and F. purpusii
are highly restricted, found in pockets of arid tropical
scrub vegetation (Henrickson, 1972) in Hidalgo (F. fasciculata)
and in Puebla and Oaxaca (F. purpusii)(Henrickson,
1969b, 1972). Extensive molecular divergence
between these two species and F. columnaris
corresponds with their highly disjunct distributions in
suggesting that a direct ancestor–descendent relationship
between subg. Bronnia and subg. Idria is untenable. A
possibility that cannot be discounted is that F. columnaris
is an allopolyploid involving a member of subg. Bronnia
and an extinct taxon, with the ITS repeat type sequenced
from F. columnaris derived from the extinct taxon.
Floral features—While Henrickson did not mention
floral pigmentation in his assessment of ancestral features
within the family, a root placement along the branch between
subg. Fouquieria and the remaining taxa (the root
most consistent with Henrickson’s hypothesis that F.
leonilae and F. ochoterenae are best representative of
ancestral conditions within the family) suggests a shift
from white/cream floral pigmentation to red pigmentation
in ML trees (Figs. 3–4), but is equivocal in parsimony
trees (Fig. 1). If subg. Idria and subg. Bronnia fall on
opposite sides of the basal dichotomy, white- to creamcolored
flowers appear ancestral in the family, with a single
derivation of red pigmentation in subg. Fouquieria
and, in the parsimony trees (Fig. 1), a reversal(s) in the
clade containing F. burragei, F. shrevei, and F. splendens.
While shifts in floral pigmentation and shape may be
indicative of pollinator shifts, there does not appear to be
strict conformity to stereotypical pollination syndromes
within Fouquieriaceae. A temporal match between hummingbird
migrations and flowering time in red-flowered
F. splendens has been shown to favorably affect seed set
(Waser, 1979). However, carpenter bees are also important
pollinators and can be the most frequent visitors
(Scott, Buchmann, and O’Rourke, 1993). Similarly, while
15 species of bees have been observed to visit the white/
cream flowered F. columnaris, observed visitors to the
white/cream flowered F. fasciculata include hummingbirds,
bees, and various other insects (Henrickson, 1972).
In general, observations of pollination in Fouquieriaceae
are limited.

588 AMERICAN JOURNAL OF BOTANY
[Vol. 86
Biogeography—Most species of Fouquieriaceae are
endemic to mainland Mexico. Those found outside of
mainland Mexico (i.e., in Baja California and the southwestern
United States) are all polyploid, with the exception
of the widespread Fouquieria splendens. While F.
diguetii (tetraploid) and F. burragei (hexaploid) are
closely related, the two hexaploids F. columnaris and F.
burragei, although geographically proximal to one another,
must represent independent origins of the hexaploid
condition. The overall distribution of Fouquieria species,
with all but one diploid species endemic to mainland
Mexico, suggests a mainland Mexican origin for the family.
The highly localized distribution of many species (e.g.,
F. shrevei, F. leonilae, F. ochoterenae, F. fasciculata, F.
purpusii) may be indicative of relictualism or neoendemism.
Considering the magnitude of the disjunction between
the hexaploid F. columnaris and the morphologically
similar species of subg. Bronnia it seems likely that
considerable extinction has occurred in the history of
Fouquieriaceae. This impression is reinforced by the isolation
of each of the three subgenera on long branches of
the clock-like ITS trees (e.g., Fig. 4).
Concluding remarks—Any discussion of character
evolution depends on an accurate tree topology and an
accurate placement of a root within that topology. Rooting
using an outgroup comparison method is problematic
when sister taxa are too divergent or unknown. Outgroup
taxa that are too divergent from the ingroup effectively
act as random sequences and tend to root an ingroup
topology along the longest branch (Wheeler, 1990). Due
to its uncertain phylogenetic placement and its apparently
high divergence from other angiosperm taxa, rooting
Fouquieriaceae with the outgroup method is problematic,
but this problem is certainly not unique.
Our strategy for identifying a root in Fouquieriaceae
was to employ the outgroup method using only the conservative
(alignable) regions within the outgroup sequences
while retaining all ingroup sequence data (Bruns
et al., 1992), a strategy shown to be preferable to retention
of all data when portions of the data are highly variable
(Smith, 1994). We rejected the possibility that the
root placement resulting from the outgroup method was
confounded by long-branch attraction (Huelsenbeck,
1997). Additionally, we found only three root placements
consistent with clock-like evolution of ITS sequences.
Our discussion of character evolution within Fouquieriaceae
is based on a set of tree topologies and root placements
that appear to be the best hypotheses given the
limitations of our data. Future examination of additional
morphological and molecular evidence may allow further
refinement of our understanding of diversification in this
fascinating family.
LITERATURE CITED
ALBERT, V.A.,AND B. D. MISHLER. 1992. On the rationale and utility
of weighting nucleotide sequence data. Cladistics 8: 73–83.
ALTESOR A., C. SILVA, AND E. EZCURRA. 1994. Allometric neoteny and
the evolution of succulence in cacti. Botanical Journal of the Linnean
Society 114: 282–292.
ANDERBERG, A. A. 1992. The circumscription of the Ericales, and their
cladistic relationships to other families of ‘‘higher’’ dicotyledons.
Systematic Botany 17: 660–675.
BALDWIN, B. G. 1992. Phylogenetic utility of the internal transcribed
spacers of nuclear ribosomal DNA in plants: an example from the
Compositae. Molecular Phylogenetics and Evolution 1: 3–16.
BESSEY, C. E. 1915. The phylogenetic taxonomy of flowering plants.
Annals of the Missouri Botanical Garden 2: 109–164.
BRUNS, T. D., R. VILGALYS, S.M.BARNS, D.GONZALEZ, D.S.HIBBETT,
D. J. LANE, L.SIMON, S.STICKEL, T.M.SZARO, W.G.WEISBURG,
AND M. S. SOGIN. 1992. Evolutionary relationships within the fungi:
analyses of nuclear small subunit rRNA sequences. Molecular
Phylogenetics and Evolution 1: 231–241.
CANTINO, P. D., R. G. OLMSTEAD, AND S. J. WAGSTAFF. 1997. A comparison
of phylogenetic nomenclature with the current system: a
botanical case study. Systematic Biology 46: 313–331.
CRONQUIST, A. 1981. An integrated system of classification of flowering
plants. Columbia University Press, New York, NY.
DAHLGREN, R., S. JENSEN, AND B. NIELSEN. 1976. Iridoid compounds
in Fouquieriaceae and notes on its possible affinities. Botaniska
notiser 129: 207–212.
DECANDOLLE, A. P. 1828. Prodromus Systematis Naturalis Regni Vegetabilis,
vol. 3. Treuttel and Wurtz, Paris.
DOYLE, J. J., AND J. L. DOYLE. 1987. A rapid DNA isolation procedure
for small quantities of fresh leaf tissue. Phytochemical Bulletin 19:
11–15.
FELSENSTEIN, J. 1988. Phylogenies from molecular sequences: inferences
and reliability. Annual Review of Genetics 22: 521–565.
HENRICKSON, J. 1967. Pollen morphology of the Fouquieriaceae. Aliso
6: 137–160.
———. 1969a. Anatomy of periderm and cortex of Fouquieriaceae.
Aliso 7: 97–126.
———. 1969b. An introduction to the Fouquieriaceae. Cactus and
Succulent Journal 61: 97–106.
———. 1969c. The succulent Fouquierias. Cactus and Succulent Journal
61: 178–184.
———. 1972. A taxonomic revision of Fouquieriaceae. Aliso 7: 439–
537.
HERSHKOVITZ, M.A., AND E. A. ZIMMER. 1997. On the evolutionary
origins of the cacti. Taxon 46: 217–232.
HIGGINS, D. G. 1994. CLUSTAL V: multiple alignment of DNA and
protein sequences. In A. M. Griffin and H. G. Griffin [eds.], Computer
analysis of sequence data, part II. Methods in molecular biology,
vol. 25, 307–318. Humana Press, Totowa, NJ.
HUELSENBECK, J. P. 1997. Is the Felsenstein zone a fly trap? Systematic
Biology 46: 69–74.
———, AND B. RANNALA. 1997. Phylogenetic methods come of age:
testing hypotheses in an evolutionary context. Science 276: 227–
232.
HUFFORD, L. 1992. Rosidae and their relationships to other nonmagnoliid
dicotyledons: a phylogenetic analysis using morphological
and chemical data. Annals of the Missouri Botanical Garden.79:
218–248.
HUMBOLDT, A., A. BONPLAND, AND C. S. KUNTH. 1823. Nova Genera
et Species Plantarum, vol. 6. Gide, Paris.
HUMPHREY, R. R. 1935. A study of Idria columnaris and Fouquieria
splendens. American Journal of Botany 22: 184–206.
———, AND D. B. MARX. 1980. Distribution of the boojum tree (Idria
columnaris) on the coast of Sonora, Mexico as influenced by climate.
Desert Plants 2: 183–187.
HUTCHINSON, J. 1926. The families of flowering plants I. dicotyledons.
MacMillan and Co., London.
JENSEN, S.R.,AND B. J. NIELSEN. 1982. Iridoid glucosides in Fouquieriaceae.
Phytochemistry 21: 1623–1629.
LLEDÓ, M. D., M. B. CRESPO, K.M.CAMERON, M.F.FAY, AND M. W.
CHASE. 1998. Systematics of Plumbaginaceae based upon cladistic
analysis of rbcL sequence data. Systematic Botany 23: 21–29.
MABRY, T. J. 1977. The order Centrospermae. Annals of the Missouri
Botanical Garden 64: 210–220.
MADDISON, W.P.,AND D. R. MADDISON. 1992. MacClade: analysis of
phylogeny and character evolution, ver. 3.0. Sinauer, Sunderland,
MA.
MARKOS, S. E. 1995. Phylogeny of Arctostaphylos hookeri (Ericaceae)

April 1999] SCHULTHEIS AND BALDWIN—MOLECULAR PHYLOGENETICS OF FOUQUIERIACEAE
589
based on ribosomal DNA sequence from the ITS region. M. A.
thesis, San Francisco State University. San Francisco, CA.
MORTON, C. M., M. W. CHASE, K.A.KRON, AND S. M. SWENSEN. 1996.
A molecular evaluation of the monophyly of the order Ebenales
based upon rbcL sequence data. Systematic Botany 21: 567–586.
NASH, G. V. 1903. A revision of the family Fouquieriaceae. Bulletin
of the Torrey Botanic Club 30: 449–459.
OLMSTEAD, R. G., B. BREMER, K.SCOTT, AND J. D. PALMER. 1993. A
parsimony analysis of the Asteridae sensu lato based on rbcL sequences.
Annals of the Missouri Botanical Garden 80: 700–722.
PORTER, J. M. 1996. Phylogeny of Polemoniaceae based on nuclear
ribosomal internal transcribed spacer DNA sequences. Aliso 15:
57–77.
ROEMER, J.J.,AND J. A. SCHULTES. 1819. Systema Vegetabilium, vol.
4. Stuttgardt.
SCOGIN, R. 1977. Anthocyanins of Fouquieriaceae. Biochemical Systematics
and Ecology 5: 265–267.
SCOTT, F. M. 1932. Some features of the anatomy of Fouquieria splendens.
American Journal of Botany 19: 673–678.
SCOTT, P. E., S. L. BUCHMANN, AND M. K. O’ROURKE. 1993. Evidence
for mutualism between a flower-piercing carpenter bee and ocotillo:
use of pollen and nectar by nesting bees. Ecological Entomology
18: 234–240.
SMITH, A. B. 1994. Rooting molecular trees: problems and strategies.
Biological Journal of the Linnean Society 51: 279–292.
SWOFFORD, D. L. 1993. PAUP: phylogenetic analysis using parsimony,
version 3.1. Computer program distributed by the Illinois Natural
History Survey, Champaign, IL.
TAKHTAJAN, A. 1969. Flowering plants: origin and dispersal. Smithsonian
Institution Press, Washington DC.
THORNE, R. F. 1969. Synopsis of a putatively phylogenetic classification
of the flowering plants. Aliso 6: 57–66.
WASER, N. M. 1979. Pollinator availability as a determinant of flowering
time in ocotillo (Fouquieria splendens). Oecologia 39: 107–
121,
WENDEL, J. F., A. SCHNABEL, AND T. SEELANAN. 1995. Bidirectional
interlocus concerted evolution following allopolyploid speciation
in cotton (Gossypium). Proceedings of the National Academy of
Sciences, USA 92: 280–284.
WHEELER, W. C. 1990. Nucleic acid sequence phylogeny and random
outgroups. Cladistics 6: 363–367.
WHITE, T. J., T. BRUNS, S.LEE, AND J. TAYLOR. 1990. Amplification
and direct sequencing of fungal ribosomal RNA genes for phylogenetics.
In M. Innis, D. Gelfand, J. Sninsky, and T. White [eds.],
PCR protocols: a guide to methods and applications, 315–322. Academic
Press, San Diego, CA.
WIGGINS, I. L. 1980. Flora of Baja California. Stanford University
Press, Stanford, CA.